@kensio/pinyinjs 1.7.4 → 1.7.5

package/dist/dictionary/locale.js

@@ -1,172 +1,15 @@
-import { isSingleCharacter, toCharacters } from "../script/characters.js";
-import { isSameReading } from "./entry.js";
-import { FrequencyTable } from "./frequency-table.js";
 /**
- * Compounds whose constituent's 國語 reading must not be carried over.
- *
- * The composition below only fires where the compound reads its constituent
- * exactly as that constituent's own entry does, which rules out most of the ways
- * it could be wrong. What it cannot rule out is a homograph: two words spelled
- * and read identically in 普通话, of which only one shifts in 國語. 相亲 is
- * `xiāngqīn` either way when it means "mutually close" and `xiàngqīn` in Taiwan
- * when it means a matchmaking meeting, and nothing in the data distinguishes the
- * two senses.
+ * Giving a compound the 國語 reading its constituents already carry.
  *
- * **Keep this small**, for the same reason
- * {@link import("./overrides.js").READING_OVERRIDES} is kept small. Three
- * exclusions against the 100 compounds the composition settles is the measured
- * ratio; a list growing much past that would mean the alignment check has
- * stopped carrying its weight and the rule itself needs revisiting.
- */
-export const LOCALE_COMPOSITION_EXCLUSIONS = [
-    {
-        word: "相亲相爱",
-        reason: "相亲 here is 相 + 亲 in the reciprocal sense, `xiāngqīn`, not the matchmaking word that reads `xiàngqīn` in 國語.",
-    },
-    {
-        word: "腹背相亲",
-        reason: "The same reciprocal 相亲, so the same reading in both locales.",
-    },
-    {
-        word: "洋为中用",
-        reason: "中 is 中国 and 用 is its own verb: 洋为中用 is not built on the word 中用, which is what reads `zhòngyòng` in 國語.",
-    },
-];
-/**
- * The exclusions as a set, for the composition to consult.
+ * Which compounds may be asked at all is `locale-candidates.ts`; what is here
+ * is what a compound inherits once it has been asked, and the rounds of the
+ * pass that let an inheritance itself be inherited.
  */
-const EXCLUDED = new Set(LOCALE_COMPOSITION_EXCLUSIONS.map((exclusion) => exclusion.word));
-/**
- * What each word in a segmentation costs on top of its frequency.
- *
- * The same charge, and for the same reason, as `READING_CHARGE` in
- * `src/decode/lattice.ts`: it orders candidate segmentations by fewest words
- * first and frequency second. That ordering is what decides 皮夹克 is 皮 + 夹克
- * rather than 皮夹 + 克, which is the difference between leaving `pí jiákè`
- * alone and mistaking it for the wallet 皮夹 that reads `pí jiá` in 國語.
- *
- * Duplicated rather than imported because the decoder sits above the dictionary
- * and must not be depended on from inside it. The two are kept in step by
- * `src/dictionary/locale.test.ts`, which asserts the segmentations this value
- * produces rather than the value itself.
- */
-const WORD_CHARGE = 16;
-/**
- * Every key an entry answers to, under either script.
- */
-function keysOf(entry) {
-    return [entry.hans, entry.hant, ...(entry.hantVariants ?? [])];
-}
-/**
- * A compound's constituent words, and the readings they carry.
- *
- * Indexed by every key an entry claims, so that a 繁體 headword segments into
- * 繁體 constituents. Headwords are set first and never displaced, which is the
- * same precedence the artifact and the build assertions use: 发 and 髮 both claim
- * 发, and the entry the key names has to win.
- */
-class ConstituentIndex {
-    #byKey;
-    #cost;
-    /**
-     * Index entries by key, quantising their frequencies the way the decoder will.
-     */
-    constructor(entries) {
-        const byKey = new Map();
-        for (const entry of entries) {
-            byKey.set(entry.hans, entry);
-        }
-        for (const entry of entries) {
-            for (const form of keysOf(entry)) {
-                if (!byKey.has(form)) {
-                    byKey.set(form, entry);
-                }
-            }
-        }
-        // Costs come from the same quantised table the artifact ships, so that a
-        // segmentation decided here is the one the decoder would have reached.
-        const table = FrequencyTable.build(entries.map((entry) => entry.frequency));
-        const byEntry = new Map();
-        for (const [at, entry] of entries.entries()) {
-            byEntry.set(entry, table.costOf(at));
-        }
-        const cost = new Map();
-        for (const [key, entry] of byKey) {
-            /* c8 ignore next -- every key was indexed from one of these entries */
-            cost.set(key, (byEntry.get(entry) ?? 0) + WORD_CHARGE);
-        }
-        this.#byKey = byKey;
-        this.#cost = cost;
-    }
-    /**
-     * Relax every word starting at one position, improving the costs it reaches.
-     *
-     * One step of the shortest-path scan, lifted out of its enclosing loop.
-     */
-    #relaxFrom(word, characters, from, best, previous) {
-        /* c8 ignore next -- the caller only relaxes from a position in range */
-        const reached = best[from] ?? Infinity;
-        let candidate = "";
-        for (let to = from; to < characters.length; to++) {
-            /* c8 ignore next -- the loop condition keeps the index in range */
-            candidate += characters[to] ?? "";
-            // The whole word is not a segmentation of itself.
-            const entry = candidate === word ? undefined : this.#byKey.get(candidate);
-            const total = reached + this.costOf(candidate);
-            /* c8 ignore next -- `to + 1` is at most `length`, which `best` holds */
-            if (entry !== undefined && total < (best[to + 1] ?? Infinity)) {
-                best[to + 1] = total;
-                previous[to + 1] = { entry, text: candidate, from, to: to + 1 };
-            }
-        }
-    }
-    /**
-     * What a key costs a segmentation that uses it.
-     */
-    costOf(key) {
-        /* c8 ignore next -- only asked about keys the index just returned */
-        return this.#cost.get(key) ?? WORD_CHARGE;
-    }
-    /**
-     * The likeliest way to cut a word into *other* dictionary words.
-     *
-     * The whole word is excluded from its own segmentation, which is the point:
-     * 垃圾分类 is a dictionary word, so a decoder handed the text would stop there
-     * and never ask what it is made of. This asks anyway.
-     *
-     * Returns undefined when no path of known words covers the word, which is
-     * common — most compounds contain a character no shorter word claims.
-     */
-    segment(word) {
-        const characters = toCharacters(word);
-        const length = characters.length;
-        const best = Array.from({ length: length + 1 }, () => Infinity);
-        const previous = Array.from({ length: length + 1 });
-        best[0] = 0;
-        for (let from = 0; from < length; from++) {
-            /* c8 ignore next -- the loop condition keeps the index in range */
-            if ((best[from] ?? Infinity) !== Infinity) {
-                this.#relaxFrom(word, characters, from, best, previous);
-            }
-        }
-        /* c8 ignore next -- `best` is built one longer than the word */
-        if ((best[length] ?? Infinity) === Infinity) {
-            return undefined;
-        }
-        const path = [];
-        let at = length;
-        while (at > 0) {
-            const step = previous[at];
-            /* c8 ignore next 3 -- a finite cost at `at` means a step reached it */
-            if (step === undefined) {
-                return undefined;
-            }
-            path.unshift(step);
-            at = step.from;
-        }
-        return path;
-    }
-}
+import { isSingleCharacter } from "../script/characters.js";
+import { isSameReading } from "./entry.js";
+import { hasMarkedConstituent, isComposable } from "./locale-candidates.js";
+import { ConstituentIndex, keysOf } from "./locale-segments.js";
+export { LOCALE_COMPOSITION_EXCLUSIONS } from "./locale-candidates.js";
 /**
  * The 國語 reading a compound inherits from its constituents, or undefined.
  *
@@ -204,38 +47,6 @@ function composedReading(entry, path, settled) {
     }
     return isContributed ? composed : undefined;
 }
-/**
- * Whether a compound can be asked what it is made of at all.
- *
- * A reading of one syllable per character is what lets a character span be cut
- * out of the compound's reading and replaced. 儿化 folds two characters into one
- * syllable and punctuation reads as none, so neither lines up and both are left
- * alone.
- */
-function isComposable(entry) {
-    const characters = toCharacters(entry.hans);
-    return (characters.length > 2 &&
-        entry.readings.cn.length === characters.length &&
-        !EXCLUDED.has(entry.hans) &&
-        !EXCLUDED.has(entry.hant));
-}
-/**
- * Whether any delta-carrying word appears inside this one.
- *
- * A cheap filter ahead of the segmentation: only a few hundred words carry a
- * delta, against 461,555 entries, and enumerating one word's substrings costs
- * far less than segmenting it.
- */
-function hasMarkedConstituent(word, marked) {
-    const characters = toCharacters(word);
-    return characters.some((_, from) => {
-        let candidate = "";
-        return characters.slice(from).some((character) => {
-            candidate += character;
-            return candidate !== word && marked.has(candidate);
-        });
-    });
-}
 /**
  * How many times the pass may feed its own output back in.
  *

