@kensio/pinyinjs 1.7.3 → 1.7.5

package/dist/dictionary/script-pairings.js

@@ -0,0 +1,120 @@
+/**
+ * Counting how the two scripts pair up, character by character.
+ *
+ * Every entry the dictionary holds is one vote for a 简体 character reading a
+ * given way being written a given 繁體 way. This is the tally, and the rules
+ * for reading a conversion out of it.
+ */
+import { toCharacters } from "../script/characters.js";
+import { conversionOf } from "./script-conversion.js";
+export { conversionOf, rivalsOf } from "./script-conversion.js";
+import { conversionKey, } from "../script/conversion.js";
+/**
+ * The key standing for "at any reading".
+ */
+export const ANY = "";
+/**
+ * How much of a reading's evidence must agree before it overrides the default.
+ *
+ * A reading that disagrees with the default only earns a line in the table if
+ * the sources are reasonably united about it. Below this it is noise — one
+ * mis-aligned entry, or a spelling nobody else uses — and storing it would
+ * make the conversion worse at the cost of bytes.
+ */
+export const READING_AGREEMENT = 0.8;
+/**
+ * How much of a character's evidence a rival form needs to be worth reporting.
+ *
+ * This decides honesty rather than output: a form listed here makes the
+ * character *ambiguous*, so every conversion of it is reported as a guess
+ * unless a word or a reading settled it. Set too low and 和 is doubtful because
+ * Unihan knows 咊; set too high and 面 looks certain when only the word tells
+ * 面 from 麵. A twentieth of the evidence is enough to be a real spelling and
+ * far more than the one-off variants clear.
+ */
+export const RIVAL_SHARE = 0.05;
+/**
+ * The pairings an entry attests, character by character.
+ *
+ * Entries whose two scripts are written with a different number of characters
+ * are skipped rather than aligned by guesswork, for the same reason
+ * `pairScripts` skips them: an invented alignment is evidence for a pairing
+ * nobody wrote.
+ */
+export function pairingsOf(entry) {
+    const hans = toCharacters(entry.hans);
+    const hant = toCharacters(entry.hant);
+    if (hans.length !== hant.length) {
+        return [];
+    }
+    const reading = entry.readings.cn;
+    // Only a reading with one syllable per character lines up; anything else
+    // (儿化 covering two, or a source disagreeing about length) is used as
+    // evidence for the pairing but not for any particular reading.
+    const isAligned = reading.length === hans.length;
+    return hans.map((character, at) => ({
+        hans: character,
+        /* c8 ignore next -- the two forms are known to be the same length */
+        hant: hant[at] ?? character,
+        key: isAligned ? conversionKey(reading[at]) : ANY,
+    }));
+}
+/**
+ * Tally one pairing under both its own reading and the any-reading key.
+ */
+export function tally(observed, from, to, key, weight) {
+    const byReading = observed.get(from) ?? new Map();
+    const keys = new Set([ANY, key]);
+    for (const at of keys) {
+        const counts = byReading.get(at) ?? new Map();
+        counts.set(to, (counts.get(to) ?? 0) + weight);
+        byReading.set(at, counts);
+    }
+    observed.set(from, byReading);
+}
+/**
+ * The commonest form in a tally, and what share of the evidence it holds.
+ *
+ * A tie prefers a form other than the character itself, and that matters more
+ * than it looks. Every character has an entry of its own where both scripts are
+ * the same string, so a character attested in exactly one cross-script word
+ * arrives here tied one-all against its own identity — 儁 against 㑺. The real
+ * pairing is the one somebody wrote down; the identity is an artefact of how
+ * the dictionary is keyed. `TraditionalTable` breaks the same tie the same way.
+ */
+export function commonest(from, counts) {
+    if (counts === undefined || counts.size === 0) {
+        return undefined;
+    }
+    let form = "";
+    let best = 0;
+    let total = 0;
+    for (const [candidate, count] of counts) {
+        total += count;
+        if (count > best ||
+            (count === best && form === from && candidate !== from)) {
+            form = candidate;
+            best = count;
+        }
+    }
+    return { form, share: best / total };
+}
+export function tableOf(observed) {
+    const table = new Map();
+    for (const [from, byReading] of observed) {
+        const conversion = conversionOf(from, byReading);
+        if (conversion !== undefined) {
+            table.set(from, conversion);
+        }
+    }
+    return table;
+}
+/**
+ * The readings a word is converted with, where they line up character by
+ * character.
+ */
+export function alignedReadings(entry) {
+    const characters = toCharacters(entry.hans).length;
+    return entry.readings.cn.length === characters ? entry.readings.cn : [];
+}
+//# sourceMappingURL=script-pairings.js.map

package/dist/dictionary/script-pairings.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"script-pairings.js","sourceRoot":"","sources":["../../src/dictionary/script-pairings.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AACvD,OAAO,EAAE,YAAY,EAAE,MAAM,wBAAwB,CAAC;AAEtD,OAAO,EAAE,YAAY,EAAE,QAAQ,EAAE,MAAM,wBAAwB,CAAC;AAChE,OAAO,EAEL,aAAa,GACd,MAAM,yBAAyB,CAAC;AAcjC;;GAEG;AACH,MAAM,CAAC,MAAM,GAAG,GAAG,EAAE,CAAC;AAEtB;;;;;;;GAOG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAAG,GAAG,CAAC;AAErC;;;;;;;;;GASG;AACH,MAAM,CAAC,MAAM,WAAW,GAAG,IAAI,CAAC;AAEhC;;;;;;;GAOG;AACH,MAAM,UAAU,UAAU,CACxB,KAAsB;IAEtB,MAAM,IAAI,GAAG,YAAY,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IACtC,MAAM,IAAI,GAAG,YAAY,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IACtC,IAAI,IAAI,CAAC,MAAM,KAAK,IAAI,CAAC,MAAM,EAAE,CAAC;QAChC,OAAO,EAAE,CAAC;IACZ,CAAC;IACD,MAAM,OAAO,GAAG,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC;IAClC,yEAAyE;IACzE,qEAAqE;IACrE,+DAA+D;IAC/D,MAAM,SAAS,GAAG,OAAO,CAAC,MAAM,KAAK,IAAI,CAAC,MAAM,CAAC;IACjD,OAAO,IAAI,CAAC,GAAG,CAAC,CAAC,SAAS,EAAE,EAAE,EAAE,EAAE,CAAC,CAAC;QAClC,IAAI,EAAE,SAAS;QACf,qEAAqE;QACrE,IAAI,EAAE,IAAI,CAAC,EAAE,CAAC,IAAI,SAAS;QAC3B,GAAG,EAAE,SAAS,CAAC,CAAC,CAAC,aAAa,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG;KAClD,CAAC,CAAC,CAAC;AACN,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,KAAK,CACnB,QAAgC,EAChC,IAAY,EACZ,EAAU,EACV,GAAW,EACX,MAAc;IAEd,MAAM,SAAS,GAAG,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,IAAI,GAAG,EAAsB,CAAC;IACtE,MAAM,IAAI,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;IACjC,KAAK,MAAM,EAAE,IAAI,IAAI,EAAE,CAAC;QACtB,MAAM,MAAM,GAAG,SAAS,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,IAAI,GAAG,EAAkB,CAAC;QAC9D,MAAM,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,CAAC,GAAG,MAAM,CAAC,CAAC;QAC/C,SAAS,CAAC,GAAG,CAAC,EAAE,EAAE,MAAM,CAAC,CAAC;IAC5B,CAAC;IACD,QAAQ,CAAC,GAAG,CAAC,IAAI,EAAE,SAAS,CAAC,CAAC;AAChC,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,UAAU,SAAS,CACvB,IAAY,EACZ,MAA8B;IAE9B,IAAI,MAAM,KAAK,SAAS,IAAI,MAAM,CAAC,IAAI,KAAK,CAAC,EAAE,CAAC;QAC9C,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,IAAI,IAAI,GAAG,EAAE,CAAC;IACd,IAAI,IAAI,GAAG,CAAC,CAAC;IACb,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,KAAK,MAAM,CAAC,SAAS,EAAE,KAAK,CAAC,IAAI,MAAM,EAAE,CAAC;QACxC,KAAK,IAAI,KAAK,CAAC;QACf,IACE,KAAK,GAAG,IAAI;YACZ,CAAC,KAAK,KAAK,IAAI,IAAI,IAAI,KAAK,IAAI,IAAI,SAAS,KAAK,IAAI,CAAC,EACvD,CAAC;YACD,IAAI,GAAG,SAAS,CAAC;YACjB,IAAI,GAAG,KAAK,CAAC;QACf,CAAC;IACH,CAAC;IACD,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,GAAG,KAAK,EAAE,CAAC;AACvC,CAAC;AAED,MAAM,UAAU,OAAO,CACrB,QAAgC;IAEhC,MAAM,KAAK,GAAG,IAAI,GAAG,EAA+B,CAAC;IACrD,KAAK,MAAM,CAAC,IAAI,EAAE,SAAS,CAAC,IAAI,QAAQ,EAAE,CAAC;QACzC,MAAM,UAAU,GAAG,YAAY,CAAC,IAAI,EAAE,SAAS,CAAC,CAAC;QACjD,IAAI,UAAU,KAAK,SAAS,EAAE,CAAC;YAC7B,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,UAAU,CAAC,CAAC;QAC9B,CAAC;IACH,CAAC;IACD,OAAO,KAAK,CAAC;AACf,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,eAAe,CAC7B,KAAsB;IAEtB,MAAM,UAAU,GAAG,YAAY,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,MAAM,CAAC;IACnD,OAAO,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,MAAM,KAAK,UAAU,CAAC,CAAC,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;AAC1E,CAAC"}

