mirror of
https://github.com/bilawalsidhu/gods-eye-view.git
synced 2026-11-18 01:02:23 +00:00
i18n: tighten the zh-Hant check and test its command line
- Without opencc-js the zh-Hant check says to run npm ci, and the other
checks still run.
- --prune handles zh-CN before zh-Hant: shippedLocales() sorts, since
git lists untracked catalogs first.
- A key missing from zh-Hant is reported once, with "run npm run i18n --
--zh-hant" when zh-CN has it; a zh-Hant entry still in English points
to zh-CN and same-as-english.json.
- 並道 -> 併道 applies only to entries whose English key says merge, so
并 (and) before 道路 or 道謝 stays 並.
- Two more corrections for strings main has added: 箇中 -> 個中 after a
count (全部 {n} 個中), and 宣佈 -> 宣布 like 發佈 and 公佈.
- zh-hant-cli.test.mjs runs the command in a throwaway repository, with
opencc-js linked as a junction, so Windows needs no symlink privilege.
zh-Hant.json does not change.
This commit is contained in:
1 parent
3822ee9f00
commit
b63205f69c
4 files changed
+239
-39
No files matched your search
@@ -10,7 +10,7 @@ npm run i18n -- --zh-hant # rewrite zh-Hant.json from zh-CN.json
|
||||
|
||||
`npm test` fails when `zh-Hant.json` differs from the conversion of `zh-CN.json`, naming the entry, and `npm run i18n -- --missing zh-Hant` points to this command instead of printing a template.
|
||||
|
||||
The command converts with OpenCC `s2tw` ([opencc-js](https://github.com/nk2028/opencc-js), `cn` → `tw`), which changes characters only. Never `s2twp`: it also swaps in Taiwan vocabulary (影片, 伺服器, 載入, 圖示), which this catalog does not use. After the conversion it fixes the characters OpenCC picks wrongly here: 年曆史 → 年歷史, 下遊 → 下游, 並道 → 併道 (not in 並道路, 並道歉, 並道理), 發佈 → 發布, 公佈 → 公布, 控制檯 → 控制台, and 周 → 週 where it means a week (1 週/秒). The list is `ZH_HANT_CORRECTIONS` in `scripts/i18n.mjs`; when the conversion picks another wrong character, add the pair there and run the command.
|
||||
The command converts with OpenCC `s2tw` ([opencc-js](https://github.com/nk2028/opencc-js), `cn` → `tw`), which changes characters only. Never `s2twp`: it also swaps in Taiwan vocabulary (影片, 伺服器, 載入, 圖示), which this catalog does not use. After the conversion it fixes the characters OpenCC picks wrongly here: 年曆史 → 年歷史, 下遊 → 下游, 並道 → 併道 (only in entries whose English key says merge), 發佈 → 發布, 公佈 → 公布, 宣佈 → 宣布, 控制檯 → 控制台, 箇中 → 個中 (全部 {n} 個中), and 周 → 週 in ` 周/秒` (1 週/秒, weeks per second). OpenCC writes 週 itself in 每週, 本週, 上週, 一週 and 兩週, but not after a number (過去 2 周, {count} 周前): a new string like that needs its own pair. The list is `ZH_HANT_CORRECTIONS` in `scripts/i18n.mjs`; when the conversion picks another wrong character, add the pair there (with a key pattern if it is right only in some entries) and run the command.
|
||||
|
||||
An entry `zh-CN` keeps in English on purpose needs the same line under `zh-Hant` in `src/i18n/same-as-english.json`.
|
||||
|
||||
|
||||
Reference in new issue
Block a user