@kensio/pinyinjs 1.7.8 → 1.7.9

package/dist/dictionary/traditional-carry.js

@@ -0,0 +1,201 @@
+/**
+ * What a 繁體 spelling takes from the 简体 word it pairs with.
+ *
+ * jieba's dictionary was counted over a 简体 corpus, and all three of the fields
+ * it settles arrive on one script and not the other. 听 is tagged `v` and
+ * counted 20,435 times. 聽 is tagged nothing, counted nowhere, and 麥 is not a
+ * name where 麦 is one. Every rule that asks what the word beside it is decides
+ * on the tag, every path the decoder weighs is priced on the count, and the
+ * capital comes straight off the proper-noun bit, so 繁體 text was read by
+ * rules that could not see it, priced by a model that had never met it and
+ * capitalised by a different answer from the one 简体 got.
+ *
+ * All three travel over the pairing the entries already hold. An entry names
+ * the 繁體 spelling of its own word. The entry keyed on that spelling is
+ * therefore the same word written the other way, and how a word is classed, how
+ * often it is met and whether it is a name are facts about the word rather than
+ * about the spelling. `corpus-mass.ts` makes the same move for the polyphone
+ * priors, where a vote cast for a 简体 word is cast again for whichever 繁體
+ * form the reading picks.
+ *
+ * A 繁體 spelling that is not a headword of its own needs none of this, because
+ * `artifact-claims.ts` lets its 简体 entry claim the key outright and the key
+ * carries that entry's fields. 銀行 is claimed by 银行 and always priced as it.
+ * What is left is the characters Unihan gives an entry to, which is why only
+ * single characters are paired here.
+ */
+import { convertCharacter } from "../script/conversion.js";
+import { isSingleCharacter } from "../script/characters.js";
+import { traditionalCharacterTable } from "./script-pairings.js";
+/**
+ * jieba's tag for a character it has counted and not classified.
+ *
+ * Written on 5,666 entries and on nothing longer than one character, 简体 and
+ * 繁體 alike: 且, 丢, 乎, 井, 今 and 仍 carry it next to 來, 個 and 這. It says
+ * that the character was seen rather than what it is. Read here as no tag at
+ * all, and left in the artifact, because whether jieba wrote `zg` or wrote
+ * nothing is a fact about the source and only this decides what to make of it.
+ *
+ * Nothing longer than one character carries it. That leaves the rules forbidding
+ * an **untagged key** alone, since each of those asks about a word of two
+ * characters or more and no such key is tagged `zg`.
+ */
+const UNCATEGORISED = "zg";
+/**
+ * Whether a tag says anything about the word.
+ */
+export function isTagged(partOfSpeech) {
+    return partOfSpeech !== "" && partOfSpeech !== UNCATEGORISED;
+}
+/**
+ * The 繁體 character a 简体 one is written as, or undefined where it is neither.
+ *
+ * Read off the aggregate table rather than off {@link DictionaryEntry.hant},
+ * and that is the whole of what keeps this honest. An entry's own 繁體 form is
+ * whichever CC-CEDICT sense matched its reading, and for a single character
+ * that can be an oddity: CC-CEDICT holds `旹 时` as an old variant of 時, so
+ * 时's entry can name 旹 while every word 时 appears in says 時. Lending 时's
+ * 103,735 occurrences to 旹 made a character nobody writes the third commonest
+ * way to write `shí`. The table counts words, so 時 wins it thousands to one.
+ */
+function traditionalOf(table, entry) {
+    if (!isSingleCharacter(entry.hans)) {
+        return undefined;
+    }
+    const written = convertCharacter(table, entry.hans, entry.readings.cn[0]);
+    return written === entry.hans ? undefined : written;
+}
+/**
+ * The 简体 entry each 繁體 character should take a field from.
+ *
+ * Where two 简体 characters are written the same way in 繁體 the commoner of
+ * them lends. Both fields are being used to guess at what a reader meant, and
+ * the commoner character is the likelier thing to have been meant. `hasField`
+ * is asked first, so that a commoner character with nothing to lend does not
+ * shut out a rarer one that has something.
+ */
+function lendersByTraditional(entries, table, hasField) {
+    const lenders = new Map();
+    for (const entry of entries) {
+        const written = hasField(entry) ? traditionalOf(table, entry) : undefined;
+        if (written === undefined) {
+            continue;
+        }
+        const held = lenders.get(written);
+        if (held === undefined || held.frequency < entry.frequency) {
+            lenders.set(written, entry);
+        }
+    }
+    return lenders;
+}
+/**
+ * Whether an entry has a tag worth lending.
+ */
+function canLendTag(entry) {
