@kensio/pinyinjs 1.7.6 → 1.7.8

package/dist/dictionary/measure-tone.d.ts

@@ -0,0 +1,36 @@
+import { type Tone } from "../tone/tone.js";
+import type { DictionaryEntry } from "./entry.js";
+/**
+ * The determiners that take 个 into a word of their own, where it is 轻声.
+ *
+ * Six characters and not a part of speech, because the tag cannot draw the
+ * line. jieba tags 这个, 那个 and 每个 `r` and tags 各个, 某个 and 整个 `r` as
+ * well, and those three are `gègè`, `mǒugè` and `zhěnggè` in both sources with
+ * nothing disagreeing. So this is a list of judgements, in the way
+ * {@link import("./neutral-senses.js").NEUTRAL_SENSE_WORDS} is, and the words
+ * it decides are the ones where a source writes the 轻声 on the determiner and
+ * the pair together.
+ */
+export declare const NEUTRAL_MEASURE_DETERMINERS: ReadonlySet<string>;
+/**
+ * The tone the 个 at a position takes, or undefined where nothing settles it.
+ */
+export declare function countingToneAt(characters: readonly string[], at: number): Tone | undefined;
+/**
+ * What the settling pass changed.
+ */
+export interface MeasureToneSettlement {
+    readonly entries: readonly DictionaryEntry[];
+    /** Entries whose 个 was retoned. */
+    readonly settled: number;
+}
+/**
+ * Give every 个 the tone its position calls for.
+ *
+ * Runs over the finished entries, before the constituent repair, so that a
+ * phrase held to the words inside it is held to a settled reading. Only the
+ * 普通话 reading is touched. A zh-TW delta is a claim about how 國語 differs
+ * from what stands beside it, and none of the 279 entries holding a 个 has one.
+ */
+export declare function settleMeasureTones(entries: readonly DictionaryEntry[]): MeasureToneSettlement;
+//# sourceMappingURL=measure-tone.d.ts.map

package/dist/dictionary/measure-tone.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"measure-tone.d.ts","sourceRoot":"","sources":["../../src/dictionary/measure-tone.ts"],"names":[],"mappings":"AA6CA,OAAO,EAAgB,KAAK,IAAI,EAAE,MAAM,iBAAiB,CAAC;AAC1D,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,YAAY,CAAC;AAYlD;;;;;;;;;;GAUG;AACH,eAAO,MAAM,2BAA2B,EAAE,WAAW,CAAC,MAAM,CAQ1D,CAAC;AAgBH;;GAEG;AACH,wBAAgB,cAAc,CAC5B,UAAU,EAAE,SAAS,MAAM,EAAE,EAC7B,EAAE,EAAE,MAAM,GACT,IAAI,GAAG,SAAS,CAQlB;AAED;;GAEG;AACH,MAAM,WAAW,qBAAqB;IACpC,QAAQ,CAAC,OAAO,EAAE,SAAS,eAAe,EAAE,CAAC;IAC7C,mCAAmC;IACnC,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;CAC1B;AAED;;;;;;;GAOG;AACH,wBAAgB,kBAAkB,CAChC,OAAO,EAAE,SAAS,eAAe,EAAE,GAClC,qBAAqB,CAsCvB"}

package/dist/dictionary/measure-tone.js

@@ -0,0 +1,139 @@
+/**
+ * Settling the tone of 个, which no source settles for itself.
+ *
+ * Every source writes 个 both ways, and where one writes the 轻声 it is a
+ * choice made an entry at a time. CC-CEDICT has 一个 `ge5` beside 一个人 `ge4`,
+ * 那个 `ge5` beside 那个人 `ge4`, 上个 `ge5` beside 上个月 `ge4`, and 每个 `ge4`
+ * beside 每个人 `ge5`. large_pinyin wanders the same way over 半个 and 半个人,
+ * and over 这个 and 这些个. No reading of sense separates them, because there is
+ * no difference of sense to read: 那个人 is 那个 followed by 人.
+ *
+ * The sources do agree on the proportion. The full tone is the large majority in
+ * all three. 现代汉语频率词典 counts 个 as `gè` 11,693 times against `ge` 1,891
+ * (Unihan's `kHanyuPinlu`), large_pinyin writes 243 against 36, and CC-CEDICT
+ * 102 against 45. So the package writes the tone and names the exception.
+ *
+ * **A numeral in front of it means the 量词 is counting**, and a counting 个 is
+ * written `gè`: 一个, 两个, 十二个, 一百个, and 一个人 and 上一个 with them.
+ * This is what the gold corpus already asserts for every 个 it holds, and what
+ * the 一个 override used to state on its own.
+ *
+ * **The exception is the determiners.** 这个, 那个 and 哪个 are headwords in
+ * 现代汉语词典 read `zhège`, `nàge` and `nǎge`, and both sources write them 轻声
+ * wherever the pair stands on its own. 每个, 上个 and 下个 are the same
+ * determiner shape and the sources split on them by length: CC-CEDICT has 上个
+ * `ge5` beside 上个月 `ge4`, and 每个 `ge4` beside 每个人 `ge5`. The 轻声 is taken
+ * for all six, which is what 那个人, 每个人 and 上个月 are said with.
+ *
+ * Everything else keeps the reading its source gave it. 几个, 半个, 多个 and
+ * 单个 have neither a numeral nor a listed determiner in front of them, so their
+ * 轻声 stands. So does the colloquial 个 of 我勒个去 and 一口吃个胖子, which
+ * counts nothing, and the 北京话 今儿个, 明儿个 and 昨儿个, whose 儿 folds into
+ * the syllable before it and leaves the reading shorter than the key.
+ *
+ * Measured over the committed dictionary it retones 25 entries of the 279 that
+ * hold a 个. Over the 88,866 lines of Tatoeba and zh.wikipedia the reading
+ * rules were sized against it moves 339 conversions, and every one is a 个:
+ * 每个 106, 那个 63, 下个月 and 下个星期 62, 上个月 and 上个星期 60, and 35
+ * where 一个 gains its tone and the 一 sandhi that follows from it. Before the
+ * pass, 2,467 of the corpus's 5,397 个 sat after a character the package read
+ * both ways; after it, 204 do, and all of those are positions no key covers,
+ * where the decode falls back to the character's own `gè`.
+ */
+import { QUANTITY_CHARACTERS } from "../numerals/characters.js";
+import { toCharacters } from "../script/characters.js";
+import { NEUTRAL_TONE } from "../tone/tone.js";
+/**
+ * The 量词 this settles, in both scripts.
+ */
+const MEASURE_CHARACTERS = new Set(["个", "個"]);
+/**
+ * The tone a counting 个 carries.
+ */
+const COUNTING_TONE = 4;
+/**
+ * The determiners that take 个 into a word of their own, where it is 轻声.
+ *
+ * Six characters and not a part of speech, because the tag cannot draw the
+ * line. jieba tags 这个, 那个 and 每个 `r` and tags 各个, 某个 and 整个 `r` as
+ * well, and those three are `gègè`, `mǒugè` and `zhěnggè` in both sources with
+ * nothing disagreeing. So this is a list of judgements, in the way
+ * {@link import("./neutral-senses.js").NEUTRAL_SENSE_WORDS} is, and the words
+ * it decides are the ones where a source writes the 轻声 on the determiner and
+ * the pair together.
+ */
+export const NEUTRAL_MEASURE_DETERMINERS = new Set([
+    "这",
+    "這",
+    "那",
+    "哪",
+    "每",
+    "上",
+    "下",
+]);
+/**
+ * Whether a syllable is the 量词 rather than 个's other reading.
+ *
+ * 独自个 is `dú zì gě`, and a `gě` retoned to `gè` would be a different word.
+ * Testing the syllable keeps it out without a list of the words it appears in.
+ */
+function isCountingSyllable(syllable) {
+    return (syllable.initial === "g" &&
+        syllable.final === "e" &&
+        (syllable.tone === COUNTING_TONE || syllable.tone === NEUTRAL_TONE));
+}
+/**
+ * The tone the 个 at a position takes, or undefined where nothing settles it.
+ */
+export function countingToneAt(characters, at) {
+    // Nothing stands in front of a 个 that starts the word, and nothing settles
+    // it there: 个人 is `gèrén` on its own account.
+    const before = characters[at - 1] ?? "";
+    if (NEUTRAL_MEASURE_DETERMINERS.has(before)) {
+        return NEUTRAL_TONE;
+    }
+    return QUANTITY_CHARACTERS.has(before) ? COUNTING_TONE : undefined;
+}
+/**
+ * Give every 个 the tone its position calls for.
+ *
+ * Runs over the finished entries, before the constituent repair, so that a
+ * phrase held to the words inside it is held to a settled reading. Only the
+ * 普通话 reading is touched. A zh-TW delta is a claim about how 國語 differs
+ * from what stands beside it, and none of the 279 entries holding a 个 has one.
+ */
+export function settleMeasureTones(entries) {
+    let settled = 0;
+    const settle = (entry) => {
+        const characters = toCharacters(entry.hans);
+        const reading = entry.readings.cn;
+        // 儿化 folds two characters into one syllable, so a reading shorter than
+        // the word says nothing about which character each syllable reads.
+        if (reading.length !== characters.length) {
+            return entry;
+        }
+        const corrected = [...reading];
+        let isMoved = false;
+        for (const [at, syllable] of reading.entries()) {
+            if (
+            /* c8 ignore next -- the reading and the characters are the same length */
+            !MEASURE_CHARACTERS.has(characters[at] ?? "") ||
+                !isCountingSyllable(syllable)) {
+                continue;
+            }
+            const tone = countingToneAt(characters, at);
+            if (tone === undefined || syllable.tone === tone) {
+                continue;
+            }
+            corrected[at] = { ...syllable, tone };
+            isMoved = true;
+        }
+        if (!isMoved) {
+            return entry;
+        }
+        settled++;
+        return { ...entry, readings: { ...entry.readings, cn: corrected } };
+    };
+    return { entries: entries.map((entry) => settle(entry)), settled };
+}
+//# sourceMappingURL=measure-tone.js.map

