@kensio/pinyinjs 1.7.12 → 1.7.14

package/dist/orthography/apostrophe.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"apostrophe.d.ts","sourceRoot":"","sources":["../../src/orthography/apostrophe.ts"],"names":[],"mappings":"AAGA;;;;;;;GAOG;AACH,MAAM,MAAM,eAAe,GAAG,QAAQ,GAAG,UAAU,GAAG,OAAO,CAAC;AAyB9D;;;;;;;GAOG;AACH,wBAAgB,QAAQ,CACtB,SAAS,EAAE,SAAS,MAAM,EAAE,EAC5B,KAAK,GAAE,eAA0B,GAChC,MAAM,CAER;AAED;;;;;;;;GAQG;AACH,wBAAgB,QAAQ,CACtB,SAAS,EAAE,SAAS,MAAM,EAAE,EAC5B,KAAK,GAAE,eAA0B,GAChC,SAAS,MAAM,EAAE,CAcnB"}
+{"version":3,"file":"apostrophe.d.ts","sourceRoot":"","sources":["../../src/orthography/apostrophe.ts"],"names":[],"mappings":"AAIA;;;;;;;GAOG;AACH,MAAM,MAAM,eAAe,GAAG,QAAQ,GAAG,UAAU,GAAG,OAAO,CAAC;AAgC9D;;;;;;;GAOG;AACH,wBAAgB,QAAQ,CACtB,SAAS,EAAE,SAAS,MAAM,EAAE,EAC5B,KAAK,GAAE,eAA0B,GAChC,MAAM,CAER;AAED;;;;;;;;GAQG;AACH,wBAAgB,QAAQ,CACtB,SAAS,EAAE,SAAS,MAAM,EAAE,EAC5B,KAAK,GAAE,eAA0B,GAChC,SAAS,MAAM,EAAE,CAcnB"}

package/dist/orthography/apostrophe.js

@@ -1,3 +1,4 @@
+import { stripToneMarks } from "../tone/tone-mark.js";
 import { isSeparableStart } from "../syllable/separation.js";
 import { splitSyllables } from "../syllable/split.js";
 /**
@@ -10,14 +11,21 @@ const APOSTROPHE = "'";
  * apostrophes.
  *
  * Asked of the reader rather than of a rule, by putting the question to the
- * same splitter that parses input: `hǎiōu` comes apart as `hǎi` + `ōu` and so
- * needs nothing, where `Xīān` reads as the single syllable `xian` and does.
+ * same splitter that parses input: `haiou` comes apart as `hai` + `ou` and so
+ * needs nothing, where `Xian` reads as the single syllable `xian` and does.
  * A word the splitter cannot read at all counts as ambiguous.
+ *
+ * Asked of the letters, with the tone marks stripped off first. The splitter
+ * reads a mark as evidence of where a syllable ends, which would answer that
+ * `Xīān` is clear enough and leave 西安 without its 隔音符号. The apostrophe is
+ * there for the letters, and a text that writes no tones at all needs it just
+ * the same.
  */
 function isUnambiguous(syllables) {
-    const split = splitSyllables(syllables.join(""));
-    return (split?.length === syllables.length &&
-        split.every((read, at) => read === syllables[at]));
+    const toneless = syllables.map((syllable) => stripToneMarks(syllable));
+    const split = splitSyllables(toneless.join(""));
+    return (split?.length === toneless.length &&
+        split.every((read, at) => read === toneless[at]));
 }
 /**
  * Join a word's syllables, writing the 隔音符号 where the orthography wants it.

package/dist/orthography/apostrophe.js.map

@@ -1 +1 @@
-{"version":3,"file":"apostrophe.js","sourceRoot":"","sources":["../../src/orthography/apostrophe.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,gBAAgB,EAAE,MAAM,2BAA2B,CAAC;AAC7D,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAYtD;;;GAGG;AACH,MAAM,UAAU,GAAG,GAAG,CAAC;AAEvB;;;;;;;;GAQG;AACH,SAAS,aAAa,CAAC,SAA4B;IACjD,MAAM,KAAK,GAAG,cAAc,CAAC,SAAS,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC;IACjD,OAAO,CACL,KAAK,EAAE,MAAM,KAAK,SAAS,CAAC,MAAM;QAClC,KAAK,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,EAAE,EAAE,EAAE,CAAC,IAAI,KAAK,SAAS,CAAC,EAAE,CAAC,CAAC,CAClD,CAAC;AACJ,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,QAAQ,CACtB,SAA4B,EAC5B,KAAK,GAAoB,QAAQ;IAEjC,OAAO,QAAQ,CAAC,SAAS,EAAE,KAAK,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;AAC7C,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,QAAQ,CACtB,SAA4B,EAC5B,KAAK,GAAoB,QAAQ;IAEjC,MAAM,MAAM,GAAG,SAAS,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,EAAE,EAAE,CAC5C,EAAE,GAAG,CAAC,IAAI,gBAAgB,CAAC,QAAQ,CAAC;QAClC,CAAC,CAAC,GAAG,UAAU,GAAG,QAAQ,EAAE;QAC5B,CAAC,CAAC,QAAQ,CACb,CAAC;IAEF,IACE,KAAK,KAAK,OAAO;QACjB,MAAM,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC,EACpD,CAAC;QACD,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,OAAO,KAAK,KAAK,UAAU,IAAI,aAAa,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,MAAM,CAAC;AAC/E,CAAC"}
+{"version":3,"file":"apostrophe.js","sourceRoot":"","sources":["../../src/orthography/apostrophe.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AACtD,OAAO,EAAE,gBAAgB,EAAE,MAAM,2BAA2B,CAAC;AAC7D,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAYtD;;;GAGG;AACH,MAAM,UAAU,GAAG,GAAG,CAAC;AAEvB;;;;;;;;;;;;;;GAcG;AACH,SAAS,aAAa,CAAC,SAA4B;IACjD,MAAM,QAAQ,GAAG,SAAS,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,cAAc,CAAC,QAAQ,CAAC,CAAC,CAAC;IACvE,MAAM,KAAK,GAAG,cAAc,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC;IAChD,OAAO,CACL,KAAK,EAAE,MAAM,KAAK,QAAQ,CAAC,MAAM;QACjC,KAAK,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,EAAE,EAAE,EAAE,CAAC,IAAI,KAAK,QAAQ,CAAC,EAAE,CAAC,CAAC,CACjD,CAAC;AACJ,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,QAAQ,CACtB,SAA4B,EAC5B,KAAK,GAAoB,QAAQ;IAEjC,OAAO,QAAQ,CAAC,SAAS,EAAE,KAAK,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;AAC7C,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,QAAQ,CACtB,SAA4B,EAC5B,KAAK,GAAoB,QAAQ;IAEjC,MAAM,MAAM,GAAG,SAAS,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,EAAE,EAAE,CAC5C,EAAE,GAAG,CAAC,IAAI,gBAAgB,CAAC,QAAQ,CAAC;QAClC,CAAC,CAAC,GAAG,UAAU,GAAG,QAAQ,EAAE;QAC5B,CAAC,CAAC,QAAQ,CACb,CAAC;IAEF,IACE,KAAK,KAAK,OAAO;QACjB,MAAM,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC,EACpD,CAAC;QACD,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,OAAO,KAAK,KAAK,UAAU,IAAI,aAAa,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,MAAM,CAAC;AAC/E,CAAC"}

