@kensio/pinyinjs 1.7.3 → 1.7.5

package/dist/decode/script.d.ts

@@ -1,77 +1,7 @@
 import type { Dictionary } from "../dictionary/dictionary.js";
-import { type ScriptTables } from "../script/conversion.js";
-import { type Script } from "../script/script.js";
-/**
- * A script and, for 繁體, a regional orthography, as a BCP-47 tag.
- *
- * BCP-47 rather than a shape of our own because these subtags already mean
- * exactly this, and callers recognise them. `zh-Hant` alone takes
- * {@link DEFAULT_REGION}, since there is no region-free 繁體 to fall back on —
- * see SCRIPTS-AND-LOCALES.md.
- */
-export declare const SCRIPT_TARGETS: readonly ["zh-Hans", "zh-Hant", "zh-Hant-TW", "zh-Hant-HK"];
-/**
- * One of the orthographies {@link toScript} writes.
- */
-export type ScriptTarget = (typeof SCRIPT_TARGETS)[number];
-/**
- * How a script conversion should be carried out.
- */
-export interface ScriptOptions {
-    /** Which orthography to write. Defaults to `zh-Hans`. */
-    readonly to?: ScriptTarget;
-    /**
-     * The script the text is written in. Detected when not given.
-     *
-     * Worth setting when the text is short enough that detection has nothing to
-     * go on — a run of characters both scripts share settles nothing, and the
-     * conversion then assumes the text needs converting.
-     */
-    readonly from?: Script;
-}
-/**
- * What settled one character's conversion, strongest first.
- *
- * - `locked` — the character has one form and there was nothing to decide.
- *   True of the great majority: simplification changed a minority of
- *   characters, and most of those are one-to-one.
- * - `word` — a word some source wrote in both scripts settled it. The strongest
- *   real evidence, because it was written rather than inferred.
- * - `reading` — the character had rival forms and the syllable it was decoded
- *   as picked between them. This is the evidence an orthographic converter does
- *   not have.
- * - `default` — rival forms existed and nothing separated them, so the
- *   commonest was taken. The only one of the four that is a guess.
- */
-export declare const SCRIPT_EVIDENCE: readonly ["locked", "word", "reading", "default"];
-/**
- * What settled one character's conversion.
- */
-export type ScriptEvidence = (typeof SCRIPT_EVIDENCE)[number];
-/**
- * Why one character came out as it did.
- *
- * The same kind of claim `ReadingConfidence` makes about a syllable, and for
- * the same reason: a conversion that cannot say which characters it guessed at
- * is asking to be trusted further than it deserves. Nothing else in this space
- * reports it.
- */
-export interface ScriptChoice {
-    /** The character as it was written. */
-    readonly from: string;
-    /** The character as it was converted. */
-    readonly to: string;
-    readonly evidence: ScriptEvidence;
-    /** The forms this character could also have taken. */
-    readonly alternatives: readonly string[];
-}
-/**
- * A converted text, with an account of every character it was unsure about.
- */
-export interface ScriptConversion {
-    readonly text: string;
-    readonly choices: readonly ScriptChoice[];
-}
+import type { ScriptTables } from "../script/conversion.js";
+import type { ScriptChoice, ScriptConversion, ScriptOptions } from "./script-types.js";
+export { SCRIPT_EVIDENCE, type ScriptChoice, type ScriptConversion, type ScriptEvidence, type ScriptOptions, SCRIPT_TARGETS, type ScriptTarget, } from "./script-types.js";
 /**
  * Convert a text between the scripts, reporting what it was unsure about.
  *

package/dist/decode/script.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"script.d.ts","sourceRoot":"","sources":["../../src/decode/script.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,6BAA6B,CAAC;AAE9D,OAAO,EAML,KAAK,YAAY,EAClB,MAAM,yBAAyB,CAAC;AAQjC,OAAO,EAAgB,KAAK,MAAM,EAAE,MAAM,qBAAqB,CAAC;AAMhE;;;;;;;GAOG;AACH,eAAO,MAAM,cAAc,YACzB,SAAS,EACT,SAAS,EACT,YAAY,EACZ,YAAY,CACJ,CAAC;AAEX;;GAEG;AACH,MAAM,MAAM,YAAY,GAAG,CAAC,OAAO,cAAc,CAAC,CAAC,MAAM,CAAC,CAAC;AAE3D;;GAEG;AACH,MAAM,WAAW,aAAa;IAC5B,yDAAyD;IACzD,QAAQ,CAAC,EAAE,CAAC,EAAE,YAAY,CAAC;IAC3B;;;;;;OAMG;IACH,QAAQ,CAAC,IAAI,CAAC,EAAE,MAAM,CAAC;CACxB;AAED;;;;;;;;;;;;;GAaG;AACH,eAAO,MAAM,eAAe,YAC1B,QAAQ,EACR,MAAM,EACN,SAAS,EACT,SAAS,CACD,CAAC;AAEX;;GAEG;AACH,MAAM,MAAM,cAAc,GAAG,CAAC,OAAO,eAAe,CAAC,CAAC,MAAM,CAAC,CAAC;AAE9D;;;;;;;GAOG;AACH,MAAM,WAAW,YAAY;IAC3B,uCAAuC;IACvC,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,yCAAyC;IACzC,QAAQ,CAAC,EAAE,EAAE,MAAM,CAAC;IACpB,QAAQ,CAAC,QAAQ,EAAE,cAAc,CAAC;IAClC,sDAAsD;IACtD,QAAQ,CAAC,YAAY,EAAE,SAAS,MAAM,EAAE,CAAC;CAC1C;AAED;;GAEG;AACH,MAAM,WAAW,gBAAgB;IAC/B,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,QAAQ,CAAC,OAAO,EAAE,SAAS,YAAY,EAAE,CAAC;CAC3C;AAyGD;;;;;;;;;;;;;;;;;GAiBG;AACH,wBAAgB,cAAc,CAC5B,UAAU,EAAE,UAAU,EACtB,MAAM,EAAE,YAAY,EACpB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,aAAkB,GAC1B,gBAAgB,CA0ClB;AAeD;;GAEG;AACH,wBAAgB,QAAQ,CACtB,UAAU,EAAE,UAAU,EACtB,MAAM,EAAE,YAAY,EACpB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,aAAkB,GAC1B,MAAM,CAER;AAED;;;;;;;;GAQG;AACH,wBAAgB,iBAAiB,CAAC,MAAM,EAAE,YAAY,GAAG,OAAO,CAE/D"}
+{"version":3,"file":"script.d.ts","sourceRoot":"","sources":["../../src/decode/script.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,6BAA6B,CAAC;AAE9D,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAQ5D,OAAO,KAAK,EACV,YAAY,EACZ,gBAAgB,EAChB,aAAa,EACd,MAAM,mBAAmB,CAAC;AAE3B,OAAO,EACL,eAAe,EACf,KAAK,YAAY,EACjB,KAAK,gBAAgB,EACrB,KAAK,cAAc,EACnB,KAAK,aAAa,EAClB,cAAc,EACd,KAAK,YAAY,GAClB,MAAM,mBAAmB,CAAC;AAC3B;;;;;;;;;;;;;;;;;GAiBG;AACH,wBAAgB,cAAc,CAC5B,UAAU,EAAE,UAAU,EACtB,MAAM,EAAE,YAAY,EACpB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,aAAkB,GAC1B,gBAAgB,CA0ClB;AAeD;;GAEG;AACH,wBAAgB,QAAQ,CACtB,UAAU,EAAE,UAAU,EACtB,MAAM,EAAE,YAAY,EACpB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,aAAkB,GAC1B,MAAM,CAER;AAED;;;;;;;;GAQG;AACH,wBAAgB,iBAAiB,CAAC,MAAM,EAAE,YAAY,GAAG,OAAO,CAE/D"}

