@kensio/pinyinjs 1.7.4 → 1.7.5

package/dist/format/html.js

@@ -1,6 +1,6 @@
 import { isUncertain } from "../decode/confidence.js";
+export { convertToAnnotatedHtml, convertToHtml, toAnnotatedHtml, } from "./annotate.js";
 import { READING_CHARGE } from "../decode/lattice.js";
-import { convertPieces, } from "../decode/convert.js";
 import { DEFAULT_LOCALE } from "../script/script.js";
 import { writeSyllable } from "../syllable/syllable.js";
 /**
@@ -62,7 +62,7 @@ const ESCAPES = new Map([
  * every attribute here is double-quoted, and the 隔音符号 of `Xī'ān` is a
  * letter of the orthography that a reader should see as one.
  */
-function escape(text) {
+export function escape(text) {
     return text.replaceAll(/[&<>"]/gu, (character) => {
         /* c8 ignore next -- the pattern only matches what the map holds */
         return ESCAPES.get(character) ?? character;
@@ -87,7 +87,7 @@ function alternativesOf(confidence, options) {
 /**
  * Mark up one piece of a conversion.
  */
-function writePiece(piece, options) {
+export function writePiece(piece, options) {
     const { toneClasses = true, markUncertain = true, lang = true } = options;
     if (piece.syllable === undefined) {
         return escape(piece.text);
@@ -122,163 +122,4 @@ function writePiece(piece, options) {
 export function toHtml(pieces, options = {}) {
     return pieces.map((piece) => writePiece(piece, options)).join("");
 }
-/**
- * What the characters being annotated declare themselves to be.
- *
- * The mirror of {@link languageTag}: inside an annotation the base is hanzi and
- * only the reading is pinyin, so on an English page the two need different tags
- * and neither can be left to inherit.
- *
- * **`zh`, with no script subtag, and that is deliberate.** `Hans` or `Hant`
- * would be better for font selection, and neither can be known here. The
- * conversion's locale does not settle it — script and locale are separate axes,
- * and a mainland edition of a classical text is 繁體 read as `zh-CN` — and
- * detecting it from the characters needs the script tables, which load
- * separately from the dictionary precisely so that converting does not pay for
- * them. A tag that is merely unspecific is worth more than one that is wrong,
- * so a caller who knows the script should wrap the output and say so there.
- */
-const BASE_TAG = "zh";
-/**
- * Gather pieces into the bases they are readings of.
- *
- * A piece naming characters opens a group and every piece after it that names
- * none reads on into the same one, which is how 玩儿 keeps `wánr` over both
- * characters and 95% keeps `bǎifēnzhījiǔshíwǔ` over all three. Pieces writing
- * no reading at all — the spaces between words, and runs that were never Han —
- * are their own groups, with nothing to annotate.
- */
-function annotationsOf(pieces) {
-    const groups = [];
-    for (const [at, piece] of pieces.entries()) {
-        const open = groups.at(-1);
-        if (open?.source !== undefined && continues(piece, pieces[at + 1])) {
-            groups[groups.length - 1] = {
-                source: open.source,
-                pieces: [...open.pieces, piece],
-            };
-            continue;
-        }
-        groups.push({ source: piece.source, pieces: [piece] });
-    }
-    return groups;
-}
-/**
- * Whether a piece belongs to the annotation the group before it opened.
- *
- * Three kinds do:
- *
- * - A **syllable naming no characters** is reading on into the ones already
- *   named, which is 儿化 and every syllable of a read number after the first.
- * - A **separator that is not a space** is a mark *inside* one orthographic
- *   word rather than a gap between two — the hyphen of `gāngān-jìngjìng`, the
- *   only one GB/T 16159 writes — so it belongs beside the syllables it divides.
- * - A **space with more of the same reading after it**. 1988 is read
- *   `yī jiǔ bā bā`, four syllables over one base, and the spaces between them
- *   are inside that reading rather than between two of them.
- *
- * That last one is why this needs the piece after it. A space is written the
- * same way whether it separates two words or two syllables of one number, and
- * what tells them apart is whether what follows is still reading the same
- * characters. Told apart wrongly, the reading escapes its own annotation:
- * 1988年 came out as `1988` over `yī` with `jiǔ bā bā` loose beside it.
- *
- * A space between two words is the case that does not continue, and dropping it
- * is the point: the hanzi it annotates has no space in it, and each base is
- * already its own group on the page.
- */
-function continues(piece, next) {
-    if (piece.syllable !== undefined) {
-        return piece.source === undefined;
-    }
-    if (piece.source !== undefined) {
-        return false;
-    }
-    return (piece.text.trim() !== "" ||
-        (next?.syllable !== undefined && next.source === undefined));
-}
-/**
- * Mark up one base and its reading.
- */
-function writeAnnotation(annotation, options) {
-    const { lang = true } = options;
-    const { source, pieces } = annotation;
-    if (pieces.every((piece) => piece.syllable === undefined)) {
-        // Nothing is read here. Either it is source text that simply has no
-        // reading — punctuation, a Latin word — and is written as the author wrote
-        // it, or it is pinyin orthography with no source behind it at all: the
-        // space between two words, and the hyphen in `gāngān-jìngjìng`. Writing
-        // those into the hanzi would annotate 干干净净 as 干干-净净.
-        return source === undefined ? "" : escape(source);
-    }
-    if (source === undefined) {
-        // A reading whose piece never named what it reads. Nothing a conversion
-        // builds looks like this — every group is opened by a piece carrying its
-        // characters — but {@link toAnnotatedHtml} takes pieces from the caller,
-        // and there is no base to hang an annotation on, so it is written as the
-        // plain markup it would have got from {@link toHtml}.
-        return pieces.map((piece) => writePiece(piece, options)).join("");
-    }
-    const reading = pieces.map((piece) => writePiece(piece, options)).join("");
-    // The tag goes on the <ruby> rather than around the base, and the syllables
-    // inside the <rt> override it with their own. `<rb>` would be the obvious
-    // place for it and is not an element any more: the WHATWG parser drops it,
-    // leaving the base as bare text, so writing one would mean emitting markup
-    // that no browser keeps.
-    const opening = lang ? `<ruby lang="${BASE_TAG}">` : "<ruby>";
-    return `${opening}${escape(source)}<rp>(</rp><rt>${reading}</rt><rp>)</rp></ruby>`;
-}
-/**
- * Mark up a converted text as hanzi with its reading above, syllable by
- * syllable.
- *
- * The output every other mode cannot give: `convert` and {@link toHtml} write
- * the pinyin *instead of* the hanzi, and a learner's text, a subtitle and a
- * dictionary entry all want both at once. The markup is `<ruby>`, which every
- * current browser lays out natively and which degrades to parentheses through
- * `<rp>` where it is not supported.
- *
- * ```html
- * <ruby lang="zh">银<rp>(</rp><rt><span …>yín</span></rt><rp>)</rp></ruby>
- * ```
- *
- * Each `<rt>` holds whatever {@link toHtml} would have written for those
- * syllables, so tone classes and uncertainty marking work inside an annotation
- * exactly as they do outside one.
- *
- * **A base is not always one character.** 玩儿 is two characters and the one
- * syllable `wánr`, and a read number reverses on the way — 95% is
- * `bǎifēnzhījiǔshíwǔ` — so both are annotated whole. Splitting them per
- * character is what everything that emits ruby from a pinyin library gets
- * wrong, and it is why {@link ConvertedPiece.source} exists.
- */
-export function toAnnotatedHtml(pieces, options = {}) {
-    return annotationsOf(pieces)
-        .map((annotation) => writeAnnotation(annotation, options))
-        .join("");
-}
-/**
- * Convert hanzi to pinyin as HTML, hanzi and reading together.
- *
- * {@link toAnnotatedHtml} applied to a fresh conversion; see it for what the
- * markup is and why a base is not always one character.
- */
-export function convertToAnnotatedHtml(dictionary, text, options = {}) {
-    return toAnnotatedHtml(convertPieces(dictionary, text, options), options);
-}
-/**
- * Convert hanzi to pinyin as HTML, one element per syllable.
- *
- * What per-syllable confidence is for, and the reason ALGORITHM.md asks for it:
- * a reading the decode was guessing at can be *shown* to be a guess, which for
- * a learner is a feature rather than a diagnostic. No styling is imposed — the
- * classes are hooks, and the page decides whether a fourth tone is red and an
- * uncertain reading is dotted underneath.
- *
- * Nothing is wrapped around the whole conversion, and anything that is not Han
- * passes through escaped rather than marked up.
- */
-export function convertToHtml(dictionary, text, options = {}) {
-    return toHtml(convertPieces(dictionary, text, options), options);
-}
 //# sourceMappingURL=html.js.map