package/dist/dictionary/script-tables.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"script-tables.d.ts","sourceRoot":"","sources":["../../src/dictionary/script-tables.ts"],"names":[],"mappings":"AAEA,OAAO,EAIL,KAAK,YAAY,EAClB,MAAM,yBAAyB,CAAC;AAEjC,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,YAAY,CAAC;AAyNlD;;;;;;;;;;;;GAYG;AACH,wBAAgB,iBAAiB,CAC/B,OAAO,EAAE,SAAS,eAAe,EAAE,GAClC,YAAY,CA6Cd"}
+{"version":3,"file":"script-tables.d.ts","sourceRoot":"","sources":["../../src/dictionary/script-tables.ts"],"names":[],"mappings":"AAEA,OAAO,EAAqB,KAAK,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAC/E,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,YAAY,CAAC;AAQlD;;;;;;;;;;;;GAYG;AACH,wBAAgB,iBAAiB,CAC/B,OAAO,EAAE,SAAS,eAAe,EAAE,GAClC,YAAY,CA6Cd"}

package/dist/dictionary/script-tables.js

@@ -1,170 +1,7 @@
 import { toCharacters } from "../script/characters.js";
 import { toCanonicalGlyphs } from "../script/glyphs.js";
-import { conversionKey, convertCharacters, } from "../script/conversion.js";
-/**
- * The key standing for "at any reading".
- */
-const ANY = "";
-/**
- * How much of a reading's evidence must agree before it overrides the default.
- *
- * A reading that disagrees with the default only earns a line in the table if
- * the sources are reasonably united about it. Below this it is noise — one
- * mis-aligned entry, or a spelling nobody else uses — and storing it would
- * make the conversion worse at the cost of bytes.
- */
-const READING_AGREEMENT = 0.8;
-/**
- * How much of a character's evidence a rival form needs to be worth reporting.
- *
- * This decides honesty rather than output: a form listed here makes the
- * character *ambiguous*, so every conversion of it is reported as a guess
- * unless a word or a reading settled it. Set too low and 和 is doubtful because
- * Unihan knows 咊; set too high and 面 looks certain when only the word tells
- * 面 from 麵. A twentieth of the evidence is enough to be a real spelling and
- * far more than the one-off variants clear.
- */
-const RIVAL_SHARE = 0.05;
-/**
- * The pairings an entry attests, character by character.
- *
- * Entries whose two scripts are written with a different number of characters
- * are skipped rather than aligned by guesswork, for the same reason
- * `pairScripts` skips them: an invented alignment is evidence for a pairing
- * nobody wrote.
- */
-function pairingsOf(entry) {
-    const hans = toCharacters(entry.hans);
-    const hant = toCharacters(entry.hant);
-    if (hans.length !== hant.length) {
-        return [];
-    }
-    const reading = entry.readings.cn;
-    // Only a reading with one syllable per character lines up; anything else
-    // (儿化 covering two, or a source disagreeing about length) is used as
-    // evidence for the pairing but not for any particular reading.
-    const isAligned = reading.length === hans.length;
-    return hans.map((character, at) => ({
-        hans: character,
-        /* c8 ignore next -- the two forms are known to be the same length */
-        hant: hant[at] ?? character,
-        key: isAligned ? conversionKey(reading[at]) : ANY,
-    }));
-}
-/**
- * Tally one pairing under both its own reading and the any-reading key.
- */
-function tally(observed, from, to, key, weight) {
-    const byReading = observed.get(from) ?? new Map();
-    const keys = new Set([ANY, key]);
-    for (const at of keys) {
-        const counts = byReading.get(at) ?? new Map();
-        counts.set(to, (counts.get(to) ?? 0) + weight);
-        byReading.set(at, counts);
-    }
-    observed.set(from, byReading);
-}
-/**
- * The commonest form in a tally, and what share of the evidence it holds.
- *
- * A tie prefers a form other than the character itself, and that matters more
- * than it looks. Every character has an entry of its own where both scripts are
- * the same string, so a character attested in exactly one cross-script word
- * arrives here tied one-all against its own identity — 儁 against 㑺. The real
- * pairing is the one somebody wrote down; the identity is an artefact of how
- * the dictionary is keyed. `TraditionalTable` breaks the same tie the same way.
- */
-function commonest(from, counts) {
-    if (counts === undefined || counts.size === 0) {
-        return undefined;
-    }
-    let form = "";
-    let best = 0;
-    let total = 0;
-    for (const [candidate, count] of counts) {
-        total += count;
-        if (count > best ||
-            (count === best && form === from && candidate !== from)) {
-            form = candidate;
-            best = count;
-        }
-    }
-    return { form, share: best / total };
-}
-/**
- * Reduce one character's observations to a default and its exceptions.
- *
- * A character with only one observed form needs no entry at all when that form
- * is itself — simplification changed a minority of characters, and storing the
- * other 40,000 as identities would be most of the file.
- */
-function conversionOf(from, byReading) {
-    const overall = commonest(from, byReading.get(ANY));
-    if (overall === undefined) {
-        return undefined;
-    }
-    const exceptions = new Map();
-    for (const [key, counts] of byReading) {
-        const at = commonest(from, counts);
-        if (key !== ANY &&
-            at !== undefined &&
-            at.form !== overall.form &&
-            at.share >= READING_AGREEMENT) {
-            exceptions.set(key, at.form);
-        }
-    }
-    const also = rivalsOf(byReading.get(ANY), overall.form, exceptions);
-    if (overall.form === from && exceptions.size === 0 && also.length === 0) {
-        return undefined;
-    }
-    return {
-        to: overall.form,
-        ...(exceptions.size > 0 && { byReading: exceptions }),
-        ...(also.length > 0 && { also }),
-    };
-}
-/**
- * The forms a character takes often enough to make it genuinely ambiguous.
- *
- * Forms a reading already accounts for are left out: those are settled evidence
- * rather than open questions, and repeating them here would report 发 as a
- * guess in 头发 when the reading decided it.
- */
-function rivalsOf(counts, chosen, exceptions) {
-    if (counts === undefined) {
-        return [];
-    }
-    let total = 0;
-    for (const count of counts.values()) {
-        total += count;
-    }
-    const settled = new Set([chosen, ...exceptions.values()]);
-    return [...counts]
-        .filter(([form, count]) => !settled.has(form) && count / total >= RIVAL_SHARE)
-        .toSorted(([, left], [, right]) => right - left)
-        .map(([form]) => form);
-}
-/**
- * Reduce every character's observations into a table.
- */
-function tableOf(observed) {
-    const table = new Map();
-    for (const [from, byReading] of observed) {
-        const conversion = conversionOf(from, byReading);
-        if (conversion !== undefined) {
-            table.set(from, conversion);
-        }
-    }
-    return table;
-}
-/**
- * The readings a word is converted with, where they line up character by
- * character.
- */
-function alignedReadings(entry) {
-    const characters = toCharacters(entry.hans).length;
-    return entry.readings.cn.length === characters ? entry.readings.cn : [];
-}
+import { convertCharacters } from "../script/conversion.js";
+import { alignedReadings, pairingsOf, tableOf, tally, } from "./script-pairings.js";
 /**
  * Build the script conversion tables from the merged dictionary.
  *

package/dist/dictionary/script-tables.js.map

@@ -1 +1 @@
-{"version":3,"file":"script-tables.js","sourceRoot":"","sources":["../../src/dictionary/script-tables.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AACvD,OAAO,EAAE,iBAAiB,EAAE,MAAM,qBAAqB,CAAC;AACxD,OAAO,EAEL,aAAa,EACb,iBAAiB,GAElB,MAAM,yBAAyB,CAAC;AAcjC;;GAEG;AACH,MAAM,GAAG,GAAG,EAAE,CAAC;AAEf;;;;;;;GAOG;AACH,MAAM,iBAAiB,GAAG,GAAG,CAAC;AAE9B;;;;;;;;;GASG;AACH,MAAM,WAAW,GAAG,IAAI,CAAC;AAEzB;;;;;;;GAOG;AACH,SAAS,UAAU,CACjB,KAAsB;IAEtB,MAAM,IAAI,GAAG,YAAY,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IACtC,MAAM,IAAI,GAAG,YAAY,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IACtC,IAAI,IAAI,CAAC,MAAM,KAAK,IAAI,CAAC,MAAM,EAAE,CAAC;QAChC,OAAO,EAAE,CAAC;IACZ,CAAC;IACD,MAAM,OAAO,GAAG,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC;IAClC,yEAAyE;IACzE,qEAAqE;IACrE,+DAA+D;IAC/D,MAAM,SAAS,GAAG,OAAO,CAAC,MAAM,KAAK,IAAI,CAAC,MAAM,CAAC;IACjD,OAAO,IAAI,CAAC,GAAG,CAAC,CAAC,SAAS,EAAE,EAAE,EAAE,EAAE,CAAC,CAAC;QAClC,IAAI,EAAE,SAAS;QACf,qEAAqE;QACrE,IAAI,EAAE,IAAI,CAAC,EAAE,CAAC,IAAI,SAAS;QAC3B,GAAG,EAAE,SAAS,CAAC,CAAC,CAAC,aAAa,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG;KAClD,CAAC,CAAC,CAAC;AACN,CAAC;AAED;;GAEG;AACH,SAAS,KAAK,CACZ,QAAgC,EAChC,IAAY,EACZ,EAAU,EACV,GAAW,EACX,MAAc;IAEd,MAAM,SAAS,GAAG,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,IAAI,GAAG,EAAsB,CAAC;IACtE,MAAM,IAAI,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;IACjC,KAAK,MAAM,EAAE,IAAI,IAAI,EAAE,CAAC;QACtB,MAAM,MAAM,GAAG,SAAS,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,IAAI,GAAG,EAAkB,CAAC;QAC9D,MAAM,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,CAAC,GAAG,MAAM,CAAC,CAAC;QAC/C,SAAS,CAAC,GAAG,CAAC,EAAE,EAAE,MAAM,CAAC,CAAC;IAC5B,CAAC;IACD,QAAQ,CAAC,GAAG,CAAC,IAAI,EAAE,SAAS,CAAC,CAAC;AAChC,CAAC;AAED;;;;;;;;;GASG;AACH,SAAS,SAAS,CAChB,IAAY,EACZ,MAA8B;IAE9B,IAAI,MAAM,KAAK,SAAS,IAAI,MAAM,CAAC,IAAI,KAAK,CAAC,EAAE,CAAC;QAC9C,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,IAAI,IAAI,GAAG,EAAE,CAAC;IACd,IAAI,IAAI,GAAG,CAAC,CAAC;IACb,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,KAAK,MAAM,CAAC,SAAS,EAAE,KAAK,CAAC,IAAI,MAAM,EAAE,CAAC;QACxC,KAAK,IAAI,KAAK,CAAC;QACf,IACE,KAAK,GAAG,IAAI;YACZ,CAAC,KAAK,KAAK,IAAI,IAAI,IAAI,KAAK,IAAI,IAAI,SAAS,KAAK,IAAI,CAAC,EACvD,CAAC;YACD,IAAI,GAAG,SAAS,CAAC;YACjB,IAAI,GAAG,KAAK,CAAC;QACf,CAAC;IACH,CAAC;IACD,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,GAAG,KAAK,EAAE,CAAC;AACvC,CAAC;AAED;;;;;;GAMG;AACH,SAAS,YAAY,CACnB,IAAY,EACZ,SAAoB;IAEpB,MAAM,OAAO,GAAG,SAAS,CAAC,IAAI,EAAE,SAAS,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC;IACpD,IAAI,OAAO,KAAK,SAAS,EAAE,CAAC;QAC1B,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,MAAM,UAAU,GAAG,IAAI,GAAG,EAAkB,CAAC;IAC7C,KAAK,MAAM,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,SAAS,EAAE,CAAC;QACtC,MAAM,EAAE,GAAG,SAAS,CAAC,IAAI,EAAE,MAAM,CAAC,CAAC;QACnC,IACE,GAAG,KAAK,GAAG;YACX,EAAE,KAAK,SAAS;YAChB,EAAE,CAAC,IAAI,KAAK,OAAO,CAAC,IAAI;YACxB,EAAE,CAAC,KAAK,IAAI,iBAAiB,EAC7B,CAAC;YACD,UAAU,CAAC,GAAG,CAAC,GAAG,EAAE,EAAE,CAAC,IAAI,CAAC,CAAC;QAC/B,CAAC;IACH,CAAC;IAED,MAAM,IAAI,GAAG,QAAQ,CAAC,SAAS,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,OAAO,CAAC,IAAI,EAAE,UAAU,CAAC,CAAC;IAEpE,IAAI,OAAO,CAAC,IAAI,KAAK,IAAI,IAAI,UAAU,CAAC,IAAI,KAAK,CAAC,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACxE,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,OAAO;QACL,EAAE,EAAE,OAAO,CAAC,IAAI;QAChB,GAAG,CAAC,UAAU,CAAC,IAAI,GAAG,CAAC,IAAI,EAAE,SAAS,EAAE,UAAU,EAAE,CAAC;QACrD,GAAG,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,IAAI,EAAE,IAAI,EAAE,CAAC;KACjC,CAAC;AACJ,CAAC;AAED;;;;;;GAMG;AACH,SAAS,QAAQ,CACf,MAA8B,EAC9B,MAAc,EACd,UAAuC;IAEvC,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;QACzB,OAAO,EAAE,CAAC;IACZ,CAAC;IACD,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,KAAK,MAAM,KAAK,IAAI,MAAM,CAAC,MAAM,EAAE,EAAE,CAAC;QACpC,KAAK,IAAI,KAAK,CAAC;IACjB,CAAC;IACD,MAAM,OAAO,GAAG,IAAI,GAAG,CAAC,CAAC,MAAM,EAAE,GAAG,UAAU,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC;IAC1D,OAAO,CAAC,GAAG,MAAM,CAAC;SACf,MAAM,CACL,CAAC,CAAC,IAAI,EAAE,KAAK,CAAC,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,KAAK,GAAG,KAAK,IAAI,WAAW,CACtE;SACA,QAAQ,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,EAAE,CAAC,EAAE,KAAK,CAAC,EAAE,EAAE,CAAC,KAAK,GAAG,IAAI,CAAC;SAC/C,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,EAAE,CAAC,IAAI,CAAC,CAAC;AAC3B,CAAC;AAED;;GAEG;AACH,SAAS,OAAO,CACd,QAAgC;IAEhC,MAAM,KAAK,GAAG,IAAI,GAAG,EAA+B,CAAC;IACrD,KAAK,MAAM,CAAC,IAAI,EAAE,SAAS,CAAC,IAAI,QAAQ,EAAE,CAAC;QACzC,MAAM,UAAU,GAAG,YAAY,CAAC,IAAI,EAAE,SAAS,CAAC,CAAC;QACjD,IAAI,UAAU,KAAK,SAAS,EAAE,CAAC;YAC7B,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,UAAU,CAAC,CAAC;QAC9B,CAAC;IACH,CAAC;IACD,OAAO,KAAK,CAAC;AACf,CAAC;AAED;;;GAGG;AACH,SAAS,eAAe,CACtB,KAAsB;IAEtB,MAAM,UAAU,GAAG,YAAY,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,MAAM,CAAC;IACnD,OAAO,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,MAAM,KAAK,UAAU,CAAC,CAAC,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;AAC1E,CAAC;AAED;;;;;;;;;;;;GAYG;AACH,MAAM,UAAU,iBAAiB,CAC/B,OAAmC;IAEnC,MAAM,mBAAmB,GAAG,IAAI,GAAG,EAAqB,CAAC;IACzD,MAAM,kBAAkB,GAAG,IAAI,GAAG,EAAqB,CAAC;IAExD,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,KAAK,MAAM,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,EAAE,IAAI,UAAU,CAAC,KAAK,CAAC,EAAE,CAAC;YACpD,KAAK,CAAC,mBAAmB,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC,CAAC,CAAC;YAC/C,KAAK,CAAC,kBAAkB,