@kensio/pinyinjs 1.7.3 → 1.7.5

package/dist/transcription/bopomofo-reverse.js

@@ -0,0 +1,48 @@
+/**
+ * The symbol tables inverted, for reading bopomofo back.
+ *
+ * Every one of these is derived from the write-side table it mirrors rather
+ * than written out again, so the two cannot disagree.
+ */
+import { toCharacters } from "../script/characters.js";
+import { FINAL_PARTS, INITIAL_SYMBOLS, MEDIAL_SYMBOLS, RHYME_SYMBOLS, TONES_BY_MARK, } from "./bopomofo-tables.js";
+/**
+ * The final each medial and rhyme pair spells, keyed as `medial|rhyme`.
+ */
+export const FINALS_BY_PARTS = new Map(Object.entries(FINAL_PARTS).map(([final, [medial, rhyme]]) => [
+    `${medial}|${rhyme}`,
+    final,
+]));
+/**
+ * The initial each symbol stands for.
+ */
+export const INITIALS_BY_SYMBOL = new Map([...INITIAL_SYMBOLS].map(([initial, symbol]) => [symbol, initial]));
+/**
+ * The medial each symbol stands for.
+ */
+export const MEDIALS_BY_SYMBOL = new Map([...MEDIAL_SYMBOLS].map(([medial, symbol]) => [symbol, medial]));
+/**
+ * The rhyme each symbol stands for.
+ */
+export const RHYMES_BY_SYMBOL = new Map([...RHYME_SYMBOLS].map(([rhyme, symbol]) => [symbol, rhyme]));
+/**
+ * Whether one character is a bopomofo letter or tone mark.
+ */
+export function isSymbol(symbol) {
+    return (INITIALS_BY_SYMBOL.has(symbol) ||
+        MEDIALS_BY_SYMBOL.has(symbol) ||
+        RHYMES_BY_SYMBOL.has(symbol) ||
+        TONES_BY_MARK.has(symbol));
+}
+/**
+ * Whether a string is written in bopomofo at all.
+ *
+ * This is the one thing bopomofo makes easy that Wade-Giles cannot: it has a
+ * script of its own, so it is never mistaken for pinyin and a caller never has
+ * to be told which system its input is in.
+ */
+export function isBopomofo(text) {
+    const trimmed = text.trim();
+    return (trimmed !== "" && toCharacters(trimmed).every((symbol) => isSymbol(symbol)));
+}
+//# sourceMappingURL=bopomofo-reverse.js.map

package/dist/transcription/bopomofo-reverse.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"bopomofo-reverse.js","sourceRoot":"","sources":["../../src/transcription/bopomofo-reverse.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAEvD,OAAO,EACL,WAAW,EACX,eAAe,EAEf,cAAc,EAEd,aAAa,EACb,aAAa,GACd,MAAM,sBAAsB,CAAC;AAE9B;;GAEG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG,IAAI,GAAG,CACpC,MAAM,CAAC,OAAO,CAAC,WAAW,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,KAAK,EAAE,CAAC,MAAM,EAAE,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC;IAC5D,GAAG,MAAM,IAAI,KAAK,EAAE;IACpB,KAAc;CACf,CAAC,CACH,CAAC;AAEF;;GAEG;AACH,MAAM,CAAC,MAAM,kBAAkB,GAAG,IAAI,GAAG,CACvC,CAAC,GAAG,eAAe,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,OAAO,EAAE,MAAM,CAAC,EAAE,EAAE,CAAC,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC,CACnE,CAAC;AAEF;;GAEG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAAG,IAAI,GAAG,CACtC,CAAC,GAAG,cAAc,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,EAAE,CAAC,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAChE,CAAC;AAEF;;GAEG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAG,IAAI,GAAG,CACrC,CAAC,GAAG,aAAa,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,KAAK,EAAE,MAAM,CAAC,EAAE,EAAE,CAAC,CAAC,MAAM,EAAE,KAAK,CAAC,CAAC,CAC7D,CAAC;AAEF;;GAEG;AACH,MAAM,UAAU,QAAQ,CAAC,MAAc;IACrC,OAAO,CACL,kBAAkB,CAAC,GAAG,CAAC,MAAM,CAAC;QAC9B,iBAAiB,CAAC,GAAG,CAAC,MAAM,CAAC;QAC7B,gBAAgB,CAAC,GAAG,CAAC,MAAM,CAAC;QAC5B,aAAa,CAAC,GAAG,CAAC,MAAM,CAAC,CAC1B,CAAC;AACJ,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,UAAU,CAAC,IAAY;IACrC,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;IAC5B,OAAO,CACL,OAAO,KAAK,EAAE,IAAI,YAAY,CAAC,OAAO,CAAC,CAAC,KAAK,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAC5E,CAAC;AACJ,CAAC"}