package/dist/syllable/run-tables.d.ts

@@ -0,0 +1,45 @@
+import { type Syllable } from "./syllable.js";
+/**
+ * How long a written syllable can be, and so how far ahead a reading looks.
+ *
+ * The longest spelling in the inventory is `zhuang` at six letters, and the
+ * rest is what hangs off one: the r of 儿化, a tone digit or the raised form of
+ * it, `u:` where ü is typed that way, and a combining mark where the input
+ * arrives decomposed. Two of those cannot both be there, which leaves the
+ * longest piece worth reading two characters short of this.
+ */
+export declare const LONGEST_SYLLABLE: number;
+/**
+ * What the two passes over a run share.
+ *
+ * Both try the same pieces, and reading one costs a Unicode normalisation and
+ * a walk through the tables, so each piece is read once and remembered.
+ * `marksBefore` counts the tone marks in front of every position, which turns
+ * counting the marks on a piece into a subtraction, and `isSeparableAt` says
+ * of every position whether a syllable starting there would need an apostrophe
+ * before it.
+ */
+export interface RunTables {
+    readonly marksBefore: readonly number[];
+    readonly isSeparableAt: readonly boolean[];
+    readonly syllables: Map<string, Syllable | undefined>;
+    readonly wellMarked: Map<string, boolean>;
+    readonly attested: Map<string, boolean>;
+}
+/**
+ * The tables for one run, read off its characters in a single pass.
+ */
+export declare function tablesFor(run: string): RunTables;
+/**
+ * Read a piece of the run as a syllable, remembering what it read.
+ */
+export declare function readPiece(tables: RunTables, piece: string): Syllable | undefined;
+/**
+ * Whether a piece is a syllable of the language, remembering the answer.
+ */
+export declare function isPieceAttested(tables: RunTables, piece: string, syllable: Syllable): boolean;
+/**
+ * Whether a piece writes its tone mark in place, remembering the answer.
+ */
+export declare function isPieceWellMarked(tables: RunTables, piece: string, syllable: Syllable): boolean;
+//# sourceMappingURL=run-tables.d.ts.map

package/dist/syllable/run-tables.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"run-tables.d.ts","sourceRoot":"","sources":["../../src/syllable/run-tables.ts"],"names":[],"mappings":"AASA,OAAO,EAGL,KAAK,QAAQ,EAGd,MAAM,eAAe,CAAC;AAEvB;;;;;;;;GAQG;AACH,eAAO,MAAM,gBAAgB,QACgD,CAAC;AAiC9E;;;;;;;;;GASG;AACH,MAAM,WAAW,SAAS;IACxB,QAAQ,CAAC,WAAW,EAAE,SAAS,MAAM,EAAE,CAAC;IACxC,QAAQ,CAAC,aAAa,EAAE,SAAS,OAAO,EAAE,CAAC;IAC3C,QAAQ,CAAC,SAAS,EAAE,GAAG,CAAC,MAAM,EAAE,QAAQ,GAAG,SAAS,CAAC,CAAC;IACtD,QAAQ,CAAC,UAAU,EAAE,GAAG,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;IAC1C,QAAQ,CAAC,QAAQ,EAAE,GAAG,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;CACzC;AAED;;GAEG;AACH,wBAAgB,SAAS,CAAC,GAAG,EAAE,MAAM,GAAG,SAAS,CAiBhD;AAED;;GAEG;AACH,wBAAgB,SAAS,CACvB,MAAM,EAAE,SAAS,EACjB,KAAK,EAAE,MAAM,GACZ,QAAQ,GAAG,SAAS,CAQtB;AAED;;GAEG;AACH,wBAAgB,eAAe,CAC7B,MAAM,EAAE,SAAS,EACjB,KAAK,EAAE,MAAM,EACb,QAAQ,EAAE,QAAQ,GACjB,OAAO,CAQT;AAED;;GAEG;AACH,wBAAgB,iBAAiB,CAC/B,MAAM,EAAE,SAAS,EACjB,KAAK,EAAE,MAAM,EACb,QAAQ,EAAE,QAAQ,GACjB,OAAO,CAQT"}

package/dist/syllable/run-tables.js

@@ -0,0 +1,103 @@
+/**
+ * What one run's search reads once and then remembers.
+ *
+ * The questions `segment-run.ts` asks of a candidate piece, and how long a
+ * piece can be worth asking about at all.
+ */
+import { countToneMarks } from "../tone/tone-mark.js";
+import { ATTESTED_SYLLABLES, DICTIONARY_SYLLABLES } from "./inventory.js";
+import { isSeparableStart } from "./separation.js";
+import { normaliseUmlaut, readSyllable, writeSyllable, writeSyllableSpelling, } from "./syllable.js";
+/**
+ * How long a written syllable can be, and so how far ahead a reading looks.
+ *
+ * The longest spelling in the inventory is `zhuang` at six letters, and the
+ * rest is what hangs off one: the r of 儿化, a tone digit or the raised form of
+ * it, `u:` where ü is typed that way, and a combining mark where the input
+ * arrives decomposed. Two of those cannot both be there, which leaves the
+ * longest piece worth reading two characters short of this.
+ */
+export const LONGEST_SYLLABLE = Math.max(...[...ATTESTED_SYLLABLES].map((spelling) => spelling.length)) + 4;
+/**
+ * Whether a syllable is one Mandarin actually uses.
+ *
+ * {@link readSyllable} asks only whether an initial and a final can be put
+ * together, which is a wider question: `tia` passes it and is not a syllable of
+ * the language. The 儿化 suffix is set aside before the check, since the
+ * inventory lists base syllables and `wanr` is `wan` with an r on it.
+ */
+function isAttestedSyllable(syllable) {
+    return DICTIONARY_SYLLABLES.has(writeSyllableSpelling({ ...syllable, erhua: false }));
+}
+/**
+ * Whether a spelling writes its tone mark where the orthography puts it.
+ *
+ * `xīa` reads as x + ia in the first tone, which pinyin writes `xiā`, so the
+ * mark sits a letter early. Read on its own that is a slip worth tolerating,
+ * and {@link readSyllable} tolerates it. Inside a longer run it is also what
+ * two syllables written without a boundary look like, `xīan1` being 西安 half
+ * typed, and the search weighs the two readings by what they cost.
+ */
+function isWellMarked(spelling, syllable) {
+    return (countToneMarks(spelling) === 0 ||
+        writeSyllable(syllable) ===
+            normaliseUmlaut(spelling).normalize("NFC").toLowerCase());
+}
+/**
+ * The tables for one run, read off its characters in a single pass.
+ */
+export function tablesFor(run) {
+    const marksBefore = [0];
+    const isSeparableAt = [];
+    let marks = 0;
+    for (let at = 0; at < run.length; at++) {
+        const character = run.charAt(at);
+        marks += countToneMarks(character);
+        marksBefore.push(marks);
+        isSeparableAt.push(isSeparableStart(character));
+    }
+    return {
+        marksBefore,
+        isSeparableAt,
+        syllables: new Map(),
+        wellMarked: new Map(),
+        attested: new Map(),
+    };
+}
+/**
+ * Read a piece of the run as a syllable, remembering what it read.
+ */
+export function readPiece(tables, piece) {
+    const read = tables.syllables.get(piece);
+    if (read !== undefined || tables.syllables.has(piece)) {
+        return read;
+    }
+    const syllable = readSyllable(piece);
+    tables.syllables.set(piece, syllable);
+    return syllable;
+}
+/**
+ * Whether a piece is a syllable of the language, remembering the answer.
+ */
+export function isPieceAttested(tables, piece, syllable) {
+    const known = tables.attested.get(piece);
+    if (known !== undefined) {
+        return known;
+    }
+    const attested = isAttestedSyllable(syllable);
+    tables.attested.set(piece, attested);
+    return attested;
+}
+/**
+ * Whether a piece writes its tone mark in place, remembering the answer.
+ */
+export function isPieceWellMarked(tables, piece, syllable) {
+    const known = tables.wellMarked.get(piece);
+    if (known !== undefined) {
+        return known;
+    }
+    const wellMarked = isWellMarked(piece, syllable);
+    tables.wellMarked.set(piece, wellMarked);
+    return wellMarked;
+}
+//# sourceMappingURL=run-tables.js.map

