Bump version for release.

SECURITY: Add timeouts to all monitoring hooks to prevent hangs on network requests.
SECURITY: For the "spot" check, use a more secure source of randomness when selecting paths to check.

.bandit

@@ -0,0 +1,3 @@
+[bandit]
+exclude=tests
+skips=S105,S404

.gitea/workflows/build.yaml

@@ -3,6 +3,8 @@ run-name: ${{ gitea.actor }} is building
 on:
   push:
     branches: [main]
+  pull_request:
+    branches: [main]
 
 jobs:
   test:
@@ -15,6 +17,7 @@ jobs:
   docs:
     needs: [test]
     runs-on: host
+    if: gitea.event_name == 'push'
     env:
       IMAGE_NAME: projects.torsion.org/borgmatic-collective/borgmatic:docs
 

.github/pull_request_template.md

@@ -0,0 +1,12 @@
+## Hold up, GitHub users
+
+Thanks for your contribution! 
+
+If you're submitting this PR on **projects.torsion.org**: Please ignore the following. You can safely delete this text before submitting your PR.
+
+If you're submitting this PR on **GitHub**: Unfortunately, we don't use GitHub pull requests to manage code contributions to this repository (and GitHub doesn't have any way to disable pull requests entirely). Instead, please see:
+
+https://torsion.org/borgmatic/#contributing
+
+... which provides full instructions on how to submit pull requests. You can even use your GitHub account to login.
+

NEWS

@@ -1,3 +1,192 @@
+2.0.7
+ * #1032: Fix a bug in which a Borg archive gets created even when a database hook fails.
+ * #1043: Support Btrfs subvolume paths in "source_directories" even when the subvolume is mounted
+   elsewhere.
+ * #1048: Ignore Btrfs subvolumes whose read-only status can't be determined.
+ * #1083: Add "debug_passphrase"/"display_passphrase" options and a "{unixtime}" placeholder in
+   support of Borg 2 features.
+ * #1099: Clarify documentation on command hooks order of execution.
+ * #1100: Fix a bug in which "borg --version" failing isn't considered a "fail" state in a command
+   hook.
+ * #1108: Fix a bug in which quoted "extra_borg_options" values containing spaces are passed to Borg
+   incorrectly.
+ * #1108: Add a "--comment" flag to the "create" action for creating an archive with a comment.
+ * Use the Bandit security analysis tool when running tests.
+ * SECURITY: Add timeouts to all monitoring hooks to prevent hangs on network requests, e.g. due to
+   a compromised monitoring server holding requests open.
+ * SECURITY: For the "spot" check, use a more secure source of randomness when selecting paths to
+   check.
+
+2.0.6
+ * #1068: Fix a warning from LVM about leaked file descriptors.
+ * #1086: Fix for the "spot" check breaking when the "--progress" flag is used.
+ * #1089: Fix for the "spot" check erroring when a checksum command errors.
+ * #1091: Fix for the "config generate" action generating invalid configuration when upgrading
+   deprecated command hooks.
+ * #1093: Fix for the LVM hook erroring when the "--dry-run" flag is used.
+ * #1094: Fix incorrect documentation about customizing Borg exit codes:
+   https://torsion.org/borgmatic/docs/how-to/customize-warnings-and-errors/
+ * #1095: Fix for the "spot" check's "xxh64sum_command" option erroring on commands containing
+   spaces.
+ * Add support for Borg 2's "s3:" and "b2:" repository URLs, so you can backup to S3 or B2 cloud
+   storage services even without using Rclone.
+ * During the "spot" check, truncate log messages containing many file paths.
+
+2.0.5
+ * #1033: Add a "password_transport" option to the MariaDB and MySQL database hooks for customizing
+   how borgmatic transmits passwords to the database client.
+ * #1078: Add "keep_3monthly" and "keep_13weekly" options for customizing "prune" action archive
+   retention.
+ * #1078: Add a "use_chunks_archive" option for controlling whether Borg uses its chunks cache
+   directory.
+ * #1078: For the "compact" action, pass "--dry-run" through to Borg.
+ * #1085: Fix a regression in which the default monitoring verbosity is 0 (warnings only) instead of
+   1 (info about steps borgmatic is taking). This prevented logs from showing up in monitoring
+   services like Healthchecks unless you had an explicit monitoring verbosity set.
+ * Move Mastodon social hosting from Fosstodon to FLOSS.social: https://floss.social/@borgmatic
+ * The borgmatic project no longer accepts pull requests on GitHub. But see
+   https://torsion.org/borgmatic/#contributing for how you can still submit pull requests. You can
+   even use your GitHub account to login.
+
+2.0.4
+ * #1072: Fix path rewriting for non-root patterns in the ZFS, Btrfs, and LVM hooks.
+ * #1073: Clarify the documentation about when an "after: error" command hook runs and how it
+   differs from other hooks:
+   https://torsion.org/borgmatic/docs/how-to/add-preparation-and-cleanup-steps-to-backups/
+ * #1075: Fix an incorrect warning about Borg placeholders being unsupported in a command hook.
+ * #1080: If the exact same "everything" command hook is present in multiple configuration files,
+   only run it once.
+
+2.0.3
+ * #1065: Fix a regression in monitoring hooks in which an error pinged the finish state instead of
+   the fail state.
+ * #1066: Add a "states" option to command hooks, so you can optionally skip an "after" hook if
+   borgmatic encounters an error.
+ * #1071: Fix an error in the LVM hook when removing a snapshot directory.
+
+2.0.2
+ * #1035: Document potential performance issues and workarounds with the ZFS, Btrfs, and LVM hooks:
+   https://torsion.org/borgmatic/docs/how-to/snapshot-your-filesystems/
+ * #1053: Display a nicer error message when the "recreate" action encounters an archive that
+   already exists.
+ * #1059: Fix a regression in which soft failure exit codes in command hooks were not respected.
+ * #1060: Fix action command hooks getting run too many times when multiple borgmatic actions are
+   executed (implicitly or explicitly).
+ * #1060: Don't run action command hooks for actions listed in the "skip_actions" option.
+ * #1062: Fix a regression that broke environment variable interpolation.
+ * #1063: List the configured "when" action names in the log entries for command hooks.
+
+2.0.1
+ * #1057: Fix argument parsing to avoid using Python 3.12+ string features. Now borgmatic will
+   work with Python 3.9, 3.10, and 3.11 again.
+
+2.0.0
+ * TL;DR: More flexible, completely revamped command hooks. All configuration options settable on
+   the command-line. New configuration options for many command-line flags (including verbosity!).
+   New "key import" and "recreate" actions. Almost everything is backwards compatible—but mind those
+   deprecation warnings!
+ * #262: Add a "default_actions" option that supports disabling default actions when borgmatic is
+   run without any command-line arguments.
+ * #303: Deprecate the "--override" flag in favor of direct command-line flags for every borgmatic
+   configuration option. See the documentation for more information:
+   https://torsion.org/borgmatic/docs/how-to/make-per-application-backups/#configuration-overrides
+ * #303: Add configuration options that serve as defaults for some (but not all) command-line
+   action flags. For example, each entry in "repositories:" now has an "encryption" option that
+   applies to the "repo-create" action, serving as a default for the "--encryption" flag. See the
+   documentation for more information: https://torsion.org/borgmatic/docs/reference/configuration/
+ * #345: Add a "key import" action to import a repository key from backup.
+ * #422: Add home directory expansion to file-based and KeePassXC credential hooks.
+ * #610: Add a "recreate" action for recreating archives, for instance for retroactively excluding
+   particular files from existing archives.
+ * #790, #821: Deprecate all "before_*", "after_*" and "on_error" command hooks in favor of more
+   flexible "commands:". See the documentation for more information:
+   https://torsion.org/borgmatic/docs/how-to/add-preparation-and-cleanup-steps-to-backups/
+ * #790: BREAKING: For both new and deprecated command hooks, run a configured "after" hook even if
+   an error occurs first. This allows you to perform cleanup steps that correspond to "before"
+   preparation commands—even when something goes wrong.
+ * #790: BREAKING: Run all command hooks (both new and deprecated) respecting the
+   "working_directory" option if configured, meaning that hook commands are run in that directory.
+ * #793: Add configuration options for all verbosity and logging flags, so you don't have to set
+   them on the command-line.
+ * #836: Add a custom command option for the SQLite hook.
+ * #837: Add custom command options for the MongoDB hook.
+ * #1010: When using Borg 2, don't pass the "--stats" flag to "borg prune".
+ * #1020: Document a database use case involving a temporary database client container:
+   https://torsion.org/borgmatic/docs/how-to/backup-your-databases/#database-containers
+ * #1037: Fix an error with the "extract" action when both a remote repository and a
+   "working_directory" are used.
+ * #1044: Fix an error in the systemd credential hook when the credential name contains a "."
+   character.
+ * #1047: Add "key-file" and "yubikey" options to the KeePassXC credential hook.
+ * #1048: Fix a "no such file or directory" error in ZFS, Btrfs, and LVM hooks with nested
+   directories that reside on separate devices/filesystems.
+ * #1050: Fix a failure in the "spot" check when the archive contains a symlink.
+ * #1051: Add configuration filename to the "Successfully ran configuration file" log message.
+
+1.9.14
+ * #409: With the PagerDuty monitoring hook, send borgmatic logs to PagerDuty so they show up in the
+   incident UI. See the documentation for more information:
+   https://torsion.org/borgmatic/docs/how-to/monitor-your-backups/#pagerduty-hook
+ * #936: Clarify Zabbix monitoring hook documentation about creating items:
+   https://torsion.org/borgmatic/docs/how-to/monitor-your-backups/#zabbix-hook
+ * #1017: Fix a regression in which some MariaDB/MySQL passwords were not escaped correctly.
+ * #1021: Fix a regression in which the "exclude_patterns" option didn't expand "~" (the user's
+   home directory). This fix means that all "patterns" and "patterns_from" also now expand "~".
+ * #1023: Fix an error in the Btrfs hook when attempting to snapshot a read-only subvolume. Now,
+   read-only subvolumes are ignored since Btrfs can't actually snapshot them.
+
+1.9.13
+ * #975: Add a "compression" option to the PostgreSQL database hook.
+ * #1001: Fix a ZFS error during snapshot cleanup.
+ * #1003: In the Zabbix monitoring hook, support Zabbix 7.2's authentication changes.
+ * #1009: Send database passwords to MariaDB and MySQL via anonymous pipe, which is more secure than
+   using an environment variable.
+ * #1013: Send database passwords to MongoDB via anonymous pipe, which is more secure than using
+   "--password" on the command-line!
+ * #1015: When ctrl-C is pressed, more strongly encourage Borg to actually exit.
+ * Add a "verify_tls" option to the Uptime Kuma monitoring hook for disabling TLS verification.
+ * Add "tls" options to the MariaDB and MySQL database hooks to enable or disable TLS encryption
+   between client and server.
+
+1.9.12
+ * #1005: Fix the credential hooks to avoid using Python 3.12+ string features. Now borgmatic will
+   work with Python 3.9, 3.10, and 3.11 again.
+
+1.9.11
+ * #795: Add credential loading from file, KeePassXC, and Docker/Podman secrets. See the
+   documentation for more information:
+   https://torsion.org/borgmatic/docs/how-to/provide-your-passwords/
+ * #996: Fix the "create" action to omit the repository label prefix from Borg's output when
+   databases are enabled.
+ * #998: Send the "encryption_passphrase" option to Borg via an anonymous pipe, which is more secure
+   than using an environment variable.
+ * #999: Fix a runtime directory error from a conflict between "extra_borg_options" and special file
+   detection.
+ * #1001: For the ZFS, Btrfs, and LVM hooks, only make snapshots for root patterns that come from
+   a borgmatic configuration option (e.g. "source_directories")—not from other hooks within
+   borgmatic.
+ * #1001: Fix a ZFS/LVM error due to colliding snapshot mount points for nested datasets or logical
+   volumes.
+ * #1001: Don't try to snapshot ZFS datasets that have the "canmount=off" property.
+ * Fix another error in the Btrfs hook when a subvolume mounted at "/" is configured in borgmatic's
+   source directories.
+
+1.9.10
+ * #966: Add a "{credential ...}" syntax for loading systemd credentials into borgmatic
+   configuration files. See the documentation for more information:
+   https://torsion.org/borgmatic/docs/how-to/provide-your-passwords/
+ * #987: Fix a "list" action error when the "encryption_passcommand" option is set.
+ * #987: When both "encryption_passcommand" and "encryption_passphrase" are configured, prefer
+   "encryption_passphrase" even if it's an empty value.
+ * #988: With the "max_duration" option or the "--max-duration" flag, run the archives and
+   repository checks separately so they don't interfere with one another. Previously, borgmatic
+   refused to run checks in this situation.
+ * #989: Fix the log message code to avoid using Python 3.10+ logging features. Now borgmatic will
+   work with Python 3.9 again.
+ * Capture and delay any log records produced before logging is fully configured, so early log
+   records don't get lost.
+ * Add support for Python 3.13.
+
 1.9.9
  * #635: Log the repository path or label on every relevant log message, not just some logs.
  * #961: When the "encryption_passcommand" option is set, call the command once from borgmatic to
@@ -173,7 +362,7 @@
    paths when a "working_directory" is set.
  * #906: Add documentation details for how to run custom database dump commands using binaries from
    running containers:
-   https://torsion.org/borgmatic/docs/how-to/backup-your-databases/#containers
+   https://torsion.org/borgmatic/docs/how-to/backup-your-databases/#database-containers
  * Fix a regression in which the "color" option had no effect.
  * Add a recent contributors section to the documentation, because credit where credit's due! See:
    https://torsion.org/borgmatic/#recent-contributors
@@ -229,7 +418,7 @@
  * Fix handling of the NO_COLOR environment variable to ignore an empty value.
  * Add documentation about backing up containerized databases by configuring borgmatic to exec into
    a container to run a dump command:
-   https://torsion.org/borgmatic/docs/how-to/backup-your-databases/#containers
+   https://torsion.org/borgmatic/docs/how-to/backup-your-databases/#database-containers
 
 1.8.9
  * #311: Add custom dump/restore command options for MySQL and MariaDB.
@@ -499,7 +688,7 @@
    at the command-line. See the configuration reference for more information:
    https://torsion.org/borgmatic/docs/reference/configuration/
  * #649: Add documentation on backing up a database running in a container:
-   https://torsion.org/borgmatic/docs/how-to/backup-your-databases/#containers
+   https://torsion.org/borgmatic/docs/how-to/backup-your-databases/#database-containers
  * #655: Fix error when databases are configured and a source directory doesn't exist.
  * Add code style plugins to enforce use of Python f-strings and prevent single-letter variables.
    To join in the pedantry, refresh your test environment with "tox --recreate".

README.md

@@ -40,8 +40,10 @@ checks:
       frequency: 2 weeks
 
 # Custom preparation scripts to run.
-before_backup:
-    - prepare-for-backup.sh
+commands:
+    - before: action
+      when: [create]
+      run: [prepare-for-backup.sh]
 
 # Databases to dump and include in backups.
 postgresql_databases:
@@ -56,6 +58,8 @@ borgmatic is powered by [Borg Backup](https://www.borgbackup.org/).
 
 ## Integrations
 
+### Data
+
 <a href="https://www.postgresql.org/"><img src="docs/static/postgresql.png" alt="PostgreSQL" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
 <a href="https://www.mysql.com/"><img src="docs/static/mysql.png" alt="MySQL" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
 <a href="https://mariadb.com/"><img src="docs/static/mariadb.png" alt="MariaDB" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
@@ -65,6 +69,11 @@ borgmatic is powered by [Borg Backup](https://www.borgbackup.org/).
 <a href="https://btrfs.readthedocs.io/"><img src="docs/static/btrfs.png" alt="Btrfs" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
 <a href="https://sourceware.org/lvm2/"><img src="docs/static/lvm.png" alt="LVM" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
 <a href="https://rclone.org"><img src="docs/static/rclone.png" alt="rclone" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
+<a href="https://www.borgbase.com/?utm_source=borgmatic"><img src="docs/static/borgbase.png" alt="BorgBase" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
+
+
+### Monitoring
+
 <a href="https://healthchecks.io/"><img src="docs/static/healthchecks.png" alt="Healthchecks" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
 <a href="https://uptime.kuma.pet/"><img src="docs/static/uptimekuma.png" alt="Uptime Kuma" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
 <a href="https://cronitor.io/"><img src="docs/static/cronitor.png" alt="Cronitor" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
@@ -76,7 +85,14 @@ borgmatic is powered by [Borg Backup](https://www.borgbackup.org/).
 <a href="https://github.com/caronc/apprise/wiki"><img src="docs/static/apprise.png" alt="Apprise" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
 <a href="https://www.zabbix.com/"><img src="docs/static/zabbix.png" alt="Zabbix" height="40px" style="margin-bottom:20px; margin-right:20px;"></a>
 <a href="https://sentry.io/"><img src="docs/static/sentry.png" alt="Sentry" height="40px" style="margin-bottom:20px; margin-right:20px;"></a>
-<a href="https://www.borgbase.com/?utm_source=borgmatic"><img src="docs/static/borgbase.png" alt="BorgBase" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
+
+
+### Credentials
+
+<a href="https://systemd.io/"><img src="docs/static/systemd.png" alt="Sentry" height="40px" style="margin-bottom:20px; margin-right:20px;"></a>
+<a href="https://www.docker.com/"><img src="docs/static/docker.png" alt="Docker" height="40px" style="margin-bottom:20px; margin-right:20px;"></a>
+<a href="https://podman.io/"><img src="docs/static/podman.png" alt="Podman" height="40px" style="margin-bottom:20px; margin-right:20px;"></a>
+<a href="https://keepassxc.org/"><img src="docs/static/keepassxc.png" alt="Podman" height="40px" style="margin-bottom:20px; margin-right:20px;"></a>
 
 
 ## Getting started
@@ -122,7 +138,8 @@ issues.
 
 ### Social
 
-Follow [borgmatic on Mastodon](https://fosstodon.org/@borgmatic).
+Follow <a rel="me" href="https://floss.social/@borgmatic">borgmatic on
+Mastodon</a>.
 
 
 ### Chat
@@ -143,21 +160,21 @@ Other questions or comments? Contact
 ### Contributing
 
 borgmatic [source code is
-available](https://projects.torsion.org/borgmatic-collective/borgmatic) and is also mirrored
-on [GitHub](https://github.com/borgmatic-collective/borgmatic) for convenience.
+available](https://projects.torsion.org/borgmatic-collective/borgmatic) and also
+has a read-only mirror on
+[GitHub](https://github.com/borgmatic-collective/borgmatic) for convenience.
 
 borgmatic is licensed under the GNU General Public License version 3 or any
 later version.
 
-If you'd like to contribute to borgmatic development, please feel free to
-submit a [Pull
-Request](https://projects.torsion.org/borgmatic-collective/borgmatic/pulls) or
+If you'd like to contribute to borgmatic development, please feel free to submit
+a [pull
+request](https://projects.torsion.org/borgmatic-collective/borgmatic/pulls) or
 open an
 [issue](https://projects.torsion.org/borgmatic-collective/borgmatic/issues) to
 discuss your idea. Note that you'll need to
 [register](https://projects.torsion.org/user/sign_up?invite_code=borgmatic)
-first. We also accept Pull Requests on GitHub, if that's more your thing. In
-general, contributions are very welcome. We don't bite!
+first. In general, contributions are very welcome. We don't bite!
 
 Also, please check out the [borgmatic development
 how-to](https://torsion.org/borgmatic/docs/how-to/develop-on-borgmatic/) for

borgmatic/actions/change_passphrase.py

@@ -16,7 +16,7 @@ def run_change_passphrase(
     remote_path,
 ):
     '''
-    Run the "key change-passprhase" action for the given repository.
+    Run the "key change-passphrase" action for the given repository.
     '''
     if (
         change_passphrase_arguments.repository is None

borgmatic/actions/check.py

@@ -6,9 +6,11 @@ import logging
 import os
 import pathlib
 import random
+import shlex
 import shutil
+import textwrap
 
-import borgmatic.actions.create
+import borgmatic.actions.pattern
 import borgmatic.borg.check
 import borgmatic.borg.create
 import borgmatic.borg.environment
@@ -170,7 +172,7 @@ def filter_checks_on_frequency(
 
             if calendar.day_name[datetime_now().weekday()] not in days:
                 logger.info(
-                    f"Skipping {check} check due to day of the week; check only runs on {'/'.join(days)} (use --force to check anyway)"
+                    f"Skipping {check} check due to day of the week; check only runs on {'/'.join(day.title() for day in days)} (use --force to check anyway)"
                 )
                 filtered_checks.remove(check)
                 continue
@@ -372,9 +374,13 @@ def collect_spot_check_source_paths(
         borgmatic.borg.create.make_base_create_command(
             dry_run=True,
             repository_path=repository['path'],
-            config=config,
-            patterns=borgmatic.actions.create.process_patterns(
-                borgmatic.actions.create.collect_patterns(config),
+            # Omit "progress" because it interferes with "list_details".
+            config=dict(
+                {option: value for option, value in config.items() if option != 'progress'},
+                list_details=True,
+            ),
+            patterns=borgmatic.actions.pattern.process_patterns(
+                borgmatic.actions.pattern.collect_patterns(config),
                 working_directory,
             ),
             local_borg_version=local_borg_version,
@@ -382,17 +388,15 @@ def collect_spot_check_source_paths(
             borgmatic_runtime_directory=borgmatic_runtime_directory,
             local_path=local_path,
             remote_path=remote_path,
-            list_files=True,
             stream_processes=stream_processes,
         )
     )
-    borg_environment = borgmatic.borg.environment.make_environment(config)
     working_directory = borgmatic.config.paths.get_working_directory(config)
 
     paths_output = borgmatic.execute.execute_command_and_capture_output(
         create_flags + create_positional_arguments,
         capture_stderr=True,
-        extra_environment=borg_environment,
+        environment=borgmatic.borg.environment.make_environment(config),
         working_directory=working_directory,
         borg_local_path=local_path,
         borg_exit_codes=config.get('borg_exit_codes'),
@@ -456,7 +460,7 @@ def collect_spot_check_archive_paths(
     )
 
 
-SAMPLE_PATHS_SUBSET_COUNT = 10000
+SAMPLE_PATHS_SUBSET_COUNT = 5000
 
 
 def compare_spot_check_hashes(
@@ -482,12 +486,14 @@ def compare_spot_check_hashes(
     sample_count = max(
         int(len(source_paths) * (min(spot_check_config['data_sample_percentage'], 100) / 100)), 1
     )
-    source_sample_paths = tuple(random.sample(source_paths, sample_count))
+    source_sample_paths = tuple(random.SystemRandom().sample(source_paths, sample_count))
     working_directory = borgmatic.config.paths.get_working_directory(config)
-    existing_source_sample_paths = {
+    hashable_source_sample_path = {
         source_path
         for source_path in source_sample_paths
-        if os.path.exists(os.path.join(working_directory or '', source_path))
+        for full_source_path in (os.path.join(working_directory or '', source_path),)
+        if os.path.exists(full_source_path)
+        if not os.path.islink(full_source_path)
     }
     logger.debug(
         f'Sampling {sample_count} source paths (~{spot_check_config["data_sample_percentage"]}%) for spot check'
@@ -508,9 +514,12 @@ def compare_spot_check_hashes(
             break
 
         hash_output = borgmatic.execute.execute_command_and_capture_output(
-            (spot_check_config.get('xxh64sum_command', 'xxh64sum'),)
+            tuple(
+                shlex.quote(part)
+                for part in shlex.split(spot_check_config.get('xxh64sum_command', 'xxh64sum'))
+            )
             + tuple(
-                path for path in source_sample_paths_subset if path in existing_source_sample_paths
+                path for path in source_sample_paths_subset if path in hashable_source_sample_path
             ),
             working_directory=working_directory,
         )
@@ -518,11 +527,13 @@ def compare_spot_check_hashes(
         source_hashes.update(
             **dict(
                 (reversed(line.split('  ', 1)) for line in hash_output.splitlines()),
-                # Represent non-existent files as having empty hashes so the comparison below still works.
+                # Represent non-existent files as having empty hashes so the comparison below still
+                # works. Same thing for filesystem links, since Borg produces empty archive hashes
+                # for them.
                 **{
                     path: ''
                     for path in source_sample_paths_subset
-                    if path not in existing_source_sample_paths
+                    if path not in hashable_source_sample_path
                 },
             )
         )
@@ -560,6 +571,9 @@ def compare_spot_check_hashes(
     return tuple(failing_paths)
 
 
+MAX_SPOT_CHECK_PATHS_LENGTH = 1000
+
+
 def spot_check(
     repository,
     config,
@@ -627,9 +641,12 @@ def spot_check(
     logger.debug(f'{len(archive_paths)} total archive paths for spot check')
 
     if len(source_paths) == 0:
-        logger.debug(
-            f'Paths in latest archive but not source paths: {", ".join(set(archive_paths)) or "none"}'
+        truncated_archive_paths = textwrap.shorten(
+            ', '.join(set(archive_paths)) or 'none',
+            width=MAX_SPOT_CHECK_PATHS_LENGTH,
+            placeholder=' ...',
         )
+        logger.debug(f'Paths in latest archive but not source paths: {truncated_archive_paths}')
         raise ValueError(
             'Spot check failed: There are no source paths to compare against the archive'
         )
@@ -640,11 +657,21 @@ def spot_check(
 
     if count_delta_percentage > spot_check_config['count_tolerance_percentage']:
         rootless_source_paths = set(path.lstrip(os.path.sep) for path in source_paths)
-        logger.debug(
-            f'Paths in source paths but not latest archive: {", ".join(rootless_source_paths - set(archive_paths)) or "none"}'
+        truncated_exclusive_source_paths = textwrap.shorten(
+            ', '.join(rootless_source_paths - set(archive_paths)) or 'none',
+            width=MAX_SPOT_CHECK_PATHS_LENGTH,
+            placeholder=' ...',
         )
         logger.debug(
-            f'Paths in latest archive but not source paths: {", ".join(set(archive_paths) - rootless_source_paths) or "none"}'
+            f'Paths in source paths but not latest archive: {truncated_exclusive_source_paths}'
+        )
+        truncated_exclusive_archive_paths = textwrap.shorten(
+            ', '.join(set(archive_paths) - rootless_source_paths) or 'none',
+            width=MAX_SPOT_CHECK_PATHS_LENGTH,
+            placeholder=' ...',
+        )
+        logger.debug(
+            f'Paths in latest archive but not source paths: {truncated_exclusive_archive_paths}'
         )
         raise ValueError(
             f'Spot check failed: {count_delta_percentage:.2f}% file count delta between source paths and latest archive (tolerance is {spot_check_config["count_tolerance_percentage"]}%)'
@@ -667,8 +694,13 @@ def spot_check(
     failing_percentage = (len(failing_paths) / len(source_paths)) * 100
 
     if failing_percentage > data_tolerance_percentage:
+        truncated_failing_paths = textwrap.shorten(
+            ', '.join(failing_paths),
+            width=MAX_SPOT_CHECK_PATHS_LENGTH,
+            placeholder=' ...',
+        )
         logger.debug(
-            f'Source paths with data not matching the latest archive: {", ".join(failing_paths)}'
+            f'Source paths with data not matching the latest archive: {truncated_failing_paths}'
         )
         raise ValueError(
             f'Spot check failed: {failing_percentage:.2f}% of source paths with data not matching the latest archive (tolerance is {data_tolerance_percentage}%)'
@@ -683,7 +715,6 @@ def run_check(
     config_filename,
     repository,
     config,
-    hook_context,
     local_borg_version,
     check_arguments,
     global_arguments,
@@ -700,15 +731,6 @@ def run_check(
     ):
         return
 
-    borgmatic.hooks.command.execute_hook(
-        config.get('before_check'),
-        config.get('umask'),
-        config_filename,
-        'pre-check',
-        global_arguments.dry_run,
-        **hook_context,
-    )
-
     logger.info('Running consistency checks')
 
     repository_id = borgmatic.borg.check.get_repository_id(
@@ -773,12 +795,3 @@ def run_check(
                 borgmatic_runtime_directory,
             )
         write_check_time(make_check_time_path(config, repository_id, 'spot'))
-
-    borgmatic.hooks.command.execute_hook(
-        config.get('after_check'),
-        config.get('umask'),
-        config_filename,
-        'post-check',
-        global_arguments.dry_run,
-        **hook_context,
-    )

borgmatic/actions/compact.py

@@ -12,7 +12,6 @@ def run_compact(
     config_filename,
     repository,
     config,
-    hook_context,
     local_borg_version,
     compact_arguments,
     global_arguments,
@@ -28,14 +27,6 @@ def run_compact(
     ):
         return
 
-    borgmatic.hooks.command.execute_hook(
-        config.get('before_compact'),
-        config.get('umask'),
-        config_filename,
-        'pre-compact',
-        global_arguments.dry_run,
-        **hook_context,
-    )
     if borgmatic.borg.feature.available(borgmatic.borg.feature.Feature.COMPACT, local_borg_version):
         logger.info(f'Compacting segments{dry_run_label}')
         borgmatic.borg.compact.compact_segments(
@@ -46,18 +37,7 @@ def run_compact(
             global_arguments,
             local_path=local_path,
             remote_path=remote_path,
-            progress=compact_arguments.progress,
             cleanup_commits=compact_arguments.cleanup_commits,
-            threshold=compact_arguments.threshold,
         )
     else:  # pragma: nocover
         logger.info('Skipping compact (only available/needed in Borg 1.2+)')
-
-    borgmatic.hooks.command.execute_hook(
-        config.get('after_compact'),
-        config.get('umask'),
-        config_filename,
-        'post-compact',
-        global_arguments.dry_run,
-        **hook_context,
-    )

borgmatic/actions/config/bootstrap.py

@@ -119,7 +119,9 @@ def run_bootstrap(bootstrap_arguments, global_arguments, local_borg_version):
         bootstrap_arguments.repository,
         archive_name,
         [config_path.lstrip(os.path.sep) for config_path in manifest_config_paths],
-        config,
+        # Only add progress here and not the extract_archive() call above, because progress
+        # conflicts with extract_to_stdout.
+        dict(config, progress=bootstrap_arguments.progress or False),
         local_borg_version,
         global_arguments,
         local_path=bootstrap_arguments.local_path,
@@ -127,5 +129,4 @@ def run_bootstrap(bootstrap_arguments, global_arguments, local_borg_version):
         extract_to_stdout=False,
         destination_path=bootstrap_arguments.destination,
         strip_components=bootstrap_arguments.strip_components,
-        progress=bootstrap_arguments.progress,
     )

borgmatic/actions/create.py

@@ -1,262 +1,22 @@
-import glob
-import itertools
 import logging
-import os
-import pathlib
 
 import borgmatic.actions.json
 import borgmatic.borg.create
-import borgmatic.borg.pattern
+import borgmatic.borg.rename
+import borgmatic.borg.repo_list
 import borgmatic.config.paths
 import borgmatic.config.validate
-import borgmatic.hooks.command
 import borgmatic.hooks.dispatch
+from borgmatic.actions import pattern
 
 logger = logging.getLogger(__name__)
 
 
-def parse_pattern(pattern_line, default_style=borgmatic.borg.pattern.Pattern_style.NONE):
-    '''
-    Given a Borg pattern as a string, parse it into a borgmatic.borg.pattern.Pattern instance and
-    return it.
-    '''
-    try:
-        (pattern_type, remainder) = pattern_line.split(' ', maxsplit=1)
-    except ValueError:
-        raise ValueError(f'Invalid pattern: {pattern_line}')
-
-    try:
-        (parsed_pattern_style, path) = remainder.split(':', maxsplit=1)
-        pattern_style = borgmatic.borg.pattern.Pattern_style(parsed_pattern_style)
-    except ValueError:
-        pattern_style = default_style
-        path = remainder
-
-    return borgmatic.borg.pattern.Pattern(
-        path,
-        borgmatic.borg.pattern.Pattern_type(pattern_type),
-        borgmatic.borg.pattern.Pattern_style(pattern_style),
-    )
-
-
-def collect_patterns(config):
-    '''
-    Given a configuration dict, produce a single sequence of patterns comprised of the configured
-    source directories, patterns, excludes, pattern files, and exclude files.
-
-    The idea is that Borg has all these different ways of specifying includes, excludes, source
-    directories, etc., but we'd like to collapse them all down to one common format (patterns) for
-    ease of manipulation within borgmatic.
-    '''
-    try:
-        return (
-            tuple(
-                borgmatic.borg.pattern.Pattern(source_directory)
-                for source_directory in config.get('source_directories', ())
-            )
-            + tuple(
-                parse_pattern(pattern_line.strip())
-                for pattern_line in config.get('patterns', ())
-                if not pattern_line.lstrip().startswith('#')
-                if pattern_line.strip()
-            )
-            + tuple(
-                parse_pattern(
-                    f'{borgmatic.borg.pattern.Pattern_type.NO_RECURSE.value} {exclude_line.strip()}',
-                    borgmatic.borg.pattern.Pattern_style.FNMATCH,
-                )
-                for exclude_line in config.get('exclude_patterns', ())
-            )
-            + tuple(
-                parse_pattern(pattern_line.strip())
-                for filename in config.get('patterns_from', ())
-                for pattern_line in open(filename).readlines()
-                if not pattern_line.lstrip().startswith('#')
-                if pattern_line.strip()
-            )
-            + tuple(
-                parse_pattern(
-                    f'{borgmatic.borg.pattern.Pattern_type.NO_RECURSE.value} {exclude_line.strip()}',
-                    borgmatic.borg.pattern.Pattern_style.FNMATCH,
-                )
-                for filename in config.get('exclude_from', ())
-                for exclude_line in open(filename).readlines()
-                if not exclude_line.lstrip().startswith('#')
-                if exclude_line.strip()
-            )
-        )
-    except (FileNotFoundError, OSError) as error:
-        logger.debug(error)
-
-        raise ValueError(f'Cannot read patterns_from/exclude_from file: {error.filename}')
-
-
-def expand_directory(directory, working_directory):
-    '''
-    Given a directory path, expand any tilde (representing a user's home directory) and any globs
-    therein. Return a list of one or more resulting paths.
-
-    Take into account the given working directory so that relative paths are supported.
-    '''
-    expanded_directory = os.path.expanduser(directory)
-
-    # This would be a lot easier to do with glob(..., root_dir=working_directory), but root_dir is
-    # only available in Python 3.10+.
-    normalized_directory = os.path.join(working_directory or '', expanded_directory)
-    glob_paths = glob.glob(normalized_directory)
-
-    if not glob_paths:
-        return [expanded_directory]
-
-    working_directory_prefix = os.path.join(working_directory or '', '')
-
-    return [
-        (
-            glob_path
-            # If these are equal, that means we didn't add any working directory prefix above.
-            if normalized_directory == expanded_directory
-            # Remove the working directory prefix that we added above in order to make glob() work.
-            # We can't use os.path.relpath() here because it collapses any use of Borg's slashdot
-            # hack.
-            else glob_path.removeprefix(working_directory_prefix)
-        )
-        for glob_path in glob_paths
-    ]
-
-
-def expand_patterns(patterns, working_directory=None, skip_paths=None):
-    '''
-    Given a sequence of borgmatic.borg.pattern.Pattern instances and an optional working directory,
-    expand tildes and globs in each root pattern. Return all the resulting patterns (not just the
-    root patterns) as a tuple.
-
-    If a set of paths are given to skip, then don't expand any patterns matching them.
-    '''
-    if patterns is None:
-        return ()
-
-    return tuple(
-        itertools.chain.from_iterable(
-            (
-                (
-                    borgmatic.borg.pattern.Pattern(
-                        expanded_path,
-                        pattern.type,
-                        pattern.style,
-                        pattern.device,
-                    )
-                    for expanded_path in expand_directory(pattern.path, working_directory)
-                )
-                if pattern.type == borgmatic.borg.pattern.Pattern_type.ROOT
-                and pattern.path not in (skip_paths or ())
-                else (pattern,)
-            )
-            for pattern in patterns
-        )
-    )
-
-
-def device_map_patterns(patterns, working_directory=None):
-    '''
-    Given a sequence of borgmatic.borg.pattern.Pattern instances and an optional working directory,
-    determine the identifier for the device on which the pattern's path resides—or None if the path
-    doesn't exist or is from a non-root pattern. Return an updated sequence of patterns with the
-    device field populated. But if the device field is already set, don't bother setting it again.
-
-    This is handy for determining whether two different pattern paths are on the same filesystem
-    (have the same device identifier).
-    '''
-    return tuple(
-        borgmatic.borg.pattern.Pattern(
-            pattern.path,
-            pattern.type,
-            pattern.style,
-            device=pattern.device
-            or (
-                os.stat(full_path).st_dev
-                if pattern.type == borgmatic.borg.pattern.Pattern_type.ROOT
-                and os.path.exists(full_path)
-                else None
-            ),
-        )
-        for pattern in patterns
-        for full_path in (os.path.join(working_directory or '', pattern.path),)
-    )
-
-
-def deduplicate_patterns(patterns):
-    '''
-    Given a sequence of borgmatic.borg.pattern.Pattern instances, return them with all duplicate
-    root child patterns removed. For instance, if two root patterns are given with paths "/foo" and
-    "/foo/bar", return just the one with "/foo". Non-root patterns are passed through without
-    modification.
-
-    The one exception to deduplication is two paths are on different filesystems (devices). In that
-    case, they won't get deduplicated, in case they both need to be passed to Borg (e.g. the
-    one_file_system option is true).
-
-    The idea is that if Borg is given a root parent pattern, then it doesn't also need to be given
-    child patterns, because it will naturally spider the contents of the parent pattern's path. And
-    there are cases where Borg coming across the same file twice will result in duplicate reads and
-    even hangs, e.g. when a database hook is using a named pipe for streaming database dumps to
-    Borg.
-    '''
-    deduplicated = {}  # Use just the keys as an ordered set.
-
-    for pattern in patterns:
-        if pattern.type != borgmatic.borg.pattern.Pattern_type.ROOT:
-            deduplicated[pattern] = True
-            continue
-
-        parents = pathlib.PurePath(pattern.path).parents
-
-        # If another directory in the given list is a parent of current directory (even n levels up)
-        # and both are on the same filesystem, then the current directory is a duplicate.
-        for other_pattern in patterns:
-            if other_pattern.type != borgmatic.borg.pattern.Pattern_type.ROOT:
-                continue
-
-            if any(
-                pathlib.PurePath(other_pattern.path) == parent
-                and pattern.device is not None
-                and other_pattern.device == pattern.device
-                for parent in parents
-            ):
-                break
-        else:
-            deduplicated[pattern] = True
-
-    return tuple(deduplicated.keys())
-
-
-def process_patterns(patterns, working_directory, skip_expand_paths=None):
-    '''
-    Given a sequence of Borg patterns and a configured working directory, expand and deduplicate any
-    "root" patterns, returning the resulting root and non-root patterns as a list.
-
-    If any paths are given to skip, don't expand them.
-    '''
-    skip_paths = set(skip_expand_paths or ())
-
-    return list(
-        deduplicate_patterns(
-            device_map_patterns(
-                expand_patterns(
-                    patterns,
-                    working_directory=working_directory,
-                    skip_paths=skip_paths,
-                )
-            )
-        )
-    )
-
-
 def run_create(
     config_filename,
     repository,
     config,
     config_paths,
-    hook_context,
     local_borg_version,
     create_arguments,
     global_arguments,
@@ -274,14 +34,15 @@ def run_create(
     ):
         return
 
-    borgmatic.hooks.command.execute_hook(
-        config.get('before_backup'),
-        config.get('umask'),
-        config_filename,
-        'pre-backup',
-        global_arguments.dry_run,
-        **hook_context,
-    )
+    if config.get('list_details') and config.get('progress'):
+        raise ValueError(
+            'With the create action, only one of --list/--files/list_details and --progress/progress can be used.'
+        )
+
+    if config.get('list_details') and create_arguments.json:
+        raise ValueError(
+            'With the create action, only one of --list/--files/list_details and --json can be used.'
+        )
 
     logger.info(f'Creating archive{dry_run_label}')
     working_directory = borgmatic.config.paths.get_working_directory(config)
@@ -294,7 +55,7 @@ def run_create(
             borgmatic_runtime_directory,
             global_arguments.dry_run,
         )
-        patterns = process_patterns(collect_patterns(config), working_directory)
+        patterns = pattern.process_patterns(pattern.collect_patterns(config), working_directory)
         active_dumps = borgmatic.hooks.dispatch.call_hooks(
             'dump_data_sources',
             config,
@@ -308,9 +69,17 @@ def run_create(
         # Process the patterns again in case any data source hooks updated them. Without this step,
         # we could end up with duplicate paths that cause Borg to hang when it tries to read from
         # the same named pipe twice.
-        patterns = process_patterns(patterns, working_directory, skip_expand_paths=config_paths)
+        patterns = pattern.process_patterns(
+            patterns, working_directory, skip_expand_paths=config_paths
+        )
         stream_processes = [process for processes in active_dumps.values() for process in processes]
 
+        # If we have stream processes, we first create an archive with .checkpoint suffix. This is
+        # to make sure we only create a real archive if all the streaming processes completed
+        # successfully (create_archive will fail if a streaming process fails, but the archive might
+        # have already been created at this point).
+        use_checkpoint = bool(stream_processes)
+
         json_output = borgmatic.borg.create.create_archive(
             global_arguments.dry_run,
             repository['path'],
@@ -319,17 +88,41 @@ def run_create(
             local_borg_version,
             global_arguments,
             borgmatic_runtime_directory,
+            archive_suffix='.checkpoint' if use_checkpoint else '',
             local_path=local_path,
             remote_path=remote_path,
-            progress=create_arguments.progress,
-            stats=create_arguments.stats,
             json=create_arguments.json,
-            list_files=create_arguments.list_files,
+            comment=create_arguments.comment,
             stream_processes=stream_processes,
         )
 
+        if use_checkpoint:
+            rename_checkpoint_archive(
+                repository['path'],
+                global_arguments,
+                config,
+                local_borg_version,
+                local_path,
+                remote_path,
+            )
+
         if json_output:
-            yield borgmatic.actions.json.parse_json(json_output, repository.get('label'))
+            output = borgmatic.actions.json.parse_json(json_output, repository.get('label'))
+            if use_checkpoint:
+                # Patch archive name and ID
+                renamed_archive = borgmatic.borg.repo_list.get_latest_archive(
+                    repository['path'],
+                    config,
+                    local_borg_version,
+                    global_arguments,
+                    local_path,
+                    remote_path,
+                )
+
+                output['archive']['name'] = renamed_archive['name']
+                output['archive']['id'] = renamed_archive['id']
+
+            yield output
 
         borgmatic.hooks.dispatch.call_hooks_even_if_unconfigured(
             'remove_data_source_dumps',
@@ -339,11 +132,53 @@ def run_create(
             global_arguments.dry_run,
         )
 
