@kensio/pinyinjs 1.7.4 → 1.7.5

package/dist/decode/run-lattice.js

@@ -0,0 +1,91 @@
+import { buildLattice } from "./lattice.js";
+export { runLattice } from "./lattice-slice.js";
+/**
+ * Building the lattice one Han run is decoded over.
+ *
+ * A run is decoded on its own and the pieces are put back together, so this
+ * is where a lattice is cut down to a run and the words are read back out of
+ * it. `decode.ts` is what asks for one.
+ */
+import { decodeSpacing } from "./decode.js";
+import { applyEdgeRules } from "./rules.js";
+/**
+ * Gather the reading units into the words the boundaries make of them.
+ *
+ * Where the two decodes disagree the readings win, which is what "advisory"
+ * means for the spacing decode: a boundary falling inside a reading unit is
+ * dropped rather than allowed to break the unit up. 玩儿 read as `wánr` is one
+ * syllable over two characters, and no spacing can be put between them.
+ */
+export function groupUnits(units, boundaries) {
+    const starts = new Set(units.map((unit) => unit.from));
+    const kept = new Set(boundaries.filter((at) => starts.has(at)));
+    const groups = [];
+    let held = [];
+    for (const unit of units) {
+        if (kept.has(unit.from) && held.length > 0) {
+            groups.push(held);
+            held = [];
+        }
+        held.push(unit);
+    }
+    if (held.length > 0) {
+        groups.push(held);
+    }
+    return groups;
+}
+/**
+ * Assemble one word from the units the boundaries kept together.
+ */
+export function wordFrom(dictionary, lattice, group) {
+    const first = group[0];
+    const last = group.at(-1);
+    /* c8 ignore next 3 -- a group is only made by pushing a unit into it */
+    if (first === undefined || last === undefined) {
+        throw new Error("a word must cover at least one reading unit");
+    }
+    const text = lattice.characters.slice(first.from, last.to).join("");
+    const entry = dictionary.lookup(text);
+    return {
+        text,
+        reading: group.flatMap((unit) => [...unit.reading]),
+        isProperNoun: entry?.isProperNoun ?? false,
+        partOfSpeech: entry?.partOfSpeech ?? "",
+        isKnown: entry !== undefined,
+    };
+}
+/**
+ * The lattice a run decodes over: stage 1 built, then stage 4 applied.
+ *
+ * The rules run between building and decoding rather than over the output,
+ * which is the whole point of them — they change what the decode gets to choose
+ * between, so everything downstream sees one lattice rather than a decode and
+ * a list of corrections to it.
+ */
+export function ruledLattice(dictionary, run, rules, hints) {
+    return applyEdgeRules(buildLattice(dictionary, run, hints), dictionary, rules);
+}
+/**
+ * Cut a decode of the whole lattice down to the run at the end of it.
+ *
+ * Everything before `at` was context rather than run — 汉字 a caller put in
+ * front so that the decode could see them — and is dropped once it has done its
+ * work.
+ */
+export function runGroups({ lattice, at }, units) {
+    return groupUnits(units.filter((unit) => unit.from >= at), decodeSpacing(lattice).filter((from) => from >= at));
+}
+/**
+ * Decode a Han run with the lattice: the recommended path.
+ *
+ * Stages 1 to 3 of ALGORITHM.md end to end — build every candidate reading,
+ * lock the positions that cannot vary, score the rest — where the greedy
+ * baseline commits to the longest match at each position and never revisits it.
+ *
+ * `before` is 汉字 standing in front of the run that the decode should see and
+ * not report: the 汉字 a number written in digits would have been written with.
+ * A run is decoded on its own without it, and a run reached that way has no idea
+ * what preceded it — 个人 alone is the word `gèrén`, while the 两 of 2个人 makes
+ * 两个 the word and leaves 人 to itself.
+ */
+//# sourceMappingURL=run-lattice.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"run-lattice.js","sourceRoot":"","sources":["../../src/decode/run-lattice.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAgB,MAAM,cAAc,CAAC;AAI1D,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AAChD;;;;;;GAMG;AACH,OAAO,EAAE,aAAa,EAAE,MAAM,aAAa,CAAC;AAG5C,OAAO,EAAE,cAAc,EAAiB,MAAM,YAAY,CAAC;AAG3D;;;;;;;GAOG;AACH,MAAM,UAAU,UAAU,CACxB,KAAsB,EACtB,UAA6B;IAE7B,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;IACvD,MAAM,IAAI,GAAG,IAAI,GAAG,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;IAChE,MAAM,MAAM,GAAa,EAAE,CAAC;IAC5B,IAAI,IAAI,GAAW,EAAE,CAAC;IAEtB,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,IAAI,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAC3C,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YAClB,IAAI,GAAG,EAAE,CAAC;QACZ,CAAC;QACD,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAClB,CAAC;IACD,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACpB,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IACpB,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,QAAQ,CACtB,UAAsB,EACtB,OAAgB,EAChB,KAA6B;IAE7B,MAAM,KAAK,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;IACvB,MAAM,IAAI,GAAG,KAAK,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;IAC1B,wEAAwE;IACxE,IAAI,KAAK,KAAK,SAAS,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;QAC9C,MAAM,IAAI,KAAK,CAAC,6CAA6C,CAAC,CAAC;IACjE,CAAC;IACD,MAAM,IAAI,GAAG,OAAO,CAAC,UAAU,CAAC,KAAK,CAAC,KAAK,CAAC,IAAI,EAAE,IAAI,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IACpE,MAAM,KAAK,GAAG,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC;IAEtC,OAAO;QACL,IAAI;QACJ,OAAO,EAAE,KAAK,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,GAAG,IAAI,CAAC,OAAO,CAAC,CAAC;QACnD,YAAY,EAAE,KAAK,EAAE,YAAY,IAAI,KAAK;QAC1C,YAAY,EAAE,KAAK,EAAE,YAAY,IAAI,EAAE;QACvC,OAAO,EAAE,KAAK,KAAK,SAAS;KAC7B,CAAC;AACJ,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,YAAY,CAC1B,UAAsB,EACtB,GAAW,EACX,KAA0B,EAC1B,KAAgC;IAEhC,OAAO,cAAc,CACnB,YAAY,CAAC,UAAU,EAAE,GAAG,EAAE,KAAK,CAAC,EACpC,UAAU,EACV,KAAK,CACN,CAAC;AACJ,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,SAAS,CACvB,EAAE,OAAO,EAAE,EAAE,EAAc,EAC3B,KAAsB;IAEtB,OAAO,UAAU,CACf,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,IAAI,EAAE,CAAC,EACvC,aAAa,CAAC,OAAO,CAAC,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,IAAI,EAAE,CAAC,CACpD,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;GAYG"}