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC,CAAC,CAAC;QAChD,CAAC;IACH,CAAC;IAED,MAAM,aAAa,GAAG,OAAO,CAAC,mBAAmB,CAAC,CAAC;IACnD,MAAM,YAAY,GAAG,OAAO,CAAC,kBAAkB,CAAC,CAAC;IAEjD,4EAA4E;IAC5E,+CAA+C;IAC/C,EAAE;IACF,4EAA4E;IAC5E,4EAA4E;IAC5E,8EAA8E;IAC9E,2EAA2E;IAC3E,6EAA6E;IAC7E,oEAAoE;IACpE,MAAM,gBAAgB,GAAG,IAAI,GAAG,EAAkB,CAAC;IACnD,MAAM,eAAe,GAAG,IAAI,GAAG,EAAkB,CAAC;IAClD,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,IAAI,YAAY,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACxC,SAAS;QACX,CAAC;QACD,MAAM,QAAQ,GAAG,eAAe,CAAC,KAAK,CAAC,CAAC;QACxC,IAAI,iBAAiB,CAAC,aAAa,EAAE,KAAK,CAAC,IAAI,EAAE,QAAQ,CAAC,KAAK,KAAK,CAAC,IAAI,EAAE,CAAC;YAC1E,gBAAgB,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,CAAC;QAC/C,CAAC;QACD,IAAI,iBAAiB,CAAC,YAAY,EAAE,KAAK,CAAC,IAAI,EAAE,QAAQ,CAAC,KAAK,KAAK,CAAC,IAAI,EAAE,CAAC;YACzE,eAAe,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,CAAC;QAC9C,CAAC;IACH,CAAC;IAED,OAAO;QACL,aAAa;QACb,YAAY;QACZ,gBAAgB;QAChB,eAAe;QACf,GAAG,oBAAoB,CAAC,OAAO,CAAC;KACjC,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AACH,SAAS,oBAAoB,CAAC,OAAmC;IAI/D,MAAM,MAAM,GAAG,IAAI,GAAG,EAAU,CAAC;IACjC,MAAM,MAAM,GAAG,IAAI,GAAG,EAAU,CAAC;IACjC,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,wEAAwE;QACxE,oEAAoE;QACpE,0EAA0E;QAC1E,uEAAuE;QACvE,2BAA2B;QAC3B,MAAM,aAAa,GAAG,iBAAiB,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QACpD,MAAM,KAAK,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,GAAG,CAAC,KAAK,CAAC,YAAY,IAAI,EAAE,CAAC,CAAC,CAAC,MAAM,CAC9D,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,KAAK,KAAK,CAAC,IAAI,IAAI,IAAI,KAAK,aAAa,CACxD,CAAC;QACF,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACvB,SAAS;QACX,CAAC;QACD,MAAM,cAAc,GAAG,YAAY,CAAC,aAAa,CAAC,CAAC;QACnD,KAAK,MAAM,SAAS,IAAI,cAAc,EAAE,CAAC;YACvC,MAAM,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QACxB,CAAC;QACD,MAAM,cAAc,GAAG,KAAK,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,EAAE,CAC5C,YAAY,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAC,CACtC,CAAC;QACF,KAAK,MAAM,SAAS,IAAI,cAAc,EAAE,CAAC;YACvC,MAAM,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QACxB,CAAC;IACH,CAAC;IAED,OAAO;QACL,QAAQ,EAAE,IAAI,GAAG,CAAC,CAAC,GAAG,MAAM,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;QAC5D,QAAQ,EAAE,IAAI,GAAG,CAAC,CAAC,GAAG,MAAM,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;KAC7D,CAAC;AACJ,CAAC"}
+{"version":3,"file":"script-tables.js","sourceRoot":"","sources":["../../src/dictionary/script-tables.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AACvD,OAAO,EAAE,iBAAiB,EAAE,MAAM,qBAAqB,CAAC;AACxD,OAAO,EAAE,iBAAiB,EAAqB,MAAM,yBAAyB,CAAC;AAE/E,OAAO,EACL,eAAe,EAEf,UAAU,EACV,OAAO,EACP,KAAK,GACN,MAAM,sBAAsB,CAAC;AAC9B;;;;;;;;;;;;GAYG;AACH,MAAM,UAAU,iBAAiB,CAC/B,OAAmC;IAEnC,MAAM,mBAAmB,GAAG,IAAI,GAAG,EAAqB,CAAC;IACzD,MAAM,kBAAkB,GAAG,IAAI,GAAG,EAAqB,CAAC;IAExD,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,KAAK,MAAM,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,EAAE,IAAI,UAAU,CAAC,KAAK,CAAC,EAAE,CAAC;YACpD,KAAK,CAAC,mBAAmB,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC,CAAC,CAAC;YAC/C,KAAK,CAAC,kBAAkB,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC,CAAC,CAAC;QAChD,CAAC;IACH,CAAC;IAED,MAAM,aAAa,GAAG,OAAO,CAAC,mBAAmB,CAAC,CAAC;IACnD,MAAM,YAAY,GAAG,OAAO,CAAC,kBAAkB,CAAC,CAAC;IAEjD,4EAA4E;IAC5E,+CAA+C;IAC/C,EAAE;IACF,4EAA4E;IAC5E,4EAA4E;IAC5E,8EAA8E;IAC9E,2EAA2E;IAC3E,6EAA6E;IAC7E,oEAAoE;IACpE,MAAM,gBAAgB,GAAG,IAAI,GAAG,EAAkB,CAAC;IACnD,MAAM,eAAe,GAAG,IAAI,GAAG,EAAkB,CAAC;IAClD,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,IAAI,YAAY,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACxC,SAAS;QACX,CAAC;QACD,MAAM,QAAQ,GAAG,eAAe,CAAC,KAAK,CAAC,CAAC;QACxC,IAAI,iBAAiB,CAAC,aAAa,EAAE,KAAK,CAAC,IAAI,EAAE,QAAQ,CAAC,KAAK,KAAK,CAAC,IAAI,EAAE,CAAC;YAC1E,gBAAgB,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,CAAC;QAC/C,CAAC;QACD,IAAI,iBAAiB,CAAC,YAAY,EAAE,KAAK,CAAC,IAAI,EAAE,QAAQ,CAAC,KAAK,KAAK,CAAC,IAAI,EAAE,CAAC;YACzE,eAAe,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,CAAC;QAC9C,CAAC;IACH,CAAC;IAED,OAAO;QACL,aAAa;QACb,YAAY;QACZ,gBAAgB;QAChB,eAAe;QACf,GAAG,oBAAoB,CAAC,OAAO,CAAC;KACjC,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AACH,SAAS,oBAAoB,CAAC,OAAmC;IAI/D,MAAM,MAAM,GAAG,IAAI,GAAG,EAAU,CAAC;IACjC,MAAM,MAAM,GAAG,IAAI,GAAG,EAAU,CAAC;IACjC,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,wEAAwE;QACxE,oEAAoE;QACpE,0EAA0E;QAC1E,uEAAuE;QACvE,2BAA2B;QAC3B,MAAM,aAAa,GAAG,iBAAiB,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QACpD,MAAM,KAAK,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,GAAG,CAAC,KAAK,CAAC,YAAY,IAAI,EAAE,CAAC,CAAC,CAAC,MAAM,CAC9D,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,KAAK,KAAK,CAAC,IAAI,IAAI,IAAI,KAAK,aAAa,CACxD,CAAC;QACF,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACvB,SAAS;QACX,CAAC;QACD,MAAM,cAAc,GAAG,YAAY,CAAC,aAAa,CAAC,CAAC;QACnD,KAAK,MAAM,SAAS,IAAI,cAAc,EAAE,CAAC;YACvC,MAAM,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QACxB,CAAC;QACD,MAAM,cAAc,GAAG,KAAK,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,EAAE,CAC5C,YAAY,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAC,CACtC,CAAC;QACF,KAAK,MAAM,SAAS,IAAI,cAAc,EAAE,CAAC;YACvC,MAAM,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QACxB,CAAC;IACH,CAAC;IAED,OAAO;QACL,QAAQ,EAAE,IAAI,GAAG,CAAC,CAAC,GAAG,MAAM,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;QAC5D,QAAQ,EAAE,IAAI,GAAG,CAAC,CAAC,GAAG,MAAM,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;KAC7D,CAAC;AACJ,CAAC"}