package/dist/syllable/run-tables.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"run-tables.js","sourceRoot":"","sources":["../../src/syllable/run-tables.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AACtD,OAAO,EAAE,kBAAkB,EAAE,oBAAoB,EAAE,MAAM,gBAAgB,CAAC;AAC1E,OAAO,EAAE,gBAAgB,EAAE,MAAM,iBAAiB,CAAC;AACnD,OAAO,EACL,eAAe,EACf,YAAY,EAEZ,aAAa,EACb,qBAAqB,GACtB,MAAM,eAAe,CAAC;AAEvB;;;;;;;;GAQG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAC3B,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,GAAG,kBAAkB,CAAC,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC;AAE9E;;;;;;;GAOG;AACH,SAAS,kBAAkB,CAAC,QAAkB;IAC5C,OAAO,oBAAoB,CAAC,GAAG,CAC7B,qBAAqB,CAAC,EAAE,GAAG,QAAQ,EAAE,KAAK,EAAE,KAAK,EAAE,CAAC,CACrD,CAAC;AACJ,CAAC;AAED;;;;;;;;GAQG;AACH,SAAS,YAAY,CAAC,QAAgB,EAAE,QAAkB;IACxD,OAAO,CACL,cAAc,CAAC,QAAQ,CAAC,KAAK,CAAC;QAC9B,aAAa,CAAC,QAAQ,CAAC;YACrB,eAAe,CAAC,QAAQ,CAAC,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC,WAAW,EAAE,CAC3D,CAAC;AACJ,CAAC;AAoBD;;GAEG;AACH,MAAM,UAAU,SAAS,CAAC,GAAW;IACnC,MAAM,WAAW,GAAa,CAAC,CAAC,CAAC,CAAC;IAClC,MAAM,aAAa,GAAc,EAAE,CAAC;IACpC,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,KAAK,IAAI,EAAE,GAAG,CAAC,EAAE,EAAE,GAAG,GAAG,CAAC,MAAM,EAAE,EAAE,EAAE,EAAE,CAAC;QACvC,MAAM,SAAS,GAAG,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC;QACjC,KAAK,IAAI,cAAc,CAAC,SAAS,CAAC,CAAC;QACnC,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;QACxB,aAAa,CAAC,IAAI,CAAC,gBAAgB,CAAC,SAAS,CAAC,CAAC,CAAC;IAClD,CAAC;IACD,OAAO;QACL,WAAW;QACX,aAAa;QACb,SAAS,EAAE,IAAI,GAAG,EAAE;QACpB,UAAU,EAAE,IAAI,GAAG,EAAE;QACrB,QAAQ,EAAE,IAAI,GAAG,EAAE;KACpB,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,SAAS,CACvB,MAAiB,EACjB,KAAa;IAEb,MAAM,IAAI,GAAG,MAAM,CAAC,SAAS,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;IACzC,IAAI,IAAI,KAAK,SAAS,IAAI,MAAM,CAAC,SAAS,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC;QACtD,OAAO,IAAI,CAAC;IACd,CAAC;IACD,MAAM,QAAQ,GAAG,YAAY,CAAC,KAAK,CAAC,CAAC;IACrC,MAAM,CAAC,SAAS,CAAC,GAAG,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC;IACtC,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,eAAe,CAC7B,MAAiB,EACjB,KAAa,EACb,QAAkB;IAElB,MAAM,KAAK,GAAG,MAAM,CAAC,QAAQ,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;IACzC,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;QACxB,OAAO,KAAK,CAAC;IACf,CAAC;IACD,MAAM,QAAQ,GAAG,kBAAkB,CAAC,QAAQ,CAAC,CAAC;IAC9C,MAAM,CAAC,QAAQ,CAAC,GAAG,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC;IACrC,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,iBAAiB,CAC/B,MAAiB,EACjB,KAAa,EACb,QAAkB;IAElB,MAAM,KAAK,GAAG,MAAM,CAAC,UAAU,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;IAC3C,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;QACxB,OAAO,KAAK,CAAC;IACf,CAAC;IACD,MAAM,UAAU,GAAG,YAAY,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC;IACjD,MAAM,CAAC,UAAU,CAAC,GAAG,CAAC,KAAK,EAAE,UAAU,CAAC,CAAC;IACzC,OAAO,UAAU,CAAC;AACpB,CAAC"}

package/dist/syllable/segment-run.d.ts

@@ -0,0 +1,10 @@
+/**
+ * Segment a run, honouring the apostrophe rule where it can be honoured.
+ *
+ * The separated reading wins every tie, so the rule holds wherever it can. It
+ * gives way only to a cheaper reading, which is a reading leaving fewer tone
+ * marks out of place: `xīan1` reads as `xīa` + `n1` under the rule, and as `xī`
+ * + `an1` without it, and the mark on `xī` is what says the second is meant.
+ */
+export declare function segmentRun(run: string): readonly string[] | undefined;
+//# sourceMappingURL=segment-run.d.ts.map

package/dist/syllable/segment-run.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"segment-run.d.ts","sourceRoot":"","sources":["../../src/syllable/segment-run.ts"],"names":[],"mappings":"AAoIA;;;;;;;GAOG;AACH,wBAAgB,UAAU,CAAC,GAAG,EAAE,MAAM,GAAG,SAAS,MAAM,EAAE,GAAG,SAAS,CAgBrE"}