package/dist/transcription/bopomofo-symbols.d.ts

@@ -0,0 +1,36 @@
+/**
+ * The bopomofo symbols, and writing a syllable in them.
+ *
+ * `bopomofo.ts` reads by inverting these same tables, so the dependency runs
+ * one way — nothing here knows how a string is parsed back.
+ */
+import type { Syllable } from "../syllable/syllable.js";
+export { EMPTY_RHYME_INITIALS, ERHUA_SYMBOL, FINAL_PARTS, INITIAL_SYMBOLS, type Medial, MEDIAL_SYMBOLS, type Rhyme, RHYME_SYMBOLS, TONES_BY_MARK, } from "./bopomofo-tables.js";
+/**
+ * How the first tone is written.
+ *
+ * Standard bopomofo leaves it unmarked, which is why `none` is the default. The
+ * cost is that a syllable whose tone was never written comes back as a first
+ * tone, since bopomofo has no way to say "no tone at all"; `mark` writes ˉ and
+ * keeps the two apart.
+ */
+export interface BopomofoOptions {
+    readonly firstTone?: "mark" | "none";
+}
+/**
+ * Write a syllable in bopomofo: 就 jiù becomes ㄐㄧㄡˋ.
+ *
+ * Total over well-formed syllables — every initial and every final has a
+ * symbol — so anything the syllable parser accepts can be written, including
+ * the syllables no dictionary uses.
+ */
+export declare function writeBopomofo(syllable: Syllable, options?: BopomofoOptions): string;
+/**
+ * Write a word in bopomofo, one syllable after another.
+ *
+ * Separated by spaces, as 教育部's dictionaries write them. A syllable boundary
+ * is findable without the space — an initial symbol can only start one — but a
+ * long unbroken run of symbols is hard to read, and the space costs nothing.
+ */
+export declare function writeBopomofoWord(syllables: readonly Syllable[], options?: BopomofoOptions): string;
+//# sourceMappingURL=bopomofo-symbols.d.ts.map

package/dist/transcription/bopomofo-symbols.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"bopomofo-symbols.d.ts","sourceRoot":"","sources":["../../src/transcription/bopomofo-symbols.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAYxD,OAAO,EACL,oBAAoB,EACpB,YAAY,EACZ,WAAW,EACX,eAAe,EACf,KAAK,MAAM,EACX,cAAc,EACd,KAAK,KAAK,EACV,aAAa,EACb,aAAa,GACd,MAAM,sBAAsB,CAAC;AAE9B;;;;;;;GAOG;AACH,MAAM,WAAW,eAAe;IAC9B,QAAQ,CAAC,SAAS,CAAC,EAAE,MAAM,GAAG,MAAM,CAAC;CACtC;AAED;;;;;;GAMG;AACH,wBAAgB,aAAa,CAC3B,QAAQ,EAAE,QAAQ,EAClB,OAAO,GAAE,eAAoB,GAC5B,MAAM,CAmBR;AAED;;;;;;GAMG;AACH,wBAAgB,iBAAiB,CAC/B,SAAS,EAAE,SAAS,QAAQ,EAAE,EAC9B,OAAO,GAAE,eAAoB,GAC5B,MAAM,CAIR"}

package/dist/transcription/bopomofo-symbols.js

@@ -0,0 +1,41 @@
+import { NEUTRAL_TONE } from "../tone/tone.js";
+import { EMPTY_RHYME_INITIALS, ERHUA_SYMBOL, FINAL_PARTS, INITIAL_SYMBOLS, MEDIAL_SYMBOLS, RHYME_SYMBOLS, TONE_MARKS, } from "./bopomofo-tables.js";
+export { EMPTY_RHYME_INITIALS, ERHUA_SYMBOL, FINAL_PARTS, INITIAL_SYMBOLS, MEDIAL_SYMBOLS, RHYME_SYMBOLS, TONES_BY_MARK, } from "./bopomofo-tables.js";
+/**
+ * Write a syllable in bopomofo: 就 jiù becomes ㄐㄧㄡˋ.
+ *
+ * Total over well-formed syllables — every initial and every final has a
+ * symbol — so anything the syllable parser accepts can be written, including
+ * the syllables no dictionary uses.
+ */
+export function writeBopomofo(syllable, options = {}) {
+    const { initial, final, tone } = syllable;
+    const [medial, rhyme] = FINAL_PARTS[final];
+    const isEmptyRhyme = final === "i" && EMPTY_RHYME_INITIALS.has(initial);
+    const written = [
+        INITIAL_SYMBOLS.get(initial) ?? "",
+        isEmptyRhyme ? "" : (MEDIAL_SYMBOLS.get(medial) ?? ""),
+        isEmptyRhyme ? "" : (RHYME_SYMBOLS.get(rhyme) ?? ""),
+    ].join("");
+    const suffix = syllable.erhua === true ? ERHUA_SYMBOL : "";
+    const mark = tone === undefined ? undefined : TONE_MARKS.get(tone);
+    if (mark === undefined || (tone === 1 && options.firstTone !== "mark")) {
+        return `${written}${suffix}`;
+    }
+    return tone === NEUTRAL_TONE
+        ? `${mark}${written}${suffix}`
+        : `${written}${mark}${suffix}`;
+}
+/**
+ * Write a word in bopomofo, one syllable after another.
+ *
+ * Separated by spaces, as 教育部's dictionaries write them. A syllable boundary
+ * is findable without the space — an initial symbol can only start one — but a
+ * long unbroken run of symbols is hard to read, and the space costs nothing.
+ */
+export function writeBopomofoWord(syllables, options = {}) {
+    return syllables
+        .map((syllable) => writeBopomofo(syllable, options))
+        .join(" ");
+}
+//# sourceMappingURL=bopomofo-symbols.js.map

