@kensio/pinyinjs 1.7.5 → 1.7.7

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

@@ -0,0 +1,17 @@
+/**
+ * Recutting a run's words for 分词连写.
+ *
+ * Grouping rewrites word boundaries and never the readings behind them, so the
+ * syllables — and the confidence beside them — survive it in order, and are
+ * simply cut in different places.
+ */
+import type { Dictionary } from "../dictionary/dictionary.js";
+import type { ScoredWord } from "./word.js";
+/**
+ * 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=run-words.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"run-words.d.ts","sourceRoot":"","sources":["../../src/decode/run-words.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,6BAA6B,CAAC;AAE9D,OAAO,KAAK,EAAe,UAAU,EAAE,MAAM,WAAW,CAAC;AAEzD;;;;;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/run-words.js

@@ -0,0 +1,29 @@
+import { applyGrouping } from "../orthography/grouping.js";
+/**
+ * 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=run-words.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"run-words.js","sourceRoot":"","sources":["../../src/decode/run-words.ts"],"names":[],"mappings":"AAQA,OAAO,EAAE,aAAa,EAAE,MAAM,4BAA4B,CAAC;AAG3D;;;;;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/decode/sandhi-input.d.ts

@@ -0,0 +1,30 @@
+/**
+ * What the sandhi pass needs from a decoded run, beyond the syllables.
+ *
+ * Two questions sandhi cannot answer from a reading alone. Third-tone sandhi
+ * applies inside a prosodic foot, so it has to be told where the words and
+ * their parts are. 一 sandhi asks what stands in front of the 一, and the answer
+ * is a character rather than a spelling. Both are things the decode knows and
+ * the syllable array does not.
+ */
+import type { Dictionary } from "../dictionary/dictionary.js";
+import type { Syllable } from "../syllable/syllable.js";
+import type { SandhiGrouping } from "./sandhi.js";
+import type { ScoredWord } from "./word.js";
+/**
+ * Where the words and their constituents fall, for third-tone sandhi.
+ *
+ * Undefined unless third-tone sandhi was asked for, since dividing a word costs
+ * dictionary lookups and nothing else reads the answer.
+ */
+export declare function groupingOf(dictionary: Dictionary, words: readonly ScoredWord[], readings: readonly (readonly Syllable[])[], isThirdTone: boolean): SandhiGrouping | undefined;
+/**
+ * One 汉字 per syllable, where a word offers one.
+ *
+ * A word whose reading is a different length from its text — 玩儿 as `wánr` —
+ * has no character to give any one of its syllables, so its syllables get none
+ * and the pass falls back to their spellings. See
+ * {@link import("./sandhi-tones.js").isCounting}.
+ */
+export declare function charactersPerSyllable(words: readonly ScoredWord[], readings: readonly (readonly Syllable[])[]): readonly (string | undefined)[];
+//# sourceMappingURL=sandhi-input.d.ts.map

package/dist/decode/sandhi-input.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"sandhi-input.d.ts","sourceRoot":"","sources":["../../src/decode/sandhi-input.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AACH,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,6BAA6B,CAAC;AAE9D,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAExD,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,aAAa,CAAC;AAClD,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,WAAW,CAAC;AAE5C;;;;;GAKG;AACH,wBAAgB,UAAU,CACxB,UAAU,EAAE,UAAU,EACtB,KAAK,EAAE,SAAS,UAAU,EAAE,EAC5B,QAAQ,EAAE,SAAS,CAAC,SAAS,QAAQ,EAAE,CAAC,EAAE,EAC1C,WAAW,EAAE,OAAO,GACnB,cAAc,GAAG,SAAS,CAQ5B;AAED;;;;;;;GAOG;AACH,wBAAgB,qBAAqB,CACnC,KAAK,EAAE,SAAS,UAAU,EAAE,EAC5B,QAAQ,EAAE,SAAS,CAAC,SAAS,QAAQ,EAAE,CAAC,EAAE,GACzC,SAAS,CAAC,MAAM,GAAG,SAAS,CAAC,EAAE,CAQjC"}

package/dist/decode/sandhi-input.js

@@ -0,0 +1,35 @@
+import { toCharacters } from "../script/characters.js";
+import { divisionOf } from "./constituents.js";
+/**
+ * Where the words and their constituents fall, for third-tone sandhi.
+ *
+ * Undefined unless third-tone sandhi was asked for, since dividing a word costs
+ * dictionary lookups and nothing else reads the answer.
+ */
+export function groupingOf(dictionary, words, readings, isThirdTone) {
+    if (!isThirdTone) {
+        return undefined;
+    }
+    return words.map((scored, index) => {
+        const reading = readings[index] ?? [];
+        return divisionOf(dictionary, scored.word.text, reading) ?? reading.length;
+    });
+}
+/**
+ * One 汉字 per syllable, where a word offers one.
+ *
+ * A word whose reading is a different length from its text — 玩儿 as `wánr` —
+ * has no character to give any one of its syllables, so its syllables get none
+ * and the pass falls back to their spellings. See
+ * {@link import("./sandhi-tones.js").isCounting}.
+ */
+export function charactersPerSyllable(words, readings) {
+    return words.flatMap((scored, index) => {
+        const reading = readings[index] ?? [];
+        const held = toCharacters(scored.word.text);
+        return held.length === reading.length
+            ? held
+            : Array.from({ length: reading.length });
+    });
+}
+//# sourceMappingURL=sandhi-input.js.map

package/dist/decode/sandhi-input.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"sandhi-input.js","sourceRoot":"","sources":["../../src/decode/sandhi-input.ts"],"names":[],"mappings":"AAUA,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAEvD,OAAO,EAAE,UAAU,EAAE,MAAM,mBAAmB,CAAC;AAI/C;;;;;GAKG;AACH,MAAM,UAAU,UAAU,CACxB,UAAsB,EACtB,KAA4B,EAC5B,QAA0C,EAC1C,WAAoB;IAEpB,IAAI,CAAC,WAAW,EAAE,CAAC;QACjB,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,OAAO,KAAK,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE;QACjC,MAAM,OAAO,GAAG,QAAQ,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC;QACtC,OAAO,UAAU,CAAC,UAAU,EAAE,MAAM,CAAC,IAAI,CAAC,IAAI,EAAE,OAAO,CAAC,IAAI,OAAO,CAAC,MAAM,CAAC;IAC7E,CAAC,CAAC,CAAC;AACL,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,qBAAqB,CACnC,KAA4B,EAC5B,QAA0C;IAE1C,OAAO,KAAK,CAAC,OAAO,CAAC,CAAC,MAAM,EAAE,KAAK,EAAmC,EAAE;QACtE,MAAM,OAAO,GAAG,QAAQ,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC;QACtC,MAAM,IAAI,GAAG,YAAY,CAAC,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAC5C,OAAO,IAAI,CAAC,MAAM,KAAK,OAAO,CAAC,MAAM;YACnC,CAAC,CAAC,IAAI;YACN,CAAC,CAAC,KAAK,CAAC,IAAI,CAAqB,EAAE,MAAM,EAAE,OAAO,CAAC,MAAM,EAAE,CAAC,CAAC;IACjE,CAAC,CAAC,CAAC;AACL,CAAC"}

