@kensio/pinyinjs 1.7.4 → 1.7.6

package/dist/decode/convert.d.ts

@@ -1,122 +1,13 @@
-import type { Dictionary } from "../dictionary/dictionary.js";
-import { type ApostropheStyle } from "../orthography/apostrophe.js";
-import { type CapitalStyle } from "../orthography/capitals.js";
-import { type PunctuationStyle } from "../orthography/punctuation.js";
-import type { Locale } from "../script/script.js";
-import { type Syllable, type ToneNotation } from "../syllable/syllable.js";
-import type { ReadingConfidence } from "./confidence.js";
-import { type ReadingHints } from "./hints.js";
-import { type SandhiOptions } from "./sandhi.js";
-/**
- * How a conversion should be carried out and written.
- */
-export interface ConvertOptions {
-    /** Which reading standard to use. Defaults to `zh-CN`. */
-    readonly locale?: Locale;
-    /** How tones are written. Defaults to diacritics. */
-    readonly notation?: ToneNotation;
-    /** Which tone sandhi to apply. */
-    readonly sandhi?: SandhiOptions;
-    /** When the 隔音符号 is written. Defaults to `always`. */
-    readonly apostrophe?: ApostropheStyle;
-    /** Which capitals are written. Defaults to `auto`. */
-    readonly capitals?: CapitalStyle;
-    /** Whether Chinese punctuation is rewritten. Defaults to `latin`. */
-    readonly punctuation?: PunctuationStyle;
-    /** Whether GB/T 16159 word grouping is applied. Defaults to true. */
-    readonly grouping?: boolean;
-    /**
-     * What to do with the digits in a text. Defaults to `read`.
-     *
-     * `read` says them: 我有3个 is `wǒ yǒu sān gè` and 1998年 is
-     * `yī jiǔ jiǔ bā nián`. `keep` leaves every digit exactly as it was written,
-     * which is what this did before there was anything to read them with.
-     */
-    readonly numbers?: NumberStyle;
-    /**
-     * Readings the caller asserts, over whatever the sources say.
-     *
-     * No rule can settle every polyphone, and some texts are genuinely ambiguous
-     * — 孩子越长越漂亮 grows where 头发越长越漂亮 lengthens — so an application
-     * that knows its own content can say what this one could only guess at.
-     *
-     * ```ts
-     * convert(dictionary, "这篇文章不太长。", { readings: { 太长: "tài cháng" } });
-     * ```
-     *
-     * A hint displaces whatever covered exactly the characters it names, and
-     * competes normally with everything else, so a bare `长` hint leaves 校长 as
-     * `xiàozhǎng` — the longer word is still the better reading of that stretch.
-     * See {@link ReadingHints}.
-     */
-    readonly readings?: ReadingHints;
-}
-/**
- * What a conversion does with the digits it meets.
- */
-export type NumberStyle = "read" | "keep";
 /**
- * One piece of a conversion: a syllable, or the text between two of them.
+ * Converting hanzi to pinyin, in every form a caller may want it.
  *
- * A conversion is assembled piece by piece and joined at the very end, so that
- * a caller wanting to render each syllable separately — colouring its tone,
- * marking it uncertain — is not left trying to find the syllables again in a
- * finished string. {@link convert} is this, joined.
+ * Six entry points over one pipeline, differing only in which decoder runs and
+ * whether the answer is joined back into a string. The pipeline itself is
+ * `convert-pipeline.ts`.
  */
-export interface ConvertedPiece {
-    readonly text: string;
-    /** The syllable this piece writes, or undefined where it writes none. */
-    readonly syllable: Syllable | undefined;
-    /**
-     * The characters this piece reads, where it names any of its own.
-     *
-     * What an annotation needs and a plain conversion throws away: to write 银行
-     * with `yín` over 银 and `háng` over 行, something has to say which characters
-     * each syllable is a reading *of*, and that is not the syllable's position.
-     * A reading is not one syllable per character — 玩儿 is two characters and
-     * the single syllable `wánr` — and a read number is not made of characters at
-     * all in the same way, 95% being `bǎifēnzhījiǔshíwǔ` over eight syllables and
-     * three written characters.
-     *
-     * So a piece either names the characters it reads or continues the ones the
-     * piece before it named, and the two are told apart by the syllable rather
-     * than by a second flag:
-     *
-     * | `syllable` | `source` | The piece |
-     * | --- | --- | --- |
-     * | undefined | undefined | writes no reading: a space, or a non-Han run |
-     * | set | set | reads exactly those characters |
-     * | set | undefined | reads on into the characters named before it |
-     *
-     * {@link import("../format/html.js").toAnnotatedHtml} is what consumes it.
-     */
-    readonly source: string | undefined;
-    /**
-     * How settled that syllable was, where the decode reported it.
-     *
-     * Only {@link convertPieces} fills this in, and only for a syllable the
-     * lattice decoded: the greedy baseline cannot say what it rejected, and a
-     * Taiwan reading that differs in length from its mainland form cannot be
-     * lined up with it.
-     */
-    readonly confidence: ReadingConfidence | undefined;
-}
-/**
- * Which characters of a word each of its syllables reads.
- *
- * One per syllable, in order, and undefined where a syllable reads on into the
- * characters the one before it took. Only 儿化 does that: measured over the
- * committed dictionary, 5,283 of 723,147 keys have fewer syllables than
- * characters, 4,024 of them because a final 儿 folds into the syllable before
- * it, and the remaining 1,259 because the word is written with punctuation —
- * which cannot reach here, since punctuation ends a Han run before the decode
- * ever sees it.
- *
- * The deficit is spent from the right, which is where 儿化 always is. 打哪儿指
- * 哪儿 is the case that needs it: two syllables short over six characters, and
- * both 儿 attach to the character in front of them rather than one taking two.
- */
-export declare function sourcesOf(word: string, reading: readonly Syllable[]): readonly (string | undefined)[];
+import type { Dictionary } from "../dictionary/dictionary.js";
+import type { ConvertedPiece, ConvertOptions } from "./pieces.js";
+export { type ConvertedPiece, type ConvertOptions, type NumberStyle, sourcesOf, } from "./pieces.js";
 /**
  * Convert hanzi to pinyin with the lattice decoder.
  *
@@ -133,8 +24,8 @@ export declare function convert(dictionary: Dictionary, text: string, options?:
  * The same conversion {@link convert} performs and the same text once joined,
  * kept in pieces so that each syllable can be rendered on its own terms.
  * Everything the decode rejected is reported with it — see
- * {@link ReadingConfidence} — which is what an output format needs to show a
- * reader where it was guessing.
+ * {@link import("./confidence.js").ReadingConfidence} — which is what an
+ * output format needs to show a reader where it was guessing.
  *
  * Costs a second sweep of the lattice, which is why it is not what
  * {@link convert} runs.