package/dist/transcription/bopomofo-symbols.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"bopomofo-symbols.js","sourceRoot":"","sources":["../../src/transcription/bopomofo-symbols.ts"],"names":[],"mappings":"AAOA,OAAO,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAC/C,OAAO,EACL,oBAAoB,EACpB,YAAY,EACZ,WAAW,EACX,eAAe,EACf,cAAc,EACd,aAAa,EACb,UAAU,GACX,MAAM,sBAAsB,CAAC;AAE9B,OAAO,EACL,oBAAoB,EACpB,YAAY,EACZ,WAAW,EACX,eAAe,EAEf,cAAc,EAEd,aAAa,EACb,aAAa,GACd,MAAM,sBAAsB,CAAC;AAc9B;;;;;;GAMG;AACH,MAAM,UAAU,aAAa,CAC3B,QAAkB,EAClB,OAAO,GAAoB,EAAE;IAE7B,MAAM,EAAE,OAAO,EAAE,KAAK,EAAE,IAAI,EAAE,GAAG,QAAQ,CAAC;IAC1C,MAAM,CAAC,MAAM,EAAE,KAAK,CAAC,GAAG,WAAW,CAAC,KAAK,CAAC,CAAC;IAC3C,MAAM,YAAY,GAAG,KAAK,KAAK,GAAG,IAAI,oBAAoB,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;IAExE,MAAM,OAAO,GAAG;QACd,eAAe,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,EAAE;QAClC,YAAY,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,cAAc,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC;QACtD,YAAY,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,aAAa,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC;KACrD,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IACX,MAAM,MAAM,GAAG,QAAQ,CAAC,KAAK,KAAK,IAAI,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,EAAE,CAAC;IAE3D,MAAM,IAAI,GAAG,IAAI,KAAK,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,UAAU,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IACnE,IAAI,IAAI,KAAK,SAAS,IAAI,CAAC,IAAI,KAAK,CAAC,IAAI,OAAO,CAAC,SAAS,KAAK,MAAM,CAAC,EAAE,CAAC;QACvE,OAAO,GAAG,OAAO,GAAG,MAAM,EAAE,CAAC;IAC/B,CAAC;IACD,OAAO,IAAI,KAAK,YAAY;QAC1B,CAAC,CAAC,GAAG,IAAI,GAAG,OAAO,GAAG,MAAM,EAAE;QAC9B,CAAC,CAAC,GAAG,OAAO,GAAG,IAAI,GAAG,MAAM,EAAE,CAAC;AACnC,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,iBAAiB,CAC/B,SAA8B,EAC9B,OAAO,GAAoB,EAAE;IAE7B,OAAO,SAAS;SACb,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,aAAa,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC;SACnD,IAAI,CAAC,GAAG,CAAC,CAAC;AACf,CAAC"}