package/dist/format/html.js.map

@@ -1 +1 @@
-{"version":3,"file":"html.js","sourceRoot":"","sources":["../../src/format/html.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAA0B,MAAM,yBAAyB,CAAC;AAC9E,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AACtD,OAAO,EAGL,aAAa,GACd,MAAM,sBAAsB,CAAC;AAE9B,OAAO,EAAE,cAAc,EAAe,MAAM,qBAAqB,CAAC;AAClE,OAAO,EAAE,aAAa,EAAE,MAAM,yBAAyB,CAAC;AA+BxD;;GAEG;AACH,MAAM,cAAc,GAAG,aAAa,CAAC;AAErC,MAAM,eAAe,GAAG,cAAc,CAAC;AAEvC;;;;;;;;;;;;;;;;;;;;;;;;;GAyBG;AACH,SAAS,WAAW,CAAC,MAAM,GAAW,cAAc;IAClD,QAAQ,MAAM,EAAE,CAAC;QACf,KAAK,OAAO,EAAE,CAAC;YACb,OAAO,mBAAmB,CAAC;QAC7B,CAAC;QACD,KAAK,OAAO,EAAE,CAAC;YACb,OAAO,mBAAmB,CAAC;QAC7B,CAAC;IACH,CAAC;AACH,CAAC;AAED;;GAEG;AACH,MAAM,OAAO,GAAG,IAAI,GAAG,CAAiB;IACtC,CAAC,GAAG,EAAE,OAAO,CAAC;IACd,CAAC,GAAG,EAAE,MAAM,CAAC;IACb,CAAC,GAAG,EAAE,MAAM,CAAC;IACb,CAAC,GAAG,EAAE,QAAQ,CAAC;CAChB,CAAC,CAAC;AAEH;;;;;;;;GAQG;AACH,SAAS,MAAM,CAAC,IAAY;IAC1B,OAAO,IAAI,CAAC,UAAU,CAAC,UAAU,EAAE,CAAC,SAAS,EAAE,EAAE;QAC/C,mEAAmE;QACnE,OAAO,OAAO,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,SAAS,CAAC;IAC7C,CAAC,CAAC,CAAC;AACL,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,cAAc,CACrB,UAA6B,EAC7B,OAAoB;IAEpB,MAAM,OAAO,GAAG,UAAU,CAAC,YAAY;SACpC,MAAM,CAAC,CAAC,WAAW,EAAE,EAAE,CAAC,WAAW,CAAC,IAAI,GAAG,cAAc,CAAC;SAC1D,GAAG,CAAC,CAAC,WAAW,EAAE,EAAE,CACnB,WAAW,CAAC,OAAO;SAChB,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,aAAa,CAAC,QAAQ,EAAE,OAAO,CAAC,QAAQ,CAAC,CAAC;SAC5D,IAAI,CAAC,EAAE,CAAC,CACZ,CAAC;IACJ,OAAO,CAAC,GAAG,IAAI,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC;AAC/B,CAAC;AAED;;GAEG;AACH,SAAS,UAAU,CAAC,KAAqB,EAAE,OAAoB;IAC7D,MAAM,EAAE,WAAW,GAAG,IAAI,EAAE,aAAa,GAAG,IAAI,EAAE,IAAI,GAAG,IAAI,EAAE,GAAG,OAAO,CAAC;IAC1E,IAAI,KAAK,CAAC,QAAQ,KAAK,SAAS,EAAE,CAAC;QACjC,OAAO,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IAC5B,CAAC;IAED,MAAM,OAAO,GAAG,CAAC,cAAc,CAAC,CAAC;IACjC,IAAI,WAAW,IAAI,KAAK,CAAC,QAAQ,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;QACrD,OAAO,CAAC,IAAI,CAAC,WAAW,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IACzD,CAAC;IACD,MAAM,YAAY,GAChB,aAAa;QACb,KAAK,CAAC,UAAU,KAAK,SAAS;QAC9B,WAAW,CAAC,KAAK,CAAC,UAAU,CAAC;QAC3B,CAAC,CAAC,cAAc,CAAC,KAAK,CAAC,UAAU,EAAE,OAAO,CAAC;QAC3C,CAAC,CAAC,EAAE,CAAC;IACT,IAAI,YAAY,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC5B,OAAO,CAAC,IAAI,CAAC,eAAe,CAAC,CAAC;IAChC,CAAC;IAED,MAAM,UAAU,GAAG,CAAC,UAAU,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;IACpD,IAAI,IAAI,EAAE,CAAC;QACT,UAAU,CAAC,IAAI,CAAC,SAAS,WAAW,CAAC,OAAO,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;IAC3D,CAAC;IACD,IAAI,YAAY,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC5B,UAAU,CAAC,IAAI,CAAC,sBAAsB,MAAM,CAAC,YAAY,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC;IAC3E,CAAC;IACD,OAAO,SAAS,UAAU,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC;AACtE,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,MAAM,CACpB,MAAiC,EACjC,OAAO,GAAgB,EAAE;IAEzB,OAAO,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,UAAU,CAAC,KAAK,EAAE,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;AACpE,CAAC;AAED;;;;;;;;;;;;;;;GAeG;AACH,MAAM,QAAQ,GAAG,IAAI,CAAC;AAWtB;;;;;;;;GAQG;AACH,SAAS,aAAa,CAAC,MAAiC;IACtD,MAAM,MAAM,GAAiB,EAAE,CAAC;IAChC,KAAK,MAAM,CAAC,EAAE,EAAE,KAAK,CAAC,IAAI,MAAM,CAAC,OAAO,EAAE,EAAE,CAAC;QAC3C,MAAM,IAAI,GAAG,MAAM,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;QAC3B,IAAI,IAAI,EAAE,MAAM,KAAK,SAAS,IAAI,SAAS,CAAC,KAAK,EAAE,MAAM,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC;YACnE,MAAM,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC,GAAG;gBAC1B,MAAM,EAAE,IAAI,CAAC,MAAM;gBACnB,MAAM,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,KAAK,CAAC;aAChC,CAAC;YACF,SAAS;QACX,CAAC;QACD,MAAM,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,KAAK,CAAC,MAAM,EAAE,MAAM,EAAE,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC;IACzD,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AACH,SAAS,SAAS,CAChB,KAAqB,EACrB,IAAgC;IAEhC,IAAI,KAAK,CAAC,QAAQ,KAAK,SAAS,EAAE,CAAC;QACjC,OAAO,KAAK,CAAC,MAAM,KAAK,SAAS,CAAC;IACpC,CAAC;IACD,IAAI,KAAK,CAAC,MAAM,KAAK,SAAS,EAAE,CAAC;QAC/B,OAAO,KAAK,CAAC;IACf,CAAC;IACD,OAAO,CACL,KAAK,CAAC,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE;QACxB,CAAC,IAAI,EAAE,QAAQ,KAAK,SAAS,IAAI,IAAI,CAAC,MAAM,KAAK,SAAS,CAAC,CAC5D,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,SAAS,eAAe,CAAC,UAAsB,EAAE,OAAoB;IACnE,MAAM,EAAE,IAAI,GAAG,IAAI,EAAE,GAAG,OAAO,CAAC;IAChC,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,GAAG,UAAU,CAAC;IACtC,IAAI,MAAM,CAAC,KAAK,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