package/dist/decode/sandhi-tones.d.ts

@@ -47,6 +47,28 @@ export declare const CONTINUES_NUMBER: Set<string>;
  * The ordinal prefix 第.
  */
 export declare const ORDINAL_PREFIX = "di";
+/**
+ * The 汉字 {@link NUMERAL_WORDS} names, for a caller that has them.
+ *
+ * The spellings are ambiguous and the characters are not, which is the whole
+ * point of carrying both. 是 is spelt `shi` and stands in front of a 一 in
+ * 那是一条狗, 这是一个好主意 and several hundred sentences like them, and every
+ * one of them lost its sandhi to a 十 that was never there. 时, 事, 使 and 试 do
+ * the same.
+ *
+ * The 繁體 forms are here because a 繁體 text reaches this pass with 繁體
+ * characters. 拾, 佰 and 仟 are the 大写 banking forms, which appear on cheques
+ * and in contracts and count exactly as their everyday spellings do.
+ */
+export declare const NUMERAL_CHARACTERS: Set<string>;
+/**
+ * The 汉字 {@link CONTINUES_NUMBER} names.
+ */
+export declare const CONTINUES_NUMBER_CHARACTERS: Set<string>;
+/**
+ * The ordinal prefix as it is written.
+ */
+export declare const ORDINAL_CHARACTER = "\u7B2C";
 /**
  * Whether a 一 is counting something, which is the only time it takes sandhi.
  *
@@ -60,10 +82,16 @@ export declare const ORDINAL_PREFIX = "di";
  *   is counting the ten.
  * - **An ordinal.** 第一个 is `dìyī gè`.
  *
- * **Read off the syllables rather than the characters**, because that is all
- * this pass has: `pinyinjs sandhi shíyī gè` never sees a hanzi. So it cannot
- * tell 十 from 时 or 第 from 地, and `docs/sandhi/` carries what that costs —
- * measured, not estimated.
+ * **Answered from the characters wherever the caller has them**, and from the
+ * spellings where it does not. `pinyinjs sandhi shíyī gè` is handed nothing but
+ * pinyin, so there the pass still cannot tell 十 from 时 or 第 from 地 and
+ * `docs/sandhi/` carries what that costs. A conversion knows which 汉字 it read,
+ * and passing them along is what stops 那是一条狗 losing its sandhi to the 是 in
+ * front of the 一.
+ *
+ * The two sides are decided separately, because a run can know one and not the
+ * other: a word whose reading has a different syllable count from its character
+ * count has no character to offer for any one syllable.
  */
-export declare function isCounting(syllables: readonly Syllable[], at: number): boolean;
+export declare function isCounting(syllables: readonly Syllable[], at: number, characters?: readonly (string | undefined)[]): boolean;
 //# sourceMappingURL=sandhi-tones.d.ts.map

package/dist/decode/sandhi-tones.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"sandhi-tones.d.ts","sourceRoot":"","sources":["../../src/decode/sandhi-tones.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,EAAE,KAAK,QAAQ,EAAyB,MAAM,yBAAyB,CAAC;AAC/E,OAAO,EAAgB,KAAK,IAAI,EAAE,MAAM,iBAAiB,CAAC;AAE1D;;GAEG;AACH,wBAAgB,IAAI,CAAC,QAAQ,EAAE,QAAQ,GAAG,OAAO,CAEhD;AAED;;GAEG;AACH,wBAAgB,IAAI,CAAC,QAAQ,EAAE,QAAQ,GAAG,OAAO,CAEhD;AAED;;;;;;GAMG;AACH,wBAAgB,YAAY,CAAC,SAAS,EAAE,IAAI,GAAG,SAAS,GAAG,IAAI,CAK9D;AAED;;;;;;;;;;;;GAYG;AACH,eAAO,MAAM,aAAa,aAAyC,CAAC;AAEpE;;;;;GAKG;AACH,eAAO,MAAM,gBAAgB,aAAsC,CAAC;AAEpE;;GAEG;AACH,eAAO,MAAM,cAAc,OAAO,CAAC;AAEnC;;;;;;;;;;;;;;;;;GAiBG;AACH,wBAAgB,UAAU,CACxB,SAAS,EAAE,SAAS,QAAQ,EAAE,EAC9B,EAAE,EAAE,MAAM,GACT,OAAO,CAUT"}
+{"version":3,"file":"sandhi-tones.d.ts","sourceRoot":"","sources":["../../src/decode/sandhi-tones.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,EAAE,KAAK,QAAQ,EAAyB,MAAM,yBAAyB,CAAC;AAC/E,OAAO,EAAgB,KAAK,IAAI,EAAE,MAAM,iBAAiB,CAAC;AAE1D;;GAEG;AACH,wBAAgB,IAAI,CAAC,QAAQ,EAAE,QAAQ,GAAG,OAAO,CAEhD;AAED;;GAEG;AACH,wBAAgB,IAAI,CAAC,QAAQ,EAAE,QAAQ,GAAG,OAAO,CAEhD;AAED;;;;;;GAMG;AACH,wBAAgB,YAAY,CAAC,SAAS,EAAE,IAAI,GAAG,SAAS,GAAG,IAAI,CAK9D;AAED;;;;;;;;;;;;GAYG;AACH,eAAO,MAAM,aAAa,aAAyC,CAAC;AAEpE;;;;;GAKG;AACH,eAAO,MAAM,gBAAgB,aAAsC,CAAC;AAEpE;;GAEG;AACH,eAAO,MAAM,cAAc,OAAO,CAAC;AAEnC;;;;;;;;;;;;GAYG;AACH,eAAO,MAAM,kBAAkB,aAS7B,CAAC;AAEH;;GAEG;AACH,eAAO,MAAM,2BAA2B,aAItC,CAAC;AAEH;;GAEG;AACH,eAAO,MAAM,iBAAiB,WAAM,CAAC;AAErC;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AACH,wBAAgB,UAAU,CACxB,SAAS,EAAE,SAAS,QAAQ,EAAE,EAC9B,EAAE,EAAE,MAAM,EACV,UAAU,GAAE,SAAS,CAAC,MAAM,GAAG,SAAS,CAAC,EAAO,GAC/C,OAAO,CAwBT"}