package/dist/transcription/bopomofo-tables.d.ts

@@ -0,0 +1,87 @@
+/**
+ * The bopomofo symbols themselves, as tables.
+ *
+ * Split from the writing so that each is one thing: this says which symbol
+ * stands for which part of a syllable, and `bopomofo-symbols.ts` puts them
+ * together.
+ */
+import type { Final, Initial } from "../syllable/phonology.js";
+/**
+ * The medial (介音) of a syllable, or the empty string for one with none.
+ */
+export type Medial = "" | "i" | "u" | "ü";
+/**
+ * The rhyme a syllable ends on, or the empty string where the medial is the
+ * whole of the final: 於 yú is ㄩ and nothing more.
+ */
+export type Rhyme = "" | "a" | "o" | "e" | "ê" | "ai" | "ei" | "ao" | "ou" | "an" | "en" | "ang" | "eng" | "er" | "m" | "n" | "ng";
+/**
+ * The symbol each initial is written with.
+ *
+ * A syllable with no initial has no entry and writes nothing, which is what the
+ * empty string stands for throughout this file.
+ */
+export declare const INITIAL_SYMBOLS: Map<Initial, string>;
+/**
+ * The symbol each medial is written with.
+ */
+export declare const MEDIAL_SYMBOLS: Map<Medial, string>;
+/**
+ * The symbol each rhyme is written with.
+ *
+ * ㄦ is here as a rhyme of its own and also serves as the 儿化 suffix, which is
+ * what bopomofo does with it: 歌儿 gēr is ㄍㄜㄦ.
+ *
+ * The three syllabic nasals at the end are the marginal case. ㄇ and ㄋ are
+ * initial letters doing rhyme duty, as bopomofo writes 呣 and 嗯; ㄫ is the
+ * obsolete letter for a syllabic ng, used here rather than ㄥ so that 嗯 ǹg and
+ * the rare 鞥 ēng do not come out identical.
+ */
+export declare const RHYME_SYMBOLS: Map<Rhyme, string>;
+/**
+ * How each final divides into a medial and a rhyme.
+ *
+ * Written out rather than split off the front of the spelling, because the
+ * division is phonological and the spelling only mostly follows it: `in` is
+ * ㄧㄣ, an i medial over the ㄣ rhyme rather than over an n of its own, and
+ * `iong` is ㄩㄥ — a ü medial despite the i in the spelling, which is why
+ * 兄 xiōng and 军 jūn share a medial.
+ */
+export declare const FINAL_PARTS: Record<Final, readonly [Medial, Rhyme]>;
+/**
+ * The initials whose syllable with an i final is written as the initial alone:
+ * 知 is ㄓ, not ㄓㄧ.
+ *
+ * The i of zhi, chi, shi, ri, zi, ci and si is not the i of ji at all — it is
+ * the syllabic continuation of the initial (空韻), and bopomofo simply does not
+ * write it. This is the one place the correspondence is not one symbol per
+ * part.
+ */
+export declare const EMPTY_RHYME_INITIALS: Set<Initial>;
+/**
+ * The symbol 儿化 is written with.
+ *
+ * ㄦ, the same letter as the rhyme of 兒 ér itself, which is why reading it
+ * takes the position into account: see {@link parseSymbols}.
+ */
+export declare const ERHUA_SYMBOL = "\u3126";
+/**
+ * The mark each tone takes.
+ *
+ * The neutral tone's dot goes *before* the syllable and the other four go
+ * after — but **before a 儿化 ㄦ rather than after it**: 哪儿 nǎr is ㄋㄚˇㄦ and
+ * 玩儿 wánr is ㄨㄢˊㄦ. The mark belongs to the syllable's nucleus, and the
+ * suffix is not part of what it marks. Wikipedia's *Bopomofo*, citing 教育部's
+ * 國語注音符號手冊, says so outright, and it is what 教育部's dictionaries and
+ * Wiktionary both write.
+ */
+export declare const TONE_MARKS: Map<1 | 2 | 3 | 4 | 5, string>;
+/**
+ * The tone each mark stands for, including the shapes a keyboard produces.
+ *
+ * Tolerant input and standard output, as the rest of the package has: `´` and
+ * `` ` `` are the ASCII neighbours of ˊ and ˋ, and are what text that was typed
+ * rather than composed tends to carry.
+ */
+export declare const TONES_BY_MARK: Map<string, 1 | 2 | 3 | 4 | 5>;
+//# sourceMappingURL=bopomofo-tables.d.ts.map

package/dist/transcription/bopomofo-tables.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"bopomofo-tables.d.ts","sourceRoot":"","sources":["../../src/transcription/bopomofo-tables.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,KAAK,EAAE,KAAK,EAAE,OAAO,EAAE,MAAM,0BAA0B,CAAC;AAG/D;;GAEG;AACH,MAAM,MAAM,MAAM,GAAG,EAAE,GAAG,GAAG,GAAG,GAAG,GAAG,GAAG,CAAC;AAE1C;;;GAGG;AACH,MAAM,MAAM,KAAK,GACb,EAAE,GACF,GAAG,GACH,GAAG,GACH,GAAG,GACH,GAAG,GACH,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,KAAK,GACL,KAAK,GACL,IAAI,GACJ,GAAG,GACH,GAAG,GACH,IAAI,CAAC;AAET;;;;;GAKG;AACH,eAAO,MAAM,eAAe,sBAsB1B,CAAC;AAEH;;GAEG;AACH,eAAO,MAAM,cAAc,qBAIzB,CAAC;AAEH;;;;;;;;;;GAUG;AACH,eAAO,MAAM,aAAa,oBAiBxB,CAAC;AAEH;;;;;;;;GAQG;AACH,eAAO,MAAM,WAAW,EAAE,MAAM,CAAC,KAAK,EAAE,SAAS,CAAC,MAAM,EAAE,KAAK,CAAC,CA0C/D,CAAC;AAEF;;;;;;;;GAQG;AACH,eAAO,MAAM,oBAAoB,cAQ/B,CAAC;AAEH;;;;;GAKG;AACH,eAAO,MAAM,YAAY,WAAM,CAAC;AAEhC;;;;;;;;;GASG;AACH,eAAO,MAAM,UAAU,gCAMrB,CAAC;AAEH;;;;;;GAMG;AACH,eAAO,MAAM,aAAa,gCAUxB,CAAC"}

