chore(i18n): fill 729 missing zh-TW translation keys

[CtriXin]

Summary

• Fill 149 missing keys from zh-CN.json via Simplified→Traditional Chinese (s2t) conversion
• Fill 580 remaining keys from vi.json (vi locale file contained Chinese values that were never synced to zh-CN/zh-TW)
• zh-TW locale now has 4365 keys vs vi's 4177 — complete vi coverage + 188 extra s2t-converted keys from zh-CN

Changes

• web/src/i18n/locales/zh-TW.json: +729 key-value pairs (1 deletion — likely formatting diff)

Risk

• Low: only adds new translation keys; no existing values modified
• s2t conversion is best-effort (~90% accuracy for common chars; full OpenCC not available in env)

Test plan

• Verify zh-TW.json is valid JSON with no duplicate keys
• Spot-check 3-5 newly added keys in zh-TW locale UI
• No regressions expected (additive change only)

Notes

• 62 pre-existing lint/formatting issues in unrelated JSX files (not introduced by this PR)
• Visual verify: stub — manual check pending for zh-TW UI rendering

Summary by CodeRabbit

• Localization
  • Expanded Traditional Chinese language support for billing tiers, token-based pricing, configuration workflows, cache pricing options, and fee breakdown information.

[coderabbitai]

Walkthrough

Expansion of Traditional Chinese (zh-TW) locale with 730 new translation keys covering token-based billing tiers, cache pricing configurations, integration flows, token estimators, fee breakdowns, webhook settings, and pricing table UI elements. One formatting line removed.

Changes

Cohort / File(s)	Summary
Traditional Chinese Localization web/src/i18n/locales/zh-TW.json	Added 730 translation keys for billing tier numbering, token-count dimensions, conditional charges, cache read/write pricing tiers, Waffo Pancake and easy-pay configurations, token estimators, fee breakdowns, Store/Product IDs, webhook public keys, time/day selectors, and pricing-table elements.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Possibly related PRs

• fix(i18n): fill missing translations in i18n. #2314: Adds extensive pricing/caching and integration translation keys to i18n locale files with overlapping billing and integration entries.
• feat(localization): added zh_TW #2913: Expands Traditional Chinese (zh-TW) localization resources and corresponding i18n wiring infrastructure.
• feat: logs cache field #2920: Introduces cache read/write translation keys and tooltips for usage-log UI rendering.

Suggested reviewers

• Calcium-Ion
• seefs001

Poem

🐰 In the realm of words, we hop and play,
Adding keys for billing's way,
Tiers and tokens, cache so neat,
Chinese speakers now compete,
730 translations, shining bright! ✨

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and specifically summarizes the main change: adding 729 missing Chinese Traditional (zh-TW) translation keys to the i18n locale file.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 3

🧹 Nitpick comments (1)

web/src/i18n/locales/zh-TW.json (1)

1-2: Recommendation: regenerate the file via project i18n tooling instead of ad-hoc imports.

The two systemic problems above (Vietnamese leakage and partial s2t) suggest the upstream pipeline (custom s2t + vi.json import) bypassed the project's standard flow. Per the repo guidelines, prefer:

bun run i18n:extract   # ensure zh-CN.json keys are current
bun run i18n:sync      # propagate missing keys to zh-TW (and other locales)
bun run i18n:lint      # validate

Then run the new zh-TW values through OpenCC with the s2twp.json config (Simplified → Traditional + Taiwan phrase mapping) so terminology (快取, 建立, 影片, 登錄, 儲值, 帳號, …) matches what's already in the file. This will also avoid the broken-plural situation where _one variants land in a different language than their _other counterparts.

As per coding guidelines: "Use CLI tools: bun run i18n:extract, bun run i18n:sync, bun run i18n:lint."