+    return isTagged(entry.partOfSpeech);
+}
+/**
+ * The tag a spelling ends up with, and whether it took one.
+ *
+ * `nr`, `ns`, `nt` and `nz` are lent along with the rest now that
+ * {@link properFor} carries the bit they go with. Lending the tag without it
+ * would have left a spelling claiming a place name the same entry denies is a
+ * proper noun, which is why they were held back until the bit travelled.
+ */
+function tagFor(entry, lender) {
+    if (isTagged(entry.partOfSpeech) || lender === undefined) {
+        return undefined;
+    }
+    return lender.partOfSpeech;
+}
+/**
+ * The count a spelling ends up with, and whether it took one.
+ *
+ * The larger of the two, rather than the lender's outright. A 繁體 spelling
+ * that jieba counted at all was counted incidentally in a corpus written the
+ * other way, and the word occurs at least as often as the commoner spelling of
+ * it was seen. Taking the smaller would price 過 by the handful of 繁體 lines
+ * jieba's corpus happens to contain.
+ *
+ * Only the form the reading picks takes it. 发 read `fā` is written 發, so 發
+ * takes 发's count and 髮 does not, which is the same line the conversion draws
+ * when it writes 出发 as 出發 and 头发 as 頭髮.
+ *
+ * **The count is carried whole and not shared out.** A share would say that
+ * 繁體 text is some fraction as common as 简体 text, and nothing measured here
+ * says what that fraction is. It would also be invisible: the buckets the
+ * decoder compares are log-spaced, one of them is a factor of 2.49, so halving
+ * a count moves it 0.76 of a bucket and a tenth of it moves it 2.5. Only a
+ * share of an order of magnitude would change a decode, and inventing one that
+ * large to fix a decode that is currently wrong would be trading one guess for
+ * another.
+ */
+function countFor(entry, lender) {
+    if (lender === undefined || lender.frequency <= entry.frequency) {
+        return undefined;
+    }
+    return lender.frequency;
+}
+/**
+ * Whether a spelling is a name, and whether that is a change.
+ *
+ * The bit is what survived `properNounOf`, so it is the answer both sources
+ * reached together rather than jieba's proposal, and it is taken whichever way
+ * it points. Demoting matters as much as promoting. 后 is a locative and 後
+ * arrived tagged `nr` with nothing under its own spelling to challenge it, so
+ * 退休後 came out `tuìxiū Hòu` where 退休后 came out `tuìxiū hòu`.
+ *
+ * This carries the 简体 answer whether or not that answer is right. jieba calls
+ * 连 a surname and 連 now agrees, where before the two differed and one of them
+ * happened to be correct. The two scripts saying one thing is what this is for.
+ * How good that one thing is for a bare character is #159.
+ */
+function properFor(entry, lender) {
+    if (lender === undefined || lender.isProperNoun === entry.isProperNoun) {
+        return undefined;
+    }
+    return lender.isProperNoun;
+}
+/**
+ * Give every 繁體 headword the tag and the count its 简体 word carries.
+ *
+ * Neither overwrites something the source stated. A spelling jieba classified
+ * keeps its own tag, and a spelling jieba counted more often than its 简体 word
+ * keeps its own count.
+ */
+export function carryToTraditional(entries) {
+    const table = traditionalCharacterTable(entries);
+    const tagLenders = lendersByTraditional(entries, table, canLendTag);
+    const countLenders = lendersByTraditional(entries, table, (entry) => entry.frequency > 0);
+    let carriedTags = 0;
+    let carriedCounts = 0;
+    let carriedCapitals = 0;
+    const carried = entries.map((entry) => {
+        const tagLender = tagLenders.get(entry.hans);
+        const partOfSpeech = tagFor(entry, tagLender);
+        const frequency = countFor(entry, countLenders.get(entry.hans));
+        const isProperNoun = properFor(entry, tagLender);
+        if (partOfSpeech === undefined &&
+            frequency === undefined &&
+            isProperNoun === undefined) {
+            return entry;
+        }
+        if (partOfSpeech !== undefined) {
+            carriedTags += 1;
+        }
+        if (frequency !== undefined) {
+            carriedCounts += 1;
+        }
+        if (isProperNoun !== undefined) {
+            carriedCapitals += 1;
+        }
+        return {
+            ...entry,
+            ...(partOfSpeech !== undefined && { partOfSpeech }),
+            ...(frequency !== undefined && { frequency }),
+            ...(isProperNoun !== undefined && { isProperNoun }),
+        };
+    });
+    return { entries: carried, carriedTags, carriedCounts, carriedCapitals };
+}
+//# sourceMappingURL=traditional-carry.js.map