package/dist/transcription/bopomofo-tables.js

@@ -0,0 +1,180 @@
+import { NEUTRAL_TONE } from "../tone/tone.js";
+/**
+ * The symbol each initial is written with.
+ *
+ * A syllable with no initial has no entry and writes nothing, which is what the
+ * empty string stands for throughout this file.
+ */
+export const INITIAL_SYMBOLS = new Map([
+    ["b", "ㄅ"],
+    ["p", "ㄆ"],
+    ["m", "ㄇ"],
+    ["f", "ㄈ"],
+    ["d", "ㄉ"],
+    ["t", "ㄊ"],
+    ["n", "ㄋ"],
+    ["l", "ㄌ"],
+    ["g", "ㄍ"],
+    ["k", "ㄎ"],
+    ["h", "ㄏ"],
+    ["j", "ㄐ"],
+    ["q", "ㄑ"],
+    ["x", "ㄒ"],
+    ["zh", "ㄓ"],
+    ["ch", "ㄔ"],
+    ["sh", "ㄕ"],
+    ["r", "ㄖ"],
+    ["z", "ㄗ"],
+    ["c", "ㄘ"],
+    ["s", "ㄙ"],
+]);
+/**
+ * The symbol each medial is written with.
+ */
+export const MEDIAL_SYMBOLS = new Map([
+    ["i", "ㄧ"],
+    ["u", "ㄨ"],
+    ["ü", "ㄩ"],
+]);
+/**
+ * The symbol each rhyme is written with.
+ *
+ * ㄦ is here as a rhyme of its own and also serves as the 儿化 suffix, which is
+ * what bopomofo does with it: 歌儿 gēr is ㄍㄜㄦ.
+ *
+ * The three syllabic nasals at the end are the marginal case. ㄇ and ㄋ are
+ * initial letters doing rhyme duty, as bopomofo writes 呣 and 嗯; ㄫ is the
+ * obsolete letter for a syllabic ng, used here rather than ㄥ so that 嗯 ǹg and
+ * the rare 鞥 ēng do not come out identical.
+ */
+export const RHYME_SYMBOLS = new Map([
+    ["a", "ㄚ"],
+    ["o", "ㄛ"],
+    ["e", "ㄜ"],
+    ["ê", "ㄝ"],
+    ["ai", "ㄞ"],
+    ["ei", "ㄟ"],
+    ["ao", "ㄠ"],
+    ["ou", "ㄡ"],
+    ["an", "ㄢ"],
+    ["en", "ㄣ"],
+    ["ang", "ㄤ"],
+    ["eng", "ㄥ"],
+    ["er", "ㄦ"],
+    ["m", "ㄇ"],
+    ["n", "ㄋ"],
+    ["ng", "ㄫ"],
+]);
+/**
+ * How each final divides into a medial and a rhyme.
+ *
+ * Written out rather than split off the front of the spelling, because the
+ * division is phonological and the spelling only mostly follows it: `in` is
+ * ㄧㄣ, an i medial over the ㄣ rhyme rather than over an n of its own, and
+ * `iong` is ㄩㄥ — a ü medial despite the i in the spelling, which is why
+ * 兄 xiōng and 军 jūn share a medial.
+ */
+export const FINAL_PARTS = {
+    a: ["", "a"],
+    o: ["", "o"],
+    e: ["", "e"],
+    ê: ["", "ê"],
+    er: ["", "er"],
+    ai: ["", "ai"],
+    ei: ["", "ei"],
+    ao: ["", "ao"],
+    ou: ["", "ou"],
+    an: ["", "an"],
+    en: ["", "en"],
+    ang: ["", "ang"],
+    eng: ["", "eng"],
+    ong: ["u", "eng"],
+    i: ["i", ""],
+    ia: ["i", "a"],
+    io: ["i", "o"],
+    ie: ["i", "ê"],
+    iao: ["i", "ao"],
+    iou: ["i", "ou"],
+    ian: ["i", "an"],
+    in: ["i", "en"],
+    iang: ["i", "ang"],
+    ing: ["i", "eng"],
+    iong: ["ü", "eng"],
+    u: ["u", ""],
+    ua: ["u", "a"],
+    uo: ["u", "o"],
+    uai: ["u", "ai"],
+    uei: ["u", "ei"],
+    uan: ["u", "an"],
+    uen: ["u", "en"],
+    uang: ["u", "ang"],
+    ueng: ["u", "eng"],
+    ü: ["ü", ""],
+    üe: ["ü", "ê"],
+    üan: ["ü", "an"],
+    ün: ["ü", "en"],
+    m: ["", "m"],
+    n: ["", "n"],
+    ng: ["", "ng"],
+};
+/**
+ * The initials whose syllable with an i final is written as the initial alone:
+ * 知 is ㄓ, not ㄓㄧ.
+ *
+ * The i of zhi, chi, shi, ri, zi, ci and si is not the i of ji at all — it is
+ * the syllabic continuation of the initial (空韻), and bopomofo simply does not
+ * write it. This is the one place the correspondence is not one symbol per
+ * part.
+ */
+export const EMPTY_RHYME_INITIALS = new Set([
+    "zh",
+    "ch",
+    "sh",
+    "r",
+    "z",
+    "c",
+    "s",
+]);
+/**
+ * The symbol 儿化 is written with.
+ *
+ * ㄦ, the same letter as the rhyme of 兒 ér itself, which is why reading it
+ * takes the position into account: see {@link parseSymbols}.
+ */
+export const ERHUA_SYMBOL = "ㄦ";
+/**
+ * The mark each tone takes.
+ *
+ * The neutral tone's dot goes *before* the syllable and the other four go
+ * after — but **before a 儿化 ㄦ rather than after it**: 哪儿 nǎr is ㄋㄚˇㄦ and
+ * 玩儿 wánr is ㄨㄢˊㄦ. The mark belongs to the syllable's nucleus, and the
+ * suffix is not part of what it marks. Wikipedia's *Bopomofo*, citing 教育部's
+ * 國語注音符號手冊, says so outright, and it is what 教育部's dictionaries and
+ * Wiktionary both write.
+ */
+export const TONE_MARKS = new Map([
+    [1, "ˉ"],
+    [2, "ˊ"],
+    [3, "ˇ"],
+    [4, "ˋ"],
+    [NEUTRAL_TONE, "˙"],
+]);
+/**
+ * The tone each mark stands for, including the shapes a keyboard produces.
+ *
+ * Tolerant input and standard output, as the rest of the package has: `´` and
+ * `` ` `` are the ASCII neighbours of ˊ and ˋ, and are what text that was typed
+ * rather than composed tends to carry.
+ */
+export const TONES_BY_MARK = new Map([
+    ["ˉ", 1],
+    ["¯", 1],
+    ["ˊ", 2],
+    ["´", 2],
+    ["ˇ", 3],
+    ["ˋ", 4],
+    ["`", 4],
+    ["˙", NEUTRAL_TONE],
+    ["·", NEUTRAL_TONE],
+]);
+//# sourceMappingURL=bopomofo-tables.js.map