package/dist/decode/script.js

@@ -1,117 +1,11 @@
 import { toCharacters } from "../script/characters.js";
-import { conversionKey, convertCharacter, formsOf, isAmbiguousCharacter, } from "../script/conversion.js";
-import { DEFAULT_REGION, isReadingSensitive, toCanonicalGlyphs, toRegionalGlyph, } from "../script/glyphs.js";
+import { toCanonicalGlyphs } from "../script/glyphs.js";
 import { detectScript } from "../script/script.js";
 import { decodeRun } from "./decode.js";
 import { READING_RULES } from "./reading-rules.js";
 import { splitRuns } from "./runs.js";
-/**
- * A script and, for 繁體, a regional orthography, as a BCP-47 tag.
- *
- * BCP-47 rather than a shape of our own because these subtags already mean
- * exactly this, and callers recognise them. `zh-Hant` alone takes
- * {@link DEFAULT_REGION}, since there is no region-free 繁體 to fall back on —
- * see SCRIPTS-AND-LOCALES.md.
- */
-export const SCRIPT_TARGETS = [
-    "zh-Hans",
-    "zh-Hant",
-    "zh-Hant-TW",
-    "zh-Hant-HK",
-];
-/**
- * What settled one character's conversion, strongest first.
- *
- * - `locked` — the character has one form and there was nothing to decide.
- *   True of the great majority: simplification changed a minority of
- *   characters, and most of those are one-to-one.
- * - `word` — a word some source wrote in both scripts settled it. The strongest
- *   real evidence, because it was written rather than inferred.
- * - `reading` — the character had rival forms and the syllable it was decoded
- *   as picked between them. This is the evidence an orthographic converter does
- *   not have.
- * - `default` — rival forms existed and nothing separated them, so the
- *   commonest was taken. The only one of the four that is a guess.
- */
-export const SCRIPT_EVIDENCE = [
-    "locked",
-    "word",
-    "reading",
-    "default",
-];
-/**
- * The script and region a target names.
- */
-function targetOf(target) {
-    if (target === "zh-Hans") {
-        return { script: "Hans", region: DEFAULT_REGION };
-    }
-    return {
-        script: "Hant",
-        region: target === "zh-Hant-HK" ? "HK" : DEFAULT_REGION,
-    };
-}
-/**
- * Convert one word, preferring what a source wrote to what its characters imply.
- */
-function convertWord(table, words, word, reading) {
-    const characters = toCharacters(word);
-    const attested = words.get(word);
-    // A word only counts as attested when it lines up character for character;
-    // anything else cannot be reported per character, which is what the choices
-    // are. Those are rare enough to fall through to the characters.
-    const attestedCharacters = attested === undefined ? undefined : toCharacters(attested);
-    const isAttested = attestedCharacters?.length === characters.length;
-    // A reading covering a different number of characters than the word has
-    // cannot be lined up with them — 儿化 writes two characters as one syllable.
-    const isAligned = reading.length === characters.length;
-    return characters.map((character, at) => {
-        const syllable = isAligned ? reading[at] : undefined;
-        const conversion = table.get(character);
-        const atReading = conversion?.byReading?.get(conversionKey(syllable));
-        const to = isAttested
-            ? /* c8 ignore next -- the lengths were just checked */
-                (attestedCharacters[at] ?? character)
-            : convertCharacter(table, character, syllable);
-        return {
-            from: character,
-            to,
-            evidence: evidenceFor(isAttested, atReading !== undefined, isAmbiguousCharacter(table, character)),
-            alternatives: formsOf(conversion).filter((form) => form !== character && form !== to),
-        };
-    });
-}
-/**
- * Rank what settled a character, strongest evidence first.
- */
-function evidenceFor(isAttested, isAtReading, isAmbiguous) {
-    if (!isAmbiguous) {
-        return "locked";
-    }
-    if (isAttested) {
-        return "word";
-    }
-    return isAtReading ? "reading" : "default";
-}
-/**
- * Apply the regional 繁體 forms, which never change a reading.
- */
-function applyRegion(choices, reading, region, isAligned) {
-    if (region === DEFAULT_REGION) {
-        return choices;
-    }
-    return choices.map((choice, at) => {
-        const syllable = isAligned ? reading[at] : undefined;
-        // A regional form that needs the reading and has none is a guess, whatever
-        // settled the script conversion before it: 著 is 着 or 著 in Hong Kong.
-        const isGuess = isReadingSensitive(choice.to) && syllable === undefined;
-        return {
-            ...choice,
-            to: toRegionalGlyph(choice.to, region, syllable),
-            evidence: isGuess ? "default" : choice.evidence,
-        };
-    });
-}
+import { applyRegion, convertWord, targetOf } from "./script-words.js";
+export { SCRIPT_EVIDENCE, SCRIPT_TARGETS, } from "./script-types.js";
 /**
  * Convert a text between the scripts, reporting what it was unsure about.
  *