package/dist/syllable/segment-run.js

@@ -0,0 +1,116 @@
+/**
+ * Reading a run of pinyin with no boundaries written in it.
+ *
+ * `split.ts` deals with the boundaries a word does write, the apostrophes and
+ * the hyphens, and hands each run between them here. What happens here is a
+ * search: every way of cutting the run into syllables, scored, with the
+ * cheapest kept. This is the syllables inside one word, and not
+ * `decode/segment.ts`, which is the words inside a text.
+ */
+import { isPieceAttested, isPieceWellMarked, LONGEST_SYLLABLE, readPiece, tablesFor, } from "./run-tables.js";
+/**
+ * Read a run of pinyin with no explicit boundaries as syllables, as cheaply as
+ * it can be read.
+ *
+ * Works longest-first, which is what the orthography assumes: an apostrophe is
+ * required exactly where the longest-first reading would be wrong, so `xian` is
+ * one syllable and 西安 must be written `Xī'ān` to be read as two. Cost breaks
+ * that habit only where a tone mark says the boundary lies elsewhere, and ties
+ * go to the longer piece, so a run with every mark in place reads as it always
+ * did.
+ *
+ * `isSeparated` carries the other half of the apostrophe rule. Longest-first
+ * alone reads `guórén` as `guór'én` and `huángěi` as `huáng'ěi`, both of which
+ * the missing apostrophe rules out — so a run is first read with a/o/e barred
+ * from starting any syllable but the first, and only falls back to plain
+ * longest-first where that finds nothing or costs more. The fallback is what
+ * keeps input tolerant: `hǎiōu` is written wrong, and still reads as two
+ * syllables.
+ *
+ * The separated pass also holds out for real syllables rather than merely
+ * well-formed ones, because barring a vowel otherwise pushes it into nonsense:
+ * `Tiānānmén` would come apart as `tiā nān mén` where `tiā` is not a syllable
+ * of the language at all.
+ *
+ * Read back to front, one suffix at a time, since without that a long run of
+ * ambiguous syllables would backtrack exponentially.
+ *
+ * Reports whether either of the separated pass's own rules ever ruled a piece
+ * out, which is what tells {@link segmentRun} that the fallback has nothing
+ * new to find.
+ */
+function readRun(run, tables, isSeparated) {
+    const readings = Array.from({
+        length: run.length + 1,
+    });
+    readings[run.length] = { syllables: [], cost: 0 };
+    let wasConstrained = false;
+    for (let at = run.length - 1; at >= 0; at--) {
+        let best;
+        const furthest = Math.min(run.length, at + LONGEST_SYLLABLE);
+        for (let end = furthest; end > at; end--) {
+            const rest = readings[end];
+            if (rest === undefined) {
+                continue;
+            }
+            // A piece costs a point at least, so a candidate that cannot come in
+            // under the best reading so far is not worth reading at all. Ties are
+            // pruned with the rest, which is what leaves them to the longer piece.
+            if (best !== undefined && rest.cost + 1 >= best.cost) {
+                continue;
+            }
+            // A syllable is written with one tone, so a second mark is a second
+            // syllable: `dìèr` is 第二 with the apostrophe left out.
+            /* c8 ignore next 2 -- unreachable: both indices are inside the run */
+            const marks = (tables.marksBefore[end] ?? 0) - (tables.marksBefore[at] ?? 0);
+            if (marks > 1) {
+                continue;
+            }
+            if (isSeparated && tables.isSeparableAt[end] === true) {
+                wasConstrained = true;
+                continue;
+            }
+            const piece = run.slice(at, end);
+            const syllable = readPiece(tables, piece);
+            if (syllable === undefined) {
+                continue;
+            }
+            if (isSeparated && !isPieceAttested(tables, piece, syllable)) {
+                wasConstrained = true;
+                continue;
+            }
+            const cost = rest.cost +
+                (marks === 0 || isPieceWellMarked(tables, piece, syllable) ? 1 : 2);
+            if (best === undefined || cost < best.cost) {
+                best = { syllables: [piece, ...rest.syllables], cost };
+            }
+        }
+        readings[at] = best;
+    }
+    return { reading: readings[0], wasConstrained };
+}
+/**
+ * Segment a run, honouring the apostrophe rule where it can be honoured.
+ *
+ * The separated reading wins every tie, so the rule holds wherever it can. It
+ * gives way only to a cheaper reading, which is a reading leaving fewer tone
+ * marks out of place: `xīan1` reads as `xīa` + `n1` under the rule, and as `xī`
+ * + `an1` without it, and the mark on `xī` is what says the second is meant.
+ */
+export function segmentRun(run) {
+    const tables = tablesFor(run);
+    const separated = readRun(run, tables, true);
+    // Where neither of the separated pass's rules ever ruled a piece out, the two
+    // passes were choosing between the same readings and reached the same one.
+    if (separated.reading !== undefined && !separated.wasConstrained) {
+        return separated.reading.syllables;
+    }
+    const loose = readRun(run, tables, false).reading;
+    if (separated.reading === undefined) {
+        return loose?.syllables;
+    }
+    return loose !== undefined && loose.cost < separated.reading.cost
+        ? loose.syllables
+        : separated.reading.syllables;
+}
+//# sourceMappingURL=segment-run.js.map

package/dist/syllable/segment-run.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"segment-run.js","sourceRoot":"","sources":["../../src/syllable/segment-run.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AACH,OAAO,EACL,eAAe,EACf,iBAAiB,EACjB,gBAAgB,EAChB,SAAS,EAET,SAAS,GACV,MAAM,iBAAiB,CAAC;AA0BzB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA8BG;AACH,SAAS,OAAO,CACd,GAAW,EACX,MAAiB,EACjB,WAAoB;IAEpB,MAAM,QAAQ,GAAG,KAAK,CAAC,IAAI,CAA2B;QACpD,MAAM,EAAE,GAAG,CAAC,MAAM,GAAG,CAAC;KACvB,CAAC,CAAC;IACH,QAAQ,CAAC,GAAG,CAAC,MAAM,CAAC,GAAG,EAAE,SAAS,EAAE,EAAE,EAAE,IAAI,EAAE,CAAC,EAAE,CAAC;IAClD,IAAI,cAAc,GAAG,KAAK,CAAC;IAE3B,KAAK,IAAI,EAAE,GAAG,GAAG,CAAC,MAAM,GAAG,CAAC,EAAE,EAAE,IAAI,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC;QAC5C,IAAI,IAA8B,CAAC;QACnC,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,MAAM,EAAE,EAAE,GAAG,gBAAgB,CAAC,CAAC;QAC7D,KAAK,IAAI,GAAG,GAAG,QAAQ,EAAE,GAAG,GAAG,EAAE,EAAE,GAAG,EAAE,EAAE,CAAC;YACzC,MAAM,IAAI,GAAG,QAAQ,CAAC,GAAG,CAAC,CAAC;YAC3B,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;gBACvB,SAAS;YACX,CAAC;YACD,qEAAqE;YACrE,sEAAsE;YACtE,uEAAuE;YACvE,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,CAAC,IAAI,GAAG,CAAC,IAAI,IAAI,CAAC,IAAI,EAAE,CAAC;gBACrD,SAAS;YACX,CAAC;YACD,oEAAoE;YACpE,uDAAuD;YACvD,sEAAsE;YACtE,MAAM,KAAK,GACT,CAAC,MAAM,CAAC,WAAW,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,MAAM,CAAC,WAAW,CAAC,EAAE,CAAC,IAAI,CAAC,CAAC,CAAC;YACjE,IAAI,KAAK,GAAG,CAAC,EAAE,CAAC;gBACd,SAAS;YACX,CAAC;YACD,IAAI,WAAW,IAAI,MAAM,CAAC,aAAa,CAAC,GAAG,CAAC,KAAK,IAAI,EAAE,CAAC;gBACtD,cAAc,GAAG,IAAI,CAAC;gBACtB,SAAS;YACX,CAAC;YACD,MAAM,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC,EAAE,EAAE,GAAG,CAAC,CAAC;YACjC,MAAM,QAAQ,GAAG,SAAS,CAAC,MAAM,EAAE,KAAK,CAAC,CAAC;YAC1C,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;gBAC3B,SAAS;YACX,CAAC;YACD,IAAI,WAAW,IAAI,CAAC,eAAe,CAAC,MAAM,EAAE,KAAK,EAAE,QAAQ,CAAC,EAAE,CAAC;gBAC7D,cAAc,GAAG,IAAI,CAAC;gBACtB,SAAS;YACX,CAAC;YACD,MAAM,IAAI,GACR,IAAI,CAAC,IAAI;gBACT,CAAC,KAAK,KAAK,CAAC,IAAI,iBAAiB,CAAC,MAAM,EAAE,KAAK,EAAE,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YACtE,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;gBAC3C,IAAI,GAAG,EAAE,SAAS,EAAE,CAAC,KAAK,EAAE,GAAG,IAAI,CAAC,SAAS,CAAC,EAAE,IAAI,EAAE,CAAC;YACzD,CAAC;QACH,CAAC;QACD,QAAQ,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC;IACtB,CAAC;IAED,OAAO,EAAE,OAAO,EAAE,QAAQ,CAAC,CAAC,CAAC,EAAE,cAAc,EAAE,CAAC;AAClD,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,UAAU,CAAC,GAAW;IACpC,MAAM,MAAM,GAAG,SAAS,CAAC,GAAG,CAAC,CAAC;IAC9B,MAAM,SAAS,GAAG,OAAO,CAAC,GAAG,EAAE,MAAM,EAAE,IAAI,CAAC,CAAC;IAC7C,8EAA8E;IAC9E,2EAA2E;IAC3E,IAAI,SAAS,CAAC,OAAO,KAAK,SAAS,IAAI,CAAC,SAAS,CAAC,cAAc,EAAE,CAAC;QACjE,OAAO,SAAS,CAAC,OAAO,CAAC,SAAS,CAAC;IACrC,CAAC;IAED,MAAM,KAAK,GAAG,OAAO,CAAC,GAAG,EAAE,MAAM,EAAE,KAAK,CAAC,CAAC,OAAO,CAAC;IAClD,IAAI,SAAS,CAAC,OAAO,KAAK,SAAS,EAAE,CAAC;QACpC,OAAO,KAAK,EAAE,SAAS,CAAC;IAC1B,CAAC;IACD,OAAO,KAAK,KAAK,SAAS,IAAI,KAAK,CAAC,IAAI,GAAG,SAAS,CAAC,OAAO,CAAC,IAAI;QAC/D,CAAC,CAAC,KAAK,CAAC,SAAS;QACjB,CAAC,CAAC,SAAS,CAAC,OAAO,CAAC,SAAS,CAAC;AAClC,CAAC"}