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

@@ -0,0 +1,69 @@
+/**
+ * Which tone 一 and 不 take, and when a 一 is being counted rather than said.
+ *
+ * Both characters change tone by what follows them, and 一 additionally does
+ * not change at all when it is a number being counted off.
+ */
+import { type Syllable } from "../syllable/syllable.js";
+import { type Tone } from "../tone/tone.js";
+/**
+ * Whether a syllable is 一, which is the only thing spelled `yi` that sandhis.
+ */
+export declare function isYi(syllable: Syllable): boolean;
+/**
+ * Whether a syllable is 不.
+ */
+export declare function isBu(syllable: Syllable): boolean;
+/**
+ * The tone 一 takes before a given following tone.
+ *
+ * Fourth tone flattens it to second; anything else raises it to fourth. Before
+ * a neutral tone or at the end of a phrase it keeps its citation first tone,
+ * since there is nothing to assimilate to.
+ */
+export declare function yiToneBefore(following: Tone | undefined): Tone;
+/**
+ * The numeral words a 一 can be the last digit of.
+ *
+ * Toneless, because what is being asked is which word it is rather than what
+ * tone it is carrying, and 十 is `shí` in every context it appears in here.
+ *
+ * **亿 is deliberately absent**, and that is measured rather than assumed: no
+ * 一 in 88,866 lines of Tatoeba and zh.wikipedia ends a number in 亿, while 意,
+ * 议, 义 and 議 all read `yì`, all precede a 一 that really is counting, and
+ * would all lose their sandhi for it — 11 conversions against nothing gained.
+ * The other three stay, because a number can end 百一 or 千一 even though this
+ * corpus has none, and the one thing they cost is 擺一個.
+ */
+export declare const NUMERAL_WORDS: Set<string>;
+/**
+ * The words that carry a number on past a 一, so that it is still counting.
+ *
+ * 一百一十's second 一 counts the ten it is followed by and takes the sandhi;
+ * 二十一's does not, because nothing numeric comes after it.
+ */
+export declare const CONTINUES_NUMBER: Set<string>;
+/**
+ * The ordinal prefix 第.
+ */
+export declare const ORDINAL_PREFIX = "di";
+/**
+ * Whether a 一 is counting something, which is the only time it takes sandhi.
+ *
+ * The rule 一 sandhi is usually stated with — fourth tone flattens it, anything
+ * else raises it — is about the 一 that *counts*. Two shapes of 一 are not
+ * counting and keep the citation tone, and both are standard:
+ *
+ * - **The last digit of a larger number.** 十一月 is `shíyīyuè` and 二十一岁 is
+ *   `èrshíyī suì`. The signal is a numeral word before it with nothing numeric
+ *   after, which is what leaves 一百一十's middle 一 alone: 十 follows it, so it
+ *   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.
+ */
+export declare function isCounting(syllables: readonly Syllable[], at: number): boolean;
+//# sourceMappingURL=sandhi-tones.d.ts.map

