@kensio/pinyinjs 1.7.4 → 1.7.6

package/dist/dictionary/traditional-ranking.js

@@ -0,0 +1,74 @@
+import { readSyllable } from "../syllable/syllable.js";
+/**
+ * Unihan's reading lists, ranked against a syllable.
+ */
+export class VariantRanking {
+    #unihanReadings;
+    #parsed = new Map();
+    constructor(unihanReadings) {
+        this.#unihanReadings = unihanReadings;
+    }
+    /**
+     * Parse a Unihan reading string, remembering the result.
+     *
+     * Worth caching: the fallback path reads the same few thousand strings once
+     * per candidate per word.
+     */
+    #syllableOf(reading) {
+        if (this.#parsed.has(reading)) {
+            return this.#parsed.get(reading);
+        }
+        const syllable = readSyllable(reading);
+        this.#parsed.set(reading, syllable);
+        return syllable;
+    }
+    /**
+     * How well a candidate character's readings match the one we want.
+     *
+     * Lower is better. A candidate whose most likely reading is the one we want
+     * beats one that merely lists it among its rarer readings, which is exactly
+     * what separates 髮 (`fà` alone) from 發 (`fā` first, `fà` third).
+     */
+    #rank(candidate, syllable) {
+        const readings = this.#unihanReadings.get(candidate)?.readings ?? [];
+        if (syllable === undefined) {
+            return readings.length === 0 ? Infinity : 0;
+        }
+        let tonelessMatch = Infinity;
+        for (const [at, reading] of readings.entries()) {
+            const parsed = this.#syllableOf(reading);
+            if (parsed === undefined) {
+                continue;
+            }
+            if (parsed.initial !== syllable.initial ||
+                parsed.final !== syllable.final) {
+                continue;
+            }
+            if (parsed.tone === syllable.tone) {
+                return at;
+            }
+            // A match on the syllable but not the tone still beats no match at all,
+            // since sources disagree about tone far more often than about spelling.
+            tonelessMatch = Math.min(tonelessMatch, readings.length + at);
+        }
+        return tonelessMatch;
+    }
+    /**
+     * The best-ranked candidate, or the character itself where there are none.
+     */
+    best(candidates, hans, syllable) {
+        let best = hans;
+        let bestRank = Infinity;
+        for (const candidate of candidates) {
+            const rank = this.#rank(candidate, syllable);
+            // Ties prefer a form other than the character itself, for the same reason
+            // mostFrequent does.
+            if (rank < bestRank || (rank === bestRank && best === hans)) {
+                best = candidate;
+                bestRank = rank;
+            }
+        }
+        return best;
+    }
+}
+//# sourceMappingURL=traditional-ranking.js.map

package/dist/dictionary/traditional-ranking.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"traditional-ranking.js","sourceRoot":"","sources":["../../src/dictionary/traditional-ranking.ts"],"names":[],"mappings":"AASA,OAAO,EAAE,YAAY,EAAiB,MAAM,yBAAyB,CAAC;AAEtE;;GAEG;AACH,MAAM,OAAO,cAAc;IAChB,eAAe,CAAsC;IACrD,OAAO,GAAG,IAAI,GAAG,EAAgC,CAAC;IAE3D,YAAY,cAAmD;QAC7D,IAAI,CAAC,eAAe,GAAG,cAAc,CAAC;IACxC,CAAC;IAED;;;;;OAKG;IACH,WAAW,CAAC,OAAe;QACzB,IAAI,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,CAAC;YAC9B,OAAO,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;QACnC,CAAC;QACD,MAAM,QAAQ,GAAG,YAAY,CAAC,OAAO,CAAC,CAAC;QACvC,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,OAAO,EAAE,QAAQ,CAAC,CAAC;QACpC,OAAO,QAAQ,CAAC;IAClB,CAAC;IAED;;;;;;OAMG;IACH,KAAK,CAAC,SAAiB,EAAE,QAA8B;QACrD,MAAM,QAAQ,GAAG,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC,SAAS,CAAC,EAAE,QAAQ,IAAI,EAAE,CAAC;QACrE,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;YAC3B,OAAO,QAAQ,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC;QAC9C,CAAC;QACD,IAAI,aAAa,GAAG,QAAQ,CAAC;QAC7B,KAAK,MAAM,CAAC,EAAE,EAAE,OAAO,CAAC,IAAI,QAAQ,CAAC,OAAO,EAAE,EAAE,CAAC;YAC/C,MAAM,MAAM,GAAG,IAAI,CAAC,WAAW,CAAC,OAAO,CAAC,CAAC;YACzC,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;gBACzB,SAAS;YACX,CAAC;YACD,IACE,MAAM,CAAC,OAAO,KAAK,QAAQ,CAAC,OAAO;gBACnC,MAAM,CAAC,KAAK,KAAK,QAAQ,CAAC,KAAK,EAC/B,CAAC;gBACD,SAAS;YACX,CAAC;YACD,IAAI,MAAM,CAAC,IAAI,KAAK,QAAQ,CAAC,IAAI,EAAE,CAAC;gBAClC,OAAO,EAAE,CAAC;YACZ,CAAC;YACD,wEAAwE;YACxE,wEAAwE;YACxE,aAAa,GAAG,IAAI,CAAC,GAAG,CAAC,aAAa,EAAE,QAAQ,CAAC,MAAM,GAAG,EAAE,CAAC,CAAC;QAChE,CAAC;QACD,OAAO,aAAa,CAAC;IACvB,CAAC;IAED;;OAEG;IACH,IAAI,CACF,UAA6B,EAC7B,IAAY,EACZ,QAA8B;QAE9B,IAAI,IAAI,GAAG,IAAI,CAAC;QAChB,IAAI,QAAQ,GAAG,QAAQ,CAAC;QACxB,KAAK,MAAM,SAAS,IAAI,UAAU,EAAE,CAAC;YACnC,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,SAAS,EAAE,QAAQ,CAAC,CAAC;YAC7C,0EAA0E;YAC1E,qBAAqB;YACrB,IAAI,IAAI,GAAG,QAAQ,IAAI,CAAC,IAAI,KAAK,QAAQ,IAAI,IAAI,KAAK,IAAI,CAAC,EAAE,CAAC;gBAC5D,IAAI,GAAG,SAAS,CAAC;gBACjB,QAAQ,GAAG,IAAI,CAAC;YAClB,CAAC;QACH,CAAC;QACD,OAAO,IAAI,CAAC;IACd,CAAC;CACF"}

package/dist/dictionary/traditional-table.d.ts

@@ -0,0 +1,51 @@
+import type { UnihanReadings, UnihanVariants } from "../sources/unihan.js";
+import type { Syllable } from "../syllable/syllable.js";
+import type { ReadCharacters } from "./reading.js";
+import { type ScriptPairing } from "./variant-counts.js";
+/**
+ * Derives the 繁體 form of a 简体 word, using the reading to disambiguate.
+ *
+ * 简体 → 繁體 is the ambiguous direction, which is what normally makes automatic
+ * conversion unreliable: 发 is either 發 (fā, to send) or 髮 (fà, hair). But the
+ * dictionary already knows the reading, and the reading settles it — 头发 is
+ * `tóu fà`, so its 发 is 髮 and the word is 頭髮.
+ *
+ * Two sources of evidence, in order:
+ *
+ * 1. **CC-CEDICT's own pairings.** It carries 124,754 entries with both scripts
+ *    written out, which is a large observed sample of how each character is
+ *    converted at each reading. This is preferred because it is evidence rather
+ *    than inference, and because it settles cases the reading cannot: 干扰 is
+ *    `gān rǎo` and stays 干擾, while 干燥 is also `gān` and becomes 乾燥.
+ * 2. **Unihan's `kTraditionalVariant`,** ranked by where the word's reading
+ *    appears in each candidate's reading list. 髮 reads only `fà`, whereas 發
+ *    reads `fā` first and `fà` only as a minor variant, so `fà` picks 髮.
+ *
+ * Characters no source knows a variant for are left alone, which is the right
+ * answer for the overwhelming majority: simplification changed a minority of
+ * characters.
+ */
+export declare class TraditionalTable {
+    #private;
+    /**
+     * Build the table from observed script pairings and Unihan's variant lists.
+     *
+     * Pairings should include the cases where the character is *unchanged*, since
+     * that is what stops a character with a rare traditional variant from always
+     * being converted to it.
+     */
+    static build(pairings: Iterable<ScriptPairing>, variants: UnihanVariants, unihanReadings: ReadonlyMap<string, UnihanReadings>): TraditionalTable;
+    /**
+     * Wrap prepared tables. Use {@link TraditionalTable.build}.
+     */
+    private constructor();
+    /**
+     * The 繁體 form of one character, given the syllable it is read as.
+     */
+    convertCharacter(hans: string, syllable: Syllable | undefined): string;
+    /**
+     * The 繁體 form of a whole word, read as the given syllables.
+     */
+    convert(aligned: readonly ReadCharacters[]): string;
+}
+//# sourceMappingURL=traditional-table.d.ts.map

package/dist/dictionary/traditional-table.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"traditional-table.d.ts","sourceRoot":"","sources":["../../src/dictionary/traditional-table.ts"],"names":[],"mappings":"AAOA,OAAO,KAAK,EAAE,cAAc,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAC3E,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AACxD,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,cAAc,CAAC;AAEnD,OAAO,EAIL,KAAK,aAAa,EAEnB,MAAM,qBAAqB,CAAC;AAE7B;;;;;;;;;;;;;;;;;;;;;;GAsBG;AACH,qBAAa,gBAAgB;;IAC3B;;;;;;OAMG;IACH,MAAM,CAAC,KAAK,CACV,QAAQ,EAAE,QAAQ,CAAC,aAAa,CAAC,EACjC,QAAQ,EAAE,cAAc,EACxB,cAAc,EAAE,WAAW,CAAC,MAAM,EAAE,cAAc,CAAC,GAClD,gBAAgB,CAalB;IAMD;;OAEG;IACH,OAAO,eAQN;IAED;;OAEG;IACH,gBAAgB,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,GAAG,SAAS,GAAG,MAAM,CAerE;IAED;;OAEG;IACH,OAAO,CAAC,OAAO,EAAE,SAAS,cAAc,EAAE,GAAG,MAAM,CAelD;CACF"}

package/dist/dictionary/traditional-table.js

@@ -0,0 +1,98 @@
+/**
+ * The 繁體 pairing table.
+ *
+ * Built from CC-CEDICT's own pairings and asked, character by character,
+ * which 繁體 form a 简体 one takes for a given reading.
+ */
+import { toCharacters } from "../script/characters.js";
+import { VariantRanking } from "./traditional-ranking.js";
+import { ANY_READING, mostFrequent, readingKey, } from "./variant-counts.js";
+/**
+ * Derives the 繁體 form of a 简体 word, using the reading to disambiguate.
+ *
+ * 简体 → 繁體 is the ambiguous direction, which is what normally makes automatic
+ * conversion unreliable: 发 is either 發 (fā, to send) or 髮 (fà, hair). But the
+ * dictionary already knows the reading, and the reading settles it — 头发 is
+ * `tóu fà`, so its 发 is 髮 and the word is 頭髮.
+ *
+ * Two sources of evidence, in order:
+ *
+ * 1. **CC-CEDICT's own pairings.** It carries 124,754 entries with both scripts
+ *    written out, which is a large observed sample of how each character is
+ *    converted at each reading. This is preferred because it is evidence rather
+ *    than inference, and because it settles cases the reading cannot: 干扰 is
+ *    `gān rǎo` and stays 干擾, while 干燥 is also `gān` and becomes 乾燥.
+ * 2. **Unihan's `kTraditionalVariant`,** ranked by where the word's reading
+ *    appears in each candidate's reading list. 髮 reads only `fà`, whereas 發
+ *    reads `fā` first and `fà` only as a minor variant, so `fà` picks 髮.
+ *
+ * Characters no source knows a variant for are left alone, which is the right
+ * answer for the overwhelming majority: simplification changed a minority of
+ * characters.
+ */
+export class TraditionalTable {
+    /**
+     * Build the table from observed script pairings and Unihan's variant lists.
+     *
+     * Pairings should include the cases where the character is *unchanged*, since
+     * that is what stops a character with a rare traditional variant from always
+     * being converted to it.
+     */
+    static build(pairings, variants, unihanReadings) {
+        const observed = new Map();
+        for (const { hans, hant, syllable } of pairings) {
+            const byReading = observed.get(hans) ?? new Map();
+            const keys = new Set([ANY_READING, readingKey(syllable)]);
+            for (const key of keys) {
+                const counts = byReading.get(key) ?? new Map();
+                counts.set(hant, (counts.get(hant) ?? 0) + 1);
+                byReading.set(key, counts);
+            }
+            observed.set(hans, byReading);
+        }
+        return new TraditionalTable(observed, variants, unihanReadings);
+    }
+    #observed;
+    #variants;
+    #ranking;
+    /**
+     * Wrap prepared tables. Use {@link TraditionalTable.build}.
+     */
+    constructor(observed, variants, unihanReadings) {
+        this.#observed = observed;
+        this.#variants = variants;
+        this.#ranking = new VariantRanking(unihanReadings);
+    }
+    /**
+     * The 繁體 form of one character, given the syllable it is read as.
+     */
+    convertCharacter(hans, syllable) {
+        const byReading = this.#observed.get(hans);
+        // Evidence at this exact reading first, then evidence at any reading.
+        const observed = mostFrequent(byReading?.get(readingKey(syllable)), hans) ??
+            mostFrequent(byReading?.get(ANY_READING), hans);
+        if (observed !== undefined) {
+            return observed;
+        }
+        return this.#ranking.best(this.#variants.traditional.get(hans) ?? [], hans, syllable);
+    }
+    /**
+     * The 繁體 form of a whole word, read as the given syllables.
+     */
+    convert(aligned) {
+        let converted = "";
+        for (const read of aligned) {
+            const group = toCharacters(read.characters);
+            for (const character of group) {
+                // A syllable covering two characters — 儿化 — is not evidence about the
+                // second of them, so only the first is disambiguated by it.
+                const syllable = group.length === 1 || character === group[0]
+                    ? read.syllable
+                    : undefined;
+                converted += this.convertCharacter(character, syllable);
+            }
+        }
+        return converted;
+    }
+}
+//# sourceMappingURL=traditional-table.js.map