,QAAQ,KAAK,SAAS,CAAC,EAAE,CAAC;QAC1D,oEAAoE;QACpE,2EAA2E;QAC3E,uEAAuE;QACvE,wEAAwE;QACxE,qDAAqD;QACrD,OAAO,MAAM,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;IACpD,CAAC;IACD,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;QACzB,wEAAwE;QACxE,yEAAyE;QACzE,yEAAyE;QACzE,yEAAyE;QACzE,sDAAsD;QACtD,OAAO,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,UAAU,CAAC,KAAK,EAAE,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IACpE,CAAC;IAED,MAAM,OAAO,GAAG,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,UAAU,CAAC,KAAK,EAAE,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IAC3E,4EAA4E;IAC5E,0EAA0E;IAC1E,2EAA2E;IAC3E,2EAA2E;IAC3E,yBAAyB;IACzB,MAAM,OAAO,GAAG,IAAI,CAAC,CAAC,CAAC,eAAe,QAAQ,IAAI,CAAC,CAAC,CAAC,QAAQ,CAAC;IAC9D,OAAO,GAAG,OAAO,GAAG,MAAM,CAAC,MAAM,CAAC,iBAAiB,OAAO,wBAAwB,CAAC;AACrF,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AACH,MAAM,UAAU,eAAe,CAC7B,MAAiC,EACjC,OAAO,GAAgB,EAAE;IAEzB,OAAO,aAAa,CAAC,MAAM,CAAC;SACzB,GAAG,CAAC,CAAC,UAAU,EAAE,EAAE,CAAC,eAAe,CAAC,UAAU,EAAE,OAAO,CAAC,CAAC;SACzD,IAAI,CAAC,EAAE,CAAC,CAAC;AACd,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,sBAAsB,CACpC,UAAsB,EACtB,IAAY,EACZ,OAAO,GAAgB,EAAE;IAEzB,OAAO,eAAe,CAAC,aAAa,CAAC,UAAU,EAAE,IAAI,EAAE,OAAO,CAAC,EAAE,OAAO,CAAC,CAAC;AAC5E,CAAC;AAED;;;;;;;;;;;GAWG;AACH,MAAM,UAAU,aAAa,CAC3B,UAAsB,EACtB,IAAY,EACZ,OAAO,GAAgB,EAAE;IAEzB,OAAO,MAAM,CAAC,aAAa,CAAC,UAAU,EAAE,IAAI,EAAE,OAAO,CAAC,EAAE,OAAO,CAAC,CAAC;AACnE,CAAC"}
+{"version":3,"file":"html.js","sourceRoot":"","sources":["../../src/format/html.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAA0B,MAAM,yBAAyB,CAAC;AAC9E,OAAO,EACL,sBAAsB,EACtB,aAAa,EACb,eAAe,GAChB,MAAM,eAAe,CAAC;AACvB,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAEtD,OAAO,EAAE,cAAc,EAAe,MAAM,qBAAqB,CAAC;AAClE,OAAO,EAAE,aAAa,EAAE,MAAM,yBAAyB,CAAC;AA+BxD;;GAEG;AACH,MAAM,cAAc,GAAG,aAAa,CAAC;AAErC,MAAM,eAAe,GAAG,cAAc,CAAC;AAEvC;;;;;;;;;;;;;;;;;;;;;;;;;GAyBG;AACH,SAAS,WAAW,CAAC,MAAM,GAAW,cAAc;IAClD,QAAQ,MAAM,EAAE,CAAC;QACf,KAAK,OAAO,EAAE,CAAC;YACb,OAAO,mBAAmB,CAAC;QAC7B,CAAC;QACD,KAAK,OAAO,EAAE,CAAC;YACb,OAAO,mBAAmB,CAAC;QAC7B,CAAC;IACH,CAAC;AACH,CAAC;AAED;;GAEG;AACH,MAAM,OAAO,GAAG,IAAI,GAAG,CAAiB;IACtC,CAAC,GAAG,EAAE,OAAO,CAAC;IACd,CAAC,GAAG,EAAE,MAAM,CAAC;IACb,CAAC,GAAG,EAAE,MAAM,CAAC;IACb,CAAC,GAAG,EAAE,QAAQ,CAAC;CAChB,CAAC,CAAC;AAEH;;;;;;;;GAQG;AACH,MAAM,UAAU,MAAM,CAAC,IAAY;IACjC,OAAO,IAAI,CAAC,UAAU,CAAC,UAAU,EAAE,CAAC,SAAS,EAAE,EAAE;QAC/C,mEAAmE;QACnE,OAAO,OAAO,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,SAAS,CAAC;IAC7C,CAAC,CAAC,CAAC;AACL,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,cAAc,CACrB,UAA6B,EAC7B,OAAoB;IAEpB,MAAM,OAAO,GAAG,UAAU,CAAC,YAAY;SACpC,MAAM,CAAC,CAAC,WAAW,EAAE,EAAE,CAAC,WAAW,CAAC,IAAI,GAAG,cAAc,CAAC;SAC1D,GAAG,CAAC,CAAC,WAAW,EAAE,EAAE,CACnB,WAAW,CAAC,OAAO;SAChB,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,aAAa,CAAC,QAAQ,EAAE,OAAO,CAAC,QAAQ,CAAC,CAAC;SAC5D,IAAI,CAAC,EAAE,CAAC,CACZ,CAAC;IACJ,OAAO,CAAC,GAAG,IAAI,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC;AAC/B,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,UAAU,CACxB,KAAqB,EACrB,OAAoB;IAEpB,MAAM,EAAE,WAAW,GAAG,IAAI,EAAE,aAAa,GAAG,IAAI,EAAE,IAAI,GAAG,IAAI,EAAE,GAAG,OAAO,CAAC;IAC1E,IAAI,KAAK,CAAC,QAAQ,KAAK,SAAS,EAAE,CAAC;QACjC,OAAO,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IAC5B,CAAC;IAED,MAAM,OAAO,GAAG,CAAC,cAAc,CAAC,CAAC;IACjC,IAAI,WAAW,IAAI,KAAK,CAAC,QAAQ,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;QACrD,OAAO,CAAC,IAAI,CAAC,WAAW,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IACzD,CAAC;IACD,MAAM,YAAY,GAChB,aAAa;QACb,KAAK,CAAC,UAAU,KAAK,SAAS;QAC9B,WAAW,CAAC,KAAK,CAAC,UAAU,CAAC;QAC3B,CAAC,CAAC,cAAc,CAAC,KAAK,CAAC,UAAU,EAAE,OAAO,CAAC;QAC3C,CAAC,CAAC,EAAE,CAAC;IACT,IAAI,YAAY,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC5B,OAAO,CAAC,IAAI,CAAC,eAAe,CAAC,CAAC;IAChC,CAAC;IAED,MAAM,UAAU,GAAG,CAAC,UAAU,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;IACpD,IAAI,IAAI,EAAE,CAAC;QACT,UAAU,CAAC,IAAI,CAAC,SAAS,WAAW,CAAC,OAAO,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;IAC3D,CAAC;IACD,IAAI,YAAY,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC5B,UAAU,CAAC,IAAI,CAAC,sBAAsB,MAAM,CAAC,YAAY,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC;IAC3E,CAAC;IACD,OAAO,SAAS,UAAU,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC;AACtE,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,MAAM,CACpB,MAAiC,EACjC,OAAO,GAAgB,EAAE;IAEzB,OAAO,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,UAAU,CAAC,KAAK,EAAE,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;AACpE,CAAC"}