-    borgmatic.hooks.command.execute_hook(
-        config.get('after_backup'),
-        config.get('umask'),
-        config_filename,
-        'post-backup',
+
+def rename_checkpoint_archive(
+    repository_path,
+    global_arguments,
+    config,
+    local_borg_version,
+    local_path,
+    remote_path,
+):
+    '''
+    Renames the latest archive to not have a '.checkpoint' suffix.
+
+    Raises ValueError if
+    - there is not latest archive
+    - the latest archive does not have a '.checkpoint' suffix
+
+    Implementation note: We cannot reliably get the just created archive name.
+    So we resort to listing the archives and picking the last one.
+
+    A similar comment applies to retrieving the ID of the renamed archive.
+    '''
+    archive = borgmatic.borg.repo_list.get_latest_archive(
+        repository_path,
+        config,
+        local_borg_version,
+        global_arguments,
+        local_path,
+        remote_path,
+        consider_checkpoints=True,
+    )
+
+    archive_name = archive['name']
+
+    if not archive_name.endswith('.checkpoint'):
+        raise ValueError(f'Latest archive did not have a .checkpoint suffix. Got: {archive_name}')
+
+    new_archive_name = archive_name.removesuffix('.checkpoint')
+
+    logger.info(f'Renaming archive {archive_name} -> {new_archive_name}')
+
+    borgmatic.borg.rename.rename_archive(
+        repository_path,
+        archive_name,
+        new_archive_name,
         global_arguments.dry_run,
-        **hook_context,
+        config,
+        local_borg_version,
+        local_path,
+        remote_path,
     )

borgmatic/actions/export_tar.py

@@ -43,6 +43,5 @@ def run_export_tar(
             local_path=local_path,
             remote_path=remote_path,
             tar_filter=export_tar_arguments.tar_filter,
-            list_files=export_tar_arguments.list_files,
             strip_components=export_tar_arguments.strip_components,
         )

borgmatic/actions/extract.py

@@ -12,7 +12,6 @@ def run_extract(
     config_filename,
     repository,
     config,
-    hook_context,
     local_borg_version,
     extract_arguments,
     global_arguments,
@@ -22,14 +21,6 @@ def run_extract(
     '''
     Run the "extract" action for the given repository.
     '''
-    borgmatic.hooks.command.execute_hook(
-        config.get('before_extract'),
-        config.get('umask'),
-        config_filename,
-        'pre-extract',
-        global_arguments.dry_run,
-        **hook_context,
-    )
     if extract_arguments.repository is None or borgmatic.config.validate.repositories_match(
         repository, extract_arguments.repository
     ):
@@ -54,13 +45,4 @@ def run_extract(
             remote_path=remote_path,
             destination_path=extract_arguments.destination,
             strip_components=extract_arguments.strip_components,
-            progress=extract_arguments.progress,
         )
-    borgmatic.hooks.command.execute_hook(
-        config.get('after_extract'),
-        config.get('umask'),
-        config_filename,
-        'post-extract',
-        global_arguments.dry_run,
-        **hook_context,
-    )

borgmatic/actions/import_key.py

@@ -0,0 +1,33 @@
+import logging
+
+import borgmatic.borg.import_key
+import borgmatic.config.validate
+
+logger = logging.getLogger(__name__)
+
+
+def run_import_key(
+    repository,
+    config,
+    local_borg_version,
+    import_arguments,
+    global_arguments,
+    local_path,
+    remote_path,
+):
+    '''
+    Run the "key import" action for the given repository.
+    '''
+    if import_arguments.repository is None or borgmatic.config.validate.repositories_match(
+        repository, import_arguments.repository
+    ):
+        logger.info('Importing repository key')
+        borgmatic.borg.import_key.import_key(
+            repository['path'],
+            config,
+            local_borg_version,
+            import_arguments,
+            global_arguments,
+            local_path=local_path,
+            remote_path=remote_path,
+        )

borgmatic/actions/pattern.py

@@ -0,0 +1,292 @@
+import glob
+import itertools
+import logging
+import os
+import pathlib
+
+import borgmatic.borg.pattern
+
+logger = logging.getLogger(__name__)
+
+
+def parse_pattern(pattern_line, default_style=borgmatic.borg.pattern.Pattern_style.NONE):
+    '''
+    Given a Borg pattern as a string, parse it into a borgmatic.borg.pattern.Pattern instance and
+    return it.
+    '''
+    try:
+        (pattern_type, remainder) = pattern_line.split(' ', maxsplit=1)
+    except ValueError:
+        raise ValueError(f'Invalid pattern: {pattern_line}')
+
+    try:
+        (parsed_pattern_style, path) = remainder.split(':', maxsplit=1)
+        pattern_style = borgmatic.borg.pattern.Pattern_style(parsed_pattern_style)
+    except ValueError:
+        pattern_style = default_style
+        path = remainder
+
+    return borgmatic.borg.pattern.Pattern(
+        path,
+        borgmatic.borg.pattern.Pattern_type(pattern_type),
+        borgmatic.borg.pattern.Pattern_style(pattern_style),
+        source=borgmatic.borg.pattern.Pattern_source.CONFIG,
+    )
+
+
+def collect_patterns(config):
+    '''
+    Given a configuration dict, produce a single sequence of patterns comprised of the configured
+    source directories, patterns, excludes, pattern files, and exclude files.
+
+    The idea is that Borg has all these different ways of specifying includes, excludes, source
+    directories, etc., but we'd like to collapse them all down to one common format (patterns) for
+    ease of manipulation within borgmatic.
+    '''
+    try:
+        return (
+            tuple(
+                borgmatic.borg.pattern.Pattern(
+                    source_directory, source=borgmatic.borg.pattern.Pattern_source.CONFIG
+                )
+                for source_directory in config.get('source_directories', ())
+            )
+            + tuple(
+                parse_pattern(pattern_line.strip())
+                for pattern_line in config.get('patterns', ())
+                if not pattern_line.lstrip().startswith('#')
+                if pattern_line.strip()
+            )
+            + tuple(
+                parse_pattern(
+                    f'{borgmatic.borg.pattern.Pattern_type.NO_RECURSE.value} {exclude_line.strip()}',
+                    borgmatic.borg.pattern.Pattern_style.FNMATCH,
+                )
+                for exclude_line in config.get('exclude_patterns', ())
+            )
+            + tuple(
+                parse_pattern(pattern_line.strip())
+                for filename in config.get('patterns_from', ())
+                for pattern_line in open(filename).readlines()
+                if not pattern_line.lstrip().startswith('#')
+                if pattern_line.strip()
+            )
+            + tuple(
+                parse_pattern(
+                    f'{borgmatic.borg.pattern.Pattern_type.NO_RECURSE.value} {exclude_line.strip()}',
+                    borgmatic.borg.pattern.Pattern_style.FNMATCH,
+                )
+                for filename in config.get('exclude_from', ())
+                for exclude_line in open(filename).readlines()
+                if not exclude_line.lstrip().startswith('#')
+                if exclude_line.strip()
+            )
+        )
+    except (FileNotFoundError, OSError) as error:
+        logger.debug(error)
+
+        raise ValueError(f'Cannot read patterns_from/exclude_from file: {error.filename}')
+
+
+def expand_directory(directory, working_directory):
+    '''
+    Given a directory path, expand any tilde (representing a user's home directory) and any globs
+    therein. Return a list of one or more resulting paths.
+
+    Take into account the given working directory so that relative paths are supported.
+    '''
+    expanded_directory = os.path.expanduser(directory)
+
+    # This would be a lot easier to do with glob(..., root_dir=working_directory), but root_dir is
+    # only available in Python 3.10+.
+    normalized_directory = os.path.join(working_directory or '', expanded_directory)
+    glob_paths = glob.glob(normalized_directory)
+
+    if not glob_paths:
+        return [expanded_directory]
+
+    working_directory_prefix = os.path.join(working_directory or '', '')
+
+    return [
+        (
+            glob_path
+            # If these are equal, that means we didn't add any working directory prefix above.
+            if normalized_directory == expanded_directory
+            # Remove the working directory prefix that we added above in order to make glob() work.
+            # We can't use os.path.relpath() here because it collapses any use of Borg's slashdot
+            # hack.
+            else glob_path.removeprefix(working_directory_prefix)
+        )
+        for glob_path in glob_paths
+    ]
+
+
+def expand_patterns(patterns, working_directory=None, skip_paths=None):
+    '''
+    Given a sequence of borgmatic.borg.pattern.Pattern instances and an optional working directory,
+    expand tildes and globs in each root pattern and expand just tildes in each non-root pattern.
+    The idea is that non-root patterns may be regular expressions or other pattern styles containing
+    "*" that borgmatic should not expand as a shell glob.
+
+    Return all the resulting patterns as a tuple.
+
+    If a set of paths are given to skip, then don't expand any patterns matching them.
+    '''
+    if patterns is None:
+        return ()
+
+    return tuple(
+        itertools.chain.from_iterable(
+            (
+                (
+                    borgmatic.borg.pattern.Pattern(
+                        expanded_path,
+                        pattern.type,
+                        pattern.style,
+                        pattern.device,
+                        pattern.source,
+                    )
+                    for expanded_path in expand_directory(pattern.path, working_directory)
+                )
+                if pattern.type == borgmatic.borg.pattern.Pattern_type.ROOT
+                and pattern.path not in (skip_paths or ())
+                else (
+                    borgmatic.borg.pattern.Pattern(
+                        os.path.expanduser(pattern.path),
+                        pattern.type,
+                        pattern.style,
+                        pattern.device,
+                        pattern.source,
+                    ),
+                )
+            )
+            for pattern in patterns
+        )
+    )
+
+
+def get_existent_path_or_parent(path):
+    '''
+    Given a path, return it if it exists. Otherwise, return the longest parent directory of the path
+    that exists. Return None if none of these paths exist.
+
+    This is used below for finding an existent path prefix of pattern's path, which is necessary if
+    the path contain globs or other special characters that we don't want to try to interpret
+    (because we want to leave that responsibility to Borg).
+    '''
+    if path.startswith('/e2e/'):
+        return None
+
+    try:
+        return next(
+            candidate_path
+            for candidate_path in (path,)
+            + tuple(str(parent) for parent in pathlib.PurePath(path).parents)
+            if os.path.exists(candidate_path)
+        )
+    except StopIteration:
+        return None
+
+
+def device_map_patterns(patterns, working_directory=None):
+    '''
+    Given a sequence of borgmatic.borg.pattern.Pattern instances and an optional working directory,
+    determine the identifier for the device on which the pattern's path resides—or None if the path
+    doesn't exist or is from a non-root pattern. Return an updated sequence of patterns with the
+    device field populated. But if the device field is already set, don't bother setting it again.
+
+    This is handy for determining whether two different pattern paths are on the same filesystem
+    (have the same device identifier).
+
+    This function only considers the start of a pattern's path—from the start of the path up until
+    there's a path component with a glob or other non-literal character. If there are no such
+    characters, the whole path is considered. The rationale is that it's not feasible for borgmatic
+    to interpret Borg's patterns to see which actual files (and therefore devices) they map to. So
+    for instance, a pattern with a path of "/var/log/*/data" would end up with its device set to the
+    device of "/var/log"—ignoring the "/*/data" part due to that glob.
+
+    The one exception is that if a regular expression pattern path starts with "^", that will get
+    stripped off for purposes of determining its device.
+    '''
+    return tuple(
+        borgmatic.borg.pattern.Pattern(
+            pattern.path,
+            pattern.type,
+            pattern.style,
+            device=pattern.device or (os.stat(existent_path).st_dev if existent_path else None),
+            source=pattern.source,
+        )
+        for pattern in patterns
+        for existent_path in (
+            get_existent_path_or_parent(
+                os.path.join(working_directory or '', pattern.path.lstrip('^'))
+            ),
+        )
+    )
+
+
+def deduplicate_patterns(patterns):
+    '''
+    Given a sequence of borgmatic.borg.pattern.Pattern instances, return them with all duplicate
+    root child patterns removed. For instance, if two root patterns are given with paths "/foo" and
+    "/foo/bar", return just the one with "/foo". Non-root patterns are passed through without
+    modification.
+
+    The one exception to deduplication is two paths are on different filesystems (devices). In that
+    case, they won't get deduplicated, in case they both need to be passed to Borg (e.g. the
+    one_file_system option is true).
+
+    The idea is that if Borg is given a root parent pattern, then it doesn't also need to be given
+    child patterns, because it will naturally spider the contents of the parent pattern's path. And
+    there are cases where Borg coming across the same file twice will result in duplicate reads and
+    even hangs, e.g. when a database hook is using a named pipe for streaming database dumps to
+    Borg.
+    '''
+    deduplicated = {}  # Use just the keys as an ordered set.
+
+    for pattern in patterns:
+        if pattern.type != borgmatic.borg.pattern.Pattern_type.ROOT:
+            deduplicated[pattern] = True
+            continue
+
+        parents = pathlib.PurePath(pattern.path).parents
+
+        # If another directory in the given list is a parent of current directory (even n levels up)
+        # and both are on the same filesystem, then the current directory is a duplicate.
+        for other_pattern in patterns:
+            if other_pattern.type != borgmatic.borg.pattern.Pattern_type.ROOT:
+                continue
+
+            if any(
+                pathlib.PurePath(other_pattern.path) == parent
+                and pattern.device is not None
+                and other_pattern.device == pattern.device
+                for parent in parents
+            ):
+                break
+        else:
+            deduplicated[pattern] = True
+
+    return tuple(deduplicated.keys())
+
+
+def process_patterns(patterns, working_directory, skip_expand_paths=None):
+    '''
+    Given a sequence of Borg patterns and a configured working directory, expand and deduplicate any
+    "root" patterns, returning the resulting root and non-root patterns as a list.
+
+    If any paths are given to skip, don't expand them.
+    '''
+    skip_paths = set(skip_expand_paths or ())
+
+    return list(
+        deduplicate_patterns(
+            device_map_patterns(
+                expand_patterns(
+                    patterns,
+                    working_directory=working_directory,
+                    skip_paths=skip_paths,
+                )
+            )
+        )
+    )

borgmatic/actions/prune.py

@@ -11,7 +11,6 @@ def run_prune(
     config_filename,
     repository,
     config,
-    hook_context,
     local_borg_version,
     prune_arguments,
     global_arguments,
@@ -27,14 +26,6 @@ def run_prune(
     ):
         return
 
-    borgmatic.hooks.command.execute_hook(
-        config.get('before_prune'),
-        config.get('umask'),
-        config_filename,
-        'pre-prune',
-        global_arguments.dry_run,
-        **hook_context,
-    )
     logger.info(f'Pruning archives{dry_run_label}')
     borgmatic.borg.prune.prune_archives(
         global_arguments.dry_run,
@@ -46,11 +37,3 @@ def run_prune(
         local_path=local_path,
         remote_path=remote_path,
     )
-    borgmatic.hooks.command.execute_hook(
-        config.get('after_prune'),
-        config.get('umask'),
-        config_filename,
-        'post-prune',
-        global_arguments.dry_run,
-        **hook_context,
-    )

borgmatic/actions/recreate.py

@@ -0,0 +1,84 @@
+import logging
+import subprocess
+
+import borgmatic.borg.info
+import borgmatic.borg.recreate
+import borgmatic.borg.repo_list
+import borgmatic.config.validate
+from borgmatic.actions.pattern import collect_patterns, process_patterns
+
+logger = logging.getLogger(__name__)
+
+
+BORG_EXIT_CODE_ARCHIVE_ALREADY_EXISTS = 30
+
+
+def run_recreate(
+    repository,
+    config,
+    local_borg_version,
+    recreate_arguments,
+    global_arguments,
+    local_path,
+    remote_path,
+):
+    '''
+    Run the "recreate" action for the given repository.
+    '''
+    if recreate_arguments.repository is None or borgmatic.config.validate.repositories_match(
+        repository, recreate_arguments.repository
+    ):
+        if recreate_arguments.archive:
+            logger.answer(f'Recreating archive {recreate_arguments.archive}')
+        else:
+            logger.answer('Recreating repository')
+
+        # Collect and process patterns.
+        processed_patterns = process_patterns(
+            collect_patterns(config), borgmatic.config.paths.get_working_directory(config)
+        )
+
+        archive = borgmatic.borg.repo_list.resolve_archive_name(
+            repository['path'],
+            recreate_arguments.archive,
+            config,
+            local_borg_version,
+            global_arguments,
+            local_path,
+            remote_path,
+        )
+
+        if archive and archive.endswith('.recreate'):
+            if recreate_arguments.archive == 'latest':
+                raise ValueError(
+                    f'The latest archive "{archive}" is leftover from a prior recreate. Delete it first or select a different archive.'
+                )
+            else:
+                raise ValueError(
+                    f'The archive "{recreate_arguments.archive}" is leftover from a prior recreate. Select a different archive.'
+                )
+
+        try:
+            borgmatic.borg.recreate.recreate_archive(
+                repository['path'],
+                archive,
+                config,
+                local_borg_version,
+                recreate_arguments,
+                global_arguments,
+                local_path=local_path,
+                remote_path=remote_path,
+                patterns=processed_patterns,
+            )
+        except subprocess.CalledProcessError as error:
+            if error.returncode == BORG_EXIT_CODE_ARCHIVE_ALREADY_EXISTS:
+                if recreate_arguments.target:
+                    raise ValueError(
+                        f'The archive "{recreate_arguments.target}" already exists. Delete it first or set a different target archive name.'
+                    )
+                elif archive:
+                    raise ValueError(
+                        f'The archive "{archive}.recreate" is leftover from a prior recreate. Delete it first or select a different archive.'
+                    )
+
+            raise

borgmatic/actions/repo_create.py

@@ -24,18 +24,38 @@ def run_repo_create(
         return
 
     logger.info('Creating repository')
+
+    encryption_mode = repo_create_arguments.encryption_mode or repository.get('encryption')
+
+    if not encryption_mode:
+        raise ValueError(
+            'With the repo-create action, either the --encryption flag or the repository encryption option is required.'
+        )
+
     borgmatic.borg.repo_create.create_repository(
         global_arguments.dry_run,
         repository['path'],
         config,
         local_borg_version,
         global_arguments,
-        repo_create_arguments.encryption_mode,
+        encryption_mode,
         repo_create_arguments.source_repository,
         repo_create_arguments.copy_crypt_key,
-        repo_create_arguments.append_only,
-        repo_create_arguments.storage_quota,
-        repo_create_arguments.make_parent_dirs,
+        (
+            repository.get('append_only')
+            if repo_create_arguments.append_only is None
+            else repo_create_arguments.append_only
+        ),
+        (
+            repository.get('storage_quota')
+            if repo_create_arguments.storage_quota is None
+            else repo_create_arguments.storage_quota
+        ),
+        (
+            repository.get('make_parent_directories')
+            if repo_create_arguments.make_parent_directories is None
+            else repo_create_arguments.make_parent_directories
+        ),
         local_path=local_path,
         remote_path=remote_path,
     )

borgmatic/actions/transfer.py

@@ -17,7 +17,13 @@ def run_transfer(
     '''
     Run the "transfer" action for the given repository.
     '''
+    if transfer_arguments.archive and config.get('match_archives'):
+        raise ValueError(
+            'With the transfer action, only one of --archive and --match-archives/match_archives can be used.'
+        )
+
     logger.info('Transferring archives to repository')
+
     borgmatic.borg.transfer.transfer_archives(
         global_arguments.dry_run,
         repository['path'],

borgmatic/borg/borg.py

@@ -60,8 +60,8 @@ def run_arbitrary_borg(
     return execute_command(
         tuple(shlex.quote(part) for part in full_command),
         output_file=DO_NOT_CAPTURE,
-        shell=True,
-        extra_environment=dict(
+        shell=True,  # noqa: S604
+        environment=dict(
             (environment.make_environment(config) or {}),
             **{
                 'BORG_REPO': repository_path,

borgmatic/borg/break_lock.py

@@ -27,17 +27,16 @@ def break_lock(
         (local_path, 'break-lock')
         + (('--remote-path', remote_path) if remote_path else ())
         + (('--umask', str(umask)) if umask else ())
-        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--log-json',) if config.get('log_json') else ())
         + (('--lock-wait', str(lock_wait)) if lock_wait else ())
         + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
         + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
         + flags.make_repository_flags(repository_path, local_borg_version)
     )
 
-    borg_environment = environment.make_environment(config)
     execute_command(
         full_command,
-        extra_environment=borg_environment,
+        environment=environment.make_environment(config),
         working_directory=borgmatic.config.paths.get_working_directory(config),
         borg_local_path=local_path,
         borg_exit_codes=config.get('borg_exit_codes'),

borgmatic/borg/change_passphrase.py

@@ -30,7 +30,7 @@ def change_passphrase(
         (local_path, 'key', 'change-passphrase')
         + (('--remote-path', remote_path) if remote_path else ())
         + (('--umask', str(umask)) if umask else ())
-        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--log-json',) if config.get('log_json') else ())
         + (('--lock-wait', str(lock_wait)) if lock_wait else ())
         + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
         + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
@@ -56,7 +56,7 @@ def change_passphrase(
         full_command,
         output_file=borgmatic.execute.DO_NOT_CAPTURE,
         output_log_level=logging.ANSWER,
-        extra_environment=environment.make_environment(config_without_passphrase),
+        environment=environment.make_environment(config_without_passphrase),
         working_directory=borgmatic.config.paths.get_working_directory(config),
         borg_local_path=local_path,
         borg_exit_codes=config.get('borg_exit_codes'),

borgmatic/borg/check.py

@@ -1,6 +1,7 @@
 import argparse
 import json
 import logging
+import shlex
 
 import borgmatic.config.paths
 from borgmatic.borg import environment, feature, flags, repo_info
@@ -32,7 +33,7 @@ def make_archive_filter_flags(local_borg_version, config, checks, check_argument
             if prefix
             else (
                 flags.make_match_archives_flags(
-                    check_arguments.match_archives or config.get('match_archives'),
+                    config.get('match_archives'),
                     config.get('archive_name_format'),
                     local_borg_version,
                 )
@@ -64,15 +65,11 @@ def make_check_name_flags(checks, archive_filter_flags):
 
         ('--repository-only',)
 
-    However, if both "repository" and "archives" are in checks, then omit them from the returned
-    flags because Borg does both checks by default. If "data" is in checks, that implies "archives".
+    However, if both "repository" and "archives" are in checks, then omit the "only" flags from the
+    returned flags because Borg does both checks by default. Note that a "data" check only works
+    along with an "archives" check.
     '''
-    if 'data' in checks:
-        data_flags = ('--verify-data',)
-        checks.update({'archives'})
-    else:
-        data_flags = ()
-
+    data_flags = ('--verify-data',) if 'data' in checks else ()
     common_flags = (archive_filter_flags if 'archives' in checks else ()) + data_flags
 
     if {'repository', 'archives'}.issubset(checks):
@@ -142,51 +139,51 @@ def check_archives(
     except StopIteration:
         repository_check_config = {}
 
-    if check_arguments.max_duration and 'archives' in checks:
-        raise ValueError('The archives check cannot run when the --max-duration flag is used')
-    if repository_check_config.get('max_duration') and 'archives' in checks:
-        raise ValueError(
-            'The archives check cannot run when the repository check has the max_duration option set'
-        )
-
     max_duration = check_arguments.max_duration or repository_check_config.get('max_duration')
+
     umask = config.get('umask')
-
-    borg_environment = environment.make_environment(config)
     borg_exit_codes = config.get('borg_exit_codes')
-
-    full_command = (
-        (local_path, 'check')
-        + (('--repair',) if check_arguments.repair else ())
-        + (('--max-duration', str(max_duration)) if max_duration else ())
-        + make_check_name_flags(checks, archive_filter_flags)
-        + (('--remote-path', remote_path) if remote_path else ())
-        + (('--umask', str(umask)) if umask else ())
-        + (('--log-json',) if global_arguments.log_json else ())
-        + (('--lock-wait', str(lock_wait)) if lock_wait else ())
-        + verbosity_flags
-        + (('--progress',) if check_arguments.progress else ())
-        + (tuple(extra_borg_options.split(' ')) if extra_borg_options else ())
-        + flags.make_repository_flags(repository_path, local_borg_version)
-    )
-
     working_directory = borgmatic.config.paths.get_working_directory(config)
 
-    # The Borg repair option triggers an interactive prompt, which won't work when output is
-    # captured. And progress messes with the terminal directly.
-    if check_arguments.repair or check_arguments.progress:
+    if 'data' in checks:
+        checks.add('archives')
+
+    grouped_checks = (checks,)
+
+    # If max_duration is set, then archives and repository checks need to be run separately, as Borg
+    # doesn't support --max-duration along with an archives checks.
+    if max_duration and 'archives' in checks and 'repository' in checks:
+        checks.remove('repository')
+        grouped_checks = (checks, {'repository'})
+
+    for checks_subset in grouped_checks:
+        full_command = (
+            (local_path, 'check')
+            + (('--repair',) if check_arguments.repair else ())
+            + (
+                ('--max-duration', str(max_duration))
+                if max_duration and 'repository' in checks_subset
+                else ()
+            )
+            + make_check_name_flags(checks_subset, archive_filter_flags)
+            + (('--remote-path', remote_path) if remote_path else ())
+            + (('--umask', str(umask)) if umask else ())
+            + (('--log-json',) if config.get('log_json') else ())
+            + (('--lock-wait', str(lock_wait)) if lock_wait else ())
+            + verbosity_flags
+            + (('--progress',) if config.get('progress') else ())
+            + (tuple(shlex.split(extra_borg_options)) if extra_borg_options else ())
+            + flags.make_repository_flags(repository_path, local_borg_version)
+        )
+
         execute_command(
             full_command,
-            output_file=DO_NOT_CAPTURE,
-            extra_environment=borg_environment,
-            working_directory=working_directory,
-            borg_local_path=local_path,
-            borg_exit_codes=borg_exit_codes,
-        )
-    else:
-        execute_command(
-            full_command,
-            extra_environment=borg_environment,
+            # The Borg repair option triggers an interactive prompt, which won't work when output is
+            # captured. And progress messes with the terminal directly.
+            output_file=(
+                DO_NOT_CAPTURE if check_arguments.repair or config.get('progress') else None
+            ),
+            environment=environment.make_environment(config),
             working_directory=working_directory,
             borg_local_path=local_path,
             borg_exit_codes=borg_exit_codes,

borgmatic/borg/compact.py

@@ -1,7 +1,8 @@
 import logging
+import shlex
 
 import borgmatic.config.paths
-from borgmatic.borg import environment, flags
+from borgmatic.borg import environment, feature, flags
 from borgmatic.execute import execute_command
 
 logger = logging.getLogger(__name__)
@@ -15,9 +16,7 @@ def compact_segments(
     global_arguments,
     local_path='borg',
     remote_path=None,
-    progress=False,
     cleanup_commits=False,
-    threshold=None,
 ):
     '''
     Given dry-run flag, a local or remote repository path, a configuration dict, and the local Borg
@@ -26,30 +25,36 @@ def compact_segments(
     umask = config.get('umask', None)
     lock_wait = config.get('lock_wait', None)
     extra_borg_options = config.get('extra_borg_options', {}).get('compact', '')
+    threshold = config.get('compact_threshold')
 
     full_command = (
         (local_path, 'compact')
         + (('--remote-path', remote_path) if remote_path else ())
         + (('--umask', str(umask)) if umask else ())
-        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--log-json',) if config.get('log_json') else ())
         + (('--lock-wait', str(lock_wait)) if lock_wait else ())
-        + (('--progress',) if progress else ())
+        + (('--progress',) if config.get('progress') else ())
         + (('--cleanup-commits',) if cleanup_commits else ())
         + (('--threshold', str(threshold)) if threshold else ())
         + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
         + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
-        + (tuple(extra_borg_options.split(' ')) if extra_borg_options else ())
+        + (
+            ('--dry-run',)
+            if dry_run and feature.available(feature.Feature.DRY_RUN_COMPACT, local_borg_version)
+            else ()
+        )
+        + (tuple(shlex.split(extra_borg_options)) if extra_borg_options else ())
         + flags.make_repository_flags(repository_path, local_borg_version)
     )
 
-    if dry_run:
+    if dry_run and not feature.available(feature.Feature.DRY_RUN_COMPACT, local_borg_version):
         logging.info('Skipping compact (dry run)')
         return
 
     execute_command(
         full_command,
         output_log_level=logging.INFO,
-        extra_environment=environment.make_environment(config),
+        environment=environment.make_environment(config),
         working_directory=borgmatic.config.paths.get_working_directory(config),
         borg_local_path=local_path,
         borg_exit_codes=config.get('borg_exit_codes'),

borgmatic/borg/create.py

@@ -1,9 +1,8 @@
-import itertools
 import logging
 import os
 import pathlib
+import shlex
 import stat
-import tempfile
 import textwrap
 
 import borgmatic.borg.pattern
@@ -20,76 +19,6 @@ from borgmatic.execute import (
 logger = logging.getLogger(__name__)
 
 
-def write_patterns_file(patterns, borgmatic_runtime_directory, patterns_file=None):
-    '''
-    Given a sequence of patterns as borgmatic.borg.pattern.Pattern instances, write them to a named
-    temporary file in the given borgmatic runtime directory and return the file object so it can
-    continue to exist on disk as long as the caller needs it.
-
-    If an optional open pattern file is given, append to it instead of making a new temporary file.
-    Return None if no patterns are provided.
-    '''
-    if not patterns:
-        return None
-
-    if patterns_file is None:
-        patterns_file = tempfile.NamedTemporaryFile('w', dir=borgmatic_runtime_directory)
-        operation_name = 'Writing'
-    else:
-        patterns_file.write('\n')
-        operation_name = 'Appending'
-
-    patterns_output = '\n'.join(
-        f'{pattern.type.value} {pattern.style.value}{":" if pattern.style.value else ""}{pattern.path}'
-        for pattern in patterns
-    )
-    logger.debug(f'{operation_name} patterns to {patterns_file.name}:\n{patterns_output}')
-
-    patterns_file.write(patterns_output)
-    patterns_file.flush()
-
-    return patterns_file
-
-
-def make_exclude_flags(config):
-    '''
-    Given a configuration dict with various exclude options, return the corresponding Borg flags as
-    a tuple.
-    '''
-    caches_flag = ('--exclude-caches',) if config.get('exclude_caches') else ()
-    if_present_flags = tuple(
-        itertools.chain.from_iterable(
-            ('--exclude-if-present', if_present)
-            for if_present in config.get('exclude_if_present', ())
-        )
-    )
-    keep_exclude_tags_flags = ('--keep-exclude-tags',) if config.get('keep_exclude_tags') else ()
-    exclude_nodump_flags = ('--exclude-nodump',) if config.get('exclude_nodump') else ()
-
-    return caches_flag + if_present_flags + keep_exclude_tags_flags + exclude_nodump_flags
-
-
-def make_list_filter_flags(local_borg_version, dry_run):
-    '''
-    Given the local Borg version and whether this is a dry run, return the corresponding flags for
-    passing to "--list --filter". The general idea is that excludes are shown for a dry run or when
-    the verbosity is debug.
-    '''
-    base_flags = 'AME'
-    show_excludes = logger.isEnabledFor(logging.DEBUG)
-
-    if feature.available(feature.Feature.EXCLUDED_FILES_MINUS, local_borg_version):
-        if show_excludes or dry_run:
-            return f'{base_flags}+-'
-        else:
-            return base_flags
-
-    if show_excludes:
-        return f'{base_flags}x-'
-    else:
-        return f'{base_flags}-'
-
-
 def special_file(path, working_directory=None):
     '''
     Return whether the given path is a special file (character device, block device, or named pipe
@@ -122,73 +51,66 @@ def collect_special_file_paths(
     config,
     local_path,
     working_directory,
-    borg_environment,
     borgmatic_runtime_directory,
 ):
     '''
     Given a dry-run flag, a Borg create command as a tuple, a configuration dict, a local Borg path,
-    a working directory, a dict of environment variables to pass to Borg, and the borgmatic runtime
-    directory, collect the paths for any special files (character devices, block devices, and named
-    pipes / FIFOs) that Borg would encounter during a create. These are all paths that could cause
-    Borg to hang if its --read-special flag is used.
+    a working directory, and the borgmatic runtime directory, collect the paths for any special
+    files (character devices, block devices, and named pipes / FIFOs) that Borg would encounter
+    during a create. These are all paths that could cause Borg to hang if its --read-special flag is
+    used.
 
     Skip looking for special files in the given borgmatic runtime directory, as borgmatic creates
-    its own special files there for database dumps. And if the borgmatic runtime directory is
-    configured to be excluded from the files Borg backs up, error, because this means Borg won't be
-    able to consume any database dumps and therefore borgmatic will hang.
+    its own special files there for database dumps and we don't want those omitted.
+
+    Additionally, if the borgmatic runtime directory is not contained somewhere in the files Borg
+    plans to backup, that means the user must have excluded the runtime directory (e.g. via
+    "exclude_patterns" or similar). Therefore, raise, because this means Borg won't be able to
+    consume any database dumps and therefore borgmatic will hang when it tries to do so.
     '''
     # Omit "--exclude-nodump" from the Borg dry run command, because that flag causes Borg to open
-    # files including any named pipe we've created.
+    # files including any named pipe we've created. And omit "--filter" because that can break the
+    # paths output parsing below such that path lines no longer start with the expected "- ".
     paths_output = execute_command_and_capture_output(
-        tuple(argument for argument in create_command if argument != '--exclude-nodump')
+        flags.omit_flag_and_value(flags.omit_flag(create_command, '--exclude-nodump'), '--filter')
         + ('--dry-run', '--list'),
         capture_stderr=True,
         working_directory=working_directory,
-        extra_environment=borg_environment,
+        environment=environment.make_environment(config),
         borg_local_path=local_path,
         borg_exit_codes=config.get('borg_exit_codes'),
     )
 
+    # These are all the individual files that Borg is planning to backup as determined by the Borg
+    # create dry run above.
     paths = tuple(
         path_line.split(' ', 1)[1]
         for path_line in paths_output.split('\n')
         if path_line and path_line.startswith('- ') or path_line.startswith('+ ')
     )
-    skip_paths = {}
+
+    # These are the subset of those files that contain the borgmatic runtime directory.
+    paths_containing_runtime_directory = {}
 
     if os.path.exists(borgmatic_runtime_directory):
-        skip_paths = {
+        paths_containing_runtime_directory = {
             path for path in paths if any_parent_directories(path, (borgmatic_runtime_directory,))
         }
 
-        if not skip_paths and not dry_run:
+        # If no paths to backup contain the runtime directory, it must've been excluded.
+        if not paths_containing_runtime_directory and not dry_run:
             raise ValueError(
                 f'The runtime directory {os.path.normpath(borgmatic_runtime_directory)} overlaps with the configured excludes or patterns with excludes. Please ensure the runtime directory is not excluded.'
             )
 
     return tuple(
-        path for path in paths if special_file(path, working_directory) if path not in skip_paths
+        path
+        for path in paths
+        if special_file(path, working_directory)
+        if path not in paths_containing_runtime_directory
     )
 
 
-def check_all_root_patterns_exist(patterns):
-    '''
-    Given a sequence of borgmatic.borg.pattern.Pattern instances, check that all root pattern
-    paths exist. If any don't, raise an exception.
-    '''
-    missing_paths = [
-        pattern.path
-        for pattern in patterns
-        if pattern.type == borgmatic.borg.pattern.Pattern_type.ROOT
-        if not os.path.exists(pattern.path)
-    ]
-
-    if missing_paths:
-        raise ValueError(
-            f"Source directories / root pattern paths do not exist: {', '.join(missing_paths)}"
-        )
-
-
 MAX_SPECIAL_FILE_PATHS_LENGTH = 1000
 
 
@@ -200,24 +122,28 @@ def make_base_create_command(
     local_borg_version,
     global_arguments,
     borgmatic_runtime_directory,
+    archive_suffix='',
     local_path='borg',
     remote_path=None,
-    progress=False,
     json=False,
-    list_files=False,
+    comment=None,
     stream_processes=None,
 ):
     '''
     Given verbosity/dry-run flags, a local or remote repository path, a configuration dict, a
-    sequence of patterns as borgmatic.borg.pattern.Pattern instances, the local Borg version,
-    global arguments as an argparse.Namespace instance, and a sequence of borgmatic source
-    directories, return a tuple of (base Borg create command flags, Borg create command positional
-    arguments, open pattern file handle).
+    sequence of patterns as borgmatic.borg.pattern.Pattern instances, the local Borg version, global
+    arguments as an argparse.Namespace instance, the borgmatic runtime directory, a string suffix to
+    add to the archive name, the local Borg path, the remote Borg path, whether to output JSON,
+    comment text to add to the created archive, and a sequence of processes streaming data to Borg,
+    return a tuple of (base Borg create command flags, Borg create command positional arguments,
+    open pattern file handle).
     '''
     if config.get('source_directories_must_exist', False):
-        check_all_root_patterns_exist(patterns)
+        borgmatic.borg.pattern.check_all_root_patterns_exist(patterns)
 
-    patterns_file = write_patterns_file(patterns, borgmatic_runtime_directory)
+    patterns_file = borgmatic.borg.pattern.write_patterns_file(
+        patterns, borgmatic_runtime_directory
+    )
     checkpoint_interval = config.get('checkpoint_interval', None)
     checkpoint_volume = config.get('checkpoint_volume', None)
     chunker_params = config.get('chunker_params', None)
@@ -226,10 +152,11 @@ def make_base_create_command(
     upload_buffer_size = config.get('upload_buffer_size', None)
     umask = config.get('umask', None)
     lock_wait = config.get('lock_wait', None)
-    list_filter_flags = make_list_filter_flags(local_borg_version, dry_run)
+    list_filter_flags = flags.make_list_filter_flags(local_borg_version, dry_run)
     files_cache = config.get('files_cache')
-    archive_name_format = config.get(
-        'archive_name_format', flags.get_default_archive_name_format(local_borg_version)
+    archive_name_format = (
+        config.get('archive_name_format', flags.get_default_archive_name_format(local_borg_version))
+        + archive_suffix
     )
     extra_borg_options = config.get('extra_borg_options', {}).get('create', '')
 
@@ -261,7 +188,8 @@ def make_base_create_command(
         tuple(local_path.split(' '))
         + ('create',)
         + (('--patterns-from', patterns_file.name) if patterns_file else ())
-        + make_exclude_flags(config)
+        + flags.make_exclude_flags(config)
+        + (('--comment', comment) if comment else ())
         + (('--checkpoint-interval', str(checkpoint_interval)) if checkpoint_interval else ())
         + (('--checkpoint-volume', str(checkpoint_volume)) if checkpoint_volume else ())
         + (('--chunker-params', chunker_params) if chunker_params else ())
@@ -278,15 +206,15 @@ def make_base_create_command(
         + (('--files-cache', files_cache) if files_cache else ())
         + (('--remote-path', remote_path) if remote_path else ())
         + (('--umask', str(umask)) if umask else ())
-        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--log-json',) if config.get('log_json') else ())
         + (('--lock-wait', str(lock_wait)) if lock_wait else ())
         + (
             ('--list', '--filter', list_filter_flags)
-            if list_files and not json and not progress
+            if config.get('list_details') and not json and not config.get('progress')
             else ()
         )
         + (('--dry-run',) if dry_run else ())
-        + (tuple(extra_borg_options.split(' ')) if extra_borg_options else ())
+        + (tuple(shlex.split(extra_borg_options)) if extra_borg_options else ())
     )
 
     create_positional_arguments = flags.make_repository_archive_flags(
@@ -299,7 +227,6 @@ def make_base_create_command(
         logger.warning(
             'Ignoring configured "read_special" value of false, as true is needed for database hooks.'
         )
-        borg_environment = environment.make_environment(config)
         working_directory = borgmatic.config.paths.get_working_directory(config)
 
         logger.debug('Collecting special file paths')
@@ -309,7 +236,6 @@ def make_base_create_command(
             config,
             local_path,
             working_directory,
-            borg_environment,
             borgmatic_runtime_directory=borgmatic_runtime_directory,
         )
 
@@ -322,12 +248,13 @@ def make_base_create_command(
             logger.warning(
                 f'Excluding special files to prevent Borg from hanging: {truncated_special_file_paths}'
             )
-            patterns_file = write_patterns_file(
+            patterns_file = borgmatic.borg.pattern.write_patterns_file(
                 tuple(
                     borgmatic.borg.pattern.Pattern(
                         special_file_path,
                         borgmatic.borg.pattern.Pattern_type.NO_RECURSE,
                         borgmatic.borg.pattern.Pattern_style.FNMATCH,
+                        source=borgmatic.borg.pattern.Pattern_source.INTERNAL,
                     )
                     for special_file_path in special_file_paths
                 ),
@@ -349,18 +276,20 @@ def create_archive(
     local_borg_version,
     global_arguments,
     borgmatic_runtime_directory,
+    archive_suffix='',
     local_path='borg',
     remote_path=None,
-    progress=False,
-    stats=False,
     json=False,
-    list_files=False,
+    comment=None,
     stream_processes=None,
 ):
     '''
     Given verbosity/dry-run flags, a local or remote repository path, a configuration dict, a
-    sequence of loaded configuration paths, the local Borg version, and global arguments as an
-    argparse.Namespace instance, create a Borg archive and return Borg's JSON output (if any).
+    sequence of loaded configuration paths, the local Borg version, global arguments as an
+    argparse.Namespace instance, the borgmatic runtime directory, a string suffix to add to the
+    archive name, the local Borg path, the remote Borg path, whether to output JSON, and comment
+    text to add to the created archive, and a sequence of processes streaming data to Borg, create a
+    Borg archive and return Borg's JSON output (if any).
 