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@web/src/i18n/locales/zh-TW.json` around lines 1 - 2, Regenerate zh-TW via the
project's i18n pipeline instead of hand-editing: run the CLI flow (bun run
i18n:extract, bun run i18n:sync, bun run i18n:lint) to ensure keys mirror zh-CN
and avoid missing/plural mismatches, then convert the resulting zh-TW strings
with OpenCC using the s2twp.json config to apply Taiwan-specific traditional
mappings (so terms like 快取/建立/影片/登錄/儲值/帳號 match existing translations and
_one/_other plural pairs remain in the same locale).

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@web/src/i18n/locales/zh-TW.json`:
- Around line 3788-4367: The zh-TW.json file contains ~561 entries with
Vietnamese values (e.g., keys "Recharge Quota", "主模型", "模型加载中..." showing
Vietnamese text) instead of Traditional Chinese; fix by re-importing the correct
translations from zh-CN.json and converting Simplified-to-Traditional with the
project's i18n tooling (run the i18n sync command such as bun run i18n:sync or
the repo's equivalent) ensuring OpenCC s2t conversion is used (not vi.json),
replace the Vietnamese values in zh-TW.json with the converted Traditional
Chinese text and verify keys like "Recharge Quota", "主模型", "模型加载中..." now
contain Traditional Chinese.
- Around line 3640-3787: The zh-TW JSON section contains simplified characters
and inconsistent terminology (e.g., keys like "阶梯内 Token 数", "同时满足", "缓存创建-1小时
(cc1h)" use 简体/混用); run the i18n tooling (bun run i18n:sync && bun run
i18n:lint) and convert these new values with OpenCC s2twp, enforcing the
project's Taiwan variants and existing terminology mapping (缓存 → 快取, 创建 → 建立, 数
→ 數, 档 → 檔, 时 → 時, 满足 → 滿足, etc.), then update the entries for the affected keys
(e.g., "阶梯内 Token 数", "同时满足", "缓存创建-5分钟 (cc5)", "缓存创建-1小时 (cc1h)", "缓存创建 Token
(cc)") to use the standardized translations used elsewhere in the file.
- Line 4160: Remove the corrupted translation entry with key "设计版本" and value
"b80c3466cb6feafeb3990c7820e10e50" from all locale JSON files where it appears;
locate occurrences by searching for the exact key or hash and delete the
key/value pair (ensure trailing commas are adjusted to keep valid JSON), then
run a quick search across the repo to confirm no code references remain and run
the i18n build/check to verify no regressions.

---

Nitpick comments:
In `@web/src/i18n/locales/zh-TW.json`:
- Around line 1-2: Regenerate zh-TW via the project's i18n pipeline instead of
hand-editing: run the CLI flow (bun run i18n:extract, bun run i18n:sync, bun run
i18n:lint) to ensure keys mirror zh-CN and avoid missing/plural mismatches, then
convert the resulting zh-TW strings with OpenCC using the s2twp.json config to
apply Taiwan-specific traditional mappings (so terms like 快取/建立/影片/登錄/儲值/帳號
match existing translations and _one/_other plural pairs remain in the same
locale).

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 19176a30-3324-49a2-8e94-4196a0f301ff

📥 Commits

Reviewing files that changed from the base of the PR and between e36d191 and fb59fa1.

📒 Files selected for processing (1)

• web/src/i18n/locales/zh-TW.json

[coderabbitai]

⚠️ Potential issue | 🔴 Critical

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
# Find newly added zh-TW values that still contain simplified-only characters,
# and count occurrences of the inconsistent 缓存/创建 vs 快取/建立 terminology.
python3 - <<'PY'
import json, re
with open('web/src/i18n/locales/zh-TW.json', encoding='utf-8') as f:
    data = json.load(f)['translation']

simp_only = set('阶档数时满维决据进条缓创输验选适该项钥钟际钱账务图视频边业级历题问统计运营议级')
bad = [(k, v) for k, v in data.items()
       if isinstance(v, str) and any(c in simp_only for c in v)]
print(f'Values with simplified-only chars: {len(bad)}')
for k, v in bad[:15]:
    print(f'  {k!r} -> {v!r}')