package/dist/dictionary/locale.js.map

@@ -1 +1 @@
-{"version":3,"file":"locale.js","sourceRoot":"","sources":["../../src/dictionary/locale.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,iBAAiB,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAE1E,OAAO,EAAwB,aAAa,EAAE,MAAM,YAAY,CAAC;AACjE,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAEtD;;;;;;;;;;;;;;;;GAgBG;AACH,MAAM,CAAC,MAAM,6BAA6B,GAGpC;IACJ;QACE,IAAI,EAAE,MAAM;QACZ,MAAM,EACJ,6GAA6G;KAChH;IACD;QACE,IAAI,EAAE,MAAM;QACZ,MAAM,EAAE,8DAA8D;KACvE;IACD;QACE,IAAI,EAAE,MAAM;QACZ,MAAM,EACJ,yGAAyG;KAC5G;CACF,CAAC;AAEF;;GAEG;AACH,MAAM,QAAQ,GAAwB,IAAI,GAAG,CAC3C,6BAA6B,CAAC,GAAG,CAAC,CAAC,SAAS,EAAE,EAAE,CAAC,SAAS,CAAC,IAAI,CAAC,CACjE,CAAC;AAEF;;;;;;;;;;;;;GAaG;AACH,MAAM,WAAW,GAAG,EAAE,CAAC;AAcvB;;GAEG;AACH,SAAS,MAAM,CAAC,KAAsB;IACpC,OAAO,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,EAAE,GAAG,CAAC,KAAK,CAAC,YAAY,IAAI,EAAE,CAAC,CAAC,CAAC;AACjE,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,gBAAgB;IACX,MAAM,CAAuC;IAC7C,KAAK,CAA8B;IAE5C;;OAEG;IACH,YAAY,OAAmC;QAC7C,MAAM,KAAK,GAAG,IAAI,GAAG,EAA2B,CAAC;QACjD,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;YAC5B,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;QAC/B,CAAC;QACD,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;YAC5B,KAAK,MAAM,IAAI,IAAI,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC;gBACjC,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;oBACrB,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;gBACzB,CAAC;YACH,CAAC;QACH,CAAC;QAED,yEAAyE;QACzE,uEAAuE;QACvE,MAAM,KAAK,GAAG,cAAc,CAAC,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC,CAAC;QAC5E,MAAM,OAAO,GAAG,IAAI,GAAG,EAA2B,CAAC;QACnD,KAAK,MAAM,CAAC,EAAE,EAAE,KAAK,CAAC,IAAI,OAAO,CAAC,OAAO,EAAE,EAAE,CAAC;YAC5C,OAAO,CAAC,GAAG,CAAC,KAAK,EAAE,KAAK,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,CAAC;QACvC,CAAC;QACD,MAAM,IAAI,GAAG,IAAI,GAAG,EAAkB,CAAC;QACvC,KAAK,MAAM,CAAC,GAAG,EAAE,KAAK,CAAC,IAAI,KAAK,EAAE,CAAC;YACjC,uEAAuE;YACvE,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,GAAG,WAAW,CAAC,CAAC;QACzD,CAAC;QAED,IAAI,CAAC,MAAM,GAAG,KAAK,CAAC;QACpB,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC;IACpB,CAAC;IAED;;;;OAIG;IACH,UAAU,CACR,IAAY,EACZ,UAA6B,EAC7B,IAAY,EACZ,IAAc,EACd,QAAiC;QAEjC,wEAAwE;QACxE,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,CAAC,IAAI,QAAQ,CAAC;QACvC,IAAI,SAAS,GAAG,EAAE,CAAC;QACnB,KAAK,IAAI,EAAE,GAAG,IAAI,EAAE,EAAE,GAAG,UAAU,CAAC,MAAM,EAAE,EAAE,EAAE,EAAE,CAAC;YACjD,mEAAmE;YACnE,SAAS,IAAI,UAAU,CAAC,EAAE,CAAC,IAAI,EAAE,CAAC;YAClC,kDAAkD;YAClD,MAAM,KAAK,GAAG,SAAS,KAAK,IAAI,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;YAC1E,MAAM,KAAK,GAAG,OAAO,GAAG,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC;YAC/C,wEAAwE;YACxE,IAAI,KAAK,KAAK,SAAS,IAAI,KAAK,GAAG,CAAC,IAAI,CAAC,EAAE,GAAG,CAAC,CAAC,IAAI,QAAQ,CAAC,EAAE,CAAC;gBAC9D,IAAI,CAAC,EAAE,GAAG,CAAC,CAAC,GAAG,KAAK,CAAC;gBACrB,QAAQ,CAAC,EAAE,GAAG,CAAC,CAAC,GAAG,EAAE,KAAK,EAAE,IAAI,EAAE,SAAS,EAAE,IAAI,EAAE,EAAE,EAAE,EAAE,GAAG,CAAC,EAAE,CAAC;YAClE,CAAC;QACH,CAAC;IACH,CAAC;IAED;;OAEG;IACH,MAAM,CAAC,GAAW;QAChB,qEAAqE;QACrE,OAAO,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,WAAW,CAAC;IAC5C,CAAC;IAED;;;;;;;;;OASG;IACH,OAAO,CAAC,IAAY;QAClB,MAAM,UAAU,GAAG,YAAY,CAAC,IAAI,CAAC,CAAC;QACtC,MAAM,MAAM,GAAG,UAAU,CAAC,MAAM,CAAC;QACjC,MAAM,IAAI,GAAG,KAAK,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,MAAM,GAAG,CAAC,EAAE,EAAE,GAAG,EAAE,CAAC,QAAQ,CAAC,CAAC;QAChE,MAAM,QAAQ,GAAG,KAAK,CAAC,IAAI,CAAsB,EAAE,MAAM,EAAE,MAAM,GAAG,CAAC,EAAE,CAAC,CAAC;QACzE,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC;QAEZ,KAAK,IAAI,IAAI,GAAG,CAAC,EAAE,IAAI,GAAG,MAAM,EAAE,IAAI,EAAE,EAAE,CAAC;YACzC,mEAAmE;YACnE,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,QAAQ,CAAC,KAAK,QAAQ,EAAE,CAAC;gBAC1C,IAAI,CAAC,UAAU,CAAC,IAAI,EAAE,UAAU,EAAE,IAAI,EAAE,IAAI,EAAE,QAAQ,CAAC,CAAC;YAC1D,CAAC;QACH,CAAC;QAED,gEAAgE;QAChE,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,QAAQ,CAAC,KAAK,QAAQ,EAAE,CAAC;YAC5C,OAAO,SAAS,CAAC;QACnB,CAAC;QAED,MAAM,IAAI,GAAc,EAAE,CAAC;QAC3B,IAAI,EAAE,GAAG,MAAM,CAAC;QAChB,OAAO,EAAE,GAAG,CAAC,EAAE,CAAC;YACd,MAAM,IAAI,GAAG,QAAQ,CAAC,EAAE,CAAC,CAAC;YAC1B,uEAAuE;YACvE,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;gBACvB,OAAO,SAAS,CAAC;YACnB,CAAC;YACD,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC;YACnB,EAAE,GAAG,IAAI,CAAC,IAAI,CAAC;QACjB,CAAC;QACD,OAAO,IAAI,CAAC;IACd,CAAC;CACF;AAED;;;;;;;;;;GAUG;AACH,SAAS,eAAe,CACtB,KAAsB,EACtB,IAAwB,EACxB,OAAiD;IAEjD,MAAM,OAAO,GAAG,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC;IAClC,MAAM,QAAQ,GAAe,EAAE,CAAC;IAChC,IAAI,aAAa,GAAG,KAAK,CAAC;IAE1B,KAAK,MAAM,IAAI,IAAI,IAAI,EAAE,CAAC;QACxB,MAAM,IAAI,GAAG,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,EAAE,CAAC,CAAC;QAC/C,4EAA4E;QAC5E,0CAA0C;QAC1C,MAAM,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC;QACtE,IAAI,MAAM,KAAK,SAAS,IAAI,iBAAiB,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YACzD,QAAQ,CAAC,IAAI,CAAC,GAAG,IAAI,CAAC,CAAC;YACvB,SAAS;QACX,CAAC;QACD,0EAA0E;QAC1E,2EAA2E;QAC3E,uEAAuE;QACvE,uEAAuE;QACvE,0CAA0C;QAC1C,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,EAAE,IAAI,CAAC,EAAE,CAAC;YACjD,OAAO,SAAS,CAAC;QACnB,CAAC;QACD,QAAQ,CAAC,IAAI,CAAC,GAAG,MAAM,CAAC,CAAC;QACzB,aAAa,GAAG,IAAI,CAAC;IACvB,CAAC;IAED,OAAO,aAAa,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,SAAS,CAAC;AAC9C,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,YAAY,CAAC,KAAsB;IAC1C,MAAM,UAAU,GAAG,YAAY,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IAC5C,OAAO,CACL,UAAU,CAAC,MAAM,GAAG,CAAC;QACrB,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,MAAM,KAAK,UAAU,CAAC,MAAM;QAC9C,CAAC,QAAQ,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC;QACzB,CAAC,QAAQ,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,CAC1B,CAAC;AACJ,CAAC;AAED;;;;;;GAMG;AACH,SAAS,oBAAoB,CAC3B,IAAY,EACZ,MAA2B;IAE3B,MAAM,UAAU,GAAG,YAAY,CAAC,IAAI,CAAC,CAAC;IACtC,OAAO,UAAU,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,EAAE;QACjC,IAAI,SAAS,GAAG,EAAE,CAAC;QACnB,OAAO,UAAU,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,CAAC,SAAS,EAAE,EAAE;YAC/C,SAAS,IAAI,SAAS,CAAC;YACvB,OAAO,SAAS,KAAK,IAAI,IAAI,MAAM,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QACrD,CAAC,CAAC,CAAC;IACL,CAAC,CAAC,CAAC;AACL,CAAC;AAWD;;;;;;;GAOG;AACH,MAAM,MAAM,GAAG,CAAC,CAAC;AAEjB;;;;;;;;;;;;;;;;;;;;;;GAsBG;AACH,MAAM,UAAU,mBAAmB,CACjC,OAAmC;IAEnC,MAAM,KAAK,GAAG,IAAI,gBAAgB,CAAC,OAAO,CAAC,CAAC;IAE5C,2EAA2E;IAC3E,yEAAyE;IACzE,2EAA2E;IAC3E,eAAe;IACf,MAAM,MAAM,GAAG,IAAI,GAAG,EAAU,CAAC;IACjC,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,IAAI,KAAK,CAAC,QAAQ,CAAC,EAAE,KAAK,SAAS,IAAI,CAAC,iBAAiB,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC;YACtE,KAAK,MAAM,GAAG,IAAI,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC;gBAChC,MAAM,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YAClB,CAAC;QACH,CAAC;IACH,CAAC;IAED,8EAA8E;IAC9E,qEAAqE;IACrE,MAAM,UAAU,GAAG,OAAO,CAAC,MAAM,CAC/B,CAAC,KAAK,EAAE,EAAE,CACR,KAAK,CAAC,QAAQ,CAAC,EAAE,KAAK,SAAS;QAC/B,YAAY,CAAC,KAAK,CAAC;QACnB,oBAAoB,CAAC,KAAK,CAAC,IAAI,EAAE,MAAM,CAAC,CAC3C,CAAC;IAEF,MAAM,OAAO,GAAG,IAAI,GAAG,EAA+B,CAAC;IAEvD;;OAEG;IACH,MAAM,UAAU,GAAG,CAAC,KAAsB,EAAW,EAAE;QACrD,MAAM,IAAI,GAAG,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QACvC,MAAM,OAAO,GACX,IAAI,KAAK,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,eAAe,CAAC,KAAK,EAAE,IAAI,EAAE,OAAO,CAAC,CAAC;QACzE,sEAAsE;QACtE,mEAAmE;QACnE,IAAI,OAAO,KAAK,SAAS,IAAI,aAAa,CAAC,OAAO,EAAE,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,EAAE,CAAC;YACvE,OAAO,KAAK,CAAC;QACf,CAAC;QACD,MAAM,IAAI,GAAG,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QACrC,IAAI,IAAI,KAAK,SAAS,IAAI,aAAa,CAAC,IAAI,EAAE,OAAO,CAAC,EAAE,CAAC;YACvD,OAAO,KAAK,CAAC;QACf,CAAC;QACD,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;QACjC,OAAO,IAAI,CAAC;IACd,CAAC,CAAC;IAEF,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC;QAC5C,MAAM,OAAO,GAAG,UAAU,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC;QACvE,IAAI,OAAO,KAAK,CAAC,EAAE,CAAC;YAClB,MAAM;QACR,CAAC;IACH,CAAC;IAED,MAAM,QAAQ,GAAG,OAAO,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE;QACrC,MAAM,OAAO,GAAG,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QACxC,OAAO,OAAO,KAAK,SAAS;YAC1B,CAAC,CAAC,KAAK;YACP,CAAC,CAAC,EAAE,GAAG,KAAK,EAAE,QAAQ,EAAE,EAAE,GAAG,KAAK,CAAC,QAAQ,EAAE,EAAE,EAAE,OAAO,EAAE,EAAE,CAAC;IACjE,CAAC,CAAC,CAAC;IAEH,OAAO,EAAE,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,OAAO,CAAC,IAAI,EAAE,CAAC;AACvD,CAAC"}