package/dist/dictionary/measure-tone.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"measure-tone.js","sourceRoot":"","sources":["../../src/dictionary/measure-tone.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAyCG;AACH,OAAO,EAAE,mBAAmB,EAAE,MAAM,2BAA2B,CAAC;AAChE,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAEvD,OAAO,EAAE,YAAY,EAAa,MAAM,iBAAiB,CAAC;AAG1D;;GAEG;AACH,MAAM,kBAAkB,GAAwB,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;AAEpE;;GAEG;AACH,MAAM,aAAa,GAAG,CAAgB,CAAC;AAEvC;;;;;;;;;;GAUG;AACH,MAAM,CAAC,MAAM,2BAA2B,GAAwB,IAAI,GAAG,CAAC;IACtE,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;CACJ,CAAC,CAAC;AAEH;;;;;GAKG;AACH,SAAS,kBAAkB,CAAC,QAAkB;IAC5C,OAAO,CACL,QAAQ,CAAC,OAAO,KAAK,GAAG;QACxB,QAAQ,CAAC,KAAK,KAAK,GAAG;QACtB,CAAC,QAAQ,CAAC,IAAI,KAAK,aAAa,IAAI,QAAQ,CAAC,IAAI,KAAK,YAAY,CAAC,CACpE,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,cAAc,CAC5B,UAA6B,EAC7B,EAAU;IAEV,2EAA2E;IAC3E,8CAA8C;IAC9C,MAAM,MAAM,GAAG,UAAU,CAAC,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;IACxC,IAAI,2BAA2B,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC;QAC5C,OAAO,YAAY,CAAC;IACtB,CAAC;IACD,OAAO,mBAAmB,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,aAAa,CAAC,CAAC,CAAC,SAAS,CAAC;AACrE,CAAC;AAWD;;;;;;;GAOG;AACH,MAAM,UAAU,kBAAkB,CAChC,OAAmC;IAEnC,IAAI,OAAO,GAAG,CAAC,CAAC;IAEhB,MAAM,MAAM,GAAG,CAAC,KAAsB,EAAmB,EAAE;QACzD,MAAM,UAAU,GAAG,YAAY,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QAC5C,MAAM,OAAO,GAAG,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC;QAClC,uEAAuE;QACvE,mEAAmE;QACnE,IAAI,OAAO,CAAC,MAAM,KAAK,UAAU,CAAC,MAAM,EAAE,CAAC;YACzC,OAAO,KAAK,CAAC;QACf,CAAC;QAED,MAAM,SAAS,GAAG,CAAC,GAAG,OAAO,CAAC,CAAC;QAC/B,IAAI,OAAO,GAAG,KAAK,CAAC;QACpB,KAAK,MAAM,CAAC,EAAE,EAAE,QAAQ,CAAC,IAAI,OAAO,CAAC,OAAO,EAAE,EAAE,CAAC;YAC/C;YACE,0EAA0E;YAC1E,CAAC,kBAAkB,CAAC,GAAG,CAAC,UAAU,CAAC,EAAE,CAAC,IAAI,EAAE,CAAC;gBAC7C,CAAC,kBAAkB,CAAC,QAAQ,CAAC,EAC7B,CAAC;gBACD,SAAS;YACX,CAAC;YACD,MAAM,IAAI,GAAG,cAAc,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC;YAC5C,IAAI,IAAI,KAAK,SAAS,IAAI,QAAQ,CAAC,IAAI,KAAK,IAAI,EAAE,CAAC;gBACjD,SAAS;YACX,CAAC;YACD,SAAS,CAAC,EAAE,CAAC,GAAG,EAAE,GAAG,QAAQ,EAAE,IAAI,EAAE,CAAC;YACtC,OAAO,GAAG,IAAI,CAAC;QACjB,CAAC;QAED,IAAI,CAAC,OAAO,EAAE,CAAC;YACb,OAAO,KAAK,CAAC;QACf,CAAC;QACD,OAAO,EAAE,CAAC;QACV,OAAO,EAAE,GAAG,KAAK,EAAE,QAAQ,EAAE,EAAE,GAAG,KAAK,CAAC,QAAQ,EAAE,EAAE,EAAE,SAAS,EAAE,EAAE,CAAC;IACtE,CAAC,CAAC;IAEF,OAAO,EAAE,OAAO,EAAE,OAAO,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,OAAO,EAAE,CAAC;AACrE,CAAC"}

package/dist/dictionary/merge-types.d.ts

@@ -36,6 +36,11 @@ export interface MergeStats {
     readonly reducedNeutrals: number;
     /** Words whose trailing 儿 was folded into the syllable before it. */
     readonly erhuaRepairs: number;
+    /**
+     * Entries whose 个 was retoned by its position — see
+     * {@link import("./measure-tone.js").settleMeasureTones}.
+     */
+    readonly settledMeasureTones: number;
     /** Words whose 繁體 form was derived rather than taken from CC-CEDICT. */
     readonly derivedTraditional: number;
     /** Entries with a 繁體 form differing from their 简体 one. */
@@ -90,9 +95,9 @@ export interface WordSources {
  * What one word contributed to the build's counts.
  *
  * The per-word view of {@link MergeStats}, minus the counts no single word can
- * move: `reducedNeutrals`, `repairedConstituents` and `composedTaiwanReadings`
- * are settled by passes over the whole dictionary, and `rejected` is counted by
- * the caller.
+ * move: `reducedNeutrals`, `settledMeasureTones`, `repairedConstituents` and
+ * `composedTaiwanReadings` are settled by passes over the whole dictionary, and
+ * `rejected` is counted by the caller.
  */
 export interface WordTally {
     readonly neutralToneCorrections: number;

package/dist/dictionary/merge-types.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"merge-types.d.ts","sourceRoot":"","sources":["../../src/dictionary/merge-types.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAC;AACxD,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AACtD,OAAO,KAAK,EAAE,cAAc,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAC3E,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AACxD,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,YAAY,CAAC;AAClD,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,wBAAwB,CAAC;AAE/D;;GAEG;AACH,MAAM,WAAW,YAAY;IAC3B,QAAQ,CAAC,cAAc,EAAE,WAAW,CAAC,MAAM,EAAE,cAAc,CAAC,CAAC;IAC7D,QAAQ,CAAC,cAAc,EAAE,cAAc,CAAC;IACxC,QAAQ,CAAC,MAAM,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,MAAM,EAAE,CAAC,CAAC;IACxD,QAAQ,CAAC,MAAM,EAAE,SAAS,WAAW,EAAE,CAAC;IACxC,QAAQ,CAAC,KAAK,EAAE,WAAW,CAAC,MAAM,EAAE,UAAU,CAAC,CAAC;CACjD;AAED;;GAEG;AACH,MAAM,WAAW,UAAU;IACzB,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAC;IAC5B,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,iEAAiE;IACjE,QAAQ,CAAC,sBAAsB,EAAE,MAAM,CAAC;IACxC;;;OAGG;IACH,QAAQ,CAAC,eAAe,EAAE,MAAM,CAAC;IACjC,qEAAqE;IACrE,QAAQ,CAAC,YAAY,EAAE,MAAM,CAAC;IAC9B,wEAAwE;IACxE,QAAQ,CAAC,kBAAkB,EAAE,MAAM,CAAC;IACpC,0DAA0D;IAC1D,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,4DAA4D;IAC5D,QAAQ,CAAC,gBAAgB,EAAE,MAAM,CAAC;IAClC;;;OAGG;IACH,QAAQ,CAAC,oBAAoB,EAAE,MAAM,CAAC;IACtC,8DAA8D;IAC9D,QAAQ,CAAC,cAAc,EAAE,MAAM,CAAC;IAChC,sEAAsE;IACtE,QAAQ,CAAC,sBAAsB,EAAE,MAAM,CAAC;IACxC;;OAEG;IACH,QAAQ,CAAC,gBAAgB,EAAE,MAAM,CAAC;IAClC,8DAA8D;IAC9D,QAAQ,CAAC,cAAc,EAAE,MAAM,CAAC;IAChC,6DAA6D;IAC7D,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC;CAC3B;AAED;;GAEG;AACH,MAAM,WAAW,WAAW;IAC1B,yCAAyC;IACzC,QAAQ,CAAC,OAAO,EAAE,SAAS,eAAe,EAAE,CAAC;IAC7C,mEAAmE;IACnE,QAAQ,CAAC,QAAQ,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,MAAM,EAAE,CAAC,CAAC;IAC1D,QAAQ,CAAC,KAAK,EAAE,UAAU,CAAC;CAC5B;AAED;;;;;;GAMG;AACH,MAAM,WAAW,WAAW;IAC1B,QAAQ,CAAC,YAAY,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,WAAW,EAAE,CAAC,CAAC;IACnE,QAAQ,CAAC,YAAY,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,WAAW,EAAE,CAAC,CAAC;IACnE,QAAQ,CAAC,MAAM,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,MAAM,EAAE,CAAC,CAAC;IACxD,QAAQ,CAAC,KAAK,EAAE,WAAW,CAAC,MAAM,EAAE,UAAU,CAAC,CAAC;IAChD,QAAQ,CAAC,cAAc,EAAE,WAAW,CAAC,MAAM,EAAE,cAAc,CAAC,CAAC;IAC7D,QAAQ,CAAC,WAAW,EAAE,gBAAgB,CAAC;IACvC,QAAQ,CAAC,QAAQ,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,QAAQ,EAAE,CAAC,CAAC;CAC7D;AAED;;;;;;;GAOG;AACH,MAAM,WAAW,SAAS;IACxB,QAAQ,CAAC,sBAAsB,EAAE,MAAM,CAAC;IACxC,QAAQ,CAAC,YAAY,EAAE,MAAM,CAAC;IAC9B,QAAQ,CAAC,kBAAkB,EAAE,MAAM,CAAC;IACpC,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,gBAAgB,EAAE,MAAM,CAAC;IAClC,QAAQ,CAAC,cAAc,EAAE,MAAM,CAAC;IAChC,QAAQ,CAAC,gBAAgB,EAAE,MAAM,CAAC;IAClC,QAAQ,CAAC,cAAc,EAAE,MAAM,CAAC;IAChC,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAC;IAC5B,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;CAC9B;AAED;;GAEG;AACH,eAAO,MAAM,QAAQ,EAAE,SAYtB,CAAC;AAEF;;GAEG;AACH,MAAM,WAAW,UAAU;IACzB,QAAQ,CAAC,KAAK,EAAE,eAAe,GAAG,SAAS,CAAC;IAC5C,8EAA8E;IAC9E,QAAQ,CAAC,QAAQ,EAAE,SAAS,MAAM,EAAE,GAAG,SAAS,CAAC;IACjD,QAAQ,CAAC,KAAK,EAAE,SAAS,CAAC;CAC3B;AAED;;GAEG;AACH,wBAAgB,QAAQ,CAAC,KAAK,EAAE,SAAS,EAAE,KAAK,EAAE,SAAS,GAAG,SAAS,CAetE"}
+{"version":3,"file":"merge-types.d.ts","sourceRoot":"","sources":["../../src/dictionary/merge-types.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAC;AACxD,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AACtD,OAAO,KAAK,EAAE,cAAc,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAC3E,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AACxD,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,YAAY,CAAC;AAClD,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,wBAAwB,CAAC;AAE/D;;GAEG;AACH,MAAM,WAAW,YAAY;IAC3B,QAAQ,CAAC,cAAc,EAAE,WAAW,CAAC,MAAM,EAAE,cAAc,CAAC,CAAC;IAC7D,QAAQ,CAAC,cAAc,EAAE,cAAc,CAAC;IACxC,QAAQ,CAAC,MAAM,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,MAAM,EAAE,CAAC,CAAC;IACxD,QAAQ,CAAC,MAAM,EAAE,SAAS,WAAW,EAAE,CAAC;IACxC,QAAQ,CAAC,KAAK,EAAE,WAAW,CAAC,MAAM,EAAE,UAAU,CAAC,CAAC;CACjD;AAED;;GAEG;AACH,MAAM,WAAW,UAAU;IACzB,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAC;IAC5B,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,iEAAiE;IACjE,QAAQ,CAAC,sBAAsB,EAAE,MAAM,CAAC;IACxC;;;OAGG;IACH,QAAQ,CAAC,eAAe,EAAE,MAAM,CAAC;IACjC,qEAAqE;IACrE,QAAQ,CAAC,YAAY,EAAE,MAAM,CAAC;IAC9B;;;OAGG;IACH,QAAQ,CAAC,mBAAmB,EAAE,MAAM,CAAC;IACrC,wEAAwE;IACxE,QAAQ,CAAC,kBAAkB,EAAE,MAAM,CAAC;IACpC,0DAA0D;IAC1D,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,4DAA4D;IAC5D,QAAQ,CAAC,gBAAgB,EAAE,MAAM,CAAC;IAClC;;;OAGG;IACH,QAAQ,CAAC,oBAAoB,EAAE,MAAM,CAAC;IACtC,8DAA8D;IAC9D,QAAQ,CAAC,cAAc,EAAE,MAAM,CAAC;IAChC,sEAAsE;IACtE,QAAQ,CAAC,sBAAsB,EAAE,MAAM,CAAC;IACxC;;OAEG;IACH,QAAQ,CAAC,gBAAgB,EAAE,MAAM,CAAC;IAClC,8DAA8D;IAC9D,QAAQ,CAAC,cAAc,EAAE,MAAM,CAAC;IAChC,6DAA6D;IAC7D,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC;CAC3B;AAED;;GAEG;AACH,MAAM,WAAW,WAAW;IAC1B,yCAAyC;IACzC,QAAQ,CAAC,OAAO,EAAE,SAAS,eAAe,EAAE,CAAC;IAC7C,mEAAmE;IACnE,QAAQ,CAAC,QAAQ,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,MAAM,EAAE,CAAC,CAAC;IAC1D,QAAQ,CAAC,KAAK,EAAE,UAAU,CAAC;CAC5B;AAED;;;;;;GAMG;AACH,MAAM,WAAW,WAAW;IAC1B,QAAQ,CAAC,YAAY,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,WAAW,EAAE,CAAC,CAAC;IACnE,QAAQ,CAAC,YAAY,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,WAAW,EAAE,CAAC,CAAC;IACnE,QAAQ,CAAC,MAAM,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,MAAM,EAAE,CAAC,CAAC;IACxD,QAAQ,CAAC,KAAK,EAAE,WAAW,CAAC,MAAM,EAAE,UAAU,CAAC,CAAC;IAChD,QAAQ,CAAC,cAAc,EAAE,WAAW,CAAC,MAAM,EAAE,cAAc,CAAC,CAAC;IAC7D,QAAQ,CAAC,WAAW,EAAE,gBAAgB,CAAC;IACvC,QAAQ,CAAC,QAAQ,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,QAAQ,EAAE,CAAC,CAAC;CAC7D;AAED;;;;;;;GAOG;AACH,MAAM,WAAW,SAAS;IACxB,QAAQ,CAAC,sBAAsB,EAAE,MAAM,CAAC;IACxC,QAAQ,CAAC,YAAY,EAAE,MAAM,CAAC;IAC9B,QAAQ,CAAC,kBAAkB,EAAE,MAAM,CAAC;IACpC,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,gBAAgB,EAAE,MAAM,CAAC;IAClC,QAAQ,CAAC,cAAc,EAAE,MAAM,CAAC;IAChC,QAAQ,CAAC,gBAAgB,EAAE,MAAM,CAAC;IAClC,QAAQ,CAAC,cAAc,EAAE,MAAM,CAAC;IAChC,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAC;IAC5B,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;CAC9B;AAED;;GAEG;AACH,eAAO,MAAM,QAAQ,EAAE,SAYtB,CAAC;AAEF;;GAEG;AACH,MAAM,WAAW,UAAU;IACzB,QAAQ,CAAC,KAAK,EAAE,eAAe,GAAG,SAAS,CAAC;IAC5C,8EAA8E;IAC9E,QAAQ,CAAC,QAAQ,EAAE,SAAS,MAAM,EAAE,GAAG,SAAS,CAAC;IACjD,QAAQ,CAAC,KAAK,EAAE,SAAS,CAAC;CAC3B;AAED;;GAEG;AACH,wBAAgB,QAAQ,CAAC,KAAK,EAAE,SAAS,EAAE,KAAK,EAAE,SAAS,GAAG,SAAS,CAetE"}

package/dist/dictionary/merge-types.js.map

@@ -1 +1 @@
-{"version":3,"file":"merge-types.js","sourceRoot":"","sources":["../../src/dictionary/merge-types.ts"],"names":[],"mappings":"AAmHA;;GAEG;AACH,MAAM,CAAC,MAAM,QAAQ,GAAc;IACjC,sBAAsB,EAAE,CAAC;IACzB,YAAY,EAAE,CAAC;IACf,kBAAkB,EAAE,CAAC;IACrB,WAAW,EAAE,CAAC;IACd,gBAAgB,EAAE,CAAC;IACnB,cAAc,EAAE,CAAC;IACjB,gBAAgB,EAAE,CAAC;IACnB,cAAc,EAAE,CAAC;IACjB,UAAU,EAAE,CAAC;IACb,WAAW,EAAE,CAAC;IACd,WAAW,EAAE,CAAC;CACf,CAAC;AAYF;;GAEG;AACH,MAAM,UAAU,QAAQ,CAAC,KAAgB,EAAE,KAAgB;IACzD,OAAO;QACL,sBAAsB,EACpB,KAAK,CAAC,sBAAsB,GAAG,KAAK,CAAC,sBAAsB;QAC7D,YAAY,EAAE,KAAK,CAAC,YAAY,GAAG,KAAK,CAAC,YAAY;QACrD,kBAAkB,EAAE,KAAK,CAAC,kBAAkB,GAAG,KAAK,CAAC,kBAAkB;QACvE,WAAW,EAAE,KAAK,CAAC,WAAW,GAAG,KAAK,CAAC,WAAW;QAClD,gBAAgB,EAAE,KAAK,CAAC,gBAAgB,GAAG,KAAK,CAAC,gBAAgB;QACjE,cAAc,EAAE,KAAK,CAAC,cAAc,GAAG,KAAK,CAAC,cAAc;QAC3D,gBAAgB,EAAE,KAAK,CAAC,gBAAgB,GAAG,KAAK,CAAC,gBAAgB;QACjE,cAAc,EAAE,KAAK,CAAC,cAAc,GAAG,KAAK,CAAC,cAAc;QAC3D,UAAU,EAAE,KAAK,CAAC,UAAU,GAAG,KAAK,CAAC,UAAU;QAC/C,WAAW,EAAE,KAAK,CAAC,WAAW,GAAG,KAAK,CAAC,WAAW;QAClD,WAAW,EAAE,KAAK,CAAC,WAAW,GAAG,KAAK,CAAC,WAAW;KACnD,CAAC;AACJ,CAAC"}
+{"version":3,"file":"merge-types.js","sourceRoot":"","sources":["../../src/dictionary/merge-types.ts"],"names":[],"mappings":"AAwHA;;GAEG;AACH,MAAM,CAAC,MAAM,QAAQ,GAAc;IACjC,sBAAsB,EAAE,CAAC;IACzB,YAAY,EAAE,CAAC;IACf,kBAAkB,EAAE,CAAC;IACrB,WAAW,EAAE,CAAC;IACd,gBAAgB,EAAE,CAAC;IACnB,cAAc,EAAE,CAAC;IACjB,gBAAgB,EAAE,CAAC;IACnB,cAAc,EAAE,CAAC;IACjB,UAAU,EAAE,CAAC;IACb,WAAW,EAAE,CAAC;IACd,WAAW,EAAE,CAAC;CACf,CAAC;AAYF;;GAEG;AACH,MAAM,UAAU,QAAQ,CAAC,KAAgB,EAAE,KAAgB;IACzD,OAAO;QACL,sBAAsB,EACpB,KAAK,CAAC,sBAAsB,GAAG,KAAK,CAAC,sBAAsB;QAC7D,YAAY,EAAE,KAAK,CAAC,YAAY,GAAG,KAAK,CAAC,YAAY;QACrD,kBAAkB,EAAE,KAAK,CAAC,kBAAkB,GAAG,KAAK,CAAC,kBAAkB;QACvE,WAAW,EAAE,KAAK,CAAC,WAAW,GAAG,KAAK,CAAC,WAAW;QAClD,gBAAgB,EAAE,KAAK,CAAC,gBAAgB,GAAG,KAAK,CAAC,gBAAgB;QACjE,cAAc,EAAE,KAAK,CAAC,cAAc,GAAG,KAAK,CAAC,cAAc;QAC3D,gBAAgB,EAAE,KAAK,CAAC,gBAAgB,GAAG,KAAK,CAAC,gBAAgB;QACjE,cAAc,EAAE,KAAK,CAAC,cAAc,GAAG,KAAK,CAAC,cAAc;QAC3D,UAAU,EAAE,KAAK,CAAC,UAAU,GAAG,KAAK,CAAC,UAAU;QAC/C,WAAW,EAAE,KAAK,CAAC,WAAW,GAAG,KAAK,CAAC,WAAW;QAClD,WAAW,EAAE,KAAK,CAAC,WAAW,GAAG,KAAK,CAAC,WAAW;KACnD,CAAC;AACJ,CAAC"}

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

@@ -1 +1 @@
-{"version":3,"file":"merge.d.ts","sourceRoot":"","sources":["../../src/dictionary/merge.ts"],"names":[],"mappings":"AAYA,OAAO,KAAK,EAAE,WAAW,EAAE,YAAY,EAAe,MAAM,kBAAkB,CAAC;AAE/E,YAAY,EAAE,WAAW,EAAE,YAAY,EAAE,UAAU,EAAE,MAAM,kBAAkB,CAAC;AAE9E;;;;;;;;GAQG;AACH,wBAAgB,YAAY,CAAC,OAAO,EAAE,YAAY,GAAG,WAAW,CAuD/D"}
+{"version":3,"file":"merge.d.ts","sourceRoot":"","sources":["../../src/dictionary/merge.ts"],"names":[],"mappings":"AAaA,OAAO,KAAK,EAAE,WAAW,EAAE,YAAY,EAAe,MAAM,kBAAkB,CAAC;AAE/E,YAAY,EAAE,WAAW,EAAE,YAAY,EAAE,UAAU,EAAE,MAAM,kBAAkB,CAAC;AAE9E;;;;;;;;GAQG;AACH,wBAAgB,YAAY,CAAC,OAAO,EAAE,YAAY,GAAG,WAAW,CA6D/D"}

package/dist/dictionary/merge.js

@@ -9,6 +9,7 @@ import { composeLocaleDeltas } from "./locale.js";
 import { indexCedict, senseLookup } from "./cedict-senses.js";
 import { buildCharacterDefaults } from "./character-defaults.js";
 import { repairConstituentReadings } from "./constituent-repair.js";
+import { settleMeasureTones } from "./measure-tone.js";
 import { mergeWords, wordList } from "./merge-walk.js";
 /**
  * Build the merged dictionary from the four parsed sources.
@@ -38,11 +39,15 @@ export function mergeSources(sources) {
         defaults,
     };
     const { entries, rejected, counts } = mergeWords(wordList(phrase, defaults, cedictByWord, cedictByHant), wordSources);
+    // ── 个 given one tone per position ─────────────────────────
+    // Before the repair below, which reads a phrase's constituents off these same
+    // entries: settling first is what lets 那个人 be held to a settled 那个.
+    const measured = settleMeasureTones(entries);
     // ── Phrase entries held to the words inside them ───────────
     // Before the locale pass, which asks whether a compound reads a constituent
     // the way that constituent's own entry reads it. A phrase entry repaired here
     // answers yes where it used to answer no.
-    const held = repairConstituentReadings(entries, senseLookup(cedictByWord, cedictByHant));
+    const held = repairConstituentReadings(measured.entries, senseLookup(cedictByWord, cedictByHant));
     // ── zh-TW deltas the sources marked only on a constituent ──
     // Last, because it segments each compound against the finished entries: the
     // readings, both scripts' keys and the frequencies all have to be settled
@@ -54,6 +59,7 @@ export function mergeSources(sources) {
         stats: {
             ...counts,
             reducedNeutrals,
+            settledMeasureTones: measured.settled,
             repairedConstituents: held.repaired,
             composedTaiwanReadings: localised.composed,
             rejected: rejected.size,

package/dist/dictionary/merge.js.map

@@ -1 +1 @@
-{"version":3,"file":"merge.js","sourceRoot":"","sources":["../../src/dictionary/merge.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,EAAE,mBAAmB,EAAE,MAAM,aAAa,CAAC;AAClD,OAAO,EAAE,WAAW,EAAE,WAAW,EAAE,MAAM,oBAAoB,CAAC;AAC9D,OAAO,EAAE,sBAAsB,EAAE,MAAM,yBAAyB,CAAC;AACjE,OAAO,EAAE,yBAAyB,EAAE,MAAM,yBAAyB,CAAC;AACpE,OAAO,EAAE,UAAU,EAAE,QAAQ,EAAE,MAAM,iBAAiB,CAAC;AAKvD;;;;;;;;GAQG;AACH,MAAM,UAAU,YAAY,CAAC,OAAqB;IAChD,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,GAAG,OAAO,CAAC;IACnC,MAAM,YAAY,GAAG,WAAW,CAAC,MAAM,EAAE,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC;IACtE,6EAA6E;IAC7E,2EAA2E;IAC3E,0EAA0E;IAC1E,kCAAkC;IAClC,MAAM,YAAY,GAAG,WAAW,CAAC,MAAM,EAAE,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC;IAEvE,MAAM,EAAE,WAAW,EAAE,QAAQ,EAAE,eAAe,EAAE,GAAG,sBAAsB,CACvE,OAAO,EACP,YAAY,EACZ,YAAY,CACb,CAAC;IAEF,MAAM,WAAW,GAAgB;QAC/B,YAAY;QACZ,YAAY;QACZ,MAAM;QACN,KAAK,EAAE,OAAO,CAAC,KAAK;QACpB,cAAc,EAAE,OAAO,CAAC,cAAc;QACtC,WAAW;QACX,QAAQ;KACT,CAAC;IACF,MAAM,EAAE,OAAO,EAAE,QAAQ,EAAE,MAAM,EAAE,GAAG,UAAU,CAC9C,QAAQ,CAAC,MAAM,EAAE,QAAQ,EAAE,YAAY,EAAE,YAAY,CAAC,EACtD,WAAW,CACZ,CAAC;IAEF,8DAA8D;IAC9D,4EAA4E;IAC5E,8EAA8E;IAC9E,0CAA0C;IAC1C,MAAM,IAAI,GAAG,yBAAyB,CACpC,OAAO,EACP,WAAW,CAAC,YAAY,EAAE,YAAY,CAAC,CACxC,CAAC;IAEF,8DAA8D;IAC9D,4EAA4E;IAC5E,0EAA0E;IAC1E,qDAAqD;IACrD,MAAM,SAAS,GAAG,mBAAmB,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAEpD,OAAO;QACL,OAAO,EAAE,SAAS,CAAC,OAAO;QAC1B,QAAQ;QACR,KAAK,EAAE;YACL,GAAG,MAAM;YACT,eAAe;YACf,oBAAoB,EAAE,IAAI,CAAC,QAAQ;YACnC,sBAAsB,EAAE,SAAS,CAAC,QAAQ;YAC1C,QAAQ,EAAE,QAAQ,CAAC,IAAI;SACxB;KACF,CAAC;AACJ,CAAC"}
+{"version":3,"file":"merge.js","sourceRoot":"","sources":["../../src/dictionary/merge.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,EAAE,mBAAmB,EAAE,MAAM,aAAa,CAAC;AAClD,OAAO,EAAE,WAAW,EAAE,WAAW,EAAE,MAAM,oBAAoB,CAAC;AAC9D,OAAO,EAAE,sBAAsB,EAAE,MAAM,yBAAyB,CAAC;AACjE,OAAO,EAAE,yBAAyB,EAAE,MAAM,yBAAyB,CAAC;AACpE,OAAO,EAAE,kBAAkB,EAAE,MAAM,mBAAmB,CAAC;AACvD,OAAO,EAAE,UAAU,EAAE,QAAQ,EAAE,MAAM,iBAAiB,CAAC;AAKvD;;;;;;;;GAQG;AACH,MAAM,UAAU,YAAY,CAAC,OAAqB;IAChD,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,GAAG,OAAO,CAAC;IACnC,MAAM,YAAY,GAAG,WAAW,CAAC,MAAM,EAAE,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC;IACtE,6EAA6E;IAC7E,2EAA2E;IAC3E,0EAA0E;IAC1E,kCAAkC;IAClC,MAAM,YAAY,GAAG,WAAW,CAAC,MAAM,EAAE,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC;IAEvE,MAAM,EAAE,WAAW,EAAE,QAAQ,EAAE,eAAe,EAAE,GAAG,sBAAsB,CACvE,OAAO,EACP,YAAY,EACZ,YAAY,CACb,CAAC;IAEF,MAAM,WAAW,GAAgB;QAC/B,YAAY;QACZ,YAAY;QACZ,MAAM;QACN,KAAK,EAAE,OAAO,CAAC,KAAK;QACpB,cAAc,EAAE,OAAO,CAAC,cAAc;QACtC,WAAW;QACX,QAAQ;KACT,CAAC;IACF,MAAM,EAAE,OAAO,EAAE,QAAQ,EAAE,MAAM,EAAE,GAAG,UAAU,CAC9C,QAAQ,CAAC,MAAM,EAAE,QAAQ,EAAE,YAAY,EAAE,YAAY,CAAC,EACtD,WAAW,CACZ,CAAC;IAEF,6DAA6D;IAC7D,8EAA8E;IAC9E,oEAAoE;IACpE,MAAM,QAAQ,GAAG,kBAAkB,CAAC,OAAO,CAAC,CAAC;IAE7C,8DAA8D;IAC9D,4EAA4E;IAC5E,8EAA8E;IAC9E,0CAA0C;IAC1C,MAAM,IAAI,GAAG,yBAAyB,CACpC,QAAQ,CAAC,OAAO,EAChB,WAAW,CAAC,YAAY,EAAE,YAAY,CAAC,CACxC,CAAC;IAEF,8DAA8D;IAC9D,4EAA4E;IAC5E,0EAA0E;IAC1E,qDAAqD;IACrD,MAAM,SAAS,GAAG,mBAAmB,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAEpD,OAAO;QACL,OAAO,EAAE,SAAS,CAAC,OAAO;QAC1B,QAAQ;QACR,KAAK,EAAE;YACL,GAAG,MAAM;YACT,eAAe;YACf,mBAAmB,EAAE,QAAQ,CAAC,OAAO;YACrC,oBAAoB,EAAE,IAAI,CAAC,QAAQ;YACnC,sBAAsB,EAAE,SAAS,CAAC,QAAQ;YAC1C,QAAQ,EAAE,QAAQ,CAAC,IAAI;SACxB;KACF,CAAC;AACJ,CAAC"}