package/dist/decode/sandhi-tones.js

@@ -56,6 +56,41 @@ export const CONTINUES_NUMBER = new Set([...NUMERAL_WORDS, "ling"]);
  * The ordinal prefix 第.
  */
 export const ORDINAL_PREFIX = "di";
+/**
+ * The 汉字 {@link NUMERAL_WORDS} names, for a caller that has them.
+ *
+ * The spellings are ambiguous and the characters are not, which is the whole
+ * point of carrying both. 是 is spelt `shi` and stands in front of a 一 in
+ * 那是一条狗, 这是一个好主意 and several hundred sentences like them, and every
+ * one of them lost its sandhi to a 十 that was never there. 时, 事, 使 and 试 do
+ * the same.
+ *
+ * The 繁體 forms are here because a 繁體 text reaches this pass with 繁體
+ * characters. 拾, 佰 and 仟 are the 大写 banking forms, which appear on cheques
+ * and in contracts and count exactly as their everyday spellings do.
+ */
+export const NUMERAL_CHARACTERS = new Set([
+    "十",
+    "百",
+    "千",
+    "万",
+    "萬",
+    "拾",
+    "佰",
+    "仟",
+]);
+/**
+ * The 汉字 {@link CONTINUES_NUMBER} names.
+ */
+export const CONTINUES_NUMBER_CHARACTERS = new Set([
+    ...NUMERAL_CHARACTERS,
+    "零",
+    "〇",
+]);
+/**
+ * The ordinal prefix as it is written.
+ */
+export const ORDINAL_CHARACTER = "第";
 /**
  * Whether a 一 is counting something, which is the only time it takes sandhi.
  *
@@ -69,20 +104,36 @@ export const ORDINAL_PREFIX = "di";
  *   is counting the ten.
  * - **An ordinal.** 第一个 is `dìyī gè`.
  *
- * **Read off the syllables rather than the characters**, because that is all
- * this pass has: `pinyinjs sandhi shíyī gè` never sees a hanzi. So it cannot
- * tell 十 from 时 or 第 from 地, and `docs/sandhi/` carries what that costs —
- * measured, not estimated.
+ * **Answered from the characters wherever the caller has them**, and from the
+ * spellings where it does not. `pinyinjs sandhi shíyī gè` is handed nothing but
+ * pinyin, so there the pass still cannot tell 十 from 时 or 第 from 地 and
+ * `docs/sandhi/` carries what that costs. A conversion knows which 汉字 it read,
+ * and passing them along is what stops 那是一条狗 losing its sandhi to the 是 in
+ * front of the 一.
+ *
+ * The two sides are decided separately, because a run can know one and not the
+ * other: a word whose reading has a different syllable count from its character
+ * count has no character to offer for any one syllable.
  */
-export function isCounting(syllables, at) {
+export function isCounting(syllables, at, characters = []) {
     const spelt = (index) => {
         const syllable = syllables[index];
         return syllable === undefined ? "" : writeSyllableSpelling(syllable);
     };
-    const before = spelt(at - 1);
-    if (before === ORDINAL_PREFIX) {
+    const before = characters[at - 1];
+    if (before === undefined
+        ? spelt(at - 1) === ORDINAL_PREFIX
+        : before === ORDINAL_CHARACTER) {
         return false;
     }
-    return !NUMERAL_WORDS.has(before) || CONTINUES_NUMBER.has(spelt(at + 1));
+    if (before === undefined
+        ? !NUMERAL_WORDS.has(spelt(at - 1))
+        : !NUMERAL_CHARACTERS.has(before)) {
+        return true;
+    }
+    const after = characters[at + 1];
+    return after === undefined
+        ? CONTINUES_NUMBER.has(spelt(at + 1))
+        : CONTINUES_NUMBER_CHARACTERS.has(after);
 }
 //# sourceMappingURL=sandhi-tones.js.map

package/dist/decode/sandhi-tones.js.map

@@ -1 +1 @@
-{"version":3,"file":"sandhi-tones.js","sourceRoot":"","sources":["../../src/decode/sandhi-tones.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,EAAiB,qBAAqB,EAAE,MAAM,yBAAyB,CAAC;AAC/E,OAAO,EAAE,YAAY,EAAa,MAAM,iBAAiB,CAAC;AAE1D;;GAEG;AACH,MAAM,UAAU,IAAI,CAAC,QAAkB;IACrC,OAAO,QAAQ,CAAC,OAAO,KAAK,EAAE,IAAI,QAAQ,CAAC,KAAK,KAAK,GAAG,CAAC;AAC3D,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,IAAI,CAAC,QAAkB;IACrC,OAAO,QAAQ,CAAC,OAAO,KAAK,GAAG,IAAI,QAAQ,CAAC,KAAK,KAAK,GAAG,CAAC;AAC5D,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,YAAY,CAAC,SAA2B;IACtD,IAAI,SAAS,KAAK,SAAS,IAAI,SAAS,KAAK,YAAY,EAAE,CAAC;QAC1D,OAAO,CAAC,CAAC;IACX,CAAC;IACD,OAAO,SAAS,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;AACjC,CAAC;AAED;;;;;;;;;;;;GAYG;AACH,MAAM,CAAC,MAAM,aAAa,GAAG,IAAI,GAAG,CAAC,CAAC,KAAK,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,CAAC,CAAC,CAAC;AAEpE;;;;;GAKG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,aAAa,EAAE,MAAM,CAAC,CAAC,CAAC;AAEpE;;GAEG;AACH,MAAM,CAAC,MAAM,cAAc,GAAG,IAAI,CAAC;AAEnC;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAM,UAAU,UAAU,CACxB,SAA8B,EAC9B,EAAU;IAEV,MAAM,KAAK,GAAG,CAAC,KAAa,EAAU,EAAE;QACtC,MAAM,QAAQ,GAAG,SAAS,CAAC,KAAK,CAAC,CAAC;QAClC,OAAO,QAAQ,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,qBAAqB,CAAC,QAAQ,CAAC,CAAC;IACvE,CAAC,CAAC;IACF,MAAM,MAAM,GAAG,KAAK,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;IAC7B,IAAI,MAAM,KAAK,cAAc,EAAE,CAAC;QAC9B,OAAO,KAAK,CAAC;IACf,CAAC;IACD,OAAO,CAAC,aAAa,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,gBAAgB,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC;AAC3E,CAAC"}
+{"version":3,"file":"sandhi-tones.js","sourceRoot":"","sources":["../../src/decode/sandhi-tones.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,EAAiB,qBAAqB,EAAE,MAAM,yBAAyB,CAAC;AAC/E,OAAO,EAAE,YAAY,EAAa,MAAM,iBAAiB,CAAC;AAE1D;;GAEG;AACH,MAAM,UAAU,IAAI,CAAC,QAAkB;IACrC,OAAO,QAAQ,CAAC,OAAO,KAAK,EAAE,IAAI,QAAQ,CAAC,KAAK,KAAK,GAAG,CAAC;AAC3D,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,IAAI,CAAC,QAAkB;IACrC,OAAO,QAAQ,CAAC,OAAO,KAAK,GAAG,IAAI,QAAQ,CAAC,KAAK,KAAK,GAAG,CAAC;AAC5D,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,YAAY,CAAC,SAA2B;IACtD,IAAI,SAAS,KAAK,SAAS,IAAI,SAAS,KAAK,YAAY,EAAE,CAAC;QAC1D,OAAO,CAAC,CAAC;IACX,CAAC;IACD,OAAO,SAAS,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;AACjC,CAAC;AAED;;;;;;;;;;;;GAYG;AACH,MAAM,CAAC,MAAM,aAAa,GAAG,IAAI,GAAG,CAAC,CAAC,KAAK,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,CAAC,CAAC,CAAC;AAEpE;;;;;GAKG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,aAAa,EAAE,MAAM,CAAC,CAAC,CAAC;AAEpE;;GAEG;AACH,MAAM,CAAC,MAAM,cAAc,GAAG,IAAI,CAAC;AAEnC;;;;;;;;;;;;GAYG;AACH,MAAM,CAAC,MAAM,kBAAkB,GAAG,IAAI,GAAG,CAAC;IACxC,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;CACJ,CAAC,CAAC;AAEH;;GAEG;AACH,MAAM,CAAC,MAAM,2BAA2B,GAAG,IAAI,GAAG,CAAC;IACjD,GAAG,kBAAkB;IACrB,GAAG;IACH,GAAG;CACJ,CAAC,CAAC;AAEH;;GAEG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAAG,GAAG,CAAC;AAErC;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AACH,MAAM,UAAU,UAAU,CACxB,SAA8B,EAC9B,EAAU,EACV,UAAU,GAAoC,EAAE;IAEhD,MAAM,KAAK,GAAG,CAAC,KAAa,EAAU,EAAE;QACtC,MAAM,QAAQ,GAAG,SAAS,CAAC,KAAK,CAAC,CAAC;QAClC,OAAO,QAAQ,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,qBAAqB,CAAC,QAAQ,CAAC,CAAC;IACvE,CAAC,CAAC;IACF,MAAM,MAAM,GAAG,UAAU,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;IAClC,IACE,MAAM,KAAK,SAAS;QAClB,CAAC,CAAC,KAAK,CAAC,EAAE,GAAG,CAAC,CAAC,KAAK,cAAc;QAClC,CAAC,CAAC,MAAM,KAAK,iBAAiB,EAChC,CAAC;QACD,OAAO,KAAK,CAAC;IACf,CAAC;IACD,IACE,MAAM,KAAK,SAAS;QAClB,CAAC,CAAC,CAAC,aAAa,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;QACnC,CAAC,CAAC,CAAC,kBAAkB,CAAC,GAAG,CAAC,MAAM,CAAC,EACnC,CAAC;QACD,OAAO,IAAI,CAAC;IACd,CAAC;IACD,MAAM,KAAK,GAAG,UAAU,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;IACjC,OAAO,KAAK,KAAK,SAAS;QACxB,CAAC,CAAC,gBAAgB,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;QACrC,CAAC,CAAC,2BAA2B,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;AAC7C,CAAC"}

