@kensio/pinyinjs 1.10.0 → 1.10.2

package/dist/dictionary/reference-readings.js

@@ -0,0 +1,112 @@
+/**
+ * Readings a standard reference dictionary gives that no source carries.
+ *
+ * Unihan, the phrase corpus and CC-CEDICT between them miss a tail of
+ * readings that print dictionaries list (a character's reading in one literary
+ * word, a surname, a place name, a dialect word). Where a source has the word
+ * and reads it from the character's everyday reading, the printed reading is
+ * the better evidence, and it replaces the corpus's.
+ */
+/**
+ * A character's readings a standard reference dictionary gives and the
+ * sources lack, as tone-marked pinyin.
+ *
+ * They follow every reading a source gives, so they never set a character's
+ * default and only ever stand as alternates. The three characters no source
+ * reads at all take the reading here as their only one. A 繁體 form is listed
+ * beside its 简体 one, since each has a Unihan entry of its own.
+ *
+ * Three dialect readings are left out, 不 `dǔn` in 不子, 泽 `duó` in 凌泽 (an
+ * icicle) and syllabic 姆 `m̄`. Each character has one reading otherwise, and a
+ * second would mark every free-standing 不, every 泽 in a given name and every
+ * 姆 in a transliterated one uncertain. 不子 carries its reading as a word.
+ */
+const CHARACTER_READINGS = new Map([
+    ["咹", ["ǎn"]],
+    ["㼎", ["báo"]],
+    ["箄", ["bēi"]],
+    ["屏", ["bīng"]], // 屏营
+    ["礤", ["cā"]],
+    ["倡", ["chāng"]], // 倡优
+    ["樘", ["chēng"]],
+    ["𡷫", ["chéng"]], // a surname
+    ["郗", ["chī"]], // a surname
+    ["佁", ["chì"]], // 佁儗
+    ["玼", ["cī"]],
+    ["粢", ["cī"]], // 粢饭
+    ["酂", ["cuán"]], // a place name
+    ["酇", ["cuán"]],
+    ["嵅", ["dǎn", "dǎng"]], // place names in Guangxi
+    ["碴", ["zhǎ"]], // 道碴
+    ["嘚", ["dè"]], // 嘚瑟
+    ["镫", ["dēng"]],
+    ["鐙", ["dēng"]],
+    ["澄", ["dēng"]],
+    ["螲", ["dié"]], // 螲𬠅
+    ["𭣧", ["dù"]],
+    ["柂", ["duò"]],
+    ["媠", ["duò"]],
+    ["欸", ["ē", "é", "ě", "è"]],
+    ["噁", ["è"]], // 二噁英
+    ["胲", ["gǎi"]],
+    ["𭥙", ["gǎng"]],
+    ["岗", ["gàng"]], // 岗尖, 岗口儿甜
+    ["崗", ["gàng"]],
+    ["瘩", ["dā"]], // 疙疙瘩瘩, 疙里疙瘩
+    ["𫵼", ["gēng"]],
+    ["𣪘", ["guǐ"]],
+    ["亢", ["háng"]], // the throat, in literary use
+    ["猲", ["hè"]],
+    ["愒", ["hè"]],
+    ["沕", ["hū"]],
+    ["鳠", ["huò"]],
+    ["鱯", ["huò"]],
+    ["雘", ["huò"]],
+    ["筊", ["jiǎo"]],
+    ["寁", ["jié"]],
+    ["蹴", ["jiù"]],
+    ["菹", ["jù"]],
+    ["朘", ["juān"]],
+    ["蠼", ["jué"]],
+    ["焅", ["kào"]],
+    ["㲉", ["kòu"]],
+    ["磈", ["kuǐ"]],
+    ["癞", ["là"]], // 癞痢
+    ["癩", ["là"]],
+    ["乐", ["lào"]], // 乐亭, a place name
+    ["樂", ["lào"]],
+    ["嘞", ["lē"]],
+    ["𥉳", ["liū"]],
+    ["蹓", ["liù"]],
+    ["碌", ["lū"]], // 骨碌碌
+    ["箵", ["xǐng"]],
+]);
+/**
+ * Words a standard reference dictionary reads, with the reading it gives, as
+ * the phrase corpus writes one. That is a syllable per character, tone-marked,
+ * and unmarked for 轻声. An 儿 read as 儿化 is `r5`, as CC-CEDICT writes it.
+ *
+ * Each either has no source at all or is read in the phrase corpus from its
+ * characters' everyday readings. 凌泽 is left out. It is a dialect word for an
+ * icicle, and as a word it would take every given name 凌泽 apart.
+ */
+const WORD_READINGS = new Map([
+    ["粢饭", ["cī", "fàn"]],
+    ["道碴", ["dào", "zhǎ"]],
+    ["螲𬠅", ["dié", "dāng"]],
+    ["不子", ["dǔn", "zi"]],
+    ["岗尖", ["gàng", "jiān"]],
+    ["岗口儿甜", ["gàng", "kǒu", "r5", "tián"]],
+    ["疙疙瘩瘩", ["gē", "ge", "dā", "dā"]],
+    ["疙里疙瘩", ["gē", "li", "gē", "dā"]],
+    ["骨碌碌", ["gū", "lū", "lū"]],
+    ["骨碌碌地", ["gū", "lū", "lū", "de"]],
+]);
+/**
+ * The readings the build adds to the sources.
+ */
+export const REFERENCE_READINGS = {
+    characters: CHARACTER_READINGS,
+    words: WORD_READINGS,
+};
+//# sourceMappingURL=reference-readings.js.map