package/dist/decode/script.js.map

@@ -1 +1 @@
-{"version":3,"file":"script.js","sourceRoot":"","sources":["../../src/decode/script.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AACvD,OAAO,EAEL,aAAa,EACb,gBAAgB,EAChB,OAAO,EACP,oBAAoB,GAErB,MAAM,yBAAyB,CAAC;AACjC,OAAO,EACL,cAAc,EACd,kBAAkB,EAElB,iBAAiB,EACjB,eAAe,GAChB,MAAM,qBAAqB,CAAC;AAC7B,OAAO,EAAE,YAAY,EAAe,MAAM,qBAAqB,CAAC;AAEhE,OAAO,EAAE,SAAS,EAAE,MAAM,aAAa,CAAC;AACxC,OAAO,EAAE,aAAa,EAAE,MAAM,oBAAoB,CAAC;AACnD,OAAO,EAAE,SAAS,EAAE,MAAM,WAAW,CAAC;AAEtC;;;;;;;GAOG;AACH,MAAM,CAAC,MAAM,cAAc,GAAG;IAC5B,SAAS;IACT,SAAS;IACT,YAAY;IACZ,YAAY;CACJ,CAAC;AAuBX;;;;;;;;;;;;;GAaG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG;IAC7B,QAAQ;IACR,MAAM;IACN,SAAS;IACT,SAAS;CACD,CAAC;AAiCX;;GAEG;AACH,SAAS,QAAQ,CAAC,MAAoB;IAIpC,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;QACzB,OAAO,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,cAAc,EAAE,CAAC;IACpD,CAAC;IACD,OAAO;QACL,MAAM,EAAE,MAAM;QACd,MAAM,EAAE,MAAM,KAAK,YAAY,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,cAAc;KACxD,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,SAAS,WAAW,CAClB,KAA+C,EAC/C,KAAkC,EAClC,IAAY,EACZ,OAA4B;IAE5B,MAAM,UAAU,GAAG,YAAY,CAAC,IAAI,CAAC,CAAC;IACtC,MAAM,QAAQ,GAAG,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IACjC,2EAA2E;IAC3E,4EAA4E;IAC5E,gEAAgE;IAChE,MAAM,kBAAkB,GACtB,QAAQ,KAAK,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,YAAY,CAAC,QAAQ,CAAC,CAAC;IAC9D,MAAM,UAAU,GAAG,kBAAkB,EAAE,MAAM,KAAK,UAAU,CAAC,MAAM,CAAC;IAEpE,wEAAwE;IACxE,2EAA2E;IAC3E,MAAM,SAAS,GAAG,OAAO,CAAC,MAAM,KAAK,UAAU,CAAC,MAAM,CAAC;IAEvD,OAAO,UAAU,CAAC,GAAG,CAAC,CAAC,SAAS,EAAE,EAAE,EAAE,EAAE;QACtC,MAAM,QAAQ,GAAG,SAAS,CAAC,CAAC,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;QACrD,MAAM,UAAU,GAAG,KAAK,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QACxC,MAAM,SAAS,GAAG,UAAU,EAAE,SAAS,EAAE,GAAG,CAAC,aAAa,CAAC,QAAQ,CAAC,CAAC,CAAC;QACtE,MAAM,EAAE,GAAG,UAAU;YACnB,CAAC,CAAC,qDAAqD;gBACrD,CAAC,kBAAkB,CAAC,EAAE,CAAC,IAAI,SAAS,CAAC;YACvC,CAAC,CAAC,gBAAgB,CAAC,KAAK,EAAE,SAAS,EAAE,QAAQ,CAAC,CAAC;QACjD,OAAO;YACL,IAAI,EAAE,SAAS;YACf,EAAE;YACF,QAAQ,EAAE,WAAW,CACnB,UAAU,EACV,SAAS,KAAK,SAAS,EACvB,oBAAoB,CAAC,KAAK,EAAE,SAAS,CAAC,CACvC;YACD,YAAY,EAAE,OAAO,CAAC,UAAU,CAAC,CAAC,MAAM,CACtC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,KAAK,SAAS,IAAI,IAAI,KAAK,EAAE,CAC5C;SACF,CAAC;IACJ,CAAC,CAAC,CAAC;AACL,CAAC;AAED;;GAEG;AACH,SAAS,WAAW,CAClB,UAAmB,EACnB,WAAoB,EACpB,WAAoB;IAEpB,IAAI,CAAC,WAAW,EAAE,CAAC;QACjB,OAAO,QAAQ,CAAC;IAClB,CAAC;IACD,IAAI,UAAU,EAAE,CAAC;QACf,OAAO,MAAM,CAAC;IAChB,CAAC;IACD,OAAO,WAAW,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC;AAC7C,CAAC;AAED;;GAEG;AACH,SAAS,WAAW,CAClB,OAAgC,EAChC,OAA4B,EAC5B,MAAc,EACd,SAAkB;IAElB,IAAI,MAAM,KAAK,cAAc,EAAE,CAAC;QAC9B,OAAO,OAAO,CAAC;IACjB,CAAC;IACD,OAAO,OAAO,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,EAAE,EAAE,EAAE;QAChC,MAAM,QAAQ,GAAG,SAAS,CAAC,CAAC,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;QACrD,2EAA2E;QAC3E,qEAAqE;QACrE,MAAM,OAAO,GAAG,kBAAkB,CAAC,MAAM,CAAC,EAAE,CAAC,IAAI,QAAQ,KAAK,SAAS,CAAC;QACxE,OAAO;YACL,GAAG,MAAM;YACT,EAAE,EAAE,eAAe,CAAC,MAAM,CAAC,EAAE,EAAE,MAAM,EAAE,QAAQ,CAAC;YAChD,QAAQ,EAAE,OAAO,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,MAAM,CAAC,QAAQ;SAChD,CAAC;IACJ,CAAC,CAAC,CAAC;AACL,CAAC;AAED;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAM,UAAU,cAAc,CAC5B,UAAsB,EACtB,MAAoB,EACpB,IAAY,EACZ,OAAO,GAAkB,EAAE;IAE3B,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,GAAG,QAAQ,CAAC,OAAO,CAAC,EAAE,IAAI,SAAS,CAAC,CAAC;IAC7D,MAAM,KAAK,GAAG,MAAM,KAAK,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,YAAY,CAAC,CAAC,CAAC,MAAM,CAAC,aAAa,CAAC;IAC7E,MAAM,KAAK,GACT,MAAM,KAAK,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,eAAe,CAAC,CAAC,CAAC,MAAM,CAAC,gBAAgB,CAAC;IAEvE,2EAA2E;IAC3E,4EAA4E;IAC5E,2EAA2E;IAC3E,gEAAgE;IAChE,2DAA2D;IAC3D,MAAM,IAAI,GACR,OAAO,CAAC,IAAI,IAAI,YAAY,CAAC,IAAI,EAAE,MAAM,CAAC,QAAQ,EAAE,MAAM,CAAC,QAAQ,CAAC,CAAC;IACvE,MAAM,YAAY,GAAG,IAAI,KAAK,MAAM,CAAC;IAErC,IAAI,SAAS,GAAG,EAAE,CAAC;IACnB,MAAM,OAAO,GAAmB,EAAE,CAAC;IAEnC,KAAK,MAAM,GAAG,IAAI,SAAS,CAAC,IAAI,CAAC,EAAE,CAAC;QAClC,IAAI,CAAC,GAAG,CAAC,KAAK,EAAE,CAAC;YACf,SAAS,IAAI,GAAG,CAAC,IAAI,CAAC;YACtB,SAAS;QACX,CAAC;QACD,sEAAsE;QACtE,kEAAkE;QAClE,KAAK,MAAM,IAAI,IAAI,SAAS,CAAC,UAAU,EAAE,GAAG,CAAC,IAAI,EAAE,aAAa,CAAC,EAAE,CAAC;YAClE,MAAM,SAAS,GAAG,iBAAiB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YAC/C,MAAM,SAAS,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,KAAK,YAAY,CAAC,SAAS,CAAC,CAAC,MAAM,CAAC;YACzE,MAAM,MAAM,GAAG,WAAW,CACxB,YAAY;gBACV,CAAC,CAAC,QAAQ,CAAC,SAAS,CAAC;gBACrB,CAAC,CAAC,WAAW,CAAC,KAAK,EAAE,KAAK,EAAE,SAAS,EAAE,IAAI,CAAC,OAAO,CAAC,EACtD,IAAI,CAAC,OAAO,EACZ,MAAM,EACN,SAAS,CACV,CAAC;YACF,OAAO,CAAC,IAAI,CAAC,GAAG,MAAM,CAAC,CAAC;YACxB,SAAS,IAAI,MAAM,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QAC1D,CAAC;IACH,CAAC;IAED,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,OAAO,EAAE,CAAC;AACtC,CAAC;AAED;;;GAGG;AACH,SAAS,QAAQ,CAAC,IAAY;IAC5B,OAAO,YAAY,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC;QAC5C,IAAI,EAAE,SAAS;QACf,EAAE,EAAE,SAAS;QACb,QAAQ,EAAE,QAAiB;QAC3B,YAAY,EAAE,EAAE;KACjB,CAAC,CAAC,CAAC;AACN,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,QAAQ,CACtB,UAAsB,EACtB,MAAoB,EACpB,IAAY,EACZ,OAAO,GAAkB,EAAE;IAE3B,OAAO,cAAc,CAAC,UAAU,EAAE,MAAM,EAAE,IAAI,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC;AAChE,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,iBAAiB,CAAC,MAAoB;IACpD,OAAO,MAAM,CAAC,QAAQ,KAAK,SAAS,CAAC;AACvC,CAAC"}
+{"version":3,"file":"script.js","sourceRoot":"","sources":["../../src/decode/script.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAEvD,OAAO,EAAE,iBAAiB,EAAE,MAAM,qBAAqB,CAAC;AACxD,OAAO,EAAE,YAAY,EAAE,MAAM,qBAAqB,CAAC;AACnD,OAAO,EAAE,SAAS,EAAE,MAAM,aAAa,CAAC;AACxC,OAAO,EAAE,aAAa,EAAE,MAAM,oBAAoB,CAAC;AACnD,OAAO,EAAE,SAAS,EAAE,MAAM,WAAW,CAAC;AAEtC,OAAO,EAAE,WAAW,EAAE,WAAW,EAAE,QAAQ,EAAE,MAAM,mBAAmB,CAAC;AAOvE,OAAO,EACL,eAAe,EAKf,cAAc,GAEf,MAAM,mBAAmB,CAAC;AAC3B;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAM,UAAU,cAAc,CAC5B,UAAsB,EACtB,MAAoB,EACpB,IAAY,EACZ,OAAO,GAAkB,EAAE;IAE3B,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,GAAG,QAAQ,CAAC,OAAO,CAAC,EAAE,IAAI,SAAS,CAAC,CAAC;IAC7D,MAAM,KAAK,GAAG,MAAM,KAAK,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,YAAY,CAAC,CAAC,CAAC,MAAM,CAAC,aAAa,CAAC;IAC7E,MAAM,KAAK,GACT,MAAM,KAAK,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,eAAe,CAAC,CAAC,CAAC,MAAM,CAAC,gBAAgB,CAAC;IAEvE,2EAA2E;IAC3E,4EAA4E;IAC5E,2EAA2E;IAC3E,gEAAgE;IAChE,2DAA2D;IAC3D,MAAM,IAAI,GACR,OAAO,CAAC,IAAI,IAAI,YAAY,CAAC,IAAI,EAAE,MAAM,CAAC,QAAQ,EAAE,MAAM,CAAC,QAAQ,CAAC,CAAC;IACvE,MAAM,YAAY,GAAG,IAAI,KAAK,MAAM,CAAC;IAErC,IAAI,SAAS,GAAG,EAAE,CAAC;IACnB,MAAM,OAAO,GAAmB,EAAE,CAAC;IAEnC,KAAK,MAAM,GAAG,IAAI,SAAS,CAAC,IAAI,CAAC,EAAE,CAAC;QAClC,IAAI,CAAC,GAAG,CAAC,KAAK,EAAE,CAAC;YACf,SAAS,IAAI,GAAG,CAAC,IAAI,CAAC;YACtB,SAAS;QACX,CAAC;QACD,sEAAsE;QACtE,kEAAkE;QAClE,KAAK,MAAM,IAAI,IAAI,SAAS,CAAC,UAAU,EAAE,GAAG,CAAC,IAAI,EAAE,aAAa,CAAC,EAAE,CAAC;YAClE,MAAM,SAAS,GAAG,iBAAiB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YAC/C,MAAM,SAAS,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,KAAK,YAAY,CAAC,SAAS,CAAC,CAAC,MAAM,CAAC;YACzE,MAAM,MAAM,GAAG,WAAW,CACxB,YAAY;gBACV,CAAC,CAAC,QAAQ,CAAC,SAAS,CAAC;gBACrB,CAAC,CAAC,WAAW,CAAC,KAAK,EAAE,KAAK,EAAE,SAAS,EAAE,IAAI,CAAC,OAAO,CAAC,EACtD,IAAI,CAAC,OAAO,EACZ,MAAM,EACN,SAAS,CACV,CAAC;YACF,OAAO,CAAC,IAAI,CAAC,GAAG,MAAM,CAAC,CAAC;YACxB,SAAS,IAAI,MAAM,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QAC1D,CAAC;IACH,CAAC;IAED,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,OAAO,EAAE,CAAC;AACtC,CAAC;AAED;;;GAGG;AACH,SAAS,QAAQ,CAAC,IAAY;IAC5B,OAAO,YAAY,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC;QAC5C,IAAI,EAAE,SAAS;QACf,EAAE,EAAE,SAAS;QACb,QAAQ,EAAE,QAAiB;QAC3B,YAAY,EAAE,EAAE;KACjB,CAAC,CAAC,CAAC;AACN,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,QAAQ,CACtB,UAAsB,EACtB,MAAoB,EACpB,IAAY,EACZ,OAAO,GAAkB,EAAE;IAE3B,OAAO,cAAc,CAAC,UAAU,EAAE,MAAM,EAAE,IAAI,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC;AAChE,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,iBAAiB,CAAC,MAAoB;IACpD,OAAO,MAAM,CAAC,QAAQ,KAAK,SAAS,CAAC;AACvC,CAAC"}

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

@@ -0,0 +1,68 @@
+import type { Syllable } from "../syllable/syllable.js";
+import type { EdgeContext, EdgeRule } from "./rules.js";
+/**
+ * Whether a syllable is `tán`, the verbal reading of 弹.
+ */
+export declare function isTan(syllable: Syllable | undefined): boolean;
+/**
+ * The character in both scripts, since 彈 is not the shape 弹 is.
+ *
+ * 得, 教 and 儿 are the same character in either script and need say it once.
+ * This one is not, and neither is the 长 above, which is silent on 長 as a
+ * result. {@link ASPECT} carries 過 and 著 for the same reason this carries 彈.
+ */
+export declare const PLUCKED: Set<string>;
+/**
+ * Whether the character at a position is 弹, in either script.
+ */
+export declare function isPluckedAt(context: EdgeContext, at: number): boolean;
+/**
+ * What a 弹 that is playing governs: the instrument, the piece or its composer.
+ *
+ * 弹吉他, 弹竖琴, 弹肖邦, 弹一曲, 给我们弹一手. The name tags are in because
+ * jieba calls 吉他 `ns` and 管风琴, 班卓琴 and 古琴 `nr` — instruments read as
+ * places and people — and `m` because a piece is as often counted as named.
+ */
+export declare const PLAYED_TAGS: Set<string>;
+/**
+ * 弹 read as `tán` where it is playing rather than a projectile.
+ *
+ * The character is stored `dàn` with `tán` as an alternate, and the sources are
+ * agreed about it: Unihan's `kHanyuPinlu` counts `dàn(313)` against `tán(50)`
+ * and `kMandarin` names `dàn`. That is a fact about a corpus in which nearly
+ * every 弹 is ammunition — 子弹, 炮弹, 导弹, 原子弹, 手榴弹 — and every one of
+ * those is a word that carries its own reading. What the default is left
+ * deciding is the 弹 that stands as a word of its own, and in running text that
+ * one is the verb: 他会弹一点儿古筝 came out `dàn`, and 弹吉他, 弹竖琴, 我在弹肖邦
+ * and 他钢琴弹得很好 with it.
+ *
+ * See {@link isPlayingAt} for what decides it. Forcing the single-character
+ * edge is not enough on its own, for the reason {@link TEACHING_JIAO} gives: a
+ * reading spanning two characters carries its own 弹 into the position, and
+ * 我的爱好是开车和弹吉他 read `dàn` off 和弹, a pair the dictionary holds with
+ * no part of speech and at the cost of a reading nobody counted. A *tagged*
+ * word ending in 弹 is left alone, which is every one that matters — 子弹, 炸弹,
+ * 导弹, 反弹, 手榴弹. Nothing has to be done about the other side, since the
+ * nominal compounds starting with 弹 are all listed: 弹匣, 弹坑, 弹壳, 弹片,
+ * 弹药, 弹道 and 弹头 reach their reading through the word, so a bare 弹 with a
+ * noun after it is not the front of one that got away.
+ *
+ * The 88,866 lines of Tatoeba and zh.wikipedia the other rules were sized
+ * against are the wrong corpus for this one on their own: 38 弹 decode as a
+ * word of their own in them, and 31 are somebody playing something. So CPP's
+ * 20,147 sentences are measured with them as plain text — the benchmark is
+ * drawn from zh.wikipedia's military articles, which is precisely where the
+ * shapes this rule can break live. Over the 109,013 lines together, 60 弹
+ * decode as a word of their own, every one read `dàn`, and 34 of the 60 are
+ * wrong.
+ *
+ * This moves 30 of them, of which 29 are right: the whole of 弹吉他, 弹一曲肖邦
+ * and 钢琴弹得很好, and none of 南部弹, 毛瑟弹, 掷弹兵, 供弹爪, 着弹点, 弹洞 or
+ * 底排弹时. The one it breaks is 拆弹专家, where a compound has an object's
+ * shape on both sides. Five it leaves alone: 开始弹, 四手联弹 and 弹起三次, which
+ * have no object to see, 用那种指法弹不会觉得费力, and 反手持法去弹班卓琴, where
+ * the decode reaches 班 rather than 班卓琴. On CPP's 40 labelled 弹 nothing
+ * moves at all, every one of them having been right through a word.
+ */
+export declare const PLAYING_TAN: EdgeRule;
+//# sourceMappingURL=tan-rule.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"tan-rule.d.ts","sourceRoot":"","sources":["../../src/decode/tan-rule.ts"],"names":[],"mappings":"AAOA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAQxD,OAAO,KAAK,EAAE,WAAW,EAAE,QAAQ,EAAE,MAAM,YAAY,CAAC;AAExD;;GAEG;AACH,wBAAgB,KAAK,CAAC,QAAQ,EAAE,QAAQ,GAAG,SAAS,GAAG,OAAO,CAI7D;AAED;;;;;;GAMG;AACH,eAAO,MAAM,OAAO,aAAsB,CAAC;AAE3C;;GAEG;AACH,wBAAgB,WAAW,CAAC,OAAO,EAAE,WAAW,EAAE,EAAE,EAAE,MAAM,GAAG,OAAO,CAErE;AAED;;;;;;GAMG;AACH,eAAO,MAAM,WAAW,aAAoD,CAAC;AAwC7E;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAuCG;AACH,eAAO,MAAM,WAAW,EAAE,QAiBzB,CAAC"}

