[stantheman0128]

Closes #901.

Problem

TSPhrases.txt was missing several phrase-level overrides requested in #901:

1. Longer 乾紅* forms for dry-red wine / dried safflower / dried jujube (干红*), beyond the short 乾紅 -> 干红 乾红 entry from t2s: 乾斷食 → 干断食, 乾紅 → 干红 #1429 / TS 可能的过度转换：乾斷，乾紅 #621.
2. Classical 谿 compounds (豁谿 / 谿刻 / 谿卡 / 谿壑) were simplified to 溪 via TSCharacters, even though these should keep 谿 (related: update TSCharacters.txt #905).
3. Under tw2sp, 列印 -> 打印 is applied by longest prefix match inside longer strings, so:

位列印度 -> 位打印度
一列印度火車 -> 一打印度火车
位列印尼 -> 位打印尼
一列印尼火車 -> 一打印尼火车

4. Personal names with an interpunct after 喬治亞 should stay 乔治亚·... in mainland mode (related: 納米比亞，喬治亞州，喬治亞共和國 轉換問題 #902).

Change

data/dictionary/TSPhrases.txt:

乾紅棗 -> 干红枣
乾紅花 -> 干红花
乾紅葡萄酒 -> 干红葡萄酒
豁谿 / 谿刻 / 谿卡 / 谿壑 -> identity (keep 谿)
列印度 / 列印尼 -> identity
喬治亞· -> 乔治亚·

TWPhrasesRev.txt / TWPhrases.txt (needed for tw2sp / reverse-mapping closure; same MaxMatch pattern as #907):

列印度 -> 列印度
列印尼 -> 列印尼

Ordinary 列印 -> 打印 is unchanged. Dictionaries sorted with data/scripts/sort.py.

Test plan

• New cases BYVoid_OpenCC_Issue_901_* in test/testcases/testcases.json
• Control: 列印 -> 打印 under tw2sp
• Local verification with OpenCC 1.4.1 wheel + branch text dictionaries (see Evidence)
• CI Bazel dictionary tests (authoritative)

Evidence

Repro before (packaged wheel dictionaries, or master text dicts for the tw2sp 列印 case):

t2s    豁谿 / 谿刻 / 谿卡 / 谿壑 -> 豁溪 / 溪刻 / 溪卡 / 溪壑
tw2sp  位列印度 -> 位打印度
tw2sp  一列印度火車 -> 一打印度火车
tw2sp  位列印尼 -> 位打印尼
tw2sp  一列印尼火車 -> 一打印尼火车

After this branch (config JSONs with ocd2 refs switched to text, driving data/dictionary/*.txt on the branch):

t2s/tw2sp  乾紅葡萄酒 -> 干红葡萄酒
t2s/tw2sp  乾紅花 -> 干红花
t2s/tw2sp  乾紅棗 -> 干红枣
t2s/tw2sp  豁谿谿刻谿卡谿壑 -> 豁谿谿刻谿卡谿壑
tw2sp      位列印度一列印度火車 -> 位列印度一列印度火车
tw2sp      位列印尼一列印尼火車 -> 位列印尼一列印尼火车
tw2sp      列印 -> 打印
tw2sp/t2s  喬治亞·馬克 -> 乔治亚·马克

Harness summary (OpenCC 1.4.1 wheel, config ocd2 references rewritten to text):

reverse mapping closure (TWPhrases): MISSING = 0
sort.py idempotence on TSPhrases/TWPhrases/TWPhrasesRev: no further diff
master vs branch on existing t2s/tw2sp expectations: new_fail = 0 (5 same pre-existing text-harness gaps on both)
branch dicts + new #901 cases: 8 passed, 0 failed

Please treat CI Bazel dictionary tests as authoritative.

AI assistance

This fix was developed with AI assistance (Cursor / Grok) and verified locally as shown above.