package/dist/format/slug-hash.d.ts

@@ -0,0 +1,51 @@
+/**
+ * The short hash a slug may carry on its end.
+ *
+ * Apart from the rest of slugging because it never sees a syllable: what is
+ * hashed is the source text, for the reason set out at {@link hashOf}, so this
+ * runs beside the conversion rather than after it.
+ */
+/**
+ * How many characters {@link SlugOptions.hash} writes when simply asked for.
+ *
+ * Four base-36 characters is 1.7 million values, which is far more than it
+ * looks: two texts only collide when their slugs *and* their hashes match, so
+ * the suffix is only ever telling apart the handful of texts sharing one slug,
+ * not the whole corpus.
+ */
+export declare const DEFAULT_HASH_LENGTH = 4;
+/**
+ * The longest hash there is anything to fill.
+ *
+ * {@link hashOf} is 32 bits wide, and 36⁷ is the last power of 36 under 2³²:
+ * asking for an eighth character would add a place that some values can never
+ * reach, which looks like more uniqueness while being none.
+ */
+export declare const LONGEST_HASH = 7;
+/**
+ * A 32-bit hash of a text, written in the letters and digits a slug allows.
+ *
+ * **The source text is what is hashed, not the pinyin.** Two texts needing a
+ * hash to tell them apart are, by definition, ones the pinyin already ran
+ * together: 权利 and 权力 are both `quan2li4`, and hashing that would give them
+ * the same suffix and settle nothing. Hashing the hanzi also means the suffix
+ * does not move when a later release reads a word differently, so a slug's tail
+ * is the stable half of it.
+ *
+ * FNV-1a with murmur3's finalizer, which is neither cryptographic nor trying to
+ * be: nothing here resists an attacker, and all that is being asked of it is
+ * that two texts a reader would call different land on different suffixes.
+ * Written out rather than reached for because the package has no runtime
+ * dependencies and the core imports no Node built-ins, and `crypto.subtle` —
+ * the one hash a browser ships — is asynchronous, which this is not.
+ *
+ * Normalised first, so that text typed as a decomposed sequence hashes as the
+ * same text once composed. 简体 and 繁體 are deliberately *not* folded together:
+ * a caller who wrote 头发 and 頭髮 separately meant them separately.
+ */
+export declare function hashOf(text: string, length: number): string;
+/**
+ * How long a hash the options asked for, or zero for none.
+ */
+export declare function hashLength(asked: number | boolean | undefined): number;
+//# sourceMappingURL=slug-hash.d.ts.map

package/dist/format/slug-hash.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"slug-hash.d.ts","sourceRoot":"","sources":["../../src/format/slug-hash.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAEH;;;;;;;GAOG;AACH,eAAO,MAAM,mBAAmB,IAAI,CAAC;AAErC;;;;;;GAMG;AACH,eAAO,MAAM,YAAY,IAAI,CAAC;AAE9B;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,wBAAgB,MAAM,CAAC,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,GAAG,MAAM,CAgB3D;AAED;;GAEG;AACH,wBAAgB,UAAU,CAAC,KAAK,EAAE,MAAM,GAAG,OAAO,GAAG,SAAS,GAAG,MAAM,CAQtE"}