package/dist/decode/sandhi.d.ts

@@ -39,14 +39,17 @@ export type SandhiGrouping = readonly (number | readonly number[])[];
  * contextual tone is put back, over the whole syllable array rather than within
  * a word, which is what lets 不 in one word assimilate to a tone in the next.
  *
- * 一 and 不 assimilate to whatever follows them and need nothing but the
- * syllables. Third-tone sandhi is not like that — its domain is the foot rather
- * than the syllable string — so `grouping` says where the words and their parts
- * are. Without one the whole array is taken for a single word, which is what a
- * caller holding nothing but a reading has.
+ * 不 assimilates to whatever follows it and needs nothing but the syllables.
+ * Third-tone sandhi is not like that — its domain is the foot rather than the
+ * syllable string — so `grouping` says where the words and their parts are.
+ * Without one the whole array is taken for a single word, which is what a
+ * caller holding nothing but a reading has. 一 sits between the two: the tone it
+ * takes is settled by the syllables, and whether it takes one at all is a
+ * question about the 汉字, so `characters` carries them where the caller has
+ * them. See {@link isCounting}.
  *
  * Never operates on a string. The old project patched output text with regexes,
  * which is what made its rules order-dependent and untestable.
  */
-export declare function applySandhi(syllables: readonly Syllable[], options?: SandhiOptions, grouping?: SandhiGrouping): readonly Syllable[];
+export declare function applySandhi(syllables: readonly Syllable[], options?: SandhiOptions, grouping?: SandhiGrouping, characters?: readonly (string | undefined)[]): readonly Syllable[];
 //# sourceMappingURL=sandhi.d.ts.map

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

@@ -1 +1 @@
-{"version":3,"file":"sandhi.d.ts","sourceRoot":"","sources":["../../src/decode/sandhi.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAExD;;GAEG;AACH,MAAM,WAAW,aAAa;IAC5B;;;;;OAKG;IACH,QAAQ,CAAC,IAAI,CAAC,EAAE,OAAO,CAAC;IACxB;;;;;;OAMG;IACH,QAAQ,CAAC,SAAS,CAAC,EAAE,OAAO,CAAC;CAC9B;AAED;;;;;;;;;GASG;AACH,MAAM,MAAM,cAAc,GAAG,SAAS,CAAC,MAAM,GAAG,SAAS,MAAM,EAAE,CAAC,EAAE,CAAC;AA6HrE;;;;;;;;;;;;;;;;;GAiBG;AACH,wBAAgB,WAAW,CACzB,SAAS,EAAE,SAAS,QAAQ,EAAE,EAC9B,OAAO,GAAE,aAAkB,EAC3B,QAAQ,CAAC,EAAE,cAAc,GACxB,SAAS,QAAQ,EAAE,CA6BrB"}
+{"version":3,"file":"sandhi.d.ts","sourceRoot":"","sources":["../../src/decode/sandhi.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAExD;;GAEG;AACH,MAAM,WAAW,aAAa;IAC5B;;;;;OAKG;IACH,QAAQ,CAAC,IAAI,CAAC,EAAE,OAAO,CAAC;IACxB;;;;;;OAMG;IACH,QAAQ,CAAC,SAAS,CAAC,EAAE,OAAO,CAAC;CAC9B;AAED;;;;;;;;;GASG;AACH,MAAM,MAAM,cAAc,GAAG,SAAS,CAAC,MAAM,GAAG,SAAS,MAAM,EAAE,CAAC,EAAE,CAAC;AA6HrE;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,wBAAgB,WAAW,CACzB,SAAS,EAAE,SAAS,QAAQ,EAAE,EAC9B,OAAO,GAAE,aAAkB,EAC3B,QAAQ,CAAC,EAAE,cAAc,EACzB,UAAU,CAAC,EAAE,SAAS,CAAC,MAAM,GAAG,SAAS,CAAC,EAAE,GAC3C,SAAS,QAAQ,EAAE,CA6BrB"}

package/dist/decode/sandhi.js

@@ -103,16 +103,19 @@ function applyThirdTone(syllables, grouping) {
  * contextual tone is put back, over the whole syllable array rather than within
  * a word, which is what lets 不 in one word assimilate to a tone in the next.
  *
- * 一 and 不 assimilate to whatever follows them and need nothing but the
- * syllables. Third-tone sandhi is not like that — its domain is the foot rather
- * than the syllable string — so `grouping` says where the words and their parts
- * are. Without one the whole array is taken for a single word, which is what a
- * caller holding nothing but a reading has.
+ * 不 assimilates to whatever follows it and needs nothing but the syllables.
+ * Third-tone sandhi is not like that — its domain is the foot rather than the
+ * syllable string — so `grouping` says where the words and their parts are.
+ * Without one the whole array is taken for a single word, which is what a
+ * caller holding nothing but a reading has. 一 sits between the two: the tone it
+ * takes is settled by the syllables, and whether it takes one at all is a
+ * question about the 汉字, so `characters` carries them where the caller has
+ * them. See {@link isCounting}.
  *
  * Never operates on a string. The old project patched output text with regexes,
  * which is what made its rules order-dependent and untestable.
  */
-export function applySandhi(syllables, options = {}, grouping) {
+export function applySandhi(syllables, options = {}, grouping, characters) {
     const { yiBu = true, thirdTone = false } = options;
     const applied = thirdTone
         ? applyThirdTone(syllables, grouping)
@@ -122,7 +125,7 @@ export function applySandhi(syllables, options = {}, grouping) {
         if (yiBu && isYi(syllable) && syllable.tone === 1) {
             // A 一 that is not counting keeps its citation tone, so it is left
             // exactly as the dictionary stored it.
-            if (isCounting(applied, at)) {
+            if (isCounting(applied, at, characters)) {
                 applied[at] = { ...syllable, tone: yiToneBefore(following?.tone) };
             }
             continue;