package/dist/dictionary/traditional-carry.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"traditional-carry.js","sourceRoot":"","sources":["../../src/dictionary/traditional-carry.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;GAyBG;AACH,OAAO,EAAE,gBAAgB,EAAE,MAAM,yBAAyB,CAAC;AAE3D,OAAO,EAAE,iBAAiB,EAAE,MAAM,yBAAyB,CAAC;AAE5D,OAAO,EAAE,yBAAyB,EAAE,MAAM,sBAAsB,CAAC;AAEjE;;;;;;;;;;;;GAYG;AACH,MAAM,aAAa,GAAG,IAAI,CAAC;AAE3B;;GAEG;AACH,MAAM,UAAU,QAAQ,CAAC,YAAoB;IAC3C,OAAO,YAAY,KAAK,EAAE,IAAI,YAAY,KAAK,aAAa,CAAC;AAC/D,CAAC;AAYD;;;;;;;;;;GAUG;AACH,SAAS,aAAa,CACpB,KAA+C,EAC/C,KAAsB;IAEtB,IAAI,CAAC,iBAAiB,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC;QACnC,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,MAAM,OAAO,GAAG,gBAAgB,CAAC,KAAK,EAAE,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;IAC1E,OAAO,OAAO,KAAK,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,OAAO,CAAC;AACtD,CAAC;AAED;;;;;;;;GAQG;AACH,SAAS,oBAAoB,CAC3B,OAAmC,EACnC,KAA+C,EAC/C,QAA6C;IAE7C,MAAM,OAAO,GAAG,IAAI,GAAG,EAA2B,CAAC;IACnD,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,MAAM,OAAO,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,aAAa,CAAC,KAAK,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;QAC1E,IAAI,OAAO,KAAK,SAAS,EAAE,CAAC;YAC1B,SAAS;QACX,CAAC;QACD,MAAM,IAAI,GAAG,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;QAClC,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,CAAC,SAAS,GAAG,KAAK,CAAC,SAAS,EAAE,CAAC;YAC3D,OAAO,CAAC,GAAG,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC;QAC9B,CAAC;IACH,CAAC;IACD,OAAO,OAAO,CAAC;AACjB,CAAC;AAED;;GAEG;AACH,SAAS,UAAU,CAAC,KAAsB;IACxC,OAAO,QAAQ,CAAC,KAAK,CAAC,YAAY,CAAC,CAAC;AACtC,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,MAAM,CACb,KAAsB,EACtB,MAAmC;IAEnC,IAAI,QAAQ,CAAC,KAAK,CAAC,YAAY,CAAC,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;QACzD,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,OAAO,MAAM,CAAC,YAAY,CAAC;AAC7B,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;GAqBG;AACH,SAAS,QAAQ,CACf,KAAsB,EACtB,MAAmC;IAEnC,IAAI,MAAM,KAAK,SAAS,IAAI,MAAM,CAAC,SAAS,IAAI,KAAK,CAAC,SAAS,EAAE,CAAC;QAChE,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,OAAO,MAAM,CAAC,SAAS,CAAC;AAC1B,CAAC;AAED;;;;;;;;;;;;;GAaG;AACH,SAAS,SAAS,CAChB,KAAsB,EACtB,MAAmC;IAEnC,IAAI,MAAM,KAAK,SAAS,IAAI,MAAM,CAAC,YAAY,KAAK,KAAK,CAAC,YAAY,EAAE,CAAC;QACvE,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,OAAO,MAAM,CAAC,YAAY,CAAC;AAC7B,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,kBAAkB,CAChC,OAAmC;IAEnC,MAAM,KAAK,GAAG,yBAAyB,CAAC,OAAO,CAAC,CAAC;IACjD,MAAM,UAAU,GAAG,oBAAoB,CAAC,OAAO,EAAE,KAAK,EAAE,UAAU,CAAC,CAAC;IACpE,MAAM,YAAY,GAAG,oBAAoB,CACvC,OAAO,EACP,KAAK,EACL,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,SAAS,GAAG,CAAC,CAC/B,CAAC;IACF,IAAI,WAAW,GAAG,CAAC,CAAC;IACpB,IAAI,aAAa,GAAG,CAAC,CAAC;IACtB,IAAI,eAAe,GAAG,CAAC,CAAC;IAExB,MAAM,OAAO,GAAG,OAAO,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE;QACpC,MAAM,SAAS,GAAG,UAAU,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QAC7C,MAAM,YAAY,GAAG,MAAM,CAAC,KAAK,EAAE,SAAS,CAAC,CAAC;QAC9C,MAAM,SAAS,GAAG,QAAQ,CAAC,KAAK,EAAE,YAAY,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC;QAChE,MAAM,YAAY,GAAG,SAAS,CAAC,KAAK,EAAE,SAAS,CAAC,CAAC;QACjD,IACE,YAAY,KAAK,SAAS;YAC1B,SAAS,KAAK,SAAS;YACvB,YAAY,KAAK,SAAS,EAC1B,CAAC;YACD,OAAO,KAAK,CAAC;QACf,CAAC;QACD,IAAI,YAAY,KAAK,SAAS,EAAE,CAAC;YAC/B,WAAW,IAAI,CAAC,CAAC;QACnB,CAAC;QACD,IAAI,SAAS,KAAK,SAAS,EAAE,CAAC;YAC5B,aAAa,IAAI,CAAC,CAAC;QACrB,CAAC;QACD,IAAI,YAAY,KAAK,SAAS,EAAE,CAAC;YAC/B,eAAe,IAAI,CAAC,CAAC;QACvB,CAAC;QACD,OAAO;YACL,GAAG,KAAK;YACR,GAAG,CAAC,YAAY,KAAK,SAAS,IAAI,EAAE,YAAY,EAAE,CAAC;YACnD,GAAG,CAAC,SAAS,KAAK,SAAS,IAAI,EAAE,SAAS,EAAE,CAAC;YAC7C,GAAG,CAAC,YAAY,KAAK,SAAS,IAAI,EAAE,YAAY,EAAE,CAAC;SACpD,CAAC;IACJ,CAAC,CAAC,CAAC;IAEH,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,WAAW,EAAE,aAAa,EAAE,eAAe,EAAE,CAAC;AAC3E,CAAC"}