package/dist/format/slug-hash.js

@@ -0,0 +1,74 @@
+/**
+ * The short hash a slug may carry on its end.
+ *
+ * Apart from the rest of slugging because it never sees a syllable: what is
+ * hashed is the source text, for the reason set out at {@link hashOf}, so this
+ * runs beside the conversion rather than after it.
+ */
+/**
+ * How many characters {@link SlugOptions.hash} writes when simply asked for.
+ *
+ * Four base-36 characters is 1.7 million values, which is far more than it
+ * looks: two texts only collide when their slugs *and* their hashes match, so
+ * the suffix is only ever telling apart the handful of texts sharing one slug,
+ * not the whole corpus.
+ */
+export const DEFAULT_HASH_LENGTH = 4;
+/**
+ * The longest hash there is anything to fill.
+ *
+ * {@link hashOf} is 32 bits wide, and 36⁷ is the last power of 36 under 2³²:
+ * asking for an eighth character would add a place that some values can never
+ * reach, which looks like more uniqueness while being none.
+ */
+export const LONGEST_HASH = 7;
+/**
+ * A 32-bit hash of a text, written in the letters and digits a slug allows.
+ *
+ * **The source text is what is hashed, not the pinyin.** Two texts needing a
+ * hash to tell them apart are, by definition, ones the pinyin already ran
+ * together: 权利 and 权力 are both `quan2li4`, and hashing that would give them
+ * the same suffix and settle nothing. Hashing the hanzi also means the suffix
+ * does not move when a later release reads a word differently, so a slug's tail
+ * is the stable half of it.
+ *
+ * FNV-1a with murmur3's finalizer, which is neither cryptographic nor trying to
+ * be: nothing here resists an attacker, and all that is being asked of it is
+ * that two texts a reader would call different land on different suffixes.
+ * Written out rather than reached for because the package has no runtime
+ * dependencies and the core imports no Node built-ins, and `crypto.subtle` —
+ * the one hash a browser ships — is asynchronous, which this is not.
+ *
+ * Normalised first, so that text typed as a decomposed sequence hashes as the
+ * same text once composed. 简体 and 繁體 are deliberately *not* folded together:
+ * a caller who wrote 头发 and 頭髮 separately meant them separately.
+ */
+export function hashOf(text, length) {
+    const bytes = new TextEncoder().encode(text.normalize("NFC"));
+    let hash = 0x81_1c_9d_c5;
+    for (const byte of bytes) {
+        hash = Math.imul(hash ^ byte, 0x01_00_01_93);
+    }
+    // FNV-1a leaves its lowest bits the least mixed, and the last characters of
+    // the base-36 spelling are exactly those bits, so this spreads them first.
+    hash ^= hash >>> 16;
+    hash = Math.imul(hash, 0x85_eb_ca_6b);
+    hash ^= hash >>> 13;
+    hash = Math.imul(hash, 0xc2_b2_ae_35);
+    hash ^= hash >>> 16;
+    const wanted = Math.min(Math.max(Math.trunc(length), 1), LONGEST_HASH);
+    return (hash >>> 0).toString(36).padStart(LONGEST_HASH, "0").slice(-wanted);
+}
+/**
+ * How long a hash the options asked for, or zero for none.
+ */
+export function hashLength(asked) {
+    if (asked === undefined || asked === false) {
+        return 0;
+    }
+    if (asked === true) {
+        return DEFAULT_HASH_LENGTH;
+    }
+    return Math.min(Math.max(Math.trunc(asked), 0), LONGEST_HASH);
+}
+//# sourceMappingURL=slug-hash.js.map

package/dist/format/slug-hash.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"slug-hash.js","sourceRoot":"","sources":["../../src/format/slug-hash.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAEH;;;;;;;GAOG;AACH,MAAM,CAAC,MAAM,mBAAmB,GAAG,CAAC,CAAC;AAErC;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,YAAY,GAAG,CAAC,CAAC;AAE9B;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,MAAM,UAAU,MAAM,CAAC,IAAY,EAAE,MAAc;IACjD,MAAM,KAAK,GAAG,IAAI,WAAW,EAAE,CAAC,MAAM,CAAC,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC,CAAC;IAC9D,IAAI,IAAI,GAAG,aAAa,CAAC;IACzB,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC,IAAI,GAAG,IAAI,EAAE,aAAa,CAAC,CAAC;IAC/C,CAAC;IACD,4EAA4E;IAC5E,2EAA2E;IAC3E,IAAI,IAAI,IAAI,KAAK,EAAE,CAAC;IACpB,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,aAAa,CAAC,CAAC;IACtC,IAAI,IAAI,IAAI,KAAK,EAAE,CAAC;IACpB,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,aAAa,CAAC,CAAC;IACtC,IAAI,IAAI,IAAI,KAAK,EAAE,CAAC;IAEpB,MAAM,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,EAAE,YAAY,CAAC,CAAC;IACvE,OAAO,CAAC,IAAI,KAAK,CAAC,CAAC,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAC,QAAQ,CAAC,YAAY,EAAE,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC,MAAM,CAAC,CAAC;AAC9E,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,UAAU,CAAC,KAAmC;IAC5D,IAAI,KAAK,KAAK,SAAS,IAAI,KAAK,KAAK,KAAK,EAAE,CAAC;QAC3C,OAAO,CAAC,CAAC;IACX,CAAC;IACD,IAAI,KAAK,KAAK,IAAI,EAAE,CAAC;QACnB,OAAO,mBAAmB,CAAC;IAC7B,CAAC;IACD,OAAO,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC,EAAE,YAAY,CAAC,CAAC;AAChE,CAAC"}

package/dist/format/slug-options.d.ts

@@ -0,0 +1,98 @@
+/**
+ * How a slug is built.
+ *
+ * Their own module because every part of slugging reads them and none of them
+ * reads the others: the words, the hash and the fitting all take the same
+ * options object, and none of the three should have to import the other two to
+ * name it.
+ */
+import type { NumberStyle } from "../decode/convert.js";
+import type { SandhiOptions } from "../decode/sandhi.js";
+import type { Locale } from "../script/script.js";
+/**
+ * Whether a slug carries its tones.
+ *
+ * `numbers` writes them as digits — 中文 is `zhong1wen2` — and `none` leaves
+ * them off. Tones are on by default because dropping them collides: there are
+ * roughly 400 toneless syllables against 1300 toned ones, so a toneless slug
+ * runs 树, 书 and 输 together as `shu`. They are not a guarantee, though —
+ * 权利 and 权力 are both `quan2li4`, and only {@link SlugOptions.hash} tells
+ * those apart.
+ */
+export type SlugTones = "numbers" | "none";
+/**
+ * Whether the syllables inside one word are separated.
+ *
+ * `join` writes a word as a word — 中文 is `zhong1wen2` — which is what 正词法
+ * asks for and what a reader expects in a URL. `separate` cuts every syllable
+ * apart, `zhong1-wen2`, for a caller matching a slug syllable by syllable.
+ */
+export type SlugSyllables = "join" | "separate";
+/**
+ * How ü is written where the alphabet has no room for it.
+ *
+ * `v` is the input-method convention and keeps 绿 `lv4` apart from 路 `lu4`;
+ * `u` is prettier and merges them.
+ */
+export type SlugUmlaut = "v" | "u";
+/**
+ * How a slug is built.
+ */
+export interface SlugOptions {
+    /** Whether tones are written. Defaults to `numbers`. */
+    readonly tones?: SlugTones;
+    /**
+     * What goes between words. Defaults to `-`.
+     *
+     * Written as given rather than checked, because the separator is the one
+     * place a caller decides what kind of identifier this is: `-` for a URL, `_`
+     * for a name in code, and the empty string for a key to match against.
+     */
+    readonly separator?: string;
+    /** Whether the syllables within a word are separated. Defaults to `join`. */
+    readonly syllables?: SlugSyllables;
+    /** How ü is written. Defaults to `v`. */
+    readonly umlaut?: SlugUmlaut;
+    /**
+     * What to do with the digits in a text. Defaults to `keep`.
+     *
+     * The opposite of what a conversion defaults to, and deliberately: a slug for
+     * iPhone 15 评测 should be `iphone-15-ping2ce4` rather than
+     * `iphone-shi2wu3-ping2ce4`, because the digits are how anyone would look for
+     * it. `read` says them, for a caller that wants the text spoken.
+     */
+    readonly numbers?: NumberStyle;
+    /**
+     * A short hash of the source text on the end. Defaults to none.
+     *
+     * `true` for {@link DEFAULT_HASH_LENGTH} characters, or a length of your own,
+     * up to {@link LONGEST_HASH}. See {@link hashOf} for what is hashed and why
+     * it is the hanzi rather than the pinyin.
+     *
+     * Unconditional rather than only where two slugs collide: a function that
+     * cannot see the rest of your corpus cannot know that they did. It narrows
+     * collisions rather than closing them, and checking the slug you generated
+     * against the ones you have stored is still yours to do.
+     */
+    readonly hash?: number | boolean;
+    /**
+     * The longest the slug may be, counting any hash. Defaults to no limit.
+     *
+     * Cut at a word boundary, so a limit lands between words rather than through
+     * the middle of one. A single word longer than the limit is cut where the
+     * limit falls, since there is nowhere else to cut it, and the hash is never
+     * what gets dropped.
+     */
+    readonly maxLength?: number;
+    /**
+     * What to write where a text slugs to nothing. Defaults to the empty string.
+     *
+     * 《》！ has no letters in it, and neither does an empty text.
+     */
+    readonly fallback?: string;
+    /** Which reading standard to use. Defaults to `zh-CN`. */
+    readonly locale?: Locale;
+    /** Which tone sandhi to apply. Defaults to what a conversion applies. */
+    readonly sandhi?: SandhiOptions;
+}
+//# sourceMappingURL=slug-options.d.ts.map

package/dist/format/slug-options.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"slug-options.d.ts","sourceRoot":"","sources":["../../src/format/slug-options.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AACH,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAC;AACxD,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,qBAAqB,CAAC;AACzD,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,qBAAqB,CAAC;AAElD;;;;;;;;;GASG;AACH,MAAM,MAAM,SAAS,GAAG,SAAS,GAAG,MAAM,CAAC;AAE3C;;;;;;GAMG;AACH,MAAM,MAAM,aAAa,GAAG,MAAM,GAAG,UAAU,CAAC;AAEhD;;;;;GAKG;AACH,MAAM,MAAM,UAAU,GAAG,GAAG,GAAG,GAAG,CAAC;AAEnC;;GAEG;AACH,MAAM,WAAW,WAAW;IAC1B,wDAAwD;IACxD,QAAQ,CAAC,KAAK,CAAC,EAAE,SAAS,CAAC;IAC3B;;;;;;OAMG;IACH,QAAQ,CAAC,SAAS,CAAC,EAAE,MAAM,CAAC;IAC5B,6EAA6E;IAC7E,QAAQ,CAAC,SAAS,CAAC,EAAE,aAAa,CAAC;IACnC,yCAAyC;IACzC,QAAQ,CAAC,MAAM,CAAC,EAAE,UAAU,CAAC;IAC7B;;;;;;;OAOG;IACH,QAAQ,CAAC,OAAO,CAAC,EAAE,WAAW,CAAC;IAC/B;;;;;;;;;;;OAWG;IACH,QAAQ,CAAC,IAAI,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC;IACjC;;;;;;;OAOG;IACH,QAAQ,CAAC,SAAS,CAAC,EAAE,MAAM,CAAC;IAC5B;;;;OAIG;IACH,QAAQ,CAAC,QAAQ,CAAC,EAAE,MAAM,CAAC;IAC3B,0DAA0D;IAC1D,QAAQ,CAAC,MAAM,CAAC,EAAE,MAAM,CAAC;IACzB,yEAAyE;IACzE,QAAQ,CAAC,MAAM,CAAC,EAAE,aAAa,CAAC;CACjC"}

package/dist/format/slug-options.js

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=slug-options.js.map

package/dist/format/slug-options.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"slug-options.js","sourceRoot":"","sources":["../../src/format/slug-options.ts"],"names":[],"mappings":""}

package/dist/format/slug-words.d.ts

@@ -0,0 +1,27 @@
+/**
+ * Writing the words of a slug, before anything is decided about its length.
+ *
+ * Everything here works in boundaries rather than in the caller's separator,
+ * for the reason given at {@link BOUNDARY}; what comes out is a list of words,
+ * and `slug.ts` decides how many of them there is room for and what goes
+ * between them.
+ */
+import type { ConvertedPiece } from "../decode/convert.js";
+import type { SlugOptions } from "./slug-options.js";
+/**
+ * Gather each run of consecutive syllables, which is one word.
+ *
+ * A conversion puts a piece carrying no syllable between two words — the space,
+ * or the hyphen 分词连写 writes inside 干干净净 — so the words are already
+ * marked out and need no second look at the text.
+ */
+export declare function slugPieces(pieces: readonly ConvertedPiece[], options: SlugOptions): string;
+/**
+ * The words a build produced, with its empty boundaries collapsed away.
+ *
+ * Everything after this works on the words rather than on the text they will
+ * become, so that a limit counts the separator the caller actually chose: a
+ * boundary is one character here and could be two, or none, once written.
+ */
+export declare function partsOf(built: string): readonly string[];
+//# sourceMappingURL=slug-words.d.ts.map

package/dist/format/slug-words.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"slug-words.d.ts","sourceRoot":"","sources":["../../src/format/slug-words.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AACH,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAG3D,OAAO,KAAK,EAAE,WAAW,EAAc,MAAM,mBAAmB,CAAC;AAsFjE;;;;;;GAMG;AACH,wBAAgB,UAAU,CACxB,MAAM,EAAE,SAAS,cAAc,EAAE,EACjC,OAAO,EAAE,WAAW,GACnB,MAAM,CAuBR;AAED;;;;;;GAMG;AACH,wBAAgB,OAAO,CAAC,KAAK,EAAE,MAAM,GAAG,SAAS,MAAM,EAAE,CAExD"}

