@kensio/pinyinjs 1.1.0 → 1.2.1

package/CHANGELOG.md

@@ -29,6 +29,15 @@ which a dictionary rebuild can change, and the artifact format under `data/`.
 
 ### Fixed
 
+- **A decimal lost the grouping of its counted part.** 一共75.5元 was
+  `yígòng qī shí wǔ diǎn wǔ yuán`, with the 75 in loose syllables, and is now
+  `yígòng qīshíwǔ diǎn wǔ yuán` — the same word 一共75元 already got. Only what
+  follows the 点 is read a digit at a time, and that part is unchanged: 3.14 is
+  still `sān diǎn yī sì`, and sandhi still stops at the point.
+- **A decimal point was read as a full stop.** 一共75.5元 came back as
+  `Yígòng …`, capitalised as though the text were a sentence, and a conversion
+  keeping its digits capitalised again after the point: 我有75.5个。was
+  `Wǒ yǒu75.5Gè.` A stop between two digits now ends no sentence.
 - **Bopomofo wrote the tone mark after the 儿化 ㄦ.** 哪儿 nǎr was `ㄋㄚㄦˇ` and
   is now `ㄋㄚˇㄦ`: the mark belongs to the nucleus and the suffix is not part of
   what it marks. A mark written after the ㄦ is still read.
@@ -50,6 +59,41 @@ which a dictionary rebuild can change, and the artifact format under `data/`.
   transcription systems_, which is where its ground truth comes from; the two
   pages differ at the medials, -ang, the empty rhyme and the diphthongs. No
   output changed — the claim did.
+- **`--system ipa` capitalised the IPA.** 我去银行。他姓王。 came out as
+  `Uo˨˩˦ ... Tʰa˥ ɕiŋ˥˩ Uaŋ˧˥`, with the sentence capital and the proper noun
+  carried over from the pinyin as though a transcription spelled words. IPA
+  letters are symbols rather than an alphabet — `[T]` is not `[t]` in a larger
+  size but a symbol the IPA does not have — so the capitals are now dropped for
+  IPA and bopomofo and kept for the three romanisations. `toTranscription`
+  takes `{ capitals: false }` for callers writing their own.
+- **A compound lost the zh-TW reading its own parts carry.** 垃圾分類 read
+  `lājīfēnlèi` under `zh-TW`, because CC-CEDICT marks `Taiwan pr.` on 垃圾 and
+  垃圾桶 and on no other compound, and the decoder prefers the longer word — so
+  the entry that had the delta was never consulted. 101 compounds now compose
+  one from their constituents, 垃圾 accounting for 14 of them and 比肩, 玳瑁,
+  从容, 骨头, 蜗牛 and 说服 for most of the rest. Three guards keep it honest:
+  the constituent has to survive segmentation (運行狀況 contains 行狀 but reads
+  運行 + 狀況), it has to be read in the compound exactly as its own entry reads
+  it, and it has to be a whole word — never a bare character, since 會's `huǐ`
+  would rewrite 三合會 as `sānhéhuǐ` and 3,743 entries would compose that way
+  against the 101 that whole words compose.
+- **A character's zh-TW delta was often a choice between its 普通话 senses.**
+  从容地 read `cōngróng dì` and 他都知道 read `tā dū zhīdào` under `zh-TW`,
+  because 地 was stored with a Taiwan reading of `dì` and 都 with `dū`. Neither
+  is a locale shift: CC-CEDICT lists 地[de5] and 地[di4] as separate entries, so
+  `dì` is what 地 reads in 普通话 when it means the ground, and the adverbial
+  particle is `de` in Taipei too. A delta the word already has as a 普通话 sense
+  is now refused — 71 characters and 3 words lose one, 都, 着, 应, 差, 称, 斗,
+  舍, 薄 and 万 among them — while 和 `hàn`, 期 `qí`, 垃 `lè` and 髮 `fǎ` stay,
+  no 普通话 sense reading any of those that way. A `Taiwan pr.` note is also
+  only read off a sense matching the reading the entry settled on, which drops
+  16 more: the note on 著's chess-move sense, which reads `zhāo`, had been
+  giving the aspect particle 着 a 國語 reading of `zhuó`.
+- **A locale delta overwrote a reading it was not measured against.** The delta
+  describes the entry's own 普通话 reading, so a polyphone the decode read some
+  other way — 得 forced to the modal `děi` rather than the particle `de` its
+  entry stores — now keeps what the decode worked out instead of being rewritten
+  from the entry.
 
 ### Changed
 
@@ -123,14 +167,14 @@ decoder internals and the build pipeline.
 
 ### Accuracy at 1.0.0
 
-Over the 105-case gold corpus, which is committed to the repository — not to
+Over the 110-case gold corpus, which is committed to the repository — not to
 the published package — and scored by `pnpm accuracy`:
 
 |                  |   lattice | greedy baseline |
 | ---------------- | --------: | --------------: |
-| exact match      | **97.1%** |           92.4% |
-| reading accuracy |     99.7% |           98.6% |
-| spacing (F1)     | **99.7%** |           97.0% |
+| exact match      | **97.3%** |           90.9% |
+| reading accuracy |     99.7% |           98.7% |
+| spacing (F1)     | **99.7%** |           96.3% |
 
 Every figure in that table is asserted against the scorer by
 `src/changelog.test.ts`, because a number nothing executes goes stale — which is