package/dist/transcription/bopomofo-tables.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"bopomofo-tables.js","sourceRoot":"","sources":["../../src/transcription/bopomofo-tables.ts"],"names":[],"mappings":"AAQA,OAAO,EAAE,YAAY,EAAa,MAAM,iBAAiB,CAAC;AA8B1D;;;;;GAKG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG,IAAI,GAAG,CAAkB;IACtD,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,IAAI,EAAE,GAAG,CAAC;IACX,CAAC,IAAI,EAAE,GAAG,CAAC;IACX,CAAC,IAAI,EAAE,GAAG,CAAC;IACX,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;CACX,CAAC,CAAC;AAEH;;GAEG;AACH,MAAM,CAAC,MAAM,cAAc,GAAG,IAAI,GAAG,CAAiB;IACpD,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;CACX,CAAC,CAAC;AAEH;;;;;;;;;;GAUG;AACH,MAAM,CAAC,MAAM,aAAa,GAAG,IAAI,GAAG,CAAgB;IAClD,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,IAAI,EAAE,GAAG,CAAC;IACX,CAAC,IAAI,EAAE,GAAG,CAAC;IACX,CAAC,IAAI,EAAE,GAAG,CAAC;IACX,CAAC,IAAI,EAAE,GAAG,CAAC;IACX,CAAC,IAAI,EAAE,GAAG,CAAC;IACX,CAAC,IAAI,EAAE,GAAG,CAAC;IACX,CAAC,KAAK,EAAE,GAAG,CAAC;IACZ,CAAC,KAAK,EAAE,GAAG,CAAC;IACZ,CAAC,IAAI,EAAE,GAAG,CAAC;IACX,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,IAAI,EAAE,GAAG,CAAC;CACZ,CAAC,CAAC;AAEH;;;;;;;;GAQG;AACH,MAAM,CAAC,MAAM,WAAW,GAA4C;IAClE,CAAC,EAAE,CAAC,EAAE,EAAE,GAAG,CAAC;IACZ,CAAC,EAAE,CAAC,EAAE,EAAE,GAAG,CAAC;IACZ,CAAC,EAAE,CAAC,EAAE,EAAE,GAAG,CAAC;IACZ,CAAC,EAAE,CAAC,EAAE,EAAE,GAAG,CAAC;IACZ,EAAE,EAAE,CAAC,EAAE,EAAE,IAAI,CAAC;IACd,EAAE,EAAE,CAAC,EAAE,EAAE,IAAI,CAAC;IACd,EAAE,EAAE,CAAC,EAAE,EAAE,IAAI,CAAC;IACd,EAAE,EAAE,CAAC,EAAE,EAAE,IAAI,CAAC;IACd,EAAE,EAAE,CAAC,EAAE,EAAE,IAAI,CAAC;IACd,EAAE,EAAE,CAAC,EAAE,EAAE,IAAI,CAAC;IACd,EAAE,EAAE,CAAC,EAAE,EAAE,IAAI,CAAC;IACd,GAAG,EAAE,CAAC,EAAE,EAAE,KAAK,CAAC;IAChB,GAAG,EAAE,CAAC,EAAE,EAAE,KAAK,CAAC;IAChB,GAAG,EAAE,CAAC,GAAG,EAAE,KAAK,CAAC;IACjB,CAAC,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC;IACZ,EAAE,EAAE,CAAC,GAAG,EAAE,GAAG,CAAC;IACd,EAAE,EAAE,CAAC,GAAG,EAAE,GAAG,CAAC;IACd,EAAE,EAAE,CAAC,GAAG,EAAE,GAAG,CAAC;IACd,GAAG,EAAE,CAAC,GAAG,EAAE,IAAI,CAAC;IAChB,GAAG,EAAE,CAAC,GAAG,EAAE,IAAI,CAAC;IAChB,GAAG,EAAE,CAAC,GAAG,EAAE,IAAI,CAAC;IAChB,EAAE,EAAE,CAAC,GAAG,EAAE,IAAI,CAAC;IACf,IAAI,EAAE,CAAC,GAAG,EAAE,KAAK,CAAC;IAClB,GAAG,EAAE,CAAC,GAAG,EAAE,KAAK,CAAC;IACjB,IAAI,EAAE,CAAC,GAAG,EAAE,KAAK,CAAC;IAClB,CAAC,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC;IACZ,EAAE,EAAE,CAAC,GAAG,EAAE,GAAG,CAAC;IACd,EAAE,EAAE,CAAC,GAAG,EAAE,GAAG,CAAC;IACd,GAAG,EAAE,CAAC,GAAG,EAAE,IAAI,CAAC;IAChB,GAAG,EAAE,CAAC,GAAG,EAAE,IAAI,CAAC;IAChB,GAAG,EAAE,CAAC,GAAG,EAAE,IAAI,CAAC;IAChB,GAAG,EAAE,CAAC,GAAG,EAAE,IAAI,CAAC;IAChB,IAAI,EAAE,CAAC,GAAG,EAAE,KAAK,CAAC;IAClB,IAAI,EAAE,CAAC,GAAG,EAAE,KAAK,CAAC;IAClB,CAAC,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC;IACZ,EAAE,EAAE,CAAC,GAAG,EAAE,GAAG,CAAC;IACd,GAAG,EAAE,CAAC,GAAG,EAAE,IAAI,CAAC;IAChB,EAAE,EAAE,CAAC,GAAG,EAAE,IAAI,CAAC;IACf,CAAC,EAAE,CAAC,EAAE,EAAE,GAAG,CAAC;IACZ,CAAC,EAAE,CAAC,EAAE,EAAE,GAAG,CAAC;IACZ,EAAE,EAAE,CAAC,EAAE,EAAE,IAAI,CAAC;CACf,CAAC;AAEF;;;;;;;;GAQG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAAG,IAAI,GAAG,CAAU;IACnD,IAAI;IACJ,IAAI;IACJ,IAAI;IACJ,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;CACJ,CAAC,CAAC;AAEH;;;;;GAKG;AACH,MAAM,CAAC,MAAM,YAAY,GAAG,GAAG,CAAC;AAEhC;;;;;;;;;GASG;AACH,MAAM,CAAC,MAAM,UAAU,GAAG,IAAI,GAAG,CAAe;IAC9C,CAAC,CAAC,EAAE,GAAG,CAAC;IACR,CAAC,CAAC,EAAE,GAAG,CAAC;IACR,CAAC,CAAC,EAAE,GAAG,CAAC;IACR,CAAC,CAAC,EAAE,GAAG,CAAC;IACR,CAAC,YAAY,EAAE,GAAG,CAAC;CACpB,CAAC,CAAC;AAEH;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,aAAa,GAAG,IAAI,GAAG,CAAe;IACjD,CAAC,GAAG,EAAE,CAAC,CAAC;IACR,CAAC,GAAG,EAAE,CAAC,CAAC;IACR,CAAC,GAAG,EAAE,CAAC,CAAC;IACR,CAAC,GAAG,EAAE,CAAC,CAAC;IACR,CAAC,GAAG,EAAE,CAAC,CAAC;IACR,CAAC,GAAG,EAAE,CAAC,CAAC;IACR,CAAC,GAAG,EAAE,CAAC,CAAC;IACR,CAAC,GAAG,EAAE,YAAY,CAAC;IACnB,CAAC,GAAG,EAAE,YAAY,CAAC;CACpB,CAAC,CAAC"}