package/dist/format/slug-words.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"slug-words.js","sourceRoot":"","sources":["../../src/format/slug-words.ts"],"names":[],"mappings":"AASA,OAAO,EAAE,QAAQ,EAAE,MAAM,8BAA8B,CAAC;AACxD,OAAO,EAAiB,aAAa,EAAE,MAAM,yBAAyB,CAAC;AAGvE;;;;;;;;;;;GAWG;AACH,MAAM,QAAQ,GAAG,GAAG,CAAC;AAErB;;GAEG;AACH,MAAM,eAAe,GAAG,cAAc,CAAC;AAEvC;;GAEG;AACH,MAAM,eAAe,GAAG,UAAU,CAAC;AAEnC;;;;;;;GAOG;AACH,SAAS,SAAS,CAAC,IAAY;IAC7B,OAAO,IAAI;SACR,WAAW,EAAE;SACb,SAAS,CAAC,KAAK,CAAC;SAChB,UAAU,CAAC,eAAe,EAAE,EAAE,CAAC;SAC/B,UAAU,CAAC,eAAe,EAAE,GAAG,EAAE,CAAC,QAAQ,CAAC,CAAC;AACjD,CAAC;AAED;;;;;;GAMG;AACH,SAAS,WAAW,CAAC,QAAgB,EAAE,MAAkB;IACvD,OAAO,QAAQ,CAAC,UAAU,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC,MAAM,CAAC,CAAC;AAChD,CAAC;AAED;;;;;;;;;;;GAWG;AACH,SAAS,QAAQ,CACf,SAA8B,EAC9B,OAAoB;IAEpB,MAAM,EAAE,KAAK,GAAG,SAAS,EAAE,SAAS,EAAE,GAAG,GAAG,MAAM,EAAE,MAAM,GAAG,GAAG,EAAE,GAAG,OAAO,CAAC;IAC7E,MAAM,SAAS,GAAG,SAAS,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CAC3C,WAAW,CACT,aAAa,CAAC,QAAQ,EAAE,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,MAAM,CAAC,EACjE,MAAM,CACP,CACF,CAAC;IACF,IAAI,GAAG,KAAK,UAAU,EAAE,CAAC;QACvB,OAAO,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;IAClC,CAAC;IACD,wEAAwE;IACxE,0EAA0E;IAC1E,OAAO,KAAK,KAAK,SAAS;QACxB,CAAC,CAAC,SAAS,CAAC,IAAI,CAAC,EAAE,CAAC;QACpB,CAAC,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAC,UAAU,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC,QAAQ,CAAC,CAAC;AAC1D,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,UAAU,CACxB,MAAiC,EACjC,OAAoB;IAEpB,MAAM,OAAO,GAAa,EAAE,CAAC;IAC7B,IAAI,IAAI,GAAe,EAAE,CAAC;IAE1B,MAAM,KAAK,GAAG,GAAS,EAAE;QACvB,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACtB,OAAO;QACT,CAAC;QAED,OAAO,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC,CAAC;QACtC,IAAI,GAAG,EAAE,CAAC;IACZ,CAAC,CAAC;IAEF,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;QAC3B,IAAI,KAAK,CAAC,QAAQ,KAAK,SAAS,EAAE,CAAC;YACjC,KAAK,EAAE,CAAC;YACR,OAAO,CAAC,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC;YACpC,SAAS;QACX,CAAC;QACD,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC;IAC5B,CAAC;IACD,KAAK,EAAE,CAAC;IACR,OAAO,OAAO,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;AAChC,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,OAAO,CAAC,KAAa;IACnC,OAAO,KAAK,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,KAAK,EAAE,CAAC,CAAC;AAC7D,CAAC"}