+{"version":3,"file":"locale.js","sourceRoot":"","sources":["../../src/dictionary/locale.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,EAAE,iBAAiB,EAAE,MAAM,yBAAyB,CAAC;AAE5D,OAAO,EAAwB,aAAa,EAAE,MAAM,YAAY,CAAC;AACjE,OAAO,EAAE,oBAAoB,EAAE,YAAY,EAAE,MAAM,wBAAwB,CAAC;AAC5E,OAAO,EAAE,gBAAgB,EAAE,MAAM,EAAgB,MAAM,sBAAsB,CAAC;AAE9E,OAAO,EAAE,6BAA6B,EAAE,MAAM,wBAAwB,CAAC;AACvE;;;;;;;;;;GAUG;AACH,SAAS,eAAe,CACtB,KAAsB,EACtB,IAAwB,EACxB,OAAiD;IAEjD,MAAM,OAAO,GAAG,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC;IAClC,MAAM,QAAQ,GAAe,EAAE,CAAC;IAChC,IAAI,aAAa,GAAG,KAAK,CAAC;IAE1B,KAAK,MAAM,IAAI,IAAI,IAAI,EAAE,CAAC;QACxB,MAAM,IAAI,GAAG,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,EAAE,CAAC,CAAC;QAC/C,4EAA4E;QAC5E,0CAA0C;QAC1C,MAAM,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC;QACtE,IAAI,MAAM,KAAK,SAAS,IAAI,iBAAiB,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YACzD,QAAQ,CAAC,IAAI,CAAC,GAAG,IAAI,CAAC,CAAC;YACvB,SAAS;QACX,CAAC;QACD,0EAA0E;QAC1E,2EAA2E;QAC3E,uEAAuE;QACvE,uEAAuE;QACvE,0CAA0C;QAC1C,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,EAAE,IAAI,CAAC,EAAE,CAAC;YACjD,OAAO,SAAS,CAAC;QACnB,CAAC;QACD,QAAQ,CAAC,IAAI,CAAC,GAAG,MAAM,CAAC,CAAC;QACzB,aAAa,GAAG,IAAI,CAAC;IACvB,CAAC;IAED,OAAO,aAAa,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,SAAS,CAAC;AAC9C,CAAC;AAWD;;;;;;;GAOG;AACH,MAAM,MAAM,GAAG,CAAC,CAAC;AAEjB;;;;;;;;;;;;;;;;;;;;;;GAsBG;AACH,MAAM,UAAU,mBAAmB,CACjC,OAAmC;IAEnC,MAAM,KAAK,GAAG,IAAI,gBAAgB,CAAC,OAAO,CAAC,CAAC;IAE5C,2EAA2E;IAC3E,yEAAyE;IACzE,2EAA2E;IAC3E,eAAe;IACf,MAAM,MAAM,GAAG,IAAI,GAAG,EAAU,CAAC;IACjC,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,IAAI,KAAK,CAAC,QAAQ,CAAC,EAAE,KAAK,SAAS,IAAI,CAAC,iBAAiB,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC;YACtE,KAAK,MAAM,GAAG,IAAI,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC;gBAChC,MAAM,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YAClB,CAAC;QACH,CAAC;IACH,CAAC;IAED,8EAA8E;IAC9E,qEAAqE;IACrE,MAAM,UAAU,GAAG,OAAO,CAAC,MAAM,CAC/B,CAAC,KAAK,EAAE,EAAE,CACR,KAAK,CAAC,QAAQ,CAAC,EAAE,KAAK,SAAS;QAC/B,YAAY,CAAC,KAAK,CAAC;QACnB,oBAAoB,CAAC,KAAK,CAAC,IAAI,EAAE,MAAM,CAAC,CAC3C,CAAC;IAEF,MAAM,OAAO,GAAG,IAAI,GAAG,EAA+B,CAAC;IAEvD;;OAEG;IACH,MAAM,UAAU,GAAG,CAAC,KAAsB,EAAW,EAAE;QACrD,MAAM,IAAI,GAAG,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QACvC,MAAM,OAAO,GACX,IAAI,KAAK,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,eAAe,CAAC,KAAK,EAAE,IAAI,EAAE,OAAO,CAAC,CAAC;QACzE,sEAAsE;QACtE,mEAAmE;QACnE,IAAI,OAAO,KAAK,SAAS,IAAI,aAAa,CAAC,OAAO,EAAE,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,EAAE,CAAC;YACvE,OAAO,KAAK,CAAC;QACf,CAAC;QACD,MAAM,IAAI,GAAG,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QACrC,IAAI,IAAI,KAAK,SAAS,IAAI,aAAa,CAAC,IAAI,EAAE,OAAO,CAAC,EAAE,CAAC;YACvD,OAAO,KAAK,CAAC;QACf,CAAC;QACD,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;QACjC,OAAO,IAAI,CAAC;IACd,CAAC,CAAC;IAEF,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC;QAC5C,MAAM,OAAO,GAAG,UAAU,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC;QACvE,IAAI,OAAO,KAAK,CAAC,EAAE,CAAC;YAClB,MAAM;QACR,CAAC;IACH,CAAC;IAED,MAAM,QAAQ,GAAG,OAAO,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE;QACrC,MAAM,OAAO,GAAG,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QACxC,OAAO,OAAO,KAAK,SAAS;YAC1B,CAAC,CAAC,KAAK;YACP,CAAC,CAAC,EAAE,GAAG,KAAK,EAAE,QAAQ,EAAE,EAAE,GAAG,KAAK,CAAC,QAAQ,EAAE,EAAE,EAAE,OAAO,EAAE,EAAE,CAAC;IACjE,CAAC,CAAC,CAAC;IAEH,OAAO,EAAE,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,OAAO,CAAC,IAAI,EAAE,CAAC;AACvD,CAAC"}

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

@@ -0,0 +1,121 @@
+/**
+ * What the merge takes in and what it hands back.
+ *
+ * Their own module so that a step of the merge can name them without
+ * importing the merge itself.
+ */
+import type { CedictEntry } from "../sources/cedict.js";
+import type { JiebaEntry } from "../sources/jieba.js";
+import type { UnihanReadings, UnihanVariants } from "../sources/unihan.js";
+import type { Syllable } from "../syllable/syllable.js";
+import type { DictionaryEntry } from "./entry.js";
+import type { TraditionalTable } from "./traditional-table.js";
+/**
+ * The parsed sources the merge combines.
+ */
+export interface MergeSources {
+    readonly unihanReadings: ReadonlyMap<string, UnihanReadings>;
+    readonly unihanVariants: UnihanVariants;
+    readonly phrase: ReadonlyMap<string, readonly string[]>;
+    readonly cedict: readonly CedictEntry[];
+    readonly jieba: ReadonlyMap<string, JiebaEntry>;
+}
+/**
+ * Counts describing what the merge did, for the build to report and check.
+ */
+export interface MergeStats {
+    readonly characters: number;
+    readonly phraseWords: number;
+    readonly cedictWords: number;
+    /** Words whose reading CC-CEDICT corrected to a neutral tone. */
+    readonly neutralToneCorrections: number;
+    /**
+     * Characters whose default was a 轻声 the frequency field only counted inside
+     * words — see {@link import("./frequency-tones.js").demoteReducedNeutrals}.
+     */
+    readonly reducedNeutrals: number;
+    /** Words whose trailing 儿 was folded into the syllable before it. */
+    readonly erhuaRepairs: 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. */
+    readonly scriptPairs: number;
+    /** Entries a source writes more than one 繁體 spelling of. */
+    readonly variantSpellings: number;
+    /** Entries carrying a zh-TW reading delta a source stated. */
+    readonly taiwanReadings: number;
+    /** Compounds given a zh-TW delta composed from their constituents. */
+    readonly composedTaiwanReadings: number;
+    /**
+     * Words jieba tagged a proper noun that CC-CEDICT's lowercase pinyin vetoed.
+     */
+    readonly properNounVetoes: number;
+    /** Entries whose parts CC-CEDICT's capitalisation divides. */
+    readonly nameBoundaries: number;
+    /** Words dropped because no source gave a usable reading. */
+    readonly rejected: number;
+}
+/**
+ * What the merge produced.
+ */
+export interface MergeResult {
+    /** Entries, ordered by their 简体 form. */
+    readonly entries: readonly DictionaryEntry[];
+    /** Words no source could be read, with the reading that failed. */
+    readonly rejected: ReadonlyMap<string, readonly string[]>;
+    readonly stats: MergeStats;
+}
+/**
+ * What the sources between them say about one word, indexed once for all of
+ * them.
+ *
+ * The per-word view of {@link MergeSources}: the same data, turned inside out
+ * so that merging a word is a handful of map reads rather than a scan.
+ */
+export interface WordSources {
+    readonly cedictByWord: ReadonlyMap<string, readonly CedictEntry[]>;
+    readonly cedictByHant: ReadonlyMap<string, readonly CedictEntry[]>;
+    readonly phrase: ReadonlyMap<string, readonly string[]>;
+    readonly jieba: ReadonlyMap<string, JiebaEntry>;
+    readonly unihanReadings: ReadonlyMap<string, UnihanReadings>;
+    readonly traditional: TraditionalTable;
+    readonly defaults: ReadonlyMap<string, readonly Syllable[]>;
+}
+/**
+ * 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` and `composedTaiwanReadings` are settled by passes
+ * over the whole dictionary, and `rejected` is counted by the caller.
+ */
+export interface WordTally {
+    readonly neutralToneCorrections: number;
+    readonly erhuaRepairs: number;
+    readonly derivedTraditional: number;
+    readonly scriptPairs: number;
+    readonly variantSpellings: number;
+    readonly taiwanReadings: number;
+    readonly properNounVetoes: number;
+    readonly nameBoundaries: number;
+    readonly characters: number;
+    readonly phraseWords: number;
+    readonly cedictWords: number;
+}
+/**
+ * A word that contributed nothing, because no source could read it.
+ */
+export declare const NO_TALLY: WordTally;
+/**
+ * One word's entry, or the reading that failed.
+ */
+export interface MergedWord {
+    readonly entry: DictionaryEntry | undefined;
+    /** The reading no source could make sense of, where that is what happened. */
+    readonly rejected: readonly string[] | undefined;
+    readonly tally: WordTally;
+}
+/**
+ * Add one word's tally to the running total.
+ */
+export declare function addTally(total: WordTally, tally: WordTally): WordTally;
+//# sourceMappingURL=merge-types.d.ts.map