     If a sequence of stream processes is given (instances of subprocess.Popen), then execute the
     create command while also triggering the given processes to produce output.
@@ -377,32 +306,30 @@ def create_archive(
         local_borg_version,
         global_arguments,
         borgmatic_runtime_directory,
+        archive_suffix,
         local_path,
         remote_path,
-        progress,
         json,
-        list_files,
+        comment,
         stream_processes,
     )
 
     if json:
         output_log_level = None
-    elif list_files or (stats and not dry_run):
+    elif config.get('list_details') or (config.get('statistics') and not dry_run):
         output_log_level = logging.ANSWER
     else:
         output_log_level = logging.INFO
 
     # The progress output isn't compatible with captured and logged output, as progress messes with
     # the terminal directly.
-    output_file = DO_NOT_CAPTURE if progress else None
-
-    borg_environment = environment.make_environment(config)
+    output_file = DO_NOT_CAPTURE if config.get('progress') else None
 
     create_flags += (
         (('--info',) if logger.getEffectiveLevel() == logging.INFO and not json else ())
-        + (('--stats',) if stats and not json and not dry_run else ())
+        + (('--stats',) if config.get('statistics') and not json and not dry_run else ())
         + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) and not json else ())
-        + (('--progress',) if progress else ())
+        + (('--progress',) if config.get('progress') else ())
         + (('--json',) if json else ())
     )
     borg_exit_codes = config.get('borg_exit_codes')
@@ -414,7 +341,7 @@ def create_archive(
             output_log_level,
             output_file,
             working_directory=working_directory,
-            extra_environment=borg_environment,
+            environment=environment.make_environment(config),
             borg_local_path=local_path,
             borg_exit_codes=borg_exit_codes,
         )
@@ -422,7 +349,7 @@ def create_archive(
         return execute_command_and_capture_output(
             create_flags + create_positional_arguments,
             working_directory=working_directory,
-            extra_environment=borg_environment,
+            environment=environment.make_environment(config),
             borg_local_path=local_path,
             borg_exit_codes=borg_exit_codes,
         )
@@ -432,7 +359,7 @@ def create_archive(
             output_log_level,
             output_file,
             working_directory=working_directory,
-            extra_environment=borg_environment,
+            environment=environment.make_environment(config),
             borg_local_path=local_path,
             borg_exit_codes=borg_exit_codes,
         )

borgmatic/borg/delete.py

@@ -32,9 +32,9 @@ def make_delete_command(
         + borgmatic.borg.flags.make_flags('dry-run', global_arguments.dry_run)
         + borgmatic.borg.flags.make_flags('remote-path', remote_path)
         + borgmatic.borg.flags.make_flags('umask', config.get('umask'))
-        + borgmatic.borg.flags.make_flags('log-json', global_arguments.log_json)
+        + borgmatic.borg.flags.make_flags('log-json', config.get('log_json'))
         + borgmatic.borg.flags.make_flags('lock-wait', config.get('lock_wait'))
-        + borgmatic.borg.flags.make_flags('list', delete_arguments.list_archives)
+        + borgmatic.borg.flags.make_flags('list', config.get('list_details'))
         + (
             (('--force',) + (('--force',) if delete_arguments.force >= 2 else ()))
             if delete_arguments.force
@@ -48,9 +48,17 @@ def make_delete_command(
             local_borg_version=local_borg_version,
             default_archive_name_format='*',
         )
+        + (('--stats',) if config.get('statistics') else ())
         + borgmatic.borg.flags.make_flags_from_arguments(
             delete_arguments,
-            excludes=('list_archives', 'force', 'match_archives', 'archive', 'repository'),
+            excludes=(
+                'list_details',
+                'statistics',
+                'force',
+                'match_archives',
+                'archive',
+                'repository',
+            ),
         )
         + borgmatic.borg.flags.make_repository_flags(repository['path'], local_borg_version)
     )
@@ -98,7 +106,7 @@ def delete_archives(
 
         repo_delete_arguments = argparse.Namespace(
             repository=repository['path'],
-            list_archives=delete_arguments.list_archives,
+            list_details=delete_arguments.list_details,
             force=delete_arguments.force,
             cache_only=delete_arguments.cache_only,
             keep_security_info=delete_arguments.keep_security_info,
@@ -128,7 +136,7 @@ def delete_archives(
     borgmatic.execute.execute_command(
         command,
         output_log_level=logging.ANSWER,
-        extra_environment=borgmatic.borg.environment.make_environment(config),
+        environment=borgmatic.borg.environment.make_environment(config),
         working_directory=borgmatic.config.paths.get_working_directory(config),
         borg_local_path=local_path,
         borg_exit_codes=config.get('borg_exit_codes'),

borgmatic/borg/environment.py

@@ -1,6 +1,7 @@
 import os
 
 import borgmatic.borg.passcommand
+import borgmatic.hooks.credential.parse
 
 OPTION_TO_ENVIRONMENT_VARIABLE = {
     'borg_base_directory': 'BORG_BASE_DIR',
@@ -9,43 +10,74 @@ OPTION_TO_ENVIRONMENT_VARIABLE = {
     'borg_files_cache_ttl': 'BORG_FILES_CACHE_TTL',
     'borg_security_directory': 'BORG_SECURITY_DIR',
     'borg_keys_directory': 'BORG_KEYS_DIR',
-    'encryption_passphrase': 'BORG_PASSPHRASE',
     'ssh_command': 'BORG_RSH',
     'temporary_directory': 'TMPDIR',
 }
 
-DEFAULT_BOOL_OPTION_TO_DOWNCASE_ENVIRONMENT_VARIABLE = {
-    'relocated_repo_access_is_ok': 'BORG_RELOCATED_REPO_ACCESS_IS_OK',
-    'unknown_unencrypted_repo_access_is_ok': 'BORG_UNKNOWN_UNENCRYPTED_REPO_ACCESS_IS_OK',
+DEFAULT_BOOL_OPTION_TO_UNCONDITIONAL_ENVIRONMENT_VARIABLE = {
+    'check_i_know_what_i_am_doing': 'BORG_CHECK_I_KNOW_WHAT_I_AM_DOING',
 }
 
-DEFAULT_BOOL_OPTION_TO_UPPERCASE_ENVIRONMENT_VARIABLE = {
-    'check_i_know_what_i_am_doing': 'BORG_CHECK_I_KNOW_WHAT_I_AM_DOING',
+DEFAULT_BOOL_OPTION_TO_ENVIRONMENT_VARIABLE = {
+    'debug_passphrase': 'BORG_DEBUG_PASSPHRASE',
+    'display_passphrase': 'BORG_DISPLAY_PASSPHRASE',
+    'relocated_repo_access_is_ok': 'BORG_RELOCATED_REPO_ACCESS_IS_OK',
+    'unknown_unencrypted_repo_access_is_ok': 'BORG_UNKNOWN_UNENCRYPTED_REPO_ACCESS_IS_OK',
+    'use_chunks_archive': 'BORG_USE_CHUNKS_ARCHIVE',
 }
 
 
 def make_environment(config):
     '''
-    Given a borgmatic configuration dict, return its options converted to a Borg environment
-    variable dict.
+    Given a borgmatic configuration dict, convert it to a Borg environment variable dict, merge it
+    with a copy of the current environment variables, and return the result.
+
+    Do not reuse this environment across multiple Borg invocations, because it can include
+    references to resources like anonymous pipes for passphrases—which can only be consumed once.
+
+    Here's how native Borg precedence works for a few of the environment variables:
+
+      1. BORG_PASSPHRASE, if set, is used first.
+      2. BORG_PASSCOMMAND is used only if BORG_PASSPHRASE isn't set.
+      3. BORG_PASSPHRASE_FD is used only if neither of the above are set.
+
+    In borgmatic, we want to simulate this precedence order, but there are some additional
+    complications. First, values can come from either configuration or from environment variables
+    set outside borgmatic; configured options should take precedence. Second, when borgmatic gets a
+    passphrase—directly from configuration or indirectly via a credential hook or a passcommand—we
+    want to pass that passphrase to Borg via an anonymous pipe (+ BORG_PASSPHRASE_FD), since that's
+    more secure than using an environment variable (BORG_PASSPHRASE).
     '''
-    environment = {}
+    environment = dict(os.environ)
 
     for option_name, environment_variable_name in OPTION_TO_ENVIRONMENT_VARIABLE.items():
         value = config.get(option_name)
 
-        if value:
+        if value is not None:
             environment[environment_variable_name] = str(value)
 
-    passphrase = borgmatic.borg.passcommand.get_passphrase_from_passcommand(config)
+    if 'encryption_passphrase' in config:
+        environment.pop('BORG_PASSPHRASE', None)
+        environment.pop('BORG_PASSCOMMAND', None)
 
-    # If the passcommand produced a passphrase, send it to Borg via an anonymous pipe.
-    if passphrase:
+    if 'encryption_passcommand' in config:
+        environment.pop('BORG_PASSCOMMAND', None)
+
+    passphrase = borgmatic.hooks.credential.parse.resolve_credential(
+        config.get('encryption_passphrase'), config
+    )
+
+    if passphrase is None:
+        passphrase = borgmatic.borg.passcommand.get_passphrase_from_passcommand(config)
+
+    # If there's a passphrase (from configuration, from a configured credential, or from a
+    # configured passcommand), send it to Borg via an anonymous pipe.
+    if passphrase is not None:
         read_file_descriptor, write_file_descriptor = os.pipe()
         os.write(write_file_descriptor, passphrase.encode('utf-8'))
         os.close(write_file_descriptor)
 
-        # This, plus subprocess.Popen(..., close_fds=False) in execute.py, is necessary for the Borg
+        # This plus subprocess.Popen(..., close_fds=False) in execute.py is necessary for the Borg
         # child process to inherit the file descriptor.
         os.set_inheritable(read_file_descriptor, True)
         environment['BORG_PASSPHRASE_FD'] = str(read_file_descriptor)
@@ -53,15 +85,15 @@ def make_environment(config):
     for (
         option_name,
         environment_variable_name,
-    ) in DEFAULT_BOOL_OPTION_TO_DOWNCASE_ENVIRONMENT_VARIABLE.items():
+    ) in DEFAULT_BOOL_OPTION_TO_ENVIRONMENT_VARIABLE.items():
         if os.environ.get(environment_variable_name) is None:
             value = config.get(option_name)
-            environment[environment_variable_name] = 'yes' if value else 'no'
+            environment[environment_variable_name] = 'YES' if value else 'NO'
 
     for (
         option_name,
         environment_variable_name,
-    ) in DEFAULT_BOOL_OPTION_TO_UPPERCASE_ENVIRONMENT_VARIABLE.items():
+    ) in DEFAULT_BOOL_OPTION_TO_UNCONDITIONAL_ENVIRONMENT_VARIABLE.items():
         value = config.get(option_name)
         if value is not None:
             environment[environment_variable_name] = 'YES' if value else 'NO'

borgmatic/borg/export_key.py

@@ -46,7 +46,7 @@ def export_key(
         (local_path, 'key', 'export')
         + (('--remote-path', remote_path) if remote_path else ())
         + (('--umask', str(umask)) if umask else ())
-        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--log-json',) if config.get('log_json') else ())
         + (('--lock-wait', str(lock_wait)) if lock_wait else ())
         + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
         + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
@@ -67,7 +67,7 @@ def export_key(
         full_command,
         output_file=output_file,
         output_log_level=logging.ANSWER,
-        extra_environment=environment.make_environment(config),
+        environment=environment.make_environment(config),
         working_directory=working_directory,
         borg_local_path=local_path,
         borg_exit_codes=config.get('borg_exit_codes'),

borgmatic/borg/export_tar.py

@@ -20,7 +20,6 @@ def export_tar_archive(
     local_path='borg',
     remote_path=None,
     tar_filter=None,
-    list_files=False,
     strip_components=None,
 ):
     '''
@@ -40,10 +39,10 @@ def export_tar_archive(
         (local_path, 'export-tar')
         + (('--remote-path', remote_path) if remote_path else ())
         + (('--umask', str(umask)) if umask else ())
-        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--log-json',) if config.get('log_json') else ())
         + (('--lock-wait', str(lock_wait)) if lock_wait else ())
         + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
-        + (('--list',) if list_files else ())
+        + (('--list',) if config.get('list_details') else ())
         + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
         + (('--dry-run',) if dry_run else ())
         + (('--tar-filter', tar_filter) if tar_filter else ())
@@ -57,7 +56,7 @@ def export_tar_archive(
         + (tuple(paths) if paths else ())
     )
 
-    if list_files:
+    if config.get('list_details'):
         output_log_level = logging.ANSWER
     else:
         output_log_level = logging.INFO
@@ -70,7 +69,7 @@ def export_tar_archive(
         full_command,
         output_file=DO_NOT_CAPTURE if destination_path == '-' else None,
         output_log_level=output_log_level,
-        extra_environment=environment.make_environment(config),
+        environment=environment.make_environment(config),
         working_directory=borgmatic.config.paths.get_working_directory(config),
         borg_local_path=local_path,
         borg_exit_codes=config.get('borg_exit_codes'),

borgmatic/borg/extract.py

@@ -44,11 +44,10 @@ def extract_last_archive_dry_run(
         return
 
     list_flag = ('--list',) if logger.isEnabledFor(logging.DEBUG) else ()
-    borg_environment = environment.make_environment(config)
     full_extract_command = (
         (local_path, 'extract', '--dry-run')
         + (('--remote-path', remote_path) if remote_path else ())
-        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--log-json',) if config.get('log_json') else ())
         + (('--lock-wait', str(lock_wait)) if lock_wait else ())
         + verbosity_flags
         + list_flag
@@ -59,7 +58,7 @@ def extract_last_archive_dry_run(
 
     execute_command(
         full_extract_command,
-        extra_environment=borg_environment,
+        environment=environment.make_environment(config),
         working_directory=borgmatic.config.paths.get_working_directory(config),
         borg_local_path=local_path,
         borg_exit_codes=config.get('borg_exit_codes'),
@@ -78,7 +77,6 @@ def extract_archive(
     remote_path=None,
     destination_path=None,
     strip_components=None,
-    progress=False,
     extract_to_stdout=False,
 ):
     '''
@@ -93,8 +91,8 @@ def extract_archive(
     umask = config.get('umask', None)
     lock_wait = config.get('lock_wait', None)
 
-    if progress and extract_to_stdout:
-        raise ValueError('progress and extract_to_stdout cannot both be set')
+    if config.get('progress') and extract_to_stdout:
+        raise ValueError('progress and extract to stdout cannot both be set')
 
     if feature.available(feature.Feature.NUMERIC_IDS, local_borg_version):
         numeric_ids_flags = ('--numeric-ids',) if config.get('numeric_ids') else ()
@@ -123,28 +121,25 @@ def extract_archive(
         + (('--remote-path', remote_path) if remote_path else ())
         + numeric_ids_flags
         + (('--umask', str(umask)) if umask else ())
-        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--log-json',) if config.get('log_json') else ())
         + (('--lock-wait', str(lock_wait)) if lock_wait else ())
         + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
         + (('--debug', '--list', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
         + (('--dry-run',) if dry_run else ())
         + (('--strip-components', str(strip_components)) if strip_components else ())
-        + (('--progress',) if progress else ())
+        + (('--progress',) if config.get('progress') else ())
         + (('--stdout',) if extract_to_stdout else ())
         + flags.make_repository_archive_flags(
             # Make the repository path absolute so the destination directory used below via changing
             # the working directory doesn't prevent Borg from finding the repo. But also apply the
             # user's configured working directory (if any) to the repo path.
-            borgmatic.config.validate.normalize_repository_path(
-                os.path.join(working_directory or '', repository)
-            ),
+            borgmatic.config.validate.normalize_repository_path(repository, working_directory),
             archive,
             local_borg_version,
         )
         + (tuple(paths) if paths else ())
     )
 
-    borg_environment = environment.make_environment(config)
     borg_exit_codes = config.get('borg_exit_codes')
     full_destination_path = (
         os.path.join(working_directory or '', destination_path) if destination_path else None
@@ -152,11 +147,11 @@ def extract_archive(
 
     # The progress output isn't compatible with captured and logged output, as progress messes with
     # the terminal directly.
-    if progress:
+    if config.get('progress'):
         return execute_command(
             full_command,
             output_file=DO_NOT_CAPTURE,
-            extra_environment=borg_environment,
+            environment=environment.make_environment(config),
             working_directory=full_destination_path,
             borg_local_path=local_path,
             borg_exit_codes=borg_exit_codes,
@@ -168,7 +163,7 @@ def extract_archive(
             full_command,
             output_file=subprocess.PIPE,
             run_to_completion=False,
-            extra_environment=borg_environment,
+            environment=environment.make_environment(config),
             working_directory=full_destination_path,
             borg_local_path=local_path,
             borg_exit_codes=borg_exit_codes,
@@ -178,7 +173,7 @@ def extract_archive(
     # if the restore paths don't exist in the archive.
     execute_command(
         full_command,
-        extra_environment=borg_environment,
+        environment=environment.make_environment(config),
         working_directory=full_destination_path,
         borg_local_path=local_path,
         borg_exit_codes=borg_exit_codes,

borgmatic/borg/feature.py

@@ -17,6 +17,8 @@ class Feature(Enum):
     MATCH_ARCHIVES = 11
     EXCLUDED_FILES_MINUS = 12
     ARCHIVE_SERIES = 13
+    NO_PRUNE_STATS = 14
+    DRY_RUN_COMPACT = 15
 
 
 FEATURE_TO_MINIMUM_BORG_VERSION = {
@@ -33,6 +35,8 @@ FEATURE_TO_MINIMUM_BORG_VERSION = {
     Feature.MATCH_ARCHIVES: parse('2.0.0b3'),  # borg --match-archives
     Feature.EXCLUDED_FILES_MINUS: parse('2.0.0b5'),  # --list --filter uses "-" for excludes
     Feature.ARCHIVE_SERIES: parse('2.0.0b11'),  # identically named archives form a series
+    Feature.NO_PRUNE_STATS: parse('2.0.0b10'),  # prune --stats is not available
+    Feature.DRY_RUN_COMPACT: parse('1.4.1'),  # borg compact --dry-run support
 }
 
 

borgmatic/borg/flags.py

@@ -156,3 +156,83 @@ def warn_for_aggressive_archive_flags(json_command, json_output):
         logger.debug(f'Cannot parse JSON output from archive command: {error}')
     except (TypeError, KeyError):
         logger.debug('Cannot parse JSON output from archive command: No "archives" key found')
+
+
+def omit_flag(arguments, flag):
+    '''
+    Given a sequence of Borg command-line arguments, return them with the given (valueless) flag
+    omitted. For instance, if the flag is "--flag" and arguments is:
+
+        ('borg', 'create', '--flag', '--other-flag')
+
+    ... then return:
+
+        ('borg', 'create', '--other-flag')
+    '''
+    return tuple(argument for argument in arguments if argument != flag)
+
+
+def omit_flag_and_value(arguments, flag):
+    '''
+    Given a sequence of Borg command-line arguments, return them with the given flag and its
+    corresponding value omitted. For instance, if the flag is "--flag" and arguments is:
+
+        ('borg', 'create', '--flag', 'value', '--other-flag')
+
+    ... or:
+
+        ('borg', 'create', '--flag=value', '--other-flag')
+
+    ... then return:
+
+        ('borg', 'create', '--other-flag')
+    '''
+    # This works by zipping together a list of overlapping pairwise arguments. E.g., ('one', 'two',
+    # 'three', 'four') becomes ((None, 'one'), ('one, 'two'), ('two', 'three'), ('three', 'four')).
+    # This makes it easy to "look back" at the previous arguments so we can exclude both a flag and
+    # its value.
+    return tuple(
+        argument
+        for (previous_argument, argument) in zip((None,) + arguments, arguments)
+        if flag not in (previous_argument, argument)
+        if not argument.startswith(f'{flag}=')
+    )
+
+
+def make_exclude_flags(config):
+    '''
+    Given a configuration dict with various exclude options, return the corresponding Borg flags as
+    a tuple.
+    '''
+    caches_flag = ('--exclude-caches',) if config.get('exclude_caches') else ()
+    if_present_flags = tuple(
+        itertools.chain.from_iterable(
+            ('--exclude-if-present', if_present)
+            for if_present in config.get('exclude_if_present', ())
+        )
+    )
+    keep_exclude_tags_flags = ('--keep-exclude-tags',) if config.get('keep_exclude_tags') else ()
+    exclude_nodump_flags = ('--exclude-nodump',) if config.get('exclude_nodump') else ()
+
+    return caches_flag + if_present_flags + keep_exclude_tags_flags + exclude_nodump_flags
+
+
+def make_list_filter_flags(local_borg_version, dry_run):
+    '''
+    Given the local Borg version and whether this is a dry run, return the corresponding flags for
+    passing to "--list --filter". The general idea is that excludes are shown for a dry run or when
+    the verbosity is debug.
+    '''
+    base_flags = 'AME'
+    show_excludes = logger.isEnabledFor(logging.DEBUG)
+
+    if feature.available(feature.Feature.EXCLUDED_FILES_MINUS, local_borg_version):
+        if show_excludes or dry_run:
+            return f'{base_flags}+-'
+        else:
+            return base_flags
+
+    if show_excludes:
+        return f'{base_flags}x-'
+    else:
+        return f'{base_flags}-'

borgmatic/borg/import_key.py

@@ -0,0 +1,70 @@
+import logging
+import os
+
+import borgmatic.config.paths
+import borgmatic.logger
+from borgmatic.borg import environment, flags
+from borgmatic.execute import DO_NOT_CAPTURE, execute_command
+
+logger = logging.getLogger(__name__)
+
+
+def import_key(
+    repository_path,
+    config,
+    local_borg_version,
+    import_arguments,
+    global_arguments,
+    local_path='borg',
+    remote_path=None,
+):
+    '''
+    Given a local or remote repository path, a configuration dict, the local Borg version, import
+    arguments, and optional local and remote Borg paths, import the repository key from the
+    path indicated in the import arguments.
+
+    If the path is empty or "-", then read the key from stdin.
+
+    Raise ValueError if the path is given and it does not exist.
+    '''
+    umask = config.get('umask', None)
+    lock_wait = config.get('lock_wait', None)
+    working_directory = borgmatic.config.paths.get_working_directory(config)
+
+    if import_arguments.path and import_arguments.path != '-':
+        if not os.path.exists(os.path.join(working_directory or '', import_arguments.path)):
+            raise ValueError(f'Path {import_arguments.path} does not exist. Aborting.')
+
+        input_file = None
+    else:
+        input_file = DO_NOT_CAPTURE
+
+    full_command = (
+        (local_path, 'key', 'import')
+        + (('--remote-path', remote_path) if remote_path else ())
+        + (('--umask', str(umask)) if umask else ())
+        + (('--log-json',) if config.get('log_json') else ())
+        + (('--lock-wait', str(lock_wait)) if lock_wait else ())
+        + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
+        + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
+        + flags.make_flags('paper', import_arguments.paper)
+        + flags.make_repository_flags(
+            repository_path,
+            local_borg_version,
+        )
+        + ((import_arguments.path,) if input_file is None else ())
+    )
+
+    if global_arguments.dry_run:
+        logger.info('Skipping key import (dry run)')
+        return
+
+    execute_command(
+        full_command,
+        input_file=input_file,
+        output_log_level=logging.INFO,
+        environment=environment.make_environment(config),
+        working_directory=working_directory,
+        borg_local_path=local_path,
+        borg_exit_codes=config.get('borg_exit_codes'),
+    )

borgmatic/borg/info.py

@@ -37,7 +37,7 @@ def make_info_command(
         )
         + flags.make_flags('remote-path', remote_path)
         + flags.make_flags('umask', config.get('umask'))
-        + flags.make_flags('log-json', global_arguments.log_json)
+        + flags.make_flags('log-json', config.get('log_json'))
         + flags.make_flags('lock-wait', config.get('lock_wait'))
         + (
             (
@@ -48,9 +48,7 @@ def make_info_command(
             if info_arguments.prefix
             else (
                 flags.make_match_archives_flags(
-                    info_arguments.match_archives
-                    or info_arguments.archive
-                    or config.get('match_archives'),
+                    info_arguments.archive or config.get('match_archives'),
                     config.get('archive_name_format'),
                     local_borg_version,
                 )
@@ -102,7 +100,7 @@ def display_archives_info(
 
     json_info = execute_command_and_capture_output(
         json_command,
-        extra_environment=environment.make_environment(config),
+        environment=environment.make_environment(config),
         working_directory=working_directory,
         borg_local_path=local_path,
         borg_exit_codes=borg_exit_codes,
@@ -116,7 +114,7 @@ def display_archives_info(
     execute_command(
         main_command,
         output_log_level=logging.ANSWER,
-        extra_environment=environment.make_environment(config),
+        environment=environment.make_environment(config),
         working_directory=working_directory,
         borg_local_path=local_path,
         borg_exit_codes=borg_exit_codes,

borgmatic/borg/list.py

@@ -48,7 +48,7 @@ def make_list_command(
         )
         + flags.make_flags('remote-path', remote_path)
         + flags.make_flags('umask', config.get('umask'))
-        + flags.make_flags('log-json', global_arguments.log_json)
+        + flags.make_flags('log-json', config.get('log_json'))
         + flags.make_flags('lock-wait', config.get('lock_wait'))
         + flags.make_flags_from_arguments(list_arguments, excludes=MAKE_FLAGS_EXCLUDES)
         + (
@@ -106,8 +106,6 @@ def capture_archive_listing(
     format to use for the output, and local and remote Borg paths, capture the
     output of listing that archive and return it as a list of file paths.
     '''
-    borg_environment = environment.make_environment(config)
-
     return tuple(
         execute_command_and_capture_output(
             make_list_command(
@@ -126,7 +124,7 @@ def capture_archive_listing(
                 local_path,
                 remote_path,
             ),
-            extra_environment=borg_environment,
+            environment=environment.make_environment(config),
             working_directory=borgmatic.config.paths.get_working_directory(config),
             borg_local_path=local_path,
             borg_exit_codes=config.get('borg_exit_codes'),
@@ -194,7 +192,6 @@ def list_archive(
             'The --json flag on the list action is not supported when using the --archive/--find flags.'
         )
 
-    borg_environment = environment.make_environment(config)
     borg_exit_codes = config.get('borg_exit_codes')
 
     # If there are any paths to find (and there's not a single archive already selected), start by
@@ -224,7 +221,7 @@ def list_archive(
                     local_path,
                     remote_path,
                 ),
-                extra_environment=borg_environment,
+                environment=environment.make_environment(config),
                 working_directory=borgmatic.config.paths.get_working_directory(config),
                 borg_local_path=local_path,
                 borg_exit_codes=borg_exit_codes,
@@ -260,7 +257,7 @@ def list_archive(
         execute_command(
             main_command,
             output_log_level=logging.ANSWER,
-            extra_environment=borg_environment,
+            environment=environment.make_environment(config),
             working_directory=borgmatic.config.paths.get_working_directory(config),
             borg_local_path=local_path,
             borg_exit_codes=borg_exit_codes,

borgmatic/borg/mount.py

@@ -30,7 +30,7 @@ def mount_archive(
         (local_path, 'mount')
         + (('--remote-path', remote_path) if remote_path else ())
         + (('--umask', str(umask)) if umask else ())
-        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--log-json',) if config.get('log_json') else ())
         + (('--lock-wait', str(lock_wait)) if lock_wait else ())
         + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
         + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
@@ -59,7 +59,6 @@ def mount_archive(
         + (tuple(mount_arguments.paths) if mount_arguments.paths else ())
     )
 
-    borg_environment = environment.make_environment(config)
     working_directory = borgmatic.config.paths.get_working_directory(config)
 
     # Don't capture the output when foreground mode is used so that ctrl-C can work properly.
@@ -67,7 +66,7 @@ def mount_archive(
         execute_command(
             full_command,
             output_file=DO_NOT_CAPTURE,
-            extra_environment=borg_environment,
+            environment=environment.make_environment(config),
             working_directory=working_directory,
             borg_local_path=local_path,
             borg_exit_codes=config.get('borg_exit_codes'),
@@ -76,7 +75,7 @@ def mount_archive(
 
     execute_command(
         full_command,
-        extra_environment=borg_environment,
+        environment=environment.make_environment(config),
         working_directory=working_directory,
         borg_local_path=local_path,
         borg_exit_codes=config.get('borg_exit_codes'),

borgmatic/borg/passcommand.py

@@ -9,21 +9,14 @@ logger = logging.getLogger(__name__)
 
 
 @functools.cache
-def run_passcommand(passcommand, passphrase_configured, working_directory):
+def run_passcommand(passcommand, working_directory):
     '''
     Run the given passcommand using the given working directory and return the passphrase produced
-    by the command. But bail first if a passphrase is already configured; this mimics Borg's
-    behavior.
+    by the command.
 
     Cache the results so that the passcommand only needs to run—and potentially prompt the user—once
     per borgmatic invocation.
     '''
-    if passcommand and passphrase_configured:
-        logger.warning(
-            'Ignoring the "encryption_passcommand" option because "encryption_passphrase" is set'
-        )
-        return None
-
     return borgmatic.execute.execute_command_and_capture_output(
         shlex.split(passcommand),
         working_directory=working_directory,
@@ -44,7 +37,4 @@ def get_passphrase_from_passcommand(config):
     if not passcommand:
         return None
 
-    passphrase = config.get('encryption_passphrase')
-    working_directory = borgmatic.config.paths.get_working_directory(config)
-
-    return run_passcommand(passcommand, bool(passphrase), working_directory)
+    return run_passcommand(passcommand, borgmatic.config.paths.get_working_directory(config))

borgmatic/borg/pattern.py

@@ -1,5 +1,12 @@
 import collections
 import enum
+import logging
+import os
+import tempfile
+
+import borgmatic.borg.pattern
+
+logger = logging.getLogger(__name__)
 
 
 # See https://borgbackup.readthedocs.io/en/stable/usage/help.html#borg-help-patterns
@@ -20,12 +27,80 @@ class Pattern_style(enum.Enum):
     PATH_FULL_MATCH = 'pf'
 
 
+class Pattern_source(enum.Enum):
+    '''
+    Where the pattern came from within borgmatic. This is important because certain use cases (like
+    filesystem snapshotting) only want to consider patterns that the user actually put in a
+    configuration file and not patterns from other sources.
+    '''
+
+    # The pattern is from a borgmatic configuration option, e.g. listed in "source_directories".
+    CONFIG = 'config'
+
+    # The pattern is generated internally within borgmatic, e.g. for special file excludes.
+    INTERNAL = 'internal'
+
+    # The pattern originates from within a borgmatic hook, e.g. a database hook that adds its dump
+    # directory.
+    HOOK = 'hook'
+
+
 Pattern = collections.namedtuple(
     'Pattern',
-    ('path', 'type', 'style', 'device'),
+    ('path', 'type', 'style', 'device', 'source'),
     defaults=(
         Pattern_type.ROOT,
         Pattern_style.NONE,
         None,
+        Pattern_source.HOOK,
     ),
 )
+
+
+def write_patterns_file(patterns, borgmatic_runtime_directory, patterns_file=None):
+    '''
+    Given a sequence of patterns as borgmatic.borg.pattern.Pattern instances, write them to a named
+    temporary file in the given borgmatic runtime directory and return the file object so it can
+    continue to exist on disk as long as the caller needs it.
+
+    If an optional open pattern file is given, append to it instead of making a new temporary file.
+    Return None if no patterns are provided.
+    '''
+    if not patterns:
+        return None
+
+    if patterns_file is None:
+        patterns_file = tempfile.NamedTemporaryFile('w', dir=borgmatic_runtime_directory)
+        operation_name = 'Writing'
+    else:
+        patterns_file.write('\n')
+        operation_name = 'Appending'
+
+    patterns_output = '\n'.join(
+        f'{pattern.type.value} {pattern.style.value}{":" if pattern.style.value else ""}{pattern.path}'
+        for pattern in patterns
+    )
+    logger.debug(f'{operation_name} patterns to {patterns_file.name}:\n{patterns_output}')
+
+    patterns_file.write(patterns_output)
+    patterns_file.flush()
+
+    return patterns_file
+
+
+def check_all_root_patterns_exist(patterns):
+    '''
+    Given a sequence of borgmatic.borg.pattern.Pattern instances, check that all root pattern
+    paths exist. If any don't, raise an exception.
+    '''
+    missing_paths = [
+        pattern.path
+        for pattern in patterns
+        if pattern.type == borgmatic.borg.pattern.Pattern_type.ROOT
+        if not os.path.exists(pattern.path)
+    ]
+
+    if missing_paths:
+        raise ValueError(
+            f"Source directories or root pattern paths do not exist: {', '.join(missing_paths)}"
+        )

borgmatic/borg/prune.py

@@ -1,4 +1,5 @@
 import logging
+import shlex
 
 import borgmatic.config.paths
 import borgmatic.logger
@@ -41,7 +42,7 @@ def make_prune_flags(config, prune_arguments, local_borg_version):
         if prefix
         else (
             flags.make_match_archives_flags(
-                prune_arguments.match_archives or config.get('match_archives'),
+                config.get('match_archives'),
                 config.get('archive_name_format'),
                 local_borg_version,
             )
@@ -73,22 +74,28 @@ def prune_archives(
         + make_prune_flags(config, prune_arguments, local_borg_version)
         + (('--remote-path', remote_path) if remote_path else ())
         + (('--umask', str(umask)) if umask else ())
-        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--log-json',) if config.get('log_json') else ())
         + (('--lock-wait', str(lock_wait)) if lock_wait else ())
-        + (('--stats',) if prune_arguments.stats and not dry_run else ())
+        + (
+            ('--stats',)
+            if config.get('statistics')
+            and not dry_run
+            and not feature.available(feature.Feature.NO_PRUNE_STATS, local_borg_version)
+            else ()
+        )
         + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
         + flags.make_flags_from_arguments(
             prune_arguments,
-            excludes=('repository', 'match_archives', 'stats', 'list_archives'),
+            excludes=('repository', 'match_archives', 'statistics', 'list_details'),
         )
-        + (('--list',) if prune_arguments.list_archives else ())
+        + (('--list',) if config.get('list_details') else ())
         + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
         + (('--dry-run',) if dry_run else ())
-        + (tuple(extra_borg_options.split(' ')) if extra_borg_options else ())
+        + (tuple(shlex.split(extra_borg_options)) if extra_borg_options else ())
         + flags.make_repository_flags(repository_path, local_borg_version)
     )
 
-    if prune_arguments.stats or prune_arguments.list_archives:
+    if config.get('statistics') or config.get('list_details'):
         output_log_level = logging.ANSWER
     else:
         output_log_level = logging.INFO
@@ -96,7 +103,7 @@ def prune_archives(
     execute_command(
         full_command,
         output_log_level=output_log_level,
-        extra_environment=environment.make_environment(config),
+        environment=environment.make_environment(config),
         working_directory=borgmatic.config.paths.get_working_directory(config),
         borg_local_path=local_path,
         borg_exit_codes=config.get('borg_exit_codes'),

borgmatic/borg/recreate.py

@@ -0,0 +1,104 @@
+import logging
+import shlex
+
+import borgmatic.borg.environment
+import borgmatic.borg.feature
+import borgmatic.config.paths
+import borgmatic.execute
+from borgmatic.borg import flags
+from borgmatic.borg.pattern import write_patterns_file
+
+logger = logging.getLogger(__name__)
+
+
+def recreate_archive(
+    repository,
+    archive,
+    config,
+    local_borg_version,
+    recreate_arguments,
+    global_arguments,
+    local_path,
+    remote_path=None,
+    patterns=None,
+):
+    '''
+    Given a local or remote repository path, an archive name, a configuration dict, the local Borg
+    version string, an argparse.Namespace of recreate arguments, an argparse.Namespace of global
+    arguments, optional local and remote Borg paths, executes the recreate command with the given
+    arguments.
+    '''
+    lock_wait = config.get('lock_wait', None)
+    exclude_flags = flags.make_exclude_flags(config)
+    compression = config.get('compression', None)
+    chunker_params = config.get('chunker_params', None)
+
+    # Available recompress MODES: "if-different", "always", "never" (default)
+    recompress = config.get('recompress', None)
+
+    # Write patterns to a temporary file and use that file with --patterns-from.
+    patterns_file = write_patterns_file(
+        patterns, borgmatic.config.paths.get_working_directory(config)
+    )
+
+    recreate_command = (
+        (local_path, 'recreate')
+        + (('--remote-path', remote_path) if remote_path else ())
+        + (('--log-json',) if config.get('log_json') else ())
+        + (('--lock-wait', str(lock_wait)) if lock_wait is not None else ())
+        + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
+        + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
+        + (('--patterns-from', patterns_file.name) if patterns_file else ())
+        + (
+            (
+                '--list',
+                '--filter',
+                flags.make_list_filter_flags(local_borg_version, global_arguments.dry_run),
+            )
+            if config.get('list_details')
+            else ()
+        )
+        # Flag --target works only for a single archive.
+        + (('--target', recreate_arguments.target) if recreate_arguments.target and archive else ())
+        + (
+            ('--comment', shlex.quote(recreate_arguments.comment))
+            if recreate_arguments.comment
+            else ()
+        )
+        + (('--timestamp', recreate_arguments.timestamp) if recreate_arguments.timestamp else ())
+        + (('--compression', compression) if compression else ())
+        + (('--chunker-params', chunker_params) if chunker_params else ())
+        + (('--recompress', recompress) if recompress else ())
+        + exclude_flags
+        + (
+            (
+                flags.make_repository_flags(repository, local_borg_version)
+                + flags.make_match_archives_flags(
+                    archive or config.get('match_archives'),
+                    config.get('archive_name_format'),
+                    local_borg_version,
+                )
+            )
+            if borgmatic.borg.feature.available(
+                borgmatic.borg.feature.Feature.SEPARATE_REPOSITORY_ARCHIVE, local_borg_version
+            )
+            else (
+                flags.make_repository_archive_flags(repository, archive, local_borg_version)
+                if archive
+                else flags.make_repository_flags(repository, local_borg_version)
+            )
+        )
+    )
+
+    if global_arguments.dry_run:
+        logger.info('Skipping the archive recreation (dry run)')
+        return
+
+    borgmatic.execute.execute_command(
+        full_command=recreate_command,
+        output_log_level=logging.INFO,
+        environment=borgmatic.borg.environment.make_environment(config),
+        working_directory=borgmatic.config.paths.get_working_directory(config),
+        borg_local_path=local_path,
+        borg_exit_codes=config.get('borg_exit_codes'),
+    )

borgmatic/borg/rename.py

@@ -0,0 +1,62 @@
+import logging
+
+import borgmatic.borg.flags
+
+logger = logging.getLogger(__name__)
+
+
+def make_rename_command(
+    dry_run,
+    repository_name,
+    old_archive_name,
+    new_archive_name,
+    config,
+    local_borg_version,
+    local_path,
+    remote_path,
+):
+    return (
+        (local_path, 'rename')
+        + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
+        + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
+        + borgmatic.borg.flags.make_flags('dry-run', dry_run)
+        + borgmatic.borg.flags.make_flags('remote-path', remote_path)
+        + borgmatic.borg.flags.make_flags('umask', config.get('umask'))
+        + borgmatic.borg.flags.make_flags('log-json', config.get('log_json'))
+        + borgmatic.borg.flags.make_flags('lock-wait', config.get('lock_wait'))
+        + borgmatic.borg.flags.make_repository_archive_flags(
+            repository_name, old_archive_name, local_borg_version
+        )
+        + (new_archive_name,)
+    )
+
+
+def rename_archive(
+    repository_name,
+    old_archive_name,
+    new_archive_name,
+    dry_run,
+    config,
+    local_borg_version,
+    local_path,
+    remote_path,
+):
+    command = make_rename_command(
+        dry_run,
+        repository_name,
+        old_archive_name,
+        new_archive_name,
+        config,
+        local_borg_version,
+        local_path,
+        remote_path,
+    )
+
+    borgmatic.execute.execute_command(
+        command,
+        output_log_level=logging.INFO,
+        environment=borgmatic.borg.environment.make_environment(config),
+        working_directory=borgmatic.config.paths.get_working_directory(config),
+        borg_local_path=local_path,
+        borg_exit_codes=config.get('borg_exit_codes'),
+    )

borgmatic/borg/repo_create.py

@@ -1,6 +1,7 @@
 import argparse
 import json
 import logging
+import shlex
 import subprocess
 
 import borgmatic.config.paths
@@ -24,7 +25,7 @@ def create_repository(
     copy_crypt_key=False,
     append_only=None,
     storage_quota=None,
-    make_parent_dirs=False,
+    make_parent_directories=False,
     local_path='borg',
     remote_path=None,
 ):
@@ -79,14 +80,14 @@ def create_repository(
         + (('--copy-crypt-key',) if copy_crypt_key else ())
         + (('--append-only',) if append_only else ())
         + (('--storage-quota', storage_quota) if storage_quota else ())
-        + (('--make-parent-dirs',) if make_parent_dirs else ())
+        + (('--make-parent-dirs',) if make_parent_directories else ())
         + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
         + (('--debug',) if logger.isEnabledFor(logging.DEBUG) else ())
-        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--log-json',) if config.get('log_json') else ())
         + (('--lock-wait', str(lock_wait)) if lock_wait else ())
         + (('--remote-path', remote_path) if remote_path else ())
         + (('--umask', str(umask)) if umask else ())
-        + (tuple(extra_borg_options.split(' ')) if extra_borg_options else ())
+        + (tuple(shlex.split(extra_borg_options)) if extra_borg_options else ())
         + flags.make_repository_flags(repository_path, local_borg_version)
     )
 
@@ -98,7 +99,7 @@ def create_repository(
     execute_command(
         repo_create_command,
         output_file=DO_NOT_CAPTURE,
-        extra_environment=environment.make_environment(config),
+        environment=environment.make_environment(config),
         working_directory=borgmatic.config.paths.get_working_directory(config),
         borg_local_path=local_path,
         borg_exit_codes=config.get('borg_exit_codes'),

borgmatic/borg/repo_delete.py

@@ -37,16 +37,16 @@ def make_repo_delete_command(
         + borgmatic.borg.flags.make_flags('dry-run', global_arguments.dry_run)
         + borgmatic.borg.flags.make_flags('remote-path', remote_path)
         + borgmatic.borg.flags.make_flags('umask', config.get('umask'))
-        + borgmatic.borg.flags.make_flags('log-json', global_arguments.log_json)
+        + borgmatic.borg.flags.make_flags('log-json', config.get('log_json'))
         + borgmatic.borg.flags.make_flags('lock-wait', config.get('lock_wait'))
-        + borgmatic.borg.flags.make_flags('list', repo_delete_arguments.list_archives)
+        + borgmatic.borg.flags.make_flags('list', config.get('list_details'))
         + (
             (('--force',) + (('--force',) if repo_delete_arguments.force >= 2 else ()))
             if repo_delete_arguments.force
             else ()
         )
         + borgmatic.borg.flags.make_flags_from_arguments(
-            repo_delete_arguments, excludes=('list_archives', 'force', 'repository')
+            repo_delete_arguments, excludes=('list_details', 'force', 'repository')
         )
         + borgmatic.borg.flags.make_repository_flags(repository['path'], local_borg_version)
     )
@@ -88,7 +88,7 @@ def delete_repository(
             if repo_delete_arguments.force or repo_delete_arguments.cache_only
             else borgmatic.execute.DO_NOT_CAPTURE
         ),
-        extra_environment=borgmatic.borg.environment.make_environment(config),
+        environment=borgmatic.borg.environment.make_environment(config),
         working_directory=borgmatic.config.paths.get_working_directory(config),
         borg_local_path=local_path,
         borg_exit_codes=config.get('borg_exit_codes'),

borgmatic/borg/repo_info.py

@@ -44,20 +44,19 @@ def display_repository_info(
         )
         + flags.make_flags('remote-path', remote_path)
         + flags.make_flags('umask', config.get('umask'))
-        + flags.make_flags('log-json', global_arguments.log_json)
+        + flags.make_flags('log-json', config.get('log_json'))
         + flags.make_flags('lock-wait', lock_wait)
         + (('--json',) if repo_info_arguments.json else ())
         + flags.make_repository_flags(repository_path, local_borg_version)
     )
 
-    extra_environment = environment.make_environment(config)
     working_directory = borgmatic.config.paths.get_working_directory(config)
     borg_exit_codes = config.get('borg_exit_codes')
 
     if repo_info_arguments.json:
         return execute_command_and_capture_output(
             full_command,
-            extra_environment=extra_environment,
+            environment=environment.make_environment(config),
             working_directory=working_directory,
             borg_local_path=local_path,
             borg_exit_codes=borg_exit_codes,
@@ -66,7 +65,7 @@ def display_repository_info(
         execute_command(
             full_command,
             output_log_level=logging.ANSWER,
-            extra_environment=extra_environment,
+            environment=environment.make_environment(config),
             working_directory=working_directory,
             borg_local_path=local_path,
             borg_exit_codes=borg_exit_codes,

borgmatic/borg/repo_list.py

@@ -1,4 +1,5 @@
 import argparse
+import json
 import logging
 
 import borgmatic.config.paths
@@ -29,6 +30,33 @@ def resolve_archive_name(
     if archive != 'latest':
         return archive
 
+    latest_archive = get_latest_archive(
+        repository_path,
+        config,
+        local_borg_version,
+        global_arguments,
+        local_path=local_path,
+        remote_path=remote_path,
+    )
+
+    return latest_archive['name']
+
+
+def get_latest_archive(
+    repository_path,
+    config,
+    local_borg_version,
+    global_arguments,
+    local_path='borg',
+    remote_path=None,
+    consider_checkpoints=False,
+):
+    '''
+    Returns a dict with information about the latest archive of a repository.
+
+    Raises ValueError if there are no archives in the repository.
+    '''
+
     full_command = (
         (
             local_path,
@@ -40,26 +68,30 @@ def resolve_archive_name(
         )
         + flags.make_flags('remote-path', remote_path)
         + flags.make_flags('umask', config.get('umask'))
-        + flags.make_flags('log-json', global_arguments.log_json)
+        + flags.make_flags('log-json', config.get('log_json'))
         + flags.make_flags('lock-wait', config.get('lock_wait'))
+        + flags.make_flags('consider-checkpoints', consider_checkpoints)
         + flags.make_flags('last', 1)
-        + ('--short',)
+        + ('--json',)
         + flags.make_repository_flags(repository_path, local_borg_version)
     )
 
-    output = execute_command_and_capture_output(
+    json_output = execute_command_and_capture_output(
         full_command,
-        extra_environment=environment.make_environment(config),
+        environment=environment.make_environment(config),
         working_directory=borgmatic.config.paths.get_working_directory(config),
         borg_local_path=local_path,
         borg_exit_codes=config.get('borg_exit_codes'),
     )
+
+    archives = json.loads(json_output)['archives']
+
     try:
-        latest_archive = output.strip().splitlines()[-1]
+        latest_archive = archives[-1]
     except IndexError:
         raise ValueError('No archives found in the repository')
 
-    logger.debug(f'Latest archive is {latest_archive}')
+    logger.debug(f'Latest archive is {latest_archive["name"]}')
 
     return latest_archive
 
@@ -102,7 +134,7 @@ def make_repo_list_command(
         )
         + flags.make_flags('remote-path', remote_path)
         + flags.make_flags('umask', config.get('umask'))
-        + flags.make_flags('log-json', global_arguments.log_json)
+        + flags.make_flags('log-json', config.get('log_json'))
         + flags.make_flags('lock-wait', config.get('lock_wait'))
         + (
             (
@@ -113,7 +145,7 @@ def make_repo_list_command(
             if repo_list_arguments.prefix
             else (
                 flags.make_match_archives_flags(
-                    repo_list_arguments.match_archives or config.get('match_archives'),
+                    config.get('match_archives'),
                     config.get('archive_name_format'),
                     local_borg_version,
                 )
@@ -140,7 +172,6 @@ def list_repository(
     return JSON output).
     '''
     borgmatic.logger.add_custom_log_levels()
-    borg_environment = environment.make_environment(config)
 
     main_command = make_repo_list_command(
         repository_path,
@@ -165,7 +196,7 @@ def list_repository(
 
     json_listing = execute_command_and_capture_output(
         json_command,
-        extra_environment=borg_environment,
+        environment=environment.make_environment(config),
         working_directory=working_directory,
         borg_local_path=local_path,
         borg_exit_codes=borg_exit_codes,
@@ -179,7 +210,7 @@ def list_repository(
     execute_command(
         main_command,
         output_log_level=logging.ANSWER,
-        extra_environment=borg_environment,
+        environment=environment.make_environment(config),
         working_directory=working_directory,
         borg_local_path=local_path,
         borg_exit_codes=borg_exit_codes,

borgmatic/borg/transfer.py

@@ -31,18 +31,23 @@ def transfer_archives(
         + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
         + flags.make_flags('remote-path', remote_path)
         + flags.make_flags('umask', config.get('umask'))
-        + flags.make_flags('log-json', global_arguments.log_json)
-        + flags.make_flags('lock-wait', config.get('lock_wait', None))
+        + flags.make_flags('log-json', config.get('log_json'))
+        + flags.make_flags('lock-wait', config.get('lock_wait'))
+        + flags.make_flags('progress', config.get('progress'))
         + (
             flags.make_flags_from_arguments(
                 transfer_arguments,
-                excludes=('repository', 'source_repository', 'archive', 'match_archives'),
+                excludes=(
+                    'repository',
+                    'source_repository',
+                    'archive',
+                    'match_archives',
+                    'progress',
+                ),
             )
             or (
                 flags.make_match_archives_flags(
-                    transfer_arguments.match_archives
-                    or transfer_arguments.archive
-                    or config.get('match_archives'),
+                    transfer_arguments.archive or config.get('match_archives'),
                     config.get('archive_name_format'),
                     local_borg_version,
                 )
@@ -56,8 +61,8 @@ def transfer_archives(
     return execute_command(
         full_command,
         output_log_level=logging.ANSWER,
-        output_file=DO_NOT_CAPTURE if transfer_arguments.progress else None,
-        extra_environment=environment.make_environment(config),
+        output_file=DO_NOT_CAPTURE if config.get('progress') else None,
+        environment=environment.make_environment(config),
         working_directory=borgmatic.config.paths.get_working_directory(config),
         borg_local_path=local_path,
         borg_exit_codes=config.get('borg_exit_codes'),

borgmatic/borg/version.py

@@ -21,7 +21,7 @@ def local_borg_version(config, local_path='borg'):
     )
     output = execute_command_and_capture_output(
         full_command,
-        extra_environment=environment.make_environment(config),
+        environment=environment.make_environment(config),
         working_directory=borgmatic.config.paths.get_working_directory(config),
         borg_local_path=local_path,
         borg_exit_codes=config.get('borg_exit_codes'),

borgmatic/commands/arguments.py

@@ -1,8 +1,13 @@
 import collections
+import io
 import itertools
+import re
 import sys
 from argparse import ArgumentParser
 
+import ruamel.yaml
+
+import borgmatic.config.schema
 from borgmatic.config import collect
 
 ACTION_ALIASES = {
@@ -27,6 +32,7 @@ ACTION_ALIASES = {
     'break-lock': [],
     'key': [],
     'borg': [],
+    'recreate': [],
 }
 
 
@@ -63,9 +69,9 @@ def get_subactions_for_actions(action_parsers):
 
 def omit_values_colliding_with_action_names(unparsed_arguments, parsed_arguments):
     '''
-    Given a sequence of string arguments and a dict from action name to parsed argparse.Namespace
-    arguments, return the string arguments with any values omitted that happen to be the same as
-    the name of a borgmatic action.
+    Given unparsed arguments as a sequence of strings and a dict from action name to parsed
+    argparse.Namespace arguments, return the string arguments with any values omitted that happen to
+    be the same as the name of a borgmatic action.
 
     This prevents, for instance, "check --only extract" from triggering the "extract" action.
     '''
@@ -282,17 +288,278 @@ def parse_arguments_for_actions(unparsed_arguments, action_parsers, global_parse
     )
 
 
-def make_parsers():
+OMITTED_FLAG_NAMES = {'match-archives', 'progress', 'statistics', 'list-details'}
+
+
+def make_argument_description(schema, flag_name):
     '''
-    Build a global arguments parser, individual action parsers, and a combined parser containing
-    both. Return them as a tuple. The global parser is useful for parsing just global arguments
-    while ignoring actions, and the combined parser is handy for displaying help that includes
-    everything: global flags, a list of actions, etc.
+    Given a configuration schema dict and a flag name for it, extend the schema's description with
+    an example or additional information as appropriate based on its type. Return the updated
+    description for use in a command-line argument.
+    '''
+    description = schema.get('description')
+    schema_type = schema.get('type')
+    example = schema.get('example')
+    pieces = [description] if description else []
+
+    if '[0]' in flag_name:
+        pieces.append(
+            ' To specify a different list element, replace the "[0]" with another array index ("[1]", "[2]", etc.).'
+        )
+
+    if example and schema_type in ('array', 'object'):
+        example_buffer = io.StringIO()
+        yaml = ruamel.yaml.YAML(typ='safe')
+        yaml.default_flow_style = True
+        yaml.dump(example, example_buffer)
+
+        pieces.append(f'Example value: "{example_buffer.getvalue().strip()}"')
+
+    return ' '.join(pieces).replace('%', '%%')
+
+
+def add_array_element_arguments(arguments_group, unparsed_arguments, flag_name):
+    r'''
+    Given an argparse._ArgumentGroup instance, a sequence of unparsed argument strings, and a dotted
+    flag name, add command-line array element flags that correspond to the given unparsed arguments.
+
+    Here's the background. We want to support flags that can have arbitrary indices like:
+
+      --foo.bar[1].baz
+
+    But argparse doesn't support that natively because the index can be an arbitrary number. We
+    won't let that stop us though, will we?
+
+    If the current flag name has an array component in it (e.g. a name with "[0]"), then make a
+    pattern that would match the flag name regardless of the number that's in it. The idea is that
+    we want to look for unparsed arguments that appear like the flag name, but instead of "[0]" they
+    have, say, "[1]" or "[123]".
+
+    Next, we check each unparsed argument against that pattern. If one of them matches, add an
+    argument flag for it to the argument parser group. Example:
+
+    Let's say flag_name is:
+
+        --foo.bar[0].baz
+
+    ... then the regular expression pattern will be:
+
+        ^--foo\.bar\[\d+\]\.baz
+
+    ... and, if that matches an unparsed argument of:
+
+        --foo.bar[1].baz
+
+    ... then an argument flag will get added equal to that unparsed argument. And so the unparsed
+    argument will match it when parsing is performed! In this manner, we're using the actual user
+    CLI input to inform what exact flags we support.
+    '''
+    if '[0]' not in flag_name or not unparsed_arguments or '--help' in unparsed_arguments:
+        return
+
+    pattern = re.compile('^--' + flag_name.replace('[0]', r'\[\d+\]').replace('.', r'\.') + '$')
+
+    try:
+        # Find an existing list index flag (and its action) corresponding to the given flag name.
+        (argument_action, existing_flag_name) = next(
+            (action, action_flag_name)
+            for action in arguments_group._group_actions
+            for action_flag_name in action.option_strings
+            if pattern.match(action_flag_name)
+            if f'--{flag_name}'.startswith(action_flag_name)
+        )
+
+        # Based on the type of the action (e.g. argparse._StoreTrueAction), look up the corresponding
+        # action registry name (e.g., "store_true") to pass to add_argument(action=...) below.
+        action_registry_name = next(
+            registry_name
+            for registry_name, action_type in arguments_group._registries['action'].items()
+            # Not using isinstance() here because we only want an exact match—no parent classes.
+            if type(argument_action) is action_type
+        )
+    except StopIteration:
+        return
+
+    for unparsed in unparsed_arguments:
+        unparsed_flag_name = unparsed.split('=', 1)[0]
+        destination_name = unparsed_flag_name.lstrip('-').replace('-', '_')
+
+        if not pattern.match(unparsed_flag_name) or unparsed_flag_name == existing_flag_name:
+            continue
+
+        if action_registry_name in ('store_true', 'store_false'):
+            arguments_group.add_argument(
+                unparsed_flag_name,
+                action=action_registry_name,
+                default=argument_action.default,
+                dest=destination_name,
+                required=argument_action.nargs,
+            )
+        else:
+            arguments_group.add_argument(
+                unparsed_flag_name,
+                action=action_registry_name,
+                choices=argument_action.choices,
+                default=argument_action.default,
+                dest=destination_name,
+                nargs=argument_action.nargs,
+                required=argument_action.nargs,
+                type=argument_action.type,
+            )
+
+
+def add_arguments_from_schema(arguments_group, schema, unparsed_arguments, names=None):
+    '''
+    Given an argparse._ArgumentGroup instance, a configuration schema dict, and a sequence of
+    unparsed argument strings, convert the entire schema into corresponding command-line flags and
+    add them to the arguments group.
+
+    For instance, given a schema of:
+
+        {
+            'type': 'object',
+            'properties': {
+                'foo': {
+                    'type': 'object',
+                    'properties': {
+                        'bar': {'type': 'integer'}
+                    }
+                }
+            }
+        }
+
+    ... the following flag will be added to the arguments group:
+
+        --foo.bar
+
+    If "foo" is instead an array of objects, both of the following will get added:
+
+        --foo
+        --foo[0].bar
+
+    And if names are also passed in, they are considered to be the name components of an option
+    (e.g. "foo" and "bar") and are used to construct a resulting flag.
+
+    Bail if the schema is not a dict.
+    '''
+    if names is None:
+        names = ()
+
+    if not isinstance(schema, dict):
+        return
+
+    schema_type = schema.get('type')
+
+    # If this option has multiple types, just use the first one (that isn't "null").
+    if isinstance(schema_type, list):
+        try:
+            schema_type = next(single_type for single_type in schema_type if single_type != 'null')
+        except StopIteration:
+            raise ValueError(f'Unknown type in configuration schema: {schema_type}')
+
+    # If this is an "object" type, recurse for each child option ("property").
+    if schema_type == 'object':
+        properties = schema.get('properties')
+
+        # If there are child properties, recurse for each one. But if there are no child properties,
+        # fall through so that a flag gets added below for the (empty) object.
+        if properties:
+            for name, child in properties.items():
+                add_arguments_from_schema(
+                    arguments_group, child, unparsed_arguments, names + (name,)
+                )
+
+            return
+
+    # If this is an "array" type, recurse for each items type child option. Don't return yet so that
+    # a flag also gets added below for the array itself.
+    if schema_type == 'array':
+        items = schema.get('items', {})
+        properties = borgmatic.config.schema.get_properties(items)
+
+        if properties:
+            for name, child in properties.items():
+                add_arguments_from_schema(
+                    arguments_group,
+                    child,
+                    unparsed_arguments,
+                    names[:-1] + (f'{names[-1]}[0]',) + (name,),
+                )
+        # If there aren't any children, then this is an array of scalars. Recurse accordingly.
+        else:
+            add_arguments_from_schema(
+                arguments_group, items, unparsed_arguments, names[:-1] + (f'{names[-1]}[0]',)
+            )
+
+    flag_name = '.'.join(names).replace('_', '-')
+
+    # Certain options already have corresponding flags on individual actions (like "create
+    # --progress"), so don't bother adding them to the global flags.
+    if not flag_name or flag_name in OMITTED_FLAG_NAMES:
+        return
+
+    metavar = names[-1].upper()
+    description = make_argument_description(schema, flag_name)
+
+    # The object=str and array=str given here is to support specifying an object or an array as a
+    # YAML string on the command-line.
+    argument_type = borgmatic.config.schema.parse_type(schema_type, object=str, array=str)
+
+    # As a UX nicety, add separate true and false flags for boolean options.
+    if schema_type == 'boolean':
+        arguments_group.add_argument(
+            f'--{flag_name}',
+            action='store_true',
+            default=None,
+            help=description,
+        )
+
+        if names[-1].startswith('no_'):
+            no_flag_name = '.'.join(names[:-1] + (names[-1][len('no_') :],)).replace('_', '-')
+        else:
+            no_flag_name = '.'.join(names[:-1] + ('no-' + names[-1],)).replace('_', '-')
+
+        arguments_group.add_argument(
+            f'--{no_flag_name}',
+            dest=flag_name.replace('-', '_'),
+            action='store_false',
+            default=None,
+            help=f'Set the --{flag_name} value to false.',
+        )
+    elif flag_name == 'verbosity':
+        arguments_group.add_argument(
+            '-v',
+            '--verbosity',
+            type=argument_type,
+            metavar=metavar,
+            help=description,
+        )
+    else:
+        arguments_group.add_argument(
+            f'--{flag_name}',
+            type=argument_type,
+            metavar=metavar,
+            help=description,
+        )
+
+    add_array_element_arguments(arguments_group, unparsed_arguments, flag_name)
+
+
+def make_parsers(schema, unparsed_arguments):
+    '''
+    Given a configuration schema dict and unparsed arguments as a sequence of strings, build a
+    global arguments parser, individual action parsers, and a combined parser containing both.
+    Return them as a tuple. The global parser is useful for parsing just global arguments while
+    ignoring actions, and the combined parser is handy for displaying help that includes everything:
+    global flags, a list of actions, etc.
     '''
     config_paths = collect.get_default_config_paths(expand_home=True)
     unexpanded_config_paths = collect.get_default_config_paths(expand_home=False)
 
-    global_parser = ArgumentParser(add_help=False)
+    # Using allow_abbrev=False here prevents the global parser from erroring about "ambiguous"
+    # options like --encryption. Such options are intended for an action parser rather than the
+    # global parser, and so we don't want to error on them here.
+    global_parser = ArgumentParser(allow_abbrev=False, add_help=False)
     global_group = global_parser.add_argument_group('global arguments')
 
     global_group.add_argument(
@@ -309,65 +576,18 @@ def make_parsers():
         action='store_true',
         help='Go through the motions, but do not actually write to any repositories',
     )
-    global_group.add_argument(
-        '-nc', '--no-color', dest='no_color', action='store_true', help='Disable colored output'
-    )
-    global_group.add_argument(
-        '-v',
-        '--verbosity',
-        type=int,
-        choices=range(-2, 3),
-        default=0,
-        help='Display verbose progress to the console: -2 (disabled), -1 (errors only), 0 (responses to actions, the default), 1 (info about steps borgmatic is taking), or 2 (debug)',
-    )
-    global_group.add_argument(
-        '--syslog-verbosity',
-        type=int,
-        choices=range(-2, 3),
-        default=-2,
-        help='Log verbose progress to syslog: -2 (disabled, the default), -1 (errors only), 0 (responses to actions), 1 (info about steps borgmatic is taking), or 2 (debug)',
-    )
-    global_group.add_argument(
-        '--log-file-verbosity',
-        type=int,
-        choices=range(-2, 3),
-        default=1,
-        help='When --log-file is given, log verbose progress to file: -2 (disabled), -1 (errors only), 0 (responses to actions), 1 (info about steps borgmatic is taking, the default), or 2 (debug)',
-    )
-    global_group.add_argument(
-        '--monitoring-verbosity',
-        type=int,
-        choices=range(-2, 3),
-        default=1,
-        help='When a monitoring integration supporting logging is configured, log verbose progress to it: -2 (disabled), -1 (errors only), responses to actions (0), 1 (info about steps borgmatic is taking, the default), or 2 (debug)',
-    )
-    global_group.add_argument(
-        '--log-file',
-        type=str,
-        help='Write log messages to this file instead of syslog',
-    )
-    global_group.add_argument(
-        '--log-file-format',
-        type=str,
-        help='Log format string used for log messages written to the log file',
-    )
-    global_group.add_argument(
-        '--log-json',
-        action='store_true',
-        help='Write log messages and console output as one JSON object per log line instead of formatted text',
-    )
     global_group.add_argument(
         '--override',
         metavar='OPTION.SUBOPTION=VALUE',
         dest='overrides',
         action='append',
-        help='Configuration file option to override with specified value, see documentation for overriding list or key/value options, can specify flag multiple times',
+        help='Deprecated. Configuration file option to override with specified value, see documentation for overriding list or key/value options, can specify flag multiple times',
     )
     global_group.add_argument(
         '--no-environment-interpolation',
         dest='resolve_env',
         action='store_false',
-        help='Do not resolve environment variables in configuration file',
+        help='Do not resolve environment variables in configuration files',
     )
     global_group.add_argument(
         '--bash-completion',
@@ -388,6 +608,7 @@ def make_parsers():
         action='store_true',
         help='Display installed version number of borgmatic and exit',
     )
+    add_arguments_from_schema(global_group, schema, unparsed_arguments)
 
     global_plus_action_parser = ArgumentParser(
         description='''
@@ -415,7 +636,6 @@ def make_parsers():
         '--encryption',
         dest='encryption_mode',
         help='Borg repository encryption mode',
-        required=True,
     )
     repo_create_group.add_argument(
         '--source-repository',
@@ -434,6 +654,7 @@ def make_parsers():
     )
     repo_create_group.add_argument(
         '--append-only',
+        default=None,
         action='store_true',
         help='Create an append-only repository',
     )
@@ -443,8 +664,10 @@ def make_parsers():
     )
     repo_create_group.add_argument(
         '--make-parent-dirs',
+        dest='make_parent_directories',
+        default=None,
         action='store_true',
-        help='Create any missing parent directories of the repository directory',
+        help='Create any missing parent directories of the repository directory [Borg 1.x only]',
     )
     repo_create_group.add_argument(
         '-h', '--help', action='help', help='Show this help message and exit'
@@ -477,7 +700,7 @@ def make_parsers():
     )
     transfer_group.add_argument(
         '--progress',
-        default=False,
+        default=None,
         action='store_true',
         help='Display progress as each archive is transferred',
     )
@@ -544,13 +767,17 @@ def make_parsers():
     )
     prune_group.add_argument(
         '--stats',
-        dest='stats',
-        default=False,
+        dest='statistics',
+        default=None,
         action='store_true',
-        help='Display statistics of the pruned archive',
+        help='Display statistics of the pruned archive [Borg 1 only]',
     )
     prune_group.add_argument(
-        '--list', dest='list_archives', action='store_true', help='List archives kept/pruned'
+        '--list',
+        dest='list_details',
+        default=None,
+        action='store_true',
+        help='List archives kept/pruned',
     )
     prune_group.add_argument(
         '--oldest',
@@ -588,8 +815,7 @@ def make_parsers():
     )
     compact_group.add_argument(
         '--progress',
-        dest='progress',
-        default=False,
+        default=None,
         action='store_true',
         help='Display progress as each segment is compacted',
     )
@@ -603,7 +829,7 @@ def make_parsers():
     compact_group.add_argument(
         '--threshold',
         type=int,
-        dest='threshold',
+        dest='compact_threshold',
         help='Minimum saved space percentage threshold for compacting a segment, defaults to 10',
     )
     compact_group.add_argument(
@@ -624,24 +850,33 @@ def make_parsers():
     )
     create_group.add_argument(
         '--progress',
-        dest='progress',
-        default=False,
+        default=None,
         action='store_true',
         help='Display progress for each file as it is backed up',
     )
     create_group.add_argument(
         '--stats',
-        dest='stats',
-        default=False,
+        dest='statistics',
+        default=None,
         action='store_true',
         help='Display statistics of archive',
     )
     create_group.add_argument(
-        '--list', '--files', dest='list_files', action='store_true', help='Show per-file details'
+        '--list',
+        '--files',
+        dest='list_details',
+        default=None,
+        action='store_true',
+        help='Show per-file details',
     )
     create_group.add_argument(
         '--json', dest='json', default=False, action='store_true', help='Output results as JSON'
     )
+    create_group.add_argument(
+        '--comment',
+        metavar='COMMENT',
+        help='Add a comment text to the archive',
+    )
     create_group.add_argument('-h', '--help', action='help', help='Show this help message and exit')
 
     check_parser = action_parsers.add_parser(
@@ -658,8 +893,7 @@ def make_parsers():
     )
     check_group.add_argument(
         '--progress',
-        dest='progress',
-        default=False,
+        default=None,
         action='store_true',
         help='Display progress for each file as it is checked',
     )
@@ -716,12 +950,15 @@ def make_parsers():
     )
     delete_group.add_argument(
         '--list',
-        dest='list_archives',
+        dest='list_details',
+        default=None,
         action='store_true',
         help='Show details for the deleted archives',
     )
     delete_group.add_argument(
         '--stats',
+        dest='statistics',
+        default=None,
         action='store_true',
         help='Display statistics for the deleted archives',
     )
@@ -826,8 +1063,7 @@ def make_parsers():
     )
     extract_group.add_argument(
         '--progress',
-        dest='progress',
-        default=False,
+        default=None,
         action='store_true',
         help='Display progress for each file as it is extracted',
     )
@@ -902,8 +1138,7 @@ def make_parsers():
     )
     config_bootstrap_group.add_argument(
         '--progress',
-        dest='progress',
-        default=False,
+        default=None,
         action='store_true',
         help='Display progress for each file as it is extracted',
     )
@@ -996,7 +1231,12 @@ def make_parsers():
         '--tar-filter', help='Name of filter program to pipe data through'
     )
     export_tar_group.add_argument(
-        '--list', '--files', dest='list_files', action='store_true', help='Show per-file details'
+        '--list',
+        '--files',
+        dest='list_details',
+        default=None,
+        action='store_true',
+        help='Show per-file details',
     )
     export_tar_group.add_argument(
         '--strip-components',
@@ -1107,7 +1347,8 @@ def make_parsers():
     )
     repo_delete_group.add_argument(
         '--list',
-        dest='list_archives',
+        dest='list_details',
+        default=None,
         action='store_true',
         help='Show details for the archives in the given repository',
     )
@@ -1479,6 +1720,31 @@ def make_parsers():
         '-h', '--help', action='help', help='Show this help message and exit'
     )
 
+    key_import_parser = key_parsers.add_parser(
+        'import',
+        help='Import a copy of the repository key from backup',
+        description='Import a copy of the repository key from backup',
+        add_help=False,
+    )
+    key_import_group = key_import_parser.add_argument_group('key import arguments')
+    key_import_group.add_argument(
+        '--paper',
+        action='store_true',
+        help='Import interactively from a backup done with --paper',
+    )
+    key_import_group.add_argument(
+        '--repository',
+        help='Path of repository to import the key from, defaults to the configured repository if there is only one, quoted globs supported',
+    )
+    key_import_group.add_argument(
+        '--path',
+        metavar='PATH',
+        help='Path to import the key from backup, defaults to stdin',
+    )
+    key_import_group.add_argument(
+        '-h', '--help', action='help', help='Show this help message and exit'
+    )
+
     key_change_passphrase_parser = key_parsers.add_parser(
         'change-passphrase',
         help='Change the passphrase protecting the repository key',
@@ -1496,6 +1762,56 @@ def make_parsers():
         '-h', '--help', action='help', help='Show this help message and exit'
     )
 
+    recreate_parser = action_parsers.add_parser(
+        'recreate',
+        aliases=ACTION_ALIASES['recreate'],
+        help='Recreate an archive in a repository (with Borg 1.2+, you must run compact afterwards to actually free space)',
+        description='Recreate an archive in a repository (with Borg 1.2+, you must run compact afterwards to actually free space)',
+        add_help=False,
+    )
+    recreate_group = recreate_parser.add_argument_group('recreate arguments')
+    recreate_group.add_argument(
+        '--repository',
+        help='Path of repository containing archive to recreate, defaults to the configured repository if there is only one, quoted globs supported',
+    )
+    recreate_group.add_argument(
+        '--archive',
+        help='Archive name, hash, or series to recreate',
+    )
+    recreate_group.add_argument(
+        '--list',
+        dest='list_details',
+        default=None,
+        action='store_true',
+        help='Show per-file details',
+    )
+    recreate_group.add_argument(
+        '--target',
+        metavar='TARGET',
+        help='Create a new archive from the specified archive (via --archive), without replacing it',
+    )
+    recreate_group.add_argument(
+        '--comment',
+        metavar='COMMENT',
+        help='Add a comment text to the archive or, if an archive is not provided, to all matching archives',
+    )
+    recreate_group.add_argument(
+        '--timestamp',
+        metavar='TIMESTAMP',
+        help='Manually override the archive creation date/time (UTC)',
+    )
+    recreate_group.add_argument(
+        '-a',
+        '--match-archives',
+        '--glob-archives',
+        dest='match_archives',
+        metavar='PATTERN',
+        help='Only consider archive names, hashes, or series matching this pattern [Borg 2.x+ only]',
+    )
+    recreate_group.add_argument(
+        '-h', '--help', action='help', help='Show this help message and exit'
+    )
+
     borg_parser = action_parsers.add_parser(
         'borg',
         aliases=ACTION_ALIASES['borg'],
@@ -1523,15 +1839,18 @@ def make_parsers():
     return global_parser, action_parsers, global_plus_action_parser
 
 
-def parse_arguments(*unparsed_arguments):
+def parse_arguments(schema, *unparsed_arguments):
     '''
-    Given command-line arguments with which this script was invoked, parse the arguments and return
-    them as a dict mapping from action name (or "global") to an argparse.Namespace instance.
+    Given a configuration schema dict and the command-line arguments with which this script was
+    invoked and unparsed arguments as a sequence of strings, parse the arguments and return them as
+    a dict mapping from action name (or "global") to an argparse.Namespace instance.
 
     Raise ValueError if the arguments cannot be parsed.
     Raise SystemExit with an error code of 0 if "--help" was requested.
     '''
-    global_parser, action_parsers, global_plus_action_parser = make_parsers()
+    global_parser, action_parsers, global_plus_action_parser = make_parsers(
+        schema, unparsed_arguments
+    )
     arguments, remaining_action_arguments = parse_arguments_for_actions(
         unparsed_arguments, action_parsers.choices, global_parser
     )
@@ -1559,15 +1878,6 @@ def parse_arguments(*unparsed_arguments):
             f"Unrecognized argument{'s' if len(unknown_arguments) > 1 else ''}: {' '.join(unknown_arguments)}"
         )
 
-    if 'create' in arguments and arguments['create'].list_files and arguments['create'].progress:
-        raise ValueError(
-            'With the create action, only one of --list (--files) and --progress flags can be used.'
-        )
-    if 'create' in arguments and arguments['create'].list_files and arguments['create'].json:
-        raise ValueError(
-            'With the create action, only one of --list (--files) and --json flags can be used.'
-        )
-
     if (
         ('list' in arguments and 'repo-info' in arguments and arguments['list'].json)
         or ('list' in arguments and 'info' in arguments and arguments['list'].json)
@@ -1575,15 +1885,6 @@ def parse_arguments(*unparsed_arguments):
     ):
         raise ValueError('With the --json flag, multiple actions cannot be used together.')
 
-    if (
-        'transfer' in arguments
-        and arguments['transfer'].archive
-        and arguments['transfer'].match_archives
-    ):
-        raise ValueError(
-            'With the transfer action, only one of --archive and --match-archives flags can be used.'
-        )
-
     if 'list' in arguments and (arguments['list'].prefix and arguments['list'].match_archives):
         raise ValueError(
             'With the list action, only one of --prefix or --match-archives flags can be used.'

borgmatic/commands/completion/bash.py

@@ -1,5 +1,7 @@
 import borgmatic.commands.arguments
 import borgmatic.commands.completion.actions
+import borgmatic.commands.completion.flag
+import borgmatic.config.validate
 
 
 def parser_flags(parser):
@@ -7,7 +9,12 @@ def parser_flags(parser):
     Given an argparse.ArgumentParser instance, return its argument flags in a space-separated
     string.
     '''
-    return ' '.join(option for action in parser._actions for option in action.option_strings)
+    return ' '.join(
+        flag_variant
+        for action in parser._actions
+        for flag_name in action.option_strings
+        for flag_variant in borgmatic.commands.completion.flag.variants(flag_name)
+    )
 
 
 def bash_completion():
@@ -19,7 +26,10 @@ def bash_completion():
         unused_global_parser,
         action_parsers,
         global_plus_action_parser,
-    ) = borgmatic.commands.arguments.make_parsers()
+    ) = borgmatic.commands.arguments.make_parsers(
+        schema=borgmatic.config.validate.load_schema(borgmatic.config.validate.schema_filename()),
+        unparsed_arguments=(),
+    )
     global_flags = parser_flags(global_plus_action_parser)
 
     # Avert your eyes.

borgmatic/commands/completion/fish.py

@@ -4,6 +4,7 @@ from textwrap import dedent
 
 import borgmatic.commands.arguments
 import borgmatic.commands.completion.actions
+import borgmatic.config.validate
 
 
 def has_file_options(action: Action):
@@ -26,9 +27,11 @@ def has_choice_options(action: Action):
 def has_unknown_required_param_options(action: Action):
     '''
     A catch-all for options that take a required parameter, but we don't know what the parameter is.
-    This should be used last. These are actions that take something like a glob, a list of numbers, or a string.
+    This should be used last. These are actions that take something like a glob, a list of numbers,
+    or a string.
 
-    Actions that match this pattern should not show the normal arguments, because those are unlikely to be valid.
+    Actions that match this pattern should not show the normal arguments, because those are unlikely
+    to be valid.
     '''
     return (
         action.required is True
@@ -52,9 +55,9 @@ def has_exact_options(action: Action):
 
 def exact_options_completion(action: Action):
     '''
-    Given an argparse.Action instance, return a completion invocation that forces file completions, options completion,
-    or just that some value follow the action, if the action takes such an argument and was the last action on the
-    command line prior to the cursor.
+    Given an argparse.Action instance, return a completion invocation that forces file completions,
+    options completion, or just that some value follow the action, if the action takes such an
+    argument and was the last action on the command line prior to the cursor.
 
     Otherwise, return an empty string.
     '''
@@ -80,8 +83,9 @@ def exact_options_completion(action: Action):
 
 def dedent_strip_as_tuple(string: str):
     '''
-    Dedent a string, then strip it to avoid requiring your first line to have content, then return a tuple of the string.
-    Makes it easier to write multiline strings for completions when you join them with a tuple.
+    Dedent a string, then strip it to avoid requiring your first line to have content, then return a
+    tuple of the string. Makes it easier to write multiline strings for completions when you join
+    them with a tuple.
     '''
     return (dedent(string).strip('\n'),)
 
@@ -95,7 +99,10 @@ def fish_completion():
         unused_global_parser,
         action_parsers,
         global_plus_action_parser,
-    ) = borgmatic.commands.arguments.make_parsers()
+    ) = borgmatic.commands.arguments.make_parsers(
+        schema=borgmatic.config.validate.load_schema(borgmatic.config.validate.schema_filename()),
+        unparsed_arguments=(),
+    )
 
     all_action_parsers = ' '.join(action for action in action_parsers.choices.keys())
 

borgmatic/commands/completion/flag.py

@@ -0,0 +1,13 @@
+def variants(flag_name):
+    '''
+    Given a flag name as a string, yield it and any variations that should be complete-able as well.
+    For instance, for a string like "--foo[0].bar", yield "--foo[0].bar", "--foo[1].bar", ...,
+    "--foo[9].bar".
+    '''
+    if '[0]' in flag_name:
+        for index in range(0, 10):
+            yield flag_name.replace('[0]', f'[{index}]')
+
+        return
+
+    yield flag_name

borgmatic/config/arguments.py

@@ -0,0 +1,176 @@
+import io
+import re
+
+import ruamel.yaml
+
+import borgmatic.config.schema
+
+LIST_INDEX_KEY_PATTERN = re.compile(r'^(?P<list_name>[a-zA-z-]+)\[(?P<index>\d+)\]$')
+
+
+def set_values(config, keys, value):
+    '''
+    Given a configuration dict, a sequence of parsed key strings, and a string value, descend into
+    the configuration hierarchy based on the given keys and set the value into the right place.
+    For example, consider these keys:
+
+        ('foo', 'bar', 'baz')
+
+    This looks up "foo" in the given configuration dict. And within that, it looks up "bar". And
+    then within that, it looks up "baz" and sets it to the given value. Another example:
+
+        ('mylist[0]', 'foo')
+
+    This looks for the zeroth element of "mylist" in the given configuration. And within that, it
+    looks up "foo" and sets it to the given value.
+    '''
+    if not keys:
+        return
+
+    first_key = keys[0]
+
+    # Support "mylist[0]" list index syntax.
+    match = LIST_INDEX_KEY_PATTERN.match(first_key)
+
+    if match:
+        list_key = match.group('list_name')
+        list_index = int(match.group('index'))
+
+        try:
+            if len(keys) == 1:
+                config[list_key][list_index] = value
+
+                return
+
+            if list_key not in config:
+                config[list_key] = []
+
+            set_values(config[list_key][list_index], keys[1:], value)
+        except (IndexError, KeyError):
+            raise ValueError(f'Argument list index {first_key} is out of range')
+
+        return
+
+    if len(keys) == 1:
+        config[first_key] = value
+
+        return
+
+    if first_key not in config:
+        config[first_key] = {}
+
+    set_values(config[first_key], keys[1:], value)
+
+
+def type_for_option(schema, option_keys):
+    '''
+    Given a configuration schema dict and a sequence of keys identifying a potentially nested
+    option, e.g. ('extra_borg_options', 'create'), return the schema type of that option as a
+    string.
+
+    Return None if the option or its type cannot be found in the schema.
+    '''
+    option_schema = schema
+
+    for key in option_keys:
+        # Support "name[0]"-style list index syntax.
+        match = LIST_INDEX_KEY_PATTERN.match(key)
+        properties = borgmatic.config.schema.get_properties(option_schema)
+
+        try:
+            if match:
+                option_schema = properties[match.group('list_name')]['items']
+            else:
+                option_schema = properties[key]
+        except KeyError:
+            return None
+
+    try:
+        return option_schema['type']
+    except KeyError:
+        return None
+
+
+def convert_value_type(value, option_type):
+    '''
+    Given a string value and its schema type as a string, determine its logical type (string,
+    boolean, integer, etc.), and return it converted to that type.
+
+    If the destination option type is a string, then leave the value as-is so that special
+    characters in it don't get interpreted as YAML during conversion.
+
+    And if the source value isn't a string, return it as-is.
+
+    Raise ruamel.yaml.error.YAMLError if there's a parse issue with the YAML.
+    Raise ValueError if the parsed value doesn't match the option type.
+    '''
+    if not isinstance(value, str):
+        return value
+
+    if option_type == 'string':
+        return value
+
+    try:
+        parsed_value = ruamel.yaml.YAML(typ='safe').load(io.StringIO(value))
+    except ruamel.yaml.error.YAMLError as error:
+        raise ValueError(f'Argument value "{value}" is invalid: {error.problem}')
+
+    if not isinstance(parsed_value, borgmatic.config.schema.parse_type(option_type)):
+        raise ValueError(f'Argument value "{value}" is not of the expected type: {option_type}')
+
+    return parsed_value
+
+
+def prepare_arguments_for_config(global_arguments, schema):
+    '''
+    Given global arguments as an argparse.Namespace and a configuration schema dict, parse each
+    argument that corresponds to an option in the schema and return a sequence of tuples (keys,
+    values) for that option, where keys is a sequence of strings. For instance, given the following
+    arguments:
+
+        argparse.Namespace(**{'my_option.sub_option': 'value1', 'other_option': 'value2'})
+
+    ... return this:
+
+        (
+            (('my_option', 'sub_option'), 'value1'),
+            (('other_option',), 'value2'),
+        )
+    '''
+    prepared_values = []
+
+    for argument_name, value in global_arguments.__dict__.items():
+        if value is None:
+            continue
+
+        keys = tuple(argument_name.split('.'))
+        option_type = type_for_option(schema, keys)
+
+        # The argument doesn't correspond to any option in the schema, so ignore it. It's
+        # probably a flag that borgmatic has on the command-line but not in configuration.
+        if option_type is None:
+            continue
+
+        prepared_values.append(
+            (
+                keys,
+                convert_value_type(value, option_type),
+            )
+        )
+
+    return tuple(prepared_values)
+
+
+def apply_arguments_to_config(config, schema, arguments):
+    '''
+    Given a configuration dict, a corresponding configuration schema dict, and arguments as a dict
+    from action name to argparse.Namespace, set those given argument values into their corresponding
+    configuration options in the configuration dict.
+
+    This supports argument flags of the from "--foo.bar.baz" where each dotted component is a nested
+    configuration object. Additionally, flags like "--foo.bar[0].baz" are supported to update a list
+    element in the configuration.
+    '''
+    for action_arguments in arguments.values():
+        for keys, value in prepare_arguments_for_config(action_arguments, schema):
+            set_values(config, keys, value)

borgmatic/config/generate.py

@@ -5,6 +5,7 @@ import re
 
 import ruamel.yaml
 
+import borgmatic.config.schema
 from borgmatic.config import load, normalize
 
 INDENT = 4
@@ -21,45 +22,59 @@ def insert_newline_before_comment(config, field_name):
     )
 
 
-def get_properties(schema):
-    '''
-    Given a schema dict, return its properties. But if it's got sub-schemas with multiple different
-    potential properties, returned their merged properties instead.
-    '''
-    if 'oneOf' in schema:
-        return dict(
-            collections.ChainMap(*[sub_schema['properties'] for sub_schema in schema['oneOf']])
-        )
-
-    return schema['properties']
+SCALAR_SCHEMA_TYPES = {'string', 'boolean', 'integer', 'number'}
 
 
-def schema_to_sample_configuration(schema, level=0, parent_is_sequence=False):
+def schema_to_sample_configuration(schema, source_config=None, level=0, parent_is_sequence=False):
     '''
-    Given a loaded configuration schema, generate and return sample config for it. Include comments
-    for each option based on the schema "description".
+    Given a loaded configuration schema and a source configuration, generate and return sample
+    config for the schema. Include comments for each option based on the schema "description".
+
+    If a source config is given, walk it alongside the given schema so that both can be taken into
+    account when commenting out particular options in add_comments_to_configuration_object().
     '''
     schema_type = schema.get('type')
     example = schema.get('example')
-    if example is not None:
-        return example
 
-    if schema_type == 'array' or (isinstance(schema_type, list) and 'array' in schema_type):
+    if borgmatic.config.schema.compare_types(schema_type, {'array'}):
         config = ruamel.yaml.comments.CommentedSeq(
-            [schema_to_sample_configuration(schema['items'], level, parent_is_sequence=True)]
+            example
+            if borgmatic.config.schema.compare_types(
+                schema['items'].get('type'), SCALAR_SCHEMA_TYPES
+            )
+            else [
+                schema_to_sample_configuration(
+                    schema['items'], source_config, level, parent_is_sequence=True
+                )
+            ]
         )
         add_comments_to_configuration_sequence(config, schema, indent=(level * INDENT))
-    elif schema_type == 'object' or (isinstance(schema_type, list) and 'object' in schema_type):
-        config = ruamel.yaml.comments.CommentedMap(
-            [
-                (field_name, schema_to_sample_configuration(sub_schema, level + 1))
-                for field_name, sub_schema in get_properties(schema).items()
-            ]
+    elif borgmatic.config.schema.compare_types(schema_type, {'object'}):
+        if source_config and isinstance(source_config, list) and isinstance(source_config[0], dict):
+            source_config = source_config[0]
+
+        config = (
+            ruamel.yaml.comments.CommentedMap(
+                [
+                    (
+                        field_name,
+                        schema_to_sample_configuration(
+                            sub_schema, (source_config or {}).get(field_name, {}), level + 1
+                        ),
+                    )
+                    for field_name, sub_schema in borgmatic.config.schema.get_properties(
+                        schema
+                    ).items()
+                ]
+            )
+            or example
         )
         indent = (level * INDENT) + (SEQUENCE_INDENT if parent_is_sequence else 0)
         add_comments_to_configuration_object(
-            config, schema, indent=indent, skip_first=parent_is_sequence
+            config, schema, source_config, indent=indent, skip_first_field=parent_is_sequence
         )
+    elif borgmatic.config.schema.compare_types(schema_type, SCALAR_SCHEMA_TYPES, match=all):
+        return example
     else:
         raise ValueError(f'Schema at level {level} is unsupported: {schema}')
 
@@ -93,17 +108,30 @@ def comment_out_optional_configuration(rendered_config):
     '''
     lines = []
     optional = False
+    indent_characters = None
+    indent_characters_at_sentinel = None
 
     for line in rendered_config.split('\n'):
+        indent_characters = len(line) - len(line.lstrip())
+
         # Upon encountering an optional configuration option, comment out lines until the next blank
         # line.
         if line.strip().startswith(f'# {COMMENTED_OUT_SENTINEL}'):
             optional = True
+            indent_characters_at_sentinel = indent_characters
             continue
 
         # Hit a blank line, so reset commenting.
         if not line.strip():
             optional = False
+            indent_characters_at_sentinel = None
+        # Dedented, so reset commenting.
+        elif (
+            indent_characters_at_sentinel is not None
+            and indent_characters < indent_characters_at_sentinel
+        ):
+            optional = False
+            indent_characters_at_sentinel = None
 
         lines.append(comment_out_line(line) if optional else line)
 
@@ -164,7 +192,7 @@ def add_comments_to_configuration_sequence(config, schema, indent=0):
         return
 
     for field_name in config[0].keys():
-        field_schema = get_properties(schema['items']).get(field_name, {})
+        field_schema = borgmatic.config.schema.get_properties(schema['items']).get(field_name, {})
         description = field_schema.get('description')
 
         # No description to use? Skip it.
@@ -178,26 +206,38 @@ def add_comments_to_configuration_sequence(config, schema, indent=0):
         return
 
 
-REQUIRED_KEYS = {'source_directories', 'repositories', 'keep_daily'}
+DEFAULT_KEYS = {'source_directories', 'repositories', 'keep_daily'}
 COMMENTED_OUT_SENTINEL = 'COMMENT_OUT'
 
 
-def add_comments_to_configuration_object(config, schema, indent=0, skip_first=False):
+def add_comments_to_configuration_object(
+    config, schema, source_config=None, indent=0, skip_first_field=False
+):
     '''
     Using descriptions from a schema as a source, add those descriptions as comments to the given
-    config mapping, before each field. Indent the comment the given number of characters.
+    configuration dict, putting them before each field. Indent the comment the given number of
+    characters.
+
+    If skip_first_field is True, omit the comment for the initial field. This is useful for
+    sequences, where the comment for the first field goes before the sequence itself.
+
+    And a sentinel for commenting out options that are neither in DEFAULT_KEYS nor the the given
+    source configuration dict. The idea is that any options used in the source configuration should
+    stay active in the generated configuration.
     '''
     for index, field_name in enumerate(config.keys()):
-        if skip_first and index == 0:
+        if skip_first_field and index == 0:
             continue
 
-        field_schema = get_properties(schema).get(field_name, {})
+        field_schema = borgmatic.config.schema.get_properties(schema).get(field_name, {})
         description = field_schema.get('description', '').strip()
 
-        # If this is an optional key, add an indicator to the comment flagging it to be commented
+        # If this isn't a default key, add an indicator to the comment, flagging it to be commented
         # out from the sample configuration. This sentinel is consumed by downstream processing that
         # does the actual commenting out.
-        if field_name not in REQUIRED_KEYS:
+        if field_name not in DEFAULT_KEYS and (
+            source_config is None or field_name not in source_config
+        ):
             description = (
                 '\n'.join((description, COMMENTED_OUT_SENTINEL))
                 if description
@@ -217,21 +257,6 @@ def add_comments_to_configuration_object(config, schema, indent=0, skip_first=Fa
 RUAMEL_YAML_COMMENTS_INDEX = 1
 
 
-def remove_commented_out_sentinel(config, field_name):
-    '''
-    Given a configuration CommentedMap and a top-level field name in it, remove any "commented out"
-    sentinel found at the end of its YAML comments. This prevents the given field name from getting
-    commented out by downstream processing that consumes the sentinel.
-    '''
-    try:
-        last_comment_value = config.ca.items[field_name][RUAMEL_YAML_COMMENTS_INDEX][-1].value
-    except KeyError:
-        return
-
-    if last_comment_value == f'# {COMMENTED_OUT_SENTINEL}\n':
-        config.ca.items[field_name][RUAMEL_YAML_COMMENTS_INDEX].pop()
-
-
 def merge_source_configuration_into_destination(destination_config, source_config):
     '''
     Deep merge the given source configuration dict into the destination configuration CommentedMap,
@@ -246,12 +271,6 @@ def merge_source_configuration_into_destination(destination_config, source_confi
         return source_config
 
     for field_name, source_value in source_config.items():
-        # Since this key/value is from the source configuration, leave it uncommented and remove any
-        # sentinel that would cause it to get commented out.
-        remove_commented_out_sentinel(
-            ruamel.yaml.comments.CommentedMap(destination_config), field_name
-        )
-
         # This is a mapping. Recurse for this key/value.
         if isinstance(source_value, collections.abc.Mapping):
             destination_config[field_name] = merge_source_configuration_into_destination(
@@ -296,8 +315,14 @@ def generate_sample_configuration(
         source_config = load.load_configuration(source_filename)
         normalize.normalize(source_filename, source_config)
 
+        # The borgmatic.config.normalize.normalize() function tacks on an empty "bootstrap" if
+        # needed, so the hook gets used by default. But we don't want it to end up in the generated
+        # config unless the user has set it explicitly, as an empty "bootstrap:" won't validate.
+        if source_config and source_config.get('bootstrap') == {}:
+            del source_config['bootstrap']
+
     destination_config = merge_source_configuration_into_destination(
-        schema_to_sample_configuration(schema), source_config
+        schema_to_sample_configuration(schema, source_config), source_config
     )
 
     if dry_run:

borgmatic/config/load.py

@@ -69,7 +69,7 @@ def include_configuration(loader, filename_node, include_directory, config_paths
         ]
 
     raise ValueError(
-        '!include value is not supported; use a single filename or a list of filenames'
+        'The value given for the !include tag is invalid; use a single filename or a list of filenames instead'
     )
 
 

borgmatic/config/normalize.py

@@ -58,6 +58,90 @@ def normalize_sections(config_filename, config):
     return []
 
 
+def make_command_hook_deprecation_log(config_filename, option_name):  # pragma: no cover
+    '''
+    Given a configuration filename and the name of a configuration option, return a deprecation
+    warning log for it.
+    '''
+    return logging.makeLogRecord(
+        dict(
+            levelno=logging.WARNING,
+            levelname='WARNING',
+            msg=f'{config_filename}: {option_name} is deprecated and support will be removed from a future release. Use commands: instead.',
+        )
+    )
+
+
+def normalize_commands(config_filename, config):
+    '''
+    Given a configuration filename and a configuration dict, transform any "before_*"- and
+    "after_*"-style command hooks into "commands:".
+    '''
+    logs = []
+
+    # Normalize "before_actions" and "after_actions".
+    for preposition in ('before', 'after'):
+        option_name = f'{preposition}_actions'
+        commands = config.pop(option_name, None)
+
+        if commands:
+            logs.append(make_command_hook_deprecation_log(config_filename, option_name))
+            config.setdefault('commands', []).append(
+                {
+                    preposition: 'repository',
+                    'run': commands,
+                }
+            )
+
+    # Normalize "before_backup", "before_prune", "after_backup", "after_prune", etc.
+    for action_name in ('create', 'prune', 'compact', 'check', 'extract'):
+        for preposition in ('before', 'after'):
+            option_name = f'{preposition}_{"backup" if action_name == "create" else action_name}'
+            commands = config.pop(option_name, None)
+
+            if not commands:
+                continue
+
+            logs.append(make_command_hook_deprecation_log(config_filename, option_name))
+            config.setdefault('commands', []).append(
+                {
+                    preposition: 'action',
+                    'when': [action_name],
+                    'run': commands,
+                }
+            )
+
+    # Normalize "on_error".
+    commands = config.pop('on_error', None)
+
+    if commands:
+        logs.append(make_command_hook_deprecation_log(config_filename, 'on_error'))
+        config.setdefault('commands', []).append(
+            {
+                'after': 'error',
+                'when': ['create', 'prune', 'compact', 'check'],
+                'run': commands,
+            }
+        )
+
+    # Normalize "before_everything" and "after_everything".
+    for preposition in ('before', 'after'):
+        option_name = f'{preposition}_everything'
+        commands = config.pop(option_name, None)
+
+        if commands:
+            logs.append(make_command_hook_deprecation_log(config_filename, option_name))
+            config.setdefault('commands', []).append(
+                {
+                    preposition: 'everything',
+                    'when': ['create'],
+                    'run': commands,
+                }
+            )
+
+    return logs
+
+
 def normalize(config_filename, config):
     '''
     Given a configuration filename and a configuration dict of its loaded contents, apply particular
@@ -67,6 +151,7 @@ def normalize(config_filename, config):
     Raise ValueError the configuration cannot be normalized.
     '''
     logs = normalize_sections(config_filename, config)
+    logs += normalize_commands(config_filename, config)
 
     if config.get('borgmatic_source_directory'):
         logs.append(
@@ -74,7 +159,7 @@ def normalize(config_filename, config):
                 dict(
                     levelno=logging.WARNING,
                     levelname='WARNING',
-                    msg=f'{config_filename}: The borgmatic_source_directory option is deprecated and will be removed from a future release. Use borgmatic_runtime_directory and borgmatic_state_directory instead.',
+                    msg=f'{config_filename}: The borgmatic_source_directory option is deprecated and will be removed from a future release. Use user_runtime_directory and user_state_directory instead.',
                 )
             )
         )
@@ -241,7 +326,11 @@ def normalize(config_filename, config):
         config['repositories'] = []
 
         for repository_dict in repositories:
-            repository_path = repository_dict['path']
+            repository_path = repository_dict.get('path')
+
+            if repository_path is None:
+                continue
+
             if '~' in repository_path:
                 logs.append(
                     logging.makeLogRecord(
@@ -267,6 +356,8 @@ def normalize(config_filename, config):
                     repository_path.startswith('ssh://')
                     or repository_path.startswith('sftp://')
                     or repository_path.startswith('rclone:')
+                    or repository_path.startswith('s3:')
+                    or repository_path.startswith('b2:')
                 ):
                     config['repositories'].append(repository_dict)
                 else:
@@ -276,7 +367,7 @@ def normalize(config_filename, config):
                             dict(
                                 levelno=logging.WARNING,
                                 levelname='WARNING',
-                                msg=f'{config_filename}: Remote repository paths without ssh:// or rclone: syntax are deprecated and support will be removed from a future release. Interpreting "{repository_path}" as "{rewritten_repository_path}"',
+                                msg=f'{config_filename}: Remote repository paths without ssh://, sftp://, rclone:, s3:, or b2:, syntax are deprecated and support will be removed from a future release. Interpreting "{repository_path}" as "{rewritten_repository_path}"',
                             )
                         )
                     )

borgmatic/config/override.py

@@ -1,7 +1,10 @@
 import io
+import logging
 
 import ruamel.yaml
 
+logger = logging.getLogger(__name__)
+
 
 def set_values(config, keys, value):
     '''
@@ -134,6 +137,11 @@ def apply_overrides(config, schema, raw_overrides):
     '''
     overrides = parse_overrides(raw_overrides, schema)
 
+    if overrides:
+        logger.warning(
+            "The --override flag is deprecated and will be removed from a future release. Instead, use a command-line flag corresponding to the configuration option you'd like to set."
+        )
+
     for keys, value in overrides:
         set_values(config, keys, value)
         set_values(config, strip_section_names(keys), value)

borgmatic/config/paths.py

@@ -103,7 +103,9 @@ class Runtime_directory:
 
             self.temporary_directory = None
         else:
-            base_directory = os.environ.get('TMPDIR') or os.environ.get('TEMP') or '/tmp'
+            base_directory = (
+                os.environ.get('TMPDIR') or os.environ.get('TEMP') or '/tmp'  # noqa: S108
+            )
 
             if not base_directory.startswith(os.path.sep):
                 raise ValueError('The temporary directory must be an absolute path')
@@ -134,7 +136,7 @@ class Runtime_directory:
         '''
         return self.runtime_path
 
-    def __exit__(self, exception, value, traceback):
+    def __exit__(self, exception_type, exception, traceback):
         '''
         Delete any temporary directory that was created as part of initialization.
         '''

borgmatic/config/schema.py

@@ -0,0 +1,72 @@
+import decimal
+import itertools
+
+
+def get_properties(schema):
+    '''
+    Given a schema dict, return its properties. But if it's got sub-schemas with multiple different
+    potential properties, return their merged properties instead (interleaved so the first
+    properties of each sub-schema come first). The idea is that the user should see all possible
+    options even if they're not all possible together.
+    '''
+    if 'oneOf' in schema:
+        return dict(
+            item
+            for item in itertools.chain(
+                *itertools.zip_longest(
+                    *[sub_schema['properties'].items() for sub_schema in schema['oneOf']]
+                )
+            )
+            if item is not None
+        )
+
+    return schema.get('properties', {})
+
+
+SCHEMA_TYPE_TO_PYTHON_TYPE = {
+    'array': list,
+    'boolean': bool,
+    'integer': int,
+    'number': decimal.Decimal,
+    'object': dict,
+    'string': str,
+}
+
+
+def parse_type(schema_type, **overrides):
+    '''
+    Given a schema type as a string, return the corresponding Python type.
+
+    If any overrides are given in the from of a schema type string to a Python type, then override
+    the default type mapping with them.
+
+    Raise ValueError if the schema type is unknown.
+    '''
+    try:
+        return dict(
+            SCHEMA_TYPE_TO_PYTHON_TYPE,
+            **overrides,
+        )[schema_type]
+    except KeyError:
+        raise ValueError(f'Unknown type in configuration schema: {schema_type}')
+
+
+def compare_types(schema_type, target_types, match=any):
+    '''
+    Given a schema type as a string or a list of strings (representing multiple types) and a set of
+    target type strings, return whether every schema type is in the set of target types.
+
+    If the schema type is a list of strings, use the given match function (such as any or all) to
+    compare elements. For instance, if match is given as all, then every element of the schema_type
+    list must be in the target types.
+    '''
+    if isinstance(schema_type, list):
+        if match(element_schema_type in target_types for element_schema_type in schema_type):
+            return True
+
+        return False
+
+    if schema_type in target_types:
+        return True
+
+    return False

borgmatic/config/validate.py

@@ -4,7 +4,7 @@ import os
 import jsonschema
 import ruamel.yaml
 
-import borgmatic.config
+import borgmatic.config.arguments
 from borgmatic.config import constants, environment, load, normalize, override
 
 
@@ -21,6 +21,18 @@ def schema_filename():
         return schema_path
 
 
+def load_schema(schema_path):  # pragma: no cover
+    '''
+    Given a schema filename path, load the schema and return it as a dict.
+
+    Raise Validation_error if the schema could not be parsed.
+    '''
+    try:
+        return load.load_configuration(schema_path)
+    except (ruamel.yaml.error.YAMLError, RecursionError) as error:
+        raise Validation_error(schema_path, (str(error),))
+
+
 def format_json_error_path_element(path_element):
     '''
     Given a path element into a JSON data structure, format it for display as a string.
@@ -84,12 +96,17 @@ def apply_logical_validation(config_filename, parsed_configuration):
             )
 
 
-def parse_configuration(config_filename, schema_filename, overrides=None, resolve_env=True):
+def parse_configuration(
+    config_filename, schema_filename, arguments, overrides=None, resolve_env=True
+):
     '''
     Given the path to a config filename in YAML format, the path to a schema filename in a YAML
-    rendition of JSON Schema format, a sequence of configuration file override strings in the form
-    of "option.suboption=value", return the parsed configuration as a data structure of nested dicts
-    and lists corresponding to the schema. Example return value:
+    rendition of JSON Schema format, arguments as dict from action name to argparse.Namespace, a
+    sequence of configuration file override strings in the form of "option.suboption=value", and
+    whether to resolve environment variables, return the parsed configuration as a data structure of
+    nested dicts and lists corresponding to the schema. Example return value.
+
+    Example return value:
 
         {
             'source_directories': ['/home', '/etc'],
@@ -112,6 +129,7 @@ def parse_configuration(config_filename, schema_filename, overrides=None, resolv
     except (ruamel.yaml.error.YAMLError, RecursionError) as error:
         raise Validation_error(config_filename, (str(error),))
 
+    borgmatic.config.arguments.apply_arguments_to_config(config, schema, arguments)
     override.apply_overrides(config, schema, overrides)
     constants.apply_constants(config, config.get('constants') if config else {})
 
@@ -124,6 +142,7 @@ def parse_configuration(config_filename, schema_filename, overrides=None, resolv
         validator = jsonschema.Draft7Validator(schema)
     except AttributeError:  # pragma: no cover
         validator = jsonschema.Draft4Validator(schema)
+
     validation_errors = tuple(validator.iter_errors(config))
 
     if validation_errors:
@@ -136,16 +155,22 @@ def parse_configuration(config_filename, schema_filename, overrides=None, resolv
     return config, config_paths, logs
 
 
-def normalize_repository_path(repository):
+def normalize_repository_path(repository, base=None):
     '''
     Given a repository path, return the absolute path of it (for local repositories).
+    Optionally, use a base path for resolving relative paths, e.g. to the configured working directory.
     '''
     # A colon in the repository could mean that it's either a file:// URL or a remote repository.
     # If it's a remote repository, we don't want to normalize it. If it's a file:// URL, we do.
     if ':' not in repository:
-        return os.path.abspath(repository)
+        return (
+            os.path.abspath(os.path.join(base, repository)) if base else os.path.abspath(repository)
+        )
     elif repository.startswith('file://'):
-        return os.path.abspath(repository.partition('file://')[-1])
+        local_path = repository.partition('file://')[-1]
+        return (
+            os.path.abspath(os.path.join(base, local_path)) if base else os.path.abspath(local_path)
+        )
     else:
         return repository
 

borgmatic/execute.py

@@ -1,7 +1,6 @@
 import collections
 import enum
 import logging
-import os
 import select
 import subprocess
 import textwrap
@@ -243,6 +242,9 @@ def mask_command_secrets(full_command):
 MAX_LOGGED_COMMAND_LENGTH = 1000
 
 
+PREFIXES_OF_ENVIRONMENT_VARIABLES_TO_LOG = ('BORG_', 'PG', 'MARIADB_', 'MYSQL_')
+
+
 def log_command(full_command, input_file=None, output_file=None, environment=None):
     '''
     Log the given command (a sequence of command/argument strings), along with its input/output file
@@ -251,14 +253,21 @@ def log_command(full_command, input_file=None, output_file=None, environment=Non
     logger.debug(
         textwrap.shorten(
             ' '.join(
-                tuple(f'{key}=***' for key in (environment or {}).keys())
+                tuple(
+                    f'{key}=***'
+                    for key in (environment or {}).keys()
+                    if any(
+                        key.startswith(prefix)
+                        for prefix in PREFIXES_OF_ENVIRONMENT_VARIABLES_TO_LOG
+                    )
+                )
                 + mask_command_secrets(full_command)
             ),
             width=MAX_LOGGED_COMMAND_LENGTH,
             placeholder=' ...',
         )
-        + (f" < {getattr(input_file, 'name', '')}" if input_file else '')
-        + (f" > {getattr(output_file, 'name', '')}" if output_file else '')
+        + (f" < {getattr(input_file, 'name', input_file)}" if input_file else '')
+        + (f" > {getattr(output_file, 'name', output_file)}" if output_file else '')
     )
 
 
@@ -274,28 +283,28 @@ def execute_command(
     output_file=None,
     input_file=None,
     shell=False,
-    extra_environment=None,
+    environment=None,
     working_directory=None,
     borg_local_path=None,
     borg_exit_codes=None,
     run_to_completion=True,
+    close_fds=False,  # Necessary for passing credentials via anonymous pipe.
 ):
     '''
     Execute the given command (a sequence of command/argument strings) and log its output at the
     given log level. If an open output file object is given, then write stdout to the file and only
     log stderr. If an open input file object is given, then read stdin from the file. If shell is
-    True, execute the command within a shell. If an extra environment dict is given, then use it to
-    augment the current environment, and pass the result into the command. If a working directory is
-    given, use that as the present working directory when running the command. If a Borg local path
-    is given, and the command matches it (regardless of arguments), treat exit code 1 as a warning
-    instead of an error. But if Borg exit codes are given as a sequence of exit code configuration
-    dicts, then use that configuration to decide what's an error and what's a warning. If run to
-    completion is False, then return the process for the command without executing it to completion.
+    True, execute the command within a shell. If an environment variables dict is given, then pass
+    it into the command. If a working directory is given, use that as the present working directory
+    when running the command. If a Borg local path is given, and the command matches it (regardless
+    of arguments), treat exit code 1 as a warning instead of an error. But if Borg exit codes are
+    given as a sequence of exit code configuration dicts, then use that configuration to decide
+    what's an error and what's a warning. If run to completion is False, then return the process for
+    the command without executing it to completion.
 
     Raise subprocesses.CalledProcessError if an error occurs while running the command.
     '''
-    log_command(full_command, input_file, output_file, extra_environment)
-    environment = {**os.environ, **extra_environment} if extra_environment else None
+    log_command(full_command, input_file, output_file, environment)
     do_not_capture = bool(output_file is DO_NOT_CAPTURE)
     command = ' '.join(full_command) if shell else full_command
 
@@ -304,11 +313,10 @@ def execute_command(
         stdin=input_file,
         stdout=None if do_not_capture else (output_file or subprocess.PIPE),
         stderr=None if do_not_capture else (subprocess.PIPE if output_file else subprocess.STDOUT),
-        shell=shell,
+        shell=shell,  # noqa: S602
         env=environment,
         cwd=working_directory,
-        # Necessary for the passcommand credential hook to work.
-        close_fds=not bool((extra_environment or {}).get('BORG_PASSPHRASE_FD')),
+        close_fds=close_fds,
     )
     if not run_to_completion:
         return process
@@ -325,39 +333,40 @@ def execute_command(
 
 def execute_command_and_capture_output(
     full_command,
+    input_file=None,
     capture_stderr=False,
     shell=False,
-    extra_environment=None,
+    environment=None,
     working_directory=None,
     borg_local_path=None,
     borg_exit_codes=None,
+    close_fds=False,  # Necessary for passing credentials via anonymous pipe.
 ):
     '''
     Execute the given command (a sequence of command/argument strings), capturing and returning its
-    output (stdout). If capture stderr is True, then capture and return stderr in addition to
-    stdout. If shell is True, execute the command within a shell. If an extra environment dict is
-    given, then use it to augment the current environment, and pass the result into the command. If
-    a working directory is given, use that as the present working directory when running the
-    command. If a Borg local path is given, and the command matches it (regardless of arguments),
-    treat exit code 1 as a warning instead of an error. But if Borg exit codes are given as a
-    sequence of exit code configuration dicts, then use that configuration to decide what's an error
-    and what's a warning.
+    output (stdout). If an input file descriptor is given, then pipe it to the command's stdin. If
+    capture stderr is True, then capture and return stderr in addition to stdout. If shell is True,
+    execute the command within a shell. If an environment variables dict is given, then pass it into
+    the command. If a working directory is given, use that as the present working directory when
+    running the command. If a Borg local path is given, and the command matches it (regardless of
+    arguments), treat exit code 1 as a warning instead of an error. But if Borg exit codes are given
+    as a sequence of exit code configuration dicts, then use that configuration to decide what's an
+    error and what's a warning.
 
     Raise subprocesses.CalledProcessError if an error occurs while running the command.
     '''
-    log_command(full_command, environment=extra_environment)
-    environment = {**os.environ, **extra_environment} if extra_environment else None
+    log_command(full_command, input_file, environment=environment)
     command = ' '.join(full_command) if shell else full_command
 
     try:
         output = subprocess.check_output(
             command,
+            stdin=input_file,
             stderr=subprocess.STDOUT if capture_stderr else None,
-            shell=shell,
+            shell=shell,  # noqa: S602
             env=environment,
             cwd=working_directory,
-            # Necessary for the passcommand credential hook to work.
-            close_fds=not bool((extra_environment or {}).get('BORG_PASSPHRASE_FD')),
+            close_fds=close_fds,
         )
     except subprocess.CalledProcessError as error:
         if (
@@ -377,10 +386,11 @@ def execute_command_with_processes(
     output_file=None,
     input_file=None,
     shell=False,
-    extra_environment=None,
+    environment=None,
     working_directory=None,
     borg_local_path=None,
     borg_exit_codes=None,
+    close_fds=False,  # Necessary for passing credentials via anonymous pipe.
 ):
     '''
     Execute the given command (a sequence of command/argument strings) and log its output at the
@@ -391,19 +401,17 @@ def execute_command_with_processes(
     If an open output file object is given, then write stdout to the file and only log stderr. But
     if output log level is None, instead suppress logging and return the captured output for (only)
     the given command. If an open input file object is given, then read stdin from the file. If
-    shell is True, execute the command within a shell. If an extra environment dict is given, then
-    use it to augment the current environment, and pass the result into the command. If a working
-    directory is given, use that as the present working directory when running the command. If a
-    Borg local path is given, then for any matching command or process (regardless of arguments),
-    treat exit code 1 as a warning instead of an error. But if Borg exit codes are given as a
-    sequence of exit code configuration dicts, then use that configuration to decide what's an error
-    and what's a warning.
+    shell is True, execute the command within a shell. If an environment variables dict is given,
+    then pass it into the command. If a working directory is given, use that as the present working
+    directory when running the command. If a Borg local path is given, then for any matching command
+    or process (regardless of arguments), treat exit code 1 as a warning instead of an error. But if
+    Borg exit codes are given as a sequence of exit code configuration dicts, then use that
+    configuration to decide what's an error and what's a warning.
 
     Raise subprocesses.CalledProcessError if an error occurs while running the command or in the
     upstream process.
     '''
-    log_command(full_command, input_file, output_file, extra_environment)
-    environment = {**os.environ, **extra_environment} if extra_environment else None
+    log_command(full_command, input_file, output_file, environment)
     do_not_capture = bool(output_file is DO_NOT_CAPTURE)
     command = ' '.join(full_command) if shell else full_command
 
@@ -415,11 +423,10 @@ def execute_command_with_processes(
             stderr=(
                 None if do_not_capture else (subprocess.PIPE if output_file else subprocess.STDOUT)
             ),
-            shell=shell,
+            shell=shell,  # noqa: S602
             env=environment,
             cwd=working_directory,
-            # Necessary for the passcommand credential hook to work.
-            close_fds=not bool((extra_environment or {}).get('BORG_PASSPHRASE_FD')),
+            close_fds=close_fds,
         )
     except (subprocess.CalledProcessError, OSError):
         # Something has gone wrong. So vent each process' output buffer to prevent it from hanging.
@@ -430,13 +437,14 @@ def execute_command_with_processes(
                 process.kill()
         raise
 
-    captured_outputs = log_outputs(
-        tuple(processes) + (command_process,),
-        (input_file, output_file),
-        output_log_level,
-        borg_local_path,
-        borg_exit_codes,
-    )
+    with borgmatic.logger.Log_prefix(None):  # Log command output without any prefix.
+        captured_outputs = log_outputs(
+            tuple(processes) + (command_process,),
+            (input_file, output_file),
+            output_log_level,
+            borg_local_path,
+            borg_exit_codes,
+        )
 
     if output_log_level is None:
         return captured_outputs.get(command_process)

borgmatic/hooks/command.py

@@ -1,15 +1,32 @@
+import functools
 import logging
 import os
 import re
 import shlex
+import subprocess
 import sys
 
 import borgmatic.execute
+import borgmatic.logger
 
 logger = logging.getLogger(__name__)
 
 
 SOFT_FAIL_EXIT_CODE = 75
+BORG_PLACEHOLDERS = {
+    '{hostname}',
+    '{fqdn}',
+    '{reverse-fqdn}',
+    '{now}',
+    '{utcnow}',
+    '{unixtime}',
+    '{user}',
+    '{pid}',
+    '{borgversion}',
+    '{borgmajor}',
+    '{borgminor}',
+    '{borgpatch}',
+}
 
 
 def interpolate_context(hook_description, command, context):
@@ -20,83 +37,226 @@ def interpolate_context(hook_description, command, context):
     for name, value in context.items():
         command = command.replace(f'{{{name}}}', shlex.quote(str(value)))
 
-    for unsupported_variable in re.findall(r'{\w+}', command):
-        logger.warning(
-            f"Variable '{unsupported_variable}' is not supported in {hook_description} hook"
-        )
+    for unsupported_variable in re.findall(r'\{\w+\}', command):
+        # Warn about variables unknown to borgmatic, but don't warn if the variable name happens to
+        # be a Borg placeholder, as Borg should hopefully consume it.
+        if unsupported_variable not in BORG_PLACEHOLDERS:
+            logger.warning(
+                f'Variable "{unsupported_variable}" is not supported in the {hook_description} hook'
+            )
 
     return command
 
 
 def make_environment(current_environment, sys_module=sys):
     '''
-    Given the existing system environment as a map from environment variable name to value, return
-    (in the same form) any extra environment variables that should be used when running command
-    hooks.
+    Given the existing system environment as a map from environment variable name to value, return a
+    copy of it, augmented with any extra environment variables that should be used when running
+    command hooks.
     '''
+    environment = dict(current_environment)
+
     # Detect whether we're running within a PyInstaller bundle. If so, set or clear LD_LIBRARY_PATH
     # based on the value of LD_LIBRARY_PATH_ORIG. This prevents library version information errors.
     if getattr(sys_module, 'frozen', False) and hasattr(sys_module, '_MEIPASS'):
-        return {'LD_LIBRARY_PATH': current_environment.get('LD_LIBRARY_PATH_ORIG', '')}
+        environment['LD_LIBRARY_PATH'] = environment.get('LD_LIBRARY_PATH_ORIG', '')
 
-    return {}
+    return environment
 
 
-def execute_hook(commands, umask, config_filename, description, dry_run, **context):
+def filter_hooks(command_hooks, before=None, after=None, action_names=None, state_names=None):
     '''
-    Given a list of hook commands to execute, a umask to execute with (or None), a config filename,
-    a hook description, and whether this is a dry run, run the given commands. Or, don't run them
-    if this is a dry run.
+    Given a sequence of command hook dicts from configuration and one or more filters (before name,
+    after name, a sequence of action names, and/or a sequence of execution result state names),
+    filter down the command hooks to just the ones that match the given filters.
+    '''
+    return tuple(
+        hook_config
+        for hook_config in command_hooks or ()
+        for config_action_names in (hook_config.get('when'),)
+        for config_state_names in (hook_config.get('states'),)
+        if before is None or hook_config.get('before') == before
+        if after is None or hook_config.get('after') == after
+        if action_names is None
+        or config_action_names is None
+        or set(config_action_names or ()).intersection(set(action_names))
+        if state_names is None
+        or config_state_names is None
+        or set(config_state_names or ()).intersection(set(state_names))
+    )
+
+
+def execute_hooks(command_hooks, umask, working_directory, dry_run, **context):
+    '''
+    Given a sequence of command hook dicts from configuration, a umask to execute with (or None), a
+    working directory to execute with, and whether this is a dry run, run the commands for each
+    hook. Or don't run them if this is a dry run.
 
     The context contains optional values interpolated by name into the hook commands.
 
-    Raise ValueError if the umask cannot be parsed.
+    Raise ValueError if the umask cannot be parsed or a hook is invalid.
     Raise subprocesses.CalledProcessError if an error occurs in a hook.
     '''
-    if not commands:
-        logger.debug(f'No commands to run for {description} hook')
-        return
+    borgmatic.logger.add_custom_log_levels()
 
     dry_run_label = ' (dry run; not actually running hooks)' if dry_run else ''
 
-    context['configuration_filename'] = config_filename
-    commands = [interpolate_context(description, command, context) for command in commands]
-
-    if len(commands) == 1:
-        logger.info(f'Running command for {description} hook{dry_run_label}')
-    else:
-        logger.info(
-            f'Running {len(commands)} commands for {description} hook{dry_run_label}',
+    for hook_config in command_hooks:
+        commands = hook_config.get('run')
+        when_description = (
+            f"{'/'.join(hook_config.get('when'))} " if hook_config.get('when') else ''
         )
 
-    if umask:
-        parsed_umask = int(str(umask), 8)
-        logger.debug(f'Set hook umask to {oct(parsed_umask)}')
-        original_umask = os.umask(parsed_umask)
-    else:
-        original_umask = None
+        if 'before' in hook_config:
+            description = f'before {when_description}{hook_config.get("before")}'
+        elif 'after' in hook_config:
+            description = f'after {when_description}{hook_config.get("after")}'
+        else:
+            raise ValueError(f'Invalid hook configuration: {hook_config}')
 
-    try:
-        for command in commands:
-            if dry_run:
-                continue
+        if not commands:
+            logger.debug(f'No commands to run for {description} hook')
+            continue
 
-            borgmatic.execute.execute_command(
-                [command],
-                output_log_level=(logging.ERROR if description == 'on-error' else logging.WARNING),
-                shell=True,
-                extra_environment=make_environment(os.environ),
+        commands = [interpolate_context(description, command, context) for command in commands]
+
+        if len(commands) == 1:
+            logger.info(f'Running {description} command hook{dry_run_label}')
+        else:
+            logger.info(
+                f'Running {len(commands)} commands for {description} hook{dry_run_label}',
             )
-    finally:
-        if original_umask:
-            os.umask(original_umask)
+
+        if umask:
+            parsed_umask = int(str(umask), 8)
+            logger.debug(f'Setting hook umask to {oct(parsed_umask)}')
+            original_umask = os.umask(parsed_umask)
+        else:
+            original_umask = None
+
+        try:
+            for command in commands:
+                if dry_run:
+                    continue
+
+                borgmatic.execute.execute_command(
+                    [command],
+                    output_log_level=(
+                        logging.ERROR if hook_config.get('after') == 'error' else logging.ANSWER
+                    ),
+                    shell=True,  # noqa: S604
+                    environment=make_environment(os.environ),
+                    working_directory=working_directory,
+                )
+        finally:
+            if original_umask:
+                os.umask(original_umask)
 
 
+class Before_after_hooks:
+    '''
+    A Python context manager for executing command hooks both before and after the wrapped code.
+
+    Example use as a context manager:
+
+       with borgmatic.hooks.command.Before_after_hooks(
+           command_hooks=config.get('commands'),
+           before_after='do_stuff',
+           umask=config.get('umask'),
+           dry_run=dry_run,
+           action_names=['create'],
+       ):
+            do()
+            some()
+            stuff()
+
+    With that context manager in place, "before" command hooks execute before the wrapped code runs,
+    and "after" command hooks execute after the wrapped code completes.
+    '''
+
+    def __init__(
+        self,
+        command_hooks,
+        before_after,
+        umask,
+        working_directory,
+        dry_run,
+        action_names=None,
+        **context,
+    ):
+        '''
+        Given a sequence of command hook configuration dicts, the before/after name, a umask to run
+        commands with, a working directory to run commands with, a dry run flag, a sequence of
+        action names, and any context for the executed commands, save those data points for use
+        below.
+        '''
+        self.command_hooks = command_hooks
+        self.before_after = before_after
+        self.umask = umask
+        self.working_directory = working_directory
+        self.dry_run = dry_run
+        self.action_names = action_names
+        self.context = context
+
+    def __enter__(self):
+        '''
+        Run the configured "before" command hooks that match the initialized data points.
+        '''
+        try:
+            execute_hooks(
+                borgmatic.hooks.command.filter_hooks(
+                    self.command_hooks,
+                    before=self.before_after,
+                    action_names=self.action_names,
+                ),
+                self.umask,
+                self.working_directory,
+                self.dry_run,
+                **self.context,
+            )
+        except (OSError, subprocess.CalledProcessError) as error:
+            if considered_soft_failure(error):
+                raise
+
+            # Trigger the after hook manually, since raising here will prevent it from being run
+            # otherwise.
+            self.__exit__(exception_type=type(error), exception=error, traceback=None)
+
+            raise ValueError(f'Error running before {self.before_after} hook: {error}')
+
+    def __exit__(self, exception_type, exception, traceback):
+        '''
+        Run the configured "after" command hooks that match the initialized data points.
+        '''
+        try:
+            execute_hooks(
+                borgmatic.hooks.command.filter_hooks(
+                    self.command_hooks,
+                    after=self.before_after,
+                    action_names=self.action_names,
+                    state_names=['fail' if exception_type else 'finish'],
+                ),
+                self.umask,
+                self.working_directory,
+                self.dry_run,
+                **self.context,
+            )
+        except (OSError, subprocess.CalledProcessError) as error:
+            if considered_soft_failure(error):
+                raise
+
+            raise ValueError(f'Error running after {self.before_after} hook: {error}')
+
+
+@functools.cache
 def considered_soft_failure(error):
     '''
     Given a configuration filename and an exception object, return whether the exception object
     represents a subprocess.CalledProcessError with a return code of SOFT_FAIL_EXIT_CODE. If so,
     that indicates that the error is a "soft failure", and should not result in an error.
+
+    The results of this function are cached so that it can be called multiple times without logging
+    multiple times.
     '''
     exit_code = getattr(error, 'returncode', None)
     if exit_code is None:

borgmatic/hooks/credential/container.py

@@ -0,0 +1,43 @@
+import logging
+import os
+import re
+
+logger = logging.getLogger(__name__)
+
+
+SECRET_NAME_PATTERN = re.compile(r'^\w+$')
+DEFAULT_SECRETS_DIRECTORY = '/run/secrets'
+
+
+def load_credential(hook_config, config, credential_parameters):
+    '''
+    Given the hook configuration dict, the configuration dict, and a credential parameters tuple
+    containing a secret name to load, read the secret from the corresponding container secrets file
+    and return it.
+
+    Raise ValueError if the credential parameters is not one element, the secret name is invalid, or
+    the secret file cannot be read.
+    '''
+    try:
+        (secret_name,) = credential_parameters
+    except ValueError:
+        name = ' '.join(credential_parameters)
+
+        raise ValueError(f'Cannot load invalid secret name: "{name}"')
+
+    if not SECRET_NAME_PATTERN.match(secret_name):
+        raise ValueError(f'Cannot load invalid secret name: "{secret_name}"')
+
+    try:
+        with open(
+            os.path.join(
+                config.get('working_directory', ''),
+                (hook_config or {}).get('secrets_directory', DEFAULT_SECRETS_DIRECTORY),
+                secret_name,
+            )
+        ) as secret_file:
+            return secret_file.read().rstrip(os.linesep)
+    except (FileNotFoundError, OSError) as error:
+        logger.warning(error)
+
+        raise ValueError(f'Cannot load secret "{secret_name}" from file: {error.filename}')

borgmatic/hooks/credential/file.py

@@ -0,0 +1,32 @@
+import logging
+import os
+
+logger = logging.getLogger(__name__)
+
+
+def load_credential(hook_config, config, credential_parameters):
+    '''
+    Given the hook configuration dict, the configuration dict, and a credential parameters tuple
+    containing a credential path to load, load the credential from file and return it.
+
+    Raise ValueError if the credential parameters is not one element or the secret file cannot be
+    read.
+    '''
+    try:
+        (credential_path,) = credential_parameters
+    except ValueError:
+        name = ' '.join(credential_parameters)
+
+        raise ValueError(f'Cannot load invalid credential: "{name}"')
+
+    expanded_credential_path = os.path.expanduser(credential_path)
+
+    try:
+        with open(
+            os.path.join(config.get('working_directory', ''), expanded_credential_path)
+        ) as credential_file:
+            return credential_file.read().rstrip(os.linesep)
+    except (FileNotFoundError, OSError) as error:
+        logger.warning(error)
+
+        raise ValueError(f'Cannot load credential file: {error.filename}')

borgmatic/hooks/credential/keepassxc.py

@@ -0,0 +1,45 @@
+import logging
+import os
+import shlex
+
+import borgmatic.execute
+
+logger = logging.getLogger(__name__)
+
+
+def load_credential(hook_config, config, credential_parameters):
+    '''
+    Given the hook configuration dict, the configuration dict, and a credential parameters tuple
+    containing a KeePassXC database path and an attribute name to load, run keepassxc-cli to fetch
+    the corresponding KeePassXC credential and return it.
+
+    Raise ValueError if keepassxc-cli can't retrieve the credential.
+    '''
+    try:
+        (database_path, attribute_name) = credential_parameters
+    except ValueError:
+        raise ValueError(f'Invalid KeePassXC credential parameters: {credential_parameters}')
+
+    expanded_database_path = os.path.expanduser(database_path)
+
+    if not os.path.exists(expanded_database_path):
+        raise ValueError(f'KeePassXC database path does not exist: {database_path}')
+
+    # Build the keepassxc-cli command.
+    command = (
+        tuple(shlex.split((hook_config or {}).get('keepassxc_cli_command', 'keepassxc-cli')))
+        + ('show', '--show-protected', '--attributes', 'Password')
+        + (
+            ('--key-file', hook_config['key_file'])
+            if hook_config and hook_config.get('key_file')
+            else ()
+        )
+        + (
+            ('--yubikey', hook_config['yubikey'])
+            if hook_config and hook_config.get('yubikey')
+            else ()
+        )
+        + (expanded_database_path, attribute_name)  # Ensure database and entry are last.
+    )
+
+    return borgmatic.execute.execute_command_and_capture_output(command).rstrip(os.linesep)

borgmatic/hooks/credential/parse.py

@@ -0,0 +1,124 @@
+import functools
+import re
+import shlex
+
+import borgmatic.hooks.dispatch
+
+IS_A_HOOK = False
+
+
+class Hash_adapter:
+    '''
+    A Hash_adapter instance wraps an unhashable object and pretends it's hashable. This is intended
+    for passing to a @functools.cache-decorated function to prevent it from complaining that an
+    argument is unhashable. It should only be used for arguments that you don't want to actually
+    impact the cache hashing, because Hash_adapter doesn't actually hash the object's contents.
+
+    Example usage:
+
+        @functools.cache
+        def func(a, b):
+            print(a, b.actual_value)
+            return a
+
+        func(5, Hash_adapter({1: 2, 3: 4}))  # Calls func(), prints, and returns.
+        func(5, Hash_adapter({1: 2, 3: 4}))  # Hits the cache and just returns the value.
+        func(5, Hash_adapter({5: 6, 7: 8}))  # Also uses cache, since the Hash_adapter is ignored.
+
+    In the above function, the "b" value is one that has been wrapped with Hash_adappter, and
+    therefore "b.actual_value" is necessary to access the original value.
+    '''
+
+    def __init__(self, actual_value):
+        self.actual_value = actual_value
+
+    def __eq__(self, other):
+        return True
+
+    def __hash__(self):
+        return 0
+
+
+UNHASHABLE_TYPES = (dict, list, set)
+
+
+def cache_ignoring_unhashable_arguments(function):
+    '''
+    A function decorator that caches calls to the decorated function but ignores any unhashable
+    arguments when performing cache lookups. This is intended to be a drop-in replacement for
+    functools.cache.
+
+    Example usage:
+
+        @cache_ignoring_unhashable_arguments
+        def func(a, b):
+            print(a, b)
+            return a
+
+        func(5, {1: 2, 3: 4})  # Calls func(), prints, and returns.
+        func(5, {1: 2, 3: 4})  # Hits the cache and just returns the value.
+        func(5, {5: 6, 7: 8})  # Also uses cache, since the unhashable value (the dict) is ignored.
+    '''
+
+    @functools.cache
+    def cached_function(*args, **kwargs):
+        return function(
+            *(arg.actual_value if isinstance(arg, Hash_adapter) else arg for arg in args),
+            **{
+                key: value.actual_value if isinstance(value, Hash_adapter) else value
+                for (key, value) in kwargs.items()
+            },
+        )
+
+    @functools.wraps(function)
+    def wrapper_function(*args, **kwargs):
+        return cached_function(
+            *(Hash_adapter(arg) if isinstance(arg, UNHASHABLE_TYPES) else arg for arg in args),
+            **{
+                key: Hash_adapter(value) if isinstance(value, UNHASHABLE_TYPES) else value
+                for (key, value) in kwargs.items()
+            },
+        )
+
+    wrapper_function.cache_clear = cached_function.cache_clear
+
+    return wrapper_function
+
+
+CREDENTIAL_PATTERN = re.compile(r'\{credential( +(?P<hook_and_parameters>.*))?\}')
+
+
+@cache_ignoring_unhashable_arguments
+def resolve_credential(value, config):
+    '''
+    Given a configuration value containing a string like "{credential hookname credentialname}" and
+    a configuration dict, resolve the credential by calling the relevant hook to get the actual
+    credential value. If the given value does not actually contain a credential tag, then return it
+    unchanged.
+
+    Cache the value (ignoring the config for purposes of caching), so repeated calls to this
+    function don't need to load the credential repeatedly.
+
+    Raise ValueError if the config could not be parsed or the credential could not be loaded.
+    '''
+    if value is None:
+        return value
+
+    matcher = CREDENTIAL_PATTERN.match(value)
+
+    if not matcher:
+        return value
+
+    hook_and_parameters = matcher.group('hook_and_parameters')
+
+    if not hook_and_parameters:
+        raise ValueError(f'Cannot load credential with invalid syntax "{value}"')
+
+    (hook_name, *credential_parameters) = shlex.split(hook_and_parameters)
+
+    if not credential_parameters:
+        raise ValueError(f'Cannot load credential with invalid syntax "{value}"')
+
+    return borgmatic.hooks.dispatch.call_hook(
+        'load_credential', config, hook_name, tuple(credential_parameters)
+    )

borgmatic/hooks/credential/systemd.py

@@ -0,0 +1,43 @@
+import logging
+import os
+import re
+
+logger = logging.getLogger(__name__)
+
+
+CREDENTIAL_NAME_PATTERN = re.compile(r'^[\w.-]+$')
+
+
+def load_credential(hook_config, config, credential_parameters):
+    '''
+    Given the hook configuration dict, the configuration dict, and a credential parameters tuple
+    containing a credential name to load, read the credential from the corresponding systemd
+    credential file and return it.
+
+    Raise ValueError if the systemd CREDENTIALS_DIRECTORY environment variable is not set, the
+    credential name is invalid, or the credential file cannot be read.
+    '''
+    try:
+        (credential_name,) = credential_parameters
+    except ValueError:
+        name = ' '.join(credential_parameters)
+
+        raise ValueError(f'Cannot load invalid credential name: "{name}"')
+
+    credentials_directory = os.environ.get('CREDENTIALS_DIRECTORY')
+
+    if not credentials_directory:
+        raise ValueError(
+            f'Cannot load credential "{credential_name}" because the systemd CREDENTIALS_DIRECTORY environment variable is not set'
+        )
+
+    if not CREDENTIAL_NAME_PATTERN.match(credential_name):
+        raise ValueError(f'Cannot load invalid credential name "{credential_name}"')
+
+    try:
+        with open(os.path.join(credentials_directory, credential_name)) as credential_file:
+            return credential_file.read().rstrip(os.linesep)
+    except (FileNotFoundError, OSError) as error:
+        logger.warning(error)
+
+        raise ValueError(f'Cannot load credential "{credential_name}" from file: {error.filename}')

borgmatic/hooks/data_source/bootstrap.py

@@ -55,9 +55,17 @@ def dump_data_sources(
             manifest_file,
         )
 
-    patterns.extend(borgmatic.borg.pattern.Pattern(config_path) for config_path in config_paths)
+    patterns.extend(
+        borgmatic.borg.pattern.Pattern(
+            config_path, source=borgmatic.borg.pattern.Pattern_source.HOOK
+        )
+        for config_path in config_paths
+    )
     patterns.append(
-        borgmatic.borg.pattern.Pattern(os.path.join(borgmatic_runtime_directory, 'bootstrap'))
+        borgmatic.borg.pattern.Pattern(
+            os.path.join(borgmatic_runtime_directory, 'bootstrap'),
+            source=borgmatic.borg.pattern.Pattern_source.HOOK,
+        )
     )
 
     return []

borgmatic/hooks/data_source/btrfs.py

@@ -1,5 +1,6 @@
 import collections
 import glob
+import itertools
 import json
 import logging
 import os
@@ -21,9 +22,44 @@ def use_streaming(hook_config, config):  # pragma: no cover
     return False
 
 
-def get_subvolume_mount_points(findmnt_command):
+def get_contained_subvolume_paths(btrfs_command, subvolume_path):
     '''
-    Given a findmnt command to run, get all sorted Btrfs subvolume mount points.
+    Given the path of a Btrfs subvolume, return it in a sequence along with the paths of its
+    contained subvolumes.
+
+    If the btrfs command errors, log that error and return an empty sequence.
+    '''
+    try:
+        btrfs_output = borgmatic.execute.execute_command_and_capture_output(
+            tuple(btrfs_command.split(' '))
+            + (
+                'subvolume',
+                'list',
+                subvolume_path,
+            ),
+            close_fds=True,
+        )
+    except subprocess.CalledProcessError as error:
+        logger.debug(
+            f'Ignoring Btrfs subvolume {subvolume_path} because of error listing its subvolumes: {error}'
+        )
+
+        return ()
+
+    return (subvolume_path,) + tuple(
+        os.path.join(subvolume_path, line.split(' ')[-1])
+        for line in btrfs_output.splitlines()
+        if line.strip()
+    )
+
+
+FINDMNT_BTRFS_ROOT_SUBVOLUME_OPTION = 'subvolid=5'
+
+
+def get_all_subvolume_paths(btrfs_command, findmnt_command):
+    '''
+    Given btrfs and findmnt commands to run, get the sorted paths for all Btrfs subvolumes on the
+    system.
     '''
     findmnt_output = borgmatic.execute.execute_command_and_capture_output(
         tuple(findmnt_command.split(' '))
@@ -32,12 +68,28 @@ def get_subvolume_mount_points(findmnt_command):
             'btrfs',
             '--json',
             '--list',  # Request a flat list instead of a nested subvolume hierarchy.
-        )
+        ),
+        close_fds=True,
     )
 
     try:
         return tuple(
-            sorted(filesystem['target'] for filesystem in json.loads(findmnt_output)['filesystems'])
+            sorted(
+                itertools.chain.from_iterable(
+                    # If findmnt gave us a Btrfs root filesystem, list the subvolumes within it.
+                    # This is necessary because findmnt only returns a subvolume's mount point
+                    # rather than its original subvolume path (which can differ). For instance,
+                    # a subvolume might exist at /mnt/subvolume but be mounted at /home/myuser.
+                    # findmnt is still useful though because it's a global way to discover all
+                    # Btrfs subvolumes—even if we have to do some additional legwork ourselves.
+                    (
+                        get_contained_subvolume_paths(btrfs_command, filesystem['target'])
+                        if FINDMNT_BTRFS_ROOT_SUBVOLUME_OPTION in filesystem['options'].split(',')
+                        else (filesystem['target'],)
+                    )
+                    for filesystem in json.loads(findmnt_output)['filesystems']
+                )
+            )
         )
     except json.JSONDecodeError as error:
         raise ValueError(f'Invalid {findmnt_command} JSON output: {error}')
@@ -48,32 +100,96 @@ def get_subvolume_mount_points(findmnt_command):
 Subvolume = collections.namedtuple('Subvolume', ('path', 'contained_patterns'), defaults=((),))
 
 
+def get_subvolume_property(btrfs_command, subvolume_path, property_name):
+    '''
+    Given a btrfs command, a subvolume path, and a property name to lookup, return the value of the
+    corresponding property.
+
+    Raise subprocess.CalledProcessError if the btrfs command errors.
+    '''
+    output = borgmatic.execute.execute_command_and_capture_output(
+        tuple(btrfs_command.split(' '))
+        + (
+            'property',
+            'get',
+            '-t',  # Type.
+            'subvol',
+            subvolume_path,
+            property_name,
+        ),
+        close_fds=True,
+    )
+
+    try:
+        value = output.strip().split('=')[1]
+    except IndexError:
+        raise ValueError(f'Invalid {btrfs_command} property output')
+
+    return {
+        'true': True,
+        'false': False,
+    }.get(value, value)
+
+
+def omit_read_only_subvolume_paths(btrfs_command, subvolume_paths):
+    '''
+    Given a Btrfs command to run and a sequence of Btrfs subvolume paths, filter them down to just
+    those that are read-write. The idea is that Btrfs can't actually snapshot a read-only subvolume,
+    so we should just ignore them.
+    '''
+    retained_subvolume_paths = []
+
+    for subvolume_path in subvolume_paths:
+        try:
+            if get_subvolume_property(btrfs_command, subvolume_path, 'ro'):
+                logger.debug(f'Ignoring Btrfs subvolume {subvolume_path} because it is read-only')
+            else:
+                retained_subvolume_paths.append(subvolume_path)
+        except subprocess.CalledProcessError as error:
+            logger.debug(
+                f'Error determining read-only status of Btrfs subvolume {subvolume_path}: {error}'
+            )
+
+    return tuple(retained_subvolume_paths)
+
+
 def get_subvolumes(btrfs_command, findmnt_command, patterns=None):
     '''
     Given a Btrfs command to run and a sequence of configured patterns, find the intersection
-    between the current Btrfs filesystem and subvolume mount points and the paths of any patterns.
-    The idea is that these pattern paths represent the requested subvolumes to snapshot.
+    between the current Btrfs filesystem and subvolume paths and the paths of any patterns.  The
+    idea is that these pattern paths represent the requested subvolumes to snapshot.
 
-    If patterns is None, then return all subvolumes, sorted by path.
+    Only include subvolumes that contain at least one root pattern sourced from borgmatic
+    configuration (as opposed to generated elsewhere in borgmatic). But if patterns is None, then
+    return all subvolumes instead, sorted by path.
 
-    Return the result as a sequence of matching subvolume mount points.
+    Return the result as a sequence of matching Subvolume instances.
     '''
     candidate_patterns = set(patterns or ())
     subvolumes = []
 
-    # For each subvolume mount point, match it against the given patterns to find the subvolumes to
+    # For each subvolume path, match it against the given patterns to find the subvolumes to
     # backup. Sort the subvolumes from longest to shortest mount points, so longer mount points get
     # a whack at the candidate pattern piñata before their parents do. (Patterns are consumed during
     # this process, so no two subvolumes end up with the same contained patterns.)
-    for mount_point in reversed(get_subvolume_mount_points(findmnt_command)):
+    for subvolume_path in reversed(
+        omit_read_only_subvolume_paths(
+            btrfs_command, get_all_subvolume_paths(btrfs_command, findmnt_command)
+        )
+    ):
         subvolumes.extend(
-            Subvolume(mount_point, contained_patterns)
+            Subvolume(subvolume_path, contained_patterns)
             for contained_patterns in (
                 borgmatic.hooks.data_source.snapshot.get_contained_patterns(
-                    mount_point, candidate_patterns
+                    subvolume_path, candidate_patterns
                 ),
             )
-            if patterns is None or contained_patterns
+            if patterns is None
+            or any(
+                pattern.type == borgmatic.borg.pattern.Pattern_type.ROOT
+                and pattern.source == borgmatic.borg.pattern.Pattern_source.CONFIG
+                for pattern in contained_patterns
+            )
         )
 
     return tuple(sorted(subvolumes, key=lambda subvolume: subvolume.path))
@@ -121,6 +237,7 @@ def make_snapshot_exclude_pattern(subvolume_path):  # pragma: no cover
         ),
         borgmatic.borg.pattern.Pattern_type.NO_RECURSE,
         borgmatic.borg.pattern.Pattern_style.FNMATCH,
+        source=borgmatic.borg.pattern.Pattern_source.HOOK,
     )
 
 
@@ -153,6 +270,7 @@ def make_borg_snapshot_pattern(subvolume_path, pattern):
         pattern.type,
         pattern.style,
         pattern.device,
+        source=borgmatic.borg.pattern.Pattern_source.HOOK,
     )
 
 
@@ -173,6 +291,7 @@ def snapshot_subvolume(btrfs_command, subvolume_path, snapshot_path):  # pragma:
             snapshot_path,
         ),
         output_log_level=logging.DEBUG,
+        close_fds=True,
     )
 
 
@@ -187,9 +306,9 @@ def dump_data_sources(
     '''
     Given a Btrfs configuration dict, a configuration dict, the borgmatic configuration file paths,
     the borgmatic runtime directory, the configured patterns, and whether this is a dry run,
-    auto-detect and snapshot any Btrfs subvolume mount points listed in the given patterns. Also
-    update those patterns, replacing subvolume mount points with corresponding snapshot directories
-    so they get stored in the Borg archive instead.
+    auto-detect and snapshot any Btrfs subvolume paths listed in the given patterns. Also update
+    those patterns, replacing subvolume paths with corresponding snapshot directories so they get
+    stored in the Borg archive instead.
 
     Return an empty sequence, since there are no ongoing dump processes from this hook.
 
@@ -198,7 +317,8 @@ def dump_data_sources(
     dry_run_label = ' (dry run; not actually snapshotting anything)' if dry_run else ''
     logger.info(f'Snapshotting Btrfs subvolumes{dry_run_label}')
 
-    # Based on the configured patterns, determine Btrfs subvolumes to backup.
+    # Based on the configured patterns, determine Btrfs subvolumes to backup. Only consider those
+    # patterns that came from actual user configuration (as opposed to, say, other hooks).
     btrfs_command = hook_config.get('btrfs_command', 'btrfs')
     findmnt_command = hook_config.get('findmnt_command', 'findmnt')
     subvolumes = get_subvolumes(btrfs_command, findmnt_command, patterns)
@@ -243,6 +363,7 @@ def delete_snapshot(btrfs_command, snapshot_path):  # pragma: no cover
             snapshot_path,
         ),
         output_log_level=logging.DEBUG,
+        close_fds=True,
     )
 
 
@@ -269,8 +390,8 @@ def remove_data_source_dumps(hook_config, config, borgmatic_runtime_directory, d
         logger.debug(error)
         return
 
-    # Reversing the sorted subvolumes ensures that we remove longer mount point paths of child
-    # subvolumes before the shorter mount point paths of parent subvolumes.
+    # Reversing the sorted subvolumes ensures that we remove longer paths of child subvolumes before
+    # the shorter paths of parent subvolumes.
     for subvolume in reversed(all_subvolumes):
         subvolume_snapshots_glob = borgmatic.config.paths.replace_temporary_subdirectory_with_glob(
             os.path.normpath(make_snapshot_path(subvolume.path)),
@@ -299,9 +420,12 @@ def remove_data_source_dumps(hook_config, config, borgmatic_runtime_directory, d
                 logger.debug(error)
                 return
 
-            # Strip off the subvolume path from the end of the snapshot path and then delete the
-            # resulting directory.
-            shutil.rmtree(snapshot_path.rsplit(subvolume.path, 1)[0])
+            # Remove the snapshot parent directory if it still exists. (It might not exist if the
+            # snapshot was for "/".)
+            snapshot_parent_dir = snapshot_path.rsplit(subvolume.path, 1)[0]
+
+            if os.path.isdir(snapshot_parent_dir):
+                shutil.rmtree(snapshot_parent_dir)
 
 
 def make_data_source_dump_patterns(

borgmatic/hooks/data_source/lvm.py

@@ -1,5 +1,6 @@
 import collections
 import glob
+import hashlib
 import json
 import logging
 import os
@@ -33,7 +34,9 @@ def get_logical_volumes(lsblk_command, patterns=None):
     between the current LVM logical volume mount points and the paths of any patterns. The idea is
     that these pattern paths represent the requested logical volumes to snapshot.
 
-    If patterns is None, include all logical volume mounts points, not just those in patterns.
+    Only include logical volumes that contain at least one root pattern sourced from borgmatic
+    configuration (as opposed to generated elsewhere in borgmatic). But if patterns is None, include
+    all logical volume mounts points instead, not just those in patterns.
 
     Return the result as a sequence of Logical_volume instances.
     '''
@@ -47,7 +50,8 @@ def get_logical_volumes(lsblk_command, patterns=None):
                     'name,path,mountpoint,type',
                     '--json',
                     '--list',
-                )
+                ),
+                close_fds=True,
             )
         )
     except json.JSONDecodeError as error:
@@ -72,7 +76,12 @@ def get_logical_volumes(lsblk_command, patterns=None):
                     device['mountpoint'], candidate_patterns
                 ),
             )
-            if not patterns or contained_patterns
+            if not patterns
+            or any(
+                pattern.type == borgmatic.borg.pattern.Pattern_type.ROOT
+                and pattern.source == borgmatic.borg.pattern.Pattern_source.CONFIG
+                for pattern in contained_patterns
+            )
         )
     except KeyError as error:
         raise ValueError(f'Invalid {lsblk_command} output: Missing key "{error}"')
@@ -101,6 +110,7 @@ def snapshot_logical_volume(
             logical_volume_device,
         ),
         output_log_level=logging.DEBUG,
+        close_fds=True,
     )
 
 
@@ -121,13 +131,18 @@ def mount_snapshot(mount_command, snapshot_device, snapshot_mount_path):  # prag
             snapshot_mount_path,
         ),
         output_log_level=logging.DEBUG,
+        close_fds=True,
     )
 
 
-def make_borg_snapshot_pattern(pattern, normalized_runtime_directory):
+MOUNT_POINT_HASH_LENGTH = 10
+
+
+def make_borg_snapshot_pattern(pattern, logical_volume, normalized_runtime_directory):
     '''
-    Given a Borg pattern as a borgmatic.borg.pattern.Pattern instance, return a new Pattern with its
-    path rewritten to be in a snapshot directory based on the given runtime directory.
+    Given a Borg pattern as a borgmatic.borg.pattern.Pattern instance and a Logical_volume
+    containing it, return a new Pattern with its path rewritten to be in a snapshot directory based
+    on both the given runtime directory and the given Logical_volume's mount point.
 
     Move any initial caret in a regular expression pattern path to the beginning, so as not to break
     the regular expression.
@@ -142,6 +157,13 @@ def make_borg_snapshot_pattern(pattern, normalized_runtime_directory):
     rewritten_path = initial_caret + os.path.join(
         normalized_runtime_directory,
         'lvm_snapshots',
+        # Including this hash prevents conflicts between snapshot patterns for different logical
+        # volumes. For instance, without this, snapshotting a logical volume at /var and another at
+        # /var/spool would result in overlapping snapshot patterns and therefore colliding mount
+        # attempts.
+        hashlib.shake_256(logical_volume.mount_point.encode('utf-8')).hexdigest(
+            MOUNT_POINT_HASH_LENGTH
+        ),
         '.',  # Borg 1.4+ "slashdot" hack.
         # Included so that the source directory ends up in the Borg archive at its "original" path.
         pattern.path.lstrip('^').lstrip(os.path.sep),
@@ -152,6 +174,7 @@ def make_borg_snapshot_pattern(pattern, normalized_runtime_directory):
         pattern.type,
         pattern.style,
         pattern.device,
+        source=borgmatic.borg.pattern.Pattern_source.HOOK,
     )
 
 
@@ -180,7 +203,8 @@ def dump_data_sources(
     dry_run_label = ' (dry run; not actually snapshotting anything)' if dry_run else ''
     logger.info(f'Snapshotting LVM logical volumes{dry_run_label}')
 
-    # List logical volumes to get their mount points.
+    # List logical volumes to get their mount points, but only consider those patterns that came
+    # from actual user configuration (as opposed to, say, other hooks).
     lsblk_command = hook_config.get('lsblk_command', 'lsblk')
     requested_logical_volumes = get_logical_volumes(lsblk_command, patterns)
 
@@ -206,18 +230,22 @@ def dump_data_sources(
             )
 
         # Get the device path for the snapshot we just created.
-        try:
-            snapshot = get_snapshots(
-                hook_config.get('lvs_command', 'lvs'), snapshot_name=snapshot_name
-            )[0]
-        except IndexError:
-            raise ValueError(f'Cannot find LVM snapshot {snapshot_name}')
+        if not dry_run:
+            try:
+                snapshot = get_snapshots(
+                    hook_config.get('lvs_command', 'lvs'), snapshot_name=snapshot_name
+                )[0]
+            except IndexError:
+                raise ValueError(f'Cannot find LVM snapshot {snapshot_name}')
 
         # Mount the snapshot into a particular named temporary directory so that the snapshot ends
         # up in the Borg archive at the "original" logical volume mount point path.
         snapshot_mount_path = os.path.join(
             normalized_runtime_directory,
             'lvm_snapshots',
+            hashlib.shake_256(logical_volume.mount_point.encode('utf-8')).hexdigest(
+                MOUNT_POINT_HASH_LENGTH
+            ),
             logical_volume.mount_point.lstrip(os.path.sep),
         )
 
@@ -233,7 +261,9 @@ def dump_data_sources(
         )
 
         for pattern in logical_volume.contained_patterns:
-            snapshot_pattern = make_borg_snapshot_pattern(pattern, normalized_runtime_directory)
+            snapshot_pattern = make_borg_snapshot_pattern(
+                pattern, logical_volume, normalized_runtime_directory
+            )
 
             # Attempt to update the pattern in place, since pattern order matters to Borg.
             try:
@@ -251,6 +281,7 @@ def unmount_snapshot(umount_command, snapshot_mount_path):  # pragma: no cover
     borgmatic.execute.execute_command(
         tuple(umount_command.split(' ')) + (snapshot_mount_path,),
         output_log_level=logging.DEBUG,
+        close_fds=True,
     )
 
 
@@ -265,6 +296,7 @@ def remove_snapshot(lvremove_command, snapshot_device_path):  # pragma: no cover
             snapshot_device_path,
         ),
         output_log_level=logging.DEBUG,
+        close_fds=True,
     )
 
 
@@ -292,7 +324,8 @@ def get_snapshots(lvs_command, snapshot_name=None):
                     'lv_name,lv_path',
                     '--select',
                     'lv_attr =~ ^s',  # Filter to just snapshots.
-                )
+                ),
+                close_fds=True,
             )
         )
     except json.JSONDecodeError as error:
@@ -337,6 +370,7 @@ def remove_data_source_dumps(hook_config, config, borgmatic_runtime_directory, d
             os.path.normpath(borgmatic_runtime_directory),
         ),
         'lvm_snapshots',
+        '*',
     )
     logger.debug(f'Looking for snapshots to remove in {snapshots_glob}{dry_run_label}')
     umount_command = hook_config.get('umount_command', 'umount')
@@ -349,7 +383,10 @@ def remove_data_source_dumps(hook_config, config, borgmatic_runtime_directory, d
             snapshot_mount_path = os.path.join(
                 snapshots_directory, logical_volume.mount_point.lstrip(os.path.sep)
             )
-            if not os.path.isdir(snapshot_mount_path):
+
+            # If the snapshot mount path is empty, this is probably just a "shadow" of a nested
+            # logical volume and therefore there's nothing to unmount.
+            if not os.path.isdir(snapshot_mount_path) or not os.listdir(snapshot_mount_path):
                 continue
 
             # This might fail if the directory is already mounted, but we swallow errors here since
@@ -374,10 +411,10 @@ def remove_data_source_dumps(hook_config, config, borgmatic_runtime_directory, d
                 return
             except subprocess.CalledProcessError as error:
                 logger.debug(error)
-                return
+                continue
 
         if not dry_run:
-            shutil.rmtree(snapshots_directory)
+            shutil.rmtree(snapshots_directory, ignore_errors=True)
 
     # Delete snapshots.
     lvremove_command = hook_config.get('lvremove_command', 'lvremove')

borgmatic/hooks/data_source/mariadb.py

@@ -1,10 +1,12 @@
 import copy
 import logging
 import os
+import re
 import shlex
 
 import borgmatic.borg.pattern
 import borgmatic.config.paths
+import borgmatic.hooks.credential.parse
 from borgmatic.execute import (
     execute_command,
     execute_command_and_capture_output,
@@ -22,14 +24,92 @@ def make_dump_path(base_directory):  # pragma: no cover
     return dump.make_data_source_dump_path(base_directory, 'mariadb_databases')
 
 
-SYSTEM_DATABASE_NAMES = ('information_schema', 'mysql', 'performance_schema', 'sys')
+DEFAULTS_EXTRA_FILE_FLAG_PATTERN = re.compile('^--defaults-extra-file=(?P<filename>.*)$')
 
 
-def database_names_to_dump(database, extra_environment, dry_run):
+def parse_extra_options(extra_options):
     '''
-    Given a requested database config, return the corresponding sequence of database names to dump.
-    In the case of "all", query for the names of databases on the configured host and return them,
-    excluding any system databases that will cause problems during restore.
+    Given an extra options string, split the options into a tuple and return it. Additionally, if
+    the first option is "--defaults-extra-file=...", then remove it from the options and return the
+    filename.
+
+    So the return value is a tuple of: (parsed options, defaults extra filename).
+
+    The intent is to support downstream merging of multiple "--defaults-extra-file"s, as
+    MariaDB/MySQL only allows one at a time.
+    '''
+    split_extra_options = tuple(shlex.split(extra_options)) if extra_options else ()
+
+    if not split_extra_options:
+        return ((), None)
+
+    match = DEFAULTS_EXTRA_FILE_FLAG_PATTERN.match(split_extra_options[0])
+
+    if not match:
+        return (split_extra_options, None)
+
+    return (split_extra_options[1:], match.group('filename'))
+
+
+def make_defaults_file_options(username=None, password=None, defaults_extra_filename=None):
+    '''
+    Given a database username and/or password, write it to an anonymous pipe and return the flags
+    for passing that file descriptor to an executed command. The idea is that this is a more secure
+    way to transmit credentials to a database client than using an environment variable.
+
+    If no username or password are given, then return the options for the given defaults extra
+    filename (if any). But if there is a username and/or password and a defaults extra filename is
+    given, then "!include" it from the generated file, effectively allowing multiple defaults extra
+    files.
+
+    Do not use the returned value for multiple different command invocations. That will not work
+    because each pipe is "used up" once read.
+    '''
+    escaped_password = None if password is None else password.replace('\\', '\\\\')
+
+    values = '\n'.join(
+        (
+            (f'user={username}' if username is not None else ''),
+            (f'password="{escaped_password}"' if escaped_password is not None else ''),
+        )
+    ).strip()
+
+    if not values:
+        if defaults_extra_filename:
+            return (f'--defaults-extra-file={defaults_extra_filename}',)
+
+        return ()
+
+    fields_message = ' and '.join(
+        field_name
+        for field_name in (
+            (f'username ({username})' if username is not None else None),
+            ('password' if password is not None else None),
+        )
+        if field_name is not None
+    )
+    include_message = f' (including {defaults_extra_filename})' if defaults_extra_filename else ''
+    logger.debug(f'Writing database {fields_message} to defaults extra file pipe{include_message}')
+
+    include = f'!include {defaults_extra_filename}\n' if defaults_extra_filename else ''
+
+    read_file_descriptor, write_file_descriptor = os.pipe()
+    os.write(write_file_descriptor, f'{include}[client]\n{values}'.encode('utf-8'))
+    os.close(write_file_descriptor)
+
+    # This plus subprocess.Popen(..., close_fds=False) in execute.py is necessary for the database
+    # client child process to inherit the file descriptor.
+    os.set_inheritable(read_file_descriptor, True)
+
+    return (f'--defaults-extra-file=/dev/fd/{read_file_descriptor}',)
+
+
+def database_names_to_dump(database, config, username, password, environment, dry_run):
+    '''
+    Given a requested database config, a configuration dict, a database username and password, an
+    environment dict, and whether this is a dry run, return the corresponding sequence of database
+    names to dump. In the case of "all", query for the names of databases on the configured host and
+    return them, excluding any system databases that will cause problems during restore.
     '''
     if database['name'] != 'all':
         return (database['name'],)
@@ -39,20 +119,29 @@ def database_names_to_dump(database, extra_environment, dry_run):
     mariadb_show_command = tuple(
         shlex.quote(part) for part in shlex.split(database.get('mariadb_command') or 'mariadb')
     )
+    extra_options, defaults_extra_filename = parse_extra_options(database.get('list_options'))
+    password_transport = database.get('password_transport', 'pipe')
     show_command = (
         mariadb_show_command
-        + (tuple(database['list_options'].split(' ')) if 'list_options' in database else ())
+        + (
+            make_defaults_file_options(username, password, defaults_extra_filename)
+            if password_transport == 'pipe'
+            else ()
+        )
+        + extra_options
         + (('--host', database['hostname']) if 'hostname' in database else ())
         + (('--port', str(database['port'])) if 'port' in database else ())
         + (('--protocol', 'tcp') if 'hostname' in database or 'port' in database else ())
-        + (('--user', database['username']) if 'username' in database else ())
+        + (('--user', username) if username and password_transport == 'environment' else ())
+        + (('--ssl',) if database.get('tls') is True else ())
+        + (('--skip-ssl',) if database.get('tls') is False else ())
         + ('--skip-column-names', '--batch')
         + ('--execute', 'show schemas')
     )
+
     logger.debug('Querying for "all" MariaDB databases to dump')
-    show_output = execute_command_and_capture_output(
-        show_command, extra_environment=extra_environment
-    )
+
+    show_output = execute_command_and_capture_output(show_command, environment=environment)
 
     return tuple(
         show_name
@@ -61,8 +150,19 @@ def database_names_to_dump(database, extra_environment, dry_run):
     )
 
 
+SYSTEM_DATABASE_NAMES = ('information_schema', 'mysql', 'performance_schema', 'sys')
+
+
 def execute_dump_command(
-    database, dump_path, database_names, extra_environment, dry_run, dry_run_label
+    database,
+    config,
+    username,
+    password,
+    dump_path,
+    database_names,
+    environment,
+    dry_run,
+    dry_run_label,
 ):
     '''
     Kick off a dump for the given MariaDB database (provided as a configuration dict) to a named
@@ -89,14 +189,23 @@ def execute_dump_command(
         shlex.quote(part)
         for part in shlex.split(database.get('mariadb_dump_command') or 'mariadb-dump')
     )
+    extra_options, defaults_extra_filename = parse_extra_options(database.get('options'))
+    password_transport = database.get('password_transport', 'pipe')
     dump_command = (
         mariadb_dump_command
-        + (tuple(database['options'].split(' ')) if 'options' in database else ())
+        + (
+            make_defaults_file_options(username, password, defaults_extra_filename)
+            if password_transport == 'pipe'
+            else ()
+        )
+        + extra_options
         + (('--add-drop-database',) if database.get('add_drop_database', True) else ())
         + (('--host', database['hostname']) if 'hostname' in database else ())
         + (('--port', str(database['port'])) if 'port' in database else ())
         + (('--protocol', 'tcp') if 'hostname' in database or 'port' in database else ())
-        + (('--user', database['username']) if 'username' in database else ())
+        + (('--user', username) if username and password_transport == 'environment' else ())
+        + (('--ssl',) if database.get('tls') is True else ())
+        + (('--skip-ssl',) if database.get('tls') is False else ())
         + ('--databases',)
         + database_names
         + ('--result-file', dump_filename)
@@ -110,7 +219,7 @@ def execute_dump_command(
 
     return execute_command(
         dump_command,
-        extra_environment=extra_environment,
+        environment=environment,
         run_to_completion=False,
     )
 
@@ -152,8 +261,23 @@ def dump_data_sources(
 
     for database in databases:
         dump_path = make_dump_path(borgmatic_runtime_directory)
-        extra_environment = {'MYSQL_PWD': database['password']} if 'password' in database else None
-        dump_database_names = database_names_to_dump(database, extra_environment, dry_run)
+        username = borgmatic.hooks.credential.parse.resolve_credential(
+            database.get('username'), config
+        )
+        password = borgmatic.hooks.credential.parse.resolve_credential(
+            database.get('password'), config
+        )
+        environment = dict(
+            os.environ,
+            **(
+                {'MYSQL_PWD': password}
+                if password and database.get('password_transport') == 'environment'
+                else {}
+            ),
+        )
+        dump_database_names = database_names_to_dump(
+            database, config, username, password, environment, dry_run
+        )
 
         if not dump_database_names:
             if dry_run:
@@ -168,9 +292,12 @@ def dump_data_sources(
                 processes.append(
                     execute_dump_command(
                         renamed_database,
+                        config,
+                        username,
+                        password,
                         dump_path,
                         (dump_name,),
-                        extra_environment,
+                        environment,
                         dry_run,
                         dry_run_label,
                     )
@@ -179,9 +306,12 @@ def dump_data_sources(
             processes.append(
                 execute_dump_command(
                     database,
+                    config,
+                    username,
+                    password,
                     dump_path,
                     dump_database_names,
-                    extra_environment,
+                    environment,
                     dry_run,
                     dry_run_label,
                 )
@@ -190,7 +320,8 @@ def dump_data_sources(
     if not dry_run:
         patterns.append(
             borgmatic.borg.pattern.Pattern(
-                os.path.join(borgmatic_runtime_directory, 'mariadb_databases')
+                os.path.join(borgmatic_runtime_directory, 'mariadb_databases'),
+                source=borgmatic.borg.pattern.Pattern_source.HOOK,
             )
         )
 
@@ -251,30 +382,47 @@ def restore_data_source_dump(
     port = str(
         connection_params['port'] or data_source.get('restore_port', data_source.get('port', ''))
     )
-    username = connection_params['username'] or data_source.get(
-        'restore_username', data_source.get('username')
+    tls = data_source.get('restore_tls', data_source.get('tls'))
+    username = borgmatic.hooks.credential.parse.resolve_credential(
+        (
+            connection_params['username']
+            or data_source.get('restore_username', data_source.get('username'))
+        ),
+        config,
     )
-    password = connection_params['password'] or data_source.get(
-        'restore_password', data_source.get('password')
+    password = borgmatic.hooks.credential.parse.resolve_credential(
+        (
+            connection_params['password']
+            or data_source.get('restore_password', data_source.get('password'))
+        ),
+        config,
     )
 
     mariadb_restore_command = tuple(
         shlex.quote(part) for part in shlex.split(data_source.get('mariadb_command') or 'mariadb')
     )
+    extra_options, defaults_extra_filename = parse_extra_options(data_source.get('restore_options'))
+    password_transport = data_source.get('password_transport', 'pipe')
     restore_command = (
         mariadb_restore_command
-        + ('--batch',)
         + (
-            tuple(data_source['restore_options'].split(' '))
-            if 'restore_options' in data_source
+            make_defaults_file_options(username, password, defaults_extra_filename)
+            if password_transport == 'pipe'
             else ()
         )
+        + extra_options
+        + ('--batch',)
         + (('--host', hostname) if hostname else ())
         + (('--port', str(port)) if port else ())
         + (('--protocol', 'tcp') if hostname or port else ())
-        + (('--user', username) if username else ())
+        + (('--user', username) if username and password_transport == 'environment' else ())
+        + (('--ssl',) if tls is True else ())
+        + (('--skip-ssl',) if tls is False else ())
+    )
+    environment = dict(
+        os.environ,
+        **({'MYSQL_PWD': password} if password and password_transport == 'environment' else {}),
     )
-    extra_environment = {'MYSQL_PWD': password} if password else None
 
     logger.debug(f"Restoring MariaDB database {data_source['name']}{dry_run_label}")
     if dry_run:
@@ -287,5 +435,5 @@ def restore_data_source_dump(
         [extract_process],
         output_log_level=logging.DEBUG,
         input_file=extract_process.stdout,
-        extra_environment=extra_environment,
+        environment=environment,
     )

borgmatic/hooks/data_source/mongodb.py

@@ -4,6 +4,7 @@ import shlex
 
 import borgmatic.borg.pattern
 import borgmatic.config.paths
+import borgmatic.hooks.credential.parse
 from borgmatic.execute import execute_command, execute_command_with_processes
 from borgmatic.hooks.data_source import dump
 
@@ -52,6 +53,7 @@ def dump_data_sources(
     logger.info(f'Dumping MongoDB databases{dry_run_label}')
 
     processes = []
+
     for database in databases:
         name = database['name']
         dump_filename = dump.make_data_source_dump_filename(
@@ -68,38 +70,78 @@ def dump_data_sources(
         if dry_run:
             continue
 
-        command = build_dump_command(database, dump_filename, dump_format)
+        command = build_dump_command(database, config, dump_filename, dump_format)
 
         if dump_format == 'directory':
             dump.create_parent_directory_for_dump(dump_filename)
-            execute_command(command, shell=True)
+            execute_command(command, shell=True)  # noqa: S604
         else:
             dump.create_named_pipe_for_dump(dump_filename)
-            processes.append(execute_command(command, shell=True, run_to_completion=False))
-
+            processes.append(
+                execute_command(command, shell=True, run_to_completion=False)  # noqa: S604
+            )
     if not dry_run:
         patterns.append(
             borgmatic.borg.pattern.Pattern(
-                os.path.join(borgmatic_runtime_directory, 'mongodb_databases')
+                os.path.join(borgmatic_runtime_directory, 'mongodb_databases'),
+                source=borgmatic.borg.pattern.Pattern_source.HOOK,
             )
         )
 
     return processes
 
 
-def build_dump_command(database, dump_filename, dump_format):
+def make_password_config_file(password):
     '''
-    Return the mongodump command from a single database configuration.
+    Given a database password, write it as a MongoDB configuration file to an anonymous pipe and
+    return its filename. The idea is that this is a more secure way to transmit a password to
+    MongoDB than providing it directly on the command-line.
+
+    Do not use the returned value for multiple different command invocations. That will not work
+    because each pipe is "used up" once read.
+    '''
+    logger.debug('Writing MongoDB password to configuration file pipe')
+
+    read_file_descriptor, write_file_descriptor = os.pipe()
+    os.write(write_file_descriptor, f'password: {password}'.encode('utf-8'))
+    os.close(write_file_descriptor)
+
+    # This plus subprocess.Popen(..., close_fds=False) in execute.py is necessary for the database
+    # client child process to inherit the file descriptor.
+    os.set_inheritable(read_file_descriptor, True)
+
+    return f'/dev/fd/{read_file_descriptor}'
+
+
+def build_dump_command(database, config, dump_filename, dump_format):
+    '''
+    Return the custom mongodump_command from a single database configuration.
     '''
     all_databases = database['name'] == 'all'
 
+    password = borgmatic.hooks.credential.parse.resolve_credential(database.get('password'), config)
+
+    dump_command = tuple(
+        shlex.quote(part) for part in shlex.split(database.get('mongodump_command') or 'mongodump')
+    )
     return (
-        ('mongodump',)
+        dump_command
         + (('--out', shlex.quote(dump_filename)) if dump_format == 'directory' else ())
         + (('--host', shlex.quote(database['hostname'])) if 'hostname' in database else ())
         + (('--port', shlex.quote(str(database['port']))) if 'port' in database else ())
-        + (('--username', shlex.quote(database['username'])) if 'username' in database else ())
-        + (('--password', shlex.quote(database['password'])) if 'password' in database else ())
+        + (
+            (
+                '--username',
+                shlex.quote(
+                    borgmatic.hooks.credential.parse.resolve_credential(
+                        database['username'], config
+                    )
+                ),
+            )
+            if 'username' in database
+            else ()
+        )
+        + (('--config', make_password_config_file(password)) if password else ())
         + (
             ('--authenticationDatabase', shlex.quote(database['authentication_database']))
             if 'authentication_database' in database
@@ -173,7 +215,7 @@ def restore_data_source_dump(
         data_source.get('hostname'),
     )
     restore_command = build_restore_command(
-        extract_process, data_source, dump_filename, connection_params
+        extract_process, data_source, config, dump_filename, connection_params
     )
 
     logger.debug(f"Restoring MongoDB database {data_source['name']}{dry_run_label}")
@@ -190,22 +232,33 @@ def restore_data_source_dump(
     )
 
 
-def build_restore_command(extract_process, database, dump_filename, connection_params):
+def build_restore_command(extract_process, database, config, dump_filename, connection_params):
     '''
-    Return the mongorestore command from a single database configuration.
+    Return the custom mongorestore_command from a single database configuration.
     '''
     hostname = connection_params['hostname'] or database.get(
         'restore_hostname', database.get('hostname')
     )
     port = str(connection_params['port'] or database.get('restore_port', database.get('port', '')))
-    username = connection_params['username'] or database.get(
-        'restore_username', database.get('username')
+    username = borgmatic.hooks.credential.parse.resolve_credential(
+        (
+            connection_params['username']
+            or database.get('restore_username', database.get('username'))
+        ),
+        config,
     )
-    password = connection_params['password'] or database.get(
-        'restore_password', database.get('password')
+    password = borgmatic.hooks.credential.parse.resolve_credential(
+        (
+            connection_params['password']
+            or database.get('restore_password', database.get('password'))
+        ),
+        config,
     )
 
-    command = ['mongorestore']
+    command = list(
+        shlex.quote(part)
+        for part in shlex.split(database.get('mongorestore_command') or 'mongorestore')
+    )
     if extract_process:
         command.append('--archive')
     else:
@@ -219,7 +272,7 @@ def build_restore_command(extract_process, database, dump_filename, connection_p
     if username:
         command.extend(('--username', username))
     if password:
-        command.extend(('--password', password))
+        command.extend(('--config', make_password_config_file(password)))
     if 'authentication_database' in database:
         command.extend(('--authenticationDatabase', database['authentication_database']))
     if 'restore_options' in database:

borgmatic/hooks/data_source/mysql.py

@@ -5,6 +5,8 @@ import shlex
 
 import borgmatic.borg.pattern
 import borgmatic.config.paths
+import borgmatic.hooks.credential.parse
+import borgmatic.hooks.data_source.mariadb
 from borgmatic.execute import (
     execute_command,
     execute_command_and_capture_output,
@@ -25,11 +27,12 @@ def make_dump_path(base_directory):  # pragma: no cover
 SYSTEM_DATABASE_NAMES = ('information_schema', 'mysql', 'performance_schema', 'sys')
 
 
-def database_names_to_dump(database, extra_environment, dry_run):
+def database_names_to_dump(database, config, username, password, environment, dry_run):
     '''
-    Given a requested database config, return the corresponding sequence of database names to dump.
-    In the case of "all", query for the names of databases on the configured host and return them,
-    excluding any system databases that will cause problems during restore.
+    Given a requested database config, a configuration dict, a database username and password, an
+    environment dict, and whether this is a dry run, return the corresponding sequence of database
+    names to dump. In the case of "all", query for the names of databases on the configured host and
+    return them, excluding any system databases that will cause problems during restore.
     '''
     if database['name'] != 'all':
         return (database['name'],)
@@ -39,20 +42,33 @@ def database_names_to_dump(database, extra_environment, dry_run):
     mysql_show_command = tuple(
         shlex.quote(part) for part in shlex.split(database.get('mysql_command') or 'mysql')
     )
+    extra_options, defaults_extra_filename = (
+        borgmatic.hooks.data_source.mariadb.parse_extra_options(database.get('list_options'))
+    )
+    password_transport = database.get('password_transport', 'pipe')
     show_command = (
         mysql_show_command
-        + (tuple(database['list_options'].split(' ')) if 'list_options' in database else ())
+        + (
+            borgmatic.hooks.data_source.mariadb.make_defaults_file_options(
+                username, password, defaults_extra_filename
+            )
+            if password_transport == 'pipe'
+            else ()
+        )
+        + extra_options
         + (('--host', database['hostname']) if 'hostname' in database else ())
         + (('--port', str(database['port'])) if 'port' in database else ())
         + (('--protocol', 'tcp') if 'hostname' in database or 'port' in database else ())
-        + (('--user', database['username']) if 'username' in database else ())
+        + (('--user', username) if username and password_transport == 'environment' else ())
+        + (('--ssl',) if database.get('tls') is True else ())
+        + (('--skip-ssl',) if database.get('tls') is False else ())
         + ('--skip-column-names', '--batch')
         + ('--execute', 'show schemas')
     )
+
     logger.debug('Querying for "all" MySQL databases to dump')
-    show_output = execute_command_and_capture_output(
-        show_command, extra_environment=extra_environment
-    )
+
+    show_output = execute_command_and_capture_output(show_command, environment=environment)
 
     return tuple(
         show_name
@@ -62,7 +78,15 @@ def database_names_to_dump(database, extra_environment, dry_run):
 
 
 def execute_dump_command(
-    database, dump_path, database_names, extra_environment, dry_run, dry_run_label
+    database,
+    config,
+    username,
+    password,
+    dump_path,
+    database_names,
+    environment,
+    dry_run,
+    dry_run_label,
 ):
     '''
     Kick off a dump for the given MySQL/MariaDB database (provided as a configuration dict) to a
@@ -88,14 +112,27 @@ def execute_dump_command(
     mysql_dump_command = tuple(
         shlex.quote(part) for part in shlex.split(database.get('mysql_dump_command') or 'mysqldump')
     )
+    extra_options, defaults_extra_filename = (
+        borgmatic.hooks.data_source.mariadb.parse_extra_options(database.get('options'))
+    )
+    password_transport = database.get('password_transport', 'pipe')
     dump_command = (
         mysql_dump_command
-        + (tuple(database['options'].split(' ')) if 'options' in database else ())
+        + (
+            borgmatic.hooks.data_source.mariadb.make_defaults_file_options(
+                username, password, defaults_extra_filename
+            )
+            if password_transport == 'pipe'
+            else ()
+        )
+        + extra_options
         + (('--add-drop-database',) if database.get('add_drop_database', True) else ())
         + (('--host', database['hostname']) if 'hostname' in database else ())
         + (('--port', str(database['port'])) if 'port' in database else ())
         + (('--protocol', 'tcp') if 'hostname' in database or 'port' in database else ())
-        + (('--user', database['username']) if 'username' in database else ())
+        + (('--user', username) if username and password_transport == 'environment' else ())
+        + (('--ssl',) if database.get('tls') is True else ())
+        + (('--skip-ssl',) if database.get('tls') is False else ())
         + ('--databases',)
         + database_names
         + ('--result-file', dump_filename)
@@ -109,7 +146,7 @@ def execute_dump_command(
 
     return execute_command(
         dump_command,
-        extra_environment=extra_environment,
+        environment=environment,
         run_to_completion=False,
     )
 
@@ -151,8 +188,23 @@ def dump_data_sources(
 
     for database in databases:
         dump_path = make_dump_path(borgmatic_runtime_directory)
-        extra_environment = {'MYSQL_PWD': database['password']} if 'password' in database else None
-        dump_database_names = database_names_to_dump(database, extra_environment, dry_run)
+        username = borgmatic.hooks.credential.parse.resolve_credential(
+            database.get('username'), config
+        )
+        password = borgmatic.hooks.credential.parse.resolve_credential(
+            database.get('password'), config
+        )
+        environment = dict(
+            os.environ,
+            **(
+                {'MYSQL_PWD': password}
+                if password and database.get('password_transport') == 'environment'
+                else {}
+            ),
+        )
+        dump_database_names = database_names_to_dump(
+            database, config, username, password, environment, dry_run
+        )
 
         if not dump_database_names:
             if dry_run:
@@ -167,9 +219,12 @@ def dump_data_sources(
                 processes.append(
                     execute_dump_command(
                         renamed_database,
+                        config,
+                        username,
+                        password,
                         dump_path,
                         (dump_name,),
-                        extra_environment,
+                        environment,
                         dry_run,
                         dry_run_label,
                     )
@@ -178,9 +233,12 @@ def dump_data_sources(
             processes.append(
                 execute_dump_command(
                     database,
+                    config,
+                    username,
+                    password,
                     dump_path,
                     dump_database_names,
-                    extra_environment,
+                    environment,
                     dry_run,
                     dry_run_label,
                 )
@@ -189,7 +247,8 @@ def dump_data_sources(
     if not dry_run:
         patterns.append(
             borgmatic.borg.pattern.Pattern(
-                os.path.join(borgmatic_runtime_directory, 'mysql_databases')
+                os.path.join(borgmatic_runtime_directory, 'mysql_databases'),
+                source=borgmatic.borg.pattern.Pattern_source.HOOK,
             )
         )
 
@@ -250,30 +309,51 @@ def restore_data_source_dump(
     port = str(
         connection_params['port'] or data_source.get('restore_port', data_source.get('port', ''))
     )
-    username = connection_params['username'] or data_source.get(
-        'restore_username', data_source.get('username')
+    tls = data_source.get('restore_tls', data_source.get('tls'))
+    username = borgmatic.hooks.credential.parse.resolve_credential(
+        (
+            connection_params['username']
+            or data_source.get('restore_username', data_source.get('username'))
+        ),
+        config,
     )
-    password = connection_params['password'] or data_source.get(
-        'restore_password', data_source.get('password')
+    password = borgmatic.hooks.credential.parse.resolve_credential(
+        (
+            connection_params['password']
+            or data_source.get('restore_password', data_source.get('password'))
+        ),
+        config,
     )
 
     mysql_restore_command = tuple(
         shlex.quote(part) for part in shlex.split(data_source.get('mysql_command') or 'mysql')
     )
+    extra_options, defaults_extra_filename = (
+        borgmatic.hooks.data_source.mariadb.parse_extra_options(data_source.get('restore_options'))
+    )
+    password_transport = data_source.get('password_transport', 'pipe')
     restore_command = (
         mysql_restore_command
-        + ('--batch',)
         + (
-            tuple(data_source['restore_options'].split(' '))
-            if 'restore_options' in data_source
+            borgmatic.hooks.data_source.mariadb.make_defaults_file_options(
+                username, password, defaults_extra_filename
+            )
+            if password_transport == 'pipe'
             else ()
         )
+        + extra_options
+        + ('--batch',)
         + (('--host', hostname) if hostname else ())
         + (('--port', str(port)) if port else ())
         + (('--protocol', 'tcp') if hostname or port else ())
-        + (('--user', username) if username else ())
+        + (('--user', username) if username and password_transport == 'environment' else ())
+        + (('--ssl',) if tls is True else ())
+        + (('--skip-ssl',) if tls is False else ())
+    )
+    environment = dict(
+        os.environ,
+        **({'MYSQL_PWD': password} if password and password_transport == 'environment' else {}),
     )
-    extra_environment = {'MYSQL_PWD': password} if password else None
 
     logger.debug(f"Restoring MySQL database {data_source['name']}{dry_run_label}")
     if dry_run:
@@ -286,5 +366,5 @@ def restore_data_source_dump(
         [extract_process],
         output_log_level=logging.DEBUG,
         input_file=extract_process.stdout,
-        extra_environment=extra_environment,
+        environment=environment,
     )

borgmatic/hooks/data_source/postgresql.py

@@ -7,6 +7,7 @@ import shlex
 
 import borgmatic.borg.pattern
 import borgmatic.config.paths
+import borgmatic.hooks.credential.parse
 from borgmatic.execute import (
     execute_command,
     execute_command_and_capture_output,
@@ -24,46 +25,52 @@ def make_dump_path(base_directory):  # pragma: no cover
     return dump.make_data_source_dump_path(base_directory, 'postgresql_databases')
 
 
-def make_extra_environment(database, restore_connection_params=None):
+def make_environment(database, config, restore_connection_params=None):
     '''
-    Make the extra_environment dict from the given database configuration. If restore connection
-    params are given, this is for a restore operation.
+    Make an environment dict from the current environment variables and the given database
+    configuration. If restore connection params are given, this is for a restore operation.
     '''
-    extra = dict()
+    environment = dict(os.environ)
 
     try:
         if restore_connection_params:
-            extra['PGPASSWORD'] = restore_connection_params.get('password') or database.get(
-                'restore_password', database['password']
+            environment['PGPASSWORD'] = borgmatic.hooks.credential.parse.resolve_credential(
+                (
+                    restore_connection_params.get('password')
+                    or database.get('restore_password', database['password'])
+                ),
+                config,
             )
         else:
-            extra['PGPASSWORD'] = database['password']
+            environment['PGPASSWORD'] = borgmatic.hooks.credential.parse.resolve_credential(
+                database['password'], config
+            )
     except (AttributeError, KeyError):
         pass
 
     if 'ssl_mode' in database:
-        extra['PGSSLMODE'] = database['ssl_mode']
+        environment['PGSSLMODE'] = database['ssl_mode']
     if 'ssl_cert' in database:
-        extra['PGSSLCERT'] = database['ssl_cert']
+        environment['PGSSLCERT'] = database['ssl_cert']
     if 'ssl_key' in database:
-        extra['PGSSLKEY'] = database['ssl_key']
+        environment['PGSSLKEY'] = database['ssl_key']
     if 'ssl_root_cert' in database:
-        extra['PGSSLROOTCERT'] = database['ssl_root_cert']
+        environment['PGSSLROOTCERT'] = database['ssl_root_cert']
     if 'ssl_crl' in database:
-        extra['PGSSLCRL'] = database['ssl_crl']
+        environment['PGSSLCRL'] = database['ssl_crl']
 
-    return extra
+    return environment
 
 
 EXCLUDED_DATABASE_NAMES = ('template0', 'template1')
 
 
-def database_names_to_dump(database, extra_environment, dry_run):
+def database_names_to_dump(database, config, environment, dry_run):
     '''
-    Given a requested database config, return the corresponding sequence of database names to dump.
-    In the case of "all" when a database format is given, query for the names of databases on the
-    configured host and return them. For "all" without a database format, just return a sequence
-    containing "all".
+    Given a requested database config and a configuration dict, return the corresponding sequence of
+    database names to dump. In the case of "all" when a database format is given, query for the
+    names of databases on the configured host and return them. For "all" without a database format,
+    just return a sequence containing "all".
     '''
     requested_name = database['name']
 
@@ -82,13 +89,18 @@ def database_names_to_dump(database, extra_environment, dry_run):
         + ('--list', '--no-password', '--no-psqlrc', '--csv', '--tuples-only')
         + (('--host', database['hostname']) if 'hostname' in database else ())
         + (('--port', str(database['port'])) if 'port' in database else ())
-        + (('--username', database['username']) if 'username' in database else ())
+        + (
+            (
+                '--username',
+                borgmatic.hooks.credential.parse.resolve_credential(database['username'], config),
+            )
+            if 'username' in database
+            else ()
+        )
         + (tuple(database['list_options'].split(' ')) if 'list_options' in database else ())
     )
     logger.debug('Querying for "all" PostgreSQL databases to dump')
-    list_output = execute_command_and_capture_output(
-        list_command, extra_environment=extra_environment
-    )
+    list_output = execute_command_and_capture_output(list_command, environment=environment)
 
     return tuple(
         row[0]
@@ -135,9 +147,9 @@ def dump_data_sources(
     logger.info(f'Dumping PostgreSQL databases{dry_run_label}')
 
     for database in databases:
-        extra_environment = make_extra_environment(database)
+        environment = make_environment(database, config)
         dump_path = make_dump_path(borgmatic_runtime_directory)
-        dump_database_names = database_names_to_dump(database, extra_environment, dry_run)
+        dump_database_names = database_names_to_dump(database, config, environment, dry_run)
 
         if not dump_database_names:
             if dry_run:
@@ -147,6 +159,7 @@ def dump_data_sources(
 
         for database_name in dump_database_names:
             dump_format = database.get('format', None if database_name == 'all' else 'custom')
+            compression = database.get('compression')
             default_dump_command = 'pg_dumpall' if database_name == 'all' else 'pg_dump'
             dump_command = tuple(
                 shlex.quote(part)
@@ -174,12 +187,20 @@ def dump_data_sources(
                 + (('--host', shlex.quote(database['hostname'])) if 'hostname' in database else ())
                 + (('--port', shlex.quote(str(database['port']))) if 'port' in database else ())
                 + (
-                    ('--username', shlex.quote(database['username']))
+                    (
+                        '--username',
+                        shlex.quote(
+                            borgmatic.hooks.credential.parse.resolve_credential(
+                                database['username'], config
+                            )
+                        ),
+                    )
                     if 'username' in database
                     else ()
                 )
                 + (('--no-owner',) if database.get('no_owner', False) else ())
                 + (('--format', shlex.quote(dump_format)) if dump_format else ())
+                + (('--compress', shlex.quote(str(compression))) if compression is not None else ())
                 + (('--file', shlex.quote(dump_filename)) if dump_format == 'directory' else ())
                 + (
                     tuple(shlex.quote(option) for option in database['options'].split(' '))
@@ -203,16 +224,16 @@ def dump_data_sources(
                 dump.create_parent_directory_for_dump(dump_filename)
                 execute_command(
                     command,
-                    shell=True,
-                    extra_environment=extra_environment,
+                    shell=True,  # noqa: S604
+                    environment=environment,
                 )
             else:
                 dump.create_named_pipe_for_dump(dump_filename)
                 processes.append(
                     execute_command(
                         command,
-                        shell=True,
-                        extra_environment=extra_environment,
+                        shell=True,  # noqa: S604
+                        environment=environment,
                         run_to_completion=False,
                     )
                 )
@@ -220,7 +241,8 @@ def dump_data_sources(
     if not dry_run:
         patterns.append(
             borgmatic.borg.pattern.Pattern(
-                os.path.join(borgmatic_runtime_directory, 'postgresql_databases')
+                os.path.join(borgmatic_runtime_directory, 'postgresql_databases'),
+                source=borgmatic.borg.pattern.Pattern_source.HOOK,
             )
         )
 
@@ -290,8 +312,12 @@ def restore_data_source_dump(
     port = str(
         connection_params['port'] or data_source.get('restore_port', data_source.get('port', ''))
     )
-    username = connection_params['username'] or data_source.get(
-        'restore_username', data_source.get('username')
+    username = borgmatic.hooks.credential.parse.resolve_credential(
+        (
+            connection_params['username']
+            or data_source.get('restore_username', data_source.get('username'))
+        ),
+        config,
     )
 
     all_databases = bool(data_source['name'] == 'all')
@@ -344,9 +370,7 @@ def restore_data_source_dump(
         )
     )
 
-    extra_environment = make_extra_environment(
-        data_source, restore_connection_params=connection_params
-    )
+    environment = make_environment(data_source, config, restore_connection_params=connection_params)
 
     logger.debug(f"Restoring PostgreSQL database {data_source['name']}{dry_run_label}")
     if dry_run:
@@ -359,6 +383,6 @@ def restore_data_source_dump(
         [extract_process] if extract_process else [],
         output_log_level=logging.DEBUG,
         input_file=extract_process.stdout if extract_process else None,
-        extra_environment=extra_environment,
+        environment=environment,
     )
-    execute_command(analyze_command, extra_environment=extra_environment)
+    execute_command(analyze_command, environment=environment)

borgmatic/hooks/data_source/snapshot.py

@@ -1,3 +1,4 @@
+import os
 import pathlib
 
 IS_A_HOOK = False
@@ -7,14 +8,18 @@ def get_contained_patterns(parent_directory, candidate_patterns):
     '''
     Given a parent directory and a set of candidate patterns potentially inside it, get the subset
     of contained patterns for which the parent directory is actually the parent, a grandparent, the
-    very same directory, etc. The idea is if, say, /var/log and /var/lib are candidate pattern
-    paths, but there's a parent directory (logical volume, dataset, subvolume, etc.) at /var, then
-    /var is what we want to snapshot.
+    very same directory, etc. The idea is if, say, "/var/log" and "/var/lib" are candidate pattern
+    paths, but there's a parent directory (logical volume, dataset, subvolume, etc.) at "/var", then
+    "/var" is what we want to snapshot.
 
-    For this to work, a candidate pattern path can't have any globs or other non-literal characters
-    in the initial portion of the path that matches the parent directory. For instance, a parent
-    directory of /var would match a candidate pattern path of /var/log/*/data, but not a pattern
-    path like /v*/log/*/data.
+    If a parent directory and a candidate pattern are on different devices, skip the pattern. That's
+    because any snapshot of a parent directory won't actually include "contained" directories if
+    they reside on separate devices.
+
+    For this function to work, a candidate pattern path can't have any globs or other non-literal
+    characters in the initial portion of the path that matches the parent directory. For instance, a
+    parent directory of "/var" would match a candidate pattern path of "/var/log/*/data", but not a
+    pattern path like "/v*/log/*/data".
 
     The one exception is that if a regular expression pattern path starts with "^", that will get
     stripped off for purposes of matching against a parent directory.
@@ -27,12 +32,17 @@ def get_contained_patterns(parent_directory, candidate_patterns):
     if not candidate_patterns:
         return ()
 
+    parent_device = os.stat(parent_directory).st_dev if os.path.exists(parent_directory) else None
+
     contained_patterns = tuple(
         candidate
         for candidate in candidate_patterns
         for candidate_path in (pathlib.PurePath(candidate.path.lstrip('^')),)
-        if pathlib.PurePath(parent_directory) == candidate_path
-        or pathlib.PurePath(parent_directory) in candidate_path.parents
+        if (
+            pathlib.PurePath(parent_directory) == candidate_path
+            or pathlib.PurePath(parent_directory) in candidate_path.parents
+        )
+        if candidate.device == parent_device
     )
     candidate_patterns -= set(contained_patterns)
 

borgmatic/hooks/data_source/sqlite.py

@@ -71,13 +71,16 @@ def dump_data_sources(
             )
             continue
 
-        command = (
-            'sqlite3',
+        sqlite_command = tuple(
+            shlex.quote(part) for part in shlex.split(database.get('sqlite_command') or 'sqlite3')
+        )
+        command = sqlite_command + (
             shlex.quote(database_path),
             '.dump',
             '>',
             shlex.quote(dump_filename),
         )
+
         logger.debug(
             f'Dumping SQLite database at {database_path} to {dump_filename}{dry_run_label}'
         )
@@ -85,12 +88,15 @@ def dump_data_sources(
             continue
 
         dump.create_named_pipe_for_dump(dump_filename)
-        processes.append(execute_command(command, shell=True, run_to_completion=False))
+        processes.append(
+            execute_command(command, shell=True, run_to_completion=False)  # noqa: S604
+        )
 
     if not dry_run:
         patterns.append(
             borgmatic.borg.pattern.Pattern(
-                os.path.join(borgmatic_runtime_directory, 'sqlite_databases')
+                os.path.join(borgmatic_runtime_directory, 'sqlite_databases'),
+                source=borgmatic.borg.pattern.Pattern_source.HOOK,
             )
         )
 
@@ -159,11 +165,11 @@ def restore_data_source_dump(
     except FileNotFoundError:  # pragma: no cover
         pass
 
-    restore_command = (
-        'sqlite3',
-        database_path,
+    sqlite_restore_command = tuple(
+        shlex.quote(part)
+        for part in shlex.split(data_source.get('sqlite_restore_command') or 'sqlite3')
     )
-
+    restore_command = sqlite_restore_command + (shlex.quote(database_path),)
     # Don't give Borg local path so as to error on warnings, as "borg extract" only gives a warning
     # if the restore paths don't exist in the archive.
     execute_command_with_processes(

borgmatic/hooks/data_source/zfs.py

@@ -1,5 +1,6 @@
 import collections
 import glob
+import hashlib
 import logging
 import os
 import shutil
@@ -38,6 +39,9 @@ def get_datasets_to_backup(zfs_command, patterns):
     pattern paths represent the requested datasets to snapshot. But also include any datasets tagged
     with a borgmatic-specific user property, whether or not they appear in the patterns.
 
+    Only include datasets that contain at least one root pattern sourced from borgmatic
+    configuration (as opposed to generated elsewhere in borgmatic).
+
     Return the result as a sequence of Dataset instances, sorted by mount point.
     '''
     list_output = borgmatic.execute.execute_command_and_capture_output(
@@ -48,8 +52,9 @@ def get_datasets_to_backup(zfs_command, patterns):
             '-t',
             'filesystem',
             '-o',
-            f'name,mountpoint,{BORGMATIC_USER_PROPERTY}',
-        )
+            f'name,mountpoint,canmount,{BORGMATIC_USER_PROPERTY}',
+        ),
+        close_fds=True,
     )
 
     try:
@@ -60,7 +65,12 @@ def get_datasets_to_backup(zfs_command, patterns):
             (
                 Dataset(dataset_name, mount_point, (user_property_value == 'auto'), ())
                 for line in list_output.splitlines()
-                for (dataset_name, mount_point, user_property_value) in (line.rstrip().split('\t'),)
+                for (dataset_name, mount_point, can_mount, user_property_value) in (
+                    line.rstrip().split('\t'),
+                )
+                # Skip datasets that are marked "canmount=off", because mounting their snapshots will
+                # result in completely empty mount points—thereby preventing us from backing them up.
+                if can_mount == 'on'
             ),
             key=lambda dataset: dataset.mount_point,
             reverse=True,
@@ -83,7 +93,12 @@ def get_datasets_to_backup(zfs_command, patterns):
                 for contained_patterns in (
                     (
                         (
-                            (borgmatic.borg.pattern.Pattern(dataset.mount_point),)
+                            (
+                                borgmatic.borg.pattern.Pattern(
+                                    dataset.mount_point,
+                                    source=borgmatic.borg.pattern.Pattern_source.HOOK,
+                                ),
+                            )
                             if dataset.auto_backup
                             else ()
                         )
@@ -92,7 +107,12 @@ def get_datasets_to_backup(zfs_command, patterns):
                         )
                     ),
                 )
-                if contained_patterns
+                if dataset.auto_backup
+                or any(
+                    pattern.type == borgmatic.borg.pattern.Pattern_type.ROOT
+                    and pattern.source == borgmatic.borg.pattern.Pattern_source.CONFIG
+                    for pattern in contained_patterns
+                )
             ),
             key=lambda dataset: dataset.mount_point,
         )
@@ -112,10 +132,20 @@ def get_all_dataset_mount_points(zfs_command):
             'filesystem',
             '-o',
             'mountpoint',
-        )
+        ),
+        close_fds=True,
     )
 
-    return tuple(sorted(line.rstrip() for line in list_output.splitlines()))
+    return tuple(
+        sorted(
+            {
+                mount_point
+                for line in list_output.splitlines()
+                for mount_point in (line.rstrip(),)
+                if mount_point != 'none'
+            }
+        )
+    )
 
 
 def snapshot_dataset(zfs_command, full_snapshot_name):  # pragma: no cover
@@ -130,6 +160,7 @@ def snapshot_dataset(zfs_command, full_snapshot_name):  # pragma: no cover
             full_snapshot_name,
         ),
         output_log_level=logging.DEBUG,
+        close_fds=True,
     )
 
 
@@ -152,13 +183,18 @@ def mount_snapshot(mount_command, full_snapshot_name, snapshot_mount_path):  # p
             snapshot_mount_path,
         ),
         output_log_level=logging.DEBUG,
+        close_fds=True,
     )
 
 
-def make_borg_snapshot_pattern(pattern, normalized_runtime_directory):
+MOUNT_POINT_HASH_LENGTH = 10
+
+
+def make_borg_snapshot_pattern(pattern, dataset, normalized_runtime_directory):
     '''
-    Given a Borg pattern as a borgmatic.borg.pattern.Pattern instance, return a new Pattern with its
-    path rewritten to be in a snapshot directory based on the given runtime directory.
+    Given a Borg pattern as a borgmatic.borg.pattern.Pattern instance and the Dataset containing it,
+    return a new Pattern with its path rewritten to be in a snapshot directory based on both the
+    given runtime directory and the given Dataset's mount point.
 
     Move any initial caret in a regular expression pattern path to the beginning, so as not to break
     the regular expression.
@@ -173,6 +209,10 @@ def make_borg_snapshot_pattern(pattern, normalized_runtime_directory):
     rewritten_path = initial_caret + os.path.join(
         normalized_runtime_directory,
         'zfs_snapshots',
+        # Including this hash prevents conflicts between snapshot patterns for different datasets.
+        # For instance, without this, snapshotting a dataset at /var and another at /var/spool would
+        # result in overlapping snapshot patterns and therefore colliding mount attempts.
+        hashlib.shake_256(dataset.mount_point.encode('utf-8')).hexdigest(MOUNT_POINT_HASH_LENGTH),
         '.',  # Borg 1.4+ "slashdot" hack.
         # Included so that the source directory ends up in the Borg archive at its "original" path.
         pattern.path.lstrip('^').lstrip(os.path.sep),
@@ -183,6 +223,7 @@ def make_borg_snapshot_pattern(pattern, normalized_runtime_directory):
         pattern.type,
         pattern.style,
         pattern.device,
+        source=borgmatic.borg.pattern.Pattern_source.HOOK,
     )
 
 
@@ -209,7 +250,8 @@ def dump_data_sources(
     dry_run_label = ' (dry run; not actually snapshotting anything)' if dry_run else ''
     logger.info(f'Snapshotting ZFS datasets{dry_run_label}')
 
-    # List ZFS datasets to get their mount points.
+    # List ZFS datasets to get their mount points, but only consider those patterns that came from
+    # actual user configuration (as opposed to, say, other hooks).
     zfs_command = hook_config.get('zfs_command', 'zfs')
     requested_datasets = get_datasets_to_backup(zfs_command, patterns)
 
@@ -234,6 +276,9 @@ def dump_data_sources(
         snapshot_mount_path = os.path.join(
             normalized_runtime_directory,
             'zfs_snapshots',
+            hashlib.shake_256(dataset.mount_point.encode('utf-8')).hexdigest(
+                MOUNT_POINT_HASH_LENGTH
+            ),
             dataset.mount_point.lstrip(os.path.sep),
         )
 
@@ -249,7 +294,9 @@ def dump_data_sources(
         )
 
         for pattern in dataset.contained_patterns:
-            snapshot_pattern = make_borg_snapshot_pattern(pattern, normalized_runtime_directory)
+            snapshot_pattern = make_borg_snapshot_pattern(
+                pattern, dataset, normalized_runtime_directory
+            )
 
             # Attempt to update the pattern in place, since pattern order matters to Borg.
             try:
@@ -267,6 +314,7 @@ def unmount_snapshot(umount_command, snapshot_mount_path):  # pragma: no cover
     borgmatic.execute.execute_command(
         tuple(umount_command.split(' ')) + (snapshot_mount_path,),
         output_log_level=logging.DEBUG,
+        close_fds=True,
     )
 
 
@@ -282,6 +330,7 @@ def destroy_snapshot(zfs_command, full_snapshot_name):  # pragma: no cover
             full_snapshot_name,
         ),
         output_log_level=logging.DEBUG,
+        close_fds=True,
     )
 
 
@@ -299,7 +348,8 @@ def get_all_snapshots(zfs_command):
             'snapshot',
             '-o',
             'name',
-        )
+        ),
+        close_fds=True,
     )
 
     return tuple(line.rstrip() for line in list_output.splitlines())
@@ -334,6 +384,7 @@ def remove_data_source_dumps(hook_config, config, borgmatic_runtime_directory, d
             os.path.normpath(borgmatic_runtime_directory),
         ),
         'zfs_snapshots',
+        '*',
     )
     logger.debug(f'Looking for snapshots to remove in {snapshots_glob}{dry_run_label}')
     umount_command = hook_config.get('umount_command', 'umount')
@@ -346,7 +397,10 @@ def remove_data_source_dumps(hook_config, config, borgmatic_runtime_directory, d
         # child datasets before the shorter mount point paths of parent datasets.
         for mount_point in reversed(dataset_mount_points):
             snapshot_mount_path = os.path.join(snapshots_directory, mount_point.lstrip(os.path.sep))
-            if not os.path.isdir(snapshot_mount_path):
+
+            # If the snapshot mount path is empty, this is probably just a "shadow" of a nested
+            # dataset and therefore there's nothing to unmount.
+            if not os.path.isdir(snapshot_mount_path) or not os.listdir(snapshot_mount_path):
                 continue
 
             # This might fail if the path is already mounted, but we swallow errors here since we'll
@@ -370,10 +424,10 @@ def remove_data_source_dumps(hook_config, config, borgmatic_runtime_directory, d
                     return
                 except subprocess.CalledProcessError as error:
                     logger.debug(error)
-                    return
+                    continue
 
         if not dry_run:
-            shutil.rmtree(snapshots_directory)
+            shutil.rmtree(snapshot_mount_path, ignore_errors=True)
 
     # Destroy snapshots.
     full_snapshot_names = get_all_snapshots(zfs_command)

borgmatic/hooks/dispatch.py

@@ -3,6 +3,7 @@ import importlib
 import logging
 import pkgutil
 
+import borgmatic.hooks.command
 import borgmatic.hooks.credential
 import borgmatic.hooks.data_source
 import borgmatic.hooks.monitoring

borgmatic/hooks/monitoring/cronhub.py

@@ -11,6 +11,7 @@ MONITOR_STATE_TO_CRONHUB = {
     monitor.State.FINISH: 'finish',
     monitor.State.FAIL: 'fail',
 }
+TIMEOUT_SECONDS = 10
 
 
 def initialize_monitor(
@@ -28,7 +29,7 @@ def ping_monitor(hook_config, config, config_filename, state, monitoring_log_lev
     filename in any log entries. If this is a dry run, then don't actually ping anything.
     '''
     if state not in MONITOR_STATE_TO_CRONHUB:
-        logger.debug(f'Ignoring unsupported monitoring {state.name.lower()} in Cronhub hook')
+        logger.debug(f'Ignoring unsupported monitoring state {state.name.lower()} in Cronhub hook')
         return
 
     dry_run_label = ' (dry run; not actually pinging)' if dry_run else ''
@@ -45,7 +46,7 @@ def ping_monitor(hook_config, config, config_filename, state, monitoring_log_lev
     if not dry_run:
         logging.getLogger('urllib3').setLevel(logging.ERROR)
         try:
-            response = requests.get(ping_url)
+            response = requests.get(ping_url, timeout=TIMEOUT_SECONDS)
             if not response.ok:
                 response.raise_for_status()
         except requests.exceptions.RequestException as error:

borgmatic/hooks/monitoring/cronitor.py

@@ -11,6 +11,7 @@ MONITOR_STATE_TO_CRONITOR = {
     monitor.State.FINISH: 'complete',
     monitor.State.FAIL: 'fail',
 }
+TIMEOUT_SECONDS = 10
 
 
 def initialize_monitor(
@@ -28,7 +29,7 @@ def ping_monitor(hook_config, config, config_filename, state, monitoring_log_lev
     filename in any log entries. If this is a dry run, then don't actually ping anything.
     '''
     if state not in MONITOR_STATE_TO_CRONITOR:
-        logger.debug(f'Ignoring unsupported monitoring {state.name.lower()} in Cronitor hook')
+        logger.debug(f'Ignoring unsupported monitoring state {state.name.lower()} in Cronitor hook')
         return
 
     dry_run_label = ' (dry run; not actually pinging)' if dry_run else ''
@@ -40,7 +41,7 @@ def ping_monitor(hook_config, config, config_filename, state, monitoring_log_lev
     if not dry_run:
         logging.getLogger('urllib3').setLevel(logging.ERROR)
         try:
-            response = requests.get(ping_url)
+            response = requests.get(ping_url, timeout=TIMEOUT_SECONDS)
             if not response.ok:
                 response.raise_for_status()
         except requests.exceptions.RequestException as error:

borgmatic/hooks/monitoring/healthchecks.py

@@ -17,6 +17,7 @@ MONITOR_STATE_TO_HEALTHCHECKS = {
 
 DEFAULT_PING_BODY_LIMIT_BYTES = 100000
 HANDLER_IDENTIFIER = 'healthchecks'
+TIMEOUT_SECONDS = 10
 
 
 def initialize_monitor(hook_config, config, config_filename, monitoring_log_level, dry_run):
@@ -84,7 +85,10 @@ def ping_monitor(hook_config, config, config_filename, state, monitoring_log_lev
         logging.getLogger('urllib3').setLevel(logging.ERROR)
         try:
             response = requests.post(
-                ping_url, data=payload.encode('utf-8'), verify=hook_config.get('verify_tls', True)
+                ping_url,
+                data=payload.encode('utf-8'),
+                verify=hook_config.get('verify_tls', True),
+                timeout=TIMEOUT_SECONDS,
             )
             if not response.ok:
                 response.raise_for_status()

borgmatic/hooks/monitoring/logs.py

@@ -38,11 +38,13 @@ class Forgetful_buffering_handler(logging.Handler):
             self.forgot = True
 
 
-def add_handler(handler):  # pragma: no cover
+def add_handler(handler):
     '''
     Add the given handler to the global logger.
     '''
-    logging.getLogger().addHandler(handler)
+    logger = logging.getLogger()
+    logger.addHandler(handler)
+    logger.setLevel(min(handler.level for handler in logger.handlers))
 
 
 def get_handler(identifier):
@@ -64,7 +66,7 @@ def get_handler(identifier):
 def format_buffered_logs_for_payload(identifier):
     '''
     Get the handler previously added to the root logger, and slurp buffered logs out of it to
-    send to Healthchecks.
+    send to the monitoring service.
     '''
     try:
         buffering_handler = get_handler(identifier)
@@ -90,3 +92,5 @@ def remove_handler(identifier):
         logger.removeHandler(get_handler(identifier))
     except ValueError:
         pass
+
+    logger.setLevel(min(handler.level for handler in logger.handlers))

borgmatic/hooks/monitoring/loki.py

@@ -15,6 +15,7 @@ MONITOR_STATE_TO_LOKI = {
     monitor.State.FINISH: 'Finished',
     monitor.State.FAIL: 'Failed',
 }
+TIMEOUT_SECONDS = 10
 
 # Threshold at which logs get flushed to loki
 MAX_BUFFER_LINES = 100
@@ -69,7 +70,9 @@ class Loki_log_buffer:
         request_header = {'Content-Type': 'application/json'}
 
         try:
-            result = requests.post(self.url, headers=request_header, data=request_body, timeout=5)
+            result = requests.post(
+                self.url, headers=request_header, data=request_body, timeout=TIMEOUT_SECONDS
+            )
             result.raise_for_status()
         except requests.RequestException:
             logger.warning('Failed to upload logs to loki')

borgmatic/hooks/monitoring/ntfy.py

@@ -2,9 +2,14 @@ import logging
 
 import requests
 
+import borgmatic.hooks.credential.parse
+
 logger = logging.getLogger(__name__)
 
 
+TIMEOUT_SECONDS = 10
+
+
 def initialize_monitor(
     ping_url, config, config_filename, monitoring_log_level, dry_run
 ):  # pragma: no cover
@@ -47,9 +52,20 @@ def ping_monitor(hook_config, config, config_filename, state, monitoring_log_lev
             'X-Tags': state_config.get('tags'),
         }
 
-        username = hook_config.get('username')
-        password = hook_config.get('password')
-        access_token = hook_config.get('access_token')
+        try:
+            username = borgmatic.hooks.credential.parse.resolve_credential(
+                hook_config.get('username'), config
+            )
+            password = borgmatic.hooks.credential.parse.resolve_credential(
+                hook_config.get('password'), config
+            )
+            access_token = borgmatic.hooks.credential.parse.resolve_credential(
+                hook_config.get('access_token'), config
+            )
+        except ValueError as error:
+            logger.warning(f'Ntfy credential error: {error}')
+            return
+
         auth = None
 
         if access_token is not None:
@@ -69,7 +85,9 @@ def ping_monitor(hook_config, config, config_filename, state, monitoring_log_lev
         if not dry_run:
             logging.getLogger('urllib3').setLevel(logging.ERROR)
             try:
-                response = requests.post(f'{base_url}/{topic}', headers=headers, auth=auth)
+                response = requests.post(
+                    f'{base_url}/{topic}', headers=headers, auth=auth, timeout=TIMEOUT_SECONDS
+                )
                 if not response.ok:
                     response.raise_for_status()
             except requests.exceptions.RequestException as error:

borgmatic/hooks/monitoring/pagerduty.py

@@ -5,20 +5,38 @@ import platform
 
 import requests
 
+import borgmatic.hooks.credential.parse
+import borgmatic.hooks.monitoring.logs
 from borgmatic.hooks.monitoring import monitor
 
 logger = logging.getLogger(__name__)
 
 EVENTS_API_URL = 'https://events.pagerduty.com/v2/enqueue'
+DEFAULT_LOGS_PAYLOAD_LIMIT_BYTES = 10000
+HANDLER_IDENTIFIER = 'pagerduty'
+TIMEOUT_SECONDS = 10
 
 
-def initialize_monitor(
-    integration_key, config, config_filename, monitoring_log_level, dry_run
-):  # pragma: no cover
+def initialize_monitor(hook_config, config, config_filename, monitoring_log_level, dry_run):
     '''
-    No initialization is necessary for this monitor.
+    Add a handler to the root logger that stores in memory the most recent logs emitted. That way,
+    we can send them all to PagerDuty upon a failure state. But skip this if the "send_logs" option
+    is false.
     '''
-    pass
+    if hook_config.get('send_logs') is False:
+        return
+
+    ping_body_limit = max(
+        DEFAULT_LOGS_PAYLOAD_LIMIT_BYTES
+        - len(borgmatic.hooks.monitoring.logs.PAYLOAD_TRUNCATION_INDICATOR),
+        0,
+    )
+
+    borgmatic.hooks.monitoring.logs.add_handler(
+        borgmatic.hooks.monitoring.logs.Forgetful_buffering_handler(
+            HANDLER_IDENTIFIER, ping_body_limit, monitoring_log_level
+        )
+    )
 
 
 def ping_monitor(hook_config, config, config_filename, state, monitoring_log_level, dry_run):
@@ -29,21 +47,30 @@ def ping_monitor(hook_config, config, config_filename, state, monitoring_log_lev
     '''
     if state != monitor.State.FAIL:
         logger.debug(
-            f'Ignoring unsupported monitoring {state.name.lower()} in PagerDuty hook',
+            f'Ignoring unsupported monitoring state {state.name.lower()} in PagerDuty hook',
         )
         return
 
     dry_run_label = ' (dry run; not actually sending)' if dry_run else ''
     logger.info(f'Sending failure event to PagerDuty {dry_run_label}')
 
-    if dry_run:
+    try:
+        integration_key = borgmatic.hooks.credential.parse.resolve_credential(
+            hook_config.get('integration_key'), config
+        )
+    except ValueError as error:
+        logger.warning(f'PagerDuty credential error: {error}')
         return
 
+    logs_payload = borgmatic.hooks.monitoring.logs.format_buffered_logs_for_payload(
+        HANDLER_IDENTIFIER
+    )
+
     hostname = platform.node()
     local_timestamp = datetime.datetime.now(datetime.timezone.utc).astimezone().isoformat()
     payload = json.dumps(
         {
-            'routing_key': hook_config['integration_key'],
+            'routing_key': integration_key,
             'event_action': 'trigger',
             'payload': {
                 'summary': f'backup failed on {hostname}',
@@ -57,15 +84,20 @@ def ping_monitor(hook_config, config, config_filename, state, monitoring_log_lev
                     'hostname': hostname,
                     'configuration filename': config_filename,
                     'server time': local_timestamp,
+                    'logs': logs_payload,
                 },
             },
         }
     )
-    logger.debug(f'Using PagerDuty payload: {payload}')
+
+    if dry_run:
+        return
 
     logging.getLogger('urllib3').setLevel(logging.ERROR)
     try:
-        response = requests.post(EVENTS_API_URL, data=payload.encode('utf-8'))
+        response = requests.post(
+            EVENTS_API_URL, data=payload.encode('utf-8'), timeout=TIMEOUT_SECONDS
+        )
         if not response.ok:
             response.raise_for_status()
     except requests.exceptions.RequestException as error:
@@ -74,6 +106,7 @@ def ping_monitor(hook_config, config, config_filename, state, monitoring_log_lev
 
 def destroy_monitor(ping_url_or_uuid, config, monitoring_log_level, dry_run):  # pragma: no cover
     '''
-    No destruction is necessary for this monitor.
+    Remove the monitor handler that was added to the root logger. This prevents the handler from
+    getting reused by other instances of this monitor.
     '''
-    pass
+    borgmatic.hooks.monitoring.logs.remove_handler(HANDLER_IDENTIFIER)

borgmatic/hooks/monitoring/pushover.py

@@ -2,10 +2,13 @@ import logging
 
 import requests
 
+import borgmatic.hooks.credential.parse
+
 logger = logging.getLogger(__name__)
 
 
 EMERGENCY_PRIORITY = 2
+TIMEOUT_SECONDS = 10
 
 
 def initialize_monitor(
@@ -32,8 +35,14 @@ def ping_monitor(hook_config, config, config_filename, state, monitoring_log_lev
 
     state_config = hook_config.get(state.name.lower(), {})
 
-    token = hook_config.get('token')
-    user = hook_config.get('user')
+    try:
+        token = borgmatic.hooks.credential.parse.resolve_credential(
+            hook_config.get('token'), config
+        )
+        user = borgmatic.hooks.credential.parse.resolve_credential(hook_config.get('user'), config)
+    except ValueError as error:
+        logger.warning(f'Pushover credential error: {error}')
+        return
 
     logger.info(f'Updating Pushover{dry_run_label}')
 
@@ -71,6 +80,7 @@ def ping_monitor(hook_config, config, config_filename, state, monitoring_log_lev
                 'https://api.pushover.net/1/messages.json',
                 headers={'Content-type': 'application/x-www-form-urlencoded'},
                 data=data,
+                timeout=TIMEOUT_SECONDS,
             )
             if not response.ok:
                 response.raise_for_status()

borgmatic/hooks/monitoring/sentry.py

@@ -6,6 +6,9 @@ import requests
 logger = logging.getLogger(__name__)
 
 
+TIMEOUT_SECONDS = 10
+
+
 def initialize_monitor(
     ping_url, config, config_filename, monitoring_log_level, dry_run
 ):  # pragma: no cover
@@ -61,7 +64,7 @@ def ping_monitor(hook_config, config, config_filename, state, monitoring_log_lev
 
     logging.getLogger('urllib3').setLevel(logging.ERROR)
     try:
-        response = requests.post(f'{cron_url}?status={status}')
+        response = requests.post(f'{cron_url}?status={status}', timeout=TIMEOUT_SECONDS)
         if not response.ok:
             response.raise_for_status()
     except requests.exceptions.RequestException as error:

borgmatic/hooks/monitoring/uptime_kuma.py

@@ -5,6 +5,9 @@ import requests
 logger = logging.getLogger(__name__)
 
 
+TIMEOUT_SECONDS = 10
+
+
 def initialize_monitor(
     push_url, config, config_filename, monitoring_log_level, dry_run
 ):  # pragma: no cover
@@ -37,7 +40,11 @@ def ping_monitor(hook_config, config, config_filename, state, monitoring_log_lev
     logging.getLogger('urllib3').setLevel(logging.ERROR)
 
     try:
-        response = requests.get(f'{push_url}?{query}')
+        response = requests.get(
+            f'{push_url}?{query}',
+            verify=hook_config.get('verify_tls', True),
+            timeout=TIMEOUT_SECONDS,
+        )
         if not response.ok:
             response.raise_for_status()
     except requests.exceptions.RequestException as error:

borgmatic/hooks/monitoring/zabbix.py

@@ -2,9 +2,14 @@ import logging
 
 import requests
 
+import borgmatic.hooks.credential.parse
+
 logger = logging.getLogger(__name__)
 
 
+TIMEOUT_SECONDS = 10
+
+
 def initialize_monitor(
     ping_url, config, config_filename, monitoring_log_level, dry_run
 ):  # pragma: no cover
@@ -14,6 +19,42 @@ def initialize_monitor(
     pass
 
 
+def send_zabbix_request(server, headers, data):
+    '''
+    Given a Zabbix server URL, HTTP headers as a dict, and valid Zabbix JSON payload data as a dict,
+    send a request to the Zabbix server via API.
+
+    Return the response "result" value or None.
+    '''
+    logging.getLogger('urllib3').setLevel(logging.ERROR)
+
+    logger.debug(f'Sending a "{data["method"]}" request to the Zabbix server')
+
+    try:
+        response = requests.post(server, headers=headers, json=data, timeout=TIMEOUT_SECONDS)
+
+        if not response.ok:
+            response.raise_for_status()
+    except requests.exceptions.RequestException as error:
+        logger.warning(f'Zabbix error: {error}')
+
+        return None
+
+    try:
+        result = response.json().get('result')
+        error_message = result['data'][0]['error']
+    except requests.exceptions.JSONDecodeError:
+        logger.warning('Zabbix error: Cannot parse API response')
+
+        return None
+    except (TypeError, KeyError, IndexError):
+        return result
+    else:
+        logger.warning(f'Zabbix error: {error_message}')
+
+        return None
+
+
 def ping_monitor(hook_config, config, config_filename, state, monitoring_log_level, dry_run):
     '''
     Update the configured Zabbix item using either the itemid, or a host and key.
@@ -34,23 +75,31 @@ def ping_monitor(hook_config, config, config_filename, state, monitoring_log_lev
         },
     )
 
+    try:
+        username = borgmatic.hooks.credential.parse.resolve_credential(
+            hook_config.get('username'), config
+        )
+        password = borgmatic.hooks.credential.parse.resolve_credential(
+            hook_config.get('password'), config
+        )
+        api_key = borgmatic.hooks.credential.parse.resolve_credential(
+            hook_config.get('api_key'), config
+        )
+    except ValueError as error:
+        logger.warning(f'Zabbix credential error: {error}')
+
+        return
+
     server = hook_config.get('server')
-    username = hook_config.get('username')
-    password = hook_config.get('password')
-    api_key = hook_config.get('api_key')
     itemid = hook_config.get('itemid')
     host = hook_config.get('host')
     key = hook_config.get('key')
     value = state_config.get('value')
     headers = {'Content-Type': 'application/json-rpc'}
 
-    logger.info(f'Updating Zabbix{dry_run_label}')
+    logger.info(f'Pinging Zabbix{dry_run_label}')
     logger.debug(f'Using Zabbix URL: {server}')
 
-    if server is None:
-        logger.warning('Server missing for Zabbix')
-        return
-
     # Determine the Zabbix method used to store the value: itemid or host/key
     if itemid is not None:
         logger.info(f'Updating {itemid} on Zabbix')
@@ -61,8 +110,8 @@ def ping_monitor(hook_config, config, config_filename, state, monitoring_log_lev
             'id': 1,
         }
 
-    elif (host and key) is not None:
-        logger.info(f'Updating Host:{host} and Key:{key} on Zabbix')
+    elif host is not None and key is not None:
+        logger.info(f'Updating Host: "{host}" and Key: "{key}" on Zabbix')
         data = {
             'jsonrpc': '2.0',
             'method': 'history.push',
@@ -72,58 +121,63 @@ def ping_monitor(hook_config, config, config_filename, state, monitoring_log_lev
 
     elif host is not None:
         logger.warning('Key missing for Zabbix')
-        return
 