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

@@ -1 +1 @@
-{"version":3,"file":"convert.d.ts","sourceRoot":"","sources":["../../src/decode/convert.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,6BAA6B,CAAC;AAG9D,OAAO,EAAE,KAAK,eAAe,EAAY,MAAM,8BAA8B,CAAC;AAC9E,OAAO,EAGL,KAAK,YAAY,EAElB,MAAM,4BAA4B,CAAC;AAEpC,OAAO,EACL,KAAK,gBAAgB,EAEtB,MAAM,+BAA+B,CAAC;AAOvC,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,qBAAqB,CAAC;AAClD,OAAO,EACL,KAAK,QAAQ,EACb,KAAK,YAAY,EAElB,MAAM,yBAAyB,CAAC;AACjC,OAAO,KAAK,EAAE,iBAAiB,EAAE,MAAM,iBAAiB,CAAC;AAGzD,OAAO,EAEL,KAAK,YAAY,EAGlB,MAAM,YAAY,CAAC;AAIpB,OAAO,EAAe,KAAK,aAAa,EAAE,MAAM,aAAa,CAAC;AAG9D;;GAEG;AACH,MAAM,WAAW,cAAc;IAC7B,0DAA0D;IAC1D,QAAQ,CAAC,MAAM,CAAC,EAAE,MAAM,CAAC;IACzB,qDAAqD;IACrD,QAAQ,CAAC,QAAQ,CAAC,EAAE,YAAY,CAAC;IACjC,kCAAkC;IAClC,QAAQ,CAAC,MAAM,CAAC,EAAE,aAAa,CAAC;IAChC,sDAAsD;IACtD,QAAQ,CAAC,UAAU,CAAC,EAAE,eAAe,CAAC;IACtC,sDAAsD;IACtD,QAAQ,CAAC,QAAQ,CAAC,EAAE,YAAY,CAAC;IACjC,qEAAqE;IACrE,QAAQ,CAAC,WAAW,CAAC,EAAE,gBAAgB,CAAC;IACxC,qEAAqE;IACrE,QAAQ,CAAC,QAAQ,CAAC,EAAE,OAAO,CAAC;IAC5B;;;;;;OAMG;IACH,QAAQ,CAAC,OAAO,CAAC,EAAE,WAAW,CAAC;IAC/B;;;;;;;;;;;;;;;OAeG;IACH,QAAQ,CAAC,QAAQ,CAAC,EAAE,YAAY,CAAC;CAClC;AAED;;GAEG;AACH,MAAM,MAAM,WAAW,GAAG,MAAM,GAAG,MAAM,CAAC;AAE1C;;;;;;;GAOG;AACH,MAAM,WAAW,cAAc;IAC7B,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,yEAAyE;IACzE,QAAQ,CAAC,QAAQ,EAAE,QAAQ,GAAG,SAAS,CAAC;IACxC;;;;;;;;;;;;;;;;;;;;;;OAsBG;IACH,QAAQ,CAAC,MAAM,EAAE,MAAM,GAAG,SAAS,CAAC;IACpC;;;;;;;OAOG;IACH,QAAQ,CAAC,UAAU,EAAE,iBAAiB,GAAG,SAAS,CAAC;CACpD;AAiFD;;;;;;;;;;;;;;GAcG;AACH,wBAAgB,SAAS,CACvB,IAAI,EAAE,MAAM,EACZ,OAAO,EAAE,SAAS,QAAQ,EAAE,GAC3B,SAAS,CAAC,MAAM,GAAG,SAAS,CAAC,EAAE,CA2BjC;AA0fD;;;;;;;;GAQG;AACH,wBAAgB,OAAO,CACrB,UAAU,EAAE,UAAU,EACtB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,cAAmB,GAC3B,MAAM,CAER;AAED;;;;;;;;;;;GAWG;AACH,wBAAgB,aAAa,CAC3B,UAAU,EAAE,UAAU,EACtB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,cAAmB,GAC3B,SAAS,cAAc,EAAE,CAE3B;AAED;;;;;;;;GAQG;AACH,wBAAgB,qBAAqB,CACnC,UAAU,EAAE,UAAU,EACtB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,cAAmB,GAC3B,SAAS,cAAc,EAAE,CAE3B;AAED;;GAEG;AACH,wBAAgB,UAAU,CAAC,MAAM,EAAE,SAAS,cAAc,EAAE,GAAG,MAAM,CAEpE;AAED;;;;;;GAMG;AACH,wBAAgB,eAAe,CAC7B,UAAU,EAAE,UAAU,EACtB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,cAAmB,GAC3B,MAAM,CAER;AAED;;;;;;GAMG;AACH,wBAAgB,qBAAqB,CACnC,UAAU,EAAE,UAAU,EACtB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,cAAmB,GAC3B,SAAS,cAAc,EAAE,CAE3B"}
+{"version":3,"file":"convert.d.ts","sourceRoot":"","sources":["../../src/decode/convert.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,6BAA6B,CAAC;AAE9D,OAAO,KAAK,EAAE,cAAc,EAAE,cAAc,EAAE,MAAM,aAAa,CAAC;AAGlE,OAAO,EACL,KAAK,cAAc,EACnB,KAAK,cAAc,EACnB,KAAK,WAAW,EAChB,SAAS,GACV,MAAM,aAAa,CAAC;AAErB;;;;;;;;GAQG;AACH,wBAAgB,OAAO,CACrB,UAAU,EAAE,UAAU,EACtB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,cAAmB,GAC3B,MAAM,CAER;AAED;;;;;;;;;;;GAWG;AACH,wBAAgB,aAAa,CAC3B,UAAU,EAAE,UAAU,EACtB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,cAAmB,GAC3B,SAAS,cAAc,EAAE,CAE3B;AAED;;;;;;;;GAQG;AACH,wBAAgB,qBAAqB,CACnC,UAAU,EAAE,UAAU,EACtB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,cAAmB,GAC3B,SAAS,cAAc,EAAE,CAE3B;AAED;;GAEG;AACH,wBAAgB,UAAU,CAAC,MAAM,EAAE,SAAS,cAAc,EAAE,GAAG,MAAM,CAEpE;AAED;;;;;;GAMG;AACH,wBAAgB,eAAe,CAC7B,UAAU,EAAE,UAAU,EACtB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,cAAmB,GAC3B,MAAM,CAER;AAED;;;;;;GAMG;AACH,wBAAgB,qBAAqB,CACnC,UAAU,EAAE,UAAU,EACtB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,cAAmB,GAC3B,SAAS,cAAc,EAAE,CAE3B"}

package/dist/decode/convert.js

@@ -1,454 +1,6 @@
-import { isErCharacter, isErFinal } from "../dictionary/erhua.js";
-import { isSameReading } from "../dictionary/entry.js";
-import { markWord } from "../orthography/apostrophe.js";
-import { capitaliseSentenceParts, capitaliseWord, isSentence, } from "../orthography/capitals.js";
-import { applyGrouping } from "../orthography/grouping.js";
-import { toLatinPunctuationParts, } from "../orthography/punctuation.js";
-import { readNumbersIn, saidNumeral, } from "../numerals/text.js";
-import { toCharacters } from "../script/characters.js";
-import { writeSyllable, } from "../syllable/syllable.js";
-import { divisionOf } from "./constituents.js";
-import { decodeRun, decodeRunScored } from "./decode.js";
-import { hintsWithin, resolveHints, } from "./hints.js";
-import { decodeGreedily } from "./greedy.js";
-import { READING_RULES } from "./reading-rules.js";
-import { splitRuns, surroundingCharacters } from "./runs.js";
-import { applySandhi } from "./sandhi.js";
-/**
- * The reading a word takes in a locale.
- *
- * `zh-TW` is stored as a delta, so a word with no Taiwan reading simply reads
- * the same in both.
- *
- * The delta is measured against the entry's own 普通话 reading, so it only
- * applies where the decode settled on that reading. A polyphone the decode read
- * some other way — 长 as `cháng` where its entry reads `zhǎng`, 差 as `chā`
- * where its entry reads `chà` — is a different sense of the word, and the delta
- * beside the entry says nothing about it. Replacing the reading there would
- * discard what the lattice worked out in favour of a reading of the wrong
- * syllable entirely.
- */
-function readingFor(dictionary, word, locale) {
-    if (locale !== "zh-TW") {
-        return word.reading;
-    }
-    const entry = dictionary.lookup(word.text);
-    if (entry?.taiwanReading === undefined) {
-        return word.reading;
-    }
-    return isSameReading(entry.reading, word.reading)
-        ? entry.taiwanReading
-        : word.reading;
-}
-/**
- * Text that writes no syllable: a space, or a run that was never Han.