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

@@ -0,0 +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,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;;;;;;GAMG;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

@@ -0,0 +1,35 @@
+/**
+ * A word that contributed nothing, because no source could read it.
+ */
+export const NO_TALLY = {
+    neutralToneCorrections: 0,
+    erhuaRepairs: 0,
+    derivedTraditional: 0,
+    scriptPairs: 0,
+    variantSpellings: 0,
+    taiwanReadings: 0,
+    properNounVetoes: 0,
+    nameBoundaries: 0,
+    characters: 0,
+    phraseWords: 0,
+    cedictWords: 0,
+};
+/**
+ * Add one word's tally to the running total.
+ */
+export function addTally(total, tally) {
+    return {
+        neutralToneCorrections: total.neutralToneCorrections + tally.neutralToneCorrections,
+        erhuaRepairs: total.erhuaRepairs + tally.erhuaRepairs,
+        derivedTraditional: total.derivedTraditional + tally.derivedTraditional,
+        scriptPairs: total.scriptPairs + tally.scriptPairs,
+        variantSpellings: total.variantSpellings + tally.variantSpellings,
+        taiwanReadings: total.taiwanReadings + tally.taiwanReadings,
+        properNounVetoes: total.properNounVetoes + tally.properNounVetoes,
+        nameBoundaries: total.nameBoundaries + tally.nameBoundaries,
+        characters: total.characters + tally.characters,
+        phraseWords: total.phraseWords + tally.phraseWords,
+        cedictWords: total.cedictWords + tally.cedictWords,
+    };
+}
+//# sourceMappingURL=merge-types.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"merge-types.js","sourceRoot":"","sources":["../../src/dictionary/merge-types.ts"],"names":[],"mappings":"AA6GA;;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-word.d.ts

@@ -0,0 +1,7 @@
+import { type MergedWord, type WordSources } from "./merge-types.js";
+export type { MergedWord, WordSources, WordTally } from "./merge-types.js";
+/**
+ * Merge one word.
+ */
+export declare function mergeWord(word: string, sources: WordSources): MergedWord;
+//# sourceMappingURL=merge-word.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"merge-word.d.ts","sourceRoot":"","sources":["../../src/dictionary/merge-word.ts"],"names":[],"mappings":"AAgBA,OAAO,EAAE,KAAK,UAAU,EAAY,KAAK,WAAW,EAAE,MAAM,kBAAkB,CAAC;AAE/E,YAAY,EAAE,UAAU,EAAE,WAAW,EAAE,SAAS,EAAE,MAAM,kBAAkB,CAAC;AAE3E;;GAEG;AACH,wBAAgB,SAAS,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,WAAW,GAAG,UAAU,CA8GxE"}

package/dist/dictionary/merge-word.js

@@ -0,0 +1,91 @@
+/**
+ * Everything the merge decides about one word.
+ *
+ * The phases run in the order MERGE.md sets out and each depends on the one
+ * before, so they stay in one function rather than becoming a pipeline of
+ * small ones. What has changed is that the counts come back as a tally
+ * instead of being written into eleven variables the caller owns.
+ */
+import { isSingleCharacter } from "../script/characters.js";
+import { isSameReading } from "./entry.js";
+import { readAlignedReading } from "./reading.js";
+import { cedictReadingsOf } from "./cedict-senses.js";
+import { traditionalFormOf } from "./traditional-form.js";
+import { taiwanReadingOf } from "./taiwan-reading.js";
+import { properNounOf } from "./proper-noun.js";
+import { settleReading } from "./word-reading.js";
+import { NO_TALLY } from "./merge-types.js";
+/**
+ * Merge one word.
+ */
+export function mergeWord(word, sources) {
+    const { cedictByWord, cedictByHant, phrase, jieba, unihanReadings, traditional, defaults, } = sources;
+    const cedictEntries = cedictByWord.get(word) ?? [];
+    const cedictReadings = cedictReadingsOf(word, cedictEntries);
+    // Every reading CC-CEDICT gives this spelling, under either script.
+    const senseReadings = [
+        ...cedictReadings,
+        ...cedictReadingsOf(word, cedictByHant.get(word) ?? []),
+    ];
+    const phraseReading = phrase.get(word);
+    const phraseAligned = phraseReading === undefined
+        ? undefined
+        : readAlignedReading(word, phraseReading);
+    const settled = settleReading(word, defaults, cedictEntries, cedictReadings, phraseAligned);
+    if (settled.reading === undefined) {
+        return {
+            entry: undefined,
+            rejected: phraseReading ?? cedictEntries[0]?.readings ?? [],
+            tally: NO_TALLY,
+        };
+    }
+    const reading = settled.reading;
+    const aligned = settled.aligned;
+    // ── 繁體: taken from CC-CEDICT, derived where it is silent ─
+    const { senses, hant, hantVariants, isDerived } = traditionalFormOf(word, cedictEntries, reading, aligned, traditional);
+    // ── zh-TW delta ───────────────────────────────────────────
+    const taiwan = taiwanReadingOf(word, senses, senseReadings, reading, unihanReadings.get(word));
+    // ── Frequency, part of speech and the proper noun bit ─────
+    const jiebaEntry = jieba.get(word);
+    const { isProperNoun, boundaries, isVetoed } = properNounOf(word, jiebaEntry, cedictEntries, senses, reading);
+    // ── Polyphone priors, for single characters only ──────────
+    const isCharacter = isSingleCharacter(word);
+    const characterReadings = defaults.get(word) ?? [];
+    const alternates = isCharacter
+        ? characterReadings
+            .filter((syllable) => !isSameReading([syllable], reading))
+            .map((syllable) => [syllable])
+        : [];
+    const entry = {
+        hans: word,
+        hant,
+        ...(hantVariants.length > 0 && { hantVariants }),
+        readings: { cn: reading, ...(taiwan !== undefined && { tw: taiwan }) },
+        frequency: jiebaEntry?.frequency ?? 0,
+        partOfSpeech: jiebaEntry?.partOfSpeech ?? "",
+        isProperNoun,
+        ...(boundaries.length > 0 && { nameBoundaries: boundaries }),
+        ...(alternates.length > 0 && { alternates }),
+    };
+    // Read off the decisions above rather than counted as they were made: every
+    // one of these is a fact about the finished entry, so a running total would
+    // only be a second place for it to be recorded.
+    return {
+        entry,
+        rejected: undefined,
+        tally: {
+            neutralToneCorrections: settled.neutralToneCorrections,
+            erhuaRepairs: settled.erhuaRepairs,
+            derivedTraditional: isDerived ? 1 : 0,
+            scriptPairs: hant === word ? 0 : 1,
+            variantSpellings: hantVariants.length > 0 ? 1 : 0,
+            taiwanReadings: taiwan === undefined ? 0 : 1,
+            properNounVetoes: isVetoed ? 1 : 0,
+            nameBoundaries: boundaries.length > 0 ? 1 : 0,
+            characters: isCharacter ? 1 : 0,
+            phraseWords: phraseReading === undefined ? 0 : 1,
+            cedictWords: cedictEntries.length > 0 ? 1 : 0,
+        },
+    };
+}
+//# sourceMappingURL=merge-word.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"merge-word.js","sourceRoot":"","sources":["../../src/dictionary/merge-word.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AACH,OAAO,EAAE,iBAAiB,EAAE,MAAM,yBAAyB,CAAC;AAC5D,OAAO,EAAwB,aAAa,EAAE,MAAM,YAAY,CAAC;AACjE,OAAO,EAAE,kBAAkB,EAAE,MAAM,cAAc,CAAC;AAClD,OAAO,EAAE,gBAAgB,EAAE,MAAM,oBAAoB,CAAC;AACtD,OAAO,EAAE,iBAAiB,EAAE,MAAM,uBAAuB,CAAC;AAC1D,OAAO,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAC;AACtD,OAAO,EAAE,YAAY,EAAE,MAAM,kBAAkB,CAAC;AAChD,OAAO,EAAE,aAAa,EAAE,MAAM,mBAAmB,CAAC;AAClD,OAAO,EAAmB,QAAQ,EAAoB,MAAM,kBAAkB,CAAC;AAI/E;;GAEG;AACH,MAAM,UAAU,SAAS,CAAC,IAAY,EAAE,OAAoB;IAC1D,MAAM,EACJ,YAAY,EACZ,YAAY,EACZ,MAAM,EACN,KAAK,EACL,cAAc,EACd,WAAW,EACX,QAAQ,GACT,GAAG,OAAO,CAAC;IAEZ,MAAM,aAAa,GAAG,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;IACnD,MAAM,cAAc,GAAG,gBAAgB,CAAC,IAAI,EAAE,aAAa,CAAC,CAAC;IAC7D,oEAAoE;IACpE,MAAM,aAAa,GAAG;QACpB,GAAG,cAAc;QACjB,GAAG,gBAAgB,CAAC,IAAI,EAAE,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;KACxD,CAAC;IACF,MAAM,aAAa,GAAG,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IACvC,MAAM,aAAa,GACjB,aAAa,KAAK,SAAS;QACzB,CAAC,CAAC,SAAS;QACX,CAAC,CAAC,kBAAkB,CAAC,IAAI,EAAE,aAAa,CAAC,CAAC;IAE9C,MAAM,OAAO,GAAG,aAAa,CAC3B,IAAI,EACJ,QAAQ,EACR,aAAa,EACb,cAAc,EACd,aAAa,CACd,CAAC;IACF,IAAI,OAAO,CAAC,OAAO,KAAK,SAAS,EAAE,CAAC;QAClC,OAAO;YACL,KAAK,EAAE,SAAS;YAChB,QAAQ,EAAE,aAAa,IAAI,aAAa,CAAC,CAAC,CAAC,EAAE,QAAQ,IAAI,EAAE;YAC3D,KAAK,EAAE,QAAQ;SAChB,CAAC;IACJ,CAAC;IACD,MAAM,OAAO,GAAG,OAAO,CAAC,OAAO,CAAC;IAChC,MAAM,OAAO,GAAG,OAAO,CAAC,OAAO,CAAC;IAEhC,4DAA4D;IAC5D,MAAM,EAAE,MAAM,EAAE,IAAI,EAAE,YAAY,EAAE,SAAS,EAAE,GAAG,iBAAiB,CACjE,IAAI,EACJ,aAAa,EACb,OAAO,EACP,OAAO,EACP,WAAW,CACZ,CAAC;IAEF,6DAA6D;IAC7D,MAAM,MAAM,GAAG,eAAe,CAC5B,IAAI,EACJ,MAAM,EACN,aAAa,EACb,OAAO,EACP,cAAc,CAAC,GAAG,CAAC,IAAI,CAAC,CACzB,CAAC;IAEF,6DAA6D;IAC7D,MAAM,UAAU,GAAG,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IACnC,MAAM,EAAE,YAAY,EAAE,UAAU,EAAE,QAAQ,EAAE,GAAG,YAAY,CACzD,IAAI,EACJ,UAAU,EACV,aAAa,EACb,MAAM,EACN,OAAO,CACR,CAAC;IAEF,6DAA6D;IAC7D,MAAM,WAAW,GAAG,iBAAiB,CAAC,IAAI,CAAC,CAAC;IAC5C,MAAM,iBAAiB,GAAG,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;IACnD,MAAM,UAAU,GAAG,WAAW;QAC5B,CAAC,CAAC,iBAAiB;aACd,MAAM,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC,aAAa,CAAC,CAAC,QAAQ,CAAC,EAAE,OAAO,CAAC,CAAC;aACzD,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC,QAAQ,CAAC,CAAC;QAClC,CAAC,CAAC,EAAE,CAAC;IAEP,MAAM,KAAK,GAAoB;QAC7B,IAAI,EAAE,IAAI;QACV,IAAI;QACJ,GAAG,CAAC,YAAY,CAAC,MAAM,GAAG,CAAC,IAAI,EAAE,YAAY,EAAE,CAAC;QAChD,QAAQ,EAAE,EAAE,EAAE,EAAE,OAAO,EAAE,GAAG,CAAC,MAAM,KAAK,SAAS,IAAI,EAAE,EAAE,EAAE,MAAM,EAAE,CAAC,EAAE;QACtE,SAAS,EAAE,UAAU,EAAE,SAAS,IAAI,CAAC;QACrC,YAAY,EAAE,UAAU,EAAE,YAAY,IAAI,EAAE;QAC5C,YAAY;QACZ,GAAG,CAAC,UAAU,CAAC,MAAM,GAAG,CAAC,IAAI,EAAE,cAAc,EAAE,UAAU,EAAE,CAAC;QAC5D,GAAG,CAAC,UAAU,CAAC,MAAM,GAAG,CAAC,IAAI,EAAE,UAAU,EAAE,CAAC;KAC7C,CAAC;IAEF,4EAA4E;IAC5E,4EAA4E;IAC5E,gDAAgD;IAChD,OAAO;QACL,KAAK;QACL,QAAQ,EAAE,SAAS;QACnB,KAAK,EAAE;YACL,sBAAsB,EAAE,OAAO,CAAC,sBAAsB;YACtD,YAAY,EAAE,OAAO,CAAC,YAAY;YAClC,kBAAkB,EAAE,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YACrC,WAAW,EAAE,IAAI,KAAK,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAClC,gBAAgB,EAAE,YAAY,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YACjD,cAAc,EAAE,MAAM,KAAK,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAC5C,gBAAgB,EAAE,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAClC,cAAc,EAAE,UAAU,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAC7C,UAAU,EAAE,WAAW,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAC/B,WAAW,EAAE,aAAa,KAAK,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAChD,WAAW,EAAE,aAAa,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;SAC9C;KACF,CAAC;AACJ,CAAC"}