kuai = sum(1 for v in data.values() if isinstance(v, str) and '快取' in v)
huan = sum(1 for v in data.values() if isinstance(v, str) and '緩存' in v or '缓存' in v)
jianli = sum(1 for v in data.values() if isinstance(v, str) and '建立' in v)
chuangjian = sum(1 for v in data.values() if isinstance(v, str) and '創建' in v or '创建' in v)
print(f'快取 occurrences: {kuai}, 緩存/缓存 occurrences: {huan}')
print(f'建立 occurrences: {jianli}, 創建/创建 occurrences: {chuangjian}')
PY

Repository: QuantumNous/new-api

Length of output: 676

---

Fix incomplete Traditional Chinese conversion and terminology inconsistency in zh-TW.json entries.

The new entries contain 86 values with simplified-only characters still present (阶, 档, 数, 时, etc.) and split terminology:

• File predominantly uses 快取 (81 times) and 建立 (86 times) for "cache" and "create"
• New entries introduce 緩存/缓存 (15 times) and 創建/创建 (11 times), creating inconsistent UI terminology

Examples:

• Line 3641: "阶梯内 Token 数" should be "階梯內 Token 數"
• Line 3642: "同时满足" should be "同時滿足"
• Cache entries use 緩存創建 instead of the existing 快取建立 pattern throughout the file

Regenerate this section using proper i18n tooling as per coding guidelines:

bun run i18n:sync
bun run i18n:lint

Run values through OpenCC s2twp (Taiwan phrase variant) to align 缓存→快取, 创建→建立, and other terminology with the rest of the file.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@web/src/i18n/locales/zh-TW.json` around lines 3640 - 3787, The zh-TW JSON
section contains simplified characters and inconsistent terminology (e.g., keys
like "阶梯内 Token 数", "同时满足", "缓存创建-1小时 (cc1h)" use 简体/混用); run the i18n tooling
(bun run i18n:sync && bun run i18n:lint) and convert these new values with
OpenCC s2twp, enforcing the project's Taiwan variants and existing terminology
mapping (缓存 → 快取, 创建 → 建立, 数 → 數, 档 → 檔, 时 → 時, 满足 → 滿足, etc.), then update the
entries for the affected keys (e.g., "阶梯内 Token 数", "同时满足", "缓存创建-5分钟 (cc5)",
"缓存创建-1小时 (cc1h)", "缓存创建 Token (cc)") to use the standardized translations used
elsewhere in the file.

[coderabbitai]

⚠️ Potential issue | 🔴 Critical

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
# Count zh-TW values that contain Vietnamese-only Latin diacritics (đ, ă, ơ, ư, ê, ô, ạ, ậ, …)
# These characters never appear in legitimate Traditional Chinese values.
python3 - <<'PY'
import json, re
with open('web/src/i18n/locales/zh-TW.json', encoding='utf-8') as f:
    data = json.load(f)['translation']
vi_re = re.compile(r'[đĐăĂâÂêÊôÔơƠưƯạảãáàậẩẫấầặẳẵắằộổỗốồớờởỡợệểễếềịỉĩíìụủũúùựửữứừỳỵỷỹý]')
hits = [(k, v) for k, v in data.items() if isinstance(v, str) and vi_re.search(v)]
print(f'Vietnamese-tainted zh-TW values: {len(hits)} / {len(data)}')
for k, v in hits[:10]:
    print(f'  {k!r} -> {v!r}')
PY

Repository: QuantumNous/new-api

Length of output: 970

---

Remove 561 Vietnamese values from zh-TW.json and replace with proper Traditional Chinese translations.