package/dist/decode/tan-rule.js

@@ -0,0 +1,131 @@
+/**
+ * Whether a 弹 is plucked or is a bullet.
+ *
+ * The two readings are a verb and a noun, so what settles it is what the word
+ * beside it is doing rather than what the character is.
+ */
+import { toCharacters } from "../script/characters.js";
+import { ASPECT, PARTICLE_TAG, tagOf, wordEndingAt, wordStartingAt, } from "./edge-context.js";
+/**
+ * Whether a syllable is `tán`, the verbal reading of 弹.
+ */
+export function isTan(syllable) {
+    return (syllable?.initial === "t" && syllable.final === "an" && syllable.tone === 2);
+}
+/**
+ * The character in both scripts, since 彈 is not the shape 弹 is.
+ *
+ * 得, 教 and 儿 are the same character in either script and need say it once.
+ * This one is not, and neither is the 长 above, which is silent on 長 as a
+ * result. {@link ASPECT} carries 過 and 著 for the same reason this carries 彈.
+ */
+export const PLUCKED = new Set(["弹", "彈"]);
+/**
+ * Whether the character at a position is 弹, in either script.
+ */
+export function isPluckedAt(context, at) {
+    return PLUCKED.has(context.characters[at] ?? "");
+}
+/**
+ * What a 弹 that is playing governs: the instrument, the piece or its composer.
+ *
+ * 弹吉他, 弹竖琴, 弹肖邦, 弹一曲, 给我们弹一手. The name tags are in because
+ * jieba calls 吉他 `ns` and 管风琴, 班卓琴 and 古琴 `nr` — instruments read as
+ * places and people — and `m` because a piece is as often counted as named.
+ */
+export const PLAYED_TAGS = new Set(["n", "nz", "nr", "ns", "nt", "ng", "m"]);
+/**
+ * Whether a word is an object rather than the far half of a compound.
+ *
+ * The one place this rule departs from {@link isTeachingAt}, and it is what
+ * makes it usable on technical prose. 弹 joins bound morphemes into nouns far
+ * more readily than 教 does — 着弹点, 掷弹兵, 供弹爪, 底排弹时, 弹洞 — and every
+ * one of those puts a single tagged character after the 弹 where an object
+ * would go. What the verb actually governs is a word: 吉他, 竖琴, 管风琴, 电子琴,
+ * 一曲, 肖邦. Requiring two characters keeps the compounds out at the cost of
+ * 反手持法去弹班卓琴, where the decode reaches 班 rather than 班卓琴 anyway.
+ *
+ * 教 could not take the same guard, since 教我 and 教你 are the commonest shape
+ * it has. Nobody plays a pronoun, so `r` is out of {@link PLAYED_TAGS} too.
+ */
+function isPlayedObject(word) {
+    return word !== undefined && toCharacters(word).length > 1;
+}
+/**
+ * Whether the 弹 at a position is playing something.
+ *
+ * Much the same shape as {@link isTeachingAt}, and for the same reason: the
+ * object is what says so. A 弹 with an instrument, a tune or a name after it is
+ * governing it, and a 弹 with 了, 过, 着 or 得 after it is a verb whatever
+ * follows that. A particle in front of it rules it out, since a verb does not
+ * follow one.
+ */
+function isPlayingAt(context, at) {
+    if (tagOf(context, wordEndingAt(context, at)).startsWith(PARTICLE_TAG)) {
+        return false;
+    }
+    if (ASPECT.has(context.characters[at + 1] ?? "")) {
+        return true;
+    }
+    const played = wordStartingAt(context, at + 1);
+    return isPlayedObject(played) && PLAYED_TAGS.has(tagOf(context, played));
+}
+/**
+ * 弹 read as `tán` where it is playing rather than a projectile.
+ *
+ * The character is stored `dàn` with `tán` as an alternate, and the sources are
+ * agreed about it: Unihan's `kHanyuPinlu` counts `dàn(313)` against `tán(50)`
+ * and `kMandarin` names `dàn`. That is a fact about a corpus in which nearly
+ * every 弹 is ammunition — 子弹, 炮弹, 导弹, 原子弹, 手榴弹 — and every one of
+ * those is a word that carries its own reading. What the default is left
+ * deciding is the 弹 that stands as a word of its own, and in running text that
+ * one is the verb: 他会弹一点儿古筝 came out `dàn`, and 弹吉他, 弹竖琴, 我在弹肖邦
+ * and 他钢琴弹得很好 with it.
+ *
+ * See {@link isPlayingAt} for what decides it. Forcing the single-character
+ * edge is not enough on its own, for the reason {@link TEACHING_JIAO} gives: a
+ * reading spanning two characters carries its own 弹 into the position, and
+ * 我的爱好是开车和弹吉他 read `dàn` off 和弹, a pair the dictionary holds with
+ * no part of speech and at the cost of a reading nobody counted. A *tagged*
+ * word ending in 弹 is left alone, which is every one that matters — 子弹, 炸弹,
+ * 导弹, 反弹, 手榴弹. Nothing has to be done about the other side, since the
+ * nominal compounds starting with 弹 are all listed: 弹匣, 弹坑, 弹壳, 弹片,
+ * 弹药, 弹道 and 弹头 reach their reading through the word, so a bare 弹 with a
+ * noun after it is not the front of one that got away.
+ *
+ * The 88,866 lines of Tatoeba and zh.wikipedia the other rules were sized
+ * against are the wrong corpus for this one on their own: 38 弹 decode as a
+ * word of their own in them, and 31 are somebody playing something. So CPP's
+ * 20,147 sentences are measured with them as plain text — the benchmark is
+ * drawn from zh.wikipedia's military articles, which is precisely where the
+ * shapes this rule can break live. Over the 109,013 lines together, 60 弹
+ * decode as a word of their own, every one read `dàn`, and 34 of the 60 are
+ * wrong.
+ *
+ * This moves 30 of them, of which 29 are right: the whole of 弹吉他, 弹一曲肖邦
+ * and 钢琴弹得很好, and none of 南部弹, 毛瑟弹, 掷弹兵, 供弹爪, 着弹点, 弹洞 or
+ * 底排弹时. The one it breaks is 拆弹专家, where a compound has an object's
+ * shape on both sides. Five it leaves alone: 开始弹, 四手联弹 and 弹起三次, which
+ * have no object to see, 用那种指法弹不会觉得费力, and 反手持法去弹班卓琴, where
+ * the decode reaches 班 rather than 班卓琴. On CPP's 40 labelled 弹 nothing
+ * moves at all, every one of them having been right through a word.
+ */
+export const PLAYING_TAN = {
+    name: "playing-tan",
+    verdictFor: (context) => {
+        const { edge } = context;
+        if (PLUCKED.has(edge.text)) {
+            return isTan(edge.reading[0]) && isPlayingAt(context, edge.from)
+                ? "force"
+                : "keep";
+        }
+        return edge.partOfSpeech === "" &&
+            isPluckedAt(context, edge.to - 1) &&
+            edge.reading.length === edge.to - edge.from &&
+            !isTan(edge.reading.at(-1)) &&
+            isPlayingAt(context, edge.to - 1)
+            ? "forbid"
+            : "keep";
+    },
+};
+//# sourceMappingURL=tan-rule.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"tan-rule.js","sourceRoot":"","sources":["../../src/decode/tan-rule.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAEvD,OAAO,EACL,MAAM,EACN,YAAY,EACZ,KAAK,EACL,YAAY,EACZ,cAAc,GACf,MAAM,mBAAmB,CAAC;AAG3B;;GAEG;AACH,MAAM,UAAU,KAAK,CAAC,QAA8B;IAClD,OAAO,CACL,QAAQ,EAAE,OAAO,KAAK,GAAG,IAAI,QAAQ,CAAC,KAAK,KAAK,IAAI,IAAI,QAAQ,CAAC,IAAI,KAAK,CAAC,CAC5E,CAAC;AACJ,CAAC;AAED;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,OAAO,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;AAE3C;;GAEG;AACH,MAAM,UAAU,WAAW,CAAC,OAAoB,EAAE,EAAU;IAC1D,OAAO,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,UAAU,CAAC,EAAE,CAAC,IAAI,EAAE,CAAC,CAAC;AACnD,CAAC;AAED;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,WAAW,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,CAAC,CAAC,CAAC;AAE7E;;;;;;;;;;;;;GAaG;AACH,SAAS,cAAc,CAAC,IAAwB;IAC9C,OAAO,IAAI,KAAK,SAAS,IAAI,YAAY,CAAC,IAAI,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC;AAC7D,CAAC;AAED;;;;;;;;GAQG;AACH,SAAS,WAAW,CAAC,OAAoB,EAAE,EAAU;IACnD,IAAI,KAAK,CAAC,OAAO,EAAE,YAAY,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC,CAAC,UAAU,CAAC,YAAY,CAAC,EAAE,CAAC;QACvE,OAAO,KAAK,CAAC;IACf,CAAC;IACD,IAAI,MAAM,CAAC,GAAG,CAAC,OAAO,CAAC,UAAU,CAAC,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC;QACjD,OAAO,IAAI,CAAC;IACd,CAAC;IACD,MAAM,MAAM,GAAG,cAAc,CAAC,OAAO,EAAE,EAAE,GAAG,CAAC,CAAC,CAAC;IAC/C,OAAO,cAAc,CAAC,MAAM,CAAC,IAAI,WAAW,CAAC,GAAG,CAAC,KAAK,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC,CAAC;AAC3E,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAuCG;AACH,MAAM,CAAC,MAAM,WAAW,GAAa;IACnC,IAAI,EAAE,aAAa;IACnB,UAAU,EAAE,CAAC,OAAoB,EAAE,EAAE;QACnC,MAAM,EAAE,IAAI,EAAE,GAAG,OAAO,CAAC;QACzB,IAAI,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YAC3B,OAAO,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,IAAI,WAAW,CAAC,OAAO,EAAE,IAAI,CAAC,IAAI,CAAC;gBAC9D,CAAC,CAAC,OAAO;gBACT,CAAC,CAAC,MAAM,CAAC;QACb,CAAC;QACD,OAAO,IAAI,CAAC,YAAY,KAAK,EAAE;YAC7B,WAAW,CAAC,OAAO,EAAE,IAAI,CAAC,EAAE,GAAG,CAAC,CAAC;YACjC,IAAI,CAAC,OAAO,CAAC,MAAM,KAAK,IAAI,CAAC,EAAE,GAAG,IAAI,CAAC,IAAI;YAC3C,CAAC,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;YAC3B,WAAW,CAAC,OAAO,EAAE,IAAI,CAAC,EAAE,GAAG,CAAC,CAAC;YACjC,CAAC,CAAC,QAAQ;YACV,CAAC,CAAC,MAAM,CAAC;IACb,CAAC;CACF,CAAC"}