- */
-function plainPiece(text) {
-    return {
-        text,
-        syllable: undefined,
-        confidence: undefined,
-        source: undefined,
-    };
-}
-/**
- * Source text that is written out but read as nothing.
- *
- * The difference from {@link plainPiece} is what the text *is*. Both write no
- * syllable, but a run that was never Han — punctuation, a Latin word, a digit
- * left unread — stands for something the author wrote, whereas the space
- * between two words and the hyphen inside 干干净净 are pinyin orthography that
- * the source has no trace of.
- *
- * A conversion joins them into one string and the distinction does not arise.
- * An annotation writes the source and the reading in different places, and
- * putting a hyphen the pinyin needs into the hanzi it annotates would be
- * inventing text: 干干净净 is `gāngān-jìngjìng` and is not written 干干-净净.
- *
- * The text and the source can differ, and the punctuation pass is why: it
- * rewrites 。 to a full stop in the `text` a conversion writes, and leaves the
- * `source` as the mark the author actually typed.
- */
-function sourcePiece(text) {
-    return {
-        text,
-        syllable: undefined,
-        confidence: undefined,
-        source: text,
-    };
-}
-/**
- * Which characters of a word each of its syllables reads.
- *
- * One per syllable, in order, and undefined where a syllable reads on into the
- * characters the one before it took. Only 儿化 does that: measured over the
- * committed dictionary, 5,283 of 723,147 keys have fewer syllables than
- * characters, 4,024 of them because a final 儿 folds into the syllable before
- * it, and the remaining 1,259 because the word is written with punctuation —
- * which cannot reach here, since punctuation ends a Han run before the decode
- * ever sees it.
- *
- * The deficit is spent from the right, which is where 儿化 always is. 打哪儿指
- * 哪儿 is the case that needs it: two syllables short over six characters, and
- * both 儿 attach to the character in front of them rather than one taking two.
- */
-export function sourcesOf(word, reading) {
-    const characters = toCharacters(word);
-    if (characters.length === reading.length) {
-        return characters;
-    }
-    // Anything this does not understand is one base for the whole word, which is
-    // always true and merely less useful than naming each syllable's characters.
-    if (characters.length < reading.length || !isErFinal(word)) {
-        return reading.map((_, at) => (at === 0 ? word : undefined));
-    }
-    const sources = [];
-    let deficit = characters.length - reading.length;
-    for (const character of characters) {
-        const previous = sources.at(-1);
-        if (deficit > 0 && previous !== undefined && isErCharacter(character)) {
-            sources[sources.length - 1] = previous + character;
-            deficit--;
-            continue;
-        }
-        sources.push(character);
-    }
-    /* c8 ignore next 3 -- the deficit is 儿 by construction, but a source that
-       wrote one somewhere else would otherwise misalign every later syllable */
-    return sources.length === reading.length
-        ? sources
-        : reading.map((_, at) => (at === 0 ? word : undefined));
-}
-/**
- * Write one decoded word as one piece per syllable.
- *
- * Only proper nouns are capitalised here; a sentence capital is applied to the
- * whole conversion afterwards, since it belongs to whichever run happens to
- * start the sentence.
- */
-function writeWord(reading, confidence, word, written) {
-    if (reading.length === 0) {
-        return [sourcePiece(word.text)];
-    }
-    // A tone number already ends its syllable, raised or not, so `xi1an1` cannot
-    // be misread and the 隔音符号 would only be noise.
-    const isNumbered = written.notation === "numbers" || written.notation === "superscript";
-    const spellings = markWord(reading.map((syllable) => writeSyllable(syllable, written.notation)), isNumbered ? "never" : written.apostrophe);
-    const isCapitalised = word.isProperNoun && written.capitals !== "none";
-    const sources = sourcesOf(word.text, reading);
-    return spellings.map((spelling, at) => ({
-        text: at === 0 && isCapitalised ? capitaliseWord(spelling) : spelling,
-        syllable: reading[at],
-        confidence: confidence[at],
-        source: sources[at],
-    }));
-}
-/**
- * A decoder that reports no confidence at all, which is every decoder but one.
- */
-function unscored(decode) {
-    return (dictionary, run, before, hints) => decode(dictionary, run, before, hints).map((word) => ({
-        word,
-        confidence: [],
-    }));
-}
-/**
- * The lattice decoder, reporting nothing about its own confidence.
- */
-const LATTICE = unscored((dictionary, run, before, hints) => decodeRun(dictionary, run, READING_RULES, before, hints));
-/**
- * The lattice decoder, with what each reading was chosen over.
- */
-const SCORED = (dictionary, run, before, hints) => decodeRunScored(dictionary, run, READING_RULES, before, hints);
-/**
- * The greedy baseline, which has nothing to report either way.
- *
- * The context in front of a run goes unused: longest-match has no way to weigh
- * one segmentation against another, which is the whole reason it is the
- * baseline, and a 汉字 it cannot report would only be another thing to trim.
- */
-const GREEDY = unscored(decodeGreedily);
-/**
- * Write one Han run's worth of decoded words.
- */
-function writeRun(dictionary, words, locale, written, sandhi) {
-    // Sandhi runs across the whole run rather than within a word, since 不 in one
-    // word assimilates to the tone starting the next. Third-tone sandhi needs to
-    // know where the words are all the same, so the grouping goes with it — and
-    // only when it is asked for, since dividing a word costs lookups.
-    const readings = words.map((scored) => readingFor(dictionary, scored.word, locale));
-    const grouping = sandhi?.thirdTone === true
-        ? words.map((scored, index) => {
-            const reading = readings[index] ?? [];
-            return (divisionOf(dictionary, scored.word.text, reading) ?? reading.length);
-        })
-        : undefined;
-    const flattened = applySandhi(readings.flat(), sandhi, grouping);
-    let at = 0;
-    const pieces = [];
-    for (const [index, scored] of words.entries()) {
-        /* c8 ignore next -- readings is built by mapping over these same words */
-        const length = readings[index]?.length ?? 0;
-        if (index > 0) {
-            // A space, unless 分词连写 wrote a hyphen: 干干净净 is one orthographic
-            // word, `gāngān-jìngjìng`, cut into two decoded ones.
-            pieces.push(plainPiece(scored.word.separator ?? " "));
-        }
-        pieces.push(...writeWord(flattened.slice(at, at + length), 
-        // A locale reading of a different length cannot be lined up with the
-        // confidence reported for the reading it replaced.
-        scored.confidence.length === length ? scored.confidence : [], scored.word, written));
-        at += length;
-    }
-    return pieces;
-}
-/**
- * Rewrite every piece's text with a pass that reads across all of them.
- */
-function rewrite(pieces, pass) {
-    const rewritten = pass(pieces.map((piece) => piece.text));
-    return pieces.map((piece, at) => {
-        /* c8 ignore next -- one part comes back for each part handed over */
-        const text = rewritten[at] ?? piece.text;
-        return text === piece.text ? piece : { ...piece, text };
-    });
-}
-/**
- * The words a Han run decodes to, with 分词连写 applied.
- *
- * Grouping rewrites word boundaries and never the readings behind them, so the
- * syllables — and the confidence beside them — survive it in order.
- */
-function wordsOf(decoded, dictionary, isGrouped) {
-    return isGrouped
-        ? regroup(decoded, applyGrouping(decoded.map((scored) => scored.word), dictionary))
-        : decoded;
-}
-/**
- * What surrounds a run once it has been decoded: the character after it, and
- * the syllable that character is read as.
- *
- * The syllable is the one part of a number's context that a decode has to
- * supply, and it is what a 一 ending the number assimilates to.