package/dist/dictionary/traditional-table.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"traditional-table.js","sourceRoot":"","sources":["../../src/dictionary/traditional-table.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAIvD,OAAO,EAAE,cAAc,EAAE,MAAM,0BAA0B,CAAC;AAC1D,OAAO,EACL,WAAW,EACX,YAAY,EACZ,UAAU,GAGX,MAAM,qBAAqB,CAAC;AAE7B;;;;;;;;;;;;;;;;;;;;;;GAsBG;AACH,MAAM,OAAO,gBAAgB;IAC3B;;;;;;OAMG;IACH,MAAM,CAAC,KAAK,CACV,QAAiC,EACjC,QAAwB,EACxB,cAAmD;QAEnD,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAsC,CAAC;QAC/D,KAAK,MAAM,EAAE,IAAI,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,QAAQ,EAAE,CAAC;YAChD,MAAM,SAAS,GAAG,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,IAAI,GAAG,EAAyB,CAAC;YACzE,MAAM,IAAI,GAAG,IAAI,GAAG,CAAC,CAAC,WAAW,EAAE,UAAU,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC;YAC1D,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;gBACvB,MAAM,MAAM,GAAG,SAAS,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,IAAI,GAAG,EAAkB,CAAC;gBAC/D,MAAM,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;gBAC9C,SAAS,CAAC,GAAG,CAAC,GAAG,EAAE,MAAM,CAAC,CAAC;YAC7B,CAAC;YACD,QAAQ,CAAC,GAAG,CAAC,IAAI,EAAE,SAAS,CAAC,CAAC;QAChC,CAAC;QACD,OAAO,IAAI,gBAAgB,CAAC,QAAQ,EAAE,QAAQ,EAAE,cAAc,CAAC,CAAC;IAClE,CAAC;IAEQ,SAAS,CAA0D;IACnE,SAAS,CAAiB;IAC1B,QAAQ,CAAiB;IAElC;;OAEG;IACH,YACE,QAAiE,EACjE,QAAwB,EACxB,cAAmD;QAEnD,IAAI,CAAC,SAAS,GAAG,QAAQ,CAAC;QAC1B,IAAI,CAAC,SAAS,GAAG,QAAQ,CAAC;QAC1B,IAAI,CAAC,QAAQ,GAAG,IAAI,cAAc,CAAC,cAAc,CAAC,CAAC;IACrD,CAAC;IAED;;OAEG;IACH,gBAAgB,CAAC,IAAY,EAAE,QAA8B;QAC3D,MAAM,SAAS,GAAG,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QAC3C,sEAAsE;QACtE,MAAM,QAAQ,GACZ,YAAY,CAAC,SAAS,EAAE,GAAG,CAAC,UAAU,CAAC,QAAQ,CAAC,CAAC,EAAE,IAAI,CAAC;YACxD,YAAY,CAAC,SAAS,EAAE,GAAG,CAAC,WAAW,CAAC,EAAE,IAAI,CAAC,CAAC;QAClD,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;YAC3B,OAAO,QAAQ,CAAC;QAClB,CAAC;QAED,OAAO,IAAI,CAAC,QAAQ,CAAC,IAAI,CACvB,IAAI,CAAC,SAAS,CAAC,WAAW,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,EAC1C,IAAI,EACJ,QAAQ,CACT,CAAC;IACJ,CAAC;IAED;;OAEG;IACH,OAAO,CAAC,OAAkC;QACxC,IAAI,SAAS,GAAG,EAAE,CAAC;QACnB,KAAK,MAAM,IAAI,IAAI,OAAO,EAAE,CAAC;YAC3B,MAAM,KAAK,GAAG,YAAY,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;YAC5C,KAAK,MAAM,SAAS,IAAI,KAAK,EAAE,CAAC;gBAC9B,sEAAsE;gBACtE,4DAA4D;gBAC5D,MAAM,QAAQ,GACZ,KAAK,CAAC,MAAM,KAAK,CAAC,IAAI,SAAS,KAAK,KAAK,CAAC,CAAC,CAAC;oBAC1C,CAAC,CAAC,IAAI,CAAC,QAAQ;oBACf,CAAC,CAAC,SAAS,CAAC;gBAChB,SAAS,IAAI,IAAI,CAAC,gBAAgB,CAAC,SAAS,EAAE,QAAQ,CAAC,CAAC;YAC1D,CAAC;QACH,CAAC;QACD,OAAO,SAAS,CAAC;IACnB,CAAC;CACF"}

package/dist/dictionary/traditional.d.ts

@@ -1,60 +1,7 @@
-import type { UnihanReadings, UnihanVariants } from "../sources/unihan.js";
-import { type Syllable } from "../syllable/syllable.js";
 import type { ReadCharacters } from "./reading.js";
-/**
- * One 简体 → 繁體 pairing observed in a source, at one reading.
- */
-export interface ScriptPairing {
-    readonly hans: string;
-    readonly hant: string;
-    readonly syllable: Syllable | undefined;
-}
-/**
- * Derives the 繁體 form of a 简体 word, using the reading to disambiguate.
- *
- * 简体 → 繁體 is the ambiguous direction, which is what normally makes automatic
- * conversion unreliable: 发 is either 發 (fā, to send) or 髮 (fà, hair). But the
- * dictionary already knows the reading, and the reading settles it — 头发 is
- * `tóu fà`, so its 发 is 髮 and the word is 頭髮.
- *
- * Two sources of evidence, in order:
- *
- * 1. **CC-CEDICT's own pairings.** It carries 124,754 entries with both scripts
- *    written out, which is a large observed sample of how each character is
- *    converted at each reading. This is preferred because it is evidence rather
- *    than inference, and because it settles cases the reading cannot: 干扰 is
- *    `gān rǎo` and stays 干擾, while 干燥 is also `gān` and becomes 乾燥.
- * 2. **Unihan's `kTraditionalVariant`,** ranked by where the word's reading
- *    appears in each candidate's reading list. 髮 reads only `fà`, whereas 發
- *    reads `fā` first and `fà` only as a minor variant, so `fà` picks 髮.
- *
- * Characters no source knows a variant for are left alone, which is the right
- * answer for the overwhelming majority: simplification changed a minority of
- * characters.
- */
-export declare class TraditionalTable {
-    #private;
-    /**
-     * Build the table from observed script pairings and Unihan's variant lists.
-     *
-     * Pairings should include the cases where the character is *unchanged*, since
-     * that is what stops a character with a rare traditional variant from always
-     * being converted to it.
-     */
-    static build(pairings: Iterable<ScriptPairing>, variants: UnihanVariants, unihanReadings: ReadonlyMap<string, UnihanReadings>): TraditionalTable;
-    /**
-     * Wrap prepared tables. Use {@link TraditionalTable.build}.
-     */
-    private constructor();
-    /**
-     * The 繁體 form of one character, given the syllable it is read as.
-     */
-    convertCharacter(hans: string, syllable: Syllable | undefined): string;
-    /**
-     * The 繁體 form of a whole word, read as the given syllables.
-     */
-    convert(aligned: readonly ReadCharacters[]): string;
-}
+import type { ScriptPairing } from "./variant-counts.js";
+export type { ScriptPairing } from "./variant-counts.js";
+export { TraditionalTable } from "./traditional-table.js";
 /**
  * Pair a CC-CEDICT entry's two scripts up character by character.
  *