package/dist/syllable/split.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"split.d.ts","sourceRoot":"","sources":["../../src/syllable/split.ts"],"names":[],"mappings":"AAEA,OAAO,EAGL,KAAK,QAAQ,EAEd,MAAM,eAAe,CAAC;AA2FvB;;;;;;GAMG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,GAAG,SAAS,MAAM,EAAE,GAAG,SAAS,CAe1E;AAED;;;;GAIG;AACH,wBAAgB,QAAQ,CAAC,IAAI,EAAE,MAAM,GAAG,SAAS,QAAQ,EAAE,GAAG,SAAS,CAgBtE"}
+{"version":3,"file":"split.d.ts","sourceRoot":"","sources":["../../src/syllable/split.ts"],"names":[],"mappings":"AACA,OAAO,EAAgB,KAAK,QAAQ,EAAE,MAAM,eAAe,CAAC;AAU5D;;;;;;GAMG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,GAAG,SAAS,MAAM,EAAE,GAAG,SAAS,CAe1E;AAED;;;;GAIG;AACH,wBAAgB,QAAQ,CAAC,IAAI,EAAE,MAAM,GAAG,SAAS,QAAQ,EAAE,GAAG,SAAS,CAgBtE"}

package/dist/syllable/split.js

@@ -1,6 +1,5 @@
-import { DICTIONARY_SYLLABLES } from "./inventory.js";
-import { isSeparableStart } from "./separation.js";
-import { isSyllable, readSyllable, writeSyllableSpelling, } from "./syllable.js";
+import { segmentRun } from "./segment-run.js";
+import { readSyllable } from "./syllable.js";
 /**
  * Characters that mark a syllable boundary explicitly.
  *
@@ -8,74 +7,6 @@ import { isSyllable, readSyllable, writeSyllableSpelling, } from "./syllable.js"
  * reduplicated word or a 成语 written 2+2.
  */
 const SYLLABLE_SEPARATORS = /['’‘-]/u;
-/**
- * Whether a spelling is a syllable Mandarin actually uses.
- *
- * {@link isSyllable} asks only whether an initial and a final can be put
- * together, which is a wider question: `tia` passes it and is not a syllable of
- * the language. The 儿化 suffix is set aside before the check, since the
- * inventory lists base syllables and `wanr` is `wan` with an r on it.
- */
-function isAttestedSyllable(text) {
-    const syllable = readSyllable(text);
-    return (syllable !== undefined &&
-        DICTIONARY_SYLLABLES.has(writeSyllableSpelling({ ...syllable, erhua: false })));
-}
-/**
- * Split a run of pinyin with no explicit boundaries into syllables.
- *
- * Works longest-first, which is what the orthography assumes: an apostrophe is
- * required exactly where the longest-first reading would be wrong, so `xian` is
- * one syllable and 西安 must be written `Xī'ān` to be read as two.
- *
- * `isSeparated` carries the other half of that rule. Longest-first alone reads
- * `guórén` as `guór'én` and `huángěi` as `huáng'ěi`, both of which the missing
- * apostrophe rules out — so a run is first segmented with a/o/e barred from
- * starting any syllable but the first, and only falls back to plain
- * longest-first when that finds nothing. The fallback is what keeps input
- * tolerant: `hǎiōu` is written wrong, and still reads as two syllables.
- *
- * The strict pass also holds out for real syllables rather than merely
- * well-formed ones, because barring a vowel otherwise pushes it into nonsense:
- * `Tiānānmén` would come apart as `tiā nān mén` where `tiā` is not a syllable
- * of the language at all.
- *
- * Memoised on the suffix being segmented, since without it a long run of
- * ambiguous syllables would backtrack exponentially.
- */
-function segmentRun(run, memo, isSeparated) {
-    if (run === "") {
-        return [];
-    }
-    const cached = memo.get(run);
-    if (cached !== undefined || memo.has(run)) {
-        return cached;
-    }
-    let found;
-    for (let length = run.length; length >= 1; length--) {
-        const head = run.slice(0, length);
-        if (isSeparated ? !isAttestedSyllable(head) : !isSyllable(head)) {
-            continue;
-        }
-        const tail = run.slice(length);
-        if (isSeparated && isSeparableStart(tail)) {
-            continue;
-        }
-        const rest = segmentRun(tail, memo, isSeparated);
-        if (rest !== undefined) {
-            found = [head, ...rest];
-            break;
-        }
-    }
-    memo.set(run, found);
-    return found;
-}
-/**
- * Segment a run, honouring the apostrophe rule where it can be honoured.
- */
-function segment(run) {
-    return segmentRun(run, new Map(), true) ?? segmentRun(run, new Map(), false);
-}
 /**
  * Split a written pinyin word into its syllables.
  *
@@ -89,7 +20,7 @@ export function splitSyllables(word) {
         if (run === "") {
             continue;
         }
-        const segmented = segment(run);
+        const segmented = segmentRun(run);
         if (segmented === undefined) {
             return undefined;
         }