package/dist/decode/sandhi.js.map

@@ -1 +1 @@
-{"version":3,"file":"sandhi.js","sourceRoot":"","sources":["../../src/decode/sandhi.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,IAAI,EAAE,UAAU,EAAE,IAAI,EAAE,YAAY,EAAE,MAAM,mBAAmB,CAAC;AAkDzE;;;;;;;GAOG;AACH,SAAS,WAAW,CAClB,MAAc,EACd,QAAyB;IAEzB,MAAM,SAAS,GAAoB;QACjC,KAAK,EAAE,KAAK,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,IAAI,CAAC,GAAG,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,EAAE,CAAC;QACrE,KAAK,EAAE,EAAE;QACT,KAAK,EAAE,EAAE;KACV,CAAC;IACF,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;QAC3B,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,IAAI,EAAE,GAAG,CAAC,CAAC;IACX,KAAK,MAAM,CAAC,KAAK,EAAE,IAAI,CAAC,IAAI,QAAQ,CAAC,OAAO,EAAE,EAAE,CAAC;QAC/C,MAAM,SAAS,GAAG,OAAO,IAAI,KAAK,QAAQ,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;QAC3D,MAAM,IAAI,GAAG,EAAE,CAAC;QAChB,KAAK,MAAM,CAAC,QAAQ,EAAE,IAAI,CAAC,IAAI,SAAS,CAAC,OAAO,EAAE,EAAE,CAAC;YACnD,KAAK,IAAI,IAAI,GAAG,CAAC,EAAE,IAAI,GAAG,IAAI,EAAE,IAAI,EAAE,EAAE,CAAC;gBACvC,KAAK,CAAC,IAAI,CAAC,EAAE,GAAG,IAAI,GAAG,CAAC,CAAC,CAAC;YAC5B,CAAC;YACD,EAAE,IAAI,IAAI,CAAC;YACX,yEAAyE;YACzE,kCAAkC;YAClC,IAAI,QAAQ,GAAG,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBACpC,KAAK,CAAC,IAAI,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;YACrB,CAAC;QACH,CAAC;QACD,IAAI,EAAE,GAAG,IAAI,KAAK,CAAC,IAAI,KAAK,GAAG,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACnD,KAAK,CAAC,IAAI,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;QACrB,CAAC;IACH,CAAC;IACD,OAAO,EAAE,KAAK,MAAM,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,SAAS,CAAC;AAC7D,CAAC;AAED;;;;;;GAMG;AACH,SAAS,KAAK,CAAC,OAAmB,EAAE,EAAU;IAC5C,MAAM,QAAQ,GAAG,OAAO,CAAC,EAAE,CAAC,CAAC;IAC7B,IAAI,QAAQ,EAAE,IAAI,KAAK,CAAC,IAAI,OAAO,CAAC,EAAE,GAAG,CAAC,CAAC,EAAE,IAAI,KAAK,CAAC,EAAE,CAAC;QACxD,OAAO,CAAC,EAAE,CAAC,GAAG,EAAE,GAAG,QAAQ,EAAE,IAAI,EAAE,CAAC,EAAE,CAAC;IACzC,CAAC;AACH,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA8BG;AACH,SAAS,cAAc,CACrB,SAA8B,EAC9B,QAAyB;IAEzB,MAAM,OAAO,GAAG,CAAC,GAAG,SAAS,CAAC,CAAC;IAC/B,MAAM,SAAS,GAAG,WAAW,CAAC,SAAS,CAAC,MAAM,EAAE,QAAQ,CAAC,CAAC;IAE1D,KAAK,MAAM,IAAI,IAAI,CAAC,SAAS,CAAC,KAAK,EAAE,SAAS,CAAC,KAAK,EAAE,SAAS,CAAC,KAAK,CAAC,EAAE,CAAC;QACvE,KAAK,MAAM,EAAE,IAAI,IAAI,EAAE,CAAC;YACtB,KAAK,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC;QACrB,CAAC;IACH,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC;AAED;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAM,UAAU,WAAW,CACzB,SAA8B,EAC9B,OAAO,GAAkB,EAAE,EAC3B,QAAyB;IAEzB,MAAM,EAAE,IAAI,GAAG,IAAI,EAAE,SAAS,GAAG,KAAK,EAAE,GAAG,OAAO,CAAC;IACnD,MAAM,OAAO,GAAG,SAAS;QACvB,CAAC,CAAC,cAAc,CAAC,SAAS,EAAE,QAAQ,CAAC;QACrC,CAAC,CAAC,CAAC,GAAG,SAAS,CAAC,CAAC;IAEnB,KAAK,MAAM,CAAC,EAAE,EAAE,QAAQ,CAAC,IAAI,OAAO,CAAC,OAAO,EAAE,EAAE,CAAC;QAC/C,MAAM,SAAS,GAAG,OAAO,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;QAElC,IAAI,IAAI,IAAI,IAAI,CAAC,QAAQ,CAAC,IAAI,QAAQ,CAAC,IAAI,KAAK,CAAC,EAAE,CAAC;YAClD,kEAAkE;YAClE,uCAAuC;YACvC,IAAI,UAAU,CAAC,OAAO,EAAE,EAAE,CAAC,EAAE,CAAC;gBAC5B,OAAO,CAAC,EAAE,CAAC,GAAG,EAAE,GAAG,QAAQ,EAAE,IAAI,EAAE,YAAY,CAAC,SAAS,EAAE,IAAI,CAAC,EAAE,CAAC;YACrE,CAAC;YACD,SAAS;QACX,CAAC;QACD,yEAAyE;QACzE,IACE,IAAI;YACJ,IAAI,CAAC,QAAQ,CAAC;YACd,QAAQ,CAAC,IAAI,KAAK,CAAC;YACnB,SAAS,EAAE,IAAI,KAAK,CAAC,EACrB,CAAC;YACD,OAAO,CAAC,EAAE,CAAC,GAAG,EAAE,GAAG,QAAQ,EAAE,IAAI,EAAE,CAAC,EAAE,CAAC;QACzC,CAAC;IACH,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC"}
+{"version":3,"file":"sandhi.js","sourceRoot":"","sources":["../../src/decode/sandhi.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,IAAI,EAAE,UAAU,EAAE,IAAI,EAAE,YAAY,EAAE,MAAM,mBAAmB,CAAC;AAkDzE;;;;;;;GAOG;AACH,SAAS,WAAW,CAClB,MAAc,EACd,QAAyB;IAEzB,MAAM,SAAS,GAAoB;QACjC,KAAK,EAAE,KAAK,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,IAAI,CAAC,GAAG,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,EAAE,CAAC;QACrE,KAAK,EAAE,EAAE;QACT,KAAK,EAAE,EAAE;KACV,CAAC;IACF,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;QAC3B,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,IAAI,EAAE,GAAG,CAAC,CAAC;IACX,KAAK,MAAM,CAAC,KAAK,EAAE,IAAI,CAAC,IAAI,QAAQ,CAAC,OAAO,EAAE,EAAE,CAAC;QAC/C,MAAM,SAAS,GAAG,OAAO,IAAI,KAAK,QAAQ,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;QAC3D,MAAM,IAAI,GAAG,EAAE,CAAC;QAChB,KAAK,MAAM,CAAC,QAAQ,EAAE,IAAI,CAAC,IAAI,SAAS,CAAC,OAAO,EAAE,EAAE,CAAC;YACnD,KAAK,IAAI,IAAI,GAAG,CAAC,EAAE,IAAI,GAAG,IAAI,EAAE,IAAI,EAAE,EAAE,CAAC;gBACvC,KAAK,CAAC,IAAI,CAAC,EAAE,GAAG,IAAI,GAAG,CAAC,CAAC,CAAC;YAC5B,CAAC;YACD,EAAE,IAAI,IAAI,CAAC;YACX,yEAAyE;YACzE,kCAAkC;YAClC,IAAI,QAAQ,GAAG,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBACpC,KAAK,CAAC,IAAI,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;YACrB,CAAC;QACH,CAAC;QACD,IAAI,EAAE,GAAG,IAAI,KAAK,CAAC,IAAI,KAAK,GAAG,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACnD,KAAK,CAAC,IAAI,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;QACrB,CAAC;IACH,CAAC;IACD,OAAO,EAAE,KAAK,MAAM,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,SAAS,CAAC;AAC7D,CAAC;AAED;;;;;;GAMG;AACH,SAAS,KAAK,CAAC,OAAmB,EAAE,EAAU;IAC5C,MAAM,QAAQ,GAAG,OAAO,CAAC,EAAE,CAAC,CAAC;IAC7B,IAAI,QAAQ,EAAE,IAAI,KAAK,CAAC,IAAI,OAAO,CAAC,EAAE,GAAG,CAAC,CAAC,EAAE,IAAI,KAAK,CAAC,EAAE,CAAC;QACxD,OAAO,CAAC,EAAE,CAAC,GAAG,EAAE,GAAG,QAAQ,EAAE,IAAI,EAAE,CAAC,EAAE,CAAC;IACzC,CAAC;AACH,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA8BG;AACH,SAAS,cAAc,CACrB,SAA8B,EAC9B,QAAyB;IAEzB,MAAM,OAAO,GAAG,CAAC,GAAG,SAAS,CAAC,CAAC;IAC/B,MAAM,SAAS,GAAG,WAAW,CAAC,SAAS,CAAC,MAAM,EAAE,QAAQ,CAAC,CAAC;IAE1D,KAAK,MAAM,IAAI,IAAI,CAAC,SAAS,CAAC,KAAK,EAAE,SAAS,CAAC,KAAK,EAAE,SAAS,CAAC,KAAK,CAAC,EAAE,CAAC;QACvE,KAAK,MAAM,EAAE,IAAI,IAAI,EAAE,CAAC;YACtB,KAAK,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC;QACrB,CAAC;IACH,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC;AAED;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,MAAM,UAAU,WAAW,CACzB,SAA8B,EAC9B,OAAO,GAAkB,EAAE,EAC3B,QAAyB,EACzB,UAA4C;IAE5C,MAAM,EAAE,IAAI,GAAG,IAAI,EAAE,SAAS,GAAG,KAAK,EAAE,GAAG,OAAO,CAAC;IACnD,MAAM,OAAO,GAAG,SAAS;QACvB,CAAC,CAAC,cAAc,CAAC,SAAS,EAAE,QAAQ,CAAC;QACrC,CAAC,CAAC,CAAC,GAAG,SAAS,CAAC,CAAC;IAEnB,KAAK,MAAM,CAAC,EAAE,EAAE,QAAQ,CAAC,IAAI,OAAO,CAAC,OAAO,EAAE,EAAE,CAAC;QAC/C,MAAM,SAAS,GAAG,OAAO,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;QAElC,IAAI,IAAI,IAAI,IAAI,CAAC,QAAQ,CAAC,IAAI,QAAQ,CAAC,IAAI,KAAK,CAAC,EAAE,CAAC;YAClD,kEAAkE;YAClE,uCAAuC;YACvC,IAAI,UAAU,CAAC,OAAO,EAAE,EAAE,EAAE,UAAU,CAAC,EAAE,CAAC;gBACxC,OAAO,CAAC,EAAE,CAAC,GAAG,EAAE,GAAG,QAAQ,EAAE,IAAI,EAAE,YAAY,CAAC,SAAS,EAAE,IAAI,CAAC,EAAE,CAAC;YACrE,CAAC;YACD,SAAS;QACX,CAAC;QACD,yEAAyE;QACzE,IACE,IAAI;YACJ,IAAI,CAAC,QAAQ,CAAC;YACd,QAAQ,CAAC,IAAI,KAAK,CAAC;YACnB,SAAS,EAAE,IAAI,KAAK,CAAC,EACrB,CAAC;YACD,OAAO,CAAC,EAAE,CAAC,GAAG,EAAE,GAAG,QAAQ,EAAE,IAAI,EAAE,CAAC,EAAE,CAAC;QACzC,CAAC;IACH,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC"}