- */
-function surrounding(runs, decoded, at) {
-    return {
-        after: {
-            character: surroundingCharacters(runs, at).following,
-            syllable: decoded[at + 1]?.[0]?.word.reading[0],
-        },
-        isAfterHan: runs[at - 1]?.isHan === true,
-    };
-}
-/**
- * The 汉字 a number in front of a Han run stands for, for that run's decode.
- *
- * Only the last segment of the run before, because only that one touches the
- * Han: the D of 3D银行 comes between them, and a decode of 银行 that saw 三
- * beside it would be reading a text nobody wrote.
- */
-function numeralBefore(segments) {
-    return segments.at(-1)?.hanzi ?? "";
-}
-/**
- * Whether a character wants a space between it and a number read out.
- *
- * A letter or a digit does; punctuation does not, so 20%。 keeps its full stop
- * against the number.
- */
-const WORDLIKE = /[\p{L}\p{N}]/u;
-/**
- * Whether two stretches take a space between them once one has been read.
- */
-function isSpaced(before, after) {
-    return WORDLIKE.test(before.at(-1) ?? "") && WORDLIKE.test(after[0] ?? "");
-}
-/**
- * Write a run whose words are already known, a word at a time.
- *
- * Each group is one orthographic word and takes the 隔音符号 within itself, so
- * a time's minutes are `sānshí` rather than `sān shí` — the same grouping the
- * number would get if the text had written 6点30分 out in 汉字.
- */
-function groupedPieces(spelled, said, words, apostrophe) {
-    const pieces = [];
-    let at = 0;
-    for (const length of words) {
-        if (at > 0) {
-            pieces.push(plainPiece(" "));
-        }
-        const group = spelled.slice(at, at + length);
-        for (const [index, text] of markWord(group, apostrophe).entries()) {
-            pieces.push({
-                text,
-                syllable: said[at + index],
-                confidence: undefined,
-                source: undefined,
-            });
-        }
-        at += length;
-    }
-    return pieces;
-}
-/**
- * Write a number's syllables as the pieces they are written with.
- *
- * A counted number is *one word*, which is what 正词法 6.1.5 asks for: 123 is
- * `yībǎi'èrshísān` and not three words, so the syllables run together and take
- * the 隔音符号 where one is needed. A number read out digit by digit is not a
- * word at all — it is digits — so those are written apart: 1998年 is
- * `yī jiǔ jiǔ bā nián`.
- */
-function numberPieces(said, segment, written) {
-    const spelled = said.map((syllable) => writeSyllable(syllable, written.notation));
-    if (segment.style === "digits") {
-        return read(spelled.flatMap((text, at) => [
-            ...(at === 0 ? [] : [plainPiece(" ")]),
-            { text, syllable: said[at], confidence: undefined, source: undefined },
-        ]), segment.text);
-    }
-    const isNumbered = written.notation === "numbers" || written.notation === "superscript";
-    const apostrophe = isNumbered ? "never" : written.apostrophe;
-    // A run that says where its words break gets them: a time is `liù diǎn
-    // sānshí fēn` and a decimal is `qīshíwǔ diǎn wǔ`, each counted part a word
-    // of its own and everything after the 点 a digit at a time.
-    if (segment.words !== undefined) {
-        return read(groupedPieces(spelled, said, segment.words, apostrophe), segment.text);
-    }
-    return read(markWord(spelled, apostrophe).map((text, at) => ({
-        text,
-        syllable: said[at],
-        confidence: undefined,
-        source: undefined,
-    })), segment.text);
-}
-/**
- * Name what a read number is a reading of, once for the whole of it.
- *
- * A number is not read character by character the way a word is: 95% is
- * `bǎifēnzhījiǔshíwǔ` over eight syllables and three written characters, and
- * the order reverses on the way, so no syllable belongs to any one of them.
- * The written form is therefore named once, by the first piece that says
- * anything, and the rest read on into it.
- */
-function read(pieces, source) {
-    const first = pieces.findIndex((piece) => piece.syllable !== undefined);
-    /* c8 ignore next 3 -- a read number has at least one syllable in it */
-    if (first === -1) {
-        return pieces;
-    }
-    return pieces.map((piece, at) => at === first ? { ...piece, source } : piece);
-}
-/**
- * A stand-in for the pinyin either side of a run, which ends in a letter.
- */
-function runEdge(isHan) {
-    return isHan ? "a" : "";
-}
-/**
- * Write a stretch that was never Han, reading the numbers in it.
- *
- * Everything that is not a number goes through exactly as written, which is
- * what this always did: digits are the only part of a non-Han run this package
- * has anything to say about. Once a number *has* been read, though, the whole
- * stretch is being said rather than shown, so its parts take the spacing of
- * words — 3D打印 is `sān D dǎyìn` — and punctuation still takes none.
- */
-function writeNumbers(text, segments, context, written, options) {
-    if (segments.every((segment) => segment.reading === undefined)) {
-        return [sourcePiece(text)];
-    }
-    const pieces = [];
-    let before = runEdge(context.isAfterHan);
-    for (const segment of segments) {
-        if (isSpaced(before, segment.text)) {
-            pieces.push(plainPiece(" "));
-        }
-        pieces.push(...(segment.reading === undefined
-            ? [sourcePiece(segment.text)]
-            : numberPieces(saidNumeral(segment, context.after.syllable, options.sandhi), segment, written)));
-        // What decides the next space is what was *written*, not what was read:
-        // 95% ends in a sign and `bǎifēnzhījiǔshíwǔ` ends in a letter.
-        before = pieces.at(-1)?.text ?? before;
-    }
-    if (isSpaced(before, runEdge(context.after.character !== ""))) {
-        pieces.push(plainPiece(" "));
-    }
-    return pieces;
-}
-/**
- * Run the pipeline over a text with a given decoder.
- */
-function convertWith(decode, dictionary, text, options) {
-    const { locale = "zh-CN", notation = "marks", apostrophe = "always", capitals = "auto", punctuation = "latin", grouping = true, numbers = "read", sandhi, readings, } = options;
-    const written = { notation, apostrophe, capitals };
-    const converted = [];
-    const runs = [...splitRuns(text)];
-    // Parsed once for the whole text rather than per run, so that a malformed
-    // hint is reported whether or not a run happens to reach it.
-    const hints = readings === undefined ? undefined : resolveHints(readings);
-    // Where each run starts, in code points, so a caller counting positions from
-    // the start of the text can be met where the run puts them.
-    const starts = [];
-    let start = 0;
-    for (const run of runs) {
-        starts.push(start);
-        start += toCharacters(run.text).length;
-    }
-    // Read before anything is decoded, because a number needs to know what
-    // follows it — 1998年 is a year and 3个 is a count — and the character that
-    // decides it is there for the reading without a decode of its own.
-    const read = runs.map((run, at) => run.isHan || numbers === "keep"
-        ? []
-        : readNumbersIn(run.text, surroundingCharacters(runs, at)));
-    // Decoded before anything is written, because a number's sandhi needs the
-    // syllable after it, and because a Han run needs the number in front of it:
-    // 2个人 is `liǎng gè rén` and 个人 on its own is the word `gèrén`.
-    const decoded = runs.map((run, at) => {
-        if (!run.isHan) {
-            return [];
-        }
-        const within = hints === undefined
-            ? undefined
-            : hintsWithin(hints, starts[at] ?? 0, toCharacters(run.text).length);
-        const said = numeralBefore(read[at - 1] ?? []);
-        return wordsOf(decode(dictionary, run.text, said, within), dictionary, grouping);
-    });
-    for (const [at, run] of runs.entries()) {
-        const words = decoded[at] ?? [];
-        if (run.isHan) {
-            converted.push(...writeRun(dictionary, words, locale, written, sandhi));
-            continue;
-        }
-        converted.push(...writeNumbers(run.text, read[at] ?? [], surrounding(runs, decoded, at), written, { sandhi }));
-    }
-    let pieces = converted;
-    // Both of these read the whole conversion rather than one run: a sentence
-    // capital belongs to whichever run happens to start the sentence, and a mark
-    // needs to know whether anything follows it before it takes a space.