package/dist/syllable/split.js.map

@@ -1 +1 @@
-{"version":3,"file":"split.js","sourceRoot":"","sources":["../../src/syllable/split.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,oBAAoB,EAAE,MAAM,gBAAgB,CAAC;AACtD,OAAO,EAAE,gBAAgB,EAAE,MAAM,iBAAiB,CAAC;AACnD,OAAO,EACL,UAAU,EACV,YAAY,EAEZ,qBAAqB,GACtB,MAAM,eAAe,CAAC;AAEvB;;;;;GAKG;AACH,MAAM,mBAAmB,GAAG,SAAS,CAAC;AAEtC;;;;;;;GAOG;AACH,SAAS,kBAAkB,CAAC,IAAY;IACtC,MAAM,QAAQ,GAAG,YAAY,CAAC,IAAI,CAAC,CAAC;IACpC,OAAO,CACL,QAAQ,KAAK,SAAS;QACtB,oBAAoB,CAAC,GAAG,CACtB,qBAAqB,CAAC,EAAE,GAAG,QAAQ,EAAE,KAAK,EAAE,KAAK,EAAE,CAAC,CACrD,CACF,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;GAqBG;AACH,SAAS,UAAU,CACjB,GAAW,EACX,IAAgD,EAChD,WAAoB;IAEpB,IAAI,GAAG,KAAK,EAAE,EAAE,CAAC;QACf,OAAO,EAAE,CAAC;IACZ,CAAC;IACD,MAAM,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;IAC7B,IAAI,MAAM,KAAK,SAAS,IAAI,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC;QAC1C,OAAO,MAAM,CAAC;IAChB,CAAC;IAED,IAAI,KAAoC,CAAC;IACzC,KAAK,IAAI,MAAM,GAAG,GAAG,CAAC,MAAM,EAAE,MAAM,IAAI,CAAC,EAAE,MAAM,EAAE,EAAE,CAAC;QACpD,MAAM,IAAI,GAAG,GAAG,CAAC,KAAK,CAAC,CAAC,EAAE,MAAM,CAAC,CAAC;QAClC,IAAI,WAAW,CAAC,CAAC,CAAC,CAAC,kBAAkB,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,IAAI,CAAC,EAAE,CAAC;YAChE,SAAS;QACX,CAAC;QACD,MAAM,IAAI,GAAG,GAAG,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;QAC/B,IAAI,WAAW,IAAI,gBAAgB,CAAC,IAAI,CAAC,EAAE,CAAC;YAC1C,SAAS;QACX,CAAC;QACD,MAAM,IAAI,GAAG,UAAU,CAAC,IAAI,EAAE,IAAI,EAAE,WAAW,CAAC,CAAC;QACjD,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;YACvB,KAAK,GAAG,CAAC,IAAI,EAAE,GAAG,IAAI,CAAC,CAAC;YACxB,MAAM;QACR,CAAC;IACH,CAAC;IAED,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,KAAK,CAAC,CAAC;IACrB,OAAO,KAAK,CAAC;AACf,CAAC;AAED;;GAEG;AACH,SAAS,OAAO,CAAC,GAAW;IAC1B,OAAO,UAAU,CAAC,GAAG,EAAE,IAAI,GAAG,EAAE,EAAE,IAAI,CAAC,IAAI,UAAU,CAAC,GAAG,EAAE,IAAI,GAAG,EAAE,EAAE,KAAK,CAAC,CAAC;AAC/E,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,cAAc,CAAC,IAAY;IACzC,MAAM,SAAS,GAAa,EAAE,CAAC;IAE/B,KAAK,MAAM,GAAG,IAAI,IAAI,CAAC,KAAK,CAAC,mBAAmB,CAAC,EAAE,CAAC;QAClD,IAAI,GAAG,KAAK,EAAE,EAAE,CAAC;YACf,SAAS;QACX,CAAC;QACD,MAAM,SAAS,GAAG,OAAO,CAAC,GAAG,CAAC,CAAC;QAC/B,IAAI,SAAS,KAAK,SAAS,EAAE,CAAC;YAC5B,OAAO,SAAS,CAAC;QACnB,CAAC;QACD,SAAS,CAAC,IAAI,CAAC,GAAG,SAAS,CAAC,CAAC;IAC/B,CAAC;IAED,OAAO,SAAS,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC;AACtD,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,QAAQ,CAAC,IAAY;IACnC,MAAM,SAAS,GAAG,cAAc,CAAC,IAAI,CAAC,CAAC;IACvC,IAAI,SAAS,KAAK,SAAS,EAAE,CAAC;QAC5B,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,MAAM,SAAS,GAAe,EAAE,CAAC;IACjC,KAAK,MAAM,QAAQ,IAAI,SAAS,EAAE,CAAC;QACjC,MAAM,QAAQ,GAAG,YAAY,CAAC,QAAQ,CAAC,CAAC;QACxC,0EAA0E;QAC1E,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;YAC3B,OAAO,SAAS,CAAC;QACnB,CAAC;QACD,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;IAC3B,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC"}
+{"version":3,"file":"split.js","sourceRoot":"","sources":["../../src/syllable/split.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,UAAU,EAAE,MAAM,kBAAkB,CAAC;AAC9C,OAAO,EAAE,YAAY,EAAiB,MAAM,eAAe,CAAC;AAE5D;;;;;GAKG;AACH,MAAM,mBAAmB,GAAG,SAAS,CAAC;AAEtC;;;;;;GAMG;AACH,MAAM,UAAU,cAAc,CAAC,IAAY;IACzC,MAAM,SAAS,GAAa,EAAE,CAAC;IAE/B,KAAK,MAAM,GAAG,IAAI,IAAI,CAAC,KAAK,CAAC,mBAAmB,CAAC,EAAE,CAAC;QAClD,IAAI,GAAG,KAAK,EAAE,EAAE,CAAC;YACf,SAAS;QACX,CAAC;QACD,MAAM,SAAS,GAAG,UAAU,CAAC,GAAG,CAAC,CAAC;QAClC,IAAI,SAAS,KAAK,SAAS,EAAE,CAAC;YAC5B,OAAO,SAAS,CAAC;QACnB,CAAC;QACD,SAAS,CAAC,IAAI,CAAC,GAAG,SAAS,CAAC,CAAC;IAC/B,CAAC;IAED,OAAO,SAAS,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC;AACtD,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,QAAQ,CAAC,IAAY;IACnC,MAAM,SAAS,GAAG,cAAc,CAAC,IAAI,CAAC,CAAC;IACvC,IAAI,SAAS,KAAK,SAAS,EAAE,CAAC;QAC5B,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,MAAM,SAAS,GAAe,EAAE,CAAC;IACjC,KAAK,MAAM,QAAQ,IAAI,SAAS,EAAE,CAAC;QACjC,MAAM,QAAQ,GAAG,YAAY,CAAC,QAAQ,CAAC,CAAC;QACxC,0EAA0E;QAC1E,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;YAC3B,OAAO,SAAS,CAAC;QACnB,CAAC;QACD,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;IAC3B,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC"}

package/dist/tone/tone-mark.d.ts

@@ -15,6 +15,14 @@ export declare function toneDiacritic(tone: Tone): string;
  * The diaeresis of ü is preserved, so ǚ becomes ü rather than u.
  */
 export declare function stripToneMarks(text: string): string;
+/**
+ * How many tone diacritics a text carries.
+ *
+ * A syllable is written with one tone, so a second mark on a run of letters is
+ * a second syllable inside it: `dìèr` is 第二 with the apostrophe left out
+ * rather than any syllable of the language.
+ */
+export declare function countToneMarks(text: string): number;
 /**
  * How each tone is written raised.
  *