package/dist/dictionary/taiwan-reading.d.ts

@@ -0,0 +1,8 @@
+import type { CedictEntry } from "../sources/cedict.js";
+import type { UnihanReadings } from "../sources/unihan.js";
+import type { Syllable } from "../syllable/syllable.js";
+/**
+ * Settle it for one word, or undefined where no source states one.
+ */
+export declare function taiwanReadingOf(word: string, senses: readonly CedictEntry[], senseReadings: readonly (readonly Syllable[])[], reading: readonly Syllable[], unihan: UnihanReadings | undefined): readonly Syllable[] | undefined;
+//# sourceMappingURL=taiwan-reading.d.ts.map

package/dist/dictionary/taiwan-reading.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"taiwan-reading.d.ts","sourceRoot":"","sources":["../../src/dictionary/taiwan-reading.ts"],"names":[],"mappings":"AAUA,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAC;AACxD,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAC3D,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAMxD;;GAEG;AACH,wBAAgB,eAAe,CAC7B,IAAI,EAAE,MAAM,EACZ,MAAM,EAAE,SAAS,WAAW,EAAE,EAC9B,aAAa,EAAE,SAAS,CAAC,SAAS,QAAQ,EAAE,CAAC,EAAE,EAC/C,OAAO,EAAE,SAAS,QAAQ,EAAE,EAC5B,MAAM,EAAE,cAAc,GAAG,SAAS,GACjC,SAAS,QAAQ,EAAE,GAAG,SAAS,CAsBjC"}