package/dist/dictionary/overrides.d.ts

@@ -29,6 +29,11 @@ export interface ReadingOverride {
  * {@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.
+ *
+ * 一个 was listed here for `yī gè` until the whole of 个 was settled by
+ * position. One word taken out of a disagreement the sources have about 262 of
+ * them is a rule missing rather than an override earned, and the rule is
+ * {@link import("./measure-tone.js").settleMeasureTones}.
  */
 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;;;;;;;;;;;;;;;;;;;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"}
+{"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;;;;;;;;;;;;;;;;;;;;;;;;GAwBG;AACH,eAAO,MAAM,iBAAiB,EAAE,SAAS,eAAe,EAOvD,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

@@ -19,6 +19,11 @@ import { NEUTRAL_TONE } from "../tone/tone.js";
  * {@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.
+ *
+ * 一个 was listed here for `yī gè` until the whole of 个 was settled by
+ * position. One word taken out of a disagreement the sources have about 262 of
+ * them is a rule missing rather than an override earned, and the rule is
+ * {@link import("./measure-tone.js").settleMeasureTones}.
  */
 export const READING_OVERRIDES = [
     {
@@ -26,11 +31,6 @@ export const READING_OVERRIDES = [
         reading: "dài fu",
         reason: "Both sources give dà fū, the ancient court title. The everyday sense, doctor, is dàifu and is overwhelmingly the more common.",
     },
-    {
-        word: "一个",
-        reading: "yī gè",
-        reason: "Not a source defect but a real disagreement: CC-CEDICT has yi1 ge5 and large_pinyin has yí gè, and both are attested speech. Step 7 prefers CC-CEDICT on neutral tones, which is right for the other 771 words it settles and wrong here — 个 counting is written with its tone. Stored underlyingly, so the runtime sandhi pass writes yí gè.",
-    },
 ];
 /**
  * Read an override's reading into syllables, throwing if it is malformed.

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;;;;;;;;;;;;;;;;;;;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"}
+{"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;;;;;;;;;;;;;;;;;;;;;;;;GAwBG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAA+B;IAC3D;QACE,IAAI,EAAE,IAAI;QACV,OAAO,EAAE,QAAQ;QACjB,MAAM,EACJ,+HAA+H;KAClI;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/dist/dictionary/script-tables.js

@@ -1,5 +1,5 @@
 import { toCharacters } from "../script/characters.js";
-import { toCanonicalGlyphs } from "../script/glyphs.js";
+import { CANONICAL_FORMS, toCanonicalGlyphs } from "../script/glyphs.js";
 import { convertCharacters } from "../script/conversion.js";
 import { alignedReadings, pairingsOf, tableOf, tally, } from "./script-pairings.js";
 /**
@@ -80,10 +80,78 @@ export function buildScriptTables(entries) {
  * 鹫峯寺 — and taking those at face value put 衞 and 峯 in the 简体-only set. A
  * text written 衞生 was then detected as 简体 and left alone when it was asked
  * for 简体, which is the opposite of what it needed.
+ *
+ * Occurrence alone is too strict, and that is what {@link writesCharacter}
+ * settles. A word list assembled from four sources holds a few headwords that
+ * failed to simplify — 见幾而作 and 极深研幾 are 简体 keys written with 幾 — and
+ * one of those was enough to take 幾 out of the 繁體-only set against the 297
+ * words that put it in. 衛 was out on the strength of a single 简体 headword
+ * against 394. So a script counts as writing a character when it writes it
+ * often enough to mean something, on the two-condition shape
+ * `deriveGlyphTables` already uses for the same kind of question.
+ *
+ * The variant forms are added afterwards, because normalising folds them away
+ * before anything is counted: every 裏 becomes 裡 on the way in, so 裏 occurs
+ * nowhere and settled nothing. Every key of `CANONICAL_FORMS` is 繁體 by
+ * construction — the variants 简体 writes were excluded from it, measured
+ * against CC-CEDICT's 简体 column — so the domain of that table is 繁體-only and
+ * joins the set wholesale.
+ *
+ * Measured against the dictionary's own paired headwords, which is 190,000
+ * labelled texts in two scripts: the multi-character pairs settle 99.13% of the
+ * time against 97.84% before, and 4 settle the wrong way. Checked against
+ * OpenCC's character tables, which no part of this is derived from, the 33
+ * characters that join the 简体 side all agree and 56 of the 60 judgeable ones
+ * joining the 繁體 side do. The four it disagrees about are 捝, 棁, 涚 and 蒀,
+ * which OpenCC lists in both its 简→繁 and 繁→简 tables and no corpus writes at
+ * all.
+ *
+ * 著 stays out of both, which is the honest answer rather than a miss: 简体
+ * writes it in its `zhù` sense — 专著, 显著, 著名 — and 繁體 writes it for that
+ * and for the aspect particle. 干, 台 and 里 are out for the same reason.
+ */
+/**
+ * How many words a script must write a character in before it counts.
+ *
+ * The absolute floor, paired with {@link WRITTEN_SHARE} below. Both come from
+ * `deriveGlyphTables`, which asks the same question about a variant spelling
+ * and answers it with the same two conditions and the same numbers.
+ */
+const WRITTEN_WORDS = 10;
+/**
+ * The share of the other script's evidence that makes a handful count anyway.
+ *
+ * Without this a rare character would belong to neither script. 齶 is written
+ * in two 繁體 words and no 简体 one, which is 繁體-only however small the number,
+ * and a bare floor would throw it away along with 幾's two stray headwords.
+ */
+const WRITTEN_SHARE = 0.05;
+/**
+ * Whether a script writes a character often enough for it to mean anything.
+ *
+ * Two conditions, asking different questions. The floor asks whether the
+ * character is written at all in that script, and the share asks whether it is
+ * written *relative to the other script* — which is what separates 幾, at two
+ * 简体 headwords against 297 繁體 ones, from 著 at 36 against 733.
  */
+/**
+ * Count each character of a word once, into a running tally.
+ */
+function countCharacters(into, word) {
+    for (const character of toCharacters(word)) {
+        into.set(character, (into.get(character) ?? 0) + 1);
+    }
+}
+function writesCharacter(counts, other, character) {
+    const held = counts.get(character) ?? 0;
+    if (held === 0) {
+        return false;
+    }
+    return (held >= WRITTEN_WORDS || held > (other.get(character) ?? 0) * WRITTEN_SHARE);
+}
 function scriptOnlyCharacters(entries) {
-    const inHans = new Set();
-    const inHant = new Set();
+    const inHans = new Map();
+    const inHant = new Map();
     for (const entry of entries) {
         // A 繁體 form that differs from the 简体 one only by glyph normalisation is
         // not evidence about script — 衞生 and 衛生 are two 繁體 spellings of one
@@ -95,18 +163,28 @@ function scriptOnlyCharacters(entries) {
         if (forms.length === 0) {
             continue;
         }
-        const hansCharacters = toCharacters(canonicalHans);
-        for (const character of hansCharacters) {
-            inHans.add(character);
+        countCharacters(inHans, canonicalHans);
+        for (const form of forms) {
+            countCharacters(inHant, toCanonicalGlyphs(form));
         }
-        const hantCharacters = forms.flatMap((form) => toCharacters(toCanonicalGlyphs(form)));
-        for (const character of hantCharacters) {
-            inHant.add(character);
+    }
+    const hansOnly = new Set();
+    const hantOnly = new Set();
+    for (const character of new Set([...inHans.keys(), ...inHant.keys()])) {
+        const hans = writesCharacter(inHans, inHant, character);
+        const hant = writesCharacter(inHant, inHans, character);
+        if (hans && !hant) {
+            hansOnly.add(character);
+        }
+        else if (hant && !hans) {
+            hantOnly.add(character);
         }
     }
-    return {
-        hansOnly: new Set([...inHans].filter((c) => !inHant.has(c))),
-        hantOnly: new Set([...inHant].filter((c) => !inHans.has(c))),
-    };
+    for (const variant of CANONICAL_FORMS.keys()) {
+        if (!hansOnly.has(variant)) {
+            hantOnly.add(variant);
+        }
+    }
+    return { hansOnly, hantOnly };
 }
 //# sourceMappingURL=script-tables.js.map

package/dist/dictionary/script-tables.js.map

@@ -1 +1 @@
-{"version":3,"file":"script-tables.js","sourceRoot":"","sources":["../../src/dictionary/script-tables.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AACvD,OAAO,EAAE,iBAAiB,EAAE,MAAM,qBAAqB,CAAC;AACxD,OAAO,EAAE,iBAAiB,EAAqB,MAAM,yBAAyB,CAAC;AAE/E,OAAO,EACL,eAAe,EAEf,UAAU,EACV,OAAO,EACP,KAAK,GACN,MAAM,sBAAsB,CAAC;AAC9B;;;;;;;;;;;;GAYG;AACH,MAAM,UAAU,iBAAiB,CAC/B,OAAmC;IAEnC,MAAM,mBAAmB,GAAG,IAAI,GAAG,EAAqB,CAAC;IACzD,MAAM,kBAAkB,GAAG,IAAI,GAAG,EAAqB,CAAC;IAExD,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,KAAK,MAAM,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,EAAE,IAAI,UAAU,CAAC,KAAK,CAAC,EAAE,CAAC;YACpD,KAAK,CAAC,mBAAmB,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC,CAAC,CAAC;YAC/C,KAAK,CAAC,kBAAkB,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC,CAAC,CAAC;QAChD,CAAC;IACH,CAAC;IAED,MAAM,aAAa,GAAG,OAAO,CAAC,mBAAmB,CAAC,CAAC;IACnD,MAAM,YAAY,GAAG,OAAO,CAAC,kBAAkB,CAAC,CAAC;IAEjD,4EAA4E;IAC5E,+CAA+C;IAC/C,EAAE;IACF,4EAA4E;IAC5E,4EAA4E;IAC5E,8EAA8E;IAC9E,2EAA2E;IAC3E,6EAA6E;IAC7E,oEAAoE;IACpE,MAAM,gBAAgB,GAAG,IAAI,GAAG,EAAkB,CAAC;IACnD,MAAM,eAAe,GAAG,IAAI,GAAG,EAAkB,CAAC;IAClD,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,IAAI,YAAY,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACxC,SAAS;QACX,CAAC;QACD,MAAM,QAAQ,GAAG,eAAe,CAAC,KAAK,CAAC,CAAC;QACxC,IAAI,iBAAiB,CAAC,aAAa,EAAE,KAAK,CAAC,IAAI,EAAE,QAAQ,CAAC,KAAK,KAAK,CAAC,IAAI,EAAE,CAAC;YAC1E,gBAAgB,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,CAAC;QAC/C,CAAC;QACD,IAAI,iBAAiB,CAAC,YAAY,EAAE,KAAK,CAAC,IAAI,EAAE,QAAQ,CAAC,KAAK,KAAK,CAAC,IAAI,EAAE,CAAC;YACzE,eAAe,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,CAAC;QAC9C,CAAC;IACH,CAAC;IAED,OAAO;QACL,aAAa;QACb,YAAY;QACZ,gBAAgB;QAChB,eAAe;QACf,GAAG,oBAAoB,CAAC,OAAO,CAAC;KACjC,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AACH,SAAS,oBAAoB,CAAC,OAAmC;IAI/D,MAAM,MAAM,GAAG,IAAI,GAAG,EAAU,CAAC;IACjC,MAAM,MAAM,GAAG,IAAI,GAAG,EAAU,CAAC;IACjC,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,wEAAwE;QACxE,oEAAoE;QACpE,0EAA0E;QAC1E,uEAAuE;QACvE,2BAA2B;QAC3B,MAAM,aAAa,GAAG,iBAAiB,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QACpD,MAAM,KAAK,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,GAAG,CAAC,KAAK,CAAC,YAAY,IAAI,EAAE,CAAC,CAAC,CAAC,MAAM,CAC9D,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,KAAK,KAAK,CAAC,IAAI,IAAI,IAAI,KAAK,aAAa,CACxD,CAAC;QACF,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACvB,SAAS;QACX,CAAC;QACD,MAAM,cAAc,GAAG,YAAY,CAAC,aAAa,CAAC,CAAC;QACnD,KAAK,MAAM,SAAS,IAAI,cAAc,EAAE,CAAC;YACvC,MAAM,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QACxB,CAAC;QACD,MAAM,cAAc,GAAG,KAAK,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,EAAE,CAC5C,YAAY,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAC,CACtC,CAAC;QACF,KAAK,MAAM,SAAS,IAAI,cAAc,EAAE,CAAC;YACvC,MAAM,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QACxB,CAAC;IACH,CAAC;IAED,OAAO;QACL,QAAQ,EAAE,IAAI,GAAG,CAAC,CAAC,GAAG,MAAM,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;QAC5D,QAAQ,EAAE,IAAI,GAAG,CAAC,CAAC,GAAG,MAAM,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;KAC7D,CAAC;AACJ,CAAC"}
+{"version":3,"file":"script-tables.js","sourceRoot":"","sources":["../../src/dictionary/script-tables.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AACvD,OAAO,EAAE,eAAe,EAAE,iBAAiB,EAAE,MAAM,qBAAqB,CAAC;AACzE,OAAO,EAAE,iBAAiB,EAAqB,MAAM,yBAAyB,CAAC;AAE/E,OAAO,EACL,eAAe,EAEf,UAAU,EACV,OAAO,EACP,KAAK,GACN,MAAM,sBAAsB,CAAC;AAC9B;;;;;;;;;;;;GAYG;AACH,MAAM,UAAU,iBAAiB,CAC/B,OAAmC;IAEnC,MAAM,mBAAmB,GAAG,IAAI,GAAG,EAAqB,CAAC;IACzD,MAAM,kBAAkB,GAAG,IAAI,GAAG,EAAqB,CAAC;IAExD,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,KAAK,MAAM,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,EAAE,IAAI,UAAU,CAAC,KAAK,CAAC,EAAE,CAAC;YACpD,KAAK,CAAC,mBAAmB,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC,CAAC,CAAC;YAC/C,KAAK,CAAC,kBAAkB,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC,CAAC,CAAC;QAChD,CAAC;IACH,CAAC;IAED,MAAM,aAAa,GAAG,OAAO,CAAC,mBAAmB,CAAC,CAAC;IACnD,MAAM,YAAY,GAAG,OAAO,CAAC,kBAAkB,CAAC,CAAC;IAEjD,4EAA4E;IAC5E,+CAA+C;IAC/C,EAAE;IACF,4EAA4E;IAC5E,4EAA4E;IAC5E,8EAA8E;IAC9E,2EAA2E;IAC3E,6EAA6E;IAC7E,oEAAoE;IACpE,MAAM,gBAAgB,GAAG,IAAI,GAAG,EAAkB,CAAC;IACnD,MAAM,eAAe,GAAG,IAAI,GAAG,EAAkB,CAAC;IAClD,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,IAAI,YAAY,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACxC,SAAS;QACX,CAAC;QACD,MAAM,QAAQ,GAAG,eAAe,CAAC,KAAK,CAAC,CAAC;QACxC,IAAI,iBAAiB,CAAC,aAAa,EAAE,KAAK,CAAC,IAAI,EAAE,QAAQ,CAAC,KAAK,KAAK,CAAC,IAAI,EAAE,CAAC;YAC1E,gBAAgB,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,CAAC;QAC/C,CAAC;QACD,IAAI,iBAAiB,CAAC,YAAY,EAAE,KAAK,CAAC,IAAI,EAAE,QAAQ,CAAC,KAAK,KAAK,CAAC,IAAI,EAAE,CAAC;YACzE,eAAe,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,CAAC;QAC9C,CAAC;IACH,CAAC;IAED,OAAO;QACL,aAAa;QACb,YAAY;QACZ,gBAAgB;QAChB,eAAe;QACf,GAAG,oBAAoB,CAAC,OAAO,CAAC;KACjC,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAoDG;AACH;;;;;;GAMG;AACH,MAAM,aAAa,GAAG,EAAE,CAAC;AAEzB;;;;;;GAMG;AACH,MAAM,aAAa,GAAG,IAAI,CAAC;AAE3B;;;;;;;GAOG;AACH;;GAEG;AACH,SAAS,eAAe,CAAC,IAAyB,EAAE,IAAY;IAC9D,KAAK,MAAM,SAAS,IAAI,YAAY,CAAC,IAAI,CAAC,EAAE,CAAC;QAC3C,IAAI,CAAC,GAAG,CAAC,SAAS,EAAE,CAAC,IAAI,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;IACtD,CAAC;AACH,CAAC;AAED,SAAS,eAAe,CACtB,MAAmC,EACnC,KAAkC,EAClC,SAAiB;IAEjB,MAAM,IAAI,GAAG,MAAM,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC;IACxC,IAAI,IAAI,KAAK,CAAC,EAAE,CAAC;QACf,OAAO,KAAK,CAAC;IACf,CAAC;IACD,OAAO,CACL,IAAI,IAAI,aAAa,IAAI,IAAI,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,GAAG,aAAa,CAC5E,CAAC;AACJ,CAAC;AAED,SAAS,oBAAoB,CAAC,OAAmC;IAI/D,MAAM,MAAM,GAAG,IAAI,GAAG,EAAkB,CAAC;IACzC,MAAM,MAAM,GAAG,IAAI,GAAG,EAAkB,CAAC;IAEzC,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,wEAAwE;QACxE,oEAAoE;QACpE,0EAA0E;QAC1E,uEAAuE;QACvE,2BAA2B;QAC3B,MAAM,aAAa,GAAG,iBAAiB,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QACpD,MAAM,KAAK,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,GAAG,CAAC,KAAK,CAAC,YAAY,IAAI,EAAE,CAAC,CAAC,CAAC,MAAM,CAC9D,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,KAAK,KAAK,CAAC,IAAI,IAAI,IAAI,KAAK,aAAa,CACxD,CAAC;QACF,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACvB,SAAS;QACX,CAAC;QACD,eAAe,CAAC,MAAM,EAAE,aAAa,CAAC,CAAC;QACvC,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;YACzB,eAAe,CAAC,MAAM,EAAE,iBAAiB,CAAC,IAAI,CAAC,CAAC,CAAC;QACnD,CAAC;IACH,CAAC;IAED,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAU,CAAC;IACnC,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAU,CAAC;IACnC,KAAK,MAAM,SAAS,IAAI,IAAI,GAAG,CAAC,CAAC,GAAG,MAAM,CAAC,IAAI,EAAE,EAAE,GAAG,MAAM,CAAC,IAAI,EAAE,CAAC,CAAC,EAAE,CAAC;QACtE,MAAM,IAAI,GAAG,eAAe,CAAC,MAAM,EAAE,MAAM,EAAE,SAAS,CAAC,CAAC;QACxD,MAAM,IAAI,GAAG,eAAe,CAAC,MAAM,EAAE,MAAM,EAAE,SAAS,CAAC,CAAC;QACxD,IAAI,IAAI,IAAI,CAAC,IAAI,EAAE,CAAC;YAClB,QAAQ,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QAC1B,CAAC;aAAM,IAAI,IAAI,IAAI,CAAC,IAAI,EAAE,CAAC;YACzB,QAAQ,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QAC1B,CAAC;IACH,CAAC;IAED,KAAK,MAAM,OAAO,IAAI,eAAe,CAAC,IAAI,EAAE,EAAE,CAAC;QAC7C,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,CAAC;YAC3B,QAAQ,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;QACxB,CAAC;IACH,CAAC;IAED,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,CAAC;AAChC,CAAC"}