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

@@ -0,0 +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"}

package/dist/decode/sandhi-tones.js

@@ -0,0 +1,88 @@
+/**
+ * Which tone 一 and 不 take, and when a 一 is being counted rather than said.
+ *
+ * Both characters change tone by what follows them, and 一 additionally does
+ * not change at all when it is a number being counted off.
+ */
+import { writeSyllableSpelling } from "../syllable/syllable.js";
+import { NEUTRAL_TONE } from "../tone/tone.js";
+/**
+ * Whether a syllable is 一, which is the only thing spelled `yi` that sandhis.
+ */
+export function isYi(syllable) {
+    return syllable.initial === "" && syllable.final === "i";
+}
+/**
+ * Whether a syllable is 不.
+ */
+export function isBu(syllable) {
+    return syllable.initial === "b" && syllable.final === "u";
+}
+/**
+ * The tone 一 takes before a given following tone.
+ *
+ * Fourth tone flattens it to second; anything else raises it to fourth. Before
+ * a neutral tone or at the end of a phrase it keeps its citation first tone,
+ * since there is nothing to assimilate to.
+ */
+export function yiToneBefore(following) {
+    if (following === undefined || following === NEUTRAL_TONE) {
+        return 1;
+    }
+    return following === 4 ? 2 : 4;
+}
+/**
+ * The numeral words a 一 can be the last digit of.
+ *
+ * Toneless, because what is being asked is which word it is rather than what
+ * tone it is carrying, and 十 is `shí` in every context it appears in here.
+ *
+ * **亿 is deliberately absent**, and that is measured rather than assumed: no
+ * 一 in 88,866 lines of Tatoeba and zh.wikipedia ends a number in 亿, while 意,
+ * 议, 义 and 議 all read `yì`, all precede a 一 that really is counting, and
+ * would all lose their sandhi for it — 11 conversions against nothing gained.
+ * The other three stay, because a number can end 百一 or 千一 even though this
+ * corpus has none, and the one thing they cost is 擺一個.
+ */
+export const NUMERAL_WORDS = new Set(["shi", "bai", "qian", "wan"]);
+/**
+ * The words that carry a number on past a 一, so that it is still counting.
+ *
+ * 一百一十's second 一 counts the ten it is followed by and takes the sandhi;
+ * 二十一's does not, because nothing numeric comes after it.
+ */
+export const CONTINUES_NUMBER = new Set([...NUMERAL_WORDS, "ling"]);
+/**
+ * The ordinal prefix 第.
+ */
+export const ORDINAL_PREFIX = "di";
+/**
+ * Whether a 一 is counting something, which is the only time it takes sandhi.
+ *
+ * The rule 一 sandhi is usually stated with — fourth tone flattens it, anything
+ * else raises it — is about the 一 that *counts*. Two shapes of 一 are not
+ * counting and keep the citation tone, and both are standard:
+ *
+ * - **The last digit of a larger number.** 十一月 is `shíyīyuè` and 二十一岁 is
+ *   `èrshíyī suì`. The signal is a numeral word before it with nothing numeric
+ *   after, which is what leaves 一百一十's middle 一 alone: 十 follows it, so it
+ *   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.
+ */
+export function isCounting(syllables, at) {
+    const spelt = (index) => {
+        const syllable = syllables[index];
+        return syllable === undefined ? "" : writeSyllableSpelling(syllable);
+    };
+    const before = spelt(at - 1);
+    if (before === ORDINAL_PREFIX) {
+        return false;
+    }
+    return !NUMERAL_WORDS.has(before) || CONTINUES_NUMBER.has(spelt(at + 1));
+}
+//# sourceMappingURL=sandhi-tones.js.map