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

@@ -1 +1 @@
-{"version":3,"file":"traditional.d.ts","sourceRoot":"","sources":["../../src/dictionary/traditional.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,cAAc,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAC3E,OAAO,EAAgB,KAAK,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AACtE,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,cAAc,CAAC;AAuBnD;;GAEG;AACH,MAAM,WAAW,aAAa;IAC5B,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,QAAQ,CAAC,QAAQ,EAAE,QAAQ,GAAG,SAAS,CAAC;CACzC;AA8BD;;;;;;;;;;;;;;;;;;;;;;GAsBG;AACH,qBAAa,gBAAgB;;IAC3B;;;;;;OAMG;IACH,MAAM,CAAC,KAAK,CACV,QAAQ,EAAE,QAAQ,CAAC,aAAa,CAAC,EACjC,QAAQ,EAAE,cAAc,EACxB,cAAc,EAAE,WAAW,CAAC,MAAM,EAAE,cAAc,CAAC,GAClD,gBAAgB,CAalB;IAOD;;OAEG;IACH,OAAO,eAQN;IAmDD;;OAEG;IACH,gBAAgB,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,GAAG,SAAS,GAAG,MAAM,CA2BrE;IAED;;OAEG;IACH,OAAO,CAAC,OAAO,EAAE,SAAS,cAAc,EAAE,GAAG,MAAM,CAelD;CACF;AAED;;;;;;;GAOG;AACH,wBAAgB,WAAW,CACzB,IAAI,EAAE,MAAM,EACZ,IAAI,EAAE,MAAM,EACZ,OAAO,EAAE,SAAS,cAAc,EAAE,GAAG,SAAS,GAC7C,SAAS,aAAa,EAAE,CA8C1B"}
+{"version":3,"file":"traditional.d.ts","sourceRoot":"","sources":["../../src/dictionary/traditional.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,cAAc,CAAC;AACnD,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,qBAAqB,CAAC;AAEzD,YAAY,EAAE,aAAa,EAAE,MAAM,qBAAqB,CAAC;AACzD,OAAO,EAAE,gBAAgB,EAAE,MAAM,wBAAwB,CAAC;AAE1D;;;;;;;GAOG;AACH,wBAAgB,WAAW,CACzB,IAAI,EAAE,MAAM,EACZ,IAAI,EAAE,MAAM,EACZ,OAAO,EAAE,SAAS,cAAc,EAAE,GAAG,SAAS,GAC7C,SAAS,aAAa,EAAE,CA8C1B"}