package/dist/tone/tone-mark.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"tone-mark.d.ts","sourceRoot":"","sources":["../../src/tone/tone-mark.ts"],"names":[],"mappings":"AAAA,OAAO,EAAgB,KAAK,IAAI,EAAE,MAAM,WAAW,CAAC;AA2CpD;;;;;;;;GAQG;AACH,wBAAgB,aAAa,CAAC,IAAI,EAAE,IAAI,GAAG,MAAM,CAEhD;AAED;;;;GAIG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAKnD;AAED;;;;;;;GAOG;AACH,eAAO,MAAM,iBAAiB,EAAE,WAAW,CAAC,IAAI,EAAE,MAAM,CAMtD,CAAC;AAiBH;;GAEG;AACH,wBAAgB,oBAAoB,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAMzD;AAED;;;;;;;;GAQG;AACH,wBAAgB,aAAa,CAAC,IAAI,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,CAQ5D;AA8BD;;;;;;;GAOG;AACH,wBAAgB,aAAa,CAC3B,QAAQ,EAAE,MAAM,EAChB,IAAI,EAAE,IAAI,GAAG,SAAS,GACrB,MAAM,CAoBR"}
+{"version":3,"file":"tone-mark.d.ts","sourceRoot":"","sources":["../../src/tone/tone-mark.ts"],"names":[],"mappings":"AAAA,OAAO,EAAgB,KAAK,IAAI,EAAE,MAAM,WAAW,CAAC;AA2CpD;;;;;;;;GAQG;AACH,wBAAgB,aAAa,CAAC,IAAI,EAAE,IAAI,GAAG,MAAM,CAEhD;AAED;;;;GAIG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAKnD;AAED;;;;;;GAMG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAEnD;AAED;;;;;;;GAOG;AACH,eAAO,MAAM,iBAAiB,EAAE,WAAW,CAAC,IAAI,EAAE,MAAM,CAMtD,CAAC;AAiBH;;GAEG;AACH,wBAAgB,oBAAoB,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAMzD;AAED;;;;;;;;GAQG;AACH,wBAAgB,aAAa,CAAC,IAAI,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,CAQ5D;AA8BD;;;;;;;GAOG;AACH,wBAAgB,aAAa,CAC3B,QAAQ,EAAE,MAAM,EAChB,IAAI,EAAE,IAAI,GAAG,SAAS,GACrB,MAAM,CAoBR"}

package/dist/tone/tone-mark.js

@@ -58,6 +58,16 @@ export function stripToneMarks(text) {
         .replaceAll(toneDiacriticPattern, "")
         .normalize("NFC");
 }
+/**
+ * How many tone diacritics a text carries.
+ *
+ * A syllable is written with one tone, so a second mark on a run of letters is
+ * a second syllable inside it: `dìèr` is 第二 with the apostrophe left out
+ * rather than any syllable of the language.
+ */
+export function countToneMarks(text) {
+    return text.normalize("NFD").match(toneDiacriticPattern)?.length ?? 0;
+}
 /**
  * How each tone is written raised.
  *