package/dist/transcription/bopomofo.d.ts

@@ -1,39 +1,21 @@
-import type { Syllable } from "../syllable/syllable.js";
-/**
- * How the first tone is written.
- *
- * Standard bopomofo leaves it unmarked, which is why `none` is the default. The
- * cost is that a syllable whose tone was never written comes back as a first
- * tone, since bopomofo has no way to say "no tone at all"; `mark` writes ˉ and
- * keeps the two apart.
- */
-export interface BopomofoOptions {
-    readonly firstTone?: "mark" | "none";
-}
-/**
- * Write a syllable in bopomofo: 就 jiù becomes ㄐㄧㄡˋ.
- *
- * Total over well-formed syllables — every initial and every final has a
- * symbol — so anything the syllable parser accepts can be written, including
- * the syllables no dictionary uses.
- */
-export declare function writeBopomofo(syllable: Syllable, options?: BopomofoOptions): string;
 /**
- * Write a word in bopomofo, one syllable after another.
+ * Bopomofo (注音符號), in both directions.
  *
- * Separated by spaces, as 教育部's dictionaries write them. A syllable boundary
- * is findable without the space — an initial symbol can only start one — but a
- * long unbroken run of symbols is hard to read, and the space costs nothing.
- */
-export declare function writeBopomofoWord(syllables: readonly Syllable[], options?: BopomofoOptions): string;
-/**
- * Whether a string is written in bopomofo at all.
+ * This is the easy half of romanisation interchange, and the reason is
+ * [phonology.ts](../syllable/phonology.ts): a parsed syllable already holds the
+ * underlying initial and final rather than the spelling, and bopomofo writes
+ * exactly those. 就 is
+ * `j` + `iou` and ㄐㄧㄡ; 军 is `j` + `ün` and ㄐㄩㄣ; 中 is `zh` + `ong` and
+ * ㄓㄨㄥ. Nothing here undoes a pinyin spelling rule, because `readSyllable`
+ * never applied one.
  *
- * This is the one thing bopomofo makes easy that Wade-Giles cannot: it has a
- * script of its own, so it is never mistaken for pinyin and a caller never has
- * to be told which system its input is in.
+ * So the mapping is compositional — an initial symbol, a medial symbol and a
+ * rhyme symbol — and it is a bijection over the whole syllable inventory. The
+ * three places it is not quite mechanical are recorded on the tables below.
  */
-export declare function isBopomofo(text: string): boolean;
+import type { Syllable } from "../syllable/syllable.js";
+export { type BopomofoOptions, writeBopomofo, writeBopomofoWord, } from "./bopomofo-symbols.js";
+export { isBopomofo } from "./bopomofo-reverse.js";
 /**
  * Read a bopomofo syllable: ㄐㄧㄡˋ becomes 就 jiù.
  *

package/dist/transcription/bopomofo.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"bopomofo.d.ts","sourceRoot":"","sources":["../../src/transcription/bopomofo.ts"],"names":[],"mappings":"AAiBA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAyNxD;;;;;;;GAOG;AACH,MAAM,WAAW,eAAe;IAC9B,QAAQ,CAAC,SAAS,CAAC,EAAE,MAAM,GAAG,MAAM,CAAC;CACtC;AAED;;;;;;GAMG;AACH,wBAAgB,aAAa,CAC3B,QAAQ,EAAE,QAAQ,EAClB,OAAO,GAAE,eAAoB,GAC5B,MAAM,CAmBR;AAED;;;;;;GAMG;AACH,wBAAgB,iBAAiB,CAC/B,SAAS,EAAE,SAAS,QAAQ,EAAE,EAC9B,OAAO,GAAE,eAAoB,GAC5B,MAAM,CAIR;AA6CD;;;;;;GAMG;AACH,wBAAgB,UAAU,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO,CAKhD;AA4HD;;;;;;;GAOG;AACH,wBAAgB,YAAY,CAAC,IAAI,EAAE,MAAM,GAAG,QAAQ,GAAG,SAAS,CAS/D"}
+{"version":3,"file":"bopomofo.d.ts","sourceRoot":"","sources":["../../src/transcription/bopomofo.ts"],"names":[],"mappings":"AACA;;;;;;;;;;;;;;GAcG;AACH,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAGxD,OAAO,EACL,KAAK,eAAe,EACpB,aAAa,EACb,iBAAiB,GAClB,MAAM,uBAAuB,CAAC;AAI/B,OAAO,EAAE,UAAU,EAAE,MAAM,uBAAuB,CAAC;AAoCnD;;;;;;;GAOG;AACH,wBAAgB,YAAY,CAAC,IAAI,EAAE,MAAM,GAAG,QAAQ,GAAG,SAAS,CAS/D"}