package/dist/decode/write-run.d.ts

@@ -0,0 +1,28 @@
+/**
+ * Writing one decoded Han run out as pieces.
+ *
+ * Sandhi runs across the whole run rather than within a word, and 分词连写
+ * rewrites the word boundaries without touching the readings, so both happen
+ * here rather than per word.
+ */
+import type { Dictionary } from "../dictionary/dictionary.js";
+import type { Locale } from "../script/script.js";
+import { type SandhiOptions } from "./sandhi.js";
+import type { ScoredWord } from "./word.js";
+import { type ConvertedPiece, type Written } from "./pieces.js";
+/**
+ * Write one Han run's worth of decoded words.
+ */
+export declare function writeRun(dictionary: Dictionary, words: readonly ScoredWord[], locale: Locale, written: Written, sandhi: SandhiOptions | undefined): readonly ConvertedPiece[];
+/**
+ * Rewrite every piece's text with a pass that reads across all of them.
+ */
+export declare function rewrite(pieces: readonly ConvertedPiece[], pass: (parts: readonly string[]) => readonly string[]): readonly ConvertedPiece[];
+/**
+ * The words a Han run decodes to, with 分词连写 applied.
+ *
+ * Grouping rewrites word boundaries and never the readings behind them, so the
+ * syllables — and the confidence beside them — survive it in order.
+ */
+export declare function wordsOf(decoded: readonly ScoredWord[], dictionary: Dictionary, isGrouped: boolean): readonly ScoredWord[];
+//# sourceMappingURL=write-run.d.ts.map

package/dist/decode/write-run.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"write-run.d.ts","sourceRoot":"","sources":["../../src/decode/write-run.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,6BAA6B,CAAC;AAE9D,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,qBAAqB,CAAC;AAElD,OAAO,EAAe,KAAK,aAAa,EAAE,MAAM,aAAa,CAAC;AAC9D,OAAO,KAAK,EAAe,UAAU,EAAE,MAAM,WAAW,CAAC;AACzD,OAAO,EACL,KAAK,cAAc,EAInB,KAAK,OAAO,EACb,MAAM,aAAa,CAAC;AAErB;;GAEG;AACH,wBAAgB,QAAQ,CACtB,UAAU,EAAE,UAAU,EACtB,KAAK,EAAE,SAAS,UAAU,EAAE,EAC5B,MAAM,EAAE,MAAM,EACd,OAAO,EAAE,OAAO,EAChB,MAAM,EAAE,aAAa,GAAG,SAAS,GAChC,SAAS,cAAc,EAAE,CA0C3B;AAED;;GAEG;AACH,wBAAgB,OAAO,CACrB,MAAM,EAAE,SAAS,cAAc,EAAE,EACjC,IAAI,EAAE,CAAC,KAAK,EAAE,SAAS,MAAM,EAAE,KAAK,SAAS,MAAM,EAAE,GACpD,SAAS,cAAc,EAAE,CAO3B;AAED;;;;;GAKG;AACH,wBAAgB,OAAO,CACrB,OAAO,EAAE,SAAS,UAAU,EAAE,EAC9B,UAAU,EAAE,UAAU,EACtB,SAAS,EAAE,OAAO,GACjB,SAAS,UAAU,EAAE,CAUvB"}