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

@@ -1 +1 @@
-{"version":3,"file":"annotate.d.ts","sourceRoot":"","sources":["../../src/format/annotate.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,EAAU,KAAK,WAAW,EAAsB,MAAM,WAAW,CAAC;AACzE,OAAO,EAAE,KAAK,cAAc,EAAiB,MAAM,sBAAsB,CAAC;AAC1E,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,6BAA6B,CAAC;AAE9D;;;;;;;;;;;;;;;GAeG;AACH,eAAO,MAAM,QAAQ,OAAO,CAAC;AAE7B;;GAEG;AACH,MAAM,WAAW,UAAU;IACzB,mEAAmE;IACnE,QAAQ,CAAC,MAAM,EAAE,MAAM,GAAG,SAAS,CAAC;IACpC,QAAQ,CAAC,MAAM,EAAE,SAAS,cAAc,EAAE,CAAC;CAC5C;AAED;;;;;;;;GAQG;AACH,wBAAgB,aAAa,CAAC,MAAM,EAAE,SAAS,cAAc,EAAE,GAAG,UAAU,EAAE,CAc7E;AA2ED;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AACH,wBAAgB,eAAe,CAC7B,MAAM,EAAE,SAAS,cAAc,EAAE,EACjC,OAAO,GAAE,WAAgB,GACxB,MAAM,CAIR;AAED;;;;;GAKG;AACH,wBAAgB,sBAAsB,CACpC,UAAU,EAAE,UAAU,EACtB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,WAAgB,GACxB,MAAM,CAER;AAED;;;;;;;;;;;GAWG;AACH,wBAAgB,aAAa,CAC3B,UAAU,EAAE,UAAU,EACtB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,WAAgB,GACxB,MAAM,CAER"}
+{"version":3,"file":"annotate.d.ts","sourceRoot":"","sources":["../../src/format/annotate.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,EAAU,KAAK,WAAW,EAAuB,MAAM,WAAW,CAAC;AAC1E,OAAO,EAAE,KAAK,cAAc,EAAiB,MAAM,sBAAsB,CAAC;AAC1E,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,6BAA6B,CAAC;AAE9D;;;;;;;;;;;;;;;GAeG;AACH,eAAO,MAAM,QAAQ,OAAO,CAAC;AAE7B;;GAEG;AACH,MAAM,WAAW,UAAU;IACzB,mEAAmE;IACnE,QAAQ,CAAC,MAAM,EAAE,MAAM,GAAG,SAAS,CAAC;IACpC,QAAQ,CAAC,MAAM,EAAE,SAAS,cAAc,EAAE,CAAC;CAC5C;AAED;;;;;;;;GAQG;AACH,wBAAgB,aAAa,CAAC,MAAM,EAAE,SAAS,cAAc,EAAE,GAAG,UAAU,EAAE,CAc7E;AA2ED;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AACH,wBAAgB,eAAe,CAC7B,MAAM,EAAE,SAAS,cAAc,EAAE,EACjC,OAAO,GAAE,WAAgB,GACxB,MAAM,CAIR;AAED;;;;;GAKG;AACH,wBAAgB,sBAAsB,CACpC,UAAU,EAAE,UAAU,EACtB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,WAAgB,GACxB,MAAM,CAER;AAED;;;;;;;;;;;GAWG;AACH,wBAAgB,aAAa,CAC3B,UAAU,EAAE,UAAU,EACtB,IAAI,EAAE,MAAM,EACZ,OAAO,GAAE,WAAgB,GACxB,MAAM,CAER"}

package/dist/format/annotate.js

@@ -4,7 +4,7 @@
  * The pinyin sits above the characters rather than beside them, so the pieces
  * are regrouped against the text they annotate.
  */
-import { escape, toHtml, writePiece } from "./html.js";
+import { escape, toHtml, writePieces } from "./html.js";
 import { convertPieces } from "../decode/convert.js";
 /**
  * What the characters being annotated declare themselves to be.
@@ -101,9 +101,9 @@ function writeAnnotation(annotation, options) {
         // 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("");
+        return writePieces(pieces, options);
     }
-    const reading = pieces.map((piece) => writePiece(piece, options)).join("");
+    const reading = writePieces(pieces, options);
     // 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,

package/dist/format/annotate.js.map

@@ -1 +1 @@
-{"version":3,"file":"annotate.js","sourceRoot":"","sources":["../../src/format/annotate.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,EAAE,MAAM,EAAoB,MAAM,EAAE,UAAU,EAAE,MAAM,WAAW,CAAC;AACzE,OAAO,EAAuB,aAAa,EAAE,MAAM,sBAAsB,CAAC;AAG1E;;;;;;;;;;;;;;;GAeG;AACH,MAAM,CAAC,MAAM,QAAQ,GAAG,IAAI,CAAC;AAW7B;;;;;;;;GAQG;AACH,MAAM,UAAU,aAAa,CAAC,MAAiC;IAC7D,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":"annotate.js","sourceRoot":"","sources":["../../src/format/annotate.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,EAAE,MAAM,EAAoB,MAAM,EAAE,WAAW,EAAE,MAAM,WAAW,CAAC;AAC1E,OAAO,EAAuB,aAAa,EAAE,MAAM,sBAAsB,CAAC;AAG1E;;;;;;;;;;;;;;;GAeG;AACH,MAAM,CAAC,MAAM,QAAQ,GAAG,IAAI,CAAC;AAW7B;;;;;;;;GAQG;AACH,MAAM,UAAU,aAAa,CAAC,MAAiC;IAC7D,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,WAAW,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;IACtC,CAAC;IAED,MAAM,OAAO,GAAG,WAAW,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;IAC7C,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"}

