@kensio/pinyinjs 1.7.4 → 1.7.5

package/dist/dictionary/assertions.js

@@ -1,320 +1,7 @@
-import { characterCount } from "../script/characters.js";
-import { DICTIONARY_SYLLABLES, isAttestedTone } from "../syllable/inventory.js";
-import { writeSyllable } from "../syllable/syllable.js";
-import { NEUTRAL_TONE } from "../tone/tone.js";
-import { traditionalForms } from "./artifact.js";
-import { NEUTRAL_SENSE_WORDS } from "./neutral-senses.js";
-/**
- * A merged dictionary, indexed for the assertions to interrogate.
- */
-export class BuiltDictionary {
-    #byKey;
-    entries;
-    /**
-     * Index entries by every key they claim, headwords first.
-     *
-     * The order matters for the same reason it does in the artifact: 发 and 髮
-     * both claim 发, and the entry the key names has to win.
-     */
-    constructor(entries) {
-        const byKey = new Map();
-        for (const entry of entries) {
-            byKey.set(entry.hans, entry);
-        }
-        for (const entry of entries) {
-            for (const form of traditionalForms(entry)) {
-                if (!byKey.has(form)) {
-                    byKey.set(form, entry);
-                }
-            }
-        }
-        this.#byKey = byKey;
-        this.entries = entries;
-    }
-    /**
-     * Every reading the dictionary holds, under every locale it holds one for.
-     */
-    *#readings() {
-        for (const entry of this.entries) {
-            yield* [
-                entry.readings.cn,
-                entry.readings.tw ?? [],
-                ...(entry.alternates ?? []),
-            ];
-        }
-    }
-    /**
-     * The entry for a word under either script, or undefined.
-     */
-    get(word) {
-        return this.#byKey.get(word);
-    }
-    /**
-     * A word's zh-CN reading in tone-marked notation, or undefined.
-     */
-    reading(word) {
-        const entry = this.#byKey.get(word);
-        return entry === undefined
-            ? undefined
-            : entry.readings.cn.map((syllable) => writeSyllable(syllable)).join(" ");
-    }
-    /**
-     * Every distinct toneless syllable the dictionary uses.
-     */
-    syllableInventory() {
-        const inventory = new Set();
-        for (const reading of this.#readings()) {
-            for (const syllable of reading) {
-                inventory.add(writeSyllable({ ...syllable, erhua: false, tone: undefined }));
-            }
-        }
-        return inventory;
-    }
-    /**
-     * Every syllable the dictionary uses that is not in the tone the inventory
-     * says that syllable is written in.
-     *
-     * The tones are read off the dictionary in the first place, so this is empty
-     * by construction — until a source refresh brings in a reading nobody has
-     * seen, which is exactly when a reader would start handing that syllable
-     * back and the table would have to be regenerated.
-     */
-    unattestedTones() {
-        const unattested = new Set();
-        for (const reading of this.#readings()) {
-            for (const syllable of reading) {
-                if (!isAttestedTone(syllable)) {
-                    unattested.add(writeSyllable({ ...syllable, erhua: false }));
-                }
-            }
-        }
-        return unattested;
-    }
-}
-/**
- * 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"]);
-/**
- * Assert one word reads a particular way.
- */
-function reads(word, expected, why) {
-    return {
-        description: `${word} → ${expected} (${why})`,
-        check: (dictionary) => {
-            const actual = dictionary.reading(word);
-            if (actual === undefined) {
-                return `${word} is missing from the dictionary`;
-            }
-            return actual === expected
-                ? undefined
-                : `${word} reads ${actual}, expected ${expected}`;
-        },
-    };
-}
-/**
- * What must hold of every build, or no artifact is produced.
- *
- * These test the **stored** form, which is deliberately not what a user sees:
- * sandhi is normalised out of the dictionary and reapplied at runtime, so
- * 一丁不识 is stored as `yī dīng bù shí` and only becomes `yì dīng bù shí` after
- * the runtime pass. The gold corpus asserts that output form. Asserting `yì`
- * here would be permanently unsatisfiable — see MERGE.md.
- *
- * Each one stands for a whole class of upstream defect that a source refresh
- * could reintroduce, so a failure here means the merge no longer repairs
- * something it used to, not merely that one word is wrong.
- */
-export const BUILD_ASSERTIONS = [
-    reads("玩儿", "wánr", "儿化 repaired from CC-CEDICT's r5"),
-    reads("女儿", "nǚ ér", "儿 keeps its own syllable here"),
-    reads("这儿", "zhèr", "儿化 despite upstream writing a tone on the er"),
-    reads("儿子", "ér zi", "儿化 exception list consulted"),
-    reads("一丁不识", "yī dīng bù shí", "一 sandhi normalised out"),
-    reads("一不小心", "yī bù xiǎo xīn", "一 and 不 sandhi normalised out"),
-    reads("大夫", "dài fu", "override applied over both sources"),
-    reads("东西", "dōng xi", "the thing, not the compass directions"),
-    reads("东西方", "dōng xī fāng", "and the directions where that is the word"),
-    reads("告诉", "gào su", "the everyday verb, not filing a complaint"),
-    reads("银行", "yín háng", "polyphone survives"),
-    reads("行长", "háng zhǎng", "polyphone survives"),
-    reads("头发", "tóu fa", "CC-CEDICT wins on the neutral tone"),
-    reads("还是", "hái shi", "CC-CEDICT wins on the neutral tone"),
-    reads("頭髮", "tóu fa", "繁體 derived using the reading, and keyed directly"),
-    reads("重複", "chóng fù", "one of two 繁體 spellings, both keyed"),
-    reads("重覆", "chóng fù", "the other, on the same entry"),
-    reads("下麵", "xià miàn", "both spellings of a word CC-CEDICT writes twice"),
-    // The 简体 phrase corpus carries a few 繁體 headwords and reads them as
-    // though the characters were 简体 — 徵 as `zhǐ` rather than as 征, 沈 as the
-    // surname rather than as 沉. Left as entries of their own they outrank the
-    // 繁體 key derived from the 简体 entry, and the word reads one way in a tier
-    // holding the phrase tail and another way in a tier without it.
-    reads("特徵", "tè zhēng", "not tè zhǐ, which is 徵 read as itself"),
-    reads("沈溺", "chén nì", "not shěn nì, which is 沈 read as the surname"),
-    reads("蝨子", "shī zi", "not shī zǐ, which is 子 read with its full tone"),
-    // kHanyuPinlu writes 李 as `li(36)`, with no tone mark. Read as 轻声 — which
-    // is what an unmarked reading means everywhere else in source data — 李华
-    // comes out `Li Huá`. The other Unihan fields all write `lǐ`, and that is
-    // what settles it. 们 is the control: genuinely neutral, and written bare by
-    // the other fields too, so it must not be "corrected".
-    reads("李", "lǐ", "a tone kHanyuPinlu left off is restored"),
-    reads("们", "men", "a genuine 轻声 is not given a tone it never had"),
-    reads("吧", "ba", "the 语气词, not the 酒吧 the corpus mass counts"),
-    reads("酒吧", "jiǔ bā", "and the word the character keeps its full tone in"),
-    // 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: ${wrong.join(", ")}`;
-        },
-    },
-    reads("西", "xī", "a 轻声 counted only inside words does not lead"),
-    reads("子", "zǐ", "and a suffix sense is not a claim about the character"),
-    reads("夫", "fū", "nor is 大夫's and 丈夫's reduction"),
-    reads("吗", "ma", "but a 语气词 CC-CEDICT reads 轻声 alone keeps it"),
-    reads("得", "de", "and so does the structural particle"),
-    {
-        // The corpus mass ranking the character defaults counts a character only
-        // inside the words the dictionary holds, and a 语气词 is never inside one:
-        // every 吧 it can reach is 酒吧, 网吧 or 吧台, and the only 呗 is 梵呗.
-        // Left to it the particle takes the reading of the noun, which is what 吧
-        // `bā` and 呗 `bài` were. A 语气词 is the one part of speech whose whole
-        // use is the bare character, so its 轻声 is what words cannot vote on.