package/dist/tone/tone-mark.js.map

@@ -1 +1 @@
-{"version":3,"file":"tone-mark.js","sourceRoot":"","sources":["../../src/tone/tone-mark.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAa,MAAM,WAAW,CAAC;AAEpD;;;;;;;;GAQG;AACH,MAAM,cAAc,GAAG;IACrB,GAAG,EAAE,2BAA2B;IAChC,GAAG,EAAE,iCAAiC;IACtC,GAAG,EAAE,0BAA0B;IAC/B,GAAG,EAAE,iCAAiC;CAC9B,CAAC;AAEX;;;;;GAKG;AACH,MAAM,oBAAoB,GAAG,UAAU,CAAC;AAExC;;GAEG;AACH,MAAM,mBAAmB,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,CAAU,CAAC;AAErD;;;GAGG;AACH,MAAM,mBAAmB,GAAG,MAAM,CAAC;AAEnC;;;GAGG;AACH,MAAM,wBAAwB,GAAG,OAAO,CAAC;AAEzC;;;;;;;;GAQG;AACH,MAAM,UAAU,aAAa,CAAC,IAAU;IACtC,OAAO,cAAc,CAAC,IAAI,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;AACxC,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,cAAc,CAAC,IAAY;IACzC,OAAO,IAAI;SACR,SAAS,CAAC,KAAK,CAAC;SAChB,UAAU,CAAC,oBAAoB,EAAE,EAAE,CAAC;SACpC,SAAS,CAAC,KAAK,CAAC,CAAC;AACtB,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAA8B,IAAI,GAAG,CAAC;IAClE,CAAC,CAAC,EAAE,GAAG,CAAC;IACR,CAAC,CAAC,EAAE,GAAG,CAAC;IACR,CAAC,CAAC,EAAE,GAAG,CAAC;IACR,CAAC,CAAC,EAAE,GAAG,CAAC;IACR,CAAC,YAAY,EAAE,GAAG,CAAC;CACpB,CAAC,CAAC;AAEH;;;;;GAKG;AACH,MAAM,iBAAiB,GAAG,IAAI,GAAG,CAAiB;IAChD,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;CACX,CAAC,CAAC;AAEH;;GAEG;AACH,MAAM,UAAU,oBAAoB,CAAC,IAAY;IAC/C,OAAO,IAAI,CAAC,UAAU,CACpB,YAAY;IACZ,mEAAmE;IACnE,CAAC,KAAK,EAAE,EAAE,CAAC,iBAAiB,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,KAAK,CACjD,CAAC;AACJ,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,aAAa,CAAC,IAAY;IACxC,MAAM,UAAU,GAAG,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;IACzC,KAAK,MAAM,CAAC,KAAK,EAAE,SAAS,CAAC,IAAI,cAAc,CAAC,OAAO,EAAE,EAAE,CAAC;QAC1D,IAAI,UAAU,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;YACnC,OAAO,CAAC,KAAK,GAAG,CAAC,CAAS,CAAC;QAC7B,CAAC;IACH,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,mBAAmB,CAAC,QAAgB;IAC3C,MAAM,SAAS,GAAG,QAAQ,CAAC,WAAW,EAAE,CAAC;IAEzC,KAAK,MAAM,KAAK,IAAI,mBAAmB,EAAE,CAAC;QACxC,MAAM,KAAK,GAAG,SAAS,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC;QACvC,IAAI,KAAK,KAAK,CAAC,CAAC,EAAE,CAAC;YACjB,OAAO,KAAK,CAAC;QACf,CAAC;IACH,CAAC;IAED,KAAK,IAAI,KAAK,GAAG,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,KAAK,IAAI,CAAC,EAAE,KAAK,EAAE,EAAE,CAAC;QAC3D,MAAM,SAAS,GAAG,SAAS,CAAC,KAAK,CAAC,CAAC;QACnC,IAAI,SAAS,KAAK,SAAS,IAAI,mBAAmB,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;YACvE,OAAO,KAAK,CAAC;QACf,CAAC;IACH,CAAC;IAED,OAAO,SAAS,CAAC,MAAM,CAAC,wBAAwB,CAAC,CAAC;AACpD,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,aAAa,CAC3B,QAAgB,EAChB,IAAsB;IAEtB,MAAM,QAAQ,GAAG,cAAc,CAAC,QAAQ,CAAC,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;IAC3D,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,KAAK,YAAY,EAAE,CAAC;QAChD,OAAO,QAAQ,CAAC;IAClB,CAAC;IAED,MAAM,WAAW,GAAG,mBAAmB,CAAC,QAAQ,CAAC,CAAC;IAClD,IAAI,WAAW,KAAK,CAAC,CAAC,EAAE,CAAC;QACvB,OAAO,QAAQ,CAAC;IAClB,CAAC;IAED,MAAM,SAAS,GAAG,cAAc,CAAC,IAAI,GAAG,CAAC,CAAC,CAAC;IAC3C,+EAA+E;IAC/E,IAAI,SAAS,KAAK,SAAS,EAAE,CAAC;QAC5B,OAAO,QAAQ,CAAC;IAClB,CAAC;IAED,OAAO,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,EAAE,WAAW,GAAG,CAAC,CAAC,GAAG,SAAS,GAAG,QAAQ,CAAC,KAAK,CACvE,WAAW,GAAG,CAAC,CAChB,EAAE,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;AACvB,CAAC"}
+{"version":3,"file":"tone-mark.js","sourceRoot":"","sources":["../../src/tone/tone-mark.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAa,MAAM,WAAW,CAAC;AAEpD;;;;;;;;GAQG;AACH,MAAM,cAAc,GAAG;IACrB,GAAG,EAAE,2BAA2B;IAChC,GAAG,EAAE,iCAAiC;IACtC,GAAG,EAAE,0BAA0B;IAC/B,GAAG,EAAE,iCAAiC;CAC9B,CAAC;AAEX;;;;;GAKG;AACH,MAAM,oBAAoB,GAAG,UAAU,CAAC;AAExC;;GAEG;AACH,MAAM,mBAAmB,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,CAAU,CAAC;AAErD;;;GAGG;AACH,MAAM,mBAAmB,GAAG,MAAM,CAAC;AAEnC;;;GAGG;AACH,MAAM,wBAAwB,GAAG,OAAO,CAAC;AAEzC;;;;;;;;GAQG;AACH,MAAM,UAAU,aAAa,CAAC,IAAU;IACtC,OAAO,cAAc,CAAC,IAAI,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;AACxC,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,cAAc,CAAC,IAAY;IACzC,OAAO,IAAI;SACR,SAAS,CAAC,KAAK,CAAC;SAChB,UAAU,CAAC,oBAAoB,EAAE,EAAE,CAAC;SACpC,SAAS,CAAC,KAAK,CAAC,CAAC;AACtB,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,cAAc,CAAC,IAAY;IACzC,OAAO,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC,KAAK,CAAC,oBAAoB,CAAC,EAAE,MAAM,IAAI,CAAC,CAAC;AACxE,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAA8B,IAAI,GAAG,CAAC;IAClE,CAAC,CAAC,EAAE,GAAG,CAAC;IACR,CAAC,CAAC,EAAE,GAAG,CAAC;IACR,CAAC,CAAC,EAAE,GAAG,CAAC;IACR,CAAC,CAAC,EAAE,GAAG,CAAC;IACR,CAAC,YAAY,EAAE,GAAG,CAAC;CACpB,CAAC,CAAC;AAEH;;;;;GAKG;AACH,MAAM,iBAAiB,GAAG,IAAI,GAAG,CAAiB;IAChD,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;CACX,CAAC,CAAC;AAEH;;GAEG;AACH,MAAM,UAAU,oBAAoB,CAAC,IAAY;IAC/C,OAAO,IAAI,CAAC,UAAU,CACpB,YAAY;IACZ,mEAAmE;IACnE,CAAC,KAAK,EAAE,EAAE,CAAC,iBAAiB,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,KAAK,CACjD,CAAC;AACJ,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,aAAa,CAAC,IAAY;IACxC,MAAM,UAAU,GAAG,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;IACzC,KAAK,MAAM,CAAC,KAAK,EAAE,SAAS,CAAC,IAAI,cAAc,CAAC,OAAO,EAAE,EAAE,CAAC;QAC1D,IAAI,UAAU,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;YACnC,OAAO,CAAC,KAAK,GAAG,CAAC,CAAS,CAAC;QAC7B,CAAC;IACH,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,mBAAmB,CAAC,QAAgB;IAC3C,MAAM,SAAS,GAAG,QAAQ,CAAC,WAAW,EAAE,CAAC;IAEzC,KAAK,MAAM,KAAK,IAAI,mBAAmB,EAAE,CAAC;QACxC,MAAM,KAAK,GAAG,SAAS,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC;QACvC,IAAI,KAAK,KAAK,CAAC,CAAC,EAAE,CAAC;YACjB,OAAO,KAAK,CAAC;QACf,CAAC;IACH,CAAC;IAED,KAAK,IAAI,KAAK,GAAG,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,KAAK,IAAI,CAAC,EAAE,KAAK,EAAE,EAAE,CAAC;QAC3D,MAAM,SAAS,GAAG,SAAS,CAAC,KAAK,CAAC,CAAC;QACnC,IAAI,SAAS,KAAK,SAAS,IAAI,mBAAmB,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;YACvE,OAAO,KAAK,CAAC;QACf,CAAC;IACH,CAAC;IAED,OAAO,SAAS,CAAC,MAAM,CAAC,wBAAwB,CAAC,CAAC;AACpD,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,aAAa,CAC3B,QAAgB,EAChB,IAAsB;IAEtB,MAAM,QAAQ,GAAG,cAAc,CAAC,QAAQ,CAAC,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;IAC3D,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,KAAK,YAAY,EAAE,CAAC;QAChD,OAAO,QAAQ,CAAC;IAClB,CAAC;IAED,MAAM,WAAW,GAAG,mBAAmB,CAAC,QAAQ,CAAC,CAAC;IAClD,IAAI,WAAW,KAAK,CAAC,CAAC,EAAE,CAAC;QACvB,OAAO,QAAQ,CAAC;IAClB,CAAC;IAED,MAAM,SAAS,GAAG,cAAc,CAAC,IAAI,GAAG,CAAC,CAAC,CAAC;IAC3C,+EAA+E;IAC/E,IAAI,SAAS,KAAK,SAAS,EAAE,CAAC;QAC5B,OAAO,QAAQ,CAAC;IAClB,CAAC;IAED,OAAO,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,EAAE,WAAW,GAAG,CAAC,CAAC,GAAG,SAAS,GAAG,QAAQ,CAAC,KAAK,CACvE,WAAW,GAAG,CAAC,CAChB,EAAE,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;AACvB,CAAC"}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@kensio/pinyinjs",
-  "version": "1.7.12",
+  "version": "1.7.14",
   "description": "Chinese hanzi to pinyin conversion",
   "repository": "https://github.com/KensioSoftware/pinyinjs",
   "homepage": "https://pinyinjs.dev",
@@ -69,7 +69,7 @@
     "conventional-changelog-conventionalcommits": "9.3.1",
     "fta-cli": "^3.0.0",
     "jiti": "^2.7.0",
-    "oxfmt": "^0.64.0",
+    "oxfmt": "^0.66.0",
     "oxlint": "^1.77.0",
     "oxlint-tsgolint": "^7.0.2001",
     "semantic-release": "25.0.9",