package/dist/format/html.d.ts

@@ -1,5 +1,14 @@
+/**
+ * A conversion as markup, and the options that decide what the markup carries.
+ *
+ * One element per syllable, which is what lets a page colour a tone or mark a
+ * guess. What goes inside one of those elements is `syllable-span.ts` and the
+ * ruby that puts the reading over the hanzi is `annotate.ts`.
+ */
 export { convertToAnnotatedHtml, convertToHtml, toAnnotatedHtml, } from "./annotate.js";
 import type { ConvertedPiece, ConvertOptions } from "../decode/convert.js";
+import type { TranscriptionSystem } from "../transcription/systems.js";
+export { writePiece } from "./syllable-span.js";
 /**
  * How a conversion is marked up.
  */
@@ -15,18 +24,32 @@ export interface HtmlOptions extends ConvertOptions {
      * Whether readings the decode was guessing at are marked. Defaults to true.
      *
      * `py-uncertain`, plus a `data-alternatives` attribute listing what was
-     * rejected. See {@link isUncertain} for what qualifies and how often it is
-     * wrong.
+     * rejected. See `isUncertain` for what qualifies and how often it is wrong.
      */
     readonly markUncertain?: boolean;
     /**
      * Whether each syllable declares what language it is in. Defaults to true.
      *
      * `lang="zh-Latn-CN-pinyin"`, or `zh-Latn-TW-pinyin` where the conversion
-     * reads `zh-TW`. See {@link languageTag} for why that matters, and what to
-     * do instead of turning it off.
+     * reads `zh-TW`. See `languageTag` for why that matters, and what to do
+     * instead of turning it off.
      */
     readonly lang?: boolean;
+    /**
+     * A system to write the reading in instead of pinyin.
+     *
+     * `BOPOMOFO`, `WADE_GILES`, `YALE`, `GWOYEU` or `IPA`, or a
+     * {@link TranscriptionSystem} of the caller's own. Left out, the reading is
+     * the pinyin the conversion wrote.
+     *
+     * The word segmentation is the conversion's either way and only the join
+     * changes, which is the division `toTranscription` makes and for the same
+     * reason. What a word is belongs to the language, and how its syllables are
+     * run together belongs to the system. Everything else a syllable carries is
+     * untouched, so tone classes, `py-uncertain` and `data-alternatives` mean
+     * what they always did.
+     */
+    readonly transcription?: TranscriptionSystem;
 }
 /**
  * Escape text for markup.
@@ -39,9 +62,21 @@ export interface HtmlOptions extends ConvertOptions {
  */
 export declare function escape(text: string): string;
 /**
- * Mark up one piece of a conversion.
+ * Mark up a run of pieces, with the system's join between two syllables.
+ *
+ * Pinyin needs none. Its syllables are run together and the marks that divide
+ * them are pieces of their own. Every other system decides the join for itself,
+ * and the join is what separates the five at this level, so it is written
+ * between the spans rather than inside one. `<span>Pei³</span>-<span>ching¹
+ * </span>` is one word in Wade-Giles and leaves each syllable its own element
+ * to carry a tone class on.
+ *
+ * A piece that wrote nothing does not break the run, which is what carries the
+ * join over pinyin's own hyphen: 干干净净 is four syllables of one word, and
+ * the mark between the second and the third is dropped rather than counted as a
+ * gap.
  */
-export declare function writePiece(piece: ConvertedPiece, options: HtmlOptions): string;
+export declare function writePieces(pieces: readonly ConvertedPiece[], options: HtmlOptions): string;
 /**
  * Mark up a converted text, one element per syllable.
  *

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

@@ -1 +1 @@
-{"version":3,"file":"html.d.ts","sourceRoot":"","sources":["../../src/format/html.ts"],"names":[],"mappings":"AACA,OAAO,EACL,sBAAsB,EACtB,aAAa,EACb,eAAe,GAChB,MAAM,eAAe,CAAC;AAEvB,OAAO,KAAK,EAAE,cAAc,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAI3E;;GAEG;AACH,MAAM,WAAW,WAAY,SAAQ,cAAc;IACjD;;;;;OAKG;IACH,QAAQ,CAAC,WAAW,CAAC,EAAE,OAAO,CAAC;IAC/B;;;;;;OAMG;IACH,QAAQ,CAAC,aAAa,CAAC,EAAE,OAAO,CAAC;IACjC;;;;;;OAMG;IACH,QAAQ,CAAC,IAAI,CAAC,EAAE,OAAO,CAAC;CACzB;AAwDD;;;;;;;;GAQG;AACH,wBAAgB,MAAM,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAK3C;AAwBD;;GAEG;AACH,wBAAgB,UAAU,CACxB,KAAK,EAAE,cAAc,EACrB,OAAO,EAAE,WAAW,GACnB,MAAM,CA4BR;AAED;;;;;GAKG;AACH,wBAAgB,MAAM,CACpB,MAAM,EAAE,SAAS,cAAc,EAAE,EACjC,OAAO,GAAE,WAAgB,GACxB,MAAM,CAER"}
+{"version":3,"file":"html.d.ts","sourceRoot":"","sources":["../../src/format/html.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,EACL,sBAAsB,EACtB,aAAa,EACb,eAAe,GAChB,MAAM,eAAe,CAAC;AACvB,OAAO,KAAK,EAAE,cAAc,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAC3E,OAAO,KAAK,EAAE,mBAAmB,EAAE,MAAM,6BAA6B,CAAC;AAGvE,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AAEhD;;GAEG;AACH,MAAM,WAAW,WAAY,SAAQ,cAAc;IACjD;;;;;OAKG;IACH,QAAQ,CAAC,WAAW,CAAC,EAAE,OAAO,CAAC;IAC/B;;;;;OAKG;IACH,QAAQ,CAAC,aAAa,CAAC,EAAE,OAAO,CAAC;IACjC;;;;;;OAMG;IACH,QAAQ,CAAC,IAAI,CAAC,EAAE,OAAO,CAAC;IACxB;;;;;;;;;;;;;OAaG;IACH,QAAQ,CAAC,aAAa,CAAC,EAAE,mBAAmB,CAAC;CAC9C;AAYD;;;;;;;;GAQG;AACH,wBAAgB,MAAM,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAK3C;AAED;;;;;;;;;;;;;;GAcG;AACH,wBAAgB,WAAW,CACzB,MAAM,EAAE,SAAS,cAAc,EAAE,EACjC,OAAO,EAAE,WAAW,GACnB,MAAM,CAgBR;AAED;;;;;GAKG;AACH,wBAAgB,MAAM,CACpB,MAAM,EAAE,SAAS,cAAc,EAAE,EACjC,OAAO,GAAE,WAAgB,GACxB,MAAM,CAER"}

package/dist/format/html.js

@@ -1,49 +1,13 @@
-import { isUncertain } from "../decode/confidence.js";
-export { convertToAnnotatedHtml, convertToHtml, toAnnotatedHtml, } from "./annotate.js";
-import { READING_CHARGE } from "../decode/lattice.js";
-import { DEFAULT_LOCALE } from "../script/script.js";
-import { writeSyllable } from "../syllable/syllable.js";
-/**
- * The class every syllable carries, whatever else it carries.
- */
-const SYLLABLE_CLASS = "py-syllable";
-const UNCERTAIN_CLASS = "py-uncertain";
 /**
- * What a syllable declares itself to be, in the reading standard it was read in.
- *
- * A syllable element holds Mandarin written in the Latin alphabet, which is
- * neither the `zh` of the surrounding page nor the `en` of a page that quotes
- * it — and nothing about `yín` on its own says so. The tag is what a screen
- * reader consults before deciding how to pronounce it, and what a browser
- * consults for hyphenation and font selection; without one, `xíng` is read as
- * whatever the page around it claims to be, which is how pinyin ends up
- * spoken as English.
- *
- * The subtags are all registered and mean exactly this: `Latn` for the script,
- * the region for the reading standard, and the `pinyin` variant, whose prefix
- * in the IANA registry is `zh-Latn`. The region is the one distinction the
- * conversion itself makes — 垃圾 is `lājī` under `zh-CN` and `lèsè` under
- * `zh-TW` — so it is read off {@link ConvertOptions.locale} rather than
- * guessed at.
+ * A conversion as markup, and the options that decide what the markup carries.
  *
- * Tone notation does not enter into it. `hang2` is pinyin spelt with a tone
- * number, not another romanisation, so it takes the same tag as `háng`.
- *
- * The cost is the tag repeated on every syllable, which is the price of
- * wrapping nothing around the whole conversion. A caller who would rather
- * declare it once can set `lang: false` and put the same tag on a wrapper of
- * their own, which is inherited by everything inside it.
+ * One element per syllable, which is what lets a page colour a tone or mark a
+ * guess. What goes inside one of those elements is `syllable-span.ts` and the
+ * ruby that puts the reading over the hanzi is `annotate.ts`.
  */
-function languageTag(locale = DEFAULT_LOCALE) {
-    switch (locale) {
-        case "zh-CN": {
-            return "zh-Latn-CN-pinyin";
-        }
-        case "zh-TW": {
-            return "zh-Latn-TW-pinyin";
-        }
-    }
-}
+export { convertToAnnotatedHtml, convertToHtml, toAnnotatedHtml, } from "./annotate.js";
+import { writePiece } from "./syllable-span.js";
+export { writePiece } from "./syllable-span.js";
 /**
  * How each character that means something in markup is written as text.
  */
@@ -69,49 +33,36 @@ export function escape(text) {
     });
 }
 /**
- * The readings a syllable was chosen over, as the reader would write them.
+ * Mark up a run of pieces, with the system's join between two syllables.
  *
- * Only the ones that made it uncertain: a rival dearer than
- * {@link READING_CHARGE} could only be taken by breaking a dictionary word
- * apart, so it was never really in the running and listing it would suggest
- * otherwise.
- */
-function alternativesOf(confidence, options) {
-    const written = confidence.alternatives
-        .filter((alternative) => alternative.cost < READING_CHARGE)
-        .map((alternative) => alternative.reading
-        .map((syllable) => writeSyllable(syllable, options.notation))
-        .join(""));
-    return [...new Set(written)];
-}
-/**
- * Mark up one piece of a conversion.
+ * Pinyin needs none. Its syllables are run together and the marks that divide
+ * them are pieces of their own. Every other system decides the join for itself,
+ * and the join is what separates the five at this level, so it is written
+ * between the spans rather than inside one. `<span>Pei³</span>-<span>ching¹
+ * </span>` is one word in Wade-Giles and leaves each syllable its own element
+ * to carry a tone class on.
+ *
+ * A piece that wrote nothing does not break the run, which is what carries the
+ * join over pinyin's own hyphen: 干干净净 is four syllables of one word, and
+ * the mark between the second and the third is dropped rather than counted as a
+ * gap.
  */
-export function writePiece(piece, options) {
-    const { toneClasses = true, markUncertain = true, lang = true } = options;
-    if (piece.syllable === undefined) {
-        return escape(piece.text);
+export function writePieces(pieces, options) {
+    const { transcription } = options;
+    if (transcription === undefined) {
+        return pieces.map((piece) => writePiece(piece, options)).join("");
     }
-    const classes = [SYLLABLE_CLASS];
-    if (toneClasses && piece.syllable.tone !== undefined) {
-        classes.push(`py-tone-${String(piece.syllable.tone)}`);
-    }
-    const alternatives = markUncertain &&
-        piece.confidence !== undefined &&
-        isUncertain(piece.confidence)
-        ? alternativesOf(piece.confidence, options)
-        : [];
-    if (alternatives.length > 0) {
-        classes.push(UNCERTAIN_CLASS);
-    }
-    const attributes = [`class="${classes.join(" ")}"`];
-    if (lang) {
-        attributes.push(`lang="${languageTag(options.locale)}"`);
-    }
-    if (alternatives.length > 0) {
-        attributes.push(`data-alternatives="${escape(alternatives.join(" "))}"`);
+    const written = [];
+    let isJoined = false;
+    for (const piece of pieces) {
+        const markup = writePiece(piece, options);
+        if (piece.syllable !== undefined && isJoined) {
+            written.push(transcription.separator);
+        }
+        written.push(markup);
+        isJoined = piece.syllable !== undefined || (isJoined && markup === "");
     }
-    return `<span ${attributes.join(" ")}>${escape(piece.text)}</span>`;
+    return written.join("");
 }
 /**
  * Mark up a converted text, one element per syllable.
@@ -120,6 +71,6 @@ export function writePiece(piece, options) {
  * once and render more than one way.
  */
 export function toHtml(pieces, options = {}) {
-    return pieces.map((piece) => writePiece(piece, options)).join("");
+    return writePieces(pieces, 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,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"}
+{"version":3,"file":"html.js","sourceRoot":"","sources":["../../src/format/html.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,EACL,sBAAsB,EACtB,aAAa,EACb,eAAe,GAChB,MAAM,eAAe,CAAC;AAGvB,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AAEhD,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AA6ChD;;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;;;;;;;;;;;;;;GAcG;AACH,MAAM,UAAU,WAAW,CACzB,MAAiC,EACjC,OAAoB;IAEpB,MAAM,EAAE,aAAa,EAAE,GAAG,OAAO,CAAC;IAClC,IAAI,aAAa,KAAK,SAAS,EAAE,CAAC;QAChC,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;IACD,MAAM,OAAO,GAAa,EAAE,CAAC;IAC7B,IAAI,QAAQ,GAAG,KAAK,CAAC;IACrB,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;QAC3B,MAAM,MAAM,GAAG,UAAU,CAAC,KAAK,EAAE,OAAO,CAAC,CAAC;QAC1C,IAAI,KAAK,CAAC,QAAQ,KAAK,SAAS,IAAI,QAAQ,EAAE,CAAC;YAC7C,OAAO,CAAC,IAAI,CAAC,aAAa,CAAC,SAAS,CAAC,CAAC;QACxC,CAAC;QACD,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QACrB,QAAQ,GAAG,KAAK,CAAC,QAAQ,KAAK,SAAS,IAAI,CAAC,QAAQ,IAAI,MAAM,KAAK,EAAE,CAAC,CAAC;IACzE,CAAC;IACD,OAAO,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;AAC1B,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,MAAM,CACpB,MAAiC,EACjC,OAAO,GAAgB,EAAE;IAEzB,OAAO,WAAW,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;AACtC,CAAC"}

package/dist/format/language-tag.d.ts

@@ -0,0 +1,57 @@
+/**
+ * What a written reading declares itself to be, in BCP 47.
+ *
+ * One place that knows the subtags and the order they go in, since the tag is
+ * the same question whether the reading is pinyin or one of the five systems
+ * that can stand in for it.
+ */
+import { type Locale } from "../script/script.js";
+import type { TranscriptionSystem } from "../transcription/systems.js";
+/**
+ * The BCP 47 region subtag for a reading standard.
+ */
+export declare function regionOf(locale: Locale): string;
+/**
+ * What a syllable declares itself to be, in the system it was written in.
+ *
+ * A syllable element holds Mandarin written in some script that is not the `zh`
+ * of the surrounding page and not the `en` of a page that quotes it, and
+ * nothing about `yín` on its own says so. The tag is what a screen reader
+ * consults before deciding how to pronounce it, and what a browser consults for
+ * hyphenation and font selection. Without one, `xíng` is read as whatever the
+ * page around it claims to be, which is how pinyin ends up spoken as English.
+ *
+ * The subtags are registered and mean exactly this. The script is `Latn` for a
+ * romanisation and `Bopo` for bopomofo, the region is the reading standard, and
+ * the variant names the system where the IANA registry has one. It has `pinyin`
+ * and `wadegile`, both with the prefix `zh-Latn`, and `fonipa` for the IPA.
+ *
+ * | System | `zh-CN` | `zh-TW` |
+ * | --- | --- | --- |
+ * | pinyin | `zh-Latn-CN-pinyin` | `zh-Latn-TW-pinyin` |
+ * | bopomofo | `zh-Bopo-CN` | `zh-Bopo-TW` |
+ * | Wade-Giles | `zh-Latn-CN-wadegile` | `zh-Latn-TW-wadegile` |
+ * | Yale | `zh-Latn-CN` | `zh-Latn-TW` |
+ * | Gwoyeu Romatzyh | `zh-Latn-CN` | `zh-Latn-TW` |
+ * | IPA | `zh-Latn-CN-fonipa` | `zh-Latn-TW-fonipa` |
+ *
+ * **The region is carried by every one of them**, because the distinction it
+ * marks is in the reading rather than in the spelling. 垃圾 is `lājī` under
+ * `zh-CN` and `lèsè` under `zh-TW`, and it is two different words in bopomofo
+ * for the same reason, so the region is read off {@link ConvertOptions.locale}
+ * rather than guessed at.
+ *
+ * Yale and Gwoyeu Romatzyh name themselves nowhere, the registry having no
+ * variant for either, so both go out saying only that they are Mandarin in the
+ * Latin alphabet. See {@link TranscriptionSystem.variant}.
+ *
+ * Tone notation does not enter into it. `hang2` is pinyin spelt with a tone
+ * number, not another romanisation, so it takes the same tag as `háng`.
+ *
+ * The cost is the tag repeated on every syllable, which is the price of
+ * wrapping nothing around the whole conversion. A caller who would rather
+ * declare it once can set `lang: false` and put the same tag on a wrapper of
+ * their own, which is inherited by everything inside it.
+ */
+export declare function languageTag(locale?: Locale, system?: TranscriptionSystem): string;
+//# sourceMappingURL=language-tag.d.ts.map

package/dist/format/language-tag.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"language-tag.d.ts","sourceRoot":"","sources":["../../src/format/language-tag.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,EAAkB,KAAK,MAAM,EAAE,MAAM,qBAAqB,CAAC;AAClE,OAAO,KAAK,EAAE,mBAAmB,EAAE,MAAM,6BAA6B,CAAC;AASvE;;GAEG;AACH,wBAAgB,QAAQ,CAAC,MAAM,EAAE,MAAM,GAAG,MAAM,CAS/C;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAyCG;AACH,wBAAgB,WAAW,CACzB,MAAM,GAAE,MAAuB,EAC/B,MAAM,CAAC,EAAE,mBAAmB,GAC3B,MAAM,CAKR"}