-        description: "no 语气词 ranks a full tone above its own 轻声",
-        check: (dictionary) => {
-            const wrong = dictionary.entries
-                .filter((entry) => entry.partOfSpeech === PARTICLE_TAG &&
-                characterCount(entry.hans) === 1 &&
-                !FULL_TONE_PARTICLES.has(entry.hans) &&
-                entry.readings.cn[0]?.tone !== NEUTRAL_TONE &&
-                (entry.alternates ?? []).some((reading) => reading[0]?.tone === NEUTRAL_TONE))
-                .map((entry) => `${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(", ")}`;
-        },
-    },
-    {
-        description: "北京 is a proper noun (jieba POS carried through)",
-        check: (dictionary) => dictionary.get("北京")?.isProperNoun === true
-            ? undefined
-            : "北京 is not marked as a proper noun",
-    },
-    {
-        // jieba tags all four of these a proper noun; CC-CEDICT writes their
-        // pinyin in lower case, and the decoder capitalises straight off this bit.
-        description: "沙发, 城市, 阿姨 and 长大 are not proper nouns",
-        check: (dictionary) => {
-            const wrong = ["沙发", "城市", "阿姨", "长大"].filter((word) => dictionary.get(word)?.isProperNoun === true);
-            return wrong.length === 0
-                ? undefined
-                : `marked as proper nouns: ${wrong.join(", ")}`;
-        },
-    },
-    {
-        // The veto only demotes. These stay proper nouns, and 人民政府 is the
-        // control that shows the veto still fires on an institution.
-        description: "齐白石, 国务院 and 湖北 stay proper nouns",
-        check: (dictionary) => {
-            const wrong = ["齐白石", "国务院", "湖北"].filter((word) => dictionary.get(word)?.isProperNoun !== true);
-            return wrong.length === 0
-                ? undefined
-                : `no longer proper nouns: ${wrong.join(", ")}`;
-        },
-    },
-    {
-        // The gap composeLocaleDeltas closes, and the two ways it must not close
-        // it. 運行狀況 contains the marked word 行狀 and cuts elsewhere; 一个个地
-        // ends in a character whose delta is a different sense of 地 entirely.
-        description: "垃圾分類 inherits 垃圾's 國語 reading, 運行狀況 inherits nothing",
-        check: (dictionary) => {
-            const taiwan = (word) => dictionary
-                .get(word)
-                ?.readings.tw?.map((syllable) => writeSyllable(syllable))
-                .join(" ");
-            const composed = taiwan("垃圾分类");
-            if (composed !== "lè sè fēn lèi") {
-                return `垃圾分类 reads ${composed ?? "nothing"} in zh-TW, expected lè sè fēn lèi`;
-            }
-            const wrong = ["运行状况", "一个个地", "相亲相爱"].filter((word) => taiwan(word) !== undefined);
-            return wrong.length === 0
-                ? undefined
-                : `composed a zh-TW reading for: ${wrong.join(", ")}`;
-        },
-    },
-    {
-        description: "头发 derives 頭髮 rather than 頭發",
-        check: (dictionary) => {
-            const hant = dictionary.get("头发")?.hant;
-            return hant === "頭髮"
-                ? undefined
-                : `头发 derived ${hant ?? "nothing"}, expected 頭髮`;
-        },
-    },
-    {
-        // Membership rather than a count. MERGE.md asked for "exactly 408 after the
-        // merge", which is the phrase corpus's own inventory — but the merged
-        // dictionary also carries Unihan's and CC-CEDICT's rare characters, whose
-        // readings are legitimately outside it (鞥 ēng, 覅 fiào, 挼 ruá). Fixing the
-        // count would either exclude those or bless whatever a refresh adds; naming
-        // the permitted set does neither.
-        description: "every syllable used is one the inventory knows",
-        check: (dictionary) => {
-            const unknown = [...dictionary.syllableInventory()].filter((syllable) => !DICTIONARY_SYLLABLES.has(syllable));
-            return unknown.length === 0
-                ? undefined
-                : `readings use ${String(unknown.length)} syllable(s) outside the inventory: ${unknown.join(", ")}`;
-        },
-    },
-    {
-        // The romanisation readers narrow an ambiguous spelling on the tone that
-        // was written — `lo²` is 羅 luó and not a 咯 that is only ever neutral —
-        // so a tone the table has never heard of is a syllable those readers will
-        // refuse to hand back. Regenerate `SYLLABLE_TONES` when this fails.
