From cba02bdd0b8d7c10b9d25a3052e60ef9670eccf8 Mon Sep 17 00:00:00 2001
From: Daniel Schadt <kingdread@gmx.de>
Date: Wed, 4 Jan 2023 00:42:33 +0100
Subject: Add fietscron script for maintenance

---
 CHANGELOG.rst                   |  1 +
 doc/administration.rst          |  1 +
 doc/administration/cronjobs.rst | 29 ++++++++++++++++
 fietsboek/scripts/fietscron.py  | 75 +++++++++++++++++++++++++++++++++++++++++
 pyproject.toml                  |  1 +
 5 files changed, 107 insertions(+)
 create mode 100644 doc/administration/cronjobs.rst
 create mode 100644 fietsboek/scripts/fietscron.py

diff --git a/CHANGELOG.rst b/CHANGELOG.rst
index ecf48e4..8668b5c 100644
--- a/CHANGELOG.rst
+++ b/CHANGELOG.rst
@@ -11,6 +11,7 @@ Added
 - The ability to load external (third party) language packs.
 - GPX metadata (track title, description & author) are now embedded into the
   actual GPX file.
+- The ``fietscron`` maintenance script.
 
 Changed
 ^^^^^^^
diff --git a/doc/administration.rst b/doc/administration.rst
index 7aa2393..706a4cb 100644
--- a/doc/administration.rst
+++ b/doc/administration.rst
@@ -9,6 +9,7 @@ Administration Guide
     administration/configuration
     administration/upgrading
     administration/backup
+    administration/cronjobs
     administration/custom-pages
 
 This guide contains information pertaining to the administration of a Fietsboek
diff --git a/doc/administration/cronjobs.rst b/doc/administration/cronjobs.rst
new file mode 100644
index 0000000..a7d1a95
--- /dev/null
+++ b/doc/administration/cronjobs.rst
@@ -0,0 +1,29 @@
+Cronjobs
+========
+
+There are some maintenance jobs that should be done at regular intervals to
+keep the instance running smoothly:
+
+* Old, abandoned uploads should be deleted from the database to free up the
+  space of the temporary GPX files.
+* Tracks that are not present in the cache should have their cache entry
+  populated by pre-calculating their length/uphill/downhill/… values. This
+  makes accessing those values faster when a user requests the track's details.
+
+Fietsboek comes with a script to deal with those jobs: ``fietscron``. You can
+run it like this::
+
+    fietscron -c path_to_config.ini
+
+You can use any mechanism you like to run this command at pre-defined intervals
+(e.g. once an hour). You can find information about this in various sources:
+
+* https://wiki.archlinux.org/title/Systemd/Timers
+* https://wiki.archlinux.org/title/Cron
+* https://wiki.gentoo.org/wiki/Cron
+
+For example, the classic *cron*-way to run ``fietscron`` once per hour would be
+to add the following to your *crontab* (using ``crontab -e``)::
+
+    0 * * * * /path/to/venv/bin/fietscron -c /path/to/settings.ini
+
diff --git a/fietsboek/scripts/fietscron.py b/fietsboek/scripts/fietscron.py
new file mode 100644
index 0000000..3b22ac2
--- /dev/null
+++ b/fietsboek/scripts/fietscron.py
@@ -0,0 +1,75 @@
+"""Script to do maintenance actions for fietsboek."""
+import datetime
+import logging
+import logging.config
+
+import click
+import pyramid.paster
+from sqlalchemy import create_engine, delete, exists, not_, select
+from sqlalchemy.engine import Engine
+from sqlalchemy.orm import Session
+
+from .. import config as mod_config
+from .. import models
+from ..data import DataManager
+
+LOGGER = logging.getLogger(__name__)
+
+
+@click.command()
+@click.option(
+    "-c",
+    "--config",
+    type=click.Path(exists=True, dir_okay=False),
+    required=True,
+    help="Path to the Fietsboek configuration file",
+)
+def cli(config):
+    """Runs regular maintenance operations on the instance.
+
+    This does the following:
+
+    \b
+    * Deletes pending uploads that are older than 24 hours.
+    * Rebuilds the cache for missing tracks.
+    """
+    logging.config.fileConfig(config)
+    settings = pyramid.paster.get_appsettings(config)
+
+    config = mod_config.parse(settings)
+    engine = create_engine(config.sqlalchemy_url)
+
+    data_manager = DataManager(config.data_dir)
+
+    LOGGER.debug("Starting maintenance tasks")
+    remove_old_uploads(engine)
+    rebuild_cache(engine, data_manager)
+
+
+def remove_old_uploads(engine: Engine):
+    """Removes old uploads from the database."""
+    LOGGER.info("Deleting old uploads")
+    limit = datetime.datetime.now() - datetime.timedelta(hours=24)
+    session = Session(engine)
+    stmt = delete(models.Upload).where(models.Upload.uploaded_at < limit)
+    session.execute(stmt)
+    session.commit()
+
+
+def rebuild_cache(engine: Engine, data_manager: DataManager):
+    """Rebuilds the cache entries that are currently missing."""
+    LOGGER.debug("Rebuilding caches")
+    session = Session(engine)
+    needed_rebuilds = select(models.Track).where(
+        not_(exists(select(models.TrackCache).where(models.TrackCache.track_id == models.Track.id)))
+    )
+    with session:
+        for track in session.execute(needed_rebuilds).scalars():
+            LOGGER.info("Rebuilding cache for track %d", track.id)
+            gpx_data = data_manager.open(track.id).decompress_gpx()
+            track.ensure_cache(gpx_data)
+            session.add(track)
+        session.commit()
+
+
+__all__ = ["cli"]
diff --git a/pyproject.toml b/pyproject.toml
index 2ad5db5..01eb607 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -92,6 +92,7 @@ types-redis = "^4.3.21.6"
 
 [tool.poetry.scripts]
 fietsctl = "fietsboek.scripts.fietsctl:main"
+fietscron = "fietsboek.scripts.fietscron:cli"
 fietsupdate = "fietsboek.updater.cli:cli"
 
 [tool.poetry.plugins."paste.app_factory"]
-- 
cgit v1.2.3