package/dist/format/language-tag.js

@@ -0,0 +1,76 @@
+/**
+ * What a written reading declares itself to be, in BCP 47.
+ *
+ * One place that knows the subtags and the order they go in, since the tag is
+ * the same question whether the reading is pinyin or one of the five systems
+ * that can stand in for it.
+ */
+import { DEFAULT_LOCALE } from "../script/script.js";
+/**
+ * The pinyin variant's script and its own subtag, which a conversion writes by
+ * default.
+ */
+const PINYIN_SCRIPT = "Latn";
+const PINYIN_VARIANT = "pinyin";
+/**
+ * The BCP 47 region subtag for a reading standard.
+ */
+export function regionOf(locale) {
+    switch (locale) {
+        case "zh-CN": {
+            return "CN";
+        }
+        case "zh-TW": {
+            return "TW";
+        }
+    }
+}
+/**
+ * What a syllable declares itself to be, in the system it was written in.
+ *
+ * A syllable element holds Mandarin written in some script that is not the `zh`
+ * of the surrounding page and not the `en` of a page that quotes it, and
+ * nothing about `yín` on its own says so. The tag is what a screen reader
+ * consults before deciding how to pronounce it, and what a browser consults for
+ * hyphenation and font selection. Without one, `xíng` is read as whatever the
+ * page around it claims to be, which is how pinyin ends up spoken as English.
+ *
+ * The subtags are registered and mean exactly this. The script is `Latn` for a
+ * romanisation and `Bopo` for bopomofo, the region is the reading standard, and
+ * the variant names the system where the IANA registry has one. It has `pinyin`
+ * and `wadegile`, both with the prefix `zh-Latn`, and `fonipa` for the IPA.
+ *
+ * | System | `zh-CN` | `zh-TW` |
+ * | --- | --- | --- |
+ * | pinyin | `zh-Latn-CN-pinyin` | `zh-Latn-TW-pinyin` |
+ * | bopomofo | `zh-Bopo-CN` | `zh-Bopo-TW` |
+ * | Wade-Giles | `zh-Latn-CN-wadegile` | `zh-Latn-TW-wadegile` |
+ * | Yale | `zh-Latn-CN` | `zh-Latn-TW` |
+ * | Gwoyeu Romatzyh | `zh-Latn-CN` | `zh-Latn-TW` |
+ * | IPA | `zh-Latn-CN-fonipa` | `zh-Latn-TW-fonipa` |
+ *
+ * **The region is carried by every one of them**, because the distinction it
+ * marks is in the reading rather than in the spelling. 垃圾 is `lājī` under
+ * `zh-CN` and `lèsè` under `zh-TW`, and it is two different words in bopomofo
+ * for the same reason, so the region is read off {@link ConvertOptions.locale}
+ * rather than guessed at.
+ *
+ * Yale and Gwoyeu Romatzyh name themselves nowhere, the registry having no
+ * variant for either, so both go out saying only that they are Mandarin in the
+ * Latin alphabet. See {@link TranscriptionSystem.variant}.
+ *
+ * Tone notation does not enter into it. `hang2` is pinyin spelt with a tone
+ * number, not another romanisation, so it takes the same tag as `háng`.
+ *
+ * The cost is the tag repeated on every syllable, which is the price of
+ * wrapping nothing around the whole conversion. A caller who would rather
+ * declare it once can set `lang: false` and put the same tag on a wrapper of
+ * their own, which is inherited by everything inside it.
+ */
+export function languageTag(locale = DEFAULT_LOCALE, system) {
+    const script = system?.script ?? PINYIN_SCRIPT;
+    const variant = system === undefined ? PINYIN_VARIANT : system.variant;
+    const tag = `zh-${script}-${regionOf(locale)}`;
+    return variant === undefined ? tag : `${tag}-${variant}`;
+}
+//# sourceMappingURL=language-tag.js.map

package/dist/format/language-tag.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"language-tag.js","sourceRoot":"","sources":["../../src/format/language-tag.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,EAAE,cAAc,EAAe,MAAM,qBAAqB,CAAC;AAGlE;;;GAGG;AACH,MAAM,aAAa,GAAG,MAAM,CAAC;AAC7B,MAAM,cAAc,GAAG,QAAQ,CAAC;AAEhC;;GAEG;AACH,MAAM,UAAU,QAAQ,CAAC,MAAc;IACrC,QAAQ,MAAM,EAAE,CAAC;QACf,KAAK,OAAO,EAAE,CAAC;YACb,OAAO,IAAI,CAAC;QACd,CAAC;QACD,KAAK,OAAO,EAAE,CAAC;YACb,OAAO,IAAI,CAAC;QACd,CAAC;IACH,CAAC;AACH,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAyCG;AACH,MAAM,UAAU,WAAW,CACzB,MAAM,GAAW,cAAc,EAC/B,MAA4B;IAE5B,MAAM,MAAM,GAAG,MAAM,EAAE,MAAM,IAAI,aAAa,CAAC;IAC/C,MAAM,OAAO,GAAG,MAAM,KAAK,SAAS,CAAC,CAAC,CAAC,cAAc,CAAC,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC;IACvE,MAAM,GAAG,GAAG,MAAM,MAAM,IAAI,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;IAC/C,OAAO,OAAO,KAAK,SAAS,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,GAAG,IAAI,OAAO,EAAE,CAAC;AAC3D,CAAC"}