package/dist/dictionary/traditional.js

@@ -1,191 +1,8 @@
-import { toCharacters } from "../script/characters.js";
-import { readSyllable } from "../syllable/syllable.js";
-/**
- * The key standing for "whatever the reading", used when the reading is unknown
- * or when no reading-specific evidence exists.
- */
-const ANY_READING = "*";
-/**
- * A reading key that ignores 儿化, which never bears on which variant is meant.
- */
-function readingKey(syllable) {
-    if (syllable === undefined) {
-        return ANY_READING;
-    }
-    return `${syllable.initial}|${syllable.final}|${String(syllable.tone)}`;
-}
-/**
- * Pick the most frequent variant, breaking ties by preferring the 繁體 form.
- *
- * The tie-break matters for a character that is its own traditional form in one
- * sense and has a distinct one in another. Where the counts cannot separate
- * them, the traditional form is the better guess in a traditional-form table,
- * and a wrong guess costs only whether a 繁體 text matches this key — never the
- * reading it is given.
- */
-function mostFrequent(counts, hans) {
-    let best;
-    let bestCount = 0;
-    const tally = counts ?? new Map();
-    for (const [variant, count] of tally) {
-        const isBetter = count > bestCount ||
-            (count === bestCount && best === hans && variant !== hans);
-        if (isBetter) {
-            best = variant;
-            bestCount = count;
-        }
-    }
-    return best;
-}
 /**
- * Derives the 繁體 form of a 简体 word, using the reading to disambiguate.
- *
- * 简体 → 繁體 is the ambiguous direction, which is what normally makes automatic
- * conversion unreliable: 发 is either 發 (fā, to send) or 髮 (fà, hair). But the
- * dictionary already knows the reading, and the reading settles it — 头发 is
- * `tóu fà`, so its 发 is 髮 and the word is 頭髮.
- *
- * Two sources of evidence, in order:
- *
- * 1. **CC-CEDICT's own pairings.** It carries 124,754 entries with both scripts
- *    written out, which is a large observed sample of how each character is
- *    converted at each reading. This is preferred because it is evidence rather
- *    than inference, and because it settles cases the reading cannot: 干扰 is
- *    `gān rǎo` and stays 干擾, while 干燥 is also `gān` and becomes 乾燥.
- * 2. **Unihan's `kTraditionalVariant`,** ranked by where the word's reading
- *    appears in each candidate's reading list. 髮 reads only `fà`, whereas 發
- *    reads `fā` first and `fà` only as a minor variant, so `fà` picks 髮.
- *
- * Characters no source knows a variant for are left alone, which is the right
- * answer for the overwhelming majority: simplification changed a minority of
- * characters.
+ * Pairing a CC-CEDICT entry's two scripts up, character by character.
  */
-export class TraditionalTable {
-    /**
-     * Build the table from observed script pairings and Unihan's variant lists.
-     *
-     * Pairings should include the cases where the character is *unchanged*, since
-     * that is what stops a character with a rare traditional variant from always
-     * being converted to it.
-     */
-    static build(pairings, variants, unihanReadings) {
-        const observed = new Map();
-        for (const { hans, hant, syllable } of pairings) {
-            const byReading = observed.get(hans) ?? new Map();
-            const keys = new Set([ANY_READING, readingKey(syllable)]);
-            for (const key of keys) {
-                const counts = byReading.get(key) ?? new Map();
-                counts.set(hant, (counts.get(hant) ?? 0) + 1);
-                byReading.set(key, counts);
-            }
-            observed.set(hans, byReading);
-        }
-        return new TraditionalTable(observed, variants, unihanReadings);
-    }
-    #observed;
-    #variants;
-    #unihanReadings;
-    #parsed = new Map();
-    /**
-     * Wrap prepared tables. Use {@link TraditionalTable.build}.
-     */
-    constructor(observed, variants, unihanReadings) {
-        this.#observed = observed;
-        this.#variants = variants;
-        this.#unihanReadings = unihanReadings;
-    }
-    /**
-     * Parse a Unihan reading string, remembering the result.
-     *
-     * Worth caching: the fallback path reads the same few thousand strings once
-     * per candidate per word.
-     */
-    #syllableOf(reading) {
-        if (this.#parsed.has(reading)) {
-            return this.#parsed.get(reading);
-        }
-        const syllable = readSyllable(reading);
-        this.#parsed.set(reading, syllable);
-        return syllable;
-    }
-    /**
-     * How well a candidate character's readings match the one we want.
-     *
-     * Lower is better. A candidate whose most likely reading is the one we want
-     * beats one that merely lists it among its rarer readings, which is exactly
-     * what separates 髮 (`fà` alone) from 發 (`fā` first, `fà` third).
-     */
-    #rank(candidate, syllable) {
-        const readings = this.#unihanReadings.get(candidate)?.readings ?? [];
-        if (syllable === undefined) {
-            return readings.length === 0 ? Infinity : 0;
-        }
-        let tonelessMatch = Infinity;
-        for (const [at, reading] of readings.entries()) {
-            const parsed = this.#syllableOf(reading);
-            if (parsed === undefined) {
-                continue;
-            }
-            if (parsed.initial !== syllable.initial ||
-                parsed.final !== syllable.final) {
-                continue;
-            }
-            if (parsed.tone === syllable.tone) {
-                return at;
-            }
-            // A match on the syllable but not the tone still beats no match at all,
-            // since sources disagree about tone far more often than about spelling.
-            tonelessMatch = Math.min(tonelessMatch, readings.length + at);
-        }
-        return tonelessMatch;
-    }
-    /**
-     * The 繁體 form of one character, given the syllable it is read as.
-     */
-    convertCharacter(hans, syllable) {
-        const byReading = this.#observed.get(hans);
-        // Evidence at this exact reading first, then evidence at any reading.
-        const observed = mostFrequent(byReading?.get(readingKey(syllable)), hans) ??
-            mostFrequent(byReading?.get(ANY_READING), hans);
-        if (observed !== undefined) {
-            return observed;
-        }
-        const candidates = this.#variants.traditional.get(hans) ?? [];
-        if (candidates.length === 0) {
-            return hans;
-        }
-        let best = hans;
-        let bestRank = Infinity;
-        for (const candidate of candidates) {
-            const rank = this.#rank(candidate, syllable);
-            // Ties prefer a form other than the character itself, for the same reason
-            // mostFrequent does.
-            if (rank < bestRank || (rank === bestRank && best === hans)) {
-                best = candidate;
-                bestRank = rank;
-            }
-        }
-        return best;
-    }
-    /**
-     * The 繁體 form of a whole word, read as the given syllables.
-     */
-    convert(aligned) {
-        let converted = "";
-        for (const read of aligned) {
-            const group = toCharacters(read.characters);
-            for (const character of group) {
-                // A syllable covering two characters — 儿化 — is not evidence about the
-                // second of them, so only the first is disambiguated by it.
-                const syllable = group.length === 1 || character === group[0]
-                    ? read.syllable
-                    : undefined;
-                converted += this.convertCharacter(character, syllable);
-            }
-        }
-        return converted;
-    }
-}
+import { toCharacters } from "../script/characters.js";
+export { TraditionalTable } from "./traditional-table.js";
 /**
  * Pair a CC-CEDICT entry's two scripts up character by character.
  *

