Management commands¶

This project ships four custom Django management commands. They cover the full content pipeline: pulling talks and speakers from Pretalx, attaching livestream URLs from a Google Sheet, backfilling recorded-video links from Vimeo, and seeding a development database with realistic fake data.

Run any command through uv in development:

uv run python manage.py <command> [flags]

In a deployed container the project is on the path, so the uv run prefix is not needed:

python manage.py <command> [flags]

Every command that imports from an external source (import_pretalx_talks, import_livestream_urls, update_video_links) supports --dry-run, which fetches and reports but never commits. Use it to preview a run before letting it touch live data. The exception is generate_fake_talks, which has no --dry-run flag because it only seeds a local development database.

import_pretalx_talks¶

Imports talks and speakers from the Pretalx REST API into the local database. This is the workhorse command and the one you run most often. It is idempotent: a second run against an unchanged Pretalx event makes no database writes and regenerates no social cards.

This page documents the flags and a few quick recipes. For the operator deep dive (sync modes, change detection, event-scoped rooms, image regeneration, the detect-and-review workflow, the email digest, and scheduling), see Pretalx sync.

Purpose¶

• Create new Talk and Speaker rows for Pretalx submissions not yet in the database.
• Update existing rows when a field or speaker association changed, writing only the fields that actually differ so manual local edits to other fields survive.
• Delete talks that disappeared from the Pretalx event.
• Generate (or refresh) the social-card image for each talk.
• Optionally run in detect-only mode that records reviewable diffs instead of mutating live data.

Usage¶

uv run python manage.py import_pretalx_talks \
  --pretalx-event-url https://pretalx.com/my-event/ \
  --event-slug my-event \
  --api-token TOKEN

When --event-slug is omitted it falls back to the DEFAULT_EVENT setting (pyconde-pydata-2026). When --api-token is omitted it uses PRETALX_API_TOKEN. The event URL falls back to the Event.pretalx_url field, then to https://pretalx.com/<event-slug>.

Flags¶

The --detect-only mode and the default mutation mode are opposites: the former records intent, the latter applies it. See the modes table for the full behavior matrix.

Required configuration¶

• PRETALX_API_TOKEN must be set (or passed via --api-token). Required.
• DEFAULT_EVENT is used as the fallback event slug when --event-slug is omitted, and by the admin "Check Pretalx now" button.

Dry-run support¶

Yes. --dry-run logs every create/update/delete it would perform and never writes to the database. --detect-only is a separate safe mode that does write PendingPretalxChange audit rows but leaves the live Talk / Speaker / Room graph untouched.

Example invocations¶

uv run python manage.py import_pretalx_talks \
  --pretalx-event-url https://pretalx.com/pyconde-pydata-2026/ \
  --event-slug pyconde-pydata-2026 \
  --event-name "PyCon DE & PyData 2026"

uv run python manage.py import_pretalx_talks --dry-run

uv run python manage.py import_pretalx_talks \
  --detect-only --event-slug pyconde-pydata-2026

uv run python manage.py import_pretalx_talks --force-images

import_livestream_urls¶

Imports per-room livestream (Vimeo embed) sessions from a Google Sheet that lists each room's stream windows. Each row becomes a Streaming record tied to a Room. The talk dashboard uses these to show "watch live" links during the session window.

Purpose¶

• Read a published Google Sheet (downloaded as xlsx) of room streaming windows.
• Keep only Vimeo rows that have an embed link and a valid start and end time.
• Replace the target event's existing Streaming rows with the ones from the sheet.

Usage¶

uv run python manage.py import_livestream_urls \
  --livestreams-sheet-id SHEET_ID \
  --livestreams-worksheet-name Sheet1

The whole command runs inside a single database transaction, so a failure rolls back cleanly.

Flags¶

Required configuration¶

• LIVESTREAMS_SHEET_ID and LIVESTREAMS_WORKSHEET_NAME (or the matching flags). The sheet must be published / shared so the xlsx export URL is reachable without authentication.
• The rooms named in the sheet must already exist in the target event. A row whose room is not found is skipped with a warning, not created. Rooms are normally created by import_pretalx_talks, so run that first.