package/dist/decode/script-region.d.ts

@@ -0,0 +1,15 @@
+/**
+ * Writing a converted word in a region's own 繁體 orthography.
+ *
+ * The last pass over a word's choices. Taiwan and Hong Kong disagree about the
+ * standard form of 58 characters and about nothing else, so this rewrites forms
+ * and never a reading.
+ */
+import { type Region } from "../script/glyphs.js";
+import type { Syllable } from "../syllable/syllable.js";
+import type { ScriptChoice } from "./script-types.js";
+/**
+ * Apply the regional 繁體 forms, which never change a reading.
+ */
+export declare function applyRegion(choices: readonly ScriptChoice[], reading: readonly Syllable[], region: Region, isAligned: boolean): readonly ScriptChoice[];
+//# sourceMappingURL=script-region.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"script-region.d.ts","sourceRoot":"","sources":["../../src/decode/script-region.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,EAGL,KAAK,MAAM,EAGZ,MAAM,qBAAqB,CAAC;AAC7B,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AACxD,OAAO,KAAK,EAAE,YAAY,EAAkB,MAAM,mBAAmB,CAAC;AAEtE;;GAEG;AACH,wBAAgB,WAAW,CACzB,OAAO,EAAE,SAAS,YAAY,EAAE,EAChC,OAAO,EAAE,SAAS,QAAQ,EAAE,EAC5B,MAAM,EAAE,MAAM,EACd,SAAS,EAAE,OAAO,GACjB,SAAS,YAAY,EAAE,CAsBzB"}