Analysis of the file reveals 561 keys (12.8% of the locale) have Vietnamese values instead of Traditional Chinese. Examples include "Recharge Quota": "Hạn ngạch nạp tiền" (Vietnamese), "主模型": "Model chính" (Vietnamese), "模型加载中...": "Đang tải mô hình..." (Vietnamese), and hundreds more using Vietnamese-specific diacritics (đ, ă, ơ, ư, ê, ô, ạ, ậ, etc.) that never appear in Chinese.

This causes zh-TW users to see Vietnamese UI text across login, account management, channels, models, requests, and other surfaces—a regression worse than untranslated fallback.

Re-import these keys from zh-CN.json and run proper OpenCC s2t (Simplified-to-Traditional) conversion using the project's i18n tooling (bun run i18n:sync or equivalent). Do not use vi.json as the source.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@web/src/i18n/locales/zh-TW.json` around lines 3788 - 4367, The zh-TW.json
file contains ~561 entries with Vietnamese values (e.g., keys "Recharge Quota",
"主模型", "模型加载中..." showing Vietnamese text) instead of Traditional Chinese; fix
by re-importing the correct translations from zh-CN.json and converting
Simplified-to-Traditional with the project's i18n tooling (run the i18n sync
command such as bun run i18n:sync or the repo's equivalent) ensuring OpenCC s2t
conversion is used (not vi.json), replace the Vietnamese values in zh-TW.json
with the converted Traditional Chinese text and verify keys like "Recharge
Quota", "主模型", "模型加载中..." now contain Traditional Chinese.

[coderabbitai]

⚠️ Potential issue | 🟠 Major

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
# Confirm the key is referenced in the frontend and check what other locales have.
rg -nP --type=js --type=jsx --type=ts --type=tsx -C2 "t\(\s*['\"]设计版本['\"]"
python3 - <<'PY'
import json, os
for f in ['zh-CN.json','zh-TW.json','en.json','vi.json','ja.json','fr.json','ru.json']:
    p = f'web/src/i18n/locales/{f}'
    if not os.path.exists(p):
        continue
    with open(p, encoding='utf-8') as fh:
        d = json.load(fh).get('translation', {})
    if '设计版本' in d:
        print(f'{f}: {d["设计版本"]!r}')
PY

Repository: QuantumNous/new-api

Length of output: 376

---

🏁 Script executed:

# Check zh-CN.json for the key
cat web/src/i18n/locales/zh-CN.json | python3 -c "
import sys, json
data = json.load(sys.stdin)
translation = data.get('translation', {})
if '设计版本' in translation:
    print(f'Found in zh-CN.json: {translation[\"设计版本\"]!r}')
else:
    print('Key not found in zh-CN.json')
"

# Broader search for the key in all files (not just JS/TS)
echo "=== Broader codebase search ==="
rg -n "设计版本" --max-count=20

Repository: QuantumNous/new-api

Length of output: 618

---

Remove the junk hash entry from zh-TW.json and all other affected locale files.

The key "设计版本" does not exist in zh-CN.json (the source language file), yet it appears in zh-TW.json, en.json, vi.json, ja.json, fr.json, and ru.json—all with the identical value "b80c3466cb6feafeb3990c7820e10e50". This hash value is not a translation; it's data corruption that must be removed. The key is not referenced anywhere in the codebase.

Delete this entry from all affected locale files:

• web/src/i18n/locales/zh-TW.json (line 4160)
• web/src/i18n/locales/en.json (line 2967)
• web/src/i18n/locales/vi.json (line 3280)
• web/src/i18n/locales/ja.json (line 2922)
• web/src/i18n/locales/fr.json (line 2953)
• web/src/i18n/locales/ru.json (line 2973)

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@web/src/i18n/locales/zh-TW.json` at line 4160, Remove the corrupted
translation entry with key "设计版本" and value "b80c3466cb6feafeb3990c7820e10e50"
from all locale JSON files where it appears; locate occurrences by searching for
the exact key or hash and delete the key/value pair (ensure trailing commas are
adjusted to keep valid JSON), then run a quick search across the repo to confirm
no code references remain and run the i18n build/check to verify no regressions.