package/dist/dictionary/taiwan-reading.js

@@ -0,0 +1,38 @@
+/**
+ * The 國語 reading an entry carries, where a source states one.
+ *
+ * Both sources hedge it. CC-CEDICT hangs `Taiwan pr.` on one sense of a
+ * headword and the others know nothing about it, and a note can sit on the
+ * right sense and still be about that sense rather than the character. What
+ * survives both tests, and is not simply another 普通话 sense of the word, is
+ * the delta.
+ */
+import { isSingleCharacter } from "../script/characters.js";
+import { isOwnSense, isSenseScopedNote } from "./cedict-senses.js";
+import { characterSyllable } from "./character-defaults.js";
+import { isSameReading } from "./entry.js";
+import { readDictionaryReading } from "./reading.js";
+/**
+ * Settle it for one word, or undefined where no source states one.
+ */
+export function taiwanReadingOf(word, senses, senseReadings, reading, unihan) {
+    const taiwanSense = senses.find((entry) => entry.taiwanReadings !== undefined);
+    const taiwanTokens = isSenseScopedNote(word, taiwanSense)
+        ? undefined
+        : taiwanSense?.taiwanReadings;
+    const unihanTaiwan = unihan?.taiwanReading;
+    let taiwan;
+    if (taiwanTokens !== undefined) {
+        taiwan = readDictionaryReading(word, taiwanTokens);
+    }
+    else if (isSingleCharacter(word) && unihanTaiwan !== undefined) {
+        const syllable = characterSyllable(word, unihanTaiwan);
+        taiwan = syllable === undefined ? undefined : [syllable];
+    }
+    if (taiwan !== undefined &&
+        (isSameReading(taiwan, reading) || isOwnSense(taiwan, senseReadings))) {
+        return undefined;
+    }
+    return taiwan;
+}
+//# sourceMappingURL=taiwan-reading.js.map

package/dist/dictionary/taiwan-reading.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"taiwan-reading.js","sourceRoot":"","sources":["../../src/dictionary/taiwan-reading.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AACH,OAAO,EAAE,iBAAiB,EAAE,MAAM,yBAAyB,CAAC;AAI5D,OAAO,EAAE,UAAU,EAAE,iBAAiB,EAAE,MAAM,oBAAoB,CAAC;AACnE,OAAO,EAAE,iBAAiB,EAAE,MAAM,yBAAyB,CAAC;AAC5D,OAAO,EAAE,aAAa,EAAE,MAAM,YAAY,CAAC;AAC3C,OAAO,EAAE,qBAAqB,EAAE,MAAM,cAAc,CAAC;AAErD;;GAEG;AACH,MAAM,UAAU,eAAe,CAC7B,IAAY,EACZ,MAA8B,EAC9B,aAA+C,EAC/C,OAA4B,EAC5B,MAAkC;IAElC,MAAM,WAAW,GAAG,MAAM,CAAC,IAAI,CAC7B,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,cAAc,KAAK,SAAS,CAC9C,CAAC;IACF,MAAM,YAAY,GAAG,iBAAiB,CAAC,IAAI,EAAE,WAAW,CAAC;QACvD,CAAC,CAAC,SAAS;QACX,CAAC,CAAC,WAAW,EAAE,cAAc,CAAC;IAChC,MAAM,YAAY,GAAG,MAAM,EAAE,aAAa,CAAC;IAC3C,IAAI,MAAuC,CAAC;IAC5C,IAAI,YAAY,KAAK,SAAS,EAAE,CAAC;QAC/B,MAAM,GAAG,qBAAqB,CAAC,IAAI,EAAE,YAAY,CAAC,CAAC;IACrD,CAAC;SAAM,IAAI,iBAAiB,CAAC,IAAI,CAAC,IAAI,YAAY,KAAK,SAAS,EAAE,CAAC;QACjE,MAAM,QAAQ,GAAG,iBAAiB,CAAC,IAAI,EAAE,YAAY,CAAC,CAAC;QACvD,MAAM,GAAG,QAAQ,KAAK,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC;IAC3D,CAAC;IACD,IACE,MAAM,KAAK,SAAS;QACpB,CAAC,aAAa,CAAC,MAAM,EAAE,OAAO,CAAC,IAAI,UAAU,CAAC,MAAM,EAAE,aAAa,CAAC,CAAC,EACrE,CAAC;QACD,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC"}

package/dist/dictionary/tone-invariants.d.ts

@@ -0,0 +1,3 @@
+import type { BuildAssertion } from "./built-dictionary.js";
+export declare const TONE_INVARIANTS: readonly BuildAssertion[];
+//# sourceMappingURL=tone-invariants.d.ts.map

package/dist/dictionary/tone-invariants.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"tone-invariants.d.ts","sourceRoot":"","sources":["../../src/dictionary/tone-invariants.ts"],"names":[],"mappings":"AASA,OAAO,KAAK,EAAE,cAAc,EAAmB,MAAM,uBAAuB,CAAC;AAkB7E,eAAO,MAAM,eAAe,EAAE,SAAS,cAAc,EAsDpD,CAAC"}

package/dist/dictionary/tone-invariants.js

@@ -0,0 +1,64 @@
+/**
+ * What must hold of the neutral tone across the dictionary.
+ *
+ * The 轻声 list and the 语气词 are the two places a source refresh could take a
+ * neutral tone away without any single word looking wrong.
+ */
+import { characterCount } from "../script/characters.js";
+import { NEUTRAL_TONE } from "../tone/tone.js";
+import { NEUTRAL_SENSE_WORDS } from "./neutral-senses.js";
+/**
+ * jieba's tag for a 语气词, which is what 吧, 呢 and 吗 are.
+ */
+const PARTICLE_TAG = "y";
+/**
+ * The 语气词 the sources themselves do not lead with a 轻声 for.
+ *
+ * The rule below is that the words a character appears in must not take the
+ * default away from its 轻声, not that a 语气词 is always neutral. For these
+ * three the reading that leads was never the neutral one and no word put it
+ * there: `kMandarin` names 呃 `è` and 哩 `lī`, and `kHanyuPinlu` ranks 呵 as
+ * `ā(392)` over `hē(64)`, all of them counting the bare character.
+ */
+const FULL_TONE_PARTICLES = new Set(["呃", "呵", "哩"]);
+export const TONE_INVARIANTS = [
+    // kHanyuPinlu writes 西 as `xi(902) xī(738)`, and the 902 are 东西: a 轻声 the
+    // field counted only inside words is not the bare character's reading. 吗 is
+    // the control — CC-CEDICT calls it a question particle in its own right, and
+    // a particle's whole use *is* the bare character.
+    {
+        // The table names words and lets CC-CEDICT supply the reading, so the way
+        // it can fail silently is a source refresh dropping or retoning the 轻声
+        // sense — after which the word would quietly go back to the full-tone
+        // homograph the list exists to get away from.
+        description: "every word on the 轻声 sense list reads with one",
+        check: (dictionary) => {
+            const wrong = NEUTRAL_SENSE_WORDS.filter((listed) => !(dictionary.get(listed.word)?.readings.cn ?? []).some((syllable) => syllable.tone === NEUTRAL_TONE)).map((listed) => `${listed.word} ${dictionary.reading(listed.word) ?? "(missing)"}`);
+            return wrong.length === 0
+                ? undefined
+                : `listed for their 轻声 sense and not reading one: ${wrong.join(", ")}`;
+        },
+    },
+    {
+        // The corpus mass ranking the character defaults counts a character only
+        // inside the words the dictionary holds, and a 语气词 is never inside one:
+        // every 吧 it can reach is 酒吧, 网吧 or 吧台, and the only 呗 is 梵呗.
+        // Left to it the particle takes the reading of the noun, which is what 吧
+        // `bā` and 呗 `bài` were. A 语气词 is the one part of speech whose whole
+        // use is the bare character, so its 轻声 is what words cannot vote on.
+        description: "no 语气词 ranks a full tone above its own 轻声",
+        check: (dictionary) => {
+            const wrong = dictionary.entries
+                .filter((entry) => entry.partOfSpeech === PARTICLE_TAG &&
+                characterCount(entry.hans) === 1 &&
+                !FULL_TONE_PARTICLES.has(entry.hans) &&
+                entry.readings.cn[0]?.tone !== NEUTRAL_TONE &&
+                (entry.alternates ?? []).some((reading) => reading[0]?.tone === NEUTRAL_TONE))
+                .map((entry) => `${entry.hans} ${dictionary.reading(entry.hans) ?? ""}`);
+            return wrong.length === 0
+                ? undefined
+                : `语气词 reading a full tone over a 轻声 they also have: ${wrong.join(", ")}`;
+        },
+    },
+];
+//# sourceMappingURL=tone-invariants.js.map