package/dist/dictionary/source.d.ts

@@ -2,6 +2,7 @@ import { type ScriptTables } from "../script/conversion.js";
 import type { DictionaryArtifact } from "./artifact.js";
 import { Dictionary } from "./dictionary.js";
 import type { Tier } from "./tiers.js";
+import { WordCounts } from "./word-counts.js";
 /**
  * Where a dictionary's files are read from.
  *
@@ -43,6 +44,27 @@ export declare function loadArtifact(source: DictionarySource, tier: Tier): Prom
  * Load a tier and wrap it in a {@link Dictionary}.
  */
 export declare function loadDictionary(source: DictionarySource, tier: Tier): Promise<Dictionary>;
+/**
+ * The file the raw corpus counts live in.
+ *
+ * `full` alone, and outside the tiers rather than one per tier: a caller
+ * ranking words wants the whole vocabulary ordered, and a ranking over part of
+ * it answers a different question.
+ */
+export declare const COUNTS_FILE = "full.counts";
+/**
+ * Load the raw corpus counts for the `full` tier.
+ *
+ * Separate from the dictionary for the same reason {@link loadScriptTables} is.
+ * Nothing on the decoding path reads counts, and the quantised frequencies the
+ * decoder does read already ship inside every tier. This file is 243 KB brotli
+ * against `full.freq`'s 148 KB, and only a caller that asks for it pays.
+ *
+ * The counts are positional over the `full` tier's keys, so pair them with a
+ * `full` {@link Dictionary} and nothing else. {@link WordCounts.size} and
+ * {@link Dictionary.size} agree where the two match.
+ */
+export declare function loadWordCounts(source: DictionarySource): Promise<WordCounts>;
 /**
  * The file the script conversion tables live in.
  *

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

@@ -1 +1 @@
-{"version":3,"file":"source.d.ts","sourceRoot":"","sources":["../../src/dictionary/source.ts"],"names":[],"mappings":"AAAA,OAAO,EAAoB,KAAK,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAC9E,OAAO,KAAK,EAAE,kBAAkB,EAAE,MAAM,eAAe,CAAC;AACxD,OAAO,EAAE,UAAU,EAAE,MAAM,iBAAiB,CAAC;AAC7C,OAAO,KAAK,EAAE,IAAI,EAAE,MAAM,YAAY,CAAC;AAEvC;;;;;;;GAOG;AACH,MAAM,WAAW,gBAAgB;IAC/B,0CAA0C;IAC1C,QAAQ,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,OAAO,CAAC,MAAM,CAAC,CAAC;IACjD,2CAA2C;IAC3C,QAAQ,CAAC,KAAK,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,OAAO,CAAC,UAAU,CAAC,CAAC;CACvD;AAED;;GAEG;AACH,wBAAgB,SAAS,CAAC,IAAI,EAAE,IAAI,GAAG;IACrC,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;IACzB,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;CAC9B,CAMA;AAED;;;;;;GAMG;AACH,wBAAgB,WAAW,CAAC,OAAO,EAAE,MAAM,GAAG,gBAAgB,CAqB7D;AAED;;;;;GAKG;AACH,wBAAsB,YAAY,CAChC,MAAM,EAAE,gBAAgB,EACxB,IAAI,EAAE,IAAI,GACT,OAAO,CAAC,kBAAkB,CAAC,CAQ7B;AAED;;GAEG;AACH,wBAAsB,cAAc,CAClC,MAAM,EAAE,gBAAgB,EACxB,IAAI,EAAE,IAAI,GACT,OAAO,CAAC,UAAU,CAAC,CAErB;AAED;;;;;;GAMG;AACH,eAAO,MAAM,WAAW,eAAe,CAAC;AAExC;;;;;;;GAOG;AACH,wBAAsB,gBAAgB,CACpC,MAAM,EAAE,gBAAgB,GACvB,OAAO,CAAC,YAAY,CAAC,CAEvB"}
+{"version":3,"file":"source.d.ts","sourceRoot":"","sources":["../../src/dictionary/source.ts"],"names":[],"mappings":"AAAA,OAAO,EAAoB,KAAK,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAC9E,OAAO,KAAK,EAAE,kBAAkB,EAAE,MAAM,eAAe,CAAC;AACxD,OAAO,EAAE,UAAU,EAAE,MAAM,iBAAiB,CAAC;AAC7C,OAAO,KAAK,EAAE,IAAI,EAAE,MAAM,YAAY,CAAC;AACvC,OAAO,EAAE,UAAU,EAAE,MAAM,kBAAkB,CAAC;AAE9C;;;;;;;GAOG;AACH,MAAM,WAAW,gBAAgB;IAC/B,0CAA0C;IAC1C,QAAQ,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,OAAO,CAAC,MAAM,CAAC,CAAC;IACjD,2CAA2C;IAC3C,QAAQ,CAAC,KAAK,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,OAAO,CAAC,UAAU,CAAC,CAAC;CACvD;AAED;;GAEG;AACH,wBAAgB,SAAS,CAAC,IAAI,EAAE,IAAI,GAAG;IACrC,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;IACzB,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;CAC9B,CAMA;AAED;;;;;;GAMG;AACH,wBAAgB,WAAW,CAAC,OAAO,EAAE,MAAM,GAAG,gBAAgB,CAqB7D;AAED;;;;;GAKG;AACH,wBAAsB,YAAY,CAChC,MAAM,EAAE,gBAAgB,EACxB,IAAI,EAAE,IAAI,GACT,OAAO,CAAC,kBAAkB,CAAC,CAQ7B;AAED;;GAEG;AACH,wBAAsB,cAAc,CAClC,MAAM,EAAE,gBAAgB,EACxB,IAAI,EAAE,IAAI,GACT,OAAO,CAAC,UAAU,CAAC,CAErB;AAED;;;;;;GAMG;AACH,eAAO,MAAM,WAAW,gBAAgB,CAAC;AAEzC;;;;;;;;;;;GAWG;AACH,wBAAsB,cAAc,CAClC,MAAM,EAAE,gBAAgB,GACvB,OAAO,CAAC,UAAU,CAAC,CAErB;AAED;;;;;;GAMG;AACH,eAAO,MAAM,WAAW,eAAe,CAAC;AAExC;;;;;;;GAOG;AACH,wBAAsB,gBAAgB,CACpC,MAAM,EAAE,gBAAgB,GACvB,OAAO,CAAC,YAAY,CAAC,CAEvB"}