package/dist/decode/script-region.js

@@ -0,0 +1,61 @@
+/**
+ * Writing a converted word in a region's own 繁體 orthography.
+ *
+ * The last pass over a word's choices. Taiwan and Hong Kong disagree about the
+ * standard form of 58 characters and about nothing else, so this rewrites forms
+ * and never a reading.
+ */
+import { DEFAULT_REGION, isReadingSensitive, regionalGlyphsOf, toRegionalGlyph, } from "../script/glyphs.js";
+/**
+ * Apply the regional 繁體 forms, which never change a reading.
+ */
+export 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;
+        const to = toRegionalGlyph(choice.to, region, syllable);
+        const alternatives = regionalAlternatives(choice, region, to, isGuess);
+        return {
+            ...choice,
+            to,
+            // A region can leave a character one form where the script left it two:
+            // Hong Kong writes 台 for both 台 and 臺, and 泄 for both 泄 and 洩. There
+            // was nothing to decide once the region is applied, whatever the script
+            // conversion thought it was deciding.
+            evidence: evidenceAfter(choice, alternatives, isGuess),
+            alternatives,
+        };
+    });
+}
+/**
+ * What settled a choice once the region has had its say.
+ */
+function evidenceAfter(choice, alternatives, isGuess) {
+    if (alternatives.length === 0) {
+        return "locked";
+    }
+    return isGuess ? "default" : choice.evidence;
+}
+/**
+ * The forms a choice was written against, in the region's own orthography.
+ *
+ * Two things happen here. The rivals the script tables named are rewritten,
+ * since a choice reporting 面 settled for Hong Kong beside a Taiwan 麵 would be
+ * writing two orthographies at once — 3,163 of the 39,430 alternatives over
+ * 88,866 lines were doing that, mostly 裡 for 裏 and 麵 for 麪. And a guess the
+ * region itself created gains the form the reading would have kept, which is
+ * the only thing 著 and 參 were ever choosing between.
+ */
+function regionalAlternatives(choice, region, to, isGuess) {
+    const forms = [
+        ...choice.alternatives.map((form) => toRegionalGlyph(form, region)),
+        ...(isGuess ? regionalGlyphsOf(choice.to, region) : []),
+    ];
+    return [...new Set(forms)].filter((form) => form !== to);
+}
+//# sourceMappingURL=script-region.js.map

package/dist/decode/script-region.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"script-region.js","sourceRoot":"","sources":["../../src/decode/script-region.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,EACL,cAAc,EACd,kBAAkB,EAElB,gBAAgB,EAChB,eAAe,GAChB,MAAM,qBAAqB,CAAC;AAI7B;;GAEG;AACH,MAAM,UAAU,WAAW,CACzB,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,MAAM,EAAE,GAAG,eAAe,CAAC,MAAM,CAAC,EAAE,EAAE,MAAM,EAAE,QAAQ,CAAC,CAAC;QACxD,MAAM,YAAY,GAAG,oBAAoB,CAAC,MAAM,EAAE,MAAM,EAAE,EAAE,EAAE,OAAO,CAAC,CAAC;QACvE,OAAO;YACL,GAAG,MAAM;YACT,EAAE;YACF,wEAAwE;YACxE,qEAAqE;YACrE,wEAAwE;YACxE,sCAAsC;YACtC,QAAQ,EAAE,aAAa,CAAC,MAAM,EAAE,YAAY,EAAE,OAAO,CAAC;YACtD,YAAY;SACb,CAAC;IACJ,CAAC,CAAC,CAAC;AACL,CAAC;AAED;;GAEG;AACH,SAAS,aAAa,CACpB,MAAoB,EACpB,YAA+B,EAC/B,OAAgB;IAEhB,IAAI,YAAY,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC9B,OAAO,QAAQ,CAAC;IAClB,CAAC;IACD,OAAO,OAAO,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,MAAM,CAAC,QAAQ,CAAC;AAC/C,CAAC;AAED;;;;;;;;;GASG;AACH,SAAS,oBAAoB,CAC3B,MAAoB,EACpB,MAAc,EACd,EAAU,EACV,OAAgB;IAEhB,MAAM,KAAK,GAAG;QACZ,GAAG,MAAM,CAAC,YAAY,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,eAAe,CAAC,IAAI,EAAE,MAAM,CAAC,CAAC;QACnE,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,gBAAgB,CAAC,MAAM,CAAC,EAAE,EAAE,MAAM,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;KACxD,CAAC;IACF,OAAO,CAAC,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,KAAK,EAAE,CAAC,CAAC;AAC3D,CAAC"}

package/dist/decode/script-types.d.ts

@@ -35,7 +35,8 @@ export interface ScriptOptions {
  *
  * - `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.
+ *   characters, and most of those are one-to-one. The one evidence that comes
+ *   with no {@link ScriptChoice.alternatives}, and the only 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
@@ -63,7 +64,15 @@ export interface ScriptChoice {
     /** The character as it was converted. */
     readonly to: string;
     readonly evidence: ScriptEvidence;
-    /** The forms this character could also have taken. */
+    /**
+     * The forms this character could also have taken.
+     *
+     * Empty exactly when the evidence is `locked`, so a caller reading one of
+     * them never has to check the other. The character's own form counts: 万 is
+     * 萬 counting and 万 in the surname 万俟, and a conversion that took 萬 was
+     * choosing between the two. So does the region's: Hong Kong writes 麪 where
+     * Taiwan writes 麵, and a Hong Kong conversion says 麪.
+     */
     readonly alternatives: readonly string[];
 }
 /**

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

@@ -1 +1 @@
-{"version":3,"file":"script-types.d.ts","sourceRoot":"","sources":["../../src/decode/script-types.ts"],"names":[],"mappings":"AAAA;;GAEG;AACH,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,qBAAqB,CAAC;AAElD;;;;;;;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"}
+{"version":3,"file":"script-types.d.ts","sourceRoot":"","sources":["../../src/decode/script-types.ts"],"names":[],"mappings":"AAAA;;GAEG;AACH,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,qBAAqB,CAAC;AAElD;;;;;;;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;;;;;;;;;;;;;;GAcG;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;;;;;;;;OAQG;IACH,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"}

package/dist/decode/script-types.js

@@ -17,7 +17,8 @@ export const SCRIPT_TARGETS = [
  *
  * - `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.
+ *   characters, and most of those are one-to-one. The one evidence that comes
+ *   with no {@link ScriptChoice.alternatives}, and the only 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

package/dist/decode/script-types.js.map

@@ -1 +1 @@
-{"version":3,"file":"script-types.js","sourceRoot":"","sources":["../../src/decode/script-types.ts"],"names":[],"mappings":"AAKA;;;;;;;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"}
+{"version":3,"file":"script-types.js","sourceRoot":"","sources":["../../src/decode/script-types.ts"],"names":[],"mappings":"AAKA;;;;;;;GAOG;AACH,MAAM,CAAC,MAAM,cAAc,GAAG;IAC5B,SAAS;IACT,SAAS;IACT,YAAY;IACZ,YAAY;CACJ,CAAC;AAuBX;;;;;;;;;;;;;;GAcG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG;IAC7B,QAAQ;IACR,MAAM;IACN,SAAS;IACT,SAAS;CACD,CAAC"}

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

@@ -23,10 +23,12 @@ export declare function targetOf(target: ScriptTarget): {
 export declare function convertWord(table: ReadonlyMap<string, CharacterConversion>, words: ReadonlyMap<string, string>, word: string, reading: readonly Syllable[]): readonly ScriptChoice[];
 /**
  * Rank what settled a character, strongest evidence first.
+ *
+ * `locked` is keyed on whether a rival form survived rather than on
+ * {@link import("../script/conversion.js").isAmbiguousCharacter}, and the two part company where a word overrides
+ * the character table. 钟 has one 繁體 form by the characters and 一见钟情 is
+ * 一見鍾情, so the choice named 鐘 as the road not taken while calling itself
+ * locked. A word settled that one, which is what `word` is for.
  */
-export declare function evidenceFor(isAttested: boolean, isAtReading: boolean, isAmbiguous: boolean): ScriptEvidence;
-/**
- * Apply the regional 繁體 forms, which never change a reading.
- */
-export declare function applyRegion(choices: readonly ScriptChoice[], reading: readonly Syllable[], region: Region, isAligned: boolean): readonly ScriptChoice[];
+export declare function evidenceFor(isAttested: boolean, isAtReading: boolean, hasRival: boolean): ScriptEvidence;
 //# sourceMappingURL=script-words.d.ts.map

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

@@ -1 +1 @@
-{"version":3,"file":"script-words.d.ts","sourceRoot":"","sources":["../../src/decode/script-words.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,KAAK,EACV,YAAY,EACZ,cAAc,EACd,YAAY,EACb,MAAM,mBAAmB,CAAC;AAE3B,OAAO,EACL,KAAK,mBAAmB,EAKzB,MAAM,yBAAyB,CAAC;AACjC,OAAO,EAGL,KAAK,MAAM,EAEZ,MAAM,qBAAqB,CAAC;AAC7B,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,qBAAqB,CAAC;AAClD,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAExD;;GAEG;AACH,wBAAgB,QAAQ,CAAC,MAAM,EAAE,YAAY,GAAG;IAC9C,MAAM,EAAE,MAAM,CAAC;IACf,MAAM,EAAE,MAAM,CAAC;CAChB,CAQA;AAED;;GAEG;AACH,wBAAgB,WAAW,CACzB,KAAK,EAAE,WAAW,CAAC,MAAM,EAAE,mBAAmB,CAAC,EAC/C,KAAK,EAAE,WAAW,CAAC,MAAM,EAAE,MAAM,CAAC,EAClC,IAAI,EAAE,MAAM,EACZ,OAAO,EAAE,SAAS,QAAQ,EAAE,GAC3B,SAAS,YAAY,EAAE,CAmCzB;AAED;;GAEG;AACH,wBAAgB,WAAW,CACzB,UAAU,EAAE,OAAO,EACnB,WAAW,EAAE,OAAO,EACpB,WAAW,EAAE,OAAO,GACnB,cAAc,CAQhB;AAED;;GAEG;AACH,wBAAgB,WAAW,CACzB,OAAO,EAAE,SAAS,YAAY,EAAE,EAChC,OAAO,EAAE,SAAS,QAAQ,EAAE,EAC5B,MAAM,EAAE,MAAM,EACd,SAAS,EAAE,OAAO,GACjB,SAAS,YAAY,EAAE,CAezB"}
+{"version":3,"file":"script-words.d.ts","sourceRoot":"","sources":["../../src/decode/script-words.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,KAAK,EACV,YAAY,EACZ,cAAc,EACd,YAAY,EACb,MAAM,mBAAmB,CAAC;AAE3B,OAAO,EACL,KAAK,mBAAmB,EAIzB,MAAM,yBAAyB,CAAC;AACjC,OAAO,EAAkB,KAAK,MAAM,EAAE,MAAM,qBAAqB,CAAC;AAClE,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,qBAAqB,CAAC;AAClD,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAExD;;GAEG;AACH,wBAAgB,QAAQ,CAAC,MAAM,EAAE,YAAY,GAAG;IAC9C,MAAM,EAAE,MAAM,CAAC;IACf,MAAM,EAAE,MAAM,CAAC;CAChB,CAQA;AAED;;GAEG;AACH,wBAAgB,WAAW,CACzB,KAAK,EAAE,WAAW,CAAC,MAAM,EAAE,mBAAmB,CAAC,EAC/C,KAAK,EAAE,WAAW,CAAC,MAAM,EAAE,MAAM,CAAC,EAClC,IAAI,EAAE,MAAM,EACZ,OAAO,EAAE,SAAS,QAAQ,EAAE,GAC3B,SAAS,YAAY,EAAE,CAuCzB;AAED;;;;;;;;GAQG;AACH,wBAAgB,WAAW,CACzB,UAAU,EAAE,OAAO,EACnB,WAAW,EAAE,OAAO,EACpB,QAAQ,EAAE,OAAO,GAChB,cAAc,CAQhB"}

package/dist/decode/script-words.js

@@ -1,6 +1,6 @@
 import { toCharacters } from "../script/characters.js";
-import { conversionKey, convertCharacter, formsOf, isAmbiguousCharacter, } from "../script/conversion.js";
-import { DEFAULT_REGION, isReadingSensitive, toRegionalGlyph, } from "../script/glyphs.js";
+import { conversionKey, convertCharacter, formsOf, } from "../script/conversion.js";
+import { DEFAULT_REGION } from "../script/glyphs.js";
 /**
  * The script and region a target names.
  */
@@ -35,19 +35,31 @@ export function convertWord(table, words, word, reading) {
             ? /* c8 ignore next -- the lengths were just checked */
                 (attestedCharacters[at] ?? character)
             : convertCharacter(table, character, syllable);
+        // Every form the table names but the one taken. The character's own form
+        // belongs among them: 万 is 萬 counting and 万 in the surname 万俟, 准 is
+        // 準 and 准 in 准將, 著 is 着 and 著 at `zhù`. Excluding it left 42 简→繁
+        // characters and 3,160 繁→简 ones reporting a guess with nothing to have
+        // guessed between.
+        const alternatives = formsOf(conversion).filter((form) => form !== to);
         return {
             from: character,
             to,
-            evidence: evidenceFor(isAttested, atReading !== undefined, isAmbiguousCharacter(table, character)),
-            alternatives: formsOf(conversion).filter((form) => form !== character && form !== to),
+            evidence: evidenceFor(isAttested, atReading !== undefined, alternatives.length > 0),
+            alternatives,
         };
     });
 }
 /**
  * Rank what settled a character, strongest evidence first.
+ *
+ * `locked` is keyed on whether a rival form survived rather than on
+ * {@link import("../script/conversion.js").isAmbiguousCharacter}, and the two part company where a word overrides
+ * the character table. 钟 has one 繁體 form by the characters and 一见钟情 is
+ * 一見鍾情, so the choice named 鐘 as the road not taken while calling itself
+ * locked. A word settled that one, which is what `word` is for.
  */
-export function evidenceFor(isAttested, isAtReading, isAmbiguous) {
-    if (!isAmbiguous) {
+export function evidenceFor(isAttested, isAtReading, hasRival) {
+    if (!hasRival) {
         return "locked";
     }
     if (isAttested) {
@@ -55,23 +67,4 @@ export function evidenceFor(isAttested, isAtReading, isAmbiguous) {
     }
     return isAtReading ? "reading" : "default";
 }
-/**
- * Apply the regional 繁體 forms, which never change a reading.
- */
-export 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,
-        };
-    });
-}
 //# sourceMappingURL=script-words.js.map