-    if (capitals === "auto" && isSentence(text)) {
-        pieces = rewrite(pieces, capitaliseSentenceParts);
-    }
-    return punctuation === "latin"
-        ? rewrite(pieces, toLatinPunctuationParts)
-        : pieces;
-}
-/**
- * Redistribute the decode's per-syllable confidence over regrouped words.
- *
- * 分词连写 moves word boundaries — 看 and 了 become 看了 — without touching a
- * reading, so the run's syllables are the same syllables in the same order and
- * are simply cut in different places.
- */
-function regroup(decoded, grouped) {
-    const confidence = decoded.flatMap((scored) => [...scored.confidence]);
-    let at = 0;
-    return grouped.map((word) => {
-        const held = confidence.slice(at, at + word.reading.length);
-        at += word.reading.length;
-        return { word, confidence: held };
-    });
-}
+import { convertWith } from "./convert-pipeline.js";
+import { GREEDY, LATTICE, SCORED } from "./decoders.js";
+export { sourcesOf, } from "./pieces.js";
 /**
  * Convert hanzi to pinyin with the lattice decoder.
  *
@@ -467,8 +19,8 @@ export function convert(dictionary, text, options = {}) {
  * The same conversion {@link convert} performs and the same text once joined,
  * kept in pieces so that each syllable can be rendered on its own terms.
  * Everything the decode rejected is reported with it — see
- * {@link ReadingConfidence} — which is what an output format needs to show a
- * reader where it was guessing.
+ * {@link import("./confidence.js").ReadingConfidence} — which is what an
+ * output format needs to show a reader where it was guessing.
  *
  * Costs a second sweep of the lattice, which is why it is not what
  * {@link convert} runs.

package/dist/decode/convert.js.map

@@ -1 +1 @@
-{"version":3,"file":"convert.js","sourceRoot":"","sources":["../../src/decode/convert.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,aAAa,EAAE,SAAS,EAAE,MAAM,wBAAwB,CAAC;AAClE,OAAO,EAAE,aAAa,EAAE,MAAM,wBAAwB,CAAC;AACvD,OAAO,EAAwB,QAAQ,EAAE,MAAM,8BAA8B,CAAC;AAC9E,OAAO,EACL,uBAAuB,EACvB,cAAc,EAEd,UAAU,GACX,MAAM,4BAA4B,CAAC;AACpC,OAAO,EAAE,aAAa,EAAE,MAAM,4BAA4B,CAAC;AAC3D,OAAO,EAEL,uBAAuB,GACxB,MAAM,+BAA+B,CAAC;AACvC,OAAO,EAEL,aAAa,EACb,WAAW,GACZ,MAAM,qBAAqB,CAAC;AAC7B,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAEvD,OAAO,EAGL,aAAa,GACd,MAAM,yBAAyB,CAAC;AAEjC,OAAO,EAAE,UAAU,EAAE,MAAM,mBAAmB,CAAC;AAC/C,OAAO,EAAE,SAAS,EAAE,eAAe,EAAE,MAAM,aAAa,CAAC;AACzD,OAAO,EACL,WAAW,EAEX,YAAY,GAEb,MAAM,YAAY,CAAC;AACpB,OAAO,EAAE,cAAc,EAAE,MAAM,aAAa,CAAC;AAC7C,OAAO,EAAE,aAAa,EAAE,MAAM,oBAAoB,CAAC;AACnD,OAAO,EAAE,SAAS,EAAE,qBAAqB,EAAgB,MAAM,WAAW,CAAC;AAC3E,OAAO,EAAE,WAAW,EAAsB,MAAM,aAAa,CAAC;AA6G9D;;;;;;;;;;;;;GAaG;AACH,SAAS,UAAU,CACjB,UAAsB,EACtB,IAAiB,EACjB,MAAc;IAEd,IAAI,MAAM,KAAK,OAAO,EAAE,CAAC;QACvB,OAAO,IAAI,CAAC,OAAO,CAAC;IACtB,CAAC;IACD,MAAM,KAAK,GAAG,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAC3C,IAAI,KAAK,EAAE,aAAa,KAAK,SAAS,EAAE,CAAC;QACvC,OAAO,IAAI,CAAC,OAAO,CAAC;IACtB,CAAC;IACD,OAAO,aAAa,CAAC,KAAK,CAAC,OAAO,EAAE,IAAI,CAAC,OAAO,CAAC;QAC/C,CAAC,CAAC,KAAK,CAAC,aAAa;QACrB,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC;AACnB,CAAC;AAED;;GAEG;AACH,SAAS,UAAU,CAAC,IAAY;IAC9B,OAAO;QACL,IAAI;QACJ,QAAQ,EAAE,SAAS;QACnB,UAAU,EAAE,SAAS;QACrB,MAAM,EAAE,SAAS;KAClB,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;;;;;;GAiBG;AACH,SAAS,WAAW,CAAC,IAAY;IAC/B,OAAO;QACL,IAAI;QACJ,QAAQ,EAAE,SAAS;QACnB,UAAU,EAAE,SAAS;QACrB,MAAM,EAAE,IAAI;KACb,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;;;GAcG;AACH,MAAM,UAAU,SAAS,CACvB,IAAY,EACZ,OAA4B;IAE5B,MAAM,UAAU,GAAG,YAAY,CAAC,IAAI,CAAC,CAAC;IACtC,IAAI,UAAU,CAAC,MAAM,KAAK,OAAO,CAAC,MAAM,EAAE,CAAC;QACzC,OAAO,UAAU,CAAC;IACpB,CAAC;IACD,6EAA6E;IAC7E,6EAA6E;IAC7E,IAAI,UAAU,CAAC,MAAM,GAAG,OAAO,CAAC,MAAM,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,EAAE,CAAC;QAC3D,OAAO,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC;IAC/D,CAAC;IAED,MAAM,OAAO,GAAa,EAAE,CAAC;IAC7B,IAAI,OAAO,GAAG,UAAU,CAAC,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC;IACjD,KAAK,MAAM,SAAS,IAAI,UAAU,EAAE,CAAC;QACnC,MAAM,QAAQ,GAAG,OAAO,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;QAChC,IAAI,OAAO,GAAG,CAAC,IAAI,QAAQ,KAAK,SAAS,IAAI,aAAa,CAAC,SAAS,CAAC,EAAE,CAAC;YACtE,OAAO,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,GAAG,QAAQ,GAAG,SAAS,CAAC;YACnD,OAAO,EAAE,CAAC;YACV,SAAS;QACX,CAAC;QACD,OAAO,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;IAC1B,CAAC;IACD;gFAC4E;IAC5E,OAAO,OAAO,CAAC,MAAM,KAAK,OAAO,CAAC,MAAM;QACtC,CAAC,CAAC,OAAO;QACT,CAAC,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC;AAC5D,CAAC;AAED;;;;;;GAMG;AACH,SAAS,SAAS,CAChB,OAA4B,EAC5B,UAAwC,EACxC,IAAiB,EACjB,OAAgB;IAEhB,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACzB,OAAO,CAAC,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;IAClC,CAAC;IACD,6EAA6E;IAC7E,+CAA+C;IAC/C,MAAM,UAAU,GACd,OAAO,CAAC,QAAQ,KAAK,SAAS,IAAI,OAAO,CAAC,QAAQ,KAAK,aAAa,CAAC;IACvE,MAAM,SAAS,GAAG,QAAQ,CACxB,OAAO,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,aAAa,CAAC,QAAQ,EAAE,OAAO,CAAC,QAAQ,CAAC,CAAC,EACpE,UAAU,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,OAAO,CAAC,UAAU,CAC1C,CAAC;IACF,MAAM,aAAa,GAAG,IAAI,CAAC,YAAY,IAAI,OAAO,CAAC,QAAQ,KAAK,MAAM,CAAC;IACvE,MAAM,OAAO,GAAG,SAAS,CAAC,IAAI,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IAE9C,OAAO,SAAS,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,EAAE,EAAE,CAAC,CAAC;QACtC,IAAI,EAAE,EAAE,KAAK,CAAC,IAAI,aAAa,CAAC,CAAC,CAAC,cAAc,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,QAAQ;QACrE,QAAQ,EAAE,OAAO,CAAC,EAAE,CAAC;QACrB,UAAU,EAAE,UAAU,CAAC,EAAE,CAAC;QAC1B,MAAM,EAAE,OAAO,CAAC,EAAE,CAAC;KACpB,CAAC,CAAC,CAAC;AACN,CAAC;AAaD;;GAEG;AACH,SAAS,QAAQ,CACf,MAK2B;IAE3B,OAAO,CAAC,UAAU,EAAE,GAAG,EAAE,MAAM,EAAE,KAAK,EAAE,EAAE,CACxC,MAAM,CAAC,UAAU,EAAE,GAAG,EAAE,MAAM,EAAE,KAAK,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC;QACpD,IAAI;QACJ,UAAU,EAAE,EAAE;KACf,CAAC,CAAC,CAAC;AACR,CAAC;AAED;;GAEG;AACH,MAAM,OAAO,GAAW,QAAQ,CAAC,CAAC,UAAU,EAAE,GAAG,EAAE,MAAM,EAAE,KAAK,EAAE,EAAE,CAClE,SAAS,CAAC,UAAU,EAAE,GAAG,EAAE,aAAa,EAAE,MAAM,EAAE,KAAK,CAAC,CACzD,CAAC;AAEF;;GAEG;AACH,MAAM,MAAM,GAAW,CAAC,UAAU,EAAE,GAAG,EAAE,MAAM,EAAE,KAAK,EAAE,EAAE,CACxD,eAAe,CAAC,UAAU,EAAE,GAAG,EAAE,aAAa,EAAE,MAAM,EAAE,KAAK,CAAC,CAAC;AAEjE;;;;;;GAMG;AACH,MAAM,MAAM,GAAW,QAAQ,CAAC,cAAc,CAAC,CAAC;AAEhD;;GAEG;AACH,SAAS,QAAQ,CACf,UAAsB,EACtB,KAA4B,EAC5B,MAAc,EACd,OAAgB,EAChB,MAAiC;IAEjC,6EAA6E;IAC7E,6EAA6E;IAC7E,4EAA4E;IAC5E,kEAAkE;IAClE,MAAM,QAAQ,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,EAAE,CACpC,UAAU,CAAC,UAAU,EAAE,MAAM,CAAC,IAAI,EAAE,MAAM,CAAC,CAC5C,CAAC;IACF,MAAM,QAAQ,GACZ,MAAM,EAAE,SAAS,KAAK,IAAI;QACxB,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE;YAC1B,MAAM,OAAO,GAAG,QAAQ,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC;YACtC,OAAO,CACL,UAAU,CAAC,UAAU,EAAE,MAAM,CAAC,IAAI,CAAC,IAAI,EAAE,OAAO,CAAC,IAAI,OAAO,CAAC,MAAM,CACpE,CAAC;QACJ,CAAC,CAAC;QACJ,CAAC,CAAC,SAAS,CAAC;IAChB,MAAM,SAAS,GAAG,WAAW,CAAC,QAAQ,CAAC,IAAI,EAAE,EAAE,MAAM,EAAE,QAAQ,CAAC,CAAC;IAEjE,IAAI,EAAE,GAAG,CAAC,CAAC;IACX,MAAM,MAAM,GAAqB,EAAE,CAAC;IACpC,KAAK,MAAM,CAAC,KAAK,EAAE,MAAM,CAAC,IAAI,KAAK,CAAC,OAAO,EAAE,EAAE,CAAC;QAC9C,0EAA0E;QAC1E,MAAM,MAAM,GAAG,QAAQ,CAAC,KAAK,CAAC,EAAE,MAAM,IAAI,CAAC,CAAC;QAC5C,IAAI,KAAK,GAAG,CAAC,EAAE,CAAC;YACd,gEAAgE;YAChE,sDAAsD;YACtD,MAAM,CAAC,IAAI,CAAC,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC,SAAS,IAAI,GAAG,CAAC,CAAC,CAAC;QACxD,CAAC;QACD,MAAM,CAAC,IAAI,CACT,GAAG,SAAS,CACV,SAAS,CAAC,KAAK,CAAC,EAAE,EAAE,EAAE,GAAG,MAAM,CAAC;QAChC,qEAAqE;QACrE,mDAAmD;QACnD,MAAM,CAAC,UAAU,CAAC,MAAM,KAAK,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,UAAU,CAAC,CAAC,CAAC,EAAE,EAC5D,MAAM,CAAC,IAAI,EACX,OAAO,CACR,CACF,CAAC;QACF,EAAE,IAAI,MAAM,CAAC;IACf,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC;AAED;;GAEG;AACH,SAAS,OAAO,CACd,MAAiC,EACjC,IAAqD;IAErD,MAAM,SAAS,GAAG,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC;IAC1D,OAAO,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,EAAE,EAAE;QAC9B,qEAAqE;QACrE,MAAM,IAAI,GAAG,SAAS,CAAC,EAAE,CAAC,IAAI,KAAK,CAAC,IAAI,CAAC;QACzC,OAAO,IAAI,KAAK,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,GAAG,KAAK,EAAE,IAAI,EAAE,CAAC;IAC1D,CAAC,CAAC,CAAC;AACL,CAAC;AAED;;;;;GAKG;AACH,SAAS,OAAO,CACd,OAA8B,EAC9B,UAAsB,EACtB,SAAkB;IAElB,OAAO,SAAS;QACd,CAAC,CAAC,OAAO,CACL,OAAO,EACP,aAAa,CACX,OAAO,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,IAAI,CAAC,EACpC,UAAU,CACX,CACF;QACH,CAAC,CAAC,OAAO,CAAC;AACd,CAAC;AAcD;;;;;;GAMG;AACH,SAAS,WAAW,CAClB,IAAwB,EACxB,OAA2C,EAC3C,EAAU;IAEV,OAAO;QACL,KAAK,EAAE;YACL,SAAS,EAAE,qBAAqB,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,SAAS;YACpD,QAAQ,EAAE,OAAO,CAAC,EAAE,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC;SAChD;QACD,UAAU,EAAE,IAAI,CAAC,EAAE,GAAG,CAAC,CAAC,EAAE,KAAK,KAAK,IAAI;KACzC,CAAC;AACJ,CAAC;AAED;;;;;;GAMG;AACH,SAAS,aAAa,CAAC,QAAmC;IACxD,OAAO,QAAQ,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,KAAK,IAAI,EAAE,CAAC;AACtC,CAAC;AAED;;;;;GAKG;AACH,MAAM,QAAQ,GAAG,eAAe,CAAC;AAEjC;;GAEG;AACH,SAAS,QAAQ,CAAC,MAAc,EAAE,KAAa;IAC7C,OAAO,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,IAAI,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;AAC7E,CAAC;AAED;;;;;;GAMG;AACH,SAAS,aAAa,CACpB,OAA0B,EAC1B,IAAyB,EACzB,KAAwB,EACxB,UAA2B;IAE3B,MAAM,MAAM,GAAqB,EAAE,CAAC;IACpC,IAAI,EAAE,GAAG,CAAC,CAAC;IACX,KAAK,MAAM,MAAM,IAAI,KAAK,EAAE,CAAC;QAC3B,IAAI,EAAE,GAAG,CAAC,EAAE,CAAC;YACX,MAAM,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC;QAC/B,CAAC;QACD,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,EAAE,EAAE,EAAE,GAAG,MAAM,CAAC,CAAC;QAC7C,KAAK,MAAM,CAAC,KAAK,EAAE,IAAI,CAAC,IAAI,QAAQ,CAAC,KAAK,EAAE,UAAU,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC;YAClE,MAAM,CAAC,IAAI,CAAC;gBACV,IAAI;gBACJ,QAAQ,EAAE,IAAI,CAAC,EAAE,GAAG,KAAK,CAAC;gBAC1B,UAAU,EAAE,SAAS;gBACrB,MAAM,EAAE,SAAS;aAClB,CAAC,CAAC;QACL,CAAC;QACD,EAAE,IAAI,MAAM,CAAC;IACf,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC;AAED;;;;;;;;GAQG;AACH,SAAS,YAAY,CACnB,IAAyB,EACzB,OAAuB,EACvB,OAAgB;IAEhB,MAAM,OAAO,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CACpC,aAAa,CAAC,QAAQ,EAAE,OAAO,CAAC,QAAQ,CAAC,CAC1C,CAAC;IACF,IAAI,OAAO,CAAC,KAAK,KAAK,QAAQ,EAAE,CAAC;QAC/B,OAAO,IAAI,CACT,OAAO,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,EAAE,EAAE,EAAE,CAAC;YAC5B,GAAG,CAAC,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC;YACtC,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,CAAC,EAAE,CAAC,EAAE,UAAU,EAAE,SAAS,EAAE,MAAM,EAAE,SAAS,EAAE;SACvE,CAAC,EACF,OAAO,CAAC,IAAI,CACb,CAAC;IACJ,CAAC;IACD,MAAM,UAAU,GACd,OAAO,CAAC,QAAQ,KAAK,SAAS,IAAI,OAAO,CAAC,QAAQ,KAAK,aAAa,CAAC;IACvE,MAAM,UAAU,GAAG,UAAU,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,OAAO,CAAC,UAAU,CAAC;IAC7D,uEAAuE;IACvE,2EAA2E;IAC3E,2DAA2D;IAC3D,IAAI,OAAO,CAAC,KAAK,KAAK,SAAS,EAAE,CAAC;QAChC,OAAO,IAAI,CACT,aAAa,CAAC,OAAO,EAAE,IAAI,EAAE,OAAO,CAAC,KAAK,EAAE,UAAU,CAAC,EACvD,OAAO,CAAC,IAAI,CACb,CAAC;IACJ,CAAC;IACD,OAAO,IAAI,CACT,QAAQ,CAAC,OAAO,EAAE,UAAU,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,EAAE,EAAE,CAAC,CAAC;QAC/C,IAAI;QACJ,QAAQ,EAAE,IAAI,CAAC,EAAE,CAAC;QAClB,UAAU,EAAE,SAAS;QACrB,MAAM,EAAE,SAAS;KAClB,CAAC,CAAC,EACH,OAAO,CAAC,IAAI,CACb,CAAC;AACJ,CAAC;AAED;;;;;;;;GAQG;AACH,SAAS,IAAI,CACX,MAAiC,EACjC,MAAc;IAEd,MAAM,KAAK,GAAG,MAAM,CAAC,SAAS,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,QAAQ,KAAK,SAAS,CAAC,CAAC;IACxE,uEAAuE;IACvE,IAAI,KAAK,KAAK,CAAC,CAAC,EAAE,CAAC;QACjB,OAAO,MAAM,CAAC;IAChB,CAAC;IACD,OAAO,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,EAAE,EAAE,CAC9B,EAAE,KAAK,KAAK,CAAC,CAAC,CAAC,EAAE,GAAG,KAAK,EAAE,MAAM,EAAE,CAAC,CAAC,CAAC,KAAK,CAC5C,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,SAAS,OAAO,CAAC,KAAc;IAC7B,OAAO,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC;AAC1B,CAAC;AAED;;;;;;;;GAQG;AACH,SAAS,YAAY,CACnB,IAAY,EACZ,QAAmC,EACnC,OAAmB,EACnB,OAAgB,EAChB,OAAuD;IAEvD,IAAI,QAAQ,CAAC,KAAK,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAC,OAAO,KAAK,SAAS,CAAC,EAAE,CAAC;QAC/D,OAAO,CAAC,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC;IAC7B,CAAC;IAED,MAAM,MAAM,GAAqB,EAAE,CAAC;IACpC,IAAI,MAAM,GAAG,OAAO,CAAC,OAAO,CAAC,UAAU,CAAC,CAAC;IAEzC,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;QAC/B,IAAI,QAAQ,CAAC,MAAM,EAAE,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC;YACnC,MAAM,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC;QAC/B,CAAC;QACD,MAAM,CAAC,IAAI,CACT,GAAG,CAAC,OAAO,CAAC,OAAO,KAAK,SAAS;YAC/B,CAAC,CAAC,CAAC,WAAW,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC;YAC7B,CAAC,CAAC,YAAY,CACV,WAAW,CAAC,OAAO,EAAE,OAAO,CAAC,KAAK,CAAC,QAAQ,EAAE,OAAO,CAAC,MAAM,CAAC,EAC5D,OAAO,EACP,OAAO,CACR,CAAC,CACP,CAAC;QACF,wEAAwE;QACxE,+DAA+D;QAC/D,MAAM,GAAG,MAAM,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,IAAI,IAAI,MAAM,CAAC;IACzC,CAAC;IACD,IAAI,QAAQ,CAAC,MAAM,EAAE,OAAO,CAAC,OAAO,CAAC,KAAK,CAAC,SAAS,KAAK,EAAE,CAAC,CAAC,EAAE,CAAC;QAC9D,MAAM,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC;IAC/B,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC;AAED;;GAEG;AACH,SAAS,WAAW,CAClB,MAAc,EACd,UAAsB,EACtB,IAAY,EACZ,OAAuB;IAEvB,MAAM,EACJ,MAAM,GAAG,OAAO,EAChB,QAAQ,GAAG,OAAO,EAClB,UAAU,GAAG,QAAQ,EACrB,QAAQ,GAAG,MAAM,EACjB,WAAW,GAAG,OAAO,EACrB,QAAQ,GAAG,IAAI,EACf,OAAO,GAAG,MAAM,EAChB,MAAM,EACN,QAAQ,GACT,GAAG,OAAO,CAAC;IACZ,MAAM,OAAO,GAAY,EAAE,QAAQ,EAAE,UAAU,EAAE,QAAQ,EAAE,CAAC;IAC5D,MAAM,SAAS,GAAqB,EAAE,CAAC;IAEvC,MAAM,IAAI,GAAG,CAAC,GAAG,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC;IAClC,0EAA0E;IAC1E,6DAA6D;IAC7D,MAAM,KAAK,GAAG,QAAQ,KAAK,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,YAAY,CAAC,QAAQ,CAAC,CAAC;IAC1E,6EAA6E;IAC7E,4DAA4D;IAC5D,MAAM,MAAM,GAAa,EAAE,CAAC;IAC5B,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;QACvB,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;QACnB,KAAK,IAAI,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,MAAM,CAAC;IACzC,CAAC;IACD,uEAAuE;IACvE,0EAA0E;IAC1E,mEAAmE;IACnE,MAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,EAAE,EAAE,CAChC,GAAG,CAAC,KAAK,IAAI,OAAO,KAAK,MAAM;QAC7B,CAAC,CAAC,EAAE;QACJ,CAAC,CAAC,aAAa,CAAC,GAAG,CAAC,IAAI,EAAE,qBAAqB,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,CAC7D,CAAC;IACF,0EAA0E;IAC1E,4EAA4E;IAC5E,+DAA+D;IAC/D,MAAM,OAAO,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,EAAE,EAAE;QACnC,IAAI,CAAC,GAAG,CAAC,KAAK,EAAE,CAAC;YACf,OAAO,EAAE,CAAC;QACZ,CAAC;QACD,MAAM,MAAM,GACV,KAAK,KAAK,SAAS;YACjB,CAAC,CAAC,SAAS;YACX,CAAC,CAAC,WAAW,CAAC,KAAK,EAAE,MAAM,CAAC,EAAE,CAAC,IAAI,CAAC,EAAE,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,MAAM,CAAC,CAAC;QACzE,MAAM,IAAI,GAAG,aAAa,CAAC,IAAI,CAAC,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;QAC/C,OAAO,OAAO,CACZ,MAAM,CAAC,UAAU,EAAE,GAAG,CAAC,IAAI,EAAE,IAAI,EAAE,MAAM,CAAC,EAC1C,UAAU,EACV,QAAQ,CACT,CAAC;IACJ,CAAC,CAAC,CAAC;IAEH,KAAK,MAAM,CAAC,EAAE,EAAE,GAAG,CAAC,IAAI,IAAI,CAAC,OAAO,EAAE,EAAE,CAAC;QACvC,MAAM,KAAK,GAAG,OAAO,CAAC,EAAE,CAAC,IAAI,EAAE,CAAC;QAChC,IAAI,GAAG,CAAC,KAAK,EAAE,CAAC;YACd,SAAS,CAAC,IAAI,CAAC,GAAG,QAAQ,CAAC,UAAU,EAAE,KAAK,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,CAAC,CAAC,CAAC;YACxE,SAAS;QACX,CAAC;QACD,SAAS,CAAC,IAAI,CACZ,GAAG,YAAY,CACb,GAAG,CAAC,IAAI,EACR,IAAI,CAAC,EAAE,CAAC,IAAI,EAAE,EACd,WAAW,CAAC,IAAI,EAAE,OAAO,EAAE,EAAE,CAAC,EAC9B,OAAO,EACP,EAAE,MAAM,EAAE,CACX,CACF,CAAC;IACJ,CAAC;IACD,IAAI,MAAM,GAA8B,SAAS,CAAC;IAElD,0EAA0E;IAC1E,6EAA6E;IAC7E,qEAAqE;IACrE,IAAI,QAAQ,KAAK,MAAM,IAAI,UAAU,CAAC,IAAI,CAAC,EAAE,CAAC;QAC5C,MAAM,GAAG,OAAO,CAAC,MAAM,EAAE,uBAAuB,CAAC,CAAC;IACpD,CAAC;IACD,OAAO,WAAW,KAAK,OAAO;QAC5B,CAAC,CAAC,OAAO,CAAC,MAAM,EAAE,uBAAuB,CAAC;QAC1C,CAAC,CAAC,MAAM,CAAC;AACb,CAAC;AAED;;;;;;GAMG;AACH,SAAS,OAAO,CACd,OAA8B,EAC9B,OAA+B;IAE/B,MAAM,UAAU,GAAG,OAAO,CAAC,OAAO,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC,GAAG,MAAM,CAAC,UAAU,CAAC,CAAC,CAAC;IACvE,IAAI,EAAE,GAAG,CAAC,CAAC;IACX,OAAO,OAAO,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE;QAC1B,MAAM,IAAI,GAAG,UAAU,CAAC,KAAK,CAAC,EAAE,EAAE,EAAE,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;QAC5D,EAAE,IAAI,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC;QAC1B,OAAO,EAAE,IAAI,EAAE,UAAU,EAAE,IAAI,EAAE,CAAC;IACpC,CAAC,CAAC,CAAC;AACL,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,OAAO,CACrB,UAAsB,EACtB,IAAY,EACZ,OAAO,GAAmB,EAAE;IAE5B,OAAO,UAAU,CAAC,WAAW,CAAC,OAAO,EAAE,UAAU,EAAE,IAAI,EAAE,OAAO,CAAC,CAAC,CAAC;AACrE,CAAC;AAED;;;;;;;;;;;GAWG;AACH,MAAM,UAAU,aAAa,CAC3B,UAAsB,EACtB,IAAY,EACZ,OAAO,GAAmB,EAAE;IAE5B,OAAO,WAAW,CAAC,MAAM,EAAE,UAAU,EAAE,IAAI,EAAE,OAAO,CAAC,CAAC;AACxD,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,qBAAqB,CACnC,UAAsB,EACtB,IAAY,EACZ,OAAO,GAAmB,EAAE;IAE5B,OAAO,WAAW,CAAC,OAAO,EAAE,UAAU,EAAE,IAAI,EAAE,OAAO,CAAC,CAAC;AACzD,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,UAAU,CAAC,MAAiC;IAC1D,OAAO,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;AACpD,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,eAAe,CAC7B,UAAsB,EACtB,IAAY,EACZ,OAAO,GAAmB,EAAE;IAE5B,OAAO,UAAU,CAAC,WAAW,CAAC,MAAM,EAAE,UAAU,EAAE,IAAI,EAAE,OAAO,CAAC,CAAC,CAAC;AACpE,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,qBAAqB,CACnC,UAAsB,EACtB,IAAY,EACZ,OAAO,GAAmB,EAAE;IAE5B,OAAO,WAAW,CAAC,MAAM,EAAE,UAAU,EAAE,IAAI,EAAE,OAAO,CAAC,CAAC;AACxD,CAAC"}
+{"version":3,"file":"convert.js","sourceRoot":"","sources":["../../src/decode/convert.ts"],"names":[],"mappings":"AAQA,OAAO,EAAE,WAAW,EAAE,MAAM,uBAAuB,CAAC;AAEpD,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,EAAE,MAAM,eAAe,CAAC;AAExD,OAAO,EAIL,SAAS,GACV,MAAM,aAAa,CAAC;AAErB;;;;;;;;GAQG;AACH,MAAM,UAAU,OAAO,CACrB,UAAsB,EACtB,IAAY,EACZ,OAAO,GAAmB,EAAE;IAE5B,OAAO,UAAU,CAAC,WAAW,CAAC,OAAO,EAAE,UAAU,EAAE,IAAI,EAAE,OAAO,CAAC,CAAC,CAAC;AACrE,CAAC;AAED;;;;;;;;;;;GAWG;AACH,MAAM,UAAU,aAAa,CAC3B,UAAsB,EACtB,IAAY,EACZ,OAAO,GAAmB,EAAE;IAE5B,OAAO,WAAW,CAAC,MAAM,EAAE,UAAU,EAAE,IAAI,EAAE,OAAO,CAAC,CAAC;AACxD,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,qBAAqB,CACnC,UAAsB,EACtB,IAAY,EACZ,OAAO,GAAmB,EAAE;IAE5B,OAAO,WAAW,CAAC,OAAO,EAAE,UAAU,EAAE,IAAI,EAAE,OAAO,CAAC,CAAC;AACzD,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,UAAU,CAAC,MAAiC;IAC1D,OAAO,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;AACpD,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,eAAe,CAC7B,UAAsB,EACtB,IAAY,EACZ,OAAO,GAAmB,EAAE;IAE5B,OAAO,UAAU,CAAC,WAAW,CAAC,MAAM,EAAE,UAAU,EAAE,IAAI,EAAE,OAAO,CAAC,CAAC,CAAC;AACpE,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,qBAAqB,CACnC,UAAsB,EACtB,IAAY,EACZ,OAAO,GAAmB,EAAE;IAE5B,OAAO,WAAW,CAAC,MAAM,EAAE,UAAU,EAAE,IAAI,EAAE,OAAO,CAAC,CAAC;AACxD,CAAC"}