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

@@ -0,0 +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"}

package/dist/decode/sandhi.d.ts

@@ -1,4 +1,4 @@
-import { type Syllable } from "../syllable/syllable.js";
+import type { Syllable } from "../syllable/syllable.js";
 /**
  * How much of the sandhi to apply.
  */

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,EAAE,KAAK,QAAQ,EAAyB,MAAM,yBAAyB,CAAC;AAG/E;;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;AAmNrE;;;;;;;;;;;;;;;;;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;;;;;;;;;;;;;;;;;GAiBG;AACH,wBAAgB,WAAW,CACzB,SAAS,EAAE,SAAS,QAAQ,EAAE,EAC9B,OAAO,GAAE,aAAkB,EAC3B,QAAQ,CAAC,EAAE,cAAc,GACxB,SAAS,QAAQ,EAAE,CA6BrB"}

package/dist/decode/sandhi.js

@@ -1,84 +1,4 @@
-import { writeSyllableSpelling } from "../syllable/syllable.js";
-import { NEUTRAL_TONE } from "../tone/tone.js";
-/**
- * Whether a syllable is 一, which is the only thing spelled `yi` that sandhis.
- */
-function isYi(syllable) {
-    return syllable.initial === "" && syllable.final === "i";
-}
-/**
- * Whether a syllable is 不.
- */
-function isBu(syllable) {
-    return syllable.initial === "b" && syllable.final === "u";
-}
-/**
- * The tone 一 takes before a given following tone.
- *
- * Fourth tone flattens it to second; anything else raises it to fourth. Before
- * a neutral tone or at the end of a phrase it keeps its citation first tone,
- * since there is nothing to assimilate to.
- */
-function yiToneBefore(following) {
-    if (following === undefined || following === NEUTRAL_TONE) {
-        return 1;
-    }
-    return following === 4 ? 2 : 4;
-}
-/**
- * The numeral words a 一 can be the last digit of.
- *
- * Toneless, because what is being asked is which word it is rather than what
- * tone it is carrying, and 十 is `shí` in every context it appears in here.
- *
- * **亿 is deliberately absent**, and that is measured rather than assumed: no
- * 一 in 88,866 lines of Tatoeba and zh.wikipedia ends a number in 亿, while 意,
- * 议, 义 and 議 all read `yì`, all precede a 一 that really is counting, and
- * would all lose their sandhi for it — 11 conversions against nothing gained.
- * The other three stay, because a number can end 百一 or 千一 even though this
- * corpus has none, and the one thing they cost is 擺一個.
- */
-const NUMERAL_WORDS = new Set(["shi", "bai", "qian", "wan"]);
-/**
- * The words that carry a number on past a 一, so that it is still counting.
- *
- * 一百一十's second 一 counts the ten it is followed by and takes the sandhi;
- * 二十一's does not, because nothing numeric comes after it.
- */
-const CONTINUES_NUMBER = new Set([...NUMERAL_WORDS, "ling"]);
-/**
- * The ordinal prefix 第.
- */
-const ORDINAL_PREFIX = "di";
-/**
- * Whether a 一 is counting something, which is the only time it takes sandhi.
- *
- * The rule 一 sandhi is usually stated with — fourth tone flattens it, anything
- * else raises it — is about the 一 that *counts*. Two shapes of 一 are not
- * counting and keep the citation tone, and both are standard:
- *
- * - **The last digit of a larger number.** 十一月 is `shíyīyuè` and 二十一岁 is
- *   `èrshíyī suì`. The signal is a numeral word before it with nothing numeric
- *   after, which is what leaves 一百一十's middle 一 alone: 十 follows it, so it
- *   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.
- */
-function isCounting(syllables, at) {
-    const spelt = (index) => {
-        const syllable = syllables[index];
-        return syllable === undefined ? "" : writeSyllableSpelling(syllable);
-    };
-    const before = spelt(at - 1);
-    if (before === ORDINAL_PREFIX) {
-        return false;
-    }
-    return !NUMERAL_WORDS.has(before) || CONTINUES_NUMBER.has(spelt(at + 1));
-}
+import { isBu, isCounting, isYi, yiToneBefore } from "./sandhi-tones.js";
 /**
  * Read a grouping as the junctions it puts in the syllable array.
  *