package/dist/dictionary/reference-readings.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"reference-readings.js","sourceRoot":"","sources":["../../src/dictionary/reference-readings.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAUH;;;;;;;;;;;;;GAaG;AACH,MAAM,kBAAkB,GAA2C,IAAI,GAAG,CAAC;IACzE,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,CAAC;IACb,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;IACd,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;IACd,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC,CAAC,EAAE,KAAK;IACtB,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,CAAC;IACb,CAAC,GAAG,EAAE,CAAC,OAAO,CAAC,CAAC,EAAE,KAAK;IACvB,CAAC,GAAG,EAAE,CAAC,OAAO,CAAC,CAAC;IAChB,CAAC,IAAI,EAAE,CAAC,OAAO,CAAC,CAAC,EAAE,YAAY;IAC/B,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC,EAAE,YAAY;IAC5B,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC,EAAE,KAAK;IACrB,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,CAAC;IACb,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,CAAC,EAAE,KAAK;IACpB,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC,CAAC,EAAE,eAAe;IAChC,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC,CAAC;IACf,CAAC,GAAG,EAAE,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC,EAAE,yBAAyB;IACjD,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC,EAAE,KAAK;IACrB,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,CAAC,EAAE,KAAK;IACpB,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC,CAAC;IACf,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC,CAAC;IACf,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC,CAAC;IACf,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC,EAAE,MAAM;IACtB,CAAC,IAAI,EAAE,CAAC,IAAI,CAAC,CAAC;IACd,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;IACd,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;IACd,CAAC,GAAG,EAAE,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;IAC3B,CAAC,GAAG,EAAE,CAAC,GAAG,CAAC,CAAC,EAAE,MAAM;IACpB,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;IACd,CAAC,IAAI,EAAE,CAAC,MAAM,CAAC,CAAC;IAChB,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC,CAAC,EAAE,WAAW;IAC5B,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC,CAAC;IACf,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,CAAC,EAAE,aAAa;IAC5B,CAAC,IAAI,EAAE,CAAC,MAAM,CAAC,CAAC;IAChB,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC,CAAC;IACf,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC,CAAC,EAAE,8BAA8B;IAC/C,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,CAAC;IACb,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,CAAC;IACb,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,CAAC;IACb,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;IACd,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;IACd,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;IACd,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC,CAAC;IACf,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;IACd,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;IACd,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,CAAC;IACb,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC,CAAC;IACf,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;IACd,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;IACd,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;IACd,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;IACd,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,CAAC,EAAE,KAAK;IACpB,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,CAAC;IACb,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC,EAAE,mBAAmB;IACnC,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;IACd,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,CAAC;IACb,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC,CAAC;IACf,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;IACd,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,CAAC,EAAE,MAAM;IACrB,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC,CAAC;CAChB,CAAC,CAAC;AAEH;;;;;;;;GAQG;AACH,MAAM,aAAa,GAA2C,IAAI,GAAG,CAAC;IACpE,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;IACrB,CAAC,IAAI,EAAE,CAAC,KAAK,EAAE,KAAK,CAAC,CAAC;IACtB,CAAC,KAAK,EAAE,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC;IACxB,CAAC,IAAI,EAAE,CAAC,KAAK,EAAE,IAAI,CAAC,CAAC;IACrB,CAAC,IAAI,EAAE,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;IACxB,CAAC,MAAM,EAAE,CAAC,MAAM,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,CAAC,CAAC;IACvC,CAAC,MAAM,EAAE,CAAC,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,CAAC;IAClC,CAAC,MAAM,EAAE,CAAC,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,CAAC;IAClC,CAAC,KAAK,EAAE,CAAC,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,CAAC;IAC3B,CAAC,MAAM,EAAE,CAAC,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,CAAC;CACnC,CAAC,CAAC;AAEH;;GAEG;AACH,MAAM,CAAC,MAAM,kBAAkB,GAAsB;IACnD,UAAU,EAAE,kBAAkB;IAC9B,KAAK,EAAE,aAAa;CACrB,CAAC"}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@kensio/pinyinjs",
-  "version": "1.10.0",
+  "version": "1.10.2",
   "description": "Chinese hanzi to pinyin conversion",
   "repository": "https://github.com/KensioSoftware/pinyinjs",
   "homepage": "https://pinyinjs.dev",
@@ -69,7 +69,7 @@
     "conventional-changelog-conventionalcommits": "9.3.1",
     "fta-cli": "^3.0.0",
     "jiti": "^2.7.0",
-    "oxfmt": "^0.73.0",
+    "oxfmt": "^0.74.0",
     "oxlint": "^1.77.0",
     "oxlint-tsgolint": "^7.0.2001",
     "semantic-release": "25.0.9",