package/dist/decode/write-run.js

@@ -0,0 +1,77 @@
+import { applyGrouping } from "../orthography/grouping.js";
+import { divisionOf } from "./constituents.js";
+import { applySandhi } from "./sandhi.js";
+import { plainPiece, readingFor, writeWord, } from "./pieces.js";
+/**
+ * Write one Han run's worth of decoded words.
+ */
+export function writeRun(dictionary, words, locale, written, sandhi) {
+    // Sandhi runs across the whole run rather than within a word, since 不 in one
+    // word assimilates to the tone starting the next. Third-tone sandhi needs to
+    // know where the words are all the same, so the grouping goes with it — and
+    // only when it is asked for, since dividing a word costs lookups.
+    const readings = words.map((scored) => readingFor(dictionary, scored.word, locale));
+    const grouping = sandhi?.thirdTone === true
+        ? words.map((scored, index) => {
+            const reading = readings[index] ?? [];
+            return (divisionOf(dictionary, scored.word.text, reading) ?? reading.length);
+        })
+        : undefined;
+    const flattened = applySandhi(readings.flat(), sandhi, grouping);
+    let at = 0;
+    const pieces = [];
+    for (const [index, scored] of words.entries()) {
+        /* c8 ignore next -- readings is built by mapping over these same words */
+        const length = readings[index]?.length ?? 0;
+        if (index > 0) {
+            // A space, unless 分词连写 wrote a hyphen: 干干净净 is one orthographic
+            // word, `gāngān-jìngjìng`, cut into two decoded ones.
+            pieces.push(plainPiece(scored.word.separator ?? " "));
+        }
+        pieces.push(...writeWord(flattened.slice(at, at + length), 
+        // A locale reading of a different length cannot be lined up with the
+        // confidence reported for the reading it replaced.
+        scored.confidence.length === length ? scored.confidence : [], scored.word, written));
+        at += length;
+    }
+    return pieces;
+}
+/**
+ * Rewrite every piece's text with a pass that reads across all of them.
+ */
+export function rewrite(pieces, pass) {
+    const rewritten = pass(pieces.map((piece) => piece.text));
+    return pieces.map((piece, at) => {
+        /* c8 ignore next -- one part comes back for each part handed over */
+        const text = rewritten[at] ?? piece.text;
+        return text === piece.text ? piece : { ...piece, text };
+    });
+}
+/**
+ * The words a Han run decodes to, with 分词连写 applied.
+ *
+ * Grouping rewrites word boundaries and never the readings behind them, so the
+ * syllables — and the confidence beside them — survive it in order.
+ */
+export function wordsOf(decoded, dictionary, isGrouped) {
+    return isGrouped
+        ? regroup(decoded, applyGrouping(decoded.map((scored) => scored.word), dictionary))
+        : decoded;
+}
+/**
+ * Redistribute the decode's per-syllable confidence over regrouped words.
+ *
+ * 分词连写 moves word boundaries — 看 and 了 become 看了 — without touching a
+ * reading, so the run's syllables are the same syllables in the same order and
+ * are simply cut in different places.
+ */
+function regroup(decoded, grouped) {
+    const confidence = decoded.flatMap((scored) => [...scored.confidence]);
+    let at = 0;
+    return grouped.map((word) => {
+        const held = confidence.slice(at, at + word.reading.length);
+        at += word.reading.length;
+        return { word, confidence: held };
+    });
+}
+//# sourceMappingURL=write-run.js.map

package/dist/decode/write-run.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"write-run.js","sourceRoot":"","sources":["../../src/decode/write-run.ts"],"names":[],"mappings":"AAQA,OAAO,EAAE,aAAa,EAAE,MAAM,4BAA4B,CAAC;AAE3D,OAAO,EAAE,UAAU,EAAE,MAAM,mBAAmB,CAAC;AAC/C,OAAO,EAAE,WAAW,EAAsB,MAAM,aAAa,CAAC;AAE9D,OAAO,EAEL,UAAU,EACV,UAAU,EACV,SAAS,GAEV,MAAM,aAAa,CAAC;AAErB;;GAEG;AACH,MAAM,UAAU,QAAQ,CACtB,UAAsB,EACtB,KAA4B,EAC5B,MAAc,EACd,OAAgB,EAChB,MAAiC;IAEjC,6EAA6E;IAC7E,6EAA6E;IAC7E,4EAA4E;IAC5E,kEAAkE;IAClE,MAAM,QAAQ,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,EAAE,CACpC,UAAU,CAAC,UAAU,EAAE,MAAM,CAAC,IAAI,EAAE,MAAM,CAAC,CAC5C,CAAC;IACF,MAAM,QAAQ,GACZ,MAAM,EAAE,SAAS,KAAK,IAAI;QACxB,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE;YAC1B,MAAM,OAAO,GAAG,QAAQ,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC;YACtC,OAAO,CACL,UAAU,CAAC,UAAU,EAAE,MAAM,CAAC,IAAI,CAAC,IAAI,EAAE,OAAO,CAAC,IAAI,OAAO,CAAC,MAAM,CACpE,CAAC;QACJ,CAAC,CAAC;QACJ,CAAC,CAAC,SAAS,CAAC;IAChB,MAAM,SAAS,GAAG,WAAW,CAAC,QAAQ,CAAC,IAAI,EAAE,EAAE,MAAM,EAAE,QAAQ,CAAC,CAAC;IAEjE,IAAI,EAAE,GAAG,CAAC,CAAC;IACX,MAAM,MAAM,GAAqB,EAAE,CAAC;IACpC,KAAK,MAAM,CAAC,KAAK,EAAE,MAAM,CAAC,IAAI,KAAK,CAAC,OAAO,EAAE,EAAE,CAAC;QAC9C,0EAA0E;QAC1E,MAAM,MAAM,GAAG,QAAQ,CAAC,KAAK,CAAC,EAAE,MAAM,IAAI,CAAC,CAAC;QAC5C,IAAI,KAAK,GAAG,CAAC,EAAE,CAAC;YACd,gEAAgE;YAChE,sDAAsD;YACtD,MAAM,CAAC,IAAI,CAAC,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC,SAAS,IAAI,GAAG,CAAC,CAAC,CAAC;QACxD,CAAC;QACD,MAAM,CAAC,IAAI,CACT,GAAG,SAAS,CACV,SAAS,CAAC,KAAK,CAAC,EAAE,EAAE,EAAE,GAAG,MAAM,CAAC;QAChC,qEAAqE;QACrE,mDAAmD;QACnD,MAAM,CAAC,UAAU,CAAC,MAAM,KAAK,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,UAAU,CAAC,CAAC,CAAC,EAAE,EAC5D,MAAM,CAAC,IAAI,EACX,OAAO,CACR,CACF,CAAC;QACF,EAAE,IAAI,MAAM,CAAC;IACf,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,OAAO,CACrB,MAAiC,EACjC,IAAqD;IAErD,MAAM,SAAS,GAAG,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC;IAC1D,OAAO,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,EAAE,EAAE;QAC9B,qEAAqE;QACrE,MAAM,IAAI,GAAG,SAAS,CAAC,EAAE,CAAC,IAAI,KAAK,CAAC,IAAI,CAAC;QACzC,OAAO,IAAI,KAAK,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,GAAG,KAAK,EAAE,IAAI,EAAE,CAAC;IAC1D,CAAC,CAAC,CAAC;AACL,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,OAAO,CACrB,OAA8B,EAC9B,UAAsB,EACtB,SAAkB;IAElB,OAAO,SAAS;QACd,CAAC,CAAC,OAAO,CACL,OAAO,EACP,aAAa,CACX,OAAO,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,IAAI,CAAC,EACpC,UAAU,CACX,CACF;QACH,CAAC,CAAC,OAAO,CAAC;AACd,CAAC;AAED;;;;;;GAMG;AACH,SAAS,OAAO,CACd,OAA8B,EAC9B,OAA+B;IAE/B,MAAM,UAAU,GAAG,OAAO,CAAC,OAAO,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC,GAAG,MAAM,CAAC,UAAU,CAAC,CAAC,CAAC;IACvE,IAAI,EAAE,GAAG,CAAC,CAAC;IACX,OAAO,OAAO,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE;QAC1B,MAAM,IAAI,GAAG,UAAU,CAAC,KAAK,CAAC,EAAE,EAAE,EAAE,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;QAC5D,EAAE,IAAI,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC;QAC1B,OAAO,EAAE,IAAI,EAAE,UAAU,EAAE,IAAI,EAAE,CAAC;IACpC,CAAC,CAAC,CAAC;AACL,CAAC"}