package/dist/dictionary/merge.d.ts

@@ -1,70 +1,13 @@
-import { type CedictEntry } from "../sources/cedict.js";
-import { type JiebaEntry } from "../sources/jieba.js";
-import { type UnihanReadings, type UnihanVariants } from "../sources/unihan.js";
-import { type DictionaryEntry } from "./entry.js";
-/**
- * The parsed sources the merge combines.
- */
-export interface MergeSources {
-    readonly unihanReadings: ReadonlyMap<string, UnihanReadings>;
-    readonly unihanVariants: UnihanVariants;
-    readonly phrase: ReadonlyMap<string, readonly string[]>;
-    readonly cedict: readonly CedictEntry[];
-    readonly jieba: ReadonlyMap<string, JiebaEntry>;
-}
-/**
- * Counts describing what the merge did, for the build to report and check.
- */
-export interface MergeStats {
-    readonly characters: number;
-    readonly phraseWords: number;
-    readonly cedictWords: number;
-    /** Words whose reading CC-CEDICT corrected to a neutral tone. */
-    readonly neutralToneCorrections: number;
-    /**
-     * Characters whose default was a 轻声 the frequency field only counted inside
-     * words — see {@link demoteReducedNeutrals}.
-     */
-    readonly reducedNeutrals: number;
-    /** Words whose trailing 儿 was folded into the syllable before it. */
-    readonly erhuaRepairs: 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. */
-    readonly scriptPairs: number;
-    /** Entries a source writes more than one 繁體 spelling of. */
-    readonly variantSpellings: number;
-    /** Entries carrying a zh-TW reading delta a source stated. */
-    readonly taiwanReadings: number;
-    /** Compounds given a zh-TW delta composed from their constituents. */
-    readonly composedTaiwanReadings: number;
-    /**
-     * Words jieba tagged a proper noun that CC-CEDICT's lowercase pinyin vetoed.
-     */
-    readonly properNounVetoes: number;
-    /** Entries whose parts CC-CEDICT's capitalisation divides. */
-    readonly nameBoundaries: number;
-    /** Words dropped because no source gave a usable reading. */
-    readonly rejected: number;
-}
-/**
- * What the merge produced.
- */
-export interface MergeResult {
-    /** Entries, ordered by their 简体 form. */
-    readonly entries: readonly DictionaryEntry[];
-    /** Words no source could be read, with the reading that failed. */
-    readonly rejected: ReadonlyMap<string, readonly string[]>;
-    readonly stats: MergeStats;
-}
+import { type MergeResult, type MergeSources } from "./merge-types.js";
+export type { MergeResult, MergeSources, MergeStats } from "./merge-types.js";
 /**
  * Build the merged dictionary from the four parsed sources.
  *
  * The order of operations is the one MERGE.md sets out, and it is not
  * arbitrary — each step depends on the one before. Spelling, tones, sandhi and
- * validation happen in {@link readAlignedReading} as each source is read; this
- * function then repairs 儿化, derives 繁體 forms, resolves the disagreements
- * between sources, and finally applies the override table.
+ * validation happen in {@link import("./reading.js").readAlignedReading} as each source is read;
+ * {@link mergeWord} then settles one word at a time, and this walks the words
+ * and gathers what each of them came to.
  */
 export declare function mergeSources(sources: MergeSources): MergeResult;
 //# sourceMappingURL=merge.d.ts.map

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

@@ -1 +1 @@
-{"version":3,"file":"merge.d.ts","sourceRoot":"","sources":["../../src/dictionary/merge.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,KAAK,WAAW,EAAoB,MAAM,sBAAsB,CAAC;AAC1E,OAAO,EAAmB,KAAK,UAAU,EAAE,MAAM,qBAAqB,CAAC;AACvE,OAAO,EAEL,KAAK,cAAc,EACnB,KAAK,cAAc,EACpB,MAAM,sBAAsB,CAAC;AAI9B,OAAO,EAAE,KAAK,eAAe,EAAiB,MAAM,YAAY,CAAC;AAgBjE;;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,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;AAieD;;;;;;;;GAQG;AACH,wBAAgB,YAAY,CAAC,OAAO,EAAE,YAAY,GAAG,WAAW,CAyX/D"}
+{"version":3,"file":"merge.d.ts","sourceRoot":"","sources":["../../src/dictionary/merge.ts"],"names":[],"mappings":"AASA,OAAO,EAEL,KAAK,WAAW,EAChB,KAAK,YAAY,EAGlB,MAAM,kBAAkB,CAAC;AAE1B,YAAY,EAAE,WAAW,EAAE,YAAY,EAAE,UAAU,EAAE,MAAM,kBAAkB,CAAC;AAE9E;;;;;;;;GAQG;AACH,wBAAgB,YAAY,CAAC,OAAO,EAAE,YAAY,GAAG,WAAW,CAkE/D"}