package/dist/dictionary/traditional.js.map

@@ -1 +1 @@
-{"version":3,"file":"traditional.js","sourceRoot":"","sources":["../../src/dictionary/traditional.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAEvD,OAAO,EAAE,YAAY,EAAiB,MAAM,yBAAyB,CAAC;AAQtE;;;GAGG;AACH,MAAM,WAAW,GAAG,GAAG,CAAC;AAExB;;GAEG;AACH,SAAS,UAAU,CAAC,QAA8B;IAChD,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;QAC3B,OAAO,WAAW,CAAC;IACrB,CAAC;IACD,OAAO,GAAG,QAAQ,CAAC,OAAO,IAAI,QAAQ,CAAC,KAAK,IAAI,MAAM,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC;AAC1E,CAAC;AAWD;;;;;;;;GAQG;AACH,SAAS,YAAY,CACnB,MAAiC,EACjC,IAAY;IAEZ,IAAI,IAAwB,CAAC;IAC7B,IAAI,SAAS,GAAG,CAAC,CAAC;IAClB,MAAM,KAAK,GAAG,MAAM,IAAI,IAAI,GAAG,EAAkB,CAAC;IAClD,KAAK,MAAM,CAAC,OAAO,EAAE,KAAK,CAAC,IAAI,KAAK,EAAE,CAAC;QACrC,MAAM,QAAQ,GACZ,KAAK,GAAG,SAAS;YACjB,CAAC,KAAK,KAAK,SAAS,IAAI,IAAI,KAAK,IAAI,IAAI,OAAO,KAAK,IAAI,CAAC,CAAC;QAC7D,IAAI,QAAQ,EAAE,CAAC;YACb,IAAI,GAAG,OAAO,CAAC;YACf,SAAS,GAAG,KAAK,CAAC;QACpB,CAAC;IACH,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;GAsBG;AACH,MAAM,OAAO,gBAAgB;IAC3B;;;;;;OAMG;IACH,MAAM,CAAC,KAAK,CACV,QAAiC,EACjC,QAAwB,EACxB,cAAmD;QAEnD,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAsC,CAAC;QAC/D,KAAK,MAAM,EAAE,IAAI,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,QAAQ,EAAE,CAAC;YAChD,MAAM,SAAS,GAAG,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,IAAI,GAAG,EAAyB,CAAC;YACzE,MAAM,IAAI,GAAG,IAAI,GAAG,CAAC,CAAC,WAAW,EAAE,UAAU,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC;YAC1D,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;gBACvB,MAAM,MAAM,GAAG,SAAS,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,IAAI,GAAG,EAAkB,CAAC;gBAC/D,MAAM,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;gBAC9C,SAAS,CAAC,GAAG,CAAC,GAAG,EAAE,MAAM,CAAC,CAAC;YAC7B,CAAC;YACD,QAAQ,CAAC,GAAG,CAAC,IAAI,EAAE,SAAS,CAAC,CAAC;QAChC,CAAC;QACD,OAAO,IAAI,gBAAgB,CAAC,QAAQ,EAAE,QAAQ,EAAE,cAAc,CAAC,CAAC;IAClE,CAAC;IAEQ,SAAS,CAA0D;IACnE,SAAS,CAAiB;IAC1B,eAAe,CAAsC;IACrD,OAAO,GAAG,IAAI,GAAG,EAAgC,CAAC;IAE3D;;OAEG;IACH,YACE,QAAiE,EACjE,QAAwB,EACxB,cAAmD;QAEnD,IAAI,CAAC,SAAS,GAAG,QAAQ,CAAC;QAC1B,IAAI,CAAC,SAAS,GAAG,QAAQ,CAAC;QAC1B,IAAI,CAAC,eAAe,GAAG,cAAc,CAAC;IACxC,CAAC;IAED;;;;;OAKG;IACH,WAAW,CAAC,OAAe;QACzB,IAAI,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,CAAC;YAC9B,OAAO,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;QACnC,CAAC;QACD,MAAM,QAAQ,GAAG,YAAY,CAAC,OAAO,CAAC,CAAC;QACvC,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,OAAO,EAAE,QAAQ,CAAC,CAAC;QACpC,OAAO,QAAQ,CAAC;IAClB,CAAC;IAED;;;;;;OAMG;IACH,KAAK,CAAC,SAAiB,EAAE,QAA8B;QACrD,MAAM,QAAQ,GAAG,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC,SAAS,CAAC,EAAE,QAAQ,IAAI,EAAE,CAAC;QACrE,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;YAC3B,OAAO,QAAQ,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC;QAC9C,CAAC;QACD,IAAI,aAAa,GAAG,QAAQ,CAAC;QAC7B,KAAK,MAAM,CAAC,EAAE,EAAE,OAAO,CAAC,IAAI,QAAQ,CAAC,OAAO,EAAE,EAAE,CAAC;YAC/C,MAAM,MAAM,GAAG,IAAI,CAAC,WAAW,CAAC,OAAO,CAAC,CAAC;YACzC,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;gBACzB,SAAS;YACX,CAAC;YACD,IACE,MAAM,CAAC,OAAO,KAAK,QAAQ,CAAC,OAAO;gBACnC,MAAM,CAAC,KAAK,KAAK,QAAQ,CAAC,KAAK,EAC/B,CAAC;gBACD,SAAS;YACX,CAAC;YACD,IAAI,MAAM,CAAC,IAAI,KAAK,QAAQ,CAAC,IAAI,EAAE,CAAC;gBAClC,OAAO,EAAE,CAAC;YACZ,CAAC;YACD,wEAAwE;YACxE,wEAAwE;YACxE,aAAa,GAAG,IAAI,CAAC,GAAG,CAAC,aAAa,EAAE,QAAQ,CAAC,MAAM,GAAG,EAAE,CAAC,CAAC;QAChE,CAAC;QACD,OAAO,aAAa,CAAC;IACvB,CAAC;IAED;;OAEG;IACH,gBAAgB,CAAC,IAAY,EAAE,QAA8B;QAC3D,MAAM,SAAS,GAAG,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QAC3C,sEAAsE;QACtE,MAAM,QAAQ,GACZ,YAAY,CAAC,SAAS,EAAE,GAAG,CAAC,UAAU,CAAC,QAAQ,CAAC,CAAC,EAAE,IAAI,CAAC;YACxD,YAAY,CAAC,SAAS,EAAE,GAAG,CAAC,WAAW,CAAC,EAAE,IAAI,CAAC,CAAC;QAClD,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;YAC3B,OAAO,QAAQ,CAAC;QAClB,CAAC;QAED,MAAM,UAAU,GAAG,IAAI,CAAC,SAAS,CAAC,WAAW,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;QAC9D,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAC5B,OAAO,IAAI,CAAC;QACd,CAAC;QAED,IAAI,IAAI,GAAG,IAAI,CAAC;QAChB,IAAI,QAAQ,GAAG,QAAQ,CAAC;QACxB,KAAK,MAAM,SAAS,IAAI,UAAU,EAAE,CAAC;YACnC,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,SAAS,EAAE,QAAQ,CAAC,CAAC;YAC7C,0EAA0E;YAC1E,qBAAqB;YACrB,IAAI,IAAI,GAAG,QAAQ,IAAI,CAAC,IAAI,KAAK,QAAQ,IAAI,IAAI,KAAK,IAAI,CAAC,EAAE,CAAC;gBAC5D,IAAI,GAAG,SAAS,CAAC;gBACjB,QAAQ,GAAG,IAAI,CAAC;YAClB,CAAC;QACH,CAAC;QACD,OAAO,IAAI,CAAC;IACd,CAAC;IAED;;OAEG;IACH,OAAO,CAAC,OAAkC;QACxC,IAAI,SAAS,GAAG,EAAE,CAAC;QACnB,KAAK,MAAM,IAAI,IAAI,OAAO,EAAE,CAAC;YAC3B,MAAM,KAAK,GAAG,YAAY,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;YAC5C,KAAK,MAAM,SAAS,IAAI,KAAK,EAAE,CAAC;gBAC9B,sEAAsE;gBACtE,4DAA4D;gBAC5D,MAAM,QAAQ,GACZ,KAAK,CAAC,MAAM,KAAK,CAAC,IAAI,SAAS,KAAK,KAAK,CAAC,CAAC,CAAC;oBAC1C,CAAC,CAAC,IAAI,CAAC,QAAQ;oBACf,CAAC,CAAC,SAAS,CAAC;gBAChB,SAAS,IAAI,IAAI,CAAC,gBAAgB,CAAC,SAAS,EAAE,QAAQ,CAAC,CAAC;YAC1D,CAAC;QACH,CAAC;QACD,OAAO,SAAS,CAAC;IACnB,CAAC;CACF;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,WAAW,CACzB,IAAY,EACZ,IAAY,EACZ,OAA8C;IAE9C,MAAM,cAAc,GAAG,YAAY,CAAC,IAAI,CAAC,CAAC;IAC1C,MAAM,cAAc,GAAG,YAAY,CAAC,IAAI,CAAC,CAAC;IAC1C,IACE,cAAc,CAAC,MAAM,KAAK,cAAc,CAAC,MAAM;QAC/C,cAAc,CAAC,MAAM,KAAK,CAAC,EAC3B,CAAC;QACD,OAAO,EAAE,CAAC;IACZ,CAAC;IAED,MAAM,QAAQ,GAAoB,EAAE,CAAC;IACrC,IAAI,EAAE,GAAG,CAAC,CAAC;IACX,MAAM,MAAM,GAAG,OAAO,IAAI,EAAE,CAAC;IAC7B,KAAK,MAAM,IAAI,IAAI,MAAM,EAAE,CAAC;QAC1B,MAAM,KAAK,GAAG,YAAY,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;QAC5C,MAAM,QAAQ,GAAG,KAAK,CAAC,OAAO,EAAE,CAAC;QACjC,KAAK,MAAM,CAAC,MAAM,EAAE,SAAS,CAAC,IAAI,QAAQ,EAAE,CAAC;YAC3C,MAAM,aAAa,GAAG,cAAc,CAAC,EAAE,GAAG,MAAM,CAAC,CAAC;YAClD,IAAI,aAAa,KAAK,SAAS,EAAE,CAAC;gBAChC,OAAO,EAAE,CAAC;YACZ,CAAC;YACD,QAAQ,CAAC,IAAI,CAAC;gBACZ,IAAI,EAAE,SAAS;gBACf,IAAI,EAAE,aAAa;gBACnB,uEAAuE;gBACvE,QAAQ,EAAE,KAAK,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,SAAS;aACzD,CAAC,CAAC;QACL,CAAC;QACD,EAAE,IAAI,KAAK,CAAC,MAAM,CAAC;IACrB,CAAC;IAED,2EAA2E;IAC3E,gEAAgE;IAChE,IAAI,OAAO,KAAK,SAAS,EAAE,CAAC;QAC1B,MAAM,MAAM,GAAG,cAAc,CAAC,OAAO,EAAE,CAAC;QACxC,KAAK,MAAM,CAAC,MAAM,EAAE,SAAS,CAAC,IAAI,MAAM,EAAE,CAAC;YACzC,QAAQ,CAAC,IAAI,CAAC;gBACZ,IAAI,EAAE,SAAS;gBACf,uEAAuE;gBACvE,IAAI,EAAE,cAAc,CAAC,MAAM,CAAC,IAAI,SAAS;gBACzC,QAAQ,EAAE,SAAS;aACpB,CAAC,CAAC;QACL,CAAC;IACH,CAAC;IAED,OAAO,QAAQ,CAAC;AAClB,CAAC"}
+{"version":3,"file":"traditional.js","sourceRoot":"","sources":["../../src/dictionary/traditional.ts"],"names":[],"mappings":"AAAA;;GAEG;AACH,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAKvD,OAAO,EAAE,gBAAgB,EAAE,MAAM,wBAAwB,CAAC;AAE1D;;;;;;;GAOG;AACH,MAAM,UAAU,WAAW,CACzB,IAAY,EACZ,IAAY,EACZ,OAA8C;IAE9C,MAAM,cAAc,GAAG,YAAY,CAAC,IAAI,CAAC,CAAC;IAC1C,MAAM,cAAc,GAAG,YAAY,CAAC,IAAI,CAAC,CAAC;IAC1C,IACE,cAAc,CAAC,MAAM,KAAK,cAAc,CAAC,MAAM;QAC/C,cAAc,CAAC,MAAM,KAAK,CAAC,EAC3B,CAAC;QACD,OAAO,EAAE,CAAC;IACZ,CAAC;IAED,MAAM,QAAQ,GAAoB,EAAE,CAAC;IACrC,IAAI,EAAE,GAAG,CAAC,CAAC;IACX,MAAM,MAAM,GAAG,OAAO,IAAI,EAAE,CAAC;IAC7B,KAAK,MAAM,IAAI,IAAI,MAAM,EAAE,CAAC;QAC1B,MAAM,KAAK,GAAG,YAAY,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;QAC5C,MAAM,QAAQ,GAAG,KAAK,CAAC,OAAO,EAAE,CAAC;QACjC,KAAK,MAAM,CAAC,MAAM,EAAE,SAAS,CAAC,IAAI,QAAQ,EAAE,CAAC;YAC3C,MAAM,aAAa,GAAG,cAAc,CAAC,EAAE,GAAG,MAAM,CAAC,CAAC;YAClD,IAAI,aAAa,KAAK,SAAS,EAAE,CAAC;gBAChC,OAAO,EAAE,CAAC;YACZ,CAAC;YACD,QAAQ,CAAC,IAAI,CAAC;gBACZ,IAAI,EAAE,SAAS;gBACf,IAAI,EAAE,aAAa;gBACnB,uEAAuE;gBACvE,QAAQ,EAAE,KAAK,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,SAAS;aACzD,CAAC,CAAC;QACL,CAAC;QACD,EAAE,IAAI,KAAK,CAAC,MAAM,CAAC;IACrB,CAAC;IAED,2EAA2E;IAC3E,gEAAgE;IAChE,IAAI,OAAO,KAAK,SAAS,EAAE,CAAC;QAC1B,MAAM,MAAM,GAAG,cAAc,CAAC,OAAO,EAAE,CAAC;QACxC,KAAK,MAAM,CAAC,MAAM,EAAE,SAAS,CAAC,IAAI,MAAM,EAAE,CAAC;YACzC,QAAQ,CAAC,IAAI,CAAC;gBACZ,IAAI,EAAE,SAAS;gBACf,uEAAuE;gBACvE,IAAI,EAAE,cAAc,CAAC,MAAM,CAAC,IAAI,SAAS;gBACzC,QAAQ,EAAE,SAAS;aACpB,CAAC,CAAC;QACL,CAAC;IACH,CAAC;IAED,OAAO,QAAQ,CAAC;AAClB,CAAC"}

