Fix a mapping

A wrong mapping is wrong for everyone, so a fix helps everyone. Every mapping lives in data/ on GitHub. Most works come from seed datasets; when one is wrong, you add a short file in data/works/ that overrides just the part that is wrong. CI checks it, and it ships in the next weekly data release.

How to contribute

1. Fork the repository on GitHub (the Fork button, top right).
2. Make a branch in your fork.
3. Change one file, usually one short YAML file in data/works/.
4. Open a pull request against courmap/courmap.

The first time you contribute, the automated checks wait for a maintainer to approve them; Xwolv or another maintainer does that, usually soon, and later pull requests from you run straight away. Nothing is wrong if the checks show as waiting.

A one-file data fix needs no clone and no Node install. Open the file under data/works/ on GitHub and click the pencil icon (“Edit this file”); GitHub forks the repository for you and walks you through the pull request. To add a new file, use “Add file” then “Create new file” in the same folder. To fix a typo or a sentence in these docs, use the “Edit page” link at the bottom of any page. Running pnpm validate first (steps below) is still the quickest way to catch a mistake, but CI runs the same checks on your pull request.

1. Find the work

Every work has a permanent id (its slug), derived from its romaji title, such as kusuriya-no-hitorigoto-2nd-season. Find it with the try-it box, or in data/seeds/slugs.jsonl by AniDB, AniList or MAL ID.

2. Write the fix

Create data/works/<id>.yaml with the id and only what changes. (You need Node 22 or later and pnpm; clone the repository first.) For example, if TMDB numbers a sequel as a continuation of season 1:

id: kusuriya-no-hitorigoto-2nd-season
projections:
  tmdb_show:
    - { season: 1, episodes: 1-24, start: 25 }
sources:
  - https://www.themoviedb.org/tv/220542/season/1

This replaces the seed’s TMDB block and keeps everything else from the seed. sources links whatever justifies the mapping.

3. Check it locally

pnpm install
pnpm validate

pnpm validate checks the schema, overlapping and missing episodes, episode counts, and that every episode translates to every provider and back. The error messages name the file, line and fix.

4. Open a pull request

CI runs the same checks and comments with the resolver corpus pass rate. A maintainer reviews it; once merged, it ships in the next weekly release.

That is the whole loop. For the hard cases (split cours, specials, long-runners, release groups with their own numbering) see Contributing mappings, and for the rules every file follows see the data format. Looking for something to fix? Fixing open issues explains each kind of bot-opened issue.