@kensio/pinyinjs 1.7.0 → 1.7.2

package/dist/dictionary/neutral-senses.js

@@ -0,0 +1,62 @@
+/**
+ * Words where CC-CEDICT's 轻声 sense is the everyday word and its full-tone
+ * sense is something else written the same way.
+ *
+ * Step 7 of the merge takes CC-CEDICT's neutral tones over the phrase corpus's
+ * full ones, and it does so against **the sense nearest the reading it already
+ * has** — otherwise 行长 would take 行's other pronunciation. That works for the
+ * 609 words it settles, and it cannot work here: the phrase corpus's reading
+ * matches the full-tone sense *exactly*, so the nearest sense is the one that
+ * already agrees and there is no neutral tone to take. 东西 is `dong1 xi1` (east
+ * and west) and `dong1 xi5` (thing), the corpus writes `dōng xī`, and the thing
+ * never got a say.
+ *
+ * **98 words are in that shape and nothing in the pipeline can rank two senses
+ * of one spelling.** jieba counts the spelling, not the sense. CPP labels 6 of
+ * the 98. Letting longer CC-CEDICT headwords vote, weighted by jieba, picks the
+ * *literal* sense for 东西 by 663 to 113 — because compounds like 东西半球 are
+ * exactly where a literal sense lives. So this is a list, and it is a list of
+ * judgements.
+ *
+ * **What is listed here.** A word whose 轻声 sense is the one 现代汉语词典 gives
+ * the headword, and whose full-tone sense is literary (生意 the life force),
+ * technical (说法 expounding the dharma), legal (告诉 filing a complaint) or a
+ * name (大方县, 金子 the Japanese surname). What is *not* listed is the larger
+ * half of the 98 where both senses are current and common — 地方 is `dìfang` the
+ * place and `dìfāng` the opposite of central, 大意 is the gist and carelessness,
+ * 多少 is how-many and an amount, 地道 is a tunnel and the real thing. Choosing
+ * for those would be trading one wrong answer for another.
+ *
+ * **No reading is written here**, only the word. The merge takes the neutral
+ * sense CC-CEDICT already holds, so a source refresh that changes a word's
+ * senses cannot leave a hand-typed reading behind — and `BUILD_ASSERTIONS`
+ * fails the build if any word here stops reading with a 轻声 at all.
+ */
+export const NEUTRAL_SENSE_WORDS = [
+    { word: "东西", displaces: "east and west, the compass directions" },
+    { word: "告诉", displaces: "to press charges, the legal term" },
+    { word: "故事", displaces: "故事 gùshì, an old practice or precedent" },
+    { word: "妻子", displaces: "wife and children, the classical sense" },
+    { word: "说法", displaces: "to expound Buddhist teachings" },
+    { word: "小子", displaces: "CC-CEDICT's own (literary) and (old) youngster" },
+    { word: "买卖", displaces: "to buy and sell, the verb" },
+    { word: "生意", displaces: "life force, vitality" },
+    { word: "本事", displaces: "the source material a work is based on" },
+    { word: "大爷", displaces: "an arrogant idler, not one's uncle" },
+    { word: "大方", displaces: "大方县, the county in Guizhou" },
+    { word: "口音", displaces: "oral speech sounds, the linguistics term" },
+    { word: "结实", displaces: "to bear fruit" },
+    { word: "把手", displaces: "to shake hands, the verb" },
+    { word: "金子", displaces: "Kaneko, the Japanese surname" },
+    { word: "出息", displaces: "to yield interest, and to exhale in Buddhism" },
+    {
+        word: "管子",
+        displaces: "管子, the Warring States politician and the book",
+    },
+    { word: "支吾", displaces: "to resist, to withstand" },
+];
+/**
+ * The listed words, for the merge to test membership against.
+ */
+export const NEUTRAL_SENSE_LOOKUP = new Set(NEUTRAL_SENSE_WORDS.map((listed) => listed.word));
+//# sourceMappingURL=neutral-senses.js.map

package/dist/dictionary/neutral-senses.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"neutral-senses.js","sourceRoot":"","sources":["../../src/dictionary/neutral-senses.ts"],"names":[],"mappings":"AAWA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAiCG;AACH,MAAM,CAAC,MAAM,mBAAmB,GAAgC;IAC9D,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE,uCAAuC,EAAE;IAClE,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE,kCAAkC,EAAE;IAC7D,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE,wCAAwC,EAAE;IACnE,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE,wCAAwC,EAAE;IACnE,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE,+BAA+B,EAAE;IAC1D,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE,gDAAgD,EAAE;IAC3E,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE,2BAA2B,EAAE;IACtD,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE,sBAAsB,EAAE;IACjD,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE,wCAAwC,EAAE;IACnE,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE,oCAAoC,EAAE;IAC/D,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE,4BAA4B,EAAE;IACvD,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE,0CAA0C,EAAE;IACrE,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE,eAAe,EAAE;IAC1C,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE,0BAA0B,EAAE;IACrD,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE,8BAA8B,EAAE;IACzD,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE,8CAA8C,EAAE;IACzE;QACE,IAAI,EAAE,IAAI;QACV,SAAS,EAAE,gDAAgD;KAC5D;IACD,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE,yBAAyB,EAAE;CACrD,CAAC;AAEF;;GAEG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAAwB,IAAI,GAAG,CAC9D,mBAAmB,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,IAAI,CAAC,CACjD,CAAC"}

package/dist/dictionary/overrides.d.ts

@@ -23,6 +23,12 @@ export interface ReadingOverride {
  * override is for the case where the sources agree with each other and are
  * still wrong, which is rare and almost always means a word whose everyday
  * sense has been displaced by an archaic one.
+ *
+ * A word CC-CEDICT reads *both* ways, where only the choice between its senses
+ * is wrong, belongs in
+ * {@link import("./neutral-senses.js").NEUTRAL_SENSE_WORDS} instead — that
+ * table names the word and lets CC-CEDICT supply the reading, so nothing is
+ * typed by hand that a source already holds.
  */
 export declare const READING_OVERRIDES: readonly ReadingOverride[];
 /**

package/dist/dictionary/overrides.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"overrides.d.ts","sourceRoot":"","sources":["../../src/dictionary/overrides.ts"],"names":[],"mappings":"AAAA,OAAO,EAAgB,KAAK,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAGtE;;GAEG;AACH,MAAM,WAAW,eAAe;IAC9B,uBAAuB;IACvB,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,4CAA4C;IAC5C,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;IACzB,0CAA0C;IAC1C,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAC;CACzB;AAED;;;;;;;;;;;;;GAaG;AACH,eAAO,MAAM,iBAAiB,EAAE,SAAS,eAAe,EAavD,CAAC;AAEF;;;;;;;GAOG;AACH,wBAAgB,mBAAmB,CACjC,QAAQ,EAAE,eAAe,GACxB,SAAS,QAAQ,EAAE,CAqBrB;AAED;;GAEG;AACH,eAAO,MAAM,iBAAiB,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,QAAQ,EAAE,CAMpE,CAAC"}
+{"version":3,"file":"overrides.d.ts","sourceRoot":"","sources":["../../src/dictionary/overrides.ts"],"names":[],"mappings":"AAAA,OAAO,EAAgB,KAAK,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAGtE;;GAEG;AACH,MAAM,WAAW,eAAe;IAC9B,uBAAuB;IACvB,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,4CAA4C;IAC5C,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;IACzB,0CAA0C;IAC1C,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAC;CACzB;AAED;;;;;;;;;;;;;;;;;;;GAmBG;AACH,eAAO,MAAM,iBAAiB,EAAE,SAAS,eAAe,EAavD,CAAC;AAEF;;;;;;;GAOG;AACH,wBAAgB,mBAAmB,CACjC,QAAQ,EAAE,eAAe,GACxB,SAAS,QAAQ,EAAE,CAqBrB;AAED;;GAEG;AACH,eAAO,MAAM,iBAAiB,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,QAAQ,EAAE,CAMpE,CAAC"}

package/dist/dictionary/overrides.js

@@ -13,6 +13,12 @@ import { NEUTRAL_TONE } from "../tone/tone.js";
  * override is for the case where the sources agree with each other and are
  * still wrong, which is rare and almost always means a word whose everyday
  * sense has been displaced by an archaic one.
+ *
+ * A word CC-CEDICT reads *both* ways, where only the choice between its senses
+ * is wrong, belongs in
+ * {@link import("./neutral-senses.js").NEUTRAL_SENSE_WORDS} instead — that
+ * table names the word and lets CC-CEDICT supply the reading, so nothing is
+ * typed by hand that a source already holds.
  */
 export const READING_OVERRIDES = [
     {

package/dist/dictionary/overrides.js.map

@@ -1 +1 @@
-{"version":3,"file":"overrides.js","sourceRoot":"","sources":["../../src/dictionary/overrides.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAiB,MAAM,yBAAyB,CAAC;AACtE,OAAO,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAc/C;;;;;;;;;;;;;GAaG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAA+B;IAC3D;QACE,IAAI,EAAE,IAAI;QACV,OAAO,EAAE,QAAQ;QACjB,MAAM,EACJ,+HAA+H;KAClI;IACD;QACE,IAAI,EAAE,IAAI;QACV,OAAO,EAAE,OAAO;QAChB,MAAM,EACJ,+UAA+U;KAClV;CACF,CAAC;AAEF;;;;;;;GAOG;AACH,MAAM,UAAU,mBAAmB,CACjC,QAAyB;IAEzB,MAAM,SAAS,GAAG,QAAQ,CAAC,OAAO;SAC/B,KAAK,CAAC,MAAM,CAAC;SACb,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,KAAK,EAAE,CAAC;SAC/B,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,YAAY,CAAC,KAAK,CAAC,CAAC,CAAC;IACvC,IAAI,SAAS,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC3B,MAAM,IAAI,KAAK,CAAC,gBAAgB,QAAQ,CAAC,IAAI,iBAAiB,CAAC,CAAC;IAClE,CAAC;IAED,MAAM,IAAI,GAAe,EAAE,CAAC;IAC5B,KAAK,MAAM,QAAQ,IAAI,SAAS,EAAE,CAAC;QACjC,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;YAC3B,MAAM,IAAI,KAAK,CACb,gBAAgB,QAAQ,CAAC,IAAI,sBAAsB,QAAQ,CAAC,OAAO,EAAE,CACtE,CAAC;QACJ,CAAC;QACD,2EAA2E;QAC3E,oEAAoE;QACpE,IAAI,CAAC,IAAI,CAAC,EAAE,GAAG,QAAQ,EAAE,IAAI,EAAE,QAAQ,CAAC,IAAI,IAAI,YAAY,EAAE,CAAC,CAAC;IAClE,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAC5B,IAAI,GAAG,CACL,iBAAiB,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC;IAClC,QAAQ,CAAC,IAAI;IACb,mBAAmB,CAAC,QAAQ,CAAC;CAC9B,CAAC,CACH,CAAC"}
+{"version":3,"file":"overrides.js","sourceRoot":"","sources":["../../src/dictionary/overrides.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAiB,MAAM,yBAAyB,CAAC;AACtE,OAAO,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAc/C;;;;;;;;;;;;;;;;;;;GAmBG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAA+B;IAC3D;QACE,IAAI,EAAE,IAAI;QACV,OAAO,EAAE,QAAQ;QACjB,MAAM,EACJ,+HAA+H;KAClI;IACD;QACE,IAAI,EAAE,IAAI;QACV,OAAO,EAAE,OAAO;QAChB,MAAM,EACJ,+UAA+U;KAClV;CACF,CAAC;AAEF;;;;;;;GAOG;AACH,MAAM,UAAU,mBAAmB,CACjC,QAAyB;IAEzB,MAAM,SAAS,GAAG,QAAQ,CAAC,OAAO;SAC/B,KAAK,CAAC,MAAM,CAAC;SACb,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,KAAK,EAAE,CAAC;SAC/B,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,YAAY,CAAC,KAAK,CAAC,CAAC,CAAC;IACvC,IAAI,SAAS,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC3B,MAAM,IAAI,KAAK,CAAC,gBAAgB,QAAQ,CAAC,IAAI,iBAAiB,CAAC,CAAC;IAClE,CAAC;IAED,MAAM,IAAI,GAAe,EAAE,CAAC;IAC5B,KAAK,MAAM,QAAQ,IAAI,SAAS,EAAE,CAAC;QACjC,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;YAC3B,MAAM,IAAI,KAAK,CACb,gBAAgB,QAAQ,CAAC,IAAI,sBAAsB,QAAQ,CAAC,OAAO,EAAE,CACtE,CAAC;QACJ,CAAC;QACD,2EAA2E;QAC3E,oEAAoE;QACpE,IAAI,CAAC,IAAI,CAAC,EAAE,GAAG,QAAQ,EAAE,IAAI,EAAE,QAAQ,CAAC,IAAI,IAAI,YAAY,EAAE,CAAC,CAAC;IAClE,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAC5B,IAAI,GAAG,CACL,iBAAiB,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC;IAClC,QAAQ,CAAC,IAAI;IACb,mBAAmB,CAAC,QAAQ,CAAC;CAC9B,CAAC,CACH,CAAC"}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@kensio/pinyinjs",
-  "version": "1.7.0",
+  "version": "1.7.2",
   "description": "Chinese hanzi to pinyin conversion",
   "repository": "https://github.com/KensioSoftware/pinyinjs",
   "homepage": "https://pinyinjs.dev",