+        return
     elif key is not None:
         logger.warning('Host missing for Zabbix')
+
         return
     else:
         logger.warning('No Zabbix itemid or host/key provided')
+
         return
 
     # Determine the authentication method: API key or username/password
     if api_key is not None:
         logger.info('Using API key auth for Zabbix')
-        headers['Authorization'] = 'Bearer ' + api_key
-
-    elif (username and password) is not None:
-        logger.info('Using user/pass auth with user {username} for Zabbix')
-        auth_data = {
+        headers['Authorization'] = f'Bearer {api_key}'
+    elif username is not None and password is not None:
+        logger.info(f'Using user/pass auth with user {username} for Zabbix')
+        login_data = {
             'jsonrpc': '2.0',
             'method': 'user.login',
             'params': {'username': username, 'password': password},
             'id': 1,
         }
+
         if not dry_run:
-            logging.getLogger('urllib3').setLevel(logging.ERROR)
-            try:
-                response = requests.post(server, headers=headers, json=auth_data)
-                data['auth'] = response.json().get('result')
-                if not response.ok:
-                    response.raise_for_status()
-            except requests.exceptions.RequestException as error:
-                logger.warning(f'Zabbix error: {error}')
+            result = send_zabbix_request(server, headers, login_data)
+
+            if not result:
                 return
 
+            headers['Authorization'] = f'Bearer {result}'
     elif username is not None:
         logger.warning('Password missing for Zabbix authentication')
-        return
 
+        return
     elif password is not None:
         logger.warning('Username missing for Zabbix authentication')
+
         return
     else:
         logger.warning('Authentication data missing for Zabbix')
+
         return
 
     if not dry_run:
-        logging.getLogger('urllib3').setLevel(logging.ERROR)
-        try:
-            response = requests.post(server, headers=headers, json=data)
-            if not response.ok:
-                response.raise_for_status()
-        except requests.exceptions.RequestException as error:
-            logger.warning(f'Zabbix error: {error}')
+        send_zabbix_request(server, headers, data)
+
+    if username is not None and password is not None:
+        logout_data = {
+            'jsonrpc': '2.0',
+            'method': 'user.logout',
+            'params': [],
+            'id': 1,
+        }
+
+        if not dry_run:
+            send_zabbix_request(server, headers, logout_data)
 
 
 def destroy_monitor(ping_url_or_uuid, config, monitoring_log_level, dry_run):  # pragma: no cover

borgmatic/logger.py

@@ -29,12 +29,13 @@ def interactive_console():
     return sys.stderr.isatty() and os.environ.get('TERM') != 'dumb'
 
 
-def should_do_markup(no_color, configs):
+def should_do_markup(configs, json_enabled):
     '''
-    Given the value of the command-line no-color argument, and a dict of configuration filename to
-    corresponding parsed configuration, determine if we should enable color marking up.
+    Given a dict of configuration filename to corresponding parsed configuration (which already have
+    any command-line overrides applied) and whether json is enabled, determine if we should enable
+    color marking up.
     '''
-    if no_color:
+    if json_enabled:
         return False
 
     if any(config.get('color', True) is False for config in configs.values()):
@@ -87,11 +88,16 @@ class Multi_stream_handler(logging.Handler):
             handler.setLevel(level)
 
 
-class Console_no_color_formatter(logging.Formatter):
-    def __init__(self, *args, **kwargs):  # pragma: no cover
-        super(Console_no_color_formatter, self).__init__(
-            '{prefix}{message}', style='{', defaults={'prefix': ''}, *args, **kwargs
-        )
+class Log_prefix_formatter(logging.Formatter):
+    def __init__(self, fmt='{prefix}{message}', style='{', *args, **kwargs):  # pragma: no cover
+        self.prefix = None
+
+        super(Log_prefix_formatter, self).__init__(fmt=fmt, style=style, *args, **kwargs)
+
+    def format(self, record):  # pragma: no cover
+        record.prefix = f'{self.prefix}: ' if self.prefix else ''
+
+        return super(Log_prefix_formatter, self).format(record)
 
 
 class Color(enum.Enum):
@@ -105,8 +111,9 @@ class Color(enum.Enum):
 
 class Console_color_formatter(logging.Formatter):
     def __init__(self, *args, **kwargs):
+        self.prefix = None
         super(Console_color_formatter, self).__init__(
-            '{prefix}{message}', style='{', defaults={'prefix': ''}, *args, **kwargs
+            '{prefix}{message}', style='{', *args, **kwargs
         )
 
     def format(self, record):
@@ -124,6 +131,7 @@ class Console_color_formatter(logging.Formatter):
             .get(record.levelno)
             .value
         )
+        record.prefix = f'{self.prefix}: ' if self.prefix else ''
 
         return color_text(color, super(Console_color_formatter, self).format(record))
 
@@ -188,28 +196,32 @@ def add_custom_log_levels():  # pragma: no cover
 
 def get_log_prefix():
     '''
-    Return the current log prefix from the defaults for the formatter on the first logging handler,
-    set by set_log_prefix(). Return None if no such prefix exists.
+    Return the current log prefix set by set_log_prefix(). Return None if no such prefix exists.
+
+    It would be a whole lot easier to use logger.Formatter(defaults=...) instead, but that argument
+    doesn't exist until Python 3.10+.
     '''
     try:
-        return next(
-            handler.formatter._style._defaults.get('prefix').rstrip().rstrip(':')
+        formatter = next(
+            handler.formatter
             for handler in logging.getLogger().handlers
+            if handler.formatter
+            if hasattr(handler.formatter, 'prefix')
         )
-    except (StopIteration, AttributeError):
+    except StopIteration:
         return None
 
+    return formatter.prefix
+
 
 def set_log_prefix(prefix):
     '''
-    Given a log prefix as a string, set it into the defaults for the formatters on all logging
-    handlers. Note that this overwrites any existing defaults.
+    Given a log prefix as a string, set it into the each handler's formatter so that it can inject
+    the prefix into each logged record.
     '''
     for handler in logging.getLogger().handlers:
-        try:
-            handler.formatter._style._defaults = {'prefix': f'{prefix}: ' if prefix else ''}
-        except AttributeError:
-            pass
+        if handler.formatter and hasattr(handler.formatter, 'prefix'):
+            handler.formatter.prefix = prefix
 
 
 class Log_prefix:
@@ -245,13 +257,74 @@ class Log_prefix:
         self.original_prefix = get_log_prefix()
         set_log_prefix(self.prefix)
 
-    def __exit__(self, exception, value, traceback):
+    def __exit__(self, exception_type, exception, traceback):
         '''
         Restore any original prefix.
         '''
         set_log_prefix(self.original_prefix)
 
 
+class Delayed_logging_handler(logging.handlers.BufferingHandler):
+    '''
+    A logging handler that buffers logs and doesn't flush them until explicitly flushed (after
+    target handlers are actually set). It's useful for holding onto messages logged before logging
+    is configured, ensuring those records eventually make their way to the relevant logging
+    handlers.
+
+    When flushing, don't forward log records to a target handler if the record's log level is below
+    that of the handler. This recreates the standard logging behavior of, say, logging.DEBUG records
+    getting suppressed if a handler's level is only set to logging.INFO.
+    '''
+
+    def __init__(self):
+        super(Delayed_logging_handler, self).__init__(capacity=0)
+
+        self.targets = None
+
+    def shouldFlush(self, record):
+        return self.targets is not None
+
+    def flush(self):
+        self.acquire()
+
+        try:
+            if not self.targets:
+                return
+
+            for record in self.buffer:
+                for target in self.targets:
+                    if record.levelno >= target.level:
+                        target.handle(record)
+
+            self.buffer.clear()
+        finally:
+            self.release()
+
+
+def configure_delayed_logging():  # pragma: no cover
+    '''
+    Configure a delayed logging handler to buffer anything that gets logged until we're ready to
+    deal with it.
+    '''
+    logging.basicConfig(
+        level=logging.DEBUG,
+        handlers=[Delayed_logging_handler()],
+    )
+
+
+def flush_delayed_logging(target_handlers):
+    '''
+    Flush any previously buffered logs to our "real" logging handlers.
+    '''
+    root_logger = logging.getLogger()
+
+    if root_logger.handlers and isinstance(root_logger.handlers[0], Delayed_logging_handler):
+        delayed_handler = root_logger.handlers[0]
+        delayed_handler.targets = target_handlers
+        delayed_handler.flush()
+        root_logger.removeHandler(delayed_handler)
+
+
 def configure_logging(
     console_log_level,
     syslog_log_level=None,
@@ -296,10 +369,9 @@ def configure_logging(
     if color_enabled:
         console_handler.setFormatter(Console_color_formatter())
     else:
-        console_handler.setFormatter(Console_no_color_formatter())
+        console_handler.setFormatter(Log_prefix_formatter())
 
     console_handler.setLevel(console_log_level)
-
     handlers = [console_handler]
 
     if syslog_log_level != logging.DISABLED:
@@ -315,10 +387,8 @@ def configure_logging(
         if syslog_path:
             syslog_handler = logging.handlers.SysLogHandler(address=syslog_path)
             syslog_handler.setFormatter(
-                logging.Formatter(
+                Log_prefix_formatter(
                     'borgmatic: {levelname} {prefix}{message}',  # noqa: FS003
-                    style='{',
-                    defaults={'prefix': ''},
                 )
             )
             syslog_handler.setLevel(syslog_log_level)
@@ -327,15 +397,15 @@ def configure_logging(
     if log_file and log_file_log_level != logging.DISABLED:
         file_handler = logging.handlers.WatchedFileHandler(log_file)
         file_handler.setFormatter(
-            logging.Formatter(
+            Log_prefix_formatter(
                 log_file_format or '[{asctime}] {levelname}: {prefix}{message}',  # noqa: FS003
-                style='{',
-                defaults={'prefix': ''},
             )
         )
         file_handler.setLevel(log_file_log_level)
         handlers.append(file_handler)
 
+    flush_delayed_logging(handlers)
+
     logging.basicConfig(
         level=min(handler.level for handler in handlers),
         handlers=handlers,

borgmatic/signals.py

@@ -24,6 +24,9 @@ def handle_signal(signal_number, frame):
         logger.critical('Exiting due to TERM signal')
         sys.exit(EXIT_CODE_FROM_SIGNAL + signal.SIGTERM)
     elif signal_number == signal.SIGINT:
+        # Borg doesn't always exit on a SIGINT, so give it a little encouragement.
+        os.killpg(os.getpgrp(), signal.SIGTERM)
+
         raise KeyboardInterrupt()
 
 

borgmatic/verbosity.py

@@ -22,3 +22,28 @@ def verbosity_to_log_level(verbosity):
         VERBOSITY_SOME: logging.INFO,
         VERBOSITY_LOTS: logging.DEBUG,
     }.get(verbosity, logging.WARNING)
+
+
+DEFAULT_VERBOSITIES = {
+    'verbosity': 0,
+    'syslog_verbosity': -2,
+    'log_file_verbosity': 1,
+    'monitoring_verbosity': 1,
+}
+
+
+def get_verbosity(configs, option_name):
+    '''
+    Given a dict from configuration filename to configuration dict, and the name of a configuration
+    verbosity option, return the maximum verbosity value from that option across the given
+    configuration files.
+    '''
+    try:
+        return max(
+            verbosity
+            for config in configs.values()
+            for verbosity in (config.get(option_name, DEFAULT_VERBOSITIES[option_name]),)
+            if verbosity is not None
+        )
+    except ValueError:
+        return DEFAULT_VERBOSITIES[option_name]

docs/Dockerfile

@@ -4,7 +4,7 @@ COPY . /app
 RUN apk add --no-cache py3-pip py3-ruamel.yaml py3-ruamel.yaml.clib
 RUN pip install --break-system-packages --no-cache /app && borgmatic config generate && chmod +r /etc/borgmatic/config.yaml
 RUN borgmatic --help > /command-line.txt \
-    && for action in repo-create transfer create prune compact check delete extract config "config bootstrap" "config generate" "config validate" export-tar mount umount repo-delete restore repo-list list repo-info info break-lock "key export" "key change-passphrase" borg; do \
+    && for action in repo-create transfer create prune compact check delete extract config "config bootstrap" "config generate" "config validate" export-tar mount umount repo-delete restore repo-list list repo-info info break-lock "key export" "key import" "key change-passphrase" recreate borg; do \
            echo -e "\n--------------------------------------------------------------------------------\n" >> /command-line.txt \
            && borgmatic $action --help >> /command-line.txt; done
 RUN /app/docs/fetch-contributors >> /contributors.html

docs/_includes/index.css

@@ -165,6 +165,7 @@ ul {
 }
 li {
 	padding: .25em 0;
+        line-height: 1.5;
 }
 li ul {
         list-style-type: disc;

docs/_includes/layouts/base.njk

@@ -4,7 +4,7 @@
 		<meta charset="utf-8">
 		<meta name="viewport" content="width=device-width, initial-scale=1.0">
                 <meta name="generator" content="{{ eleventy.generator }}">
-		<link rel="icon" href="docs/static/borgmatic.png" type="image/x-icon">
+		<link rel="icon" href="https://torsion.org/borgmatic/docs/static/borgmatic.png" type="image/x-icon">
 		<title>{{ subtitle + ' - ' if subtitle}}{{ title }}</title>
 {%- set css %}
 {% include 'index.css' %}

docs/fetch-contributors

@@ -26,8 +26,7 @@ def list_merged_pulls(url):
 
 
 def list_contributing_issues(url):
-    # labels = bug, design finalized, etc.
-    response = requests.get(f'{url}?labels=19,20,22,23,32,52,53,54', headers={'Accept': 'application/json', 'Content-Type': 'application/json'})
+    response = requests.get(url, headers={'Accept': 'application/json', 'Content-Type': 'application/json'})
 
     if not response.ok:
         response.raise_for_status()
@@ -39,7 +38,7 @@ PULLS_API_ENDPOINT_URLS = (
     'https://projects.torsion.org/api/v1/repos/borgmatic-collective/borgmatic/pulls',
     'https://api.github.com/repos/borgmatic-collective/borgmatic/pulls',
 )
-ISSUES_API_ENDPOINT_URL = 'https://projects.torsion.org/api/v1/repos/borgmatic-collective/borgmatic/issues'
+ISSUES_API_ENDPOINT_URL = 'https://projects.torsion.org/api/v1/repos/borgmatic-collective/borgmatic/issues?state=all'
 RECENT_CONTRIBUTORS_CUTOFF_DAYS = 365
 
 

docs/how-to/add-preparation-and-cleanup-steps-to-backups.md

@@ -7,18 +7,148 @@ eleventyNavigation:
 ---
 ## Preparation and cleanup hooks
 
-If you find yourself performing preparation tasks before your backup runs, or
-cleanup work afterwards, borgmatic hooks may be of interest. Hooks are shell
-commands that borgmatic executes for you at various points as it runs, and
-they're configured in the `hooks` section of your configuration file. But if
-you're looking to backup a database, it's probably easier to use the [database
-backup
-feature](https://torsion.org/borgmatic/docs/how-to/backup-your-databases/)
-instead.
+If you find yourself performing preparation tasks before your backup runs or
+doing cleanup work afterwards, borgmatic command hooks may be of interest. These
+are custom shell commands you can configure borgmatic to execute at various
+points as it runs.
 
-You can specify `before_backup` hooks to perform preparation steps before
+(But if you're looking to backup a database, it's probably easier to use the
+[database backup
+feature](https://torsion.org/borgmatic/docs/how-to/backup-your-databases/)
+instead.)
+
+<span class="minilink minilink-addedin">New in version 2.0.0</span> Command
+hooks are now configured via a list of `commands:` in your borgmatic
+configuration file. For example:
+
+```yaml
+commands:
+    - before: action
+      when: [create]
+      run:
+          - echo "Before create!"
+    - after: action
+      when:
+          - create
+          - prune
+      run:
+          - echo "After create or prune!"
+    - after: error
+      run:
+          - echo "Something went wrong!"
+```
+
+If you're coming from an older version of borgmatic, there is tooling to help
+you [upgrade your
+configuration](https://torsion.org/borgmatic/docs/how-to/upgrade/#upgrading-your-configuration)
+to this new command hook format.
+
+Note that if a `run:` command contains a special YAML character such as a colon,
+you may need to quote the entire string (or use a [multiline
+string](https://yaml-multiline.info/)) to avoid an error:
+
+```yaml
+commands:
+    - before: action
+      when: [create]
+      run:
+          - "echo Backup: start"
+```
+
+By default, an `after` command hook runs even if an error occurs in the
+corresponding `before` hook or between those two hooks. This allows you to
+perform cleanup steps that correspond to `before` preparation commands—even when
+something goes wrong. You may notice that this is a departure from the way that
+the deprecated `after_*` hooks worked in borgmatic prior to version 2.0.0.
+
+<span class="minilink minilink-addedin">New in version 2.0.3</span> You can
+customize this behavior with the `states` option. For instance, here's an
+example of an `after` hook that only triggers on success and not on error:
+
+```yaml
+commands:
+    - after: action
+      when: [create]
+      states: [finish]
+      run:
+          - echo "After successful create!"
+```
+
+Each command in the `commands:` list has the following options:
+
+ * `before` or `after`: Name for the point in borgmatic's execution that the commands should be run before or after, one of:
+    * `action` runs before each action for each repository. This replaces the deprecated `before_create`, `after_prune`, etc.
+    * `repository` runs before or after all actions for each repository. This replaces the deprecated `before_actions` and `after_actions`.
+    * `configuration` runs before or after all actions and repositories in the current configuration file.
+    * `everything` runs before or after all configuration files. Errors here do not trigger `error` hooks or the `fail` state in monitoring hooks. This replaces the deprecated `before_everything` and `after_everything`.
+    * `error` runs after an error occurs—and it's only available for `after`. This replaces the deprecated `on_error` hook.
+ * `when`: Only trigger the hook when borgmatic is run with particular actions (`create`, `prune`, etc.) listed here. Defaults to running for all actions.
+ * `states`: <span class="minilink minilink-addedin">New in version 2.0.3</span> Only trigger the hook if borgmatic encounters one of the states (execution results) listed here. This state is evaluated only for the scope of the configured `action`, `repository`, etc., rather than for the entire borgmatic run. Only available for `after` hooks. Defaults to running the hook for all states. One or more of:
+    * `finish`: No errors occurred.
+    * `fail`: An error occurred.
+ * `run`: List of one or more shell commands or scripts to run when this command hook is triggered.
+
+When command hooks run, they respect the `working_directory` option if it is
+configured, meaning that the hook commands are run in that directory.
+
+<span class="minilink minilink-addedin">New in version 2.0.4</span>If the exact
+same `everything` command hook is present in multiple configuration files,
+borgmatic only runs it once.
+
+borgmatic's `--repository` flag does not impact which command hooks get run. But
+you can use the `--config` flag to limit the configuration files (and thus
+command hooks) used.
+
+
+### Order of execution
+
+Here's a way of visualizing how all of these command hooks slot into borgmatic's
+execution.
+
+Let's say you've got a borgmatic configuration file with a configured
+repository. And suppose you configure several command hooks and then run
+borgmatic for the `create` and `prune` actions. Here's the order of execution:
+
+ * Run `before: everything` hooks (from all configuration files).
+    * Run `before: configuration` hooks (from the first configuration file).
+        * Run `before: repository` hooks (for the first repository).
+            * Run `before: action` hooks for `create`.
+            * Actually run the `create` action (e.g. `borg create`).
+            * Run `after: action` hooks for `create`.
+            * Run `before: action` hooks for `prune`.
+            * Actually run the `prune` action (e.g. `borg prune`).
+            * Run `after: action` hooks for `prune`.
+        * Run `after: repository` hooks (for the first repository).
+    * Run `after: configuration` hooks (from the first configuration file).
+    * Run `after: error` hooks (if an error occurs).
+ * Run `after: everything` hooks (from all configuration files).
+
+This same order of execution extends to multiple repositories and/or
+configuration files.
+
+Based on the above, you can see the difference between, say, an `after: action`
+hook with `states: [fail]` and an `after: error` hook. The `after: action hook`
+runs immediately after the create action fails for a particular repository—so
+before any subsequent actions for that repository or other repositories even
+have a chance to run. Whereas the `after: error` hook doesn't run until all
+actions for—and repositories in—a configuration file have had a chance to
+execute.
+
+And if there are multiple hooks defined for a particular step (e.g. `before:
+action` for `create`), then those hooks are run in the order they're defined in
+configuration.
+
+
+### Deprecated command hooks
+
+<span class="minilink minilink-addedin">Prior to version 2.0.0</span> The
+command hooks worked a little differently. In these older versions of borgmatic,
+you can specify `before_backup` hooks to perform preparation steps before
 running backups and specify `after_backup` hooks to perform cleanup steps
-afterwards. Here's an example:
+afterwards. These deprecated command hooks still work in version 2.0.0+,
+although see below about a few semantic differences starting in that version.
+
+Here's an example of these deprecated hooks:
 
 ```yaml
 before_backup:
@@ -43,6 +173,15 @@ instance, `before_prune` runs before a `prune` action for a repository, while
 <span class="minilink minilink-addedin">Prior to version 1.8.0</span> Put
 these options in the `hooks:` section of your configuration.
 
+<span class="minilink minilink-addedin">New in version 2.0.0</span> An `after_*`
+command hook runs even if an error occurs in the corresponding `before_*` hook
+or between those two hooks. This allows you to perform cleanup steps that
+correspond to `before_*` preparation commands—even when something goes wrong.
+
+<span class="minilink minilink-addedin">New in version 2.0.0</span> When command
+hooks run, they respect the `working_directory` option if it is configured,
+meaning that the hook commands are run in that directory.
+
 <span class="minilink minilink-addedin">New in version 1.7.0</span> The
 `before_actions` and `after_actions` hooks run before/after all the actions
 (like `create`, `prune`, etc.) for each repository. These hooks are a good
@@ -57,49 +196,13 @@ but not if an error occurs in a previous hook or in the backups themselves.
 (Prior to borgmatic 1.6.0, these hooks instead ran once per configuration file
 rather than once per repository.)
 
-
-## Variable interpolation
-
-The before and after action hooks support interpolating particular runtime
-variables into the hook command. Here's an example that assumes you provide a
-separate shell script:
-
-```yaml
-after_prune:
-    - record-prune.sh "{configuration_filename}" "{repository}"
-```
-
-<span class="minilink minilink-addedin">Prior to version 1.8.0</span> Put
-this option in the `hooks:` section of your configuration.
-
-In this example, when the hook is triggered, borgmatic interpolates runtime
-values into the hook command: the borgmatic configuration filename and the
-paths of the current Borg repository. Here's the full set of supported
-variables you can use here:
-
- * `configuration_filename`: borgmatic configuration filename in which the
-   hook was defined
- * `log_file`
-   <span class="minilink minilink-addedin">New in version 1.7.12</span>:
-   path of the borgmatic log file, only set when the `--log-file` flag is used
- * `repository`: path of the current repository as configured in the current
-   borgmatic configuration file
- * `repository_label` <span class="minilink minilink-addedin">New in version
-   1.8.12</span>: label of the current repository as configured in the current
-   borgmatic configuration file
-
-Note that you can also interpolate in [arbitrary environment
-variables](https://torsion.org/borgmatic/docs/how-to/provide-your-passwords/).
-
-
-## Global hooks
-
 You can also use `before_everything` and `after_everything` hooks to perform
 global setup or cleanup:
 
 ```yaml
 before_everything:
     - set-up-stuff-globally
+
 after_everything:
     - clean-up-stuff-globally
 ```
@@ -117,13 +220,102 @@ but only if there is a `create` action. It runs even if an error occurs during
 a backup or a backup hook, but not if an error occurs during a
 `before_everything` hook.
 
+`on_error` hooks run when an error occurs, but only if there is a `create`,
+`prune`, `compact`, or `check` action. For instance, borgmatic can run
+configurable shell commands to fire off custom error notifications or take other
+actions, so you can get alerted as soon as something goes wrong. Here's a
+not-so-useful example:
 
-## Error hooks
+```yaml
+on_error:
+    - echo "Error while creating a backup or running a backup hook."
+```
 
-borgmatic also runs `on_error` hooks if an error occurs, either when creating
-a backup or running a backup hook. See the [monitoring and alerting
-documentation](https://torsion.org/borgmatic/docs/how-to/monitor-your-backups/)
-for more information.
+<span class="minilink minilink-addedin">Prior to version 1.8.0</span> Put
+this option in the `hooks:` section of your configuration.
+
+The `on_error` hook supports interpolating particular runtime variables into
+the hook command. Here's an example that assumes you provide a separate shell
+script to handle the alerting:
+
+```yaml
+on_error:
+    - send-text-message.sh
+```
+
+borgmatic does not run `on_error` hooks if an error occurs within a
+`before_everything` or `after_everything` hook.
+
+
+## Variable interpolation
+
+The command action hooks support interpolating particular runtime variables into
+the commands that are run. Here's are a couple examples that assume you provide
+separate shell scripts:
+
+```yaml
+commands:
+    - after: action
+      when: [prune]
+      run:
+          - record-prune.sh {configuration_filename} {repository}
+    - after: error
+      when: [create]
+      run:
+          - send-text-message.sh {configuration_filename} {repository}
+```
+
+In this example, when the hook is triggered, borgmatic interpolates runtime
+values into each hook command: the borgmatic configuration filename and the
+paths of the current Borg repository.
+
+Here's the full set of supported variables you can use here:
+
+ * `configuration_filename`: borgmatic configuration filename in which the
+   hook was defined
+ * `log_file`
+   <span class="minilink minilink-addedin">New in version 1.7.12</span>:
+   path of the borgmatic log file, only set when the `--log-file` flag is used
+ * `repository`: path of the current repository as configured in the current
+   borgmatic configuration file, if applicable to the current hook
+ * `repository_label` <span class="minilink minilink-addedin">New in version
+   1.8.12</span>: label of the current repository as configured in the current
+   borgmatic configuration file, if applicable to the current hook
+ * `error`: the error message itself, only applies to `error` hooks
+ * `output`: output of the command that failed, only applies to `error` hooks
+   (may be blank if an error occurred without running a command)
+
+Not all command hooks support all variables. For instance, the `everything` and
+`configuration` hooks don't support repository variables because those hooks
+don't run in the context of a single repository. But the deprecated command
+hooks (`before_backup`, `on_error`, etc.) do generally support variable
+interpolation.
+
+borgmatic automatically escapes these interpolated values to prevent shell
+injection attacks. One implication is that you shouldn't wrap the interpolated
+values in your own quotes, as that will interfere with the quoting performed by
+borgmatic and result in your command receiving incorrect arguments. For
+instance, this won't work:
+
+```yaml
+commands:
+    - after: error
+      run:
+          # Don't do this! It won't work, as the {error} value is already quoted.
+          - send-text-message.sh "Uh oh: {error}"
+```
+
+Do this instead:
+
+```yaml
+commands:
+    - after: error
+      run:
+          - send-text-message.sh {error}
+```
+
+Note that you can also interpolate [arbitrary environment
+variables](https://torsion.org/borgmatic/docs/how-to/provide-your-passwords/).
 
 
 ## Hook output

docs/how-to/backup-to-a-removable-drive-or-an-intermittent-server.md

@@ -29,17 +29,14 @@ concept of "soft failure" come in.
 
 This feature leverages [borgmatic command
 hooks](https://torsion.org/borgmatic/docs/how-to/add-preparation-and-cleanup-steps-to-backups/),
-so first familiarize yourself with them. The idea is that you write a simple
-test in the form of a borgmatic hook to see if backups should proceed or not.
+so familiarize yourself with them first. The idea is that you write a simple
+test in the form of a borgmatic command hook to see if backups should proceed or
+not.
 
 The way the test works is that if any of your hook commands return a special
 exit status of 75, that indicates to borgmatic that it's a temporary failure,
 and borgmatic should skip all subsequent actions for the current repository.
 
-<span class="minilink minilink-addedin">Prior to version 1.9.0</span> Soft
-failures skipped subsequent actions for *all* repositories in the
-configuration file, rather than just for the current repository.
-
 If you return any status besides 75, then it's a standard success or error.
 (Zero is success; anything else other than 75 is an error).
 
@@ -62,33 +59,37 @@ these options in the `location:` section of your configuration.
 <span class="minilink minilink-addedin">Prior to version 1.7.10</span> Omit
 the `path:` portion of the `repositories` list.
 
-Then, write a `before_backup` hook in that same configuration file that uses
-the external `findmnt` utility to see whether the drive is mounted before
-proceeding.
+Then, make a command hook in that same configuration file that uses the external
+`findmnt` utility to see whether the drive is mounted before proceeding.
 
 ```yaml
-before_backup:
+commands:
+    - before: repository
+      run:
+          - findmnt /mnt/removable > /dev/null || exit 75
+```
+
+<span class="minilink minilink-addedin">Prior to version 2.0.0</span> Use the
+deprecated `before_actions` hook instead:
+
+```yaml
+before_actions:
     - findmnt /mnt/removable > /dev/null || exit 75
 ```
 
 <span class="minilink minilink-addedin">Prior to version 1.8.0</span> Put this
 option in the `hooks:` section of your configuration.
 
+<span class="minilink minilink-addedin">Prior to version 1.7.0</span> Use
+`before_create` or similar instead of `before_actions`, which was introduced in
+borgmatic 1.7.0.
+
 What this does is check if the `findmnt` command errors when probing for a
 particular mount point. If it does error, then it returns exit code 75 to
 borgmatic. borgmatic logs the soft failure, skips all further actions for the
 current repository, and proceeds onward to any other repositories and/or
 configuration files you may have.
 
-If you'd prefer not to use a separate configuration file, and you'd rather
-have multiple repositories in a single configuration file, you can make your
-`before_backup` soft failure test [vary by
-repository](https://torsion.org/borgmatic/docs/how-to/add-preparation-and-cleanup-steps-to-backups/#variable-interpolation).
-That might require calling out to a separate script though.
-
-Note that `before_backup` only runs on the `create` action. See below about
-optionally using `before_actions` instead.
-
 You can imagine a similar check for the sometimes-online server case:
 
 ```yaml
@@ -98,50 +99,50 @@ source_directories:
 repositories:
     - path: ssh://me@buddys-server.org/./backup.borg
 
-before_backup:
-    - ping -q -c 1 buddys-server.org > /dev/null || exit 75
+commands:
+    - before: repository
+      run:
+          - ping -q -c 1 buddys-server.org > /dev/null || exit 75
 ```
 
 Or to only run backups if the battery level is high enough:
 
 ```yaml
-before_backup:
-    - is_battery_percent_at_least.sh 25
+commands:
+    - before: repository
+      run:
+          - is_battery_percent_at_least.sh 25
 ```
 
-(Writing the battery script is left as an exercise to the reader.)
-
-<span class="minilink minilink-addedin">New in version 1.7.0</span> The
-`before_actions` and `after_actions` hooks run before/after all the actions
-(like `create`, `prune`, etc.) for each repository. So if you'd like your soft
-failure command hook to run regardless of action, consider using
-`before_actions` instead of `before_backup`.
+Writing the battery script is left as an exercise to the reader.
 
 
 ## Caveats and details
 
 There are some caveats you should be aware of with this feature.
 
- * You'll generally want to put a soft failure command in the `before_backup`
+ * You'll generally want to put a soft failure command in a `before` command
    hook, so as to gate whether the backup action occurs. While a soft failure is
-   also supported in the `after_backup` hook, returning a soft failure there
+   also supported in an `after` command hook, returning a soft failure there
    won't prevent any actions from occurring, because they've already occurred!
-   Similarly, you can return a soft failure from an `on_error` hook, but at
+   Similarly, you can return a soft failure from an `error` command hook, but at
    that point it's too late to prevent the error.
  * Returning a soft failure does prevent further commands in the same hook from
-   executing. So, like a standard error, it is an "early out". Unlike a standard
+   executing. So, like a standard error, it is an "early out." Unlike a standard
    error, borgmatic does not display it in angry red text or consider it a
    failure.
- * Any given soft failure only applies to the a single borgmatic repository
-   (as of borgmatic 1.9.0). So if you have other repositories you don't want
-   soft-failed, then make your soft fail test [vary by
-   repository](https://torsion.org/borgmatic/docs/how-to/add-preparation-and-cleanup-steps-to-backups/#variable-interpolation)—or
-   put anything that you don't want soft-failed (like always-online cloud
-   backups) in separate configuration files from your soft-failing
-   repositories.
+ * <span class="minilink minilink-addedin">New in version 1.9.0</span> Soft
+   failures in `action` or `before_*` command hooks only skip the current
+   repository rather than all repositories in a configuration file.
+ * If you're writing a soft failure script that you want to vary based on the
+   current repository, for instance so you can have multiple repositories in a
+   single configuration file, have a look at [command hook variable
+   interpolation](https://torsion.org/borgmatic/docs/how-to/add-preparation-and-cleanup-steps-to-backups/#variable-interpolation).
+   And there's always still the option of putting anything that you don't want
+   soft-failed (like always-online cloud backups) in separate configuration
+   files from your soft-failing repositories.
  * The soft failure doesn't have to test anything related to a repository. You
-   can even perform a test to make sure that individual source directories are
-   mounted and available. Use your imagination!
- * The soft failure feature also works for before/after hooks for other
-   actions as well. But it is not implemented for `before_everything` or
-   `after_everything`.
+   can even perform a test that individual source directories are mounted and
+   available. Use your imagination!
+ * Soft failures are not currently implemented for `everything`,
+   `before_everything`, or `after_everything` command hooks.

docs/how-to/backup-your-databases.md

@@ -191,16 +191,16 @@ mysql_databases:
       format: sql
 ```
 
-### Containers
+### Database containers
 
-If your database is running within a container and borgmatic is too, no
+If your database server is running within a container and borgmatic is too, no
 problem—configure borgmatic to connect to the container's name on its exposed
 port. For instance:
 
 ```yaml
 postgresql_databases:
     - name: users
-      hostname: your-database-container-name
+      hostname: your-database-server-container-name
       port: 5433
       username: postgres
       password: trustsome1
@@ -209,59 +209,65 @@ postgresql_databases:
 <span class="minilink minilink-addedin">Prior to version 1.8.0</span> Put
 these options in the `hooks:` section of your configuration.
 
+
+#### Database client on the host
+
 But what if borgmatic is running on the host? You can still connect to a
-database container if its ports are properly exposed to the host. For
+database server container if its ports are properly exposed to the host. For
 instance, when running the database container, you can specify `--publish
 127.0.0.1:5433:5432` so that it exposes the container's port 5432 to port 5433
-on the host (only reachable on localhost, in this case). Or the same thing
-with Docker Compose:
+on the host (only reachable on localhost, in this case). Or the same thing with
+Docker Compose:
 
 ```yaml
 services:
-   your-database-container-name:
+   your-database-server-container-name:
        image: postgres
        ports:
            - 127.0.0.1:5433:5432
 ```
 
-And then you can connect to the database from borgmatic running on the host:
+And then you can configure borgmatic running on the host to connect to the
+database:
 
 ```yaml
-hooks:
-    postgresql_databases:
-        - name: users
-          hostname: 127.0.0.1
-          port: 5433
-          username: postgres
-          password: trustsome1
+postgresql_databases:
+    - name: users
+      hostname: 127.0.0.1
+      port: 5433
+      username: postgres
+      password: trustsome1
 ```
 
 Alter the ports in these examples to suit your particular database system.
 
+
+#### Database client in a running container
+
 Normally, borgmatic dumps a database by running a database dump command (e.g.
 `pg_dump`) on the host or wherever borgmatic is running, and this command
-connects to your containerized database via the given `hostname` and `port`.
-But if you don't have any database dump commands installed on your host and
-you'd rather use the commands inside your database container itself, borgmatic
+connects to your containerized database via the given `hostname` and `port`. But
+if you don't have any database dump commands installed on your host and you'd
+rather use the commands inside your running database container itself, borgmatic
 supports that too. For that, configure borgmatic to `exec` into your container
 to run the dump command.
 
 For instance, if using Docker and PostgreSQL, something like this might work:
 
 ```yaml
-hooks:
-    postgresql_databases:
-        - name: users
-          hostname: 127.0.0.1
-          port: 5433
-          username: postgres
-          password: trustsome1
-          pg_dump_command: docker exec my_pg_container pg_dump
+postgresql_databases:
+    - name: users
+      hostname: 127.0.0.1
+      port: 5433
+      username: postgres
+      password: trustsome1
+      pg_dump_command: docker exec my_pg_container pg_dump
 ```
 
-... where `my_pg_container` is the name of your database container. In this
-example, you'd also need to set the `pg_restore_command` and `psql_command`
-options.
+... where `my_pg_container` is the name of your running database container.
+Running `pg_dump` this way takes advantage of the localhost "trust"
+authentication within that container. In this example, you'd also need to set
+the `pg_restore_command` and `psql_command` options.
 
 If you choose to use the `pg_dump` command within the container, and you're
 using the `directory` format in particular, you'll also need to mount the
@@ -280,6 +286,48 @@ services:
       - /run/user/1000:/run/user/1000
 ```
 
+And here's an example of using a MariaDB database client within a running Docker
+container:
+
+```yaml
+mariadb_databases:
+    - name: users
+      hostname: 127.0.0.1
+      username: example
+      password: trustsome1
+      password_transport: environment
+      mariadb_dump_command: docker exec --env MYSQL_PWD my_mariadb_container mariadb-dump
+```
+
+The `password_transport: environment` option tells borgmatic to transmit the
+password via environment variable instead of the default behavior of using an
+anonymous pipe. The environment variable transport is potentially less secure
+than the pipe, but it's necessary when the database client is running in a
+container separate from borgmatic.
+
+A similar approach can work with MySQL, using `mysql_dump_command` instead of
+`mariadb_dump_command` to run `mysqldump` in a container.
+
+
+#### Database client in a temporary container
+
+Another variation: If you're running borgmatic on the host but want to spin up a
+temporary `pg_dump` container whenever borgmatic dumps a database, for
+instance to make use of a `pg_dump` version not present on the host, try
+something like this:
+
+```yaml
+postgresql_databases:
+    - name: users
+      hostname: your-database-hostname
+      username: postgres
+      password: trustsome1
+      pg_dump_command: docker run --rm --env PGPASSWORD postgres:17-alpine pg_dump
+```
+
+The `--env PGPASSWORD` is necessary here for borgmatic to provide your database
+password to the temporary `pg_dump` container.
+
 Similar command override options are available for (some of) the other
 supported database types as well. See the [configuration
 reference](https://torsion.org/borgmatic/docs/reference/configuration/) for
@@ -309,10 +357,8 @@ hooks:
 ### External passwords
 
 If you don't want to keep your database passwords in your borgmatic
-configuration file, you can instead pass them in via [environment
-variables](https://torsion.org/borgmatic/docs/how-to/provide-your-passwords/)
-or command-line [configuration
-overrides](https://torsion.org/borgmatic/docs/how-to/make-per-application-backups/#configuration-overrides).
+configuration file, you can instead pass them in [from external credential
+sources](https://torsion.org/borgmatic/docs/how-to/provide-your-passwords/).
 
 
 ### Configuration backups
@@ -627,9 +673,10 @@ With PostgreSQL, MariaDB, and MySQL, if you're getting authentication errors
 when borgmatic tries to connect to your database, a natural reaction is to
 increase your borgmatic verbosity with `--verbosity 2` and go looking in the
 logs. You'll notice though that your database password does not show up in the
-logs. But this is likely not the cause of the authentication problem unless
-you mistyped your password; borgmatic passes your password to the database via
-an environment variable that does not appear in the logs.
+logs. But this is likely not the cause of the authentication problem unless you
+mistyped your password; borgmatic passes your password to the database via an
+environment variable or anonymous pipe, so the password does not appear in the
+logs.
 
 The cause of an authentication error is often on the database side—in the
 configuration of which users are allowed to connect and how they are

docs/how-to/customize-warnings-and-errors.md

@@ -12,52 +12,18 @@ numeric ID indicating success, warning, or error. borgmatic consumes this exit
 code to decide how to respond. Normally, a Borg error results in a borgmatic
 error, while a Borg warning or success doesn't.
 
-But if that default behavior isn't sufficient for your needs, you can
-customize how borgmatic interprets [Borg's exit
+<span class="minilink minilink-addedin">With Borg version 1.4+</span> If the
+default behavior isn't sufficient for your needs, you can customize how
+borgmatic interprets [Borg's exit
 codes](https://borgbackup.readthedocs.io/en/stable/usage/general.html#return-codes).
-For instance, to elevate Borg warnings to errors, thereby causing borgmatic to
-error on them, use the following borgmatic configuration:
-
-```yaml
-borg_exit_codes:
-   - exit_code: 1
-     treat_as: error
-```
-
-Be aware though that Borg exits with a warning code for a variety of benign
-situations such as files changing while they're being read, so this example
-may not meet your needs. Keep reading though for more granular exit code
-configuration.
-
-Here's an example that squashes Borg errors to warnings:
-
-```yaml
-borg_exit_codes:
-   - exit_code: 2
-     treat_as: warning
-```
-
-Be careful with this example though, because it prevents borgmatic from
-erroring when Borg errors, which may not be desirable.
-
-
-### More granular configuration
-
-<span class="minilink minilink-addedin">New in Borg version 1.4</span> Borg
-support for [more granular exit
-codes](https://borgbackup.readthedocs.io/en/1.4-maint/usage/general.html#return-codes)
-means that you can configure borgmatic to respond to specific Borg conditions.
-See the full list of [Borg 1.4 error and warning exit
-codes](https://borgbackup.readthedocs.io/en/stable/internals/frontends.html#message-ids).
-The `rc:` numeric value there tells you the exit code for each.
 
 For instance, this borgmatic configuration elevates all Borg backup file
 permission warnings (exit code `105`)—and only those warnings—to errors:
 
 ```yaml
 borg_exit_codes:
-   - exit_code: 105
-     treat_as: error
+    - code: 105
+      treat_as: error
 ```
 
 The following configuration does that *and* elevates backup file not found
@@ -65,16 +31,20 @@ warnings (exit code `107`) to errors as well:
 
 ```yaml
 borg_exit_codes:
-   - exit_code: 105
-     treat_as: error
-   - exit_code: 107
-     treat_as: error
+    - code: 105
+      treat_as: error
+    - code: 107
+      treat_as: error
 ```
 
+See the full list of [Borg 1.4 error and warning exit
+codes](https://borgbackup.readthedocs.io/en/stable/internals/frontends.html#message-ids).
+The `rc:` numeric value there tells you the exit code for each.
+
 If you don't know the exit code for a particular Borg error or warning you're
-experiencing, you can usually find it in your borgmatic output when
-`--verbosity 2` is enabled. For instance, here's a snippet of that output when
-a backup file is not found:
+experiencing, you can usually find it in your borgmatic output when `--verbosity
+2` is enabled. For instance, here's a snippet of that output when a backup file
+is not found:
 
 ```
 /noexist: stat: [Errno 2] No such file or directory: '/noexist'
@@ -84,3 +54,30 @@ terminating with warning status, rc 107
 
 So if you want to configure borgmatic to treat this as an error instead of a
 warning, the exit status to use is `107`.
+
+<span class="minilink minilink-addedin">With Borg version 1.2 and earlier</span>
+Older versions of Borg didn't support granular exit codes, but still
+distinguished between Borg errors and warnings. For instance, to elevate Borg
+warnings to errors, thereby causing borgmatic to error on them, use the
+following borgmatic configuration with Borg 1.2 or below:
+
+```yaml
+borg_exit_codes:
+    - code: 1
+      treat_as: error
+```
+
+Be aware though that Borg exits with a warning code for a variety of benign
+situations such as files changing while they're being read, so this example
+may not meet your needs.
+
+Here's another Borg 1.2 example that squashes Borg errors to warnings:
+
+```yaml
+borg_exit_codes:
+    - code: 2
+      treat_as: warning
+```
+
+Be careful with this example though, because it prevents borgmatic from
+erroring when Borg errors, which may not be desirable.