Work on improving 1-1 matching accuracy and coverage

[e0406370]

Build on the MVP from #1 to refine 1-1 matching logic and expand coverage by incorporating Filmarks as an additional data source.

[e0406370]

Current limitations:

[#12, FIN] [#21, FIN]

• Title may contain full-width characters instead of half-width characters (e.g. Ｎ vs. N).

  • Querying Ｎのために into the search EP does not return any results while Nのために returns the show as the first result.

• Moreover, some characters may be represented differently on TVer compared to Filmarks in terms of Unicode, so it may not be possible to account for all scenarios in the matching logic.

Solution:

• Normalise the title query as much as possible.
• May need to test if the same issue occurs with series that have full-width non-alphabet characters in their titles.

---

• TVer combines multiple seasons of a show into a single series page (e.g. ドクターX, 相棒, 科捜研の女).
  • The title element displays only the title, while the specific season or year is available only through the season filter list.
  • This means that with the existing logic, the set of score and link displayed on the series page can be inaccurate since it is not possible to distinguish between individual seasons.
  • A similar issue arises with shows that have adaptations sharing the same title but aired in different years.

Solution:

• For ongoing series, extract the year from the broadcastDateLabel in each season's episode data and use it as an additional match identifier.
• Generate multiple score-link pairs in the DOM, each associated with its corresponding year.

---

• Some non-Japanese shows are airing on TVer with Japanese titles, which are unlikely to produce a one-one match with MyDramaList search EP as the shows are listed there under their native titles instead.

Solution:

• Possible to utilise Filmarks here to retrieve the native title from its drama info EP (connect to Filmarks first to get its native title then use it in the MyDramaList search EP).
• Example:

1. Extract クリーニングアップ from title element in TVer series page.
   https://tver.jp/series/sr0qdwkaog
2. Query into Filmarks search EP and extract series id + season id from best match.
   https://filmarks.com/search/dramas?q=クリーニングアップ
3. Insert series id + season id retrieved from previous step into Filmarks dramas EP and extract native title (클리닝 업).
   https://filmarks.com/dramas/12258/16910
4. Query into MyDramaList search EP using the native title retrieved from previous step.
   https://mydramalist.com/search?q=클리닝 업
   ...

• Note that in Filmarks, the element responsible for the native title will not be present if 製作国 is 日本.

---

[#11, FIN]

• The userscript currently targets only series that are listed as "Japanese Drama" on MyDramaList.
  • However, there are also variety shows airing on TVer which are tagged on MDL as "Japanese TV Show".
  • Note that specials (tagged as "Japanese Special" on MDL) are not included as they could interfere with correct matching.

Solution:

• Include "Japanese TV Show" as an additional match identifier.

[e0406370]

Possible improvements:

• [#9, FIN] Implement lightweight caching using localStorage to avoid duplicate lookups and minimise API calls.
• [#9, FIN] Alternatively, implement GM_getValue / GM_setValue custom methods offered by Tampermonkey.
• [#19, FIN] Include a loading spinner in the injected HTML while data is being fetched from the API.
• [#23, FIN] Include a green tick if there is matching data or a red cross if there is no matching data.
• [#21, FIN] Incorporate Filmarks as an additional data source.

[e0406370]

[#9, FIN]

Series data is cached using GM.getValue / GM.setValue custom Tampermonkey methods with weekly TTL imposed.

[e0406370]

[#13, FIN]

Observed that navigating from T => Series (1) => T => Series (2) using the forward and back buttons will display the MDL series data on (1) but not on (2).

[e0406370]

[#14, FIN]

Observed that navigating from Series A => Series B directly using the embedded search bar then navigating back to Series A　creates an additional series data element. The same behaviour replicates upon navigating A <=> B alternately.

[e0406370]

[#21, FIN]

In order to incorporate Filmarks as an additional data source, the search API endpoint for Filmarks needs to be created from scratch as there is no public or non-official API at this time of writing.

EDIT: API created here => https://github.com/e0406370/markuapi