package/dist/dictionary/variant-counts.d.ts

@@ -0,0 +1,39 @@
+/**
+ * Counting which 繁體 form a 简体 character is paired with.
+ *
+ * The tally the pairing table is built from, and the rule for reading a
+ * winner out of it.
+ */
+import type { Syllable } from "../syllable/syllable.js";
+/**
+ * How often one 简体 character was written as a given 繁體 one.
+ */
+export type VariantCounts = Map<string, number>;
+/**
+ * The key standing for "whatever the reading", used when the reading is unknown
+ * or when no reading-specific evidence exists.
+ */
+export declare const ANY_READING = "*";
+/**
+ * A reading key that ignores 儿化, which never bears on which variant is meant.
+ */
+export declare function readingKey(syllable: Syllable | undefined): string;
+/**
+ * One 简体 → 繁體 pairing observed in a source, at one reading.
+ */
+export interface ScriptPairing {
+    readonly hans: string;
+    readonly hant: string;
+    readonly syllable: Syllable | undefined;
+}
+/**
+ * Pick the most frequent variant, breaking ties by preferring the 繁體 form.
+ *
+ * The tie-break matters for a character that is its own traditional form in one
+ * sense and has a distinct one in another. Where the counts cannot separate
+ * them, the traditional form is the better guess in a traditional-form table,
+ * and a wrong guess costs only whether a 繁體 text matches this key — never the
+ * reading it is given.
+ */
+export declare function mostFrequent(counts: VariantCounts | undefined, hans: string): string | undefined;
+//# sourceMappingURL=variant-counts.d.ts.map

package/dist/dictionary/variant-counts.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"variant-counts.d.ts","sourceRoot":"","sources":["../../src/dictionary/variant-counts.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAExD;;GAEG;AACH,MAAM,MAAM,aAAa,GAAG,GAAG,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;AAEhD;;;GAGG;AACH,eAAO,MAAM,WAAW,MAAM,CAAC;AAE/B;;GAEG;AACH,wBAAgB,UAAU,CAAC,QAAQ,EAAE,QAAQ,GAAG,SAAS,GAAG,MAAM,CAKjE;AAED;;GAEG;AACH,MAAM,WAAW,aAAa;IAC5B,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,QAAQ,CAAC,QAAQ,EAAE,QAAQ,GAAG,SAAS,CAAC;CACzC;AAED;;;;;;;;GAQG;AACH,wBAAgB,YAAY,CAC1B,MAAM,EAAE,aAAa,GAAG,SAAS,EACjC,IAAI,EAAE,MAAM,GACX,MAAM,GAAG,SAAS,CAcpB"}

package/dist/dictionary/variant-counts.js

@@ -0,0 +1,38 @@
+/**
+ * The key standing for "whatever the reading", used when the reading is unknown
+ * or when no reading-specific evidence exists.
+ */
+export const ANY_READING = "*";
+/**
+ * A reading key that ignores 儿化, which never bears on which variant is meant.
+ */
+export function readingKey(syllable) {
+    if (syllable === undefined) {
+        return ANY_READING;
+    }
+    return `${syllable.initial}|${syllable.final}|${String(syllable.tone)}`;
+}
+/**
+ * Pick the most frequent variant, breaking ties by preferring the 繁體 form.
+ *
+ * The tie-break matters for a character that is its own traditional form in one
+ * sense and has a distinct one in another. Where the counts cannot separate
+ * them, the traditional form is the better guess in a traditional-form table,
+ * and a wrong guess costs only whether a 繁體 text matches this key — never the
+ * reading it is given.
+ */
+export function mostFrequent(counts, hans) {
+    let best;
+    let bestCount = 0;
+    const tally = counts ?? new Map();
+    for (const [variant, count] of tally) {
+        const isBetter = count > bestCount ||
+            (count === bestCount && best === hans && variant !== hans);
+        if (isBetter) {
+            best = variant;
+            bestCount = count;
+        }
+    }
+    return best;
+}
+//# sourceMappingURL=variant-counts.js.map

package/dist/dictionary/variant-counts.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"variant-counts.js","sourceRoot":"","sources":["../../src/dictionary/variant-counts.ts"],"names":[],"mappings":"AAaA;;;GAGG;AACH,MAAM,CAAC,MAAM,WAAW,GAAG,GAAG,CAAC;AAE/B;;GAEG;AACH,MAAM,UAAU,UAAU,CAAC,QAA8B;IACvD,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;QAC3B,OAAO,WAAW,CAAC;IACrB,CAAC;IACD,OAAO,GAAG,QAAQ,CAAC,OAAO,IAAI,QAAQ,CAAC,KAAK,IAAI,MAAM,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC;AAC1E,CAAC;AAWD;;;;;;;;GAQG;AACH,MAAM,UAAU,YAAY,CAC1B,MAAiC,EACjC,IAAY;IAEZ,IAAI,IAAwB,CAAC;IAC7B,IAAI,SAAS,GAAG,CAAC,CAAC;IAClB,MAAM,KAAK,GAAG,MAAM,IAAI,IAAI,GAAG,EAAkB,CAAC;IAClD,KAAK,MAAM,CAAC,OAAO,EAAE,KAAK,CAAC,IAAI,KAAK,EAAE,CAAC;QACrC,MAAM,QAAQ,GACZ,KAAK,GAAG,SAAS;YACjB,CAAC,KAAK,KAAK,SAAS,IAAI,IAAI,KAAK,IAAI,IAAI,OAAO,KAAK,IAAI,CAAC,CAAC;QAC7D,IAAI,QAAQ,EAAE,CAAC;YACb,IAAI,GAAG,OAAO,CAAC;YACf,SAAS,GAAG,KAAK,CAAC;QACpB,CAAC;IACH,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC"}

package/dist/dictionary/word-reading.d.ts

@@ -0,0 +1,26 @@
+/**
+ * Settling which reading a word takes.
+ *
+ * The phrase corpus leads because it has three times CC-CEDICT's coverage;
+ * CC-CEDICT corrects it on the neutral tone, which is what the two of them
+ * actually disagree about; 儿化 is repaired before anything compares two
+ * readings; and the override table has the last word.
+ */
+import type { CedictEntry } from "../sources/cedict.js";
+import type { Syllable } from "../syllable/syllable.js";
+import type { ReadCharacters } from "./reading.js";
+/**
+ * A word's settled reading, and what settling it cost the counts.
+ */
+export interface SettledReading {
+    /** Undefined where no source gave a usable reading. */
+    readonly reading: readonly Syllable[] | undefined;
+    readonly aligned: readonly ReadCharacters[] | undefined;
+    readonly neutralToneCorrections: number;
+    readonly erhuaRepairs: number;
+}
+/**
+ * Settle it.
+ */
+export declare function settleReading(word: string, defaults: ReadonlyMap<string, readonly Syllable[]>, cedictEntries: readonly CedictEntry[], cedictReadings: readonly (readonly Syllable[])[], phraseAligned: readonly ReadCharacters[] | undefined): SettledReading;
+//# sourceMappingURL=word-reading.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"word-reading.d.ts","sourceRoot":"","sources":["../../src/dictionary/word-reading.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AACH,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAC;AACxD,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAYxD,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,cAAc,CAAC;AAEnD;;GAEG;AACH,MAAM,WAAW,cAAc;IAC7B,uDAAuD;IACvD,QAAQ,CAAC,OAAO,EAAE,SAAS,QAAQ,EAAE,GAAG,SAAS,CAAC;IAClD,QAAQ,CAAC,OAAO,EAAE,SAAS,cAAc,EAAE,GAAG,SAAS,CAAC;IACxD,QAAQ,CAAC,sBAAsB,EAAE,MAAM,CAAC;IACxC,QAAQ,CAAC,YAAY,EAAE,MAAM,CAAC;CAC/B;AAED;;GAEG;AACH,wBAAgB,aAAa,CAC3B,IAAI,EAAE,MAAM,EACZ,QAAQ,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,QAAQ,EAAE,CAAC,EAClD,aAAa,EAAE,SAAS,WAAW,EAAE,EACrC,cAAc,EAAE,SAAS,CAAC,SAAS,QAAQ,EAAE,CAAC,EAAE,EAChD,aAAa,EAAE,SAAS,cAAc,EAAE,GAAG,SAAS,GACnD,cAAc,CAkFhB"}