package/dist/dictionary/artifact-claims.d.ts

@@ -0,0 +1,25 @@
+/**
+ * Which entry owns each key, since an entry claims both its scripts.
+ *
+ * Settled before a single line is written, and separately from writing them,
+ * because it is the one place the artifact resolves a disagreement rather than
+ * recording one: 发 and 髮 both claim 发 and read it differently, and something
+ * has to say which of them the key means.
+ */
+import type { DictionaryEntry } from "./entry.js";
+/**
+ * Decide which entry owns each key, since an entry claims both its scripts.
+ *
+ * Entries overlap: 头 claims 頭 as its 繁體 key, and 頭 is also a character
+ * entry of its own. They agree, so the conflict is harmless there — but 发 and
+ * 髮 both claim 发, and they emphatically do not agree, one reading `fā` and the
+ * other `fà`. The entry whose own headword *is* the key wins, which is what
+ * keeps 发 on its own reading rather than the one that reached it sideways.
+ *
+ * An entry claims every 繁體 spelling a source attests for it, not only the one
+ * it stores as {@link DictionaryEntry.hant}: 重复 claims 重複 and 重覆 alike,
+ * since a reader typing either expects `chóngfù` rather than the character by
+ * character reading a missing key falls back to.
+ */
+export declare function claimKeys(entries: readonly DictionaryEntry[]): ReadonlyMap<string, DictionaryEntry>;
+//# sourceMappingURL=artifact-claims.d.ts.map

package/dist/dictionary/artifact-claims.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"artifact-claims.d.ts","sourceRoot":"","sources":["../../src/dictionary/artifact-claims.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AACH,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,YAAY,CAAC;AAuBlD;;;;;;;;;;;;;GAaG;AACH,wBAAgB,SAAS,CACvB,OAAO,EAAE,SAAS,eAAe,EAAE,GAClC,WAAW,CAAC,MAAM,EAAE,eAAe,CAAC,CAiBtC"}

package/dist/dictionary/artifact-claims.js

@@ -0,0 +1,47 @@
+import { traditionalForms } from "./artifact-format.js";
+/**
+ * Let one entry claim each of its 繁體 spellings, where nothing outranks it.
+ *
+ * A spelling equal to the 简体 headword is already claimed, and a spelling that
+ * is some entry's own headword belongs to that entry — 发 and 髮 disagree about
+ * the reading, and the one the key names has to win.
+ */
+function claimTraditional(claimed, entry) {
+    for (const form of traditionalForms(entry)) {
+        const held = claimed.get(form);
+        const isTaken = form === entry.hans || held?.hans === form;
+        if (!isTaken && (held === undefined || entry.frequency > held.frequency)) {
+            claimed.set(form, entry);
+        }
+    }
+}
+/**
+ * Decide which entry owns each key, since an entry claims both its scripts.
+ *
+ * Entries overlap: 头 claims 頭 as its 繁體 key, and 頭 is also a character
+ * entry of its own. They agree, so the conflict is harmless there — but 发 and
+ * 髮 both claim 发, and they emphatically do not agree, one reading `fā` and the
+ * other `fà`. The entry whose own headword *is* the key wins, which is what
+ * keeps 发 on its own reading rather than the one that reached it sideways.
+ *
+ * An entry claims every 繁體 spelling a source attests for it, not only the one
+ * it stores as {@link DictionaryEntry.hant}: 重复 claims 重複 and 重覆 alike,
+ * since a reader typing either expects `chóngfù` rather than the character by
+ * character reading a missing key falls back to.
+ */
+export function claimKeys(entries) {
+    const claimed = new Map();
+    // Headwords first, so that the second pass can never displace one: a 繁體
+    // alias reaching a key sideways must not outrank the entry that key names.
+    for (const entry of entries) {
+        const held = claimed.get(entry.hans);
+        if (held === undefined || entry.frequency > held.frequency) {
+            claimed.set(entry.hans, entry);
+        }
+    }
+    for (const entry of entries) {
+        claimTraditional(claimed, entry);
+    }
+    return claimed;
+}
+//# sourceMappingURL=artifact-claims.js.map

package/dist/dictionary/artifact-claims.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"artifact-claims.js","sourceRoot":"","sources":["../../src/dictionary/artifact-claims.ts"],"names":[],"mappings":"AASA,OAAO,EAAE,gBAAgB,EAAE,MAAM,sBAAsB,CAAC;AAExD;;;;;;GAMG;AACH,SAAS,gBAAgB,CACvB,OAAqC,EACrC,KAAsB;IAEtB,KAAK,MAAM,IAAI,IAAI,gBAAgB,CAAC,KAAK,CAAC,EAAE,CAAC;QAC3C,MAAM,IAAI,GAAG,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QAC/B,MAAM,OAAO,GAAG,IAAI,KAAK,KAAK,CAAC,IAAI,IAAI,IAAI,EAAE,IAAI,KAAK,IAAI,CAAC;QAC3D,IAAI,CAAC,OAAO,IAAI,CAAC,IAAI,KAAK,SAAS,IAAI,KAAK,CAAC,SAAS,GAAG,IAAI,CAAC,SAAS,CAAC,EAAE,CAAC;YACzE,OAAO,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;QAC3B,CAAC;IACH,CAAC;AACH,CAAC;AAED;;;;;;;;;;;;;GAaG;AACH,MAAM,UAAU,SAAS,CACvB,OAAmC;IAEnC,MAAM,OAAO,GAAG,IAAI,GAAG,EAA2B,CAAC;IAEnD,wEAAwE;IACxE,2EAA2E;IAC3E,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,MAAM,IAAI,GAAG,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QACrC,IAAI,IAAI,KAAK,SAAS,IAAI,KAAK,CAAC,SAAS,GAAG,IAAI,CAAC,SAAS,EAAE,CAAC;YAC3D,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;QACjC,CAAC;IACH,CAAC;IAED,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,gBAAgB,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC;IACnC,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC"}