package/dist/decode/sandhi.js.map

@@ -1 +1 @@
-{"version":3,"file":"sandhi.js","sourceRoot":"","sources":["../../src/decode/sandhi.ts"],"names":[],"mappings":"AAAA,OAAO,EAAiB,qBAAqB,EAAE,MAAM,yBAAyB,CAAC;AAC/E,OAAO,EAAE,YAAY,EAAa,MAAM,iBAAiB,CAAC;AAmC1D;;GAEG;AACH,SAAS,IAAI,CAAC,QAAkB;IAC9B,OAAO,QAAQ,CAAC,OAAO,KAAK,EAAE,IAAI,QAAQ,CAAC,KAAK,KAAK,GAAG,CAAC;AAC3D,CAAC;AAED;;GAEG;AACH,SAAS,IAAI,CAAC,QAAkB;IAC9B,OAAO,QAAQ,CAAC,OAAO,KAAK,GAAG,IAAI,QAAQ,CAAC,KAAK,KAAK,GAAG,CAAC;AAC5D,CAAC;AAED;;;;;;GAMG;AACH,SAAS,YAAY,CAAC,SAA2B;IAC/C,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,aAAa,GAAG,IAAI,GAAG,CAAC,CAAC,KAAK,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,CAAC,CAAC,CAAC;AAE7D;;;;;GAKG;AACH,MAAM,gBAAgB,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,aAAa,EAAE,MAAM,CAAC,CAAC,CAAC;AAE7D;;GAEG;AACH,MAAM,cAAc,GAAG,IAAI,CAAC;AAE5B;;;;;;;;;;;;;;;;;GAiBG;AACH,SAAS,UAAU,CAAC,SAA8B,EAAE,EAAU;IAC5D,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;AAkBD;;;;;;;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;;;;;;;;;;;;;;;;;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"}

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

@@ -0,0 +1,76 @@
+/**
+ * What a script conversion is asked for and what it hands back.
+ */
+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[];
+}
+//# sourceMappingURL=script-types.d.ts.map

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

@@ -0,0 +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"}

package/dist/decode/script-types.js

@@ -0,0 +1,35 @@
+/**
+ * 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",
+];
+//# sourceMappingURL=script-types.js.map

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

@@ -0,0 +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"}

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

@@ -0,0 +1,32 @@
+/**
+ * Converting one word between the scripts, and what the evidence was.
+ *
+ * A word is converted whole where the tables know it and character by
+ * character otherwise, and each choice carries what settled it so a caller
+ * can show where the conversion was guessing.
+ */
+import type { ScriptChoice, ScriptEvidence, ScriptTarget } from "./script-types.js";
+import { type CharacterConversion } from "../script/conversion.js";
+import { type Region } from "../script/glyphs.js";
+import type { Script } from "../script/script.js";
+import type { Syllable } from "../syllable/syllable.js";
+/**
+ * The script and region a target names.
+ */
+export declare function targetOf(target: ScriptTarget): {
+    script: Script;
+    region: Region;
+};
+/**
+ * Convert one word, preferring what a source wrote to what its characters imply.
+ */
+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.
+ */
+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[];
+//# sourceMappingURL=script-words.d.ts.map

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

@@ -0,0 +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"}

package/dist/decode/script-words.js

@@ -0,0 +1,77 @@
+import { toCharacters } from "../script/characters.js";
+import { conversionKey, convertCharacter, formsOf, isAmbiguousCharacter, } from "../script/conversion.js";
+import { DEFAULT_REGION, isReadingSensitive, toRegionalGlyph, } from "../script/glyphs.js";
+/**
+ * The script and region a target names.
+ */
+export 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.
+ */
+export 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.
+ */
+export 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.
+ */
+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

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

@@ -0,0 +1 @@
+{"version":3,"file":"script-words.js","sourceRoot":"","sources":["../../src/decode/script-words.ts"],"names":[],"mappings":"AAYA,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AACvD,OAAO,EAEL,aAAa,EACb,gBAAgB,EAChB,OAAO,EACP,oBAAoB,GACrB,MAAM,yBAAyB,CAAC;AACjC,OAAO,EACL,cAAc,EACd,kBAAkB,EAElB,eAAe,GAChB,MAAM,qBAAqB,CAAC;AAI7B;;GAEG;AACH,MAAM,UAAU,QAAQ,CAAC,MAAoB;IAI3C,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,MAAM,UAAU,WAAW,CACzB,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,MAAM,UAAU,WAAW,CACzB,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,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,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"}