package/dist/dictionary/source.js

@@ -1,5 +1,6 @@
 import { readScriptTables } from "../script/conversion.js";
 import { Dictionary } from "./dictionary.js";
+import { WordCounts } from "./word-counts.js";
 /**
  * The three files a tier is made of.
  */
@@ -58,6 +59,29 @@ export async function loadArtifact(source, tier) {
 export async function loadDictionary(source, tier) {
     return Dictionary.from(await loadArtifact(source, tier));
 }
+/**
+ * The file the raw corpus counts live in.
+ *
+ * `full` alone, and outside the tiers rather than one per tier: a caller
+ * ranking words wants the whole vocabulary ordered, and a ranking over part of
+ * it answers a different question.
+ */
+export const COUNTS_FILE = "full.counts";
+/**
+ * Load the raw corpus counts for the `full` tier.
+ *
+ * Separate from the dictionary for the same reason {@link loadScriptTables} is.
+ * Nothing on the decoding path reads counts, and the quantised frequencies the
+ * decoder does read already ship inside every tier. This file is 243 KB brotli
+ * against `full.freq`'s 148 KB, and only a caller that asks for it pays.
+ *
+ * The counts are positional over the `full` tier's keys, so pair them with a
+ * `full` {@link Dictionary} and nothing else. {@link WordCounts.size} and
+ * {@link Dictionary.size} agree where the two match.
+ */
+export async function loadWordCounts(source) {
+    return WordCounts.from(await source.bytes(COUNTS_FILE));
+}
 /**
  * The file the script conversion tables live in.
  *

package/dist/dictionary/source.js.map

@@ -1 +1 @@
-{"version":3,"file":"source.js","sourceRoot":"","sources":["../../src/dictionary/source.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,gBAAgB,EAAqB,MAAM,yBAAyB,CAAC;AAE9E,OAAO,EAAE,UAAU,EAAE,MAAM,iBAAiB,CAAC;AAkB7C;;GAEG;AACH,MAAM,UAAU,SAAS,CAAC,IAAU;IAKlC,OAAO;QACL,IAAI,EAAE,GAAG,IAAI,OAAO;QACpB,OAAO,EAAE,GAAG,IAAI,UAAU;QAC1B,WAAW,EAAE,GAAG,IAAI,OAAO;KAC5B,CAAC;AACJ,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,WAAW,CAAC,OAAe;IACzC,MAAM,EAAE,GAAG,CAAC,IAAY,EAAU,EAAE,CAAC,GAAG,OAAO,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC,IAAI,IAAI,EAAE,CAAC;IAC9E,MAAM,GAAG,GAAG,KAAK,EAAE,IAAY,EAAqB,EAAE;QACpD,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,EAAE,CAAC,IAAI,CAAC,CAAC,CAAC;QACvC,IAAI,CAAC,QAAQ,CAAC,EAAE,EAAE,CAAC;YACjB,MAAM,IAAI,KAAK,CACb,YAAY,EAAE,CAAC,IAAI,CAAC,YAAY,MAAM,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,QAAQ,CAAC,UAAU,EAAE,CACjF,CAAC;QACJ,CAAC;QACD,OAAO,QAAQ,CAAC;IAClB,CAAC,CAAC;IACF,OAAO;QACL,IAAI,EAAE,KAAK,EAAE,IAAY,EAAmB,EAAE;YAC5C,MAAM,QAAQ,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,CAAC;YACjC,OAAO,QAAQ,CAAC,IAAI,EAAE,CAAC;QACzB,CAAC;QACD,KAAK,EAAE,KAAK,EAAE,IAAY,EAAuB,EAAE;YACjD,MAAM,QAAQ,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,CAAC;YACjC,OAAO,IAAI,UAAU,CAAC,MAAM,QAAQ,CAAC,WAAW,EAAE,CAAC,CAAC;QACtD,CAAC;KACF,CAAC;AACJ,CAAC;AAED;;;;;GAKG;AACH,MAAM,CAAC,KAAK,UAAU,YAAY,CAChC,MAAwB,EACxB,IAAU;IAEV,MAAM,KAAK,GAAG,SAAS,CAAC,IAAI,CAAC,CAAC;IAC9B,MAAM,CAAC,IAAI,EAAE,OAAO,EAAE,WAAW,CAAC,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC;QACrD,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC;QACvB,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC;QAC1B,MAAM,CAAC,KAAK,CAAC,KAAK,CAAC,WAAW,CAAC;KAChC,CAAC,CAAC;IACH,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,WAAW,EAAE,CAAC;AACxC,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,KAAK,UAAU,cAAc,CAClC,MAAwB,EACxB,IAAU;IAEV,OAAO,UAAU,CAAC,IAAI,CAAC,MAAM,YAAY,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC,CAAC;AAC3D,CAAC;AAED;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,WAAW,GAAG,YAAY,CAAC;AAExC;;;;;;;GAOG;AACH,MAAM,CAAC,KAAK,UAAU,gBAAgB,CACpC,MAAwB;IAExB,OAAO,gBAAgB,CAAC,MAAM,MAAM,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC,CAAC;AAC1D,CAAC"}
+{"version":3,"file":"source.js","sourceRoot":"","sources":["../../src/dictionary/source.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,gBAAgB,EAAqB,MAAM,yBAAyB,CAAC;AAE9E,OAAO,EAAE,UAAU,EAAE,MAAM,iBAAiB,CAAC;AAE7C,OAAO,EAAE,UAAU,EAAE,MAAM,kBAAkB,CAAC;AAiB9C;;GAEG;AACH,MAAM,UAAU,SAAS,CAAC,IAAU;IAKlC,OAAO;QACL,IAAI,EAAE,GAAG,IAAI,OAAO;QACpB,OAAO,EAAE,GAAG,IAAI,UAAU;QAC1B,WAAW,EAAE,GAAG,IAAI,OAAO;KAC5B,CAAC;AACJ,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,WAAW,CAAC,OAAe;IACzC,MAAM,EAAE,GAAG,CAAC,IAAY,EAAU,EAAE,CAAC,GAAG,OAAO,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC,IAAI,IAAI,EAAE,CAAC;IAC9E,MAAM,GAAG,GAAG,KAAK,EAAE,IAAY,EAAqB,EAAE;QACpD,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,EAAE,CAAC,IAAI,CAAC,CAAC,CAAC;QACvC,IAAI,CAAC,QAAQ,CAAC,EAAE,EAAE,CAAC;YACjB,MAAM,IAAI,KAAK,CACb,YAAY,EAAE,CAAC,IAAI,CAAC,YAAY,MAAM,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,QAAQ,CAAC,UAAU,EAAE,CACjF,CAAC;QACJ,CAAC;QACD,OAAO,QAAQ,CAAC;IAClB,CAAC,CAAC;IACF,OAAO;QACL,IAAI,EAAE,KAAK,EAAE,IAAY,EAAmB,EAAE;YAC5C,MAAM,QAAQ,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,CAAC;YACjC,OAAO,QAAQ,CAAC,IAAI,EAAE,CAAC;QACzB,CAAC;QACD,KAAK,EAAE,KAAK,EAAE,IAAY,EAAuB,EAAE;YACjD,MAAM,QAAQ,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,CAAC;YACjC,OAAO,IAAI,UAAU,CAAC,MAAM,QAAQ,CAAC,WAAW,EAAE,CAAC,CAAC;QACtD,CAAC;KACF,CAAC;AACJ,CAAC;AAED;;;;;GAKG;AACH,MAAM,CAAC,KAAK,UAAU,YAAY,CAChC,MAAwB,EACxB,IAAU;IAEV,MAAM,KAAK,GAAG,SAAS,CAAC,IAAI,CAAC,CAAC;IAC9B,MAAM,CAAC,IAAI,EAAE,OAAO,EAAE,WAAW,CAAC,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC;QACrD,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC;QACvB,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC;QAC1B,MAAM,CAAC,KAAK,CAAC,KAAK,CAAC,WAAW,CAAC;KAChC,CAAC,CAAC;IACH,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,WAAW,EAAE,CAAC;AACxC,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,KAAK,UAAU,cAAc,CAClC,MAAwB,EACxB,IAAU;IAEV,OAAO,UAAU,CAAC,IAAI,CAAC,MAAM,YAAY,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC,CAAC;AAC3D,CAAC;AAED;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,WAAW,GAAG,aAAa,CAAC;AAEzC;;;;;;;;;;;GAWG;AACH,MAAM,CAAC,KAAK,UAAU,cAAc,CAClC,MAAwB;IAExB,OAAO,UAAU,CAAC,IAAI,CAAC,MAAM,MAAM,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC,CAAC;AAC1D,CAAC;AAED;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,WAAW,GAAG,YAAY,CAAC;AAExC;;;;;;;GAOG;AACH,MAAM,CAAC,KAAK,UAAU,gBAAgB,CACpC,MAAwB;IAExB,OAAO,gBAAgB,CAAC,MAAM,MAAM,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC,CAAC;AAC1D,CAAC"}