Skip to content

feat(extras): promote 老抽 / プーアール茶 from candidate scan#243

Merged
send merged 2 commits intomainfrom
feat/extras-promote-batch1
May 9, 2026
Merged

feat(extras): promote 老抽 / プーアール茶 from candidate scan#243
send merged 2 commits intomainfrom
feat/extras-promote-batch1

Conversation

@send
Copy link
Copy Markdown
Owner

@send send commented May 9, 2026

Summary

PR #242 で merge した `dictool candidates mine` の出力をスキャンし、Mozc 辞書の真の gap を 2 件のみ promote。

reading surface 現状の Mozc top-1 補完理由
ろうちゅう 老抽 郎中 (final=19063) 老抽 は build dict に無し。中華料理の濃口醤油
ぷーあーるちゃ プーアール茶 プーアールちゃ (mixed-script) compound 単語化で正しく茶に固定

ビルドログ: `Merged 'extras': +7 readings, +9 entries, 0 replaced` で衝突なしを確認。

Scan で除外した候補 (memo)

  • `らおちゅう→老抽`: Mozc 既存の `らおちゅう→老酒` (final=13716) を demote してしまうため見送り
  • `てんめんじゃん→甜麺醤`: Mozc に既に top-1 として存在 (final=12041)、redundant
  • `とうち→豆豉`: 大幅に common な `当地` を demote する副作用

Accuracy

`mise run accuracy`: 77/77 active pass (skip 1 = 既存)、extras カテゴリ 7→9 cases。`mise run accuracy-history`: 6/6 pass。

Test plan

  • `cargo fmt --check` / `cargo clippy --all-features -- -D warnings` / `cargo test --workspace --all-features`
  • `mise run accuracy` 全 pass (extras +2 ケース追加)
  • `mise run accuracy-history` 全 pass
  • `mise run build && mise run install && mise run reload`
  • 実機で `ろうちゅう` → `老抽`、`ぷーあーるちゃ` → `プーアール茶` が top-1 候補に出ることを確認

🤖 Generated with Claude Code

Scanned `dictool candidates mine` (PR #242) output and hand-picked 2
food entries that are genuine Mozc gaps:

- ろうちゅう → 老抽 (中華の濃口醤油)
  Mozc top-1 was 郎中 (final=19063); 老抽 was absent from the build dict.

- ぷーあーるちゃ → プーアール茶
  Mozc has no compound entry; lattice produced "プーアールちゃ" (mixed
  script) as top-1. Promoting to a single-word entry pins 茶 as the
  tail kanji.

Skipped during scan:
- らおちゅう→老抽 (would demote existing らおちゅう→老酒)
- てんめんじゃん→甜麺醤 (Mozc already top-1)
- とうち→豆豉 (would demote much-more-common 当地)

Build merge log confirms no surface collisions (`+7 readings, +9
entries, 0 replaced`). Accuracy: 77/77 active cases pass.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Copilot AI review requested due to automatic review settings May 9, 2026 07:26
Copy link
Copy Markdown

Copilot AI left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

This PR promotes two curated “extras” dictionary entries (老抽 / プーアール茶) identified via candidate-scan output, and adds corresponding accuracy corpus cases to ensure they remain top-1 conversions.

Changes:

  • Add ろうちゅう→老抽 and ぷーあーるちゃ→プーアール茶 to the curated extras food TSV.
  • Add two extras cases to the conversion accuracy corpus to lock in expected top-1 outputs.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File Description
engine/testcorpus/accuracy-corpus.toml Adds 2 new extras accuracy cases for the promoted entries.
engine/crates/lex-cli/src/dict_source/extras/food.tsv Promotes 2 curated food-related entries into the extras dictionary source.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Comment thread engine/testcorpus/accuracy-corpus.toml
Comment thread engine/crates/lex-cli/src/dict_source/extras/food.tsv
- food.tsv: split "中国茶" out from "日本茶ブランド" header so プーアール茶
  doesn't sit under a 日本茶-only section
- accuracy-corpus.toml: reorder so 老抽 / プーアール茶 (food) stay
  contiguous with the other food cases instead of being split by 雲母坂
  (geography)

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Copy link
Copy Markdown

Copilot AI left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated no new comments.

@send send merged commit f696738 into main May 9, 2026
14 checks passed
@send send deleted the feat/extras-promote-batch1 branch May 9, 2026 07:47
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants