@kensio/pinyinjs 1.9.0 → 1.9.2

package/dist/decode/dominant-rule.js

@@ -0,0 +1,122 @@
+/**
+ * Whether a pair nobody counted may read a function word its rare way.
+ *
+ * The reading decode prices a word by the edge, so a pair the dictionary holds
+ * beats the same two characters read apart whatever the pair says about them.
+ * Some pairs are a sense the language barely uses, held by a source and
+ * counted by no corpus, and they turn a common function word into its rare
+ * reading.
+ */
+import { costOfBucket } from "../dictionary/frequency-table.js";
+import { toCharacters } from "../script/characters.js";
+import { isOnSide, RARE_READINGS } from "./dominant-readings.js";
+import { startsLongerWord, tagOf, wordStartingAt, } from "./rules.js";
+/**
+ * The highest frequency bucket a word nobody counted lands in.
+ *
+ * A key jieba never listed has a count of zero and lands in bucket 0. A key
+ * jieba lists without counting carries its default of 3, and bucket 2 runs
+ * from 3 to 8. {@link import("../dictionary/frequency-table.js").UNCOUNTED_NAME}
+ * gives a name the same default.
+ */
+const UNCOUNTED_BUCKET = 2;
+/**
+ * What an edge in {@link UNCOUNTED_BUCKET} costs. Rarer edges cost more.
+ */
+const UNCOUNTED_COST = costOfBucket(UNCOUNTED_BUCKET);
+/**
+ * Whether two syllables are the same reading, tone and all.
+ */
+function isSame(left, right) {
+    return (right !== undefined &&
+        left.initial === right.initial &&
+        left.final === right.final &&
+        left.tone === right.tone);
+}
+/**
+ * The tag prefix jieba gives a noun.
+ */
+const NOUN_TAG = "n";
+/**
+ * Whether a pair ending in 都 is a capital's name where it stands.
+ *
+ * 都 is `dū` at the end of a capital's name and `dōu` in front of what is said
+ * of everything before it. A name ends the run (大家习惯把上海称为魔都) or
+ * stands in front of a noun (金中都故城). The adverb has a predicate after it,
+ * as in 书架上都是书 and 猫在黑暗中都是灰色的.
+ */
+function isCapital(context) {
+    const { characters, edge } = context;
+    return (edge.to === characters.length ||
+        tagOf(context, wordStartingAt(context, edge.to)).startsWith(NOUN_TAG));
+}
+/**
+ * The tag prefix jieba gives a verb.
+ */
+const VERB_TAG = "v";
+/**
+ * Whether a pair opening with 更 is a compound of its own.
+ *
+ * 更 the adverb raises what follows it: 会更香, 更易使出, 更适用. 更 the verb
+ * *to change* makes compounds with a verb beside it, and 更变 is one no corpus
+ * counted. A bare verb after the 更 is the compound's second half. A verb
+ * opening a longer word (适用) belongs to that word, and the 更 in front of it
+ * is the adverb.
+ */
+function isGengCompound(context) {
+    const { characters, edge } = context;
+    const second = edge.from + 1;
+    return (tagOf(context, characters[second]).startsWith(VERB_TAG) &&
+        !startsLongerWord(context, second));
+}
+/**
+ * Whether a pair reads a function word its rare way on the word's own side.
+ */
+function readsRarely(context, characters) {
+    return characters.some((character, index) => {
+        const rare = RARE_READINGS.get(character);
+        if (rare === undefined || !isOnSide(rare.side, index)) {
+            return false;
+        }
+        if ((character === "都" && isCapital(context)) ||
+            (character === "更" && isGengCompound(context))) {
+            return false;
+        }
+        return rare.readings.some((reading) => isSame(reading, context.edge.reading[index]));
+    });
+}
+/**
+ * A pair nobody counted that reads a function word its rare way is taken off
+ * the lattice.
+ *
+ * The reading decode charges per edge, and that charge is what lets 地气 read
+ * `dì qì` against the 地 and 气 around it. The charge is the same for a pair
+ * no corpus counts. 破的 is CC-CEDICT's *to hit the target*, and it read
+ * 气球胀得太大，会破的 as `huì pò dì`. 更香 is `large_pinyin.txt`'s incense
+ * clock, 家给 is the 家给人足 of an idiom, 赚得 is that file's reading of a word
+ * CC-CEDICT reads `zhuàn dé`, and 上都 and 中都 are the capitals of two
+ * dynasties. Each sat in an everyday sentence and took its 的, 更, 给, 赚 or 都.
+ *
+ * Taking the edge away leaves each character to its own reading, which is the
+ * usual one. A counted word keeps its reading (目的, 首都, 京都), and so does a
+ * pair of more than two characters (知更鸟, 漫无目的地).
+ *
+ * Over 88,866 lines of Tatoeba and zh.wikipedia this changes 25 runs in the
+ * full tier, and all 25 are corrections. Over both CPP splits it changes 32,
+ * and two are wrong (官员俸给 and 北魏国都代郡). CPP's labelled accuracy holds
+ * at 91.86%. The standard and core tiers hold none of these pairs.
+ */
+export const DOMINANT_READING = {
+    name: "dominant-reading",
+    verdictFor: (context) => {
+        const { edge } = context;
+        const characters = toCharacters(edge.text);
+        if (characters.length !== 2 ||
+            edge.reading.length !== characters.length ||
+            edge.cost < UNCOUNTED_COST) {
+            return "keep";
+        }
+        return readsRarely(context, characters) ? "forbid" : "keep";
+    },
+};
+//# sourceMappingURL=dominant-rule.js.map

package/dist/decode/dominant-rule.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"dominant-rule.js","sourceRoot":"","sources":["../../src/decode/dominant-rule.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AACH,OAAO,EAAE,YAAY,EAAE,MAAM,kCAAkC,CAAC;AAChE,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAEvD,OAAO,EAAE,QAAQ,EAAE,aAAa,EAAE,MAAM,wBAAwB,CAAC;AACjE,OAAO,EAGL,gBAAgB,EAChB,KAAK,EACL,cAAc,GACf,MAAM,YAAY,CAAC;AAEpB;;;;;;;GAOG;AACH,MAAM,gBAAgB,GAAG,CAAC,CAAC;AAE3B;;GAEG;AACH,MAAM,cAAc,GAAG,YAAY,CAAC,gBAAgB,CAAC,CAAC;AAEtD;;GAEG;AACH,SAAS,MAAM,CAAC,IAAc,EAAE,KAA2B;IACzD,OAAO,CACL,KAAK,KAAK,SAAS;QACnB,IAAI,CAAC,OAAO,KAAK,KAAK,CAAC,OAAO;QAC9B,IAAI,CAAC,KAAK,KAAK,KAAK,CAAC,KAAK;QAC1B,IAAI,CAAC,IAAI,KAAK,KAAK,CAAC,IAAI,CACzB,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,QAAQ,GAAG,GAAG,CAAC;AAErB;;;;;;;GAOG;AACH,SAAS,SAAS,CAAC,OAAoB;IACrC,MAAM,EAAE,UAAU,EAAE,IAAI,EAAE,GAAG,OAAO,CAAC;IACrC,OAAO,CACL,IAAI,CAAC,EAAE,KAAK,UAAU,CAAC,MAAM;QAC7B,KAAK,CAAC,OAAO,EAAE,cAAc,CAAC,OAAO,EAAE,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC,UAAU,CAAC,QAAQ,CAAC,CACtE,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,QAAQ,GAAG,GAAG,CAAC;AAErB;;;;;;;;GAQG;AACH,SAAS,cAAc,CAAC,OAAoB;IAC1C,MAAM,EAAE,UAAU,EAAE,IAAI,EAAE,GAAG,OAAO,CAAC;IACrC,MAAM,MAAM,GAAG,IAAI,CAAC,IAAI,GAAG,CAAC,CAAC;IAC7B,OAAO,CACL,KAAK,CAAC,OAAO,EAAE,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,UAAU,CAAC,QAAQ,CAAC;QACvD,CAAC,gBAAgB,CAAC,OAAO,EAAE,MAAM,CAAC,CACnC,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,SAAS,WAAW,CAClB,OAAoB,EACpB,UAA6B;IAE7B,OAAO,UAAU,CAAC,IAAI,CAAC,CAAC,SAAS,EAAE,KAAK,EAAE,EAAE;QAC1C,MAAM,IAAI,GAAG,aAAa,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QAC1C,IAAI,IAAI,KAAK,SAAS,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,EAAE,KAAK,CAAC,EAAE,CAAC;YACtD,OAAO,KAAK,CAAC;QACf,CAAC;QACD,IACE,CAAC,SAAS,KAAK,GAAG,IAAI,SAAS,CAAC,OAAO,CAAC,CAAC;YACzC,CAAC,SAAS,KAAK,GAAG,IAAI,cAAc,CAAC,OAAO,CAAC,CAAC,EAC9C,CAAC;YACD,OAAO,KAAK,CAAC;QACf,CAAC;QACD,OAAO,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,OAAO,EAAE,EAAE,CACpC,MAAM,CAAC,OAAO,EAAE,OAAO,CAAC,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAC7C,CAAC;IACJ,CAAC,CAAC,CAAC;AACL,CAAC;AAED;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAa;IACxC,IAAI,EAAE,kBAAkB;IACxB,UAAU,EAAE,CAAC,OAAoB,EAAE,EAAE;QACnC,MAAM,EAAE,IAAI,EAAE,GAAG,OAAO,CAAC;QACzB,MAAM,UAAU,GAAG,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAC3C,IACE,UAAU,CAAC,MAAM,KAAK,CAAC;YACvB,IAAI,CAAC,OAAO,CAAC,MAAM,KAAK,UAAU,CAAC,MAAM;YACzC,IAAI,CAAC,IAAI,GAAG,cAAc,EAC1B,CAAC;YACD,OAAO,MAAM,CAAC;QAChB,CAAC;QACD,OAAO,WAAW,CAAC,OAAO,EAAE,UAAU,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,MAAM,CAAC;IAC9D,CAAC;CACF,CAAC"}

package/dist/decode/guo-rule.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"guo-rule.d.ts","sourceRoot":"","sources":["../../src/decode/guo-rule.ts"],"names":[],"mappings":"AAQA,OAAO,EAEL,KAAK,QAAQ,EAGd,MAAM,YAAY,CAAC;AAoEpB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4DG;AACH,eAAO,MAAM,gBAAgB,EAAE,QAa9B,CAAC"}
+{"version":3,"file":"guo-rule.d.ts","sourceRoot":"","sources":["../../src/decode/guo-rule.ts"],"names":[],"mappings":"AAQA,OAAO,EAEL,KAAK,QAAQ,EAGd,MAAM,YAAY,CAAC;AAiFpB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4DG;AACH,eAAO,MAAM,gBAAgB,EAAE,QAa9B,CAAC"}

package/dist/decode/guo-rule.js

@@ -20,6 +20,15 @@ const MARKER = new Set(["过", "過"]);
  * shape, and jieba tags 检查 `vn`.
  */
 const VERB_TAG = "v";
+/**
+ * The negator jieba tags a verb, in both scripts.
+ *
+ * 没 is *not have* as well as *not*, so the tag is right, and the 过 after it
+ * is still the verb. 考试没过 failed the exam: aspect comes after the verb it
+ * marks, and 没 is the negation in front of one, so 没去过 has its 过 after
+ * 去. 没有过 is untouched, since 有 is the verb there.
+ */
+const NEGATOR = new Set(["没", "沒"]);
 /**
  * Whether a syllable is toneless `guo`, the reading the marker takes.
  */
@@ -36,6 +45,9 @@ function isGuo(syllable) {
  * 巴黎.
  */
 function isVerbEndingAt(context, at) {
+    if (NEGATOR.has(context.characters[at - 1] ?? "")) {
+        return false;
+    }
     return wordsEndingAt(context, at).some((word) => tagOf(context, word).startsWith(VERB_TAG));
 }
 /**

package/dist/decode/guo-rule.js.map

@@ -1 +1 @@
-{"version":3,"file":"guo-rule.js","sourceRoot":"","sources":["../../src/decode/guo-rule.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAEvD,OAAO,EAGL,KAAK,EACL,aAAa,GACd,MAAM,YAAY,CAAC;AAEpB;;;;;GAKG;AACH,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;AAEnC;;;;;GAKG;AACH,MAAM,QAAQ,GAAG,GAAG,CAAC;AAErB;;GAEG;AACH,SAAS,KAAK,CAAC,QAA8B;IAC3C,OAAO,CACL,QAAQ,EAAE,OAAO,KAAK,GAAG,IAAI,QAAQ,CAAC,KAAK,KAAK,IAAI,IAAI,QAAQ,CAAC,IAAI,KAAK,CAAC,CAC5E,CAAC;AACJ,CAAC;AAED;;;;;;;;GAQG;AACH,SAAS,cAAc,CAAC,OAAoB,EAAE,EAAU;IACtD,OAAO,aAAa,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAC9C,KAAK,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC,UAAU,CAAC,QAAQ,CAAC,CAC1C,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;;GAaG;AACH,SAAS,UAAU,CAAC,OAAoB,EAAE,IAAyB;IACjE,IAAI,IAAI,CAAC,EAAE,GAAG,IAAI,CAAC,IAAI,KAAK,CAAC,IAAI,IAAI,CAAC,OAAO,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC3D,OAAO,KAAK,CAAC;IACf,CAAC;IACD,OAAO,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,CACjC,CAAC,SAAS,EAAE,MAAM,EAAE,EAAE,CACpB,MAAM,CAAC,GAAG,CAAC,SAAS,CAAC;QACrB,CAAC,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;QAC5B,cAAc,CAAC,OAAO,EAAE,IAAI,CAAC,IAAI,GAAG,MAAM,CAAC,CAC9C,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4DG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAa;IACxC,IAAI,EAAE,kBAAkB;IACxB,UAAU,EAAE,CAAC,OAAoB,EAAE,EAAE;QACnC,MAAM,EAAE,IAAI,EAAE,GAAG,OAAO,CAAC;QACzB,IAAI,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YAC1B,OAAO,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,IAAI,cAAc,CAAC,OAAO,EAAE,IAAI,CAAC,IAAI,CAAC;gBACjE,CAAC,CAAC,OAAO;gBACT,CAAC,CAAC,MAAM,CAAC;QACb,CAAC;QACD,OAAO,IAAI,CAAC,YAAY,KAAK,EAAE,IAAI,UAAU,CAAC,OAAO,EAAE,IAAI,CAAC;YAC1D,CAAC,CAAC,QAAQ;YACV,CAAC,CAAC,MAAM,CAAC;IACb,CAAC;CACF,CAAC"}
+{"version":3,"file":"guo-rule.js","sourceRoot":"","sources":["../../src/decode/guo-rule.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAEvD,OAAO,EAGL,KAAK,EACL,aAAa,GACd,MAAM,YAAY,CAAC;AAEpB;;;;;GAKG;AACH,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;AAEnC;;;;;GAKG;AACH,MAAM,QAAQ,GAAG,GAAG,CAAC;AAErB;;;;;;;GAOG;AACH,MAAM,OAAO,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;AAEpC;;GAEG;AACH,SAAS,KAAK,CAAC,QAA8B;IAC3C,OAAO,CACL,QAAQ,EAAE,OAAO,KAAK,GAAG,IAAI,QAAQ,CAAC,KAAK,KAAK,IAAI,IAAI,QAAQ,CAAC,IAAI,KAAK,CAAC,CAC5E,CAAC;AACJ,CAAC;AAED;;;;;;;;GAQG;AACH,SAAS,cAAc,CAAC,OAAoB,EAAE,EAAU;IACtD,IAAI,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,UAAU,CAAC,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC;QAClD,OAAO,KAAK,CAAC;IACf,CAAC;IACD,OAAO,aAAa,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAC9C,KAAK,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC,UAAU,CAAC,QAAQ,CAAC,CAC1C,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;;GAaG;AACH,SAAS,UAAU,CAAC,OAAoB,EAAE,IAAyB;IACjE,IAAI,IAAI,CAAC,EAAE,GAAG,IAAI,CAAC,IAAI,KAAK,CAAC,IAAI,IAAI,CAAC,OAAO,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC3D,OAAO,KAAK,CAAC;IACf,CAAC;IACD,OAAO,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,CACjC,CAAC,SAAS,EAAE,MAAM,EAAE,EAAE,CACpB,MAAM,CAAC,GAAG,CAAC,SAAS,CAAC;QACrB,CAAC,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;QAC5B,cAAc,CAAC,OAAO,EAAE,IAAI,CAAC,IAAI,GAAG,MAAM,CAAC,CAC9C,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4DG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAa;IACxC,IAAI,EAAE,kBAAkB;IACxB,UAAU,EAAE,CAAC,OAAoB,EAAE,EAAE;QACnC,MAAM,EAAE,IAAI,EAAE,GAAG,OAAO,CAAC;QACzB,IAAI,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YAC1B,OAAO,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,IAAI,cAAc,CAAC,OAAO,EAAE,IAAI,CAAC,IAAI,CAAC;gBACjE,CAAC,CAAC,OAAO;gBACT,CAAC,CAAC,MAAM,CAAC;QACb,CAAC;QACD,OAAO,IAAI,CAAC,YAAY,KAAK,EAAE,IAAI,UAAU,CAAC,OAAO,EAAE,IAAI,CAAC;YAC1D,CAAC,CAAC,QAAQ;YACV,CAAC,CAAC,MAAM,CAAC;IACb,CAAC;CACF,CAAC"}

package/dist/decode/hao-rule.d.ts

@@ -0,0 +1,16 @@
+import { type EdgeRule } from "./rules.js";
+/**
+ * 好学 taken apart where it means easy to learn.
+ *
+ * 德语好学吗 and 法语好学吗 came out `hào xué`, asking whether German is
+ * studious. Taking the edge away leaves 好 and 学 to their own readings,
+ * `hǎo` and `xué`. 聪明好学, 勤而好学 and 沉静好学 are said of a person and
+ * keep the word.
+ *
+ * Over 88,866 lines of Tatoeba and zh.wikipedia this changes 5 readings in the
+ * full tier and 3 in the standard tier, and all are corrections: 德语好学吗,
+ * 義大利語好學嗎 and 只要你好好学中文 among them. Nothing moves in the core tier
+ * or on either CPP split.
+ */
+export declare const EASY_HAO: EdgeRule;
+//# sourceMappingURL=hao-rule.d.ts.map

package/dist/decode/hao-rule.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"hao-rule.d.ts","sourceRoot":"","sources":["../../src/decode/hao-rule.ts"],"names":[],"mappings":"AAQA,OAAO,EAAoB,KAAK,QAAQ,EAAiB,MAAM,YAAY,CAAC;AAsC5E;;;;;;;;;;;;GAYG;AACH,eAAO,MAAM,QAAQ,EAAE,QAQtB,CAAC"}

package/dist/decode/hao-rule.js

@@ -0,0 +1,61 @@
+/**
+ * Whether a 好学 is studious or is easy to learn.
+ *
+ * The dictionary holds 好学 as `hào xué`, *studious*, the sense jieba counted.
+ * CC-CEDICT also gives `hǎo xué`, *easy to learn*, and the merge kept the
+ * first. What decides between them is what the 好学 is said of.
+ */
+import { toCharacters } from "../script/characters.js";
+import { wordsEndingAt } from "./rules.js";
+/**
+ * The word, in both scripts.
+ */
+const STUDIOUS = new Set(["好学", "好學"]);
+/**
+ * The characters that close the name of a language.
+ *
+ * 德语, 法语, 拉丁语, 中文 and 英文. A language is learned and cannot be
+ * studious, so the 好学 after one is easy to learn.
+ */
+const LANGUAGE_ENDINGS = new Set(["语", "語", "文"]);
+/**
+ * The character a reduplicated 好好 repeats.
+ */
+const GOOD = "好";
+/**
+ * Whether the 好学 starting at a position is said of a language, or is the
+ * second 好 of 好好.
+ *
+ * 只要你好好学中文 has 好好 and then 学. A 好 in front of the 好学 belongs to
+ * the first 好, and the pair would read it `hǎohào xué`.
+ */
+function isEasyToLearnAt(context, at) {
+    if (context.characters[at - 1] === GOOD) {
+        return true;
+    }
+    return wordsEndingAt(context, at).some((word) => toCharacters(word).length > 1 &&
+        LANGUAGE_ENDINGS.has(toCharacters(word).at(-1) ?? ""));
+}
+/**
+ * 好学 taken apart where it means easy to learn.
+ *
+ * 德语好学吗 and 法语好学吗 came out `hào xué`, asking whether German is
+ * studious. Taking the edge away leaves 好 and 学 to their own readings,
+ * `hǎo` and `xué`. 聪明好学, 勤而好学 and 沉静好学 are said of a person and
+ * keep the word.
+ *
+ * Over 88,866 lines of Tatoeba and zh.wikipedia this changes 5 readings in the
+ * full tier and 3 in the standard tier, and all are corrections: 德语好学吗,
+ * 義大利語好學嗎 and 只要你好好学中文 among them. Nothing moves in the core tier
+ * or on either CPP split.
+ */
+export const EASY_HAO = {
+    name: "easy-hao",
+    verdictFor: (context) => {
+        const { edge } = context;
+        return STUDIOUS.has(edge.text) && isEasyToLearnAt(context, edge.from)
+            ? "forbid"
+            : "keep";
+    },
+};
+//# sourceMappingURL=hao-rule.js.map

package/dist/decode/hao-rule.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"hao-rule.js","sourceRoot":"","sources":["../../src/decode/hao-rule.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AACvD,OAAO,EAAmC,aAAa,EAAE,MAAM,YAAY,CAAC;AAE5E;;GAEG;AACH,MAAM,QAAQ,GAAG,IAAI,GAAG,CAAC,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC,CAAC;AAEvC;;;;;GAKG;AACH,MAAM,gBAAgB,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;AAElD;;GAEG;AACH,MAAM,IAAI,GAAG,GAAG,CAAC;AAEjB;;;;;;GAMG;AACH,SAAS,eAAe,CAAC,OAAoB,EAAE,EAAU;IACvD,IAAI,OAAO,CAAC,UAAU,CAAC,EAAE,GAAG,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC;QACxC,OAAO,IAAI,CAAC;IACd,CAAC;IACD,OAAO,aAAa,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC,IAAI,CACpC,CAAC,IAAI,EAAE,EAAE,CACP,YAAY,CAAC,IAAI,CAAC,CAAC,MAAM,GAAG,CAAC;QAC7B,gBAAgB,CAAC,GAAG,CAAC,YAAY,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CACxD,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;GAYG;AACH,MAAM,CAAC,MAAM,QAAQ,GAAa;IAChC,IAAI,EAAE,UAAU;IAChB,UAAU,EAAE,CAAC,OAAoB,EAAE,EAAE;QACnC,MAAM,EAAE,IAAI,EAAE,GAAG,OAAO,CAAC;QACzB,OAAO,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,eAAe,CAAC,OAAO,EAAE,IAAI,CAAC,IAAI,CAAC;YACnE,CAAC,CAAC,QAAQ;YACV,CAAC,CAAC,MAAM,CAAC;IACb,CAAC;CACF,CAAC"}

package/dist/decode/hua-rule.d.ts

@@ -0,0 +1,17 @@
+import { type EdgeRule } from "./rules.js";
+/**
+ * 的话 is taken off the lattice where the 话 is the noun its 的 modifies.
+ *
+ * The conditional is the default, since it is what the word is, and the rule
+ * names the contexts where it cannot be: a 吗 or 里 after the 话, a time the
+ * words were said, a verb of speaking before the 的, a kind of words that verb
+ * is saying, a 句 counting them, or a person whose words they are. Taking the edge away leaves 的 and 话
+ * to read as themselves, which they already did, so this changes the spacing
+ * and the segmentation rather than any syllable: `shuō de huà`, not
+ * `shuō dehuà`.
+ *
+ * Over the 472 distinct lines of Tatoeba, zh.wikipedia and the Chinese Boost
+ * example sentences that hold 的话, the full tier read 463 runs as the word.
+ */
+export declare const NOUN_HUA: EdgeRule;
+//# sourceMappingURL=hua-rule.d.ts.map

package/dist/decode/hua-rule.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"hua-rule.d.ts","sourceRoot":"","sources":["../../src/decode/hua-rule.ts"],"names":[],"mappings":"AAUA,OAAO,EAEL,KAAK,QAAQ,EAId,MAAM,YAAY,CAAC;AAsDpB;;;;;;;;;;;;;GAaG;AACH,eAAO,MAAM,QAAQ,EAAE,QAkBtB,CAAC"}

package/dist/decode/hua-rule.js

@@ -0,0 +1,83 @@
+/**
+ * The 话 after a structural 的, which is the noun *words* and not half of 的话.
+ *
+ * 的话 is a word, the clause-final *if* of 要是下雨的话, and every source counts
+ * it. The same two characters are just as often the particle 的 and the noun 话
+ * it modifies: 他说的话, 听妈妈的话. The decode had no way to tell the two
+ * apart and read every one of them as the conditional, so 你听得懂从我嘴里出来的话吗
+ * came out `chūlái dehuà ma`.
+ */
+import { toCharacters } from "../script/characters.js";
+import { tagOf, wordEndingAt, wordsEndingAt, } from "./rules.js";
+import { DE_HUA, AFTER_NOUN, PERSONS, OWNER_TAGS, RECEIVING, NEGATION, RECEIVING_TAIL, EARLIER, } from "./hua-words.js";
+import { at, followsSpeaking, isCounted, saysKind } from "./hua-speaking.js";
+/**
+ * Whether the 的 marks whose words they are, 他的话 and 听老师的话.
+ *
+ * The owner is a person, and it either opens the run with something said about
+ * the words after them (他的话让我吃惊) or is the object of a verb that takes
+ * words (听妈妈的话, 相信你的话, 听她叔叔的话). A noun owner is a word of two
+ * characters or more, since 热的话就脱衣服 is tagged as a noun and is the
+ * conditional.
+ *
+ * A negated verb counts as the verb. 不听 is a word of its own, and
+ * 为什么不听我的话 has to find 听 in it.
+ */
+function isOwned(context, from, to) {
+    const owner = wordEndingAt(context, from, 3);
+    if (owner === undefined) {
+        return false;
+    }
+    const isPerson = PERSONS.has(owner);
+    if (!isPerson &&
+        (toCharacters(owner).length < 2 || !OWNER_TAGS.test(tagOf(context, owner)))) {
+        return false;
+    }
+    let start = from - toCharacters(owner).length;
+    if (start === 0) {
+        return to < context.characters.length;
+    }
+    const possessor = wordEndingAt(context, start, 2);
+    if (!isPerson && possessor !== undefined && PERSONS.has(possessor)) {
+        start -= toCharacters(possessor).length;
+    }
+    let verb = start;
+    while (verb > start - 2 && RECEIVING_TAIL.has(at(context, verb - 1))) {
+        verb -= 1;
+    }
+    return wordsEndingAt(context, verb).some((word) => RECEIVING.has(word) || RECEIVING.has(word.replace(NEGATION, "")));
+}
+/**
+ * 的话 is taken off the lattice where the 话 is the noun its 的 modifies.
+ *
+ * The conditional is the default, since it is what the word is, and the rule
+ * names the contexts where it cannot be: a 吗 or 里 after the 话, a time the
+ * words were said, a verb of speaking before the 的, a kind of words that verb
+ * is saying, a 句 counting them, or a person whose words they are. Taking the edge away leaves 的 and 话
+ * to read as themselves, which they already did, so this changes the spacing
+ * and the segmentation rather than any syllable: `shuō de huà`, not
+ * `shuō dehuà`.
+ *
+ * Over the 472 distinct lines of Tatoeba, zh.wikipedia and the Chinese Boost
+ * example sentences that hold 的话, the full tier read 463 runs as the word.
+ */
+export const NOUN_HUA = {
+    name: "noun-hua",
+    verdictFor: (context) => {
+        const { edge } = context;
+        if (!DE_HUA.has(edge.text)) {
+            return "keep";
+        }
+        const { from, to } = edge;
+        const before = wordEndingAt(context, from, 2);
+        return AFTER_NOUN.has(at(context, to)) ||
+            (before !== undefined && EARLIER.has(before)) ||
+            followsSpeaking(context, from) ||
+            saysKind(context, from) ||
+            isCounted(context, from) ||
+            isOwned(context, from, to)
+            ? "forbid"
+            : "keep";
+    },
+};
+//# sourceMappingURL=hua-rule.js.map

package/dist/decode/hua-rule.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"hua-rule.js","sourceRoot":"","sources":["../../src/decode/hua-rule.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AACH,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AACvD,OAAO,EAGL,KAAK,EACL,YAAY,EACZ,aAAa,GACd,MAAM,YAAY,CAAC;AACpB,OAAO,EACL,MAAM,EACN,UAAU,EACV,OAAO,EACP,UAAU,EACV,SAAS,EACT,QAAQ,EACR,cAAc,EACd,OAAO,GACR,MAAM,gBAAgB,CAAC;AACxB,OAAO,EAAE,EAAE,EAAE,eAAe,EAAE,SAAS,EAAE,QAAQ,EAAE,MAAM,mBAAmB,CAAC;AAE7E;;;;;;;;;;;GAWG;AACH,SAAS,OAAO,CAAC,OAAoB,EAAE,IAAY,EAAE,EAAU;IAC7D,MAAM,KAAK,GAAG,YAAY,CAAC,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC;IAC7C,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;QACxB,OAAO,KAAK,CAAC;IACf,CAAC;IACD,MAAM,QAAQ,GAAG,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;IACpC,IACE,CAAC,QAAQ;QACT,CAAC,YAAY,CAAC,KAAK,CAAC,CAAC,MAAM,GAAG,CAAC,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC,KAAK,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC,CAAC,EAC3E,CAAC;QACD,OAAO,KAAK,CAAC;IACf,CAAC;IACD,IAAI,KAAK,GAAG,IAAI,GAAG,YAAY,CAAC,KAAK,CAAC,CAAC,MAAM,CAAC;IAC9C,IAAI,KAAK,KAAK,CAAC,EAAE,CAAC;QAChB,OAAO,EAAE,GAAG,OAAO,CAAC,UAAU,CAAC,MAAM,CAAC;IACxC,CAAC;IACD,MAAM,SAAS,GAAG,YAAY,CAAC,OAAO,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC;IAClD,IAAI,CAAC,QAAQ,IAAI,SAAS,KAAK,SAAS,IAAI,OAAO,CAAC,GAAG,CAAC,SAAS,CAAC,EAAE,CAAC;QACnE,KAAK,IAAI,YAAY,CAAC,SAAS,CAAC,CAAC,MAAM,CAAC;IAC1C,CAAC;IACD,IAAI,IAAI,GAAG,KAAK,CAAC;IACjB,OAAO,IAAI,GAAG,KAAK,GAAG,CAAC,IAAI,cAAc,CAAC,GAAG,CAAC,EAAE,CAAC,OAAO,EAAE,IAAI,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC;QACrE,IAAI,IAAI,CAAC,CAAC;IACZ,CAAC;IACD,OAAO,aAAa,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC,IAAI,CACtC,CAAC,IAAI,EAAE,EAAE,CAAC,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC,CAC3E,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;;GAaG;AACH,MAAM,CAAC,MAAM,QAAQ,GAAa;IAChC,IAAI,EAAE,UAAU;IAChB,UAAU,EAAE,CAAC,OAAoB,EAAE,EAAE;QACnC,MAAM,EAAE,IAAI,EAAE,GAAG,OAAO,CAAC;QACzB,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YAC3B,OAAO,MAAM,CAAC;QAChB,CAAC;QACD,MAAM,EAAE,IAAI,EAAE,EAAE,EAAE,GAAG,IAAI,CAAC;QAC1B,MAAM,MAAM,GAAG,YAAY,CAAC,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC;QAC9C,OAAO,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC;YACpC,CAAC,MAAM,KAAK,SAAS,IAAI,OAAO,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;YAC7C,eAAe,CAAC,OAAO,EAAE,IAAI,CAAC;YAC9B,QAAQ,CAAC,OAAO,EAAE,IAAI,CAAC;YACvB,SAAS,CAAC,OAAO,EAAE,IAAI,CAAC;YACxB,OAAO,CAAC,OAAO,EAAE,IAAI,EAAE,EAAE,CAAC;YAC1B,CAAC,CAAC,QAAQ;YACV,CAAC,CAAC,MAAM,CAAC;IACb,CAAC;CACF,CAAC"}

package/dist/decode/hua-speaking.d.ts

@@ -0,0 +1,29 @@
+import { type EdgeContext } from "./rules.js";
+/**
+ * The character at a position, or the empty string beyond either end.
+ */
+export declare function at(context: EdgeContext, index: number): string;
+/**
+ * Whether the 的 is the object marker of a verb of speaking.
+ *
+ * 说的话, 所说的话, 说过的话 and 说出来的话 all qualify, unless a manner stands in
+ * front of the verb or somebody is asked for an opinion: 这么说的话 is *if you
+ * put it like that* and 要我说的话 is *if you ask me*.
+ */
+export declare function followsSpeaking(context: EdgeContext, from: number): boolean;
+/**
+ * Whether the 的 follows a kind of words a verb of speaking is saying.
+ *
+ * 说这样的话, 说过那样的话 and 讲同样的话.
+ */
+export declare function saysKind(context: EdgeContext, from: number): boolean;
+/**
+ * Whether a 句, or a 些 after a verb of speaking, counts the words.
+ *
+ * 句 is the classifier of 话 and of little else a 的话 could follow, so one in
+ * the few characters before the 的 is counting the head: 一句感谢的话,
+ * 说句简单的话. 些 counts anything, 有些人不来的话 among them, so it counts the
+ * words only when it is what is said: 说些傻乎乎的话, 说一些下流的话.
+ */
+export declare function isCounted(context: EdgeContext, from: number): boolean;
+//# sourceMappingURL=hua-speaking.d.ts.map

package/dist/decode/hua-speaking.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"hua-speaking.d.ts","sourceRoot":"","sources":["../../src/decode/hua-speaking.ts"],"names":[],"mappings":"AAKA,OAAO,EAAE,KAAK,WAAW,EAAgB,MAAM,YAAY,CAAC;AAU5D;;GAEG;AACH,wBAAgB,EAAE,CAAC,OAAO,EAAE,WAAW,EAAE,KAAK,EAAE,MAAM,GAAG,MAAM,CAE9D;AAED;;;;;;GAMG;AACH,wBAAgB,eAAe,CAAC,OAAO,EAAE,WAAW,EAAE,IAAI,EAAE,MAAM,GAAG,OAAO,CAiB3E;AAED;;;;GAIG;AACH,wBAAgB,QAAQ,CAAC,OAAO,EAAE,WAAW,EAAE,IAAI,EAAE,MAAM,GAAG,OAAO,CAUpE;AAED;;;;;;;GAOG;AACH,wBAAgB,SAAS,CAAC,OAAO,EAAE,WAAW,EAAE,IAAI,EAAE,MAAM,GAAG,OAAO,CAerE"}

package/dist/decode/hua-speaking.js

@@ -0,0 +1,79 @@
+/**
+ * What a verb of speaking says about the 话 after a 的: 说的话, 说这样的话,
+ * 说句简单的话.
+ */
+import { toCharacters } from "../script/characters.js";
+import { wordEndingAt } from "./rules.js";
+import { SPEAKING, NOT_SPEAKING, ASKING, SPEAKING_TAIL, MANNER, KIND, } from "./hua-words.js";
+/**
+ * The character at a position, or the empty string beyond either end.
+ */
+export function at(context, index) {
+    return context.characters[index] ?? "";
+}
+/**
+ * Whether the 的 is the object marker of a verb of speaking.
+ *
+ * 说的话, 所说的话, 说过的话 and 说出来的话 all qualify, unless a manner stands in
+ * front of the verb or somebody is asked for an opinion: 这么说的话 is *if you
+ * put it like that* and 要我说的话 is *if you ask me*.
+ */
+export function followsSpeaking(context, from) {
+    let verb = from - 1;
+    while (verb > from - 4 && SPEAKING_TAIL.has(at(context, verb))) {
+        verb -= 1;
+    }
+    if (!SPEAKING.has(at(context, verb))) {
+        return false;
+    }
+    const word = wordEndingAt(context, verb + 1, 2);
+    if (word !== undefined && NOT_SPEAKING.has(word)) {
+        return false;
+    }
+    if (ASKING.has(at(context, verb - 2)) && at(context, verb - 1) === "我") {
+        return false;
+    }
+    const manner = wordEndingAt(context, verb, 2);
+    return manner === undefined || !MANNER.has(manner);
+}
+/**
+ * Whether the 的 follows a kind of words a verb of speaking is saying.
+ *
+ * 说这样的话, 说过那样的话 and 讲同样的话.
+ */
+export function saysKind(context, from) {
+    const kind = wordEndingAt(context, from, 2);
+    if (kind === undefined || !KIND.has(kind)) {
+        return false;
+    }
+    let verb = from - toCharacters(kind).length - 1;
+    if (SPEAKING_TAIL.has(at(context, verb))) {
+        verb -= 1;
+    }
+    return SPEAKING.has(at(context, verb));
+}
+/**
+ * Whether a 句, or a 些 after a verb of speaking, counts the words.
+ *
+ * 句 is the classifier of 话 and of little else a 的话 could follow, so one in
+ * the few characters before the 的 is counting the head: 一句感谢的话,
+ * 说句简单的话. 些 counts anything, 有些人不来的话 among them, so it counts the
+ * words only when it is what is said: 说些傻乎乎的话, 说一些下流的话.
+ */
+export function isCounted(context, from) {
+    for (let index = from - 1; index >= Math.max(0, from - 4); index--) {
+        const character = at(context, index);
+        if (character === "的") {
+            return false;
+        }
+        if (character === "句") {
+            return true;
+        }
+        if (character === "些") {
+            const verb = at(context, index - 1) === "一" ? index - 2 : index - 1;
+            return SPEAKING.has(at(context, verb));
+        }
+    }
+    return false;
+}
+//# sourceMappingURL=hua-speaking.js.map

package/dist/decode/hua-speaking.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"hua-speaking.js","sourceRoot":"","sources":["../../src/decode/hua-speaking.ts"],"names":[],"mappings":"AAAA;;;GAGG;AACH,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AACvD,OAAO,EAAoB,YAAY,EAAE,MAAM,YAAY,CAAC;AAC5D,OAAO,EACL,QAAQ,EACR,YAAY,EACZ,MAAM,EACN,aAAa,EACb,MAAM,EACN,IAAI,GACL,MAAM,gBAAgB,CAAC;AAExB;;GAEG;AACH,MAAM,UAAU,EAAE,CAAC,OAAoB,EAAE,KAAa;IACpD,OAAO,OAAO,CAAC,UAAU,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC;AACzC,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,eAAe,CAAC,OAAoB,EAAE,IAAY;IAChE,IAAI,IAAI,GAAG,IAAI,GAAG,CAAC,CAAC;IACpB,OAAO,IAAI,GAAG,IAAI,GAAG,CAAC,IAAI,aAAa,CAAC,GAAG,CAAC,EAAE,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC,EAAE,CAAC;QAC/D,IAAI,IAAI,CAAC,CAAC;IACZ,CAAC;IACD,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC,EAAE,CAAC;QACrC,OAAO,KAAK,CAAC;IACf,CAAC;IACD,MAAM,IAAI,GAAG,YAAY,CAAC,OAAO,EAAE,IAAI,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC;IAChD,IAAI,IAAI,KAAK,SAAS,IAAI,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;QACjD,OAAO,KAAK,CAAC;IACf,CAAC;IACD,IAAI,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,OAAO,EAAE,IAAI,GAAG,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,OAAO,EAAE,IAAI,GAAG,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;QACvE,OAAO,KAAK,CAAC;IACf,CAAC;IACD,MAAM,MAAM,GAAG,YAAY,CAAC,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC;IAC9C,OAAO,MAAM,KAAK,SAAS,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;AACrD,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,QAAQ,CAAC,OAAoB,EAAE,IAAY;IACzD,MAAM,IAAI,GAAG,YAAY,CAAC,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC;IAC5C,IAAI,IAAI,KAAK,SAAS,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;QAC1C,OAAO,KAAK,CAAC;IACf,CAAC;IACD,IAAI,IAAI,GAAG,IAAI,GAAG,YAAY,CAAC,IAAI,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC;IAChD,IAAI,aAAa,CAAC,GAAG,CAAC,EAAE,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC,EAAE,CAAC;QACzC,IAAI,IAAI,CAAC,CAAC;IACZ,CAAC;IACD,OAAO,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC,CAAC;AACzC,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,SAAS,CAAC,OAAoB,EAAE,IAAY;IAC1D,KAAK,IAAI,KAAK,GAAG,IAAI,GAAG,CAAC,EAAE,KAAK,IAAI,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,GAAG,CAAC,CAAC,EAAE,KAAK,EAAE,EAAE,CAAC;QACnE,MAAM,SAAS,GAAG,EAAE,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC;QACrC,IAAI,SAAS,KAAK,GAAG,EAAE,CAAC;YACtB,OAAO,KAAK,CAAC;QACf,CAAC;QACD,IAAI,SAAS,KAAK,GAAG,EAAE,CAAC;YACtB,OAAO,IAAI,CAAC;QACd,CAAC;QACD,IAAI,SAAS,KAAK,GAAG,EAAE,CAAC;YACtB,MAAM,IAAI,GAAG,EAAE,CAAC,OAAO,EAAE,KAAK,GAAG,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC;YACpE,OAAO,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC,CAAC;QACzC,CAAC;IACH,CAAC;IACD,OAAO,KAAK,CAAC;AACf,CAAC"}

package/dist/decode/hua-words.d.ts

@@ -0,0 +1,73 @@
+/**
+ * The words {@link import("./hua-rule.js").NOUN_HUA} reads the context of a
+ * 的话 by, in both scripts.
+ */
+/** The two writings of the word, which are all this rule looks at. */
+export declare const DE_HUA: Set<string>;
+/**
+ * What follows the noun 话 and cannot follow the conditional.
+ *
+ * A conditional clause is followed by the clause it conditions, so it is never
+ * the thing a 吗 asks about and has no inside for 里 to name: 出来的话吗,
+ * 他的话里有文章.
+ */
+export declare const AFTER_NOUN: Set<string>;
+/** The verbs of speaking whose object 话 is: 说的话, 讲过的话. */
+export declare const SPEAKING: Set<string>;
+/**
+ * Words ending in a verb of speaking that are not that verb.
+ *
+ * 听讲 is attending a lecture and 小说 is a novel, and 去听讲的话 is a
+ * conditional.
+ */
+export declare const NOT_SPEAKING: Set<string>;
+/**
+ * What asks for an opinion before 我说, 要我说的话 being *if you ask me*.
+ */
+export declare const ASKING: Set<string>;
+/**
+ * What can stand between a verb of speaking and the 的 of its object.
+ *
+ * 过, 出来, 了 and 不完 are the verb's own, and 所 makes the verb the modifier.
+ */
+export declare const SPEAKING_TAIL: Set<string>;
+/**
+ * The manners a speaker can ask after, which make 说的话 the conditional.
+ *
+ * 我这么说的话是不是像个傻瓜 is *if I put it like this*, and the 说 here is
+ * saying it that way rather than saying the words.
+ */
+export declare const MANNER: Set<string>;
+/**
+ * The modifiers that pick out words of a kind after a verb of speaking.
+ *
+ * 说这样的话 is saying such things, where 这样的话 alone is *in that case*.
+ */
+export declare const KIND: Set<string>;
+/**
+ * The pronouns that make 的话 someone's words: 他的话, 听妈妈的话.
+ *
+ * Personal pronouns only. 这样 is tagged a pronoun as well, and 这样的话 is the
+ * conditional far more often than it is anything else.
+ */
+export declare const PERSONS: Set<string>;
+/** The tags of a noun that can own its words: 老师, 汤姆. */
+export declare const OWNER_TAGS: RegExp;
+/**
+ * The verbs whose object is somebody's words, 听他的话 and 相信你的话.
+ *
+ * Each takes what was said rather than an action, which is why 帮助我的话 and
+ * 陪我的话 stay conditional while 相信我的话 does not. 把, 对 and 因为 are in for
+ * the same reason: 把你的话转告给他, 对我的话充耳不闻 and 因为他的话 have the
+ * words as their object.
+ */
+export declare const RECEIVING: Set<string>;
+/** What can negate a receiving verb inside a word of its own: 不听, 没听. */
+export declare const NEGATION: RegExp;
+/** The characters a receiving verb can carry before its object: 听了, 听完了. */
+export declare const RECEIVING_TAIL: Set<string>;
+/**
+ * When the words were said, which a conditional cannot be about: 刚才的话.
+ */
+export declare const EARLIER: Set<string>;
+//# sourceMappingURL=hua-words.d.ts.map

package/dist/decode/hua-words.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"hua-words.d.ts","sourceRoot":"","sources":["../../src/decode/hua-words.ts"],"names":[],"mappings":"AAAA;;;GAGG;AAEH,sEAAsE;AACtE,eAAO,MAAM,MAAM,aAAwB,CAAC;AAE5C;;;;;;GAMG;AACH,eAAO,MAAM,UAAU,aAAqC,CAAC;AAE7D,0DAA0D;AAC1D,eAAO,MAAM,QAAQ,aAAgC,CAAC;AAEtD;;;;;GAKG;AACH,eAAO,MAAM,YAAY,aASvB,CAAC;AAEH;;GAEG;AACH,eAAO,MAAM,MAAM,aAAgC,CAAC;AAEpD;;;;GAIG;AACH,eAAO,MAAM,aAAa,aASxB,CAAC;AAEH;;;;;GAKG;AACH,eAAO,MAAM,MAAM,aASjB,CAAC;AAEH;;;;GAIG;AACH,eAAO,MAAM,IAAI,aAQf,CAAC;AAEH;;;;;GAKG;AACH,eAAO,MAAM,OAAO,aAoBlB,CAAC;AAEH,yDAAyD;AACzD,eAAO,MAAM,UAAU,QAAyB,CAAC;AAEjD;;;;;;;GAOG;AACH,eAAO,MAAM,SAAS,aAiCpB,CAAC;AAEH,yEAAyE;AACzE,eAAO,MAAM,QAAQ,QAAY,CAAC;AAElC,4EAA4E;AAC5E,eAAO,MAAM,cAAc,aAAsB,CAAC;AAElD;;GAEG;AACH,eAAO,MAAM,OAAO,aAAoC,CAAC"}