package/dist/dictionary/tone-invariants.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"tone-invariants.js","sourceRoot":"","sources":["../../src/dictionary/tone-invariants.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,EAAE,cAAc,EAAE,MAAM,yBAAyB,CAAC;AACzD,OAAO,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAC/C,OAAO,EAAE,mBAAmB,EAAE,MAAM,qBAAqB,CAAC;AAG1D;;GAEG;AACH,MAAM,YAAY,GAAG,GAAG,CAAC;AAEzB;;;;;;;;GAQG;AACH,MAAM,mBAAmB,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;AAErD,MAAM,CAAC,MAAM,eAAe,GAA8B;IACxD,0EAA0E;IAC1E,4EAA4E;IAC5E,6EAA6E;IAC7E,kDAAkD;IAClD;QACE,0EAA0E;QAC1E,uEAAuE;QACvE,sEAAsE;QACtE,8CAA8C;QAC9C,WAAW,EAAE,gDAAgD;QAC7D,KAAK,EAAE,CAAC,UAA2B,EAAsB,EAAE;YACzD,MAAM,KAAK,GAAG,mBAAmB,CAAC,MAAM,CACtC,CAAC,MAAM,EAAE,EAAE,CACT,CAAC,CAAC,UAAU,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,QAAQ,CAAC,EAAE,IAAI,EAAE,CAAC,CAAC,IAAI,CACpD,CAAC,QAAQ,EAAE,EAAE,CAAC,QAAQ,CAAC,IAAI,KAAK,YAAY,CAC7C,CACJ,CAAC,GAAG,CACH,CAAC,MAAM,EAAE,EAAE,CACT,GAAG,MAAM,CAAC,IAAI,IAAI,UAAU,CAAC,OAAO,CAAC,MAAM,CAAC,IAAI,CAAC,IAAI,WAAW,EAAE,CACrE,CAAC;YACF,OAAO,KAAK,CAAC,MAAM,KAAK,CAAC;gBACvB,CAAC,CAAC,SAAS;gBACX,CAAC,CAAC,kDAAkD,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;QAC3E,CAAC;KACF;IACD;QACE,yEAAyE;QACzE,wEAAwE;QACxE,8DAA8D;QAC9D,yEAAyE;QACzE,qEAAqE;QACrE,qEAAqE;QACrE,WAAW,EAAE,2CAA2C;QACxD,KAAK,EAAE,CAAC,UAA2B,EAAsB,EAAE;YACzD,MAAM,KAAK,GAAG,UAAU,CAAC,OAAO;iBAC7B,MAAM,CACL,CAAC,KAAK,EAAE,EAAE,CACR,KAAK,CAAC,YAAY,KAAK,YAAY;gBACnC,cAAc,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC;gBAChC,CAAC,mBAAmB,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC;gBACpC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,IAAI,KAAK,YAAY;gBAC3C,CAAC,KAAK,CAAC,UAAU,IAAI,EAAE,CAAC,CAAC,IAAI,CAC3B,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,IAAI,KAAK,YAAY,CAC/C,CACJ;iBACA,GAAG,CACF,CAAC,KAAK,EAAE,EAAE,CAAC,GAAG,KAAK,CAAC,IAAI,IAAI,UAAU,CAAC,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,EAAE,EAAE,CACnE,CAAC;YACJ,OAAO,KAAK,CAAC,MAAM,KAAK,CAAC;gBACvB,CAAC,CAAC,SAAS;gBACX,CAAC,CAAC,qDAAqD,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;QAC9E,CAAC;KACF;CACF,CAAC"}

package/dist/dictionary/traditional-form.d.ts

@@ -0,0 +1,21 @@
+import type { CedictEntry } from "../sources/cedict.js";
+import type { Syllable } from "../syllable/syllable.js";
+import type { ReadCharacters } from "./reading.js";
+import type { TraditionalTable } from "./traditional.js";
+/**
+ * The 繁體 form of one entry, and what the merge counts about it.
+ */
+export interface TraditionalForm {
+    /** The CC-CEDICT senses matching the reading, most useful first. */
+    readonly senses: readonly CedictEntry[];
+    readonly hant: string;
+    /** Other 繁體 spellings a source writes of the same word. */
+    readonly hantVariants: readonly string[];
+    /** Whether `hant` was derived rather than taken from CC-CEDICT. */
+    readonly isDerived: boolean;
+}
+/**
+ * Settle all of it for one word.
+ */
+export declare function traditionalFormOf(word: string, cedictEntries: readonly CedictEntry[], reading: readonly Syllable[], aligned: readonly ReadCharacters[] | undefined, traditional: TraditionalTable): TraditionalForm;
+//# sourceMappingURL=traditional-form.d.ts.map

package/dist/dictionary/traditional-form.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"traditional-form.d.ts","sourceRoot":"","sources":["../../src/dictionary/traditional-form.ts"],"names":[],"mappings":"AASA,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAC;AACxD,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAExD,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,cAAc,CAAC;AACnD,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,kBAAkB,CAAC;AAEzD;;GAEG;AACH,MAAM,WAAW,eAAe;IAC9B,oEAAoE;IACpE,QAAQ,CAAC,MAAM,EAAE,SAAS,WAAW,EAAE,CAAC;IACxC,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,2DAA2D;IAC3D,QAAQ,CAAC,YAAY,EAAE,SAAS,MAAM,EAAE,CAAC;IACzC,mEAAmE;IACnE,QAAQ,CAAC,SAAS,EAAE,OAAO,CAAC;CAC7B;AAED;;GAEG;AACH,wBAAgB,iBAAiB,CAC/B,IAAI,EAAE,MAAM,EACZ,aAAa,EAAE,SAAS,WAAW,EAAE,EACrC,OAAO,EAAE,SAAS,QAAQ,EAAE,EAC5B,OAAO,EAAE,SAAS,cAAc,EAAE,GAAG,SAAS,EAC9C,WAAW,EAAE,gBAAgB,GAC5B,eAAe,CA6CjB"}

package/dist/dictionary/traditional-form.js

@@ -0,0 +1,59 @@
+/**
+ * Which 繁體 spelling an entry takes, and which others reach it.
+ *
+ * CC-CEDICT decides wherever it has the word, since it pairs the two scripts
+ * itself; the derivation only answers where it is silent. Both are settled
+ * against the sense matching the reading, because a word read two ways can be
+ * written two ways — 万 is 萬 when read wàn and stays 万 when read mò.
+ */
+import { toCanonicalGlyphs } from "../script/glyphs.js";
+import { sensesForReading } from "./reading-agreement.js";
+/**
+ * Settle all of it for one word.
+ */
+export function traditionalFormOf(word, cedictEntries, reading, aligned, traditional) {
+    let isDerived = false;
+    // Which sense matters: 万 is 萬 when read wàn but stays 万 when read mò, and
+    // CC-CEDICT carries both as separate entries. Taking whichever came first
+    // in the file would pair the chosen reading with another sense's script.
+    const senses = sensesForReading(word, cedictEntries, reading);
+    const sense = senses[0];
+    let hant;
+    if (sense === undefined) {
+        // Canonicalised because the derivation settles each word against Unihan's
+        // variant lists with no single standard behind it, so 里 derives as 裏
+        // where the corpus overwhelmingly writes 裡. Both are the same character
+        // with the same reading, and a key written in the form the lookup path
+        // normalises *away from* can never be found — see the note below.
+        hant = toCanonicalGlyphs(traditional.convert(aligned ?? []));
+        isDerived = hant !== word;
+    }
+    else {
+        hant = toCanonicalGlyphs(sense.traditional);
+    }
+    // Other spellings of the same word, so that a 繁體 reader of either finds
+    // it. Only what a source writes out: composing spellings from per-character
+    // variants instead would add 229,482 keys to the full tier, almost all of
+    // them forms nobody writes — 方麵 for 方面, 公裡 for 公里 — because the
+    // reading that picks the right variant for one word does not generalise to
+    // every word the character appears in.
+    //
+    // Canonicalised for the same reason `hant` is, and deduplicated after:
+    // 裏面 and 裡面 are one spelling once the glyph forms are folded, and
+    // keying both would claim a variant that does not exist.
+    //
+    // The 简体 headword's own canonical spelling joins them where it differs,
+    // and that is not a nicety: the lookup path normalises 繁體 glyph forms
+    // before it searches, so a key written in a form it normalises *away from*
+    // can never be found. The phrase corpus carries a few hundred headwords
+    // spelled with 峯, 藴 or 枱 — 鹫峯寺, 义藴, 写字枱 — and without this they
+    // are entries nothing can reach.
+    const hantVariants = [
+        ...new Set([
+            ...senses.map((other) => toCanonicalGlyphs(other.traditional)),
+            toCanonicalGlyphs(word),
+        ]),
+    ].filter((form) => form !== hant && form !== word);
+    return { senses, hant, hantVariants, isDerived };
+}
+//# sourceMappingURL=traditional-form.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"traditional-form.js","sourceRoot":"","sources":["../../src/dictionary/traditional-form.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AACH,OAAO,EAAE,iBAAiB,EAAE,MAAM,qBAAqB,CAAC;AAGxD,OAAO,EAAE,gBAAgB,EAAE,MAAM,wBAAwB,CAAC;AAiB1D;;GAEG;AACH,MAAM,UAAU,iBAAiB,CAC/B,IAAY,EACZ,aAAqC,EACrC,OAA4B,EAC5B,OAA8C,EAC9C,WAA6B;IAE7B,IAAI,SAAS,GAAG,KAAK,CAAC;IACtB,0EAA0E;IAC1E,0EAA0E;IAC1E,yEAAyE;IACzE,MAAM,MAAM,GAAG,gBAAgB,CAAC,IAAI,EAAE,aAAa,EAAE,OAAO,CAAC,CAAC;IAC9D,MAAM,KAAK,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC;IACxB,IAAI,IAAY,CAAC;IACjB,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;QACxB,0EAA0E;QAC1E,qEAAqE;QACrE,wEAAwE;QACxE,uEAAuE;QACvE,kEAAkE;QAClE,IAAI,GAAG,iBAAiB,CAAC,WAAW,CAAC,OAAO,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,CAAC;QAC7D,SAAS,GAAG,IAAI,KAAK,IAAI,CAAC;IAC5B,CAAC;SAAM,CAAC;QACN,IAAI,GAAG,iBAAiB,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC;IAC9C,CAAC;IAED,wEAAwE;IACxE,4EAA4E;IAC5E,0EAA0E;IAC1E,gEAAgE;IAChE,2EAA2E;IAC3E,uCAAuC;IACvC,EAAE;IACF,uEAAuE;IACvE,kEAAkE;IAClE,yDAAyD;IACzD,EAAE;IACF,wEAAwE;IACxE,sEAAsE;IACtE,2EAA2E;IAC3E,wEAAwE;IACxE,gEAAgE;IAChE,iCAAiC;IACjC,MAAM,YAAY,GAAG;QACnB,GAAG,IAAI,GAAG,CAAC;YACT,GAAG,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,iBAAiB,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC;YAC9D,iBAAiB,CAAC,IAAI,CAAC;SACxB,CAAC;KACH,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,KAAK,IAAI,IAAI,IAAI,KAAK,IAAI,CAAC,CAAC;IAEnD,OAAO,EAAE,MAAM,EAAE,IAAI,EAAE,YAAY,EAAE,SAAS,EAAE,CAAC;AACnD,CAAC"}

package/dist/dictionary/traditional-ranking.d.ts

@@ -0,0 +1,22 @@
+/**
+ * Choosing between the 繁體 variants Unihan lists, by how a character is read.
+ *
+ * The second of the two sources of evidence `TraditionalTable` weighs, and the
+ * only one that is inference rather than observation. Kept apart because it is
+ * a different question from the pairings above it: not "how has this character
+ * been converted before" but "which of these candidates is read this way".
+ */
+import type { UnihanReadings } from "../sources/unihan.js";
+import { type Syllable } from "../syllable/syllable.js";
+/**
+ * Unihan's reading lists, ranked against a syllable.
+ */
+export declare class VariantRanking {
+    #private;
+    constructor(unihanReadings: ReadonlyMap<string, UnihanReadings>);
+    /**
+     * The best-ranked candidate, or the character itself where there are none.
+     */
+    best(candidates: readonly string[], hans: string, syllable: Syllable | undefined): string;
+}
+//# sourceMappingURL=traditional-ranking.d.ts.map

package/dist/dictionary/traditional-ranking.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"traditional-ranking.d.ts","sourceRoot":"","sources":["../../src/dictionary/traditional-ranking.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AACH,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAC3D,OAAO,EAAgB,KAAK,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAEtE;;GAEG;AACH,qBAAa,cAAc;;IAIzB,YAAY,cAAc,EAAE,WAAW,CAAC,MAAM,EAAE,cAAc,CAAC,EAE9D;IAmDD;;OAEG;IACH,IAAI,CACF,UAAU,EAAE,SAAS,MAAM,EAAE,EAC7B,IAAI,EAAE,MAAM,EACZ,QAAQ,EAAE,QAAQ,GAAG,SAAS,GAC7B,MAAM,CAaR;CACF"}