Dry-run support¶

Yes. --dry-run prints each row it would process (matched room, start, end) and a count of rows it would skip because their room was not found. No streamings are deleted or created.

Expected sheet columns¶

The cleaner keeps rows where Vimeo / Restream equals Vimeo and the Embed Link is non-empty, then uses these columns:

• Room - matched against an existing Room.name (scoped to the event when one is given).
• Start Time and End Time - parsed as Europe/Berlin local time. Blank or unparseable cells are dropped (with a warning) rather than aborting the whole import.
• Embed Link - stored as the streaming's video link.

Example invocations¶

uv run python manage.py import_livestream_urls \
  --livestreams-sheet-id 1AbC...XyZ \
  --livestreams-worksheet-name "Streams"

uv run python manage.py import_livestream_urls --dry-run

uv run python manage.py import_livestream_urls \
  --event-slug pyconde-pydata-2026

update_video_links¶

Backfills recorded-video links onto talks once the conference recordings are uploaded to Vimeo. It fetches every video in the configured Vimeo project folders and matches each one to a talk by its Pretalx code.

Purpose¶

• List all videos in one or more Vimeo project folders (paginated, 100 per page).
• Match each video to a talk. The match assumes the video name starts with the Pretalx code, followed by a - separator and the title ({pretalx_code}-{title}).
• Set the matched talk's video_link to the Vimeo player embed URL and reset video_start_time to 0.

Usage¶

uv run python manage.py update_video_links \
  --vimeo-access-token TOKEN \
  --vimeo-project-ids 123,456

Runs inside a single transaction.

Flags¶

Required configuration¶

• VIMEO_ACCESS_TOKEN (or --vimeo-access-token).
• VIMEO_PROJECT_IDS (or --vimeo-project-ids) listing the project folders that hold the talk recordings.

Dry-run support¶

Yes. --dry-run fetches the videos and reports how many it found, but does not update any talk.

Example invocations¶

uv run python manage.py update_video_links \
  --vimeo-access-token "$VIMEO_ACCESS_TOKEN" \
  --vimeo-project-ids 123,456

uv run python manage.py update_video_links --dry-run

generate_fake_talks¶

Generates realistic fake talks, speakers, rooms, and streaming sessions for local development and testing. It is run automatically by the dev setup script when GEN_FAKE_DATA=true. The generated schedule deliberately includes a finished talk, a talk happening right now, and an upcoming talk so the dashboard's time-based states are easy to see.

Purpose¶

• Build an event (a synthetic fake-event when no --event-slug is given) with rooms in three categories: plenary, talks, and tutorials.
• Create streaming sessions across the conference days, guaranteeing at least one session covers the current moment.
• Generate a pool of speakers and assign 1-3 of them to each talk.
• Create talks at conflict-free time slots with plausible titles, tracks, durations, and optional Slido/video links.

Usage¶

uv run python manage.py generate_fake_talks

With no flags it creates 100 talks over 3 days starting yesterday, scoped to DEFAULT_EVENT (or the synthetic fake-event when that setting is empty).

Flags¶

The built-in default track list is: MLOps & DevOps, Security, Django & Web, Natural Language Processing, Machine Learning, Data Handling & Engineering, Computer Vision, and Programming & Software Engineering.

Required configuration¶

None. The command needs no API tokens or external services. It only depends on the faker package, which is part of the project dependencies.

Dry-run support¶

No. This command does not have a --dry-run flag because it only writes to a local development database. Use --clear-existing to start from a clean slate, and --seed for reproducible runs.

Example invocations¶

uv run python manage.py generate_fake_talks

uv run python manage.py generate_fake_talks --count 20 --seed 42 --clear-existing

uv run python manage.py generate_fake_talks \
  --date 2026-04-22 --days 4 \
  --rooms-talks "Room A,Room B,Room C"

uv run python manage.py generate_fake_talks \
  --event-slug demo-2026 --event-name "Demo 2026"

See also¶

• Pretalx sync for the full importer operator manual.
• The command sources live under talks/management/commands/.