package/dist/format/slug.d.ts

@@ -1,132 +1,7 @@
-import { type NumberStyle } from "../decode/convert.js";
-import type { SandhiOptions } from "../decode/sandhi.js";
 import type { Dictionary } from "../dictionary/dictionary.js";
-import type { Locale } from "../script/script.js";
-/**
- * Whether a slug carries its tones.
- *
- * `numbers` writes them as digits — 中文 is `zhong1wen2` — and `none` leaves
- * them off. Tones are on by default because dropping them collides: there are
- * roughly 400 toneless syllables against 1300 toned ones, so a toneless slug
- * runs 树, 书 and 输 together as `shu`. They are not a guarantee, though —
- * 权利 and 权力 are both `quan2li4`, and only {@link SlugOptions.hash} tells
- * those apart.
- */
-export type SlugTones = "numbers" | "none";
-/**
- * Whether the syllables inside one word are separated.
- *
- * `join` writes a word as a word — 中文 is `zhong1wen2` — which is what 正词法
- * asks for and what a reader expects in a URL. `separate` cuts every syllable
- * apart, `zhong1-wen2`, for a caller matching a slug syllable by syllable.
- */
-export type SlugSyllables = "join" | "separate";
-/**
- * How ü is written where the alphabet has no room for it.
- *
- * `v` is the input-method convention and keeps 绿 `lv4` apart from 路 `lu4`;
- * `u` is prettier and merges them.
- */
-export type SlugUmlaut = "v" | "u";
-/**
- * How a slug is built.
- */
-export interface SlugOptions {
-    /** Whether tones are written. Defaults to `numbers`. */
-    readonly tones?: SlugTones;
-    /**
-     * What goes between words. Defaults to `-`.
-     *
-     * Written as given rather than checked, because the separator is the one
-     * place a caller decides what kind of identifier this is: `-` for a URL, `_`
-     * for a name in code, and the empty string for a key to match against.
-     */
-    readonly separator?: string;
-    /** Whether the syllables within a word are separated. Defaults to `join`. */
-    readonly syllables?: SlugSyllables;
-    /** How ü is written. Defaults to `v`. */
-    readonly umlaut?: SlugUmlaut;
-    /**
-     * What to do with the digits in a text. Defaults to `keep`.
-     *
-     * The opposite of what a conversion defaults to, and deliberately: a slug for
-     * iPhone 15 评测 should be `iphone-15-ping2ce4` rather than
-     * `iphone-shi2wu3-ping2ce4`, because the digits are how anyone would look for
-     * it. `read` says them, for a caller that wants the text spoken.
-     */
-    readonly numbers?: NumberStyle;
-    /**
-     * A short hash of the source text on the end. Defaults to none.
-     *
-     * `true` for {@link DEFAULT_HASH_LENGTH} characters, or a length of your own,
-     * up to {@link LONGEST_HASH}. See {@link hashOf} for what is hashed and why
-     * it is the hanzi rather than the pinyin.
-     *
-     * Unconditional rather than only where two slugs collide: a function that
-     * cannot see the rest of your corpus cannot know that they did. It narrows
-     * collisions rather than closing them, and checking the slug you generated
-     * against the ones you have stored is still yours to do.
-     */
-    readonly hash?: number | boolean;
-    /**
-     * The longest the slug may be, counting any hash. Defaults to no limit.
-     *
-     * Cut at a word boundary, so a limit lands between words rather than through
-     * the middle of one. A single word longer than the limit is cut where the
-     * limit falls, since there is nowhere else to cut it, and the hash is never
-     * what gets dropped.
-     */
-    readonly maxLength?: number;
-    /**
-     * What to write where a text slugs to nothing. Defaults to the empty string.
-     *
-     * 《》！ has no letters in it, and neither does an empty text.
-     */
-    readonly fallback?: string;
-    /** Which reading standard to use. Defaults to `zh-CN`. */
-    readonly locale?: Locale;
-    /** Which tone sandhi to apply. Defaults to what a conversion applies. */
-    readonly sandhi?: SandhiOptions;
-}
-/**
- * How many characters {@link SlugOptions.hash} writes when simply asked for.
- *
- * Four base-36 characters is 1.7 million values, which is far more than it
- * looks: two texts only collide when their slugs *and* their hashes match, so
- * the suffix is only ever telling apart the handful of texts sharing one slug,
- * not the whole corpus.
- */
-export declare const DEFAULT_HASH_LENGTH = 4;
-/**
- * The longest hash there is anything to fill.
- *
- * {@link hashOf} is 32 bits wide, and 36⁷ is the last power of 36 under 2³²:
- * asking for an eighth character would add a place that some values can never
- * reach, which looks like more uniqueness while being none.
- */
-export declare const LONGEST_HASH = 7;
-/**
- * A 32-bit hash of a text, written in the letters and digits a slug allows.
- *
- * **The source text is what is hashed, not the pinyin.** Two texts needing a
- * hash to tell them apart are, by definition, ones the pinyin already ran
- * together: 权利 and 权力 are both `quan2li4`, and hashing that would give them
- * the same suffix and settle nothing. Hashing the hanzi also means the suffix
- * does not move when a later release reads a word differently, so a slug's tail
- * is the stable half of it.
- *
- * FNV-1a with murmur3's finalizer, which is neither cryptographic nor trying to
- * be: nothing here resists an attacker, and all that is being asked of it is
- * that two texts a reader would call different land on different suffixes.
- * Written out rather than reached for because the package has no runtime
- * dependencies and the core imports no Node built-ins, and `crypto.subtle` —
- * the one hash a browser ships — is asynchronous, which this is not.
- *
- * Normalised first, so that text typed as a decomposed sequence hashes as the
- * same text once composed. 简体 and 繁體 are deliberately *not* folded together:
- * a caller who wrote 头发 and 頭髮 separately meant them separately.
- */
-export declare function hashOf(text: string, length: number): string;
+import type { SlugOptions } from "./slug-options.js";
+export { DEFAULT_HASH_LENGTH, hashOf, LONGEST_HASH } from "./slug-hash.js";
+export type { SlugOptions, SlugSyllables, SlugTones, SlugUmlaut, } from "./slug-options.js";
 /**
  * Convert hanzi to a URL-safe slug.
  *

package/dist/format/slug.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"slug.d.ts","sourceRoot":"","sources":["../../src/format/slug.ts"],"names":[],"mappings":"AAAA,OAAO,EAGL,KAAK,WAAW,EACjB,MAAM,sBAAsB,CAAC;AAC9B,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,qBAAqB,CAAC;AACzD,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,6BAA6B,CAAC;AAE9D,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,qBAAqB,CAAC;AAGlD;;;;;;;;;GASG;AACH,MAAM,MAAM,SAAS,GAAG,SAAS,GAAG,MAAM,CAAC;AAE3C;;;;;;GAMG;AACH,MAAM,MAAM,aAAa,GAAG,MAAM,GAAG,UAAU,CAAC;AAEhD;;;;;GAKG;AACH,MAAM,MAAM,UAAU,GAAG,GAAG,GAAG,GAAG,CAAC;AAEnC;;GAEG;AACH,MAAM,WAAW,WAAW;IAC1B,wDAAwD;IACxD,QAAQ,CAAC,KAAK,CAAC,EAAE,SAAS,CAAC;IAC3B;;;;;;OAMG;IACH,QAAQ,CAAC,SAAS,CAAC,EAAE,MAAM,CAAC;IAC5B,6EAA6E;IAC7E,QAAQ,CAAC,SAAS,CAAC,EAAE,aAAa,CAAC;IACnC,yCAAyC;IACzC,QAAQ,CAAC,MAAM,CAAC,EAAE,UAAU,CAAC;IAC7B;;;;;;;OAOG;IACH,QAAQ,CAAC,OAAO,CAAC,EAAE,WAAW,CAAC;IAC/B;;;;;;;;;;;OAWG;IACH,QAAQ,CAAC,IAAI,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC;IACjC;;;;;;;OAOG;IACH,QAAQ,CAAC,SAAS,CAAC,EAAE,MAAM,CAAC;IAC5B;;;;OAIG;IACH,QAAQ,CAAC,QAAQ,CAAC,EAAE,MAAM,CAAC;IAC3B,0DAA0D;IAC1D,QAAQ,CAAC,MAAM,CAAC,EAAE,MAAM,CAAC;IACzB,yEAAyE;IACzE,QAAQ,CAAC,MAAM,CAAC,EAAE,aAAa,CAAC;CACjC;AAED;;;;;;;GAOG;AACH,eAAO,MAAM,mBAAmB,IAAI,CAAC;AAErC;;;;;;GAMG;AACH,eAAO,MAAM,YAAY,IAAI,CAAC;AAwK9B;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,wBAAgB,MAAM,CAAC,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,GAAG,MAAM,CAgB3D;AA0BD;;;;;;;;;;;;;;;;;;;;;;;;GAwBG;AACH,wBAAgB,IAAI,CAClB,UAAU,EAAE,UAAU,EACtB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,WAAgB,GACxB,MAAM,CAuBR"}
+{"version":3,"file":"slug.d.ts","sourceRoot":"","sources":["../../src/format/slug.ts"],"names":[],"mappings":"AAQA,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,6BAA6B,CAAC;AAE9D,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAGrD,OAAO,EAAE,mBAAmB,EAAE,MAAM,EAAE,YAAY,EAAE,MAAM,gBAAgB,CAAC;AAC3E,YAAY,EACV,WAAW,EACX,aAAa,EACb,SAAS,EACT,UAAU,GACX,MAAM,mBAAmB,CAAC;AAiD3B;;;;;;;;;;;;;;;;;;;;;;;;GAwBG;AACH,wBAAgB,IAAI,CAClB,UAAU,EAAE,UAAU,EACtB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,WAAgB,GACxB,MAAM,CAuBR"}