-        description: "every syllable is used in a tone the inventory knows",
-        check: (dictionary) => {
-            const unattested = [...dictionary.unattestedTones()];
-            return unattested.length === 0
-                ? undefined
-                : `readings use ${String(unattested.length)} syllable(s) in a tone the inventory does not list: ${unattested.join(", ")}`;
-        },
-    },
-    {
-        description: "every entry's syllable count matches its character count",
-        check: (dictionary) => {
-            for (const entry of dictionary.entries) {
-                const characters = characterCount(entry.hans);
-                const erhua = entry.readings.cn.filter((syllable) => syllable.erhua === true).length;
-                // Punctuation is written but unread, so it is allowed to have no
-                // syllable; a reading may therefore be shorter but never longer.
-                if (entry.readings.cn.length + erhua > characters) {
-                    return `${entry.hans} has ${String(entry.readings.cn.length)} syllables for ${String(characters)} characters`;
-                }
-            }
-            return;
-        },
-    },
-];
+import { BuiltDictionary } from "./built-dictionary.js";
+import { BUILD_ASSERTIONS } from "./build-assertions.js";
+export { BuiltDictionary } from "./built-dictionary.js";
+export { BUILD_ASSERTIONS } from "./build-assertions.js";
 /**
  * Run every build assertion, returning the failures.
  */

package/dist/dictionary/assertions.js.map

@@ -1 +1 @@
-{"version":3,"file":"assertions.js","sourceRoot":"","sources":["../../src/dictionary/assertions.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,cAAc,EAAE,MAAM,yBAAyB,CAAC;AACzD,OAAO,EAAE,oBAAoB,EAAE,cAAc,EAAE,MAAM,0BAA0B,CAAC;AAChF,OAAO,EAAiB,aAAa,EAAE,MAAM,yBAAyB,CAAC;AACvE,OAAO,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAC/C,OAAO,EAAE,gBAAgB,EAAE,MAAM,eAAe,CAAC;AAEjD,OAAO,EAAE,mBAAmB,EAAE,MAAM,qBAAqB,CAAC;AAW1D;;GAEG;AACH,MAAM,OAAO,eAAe;IACjB,MAAM,CAAuC;IAE7C,OAAO,CAA6B;IAE7C;;;;;OAKG;IACH,YAAY,OAAmC;QAC7C,MAAM,KAAK,GAAG,IAAI,GAAG,EAA2B,CAAC;QACjD,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;YAC5B,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;QAC/B,CAAC;QACD,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;YAC5B,KAAK,MAAM,IAAI,IAAI,gBAAgB,CAAC,KAAK,CAAC,EAAE,CAAC;gBAC3C,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;oBACrB,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;gBACzB,CAAC;YACH,CAAC;QACH,CAAC;QACD,IAAI,CAAC,MAAM,GAAG,KAAK,CAAC;QACpB,IAAI,CAAC,OAAO,GAAG,OAAO,CAAC;IACzB,CAAC;IAED;;OAEG;IACH,CAAC,SAAS;QACR,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,OAAO,EAAE,CAAC;YACjC,KAAK,CAAC,CAAC;gBACL,KAAK,CAAC,QAAQ,CAAC,EAAE;gBACjB,KAAK,CAAC,QAAQ,CAAC,EAAE,IAAI,EAAE;gBACvB,GAAG,CAAC,KAAK,CAAC,UAAU,IAAI,EAAE,CAAC;aAC5B,CAAC;QACJ,CAAC;IACH,CAAC;IAED;;OAEG;IACH,GAAG,CAAC,IAAY;QACd,OAAO,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IAC/B,CAAC;IAED;;OAEG;IACH,OAAO,CAAC,IAAY;QAClB,MAAM,KAAK,GAAG,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QACpC,OAAO,KAAK,KAAK,SAAS;YACxB,CAAC,CAAC,SAAS;YACX,CAAC,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,aAAa,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;IAC7E,CAAC;IAED;;OAEG;IACH,iBAAiB;QACf,MAAM,SAAS,GAAG,IAAI,GAAG,EAAU,CAAC;QACpC,KAAK,MAAM,OAAO,IAAI,IAAI,CAAC,SAAS,EAAE,EAAE,CAAC;YACvC,KAAK,MAAM,QAAQ,IAAI,OAAO,EAAE,CAAC;gBAC/B,SAAS,CAAC,GAAG,CACX,aAAa,CAAC,EAAE,GAAG,QAAQ,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,SAAS,EAAE,CAAC,CAC9D,CAAC;YACJ,CAAC;QACH,CAAC;QACD,OAAO,SAAS,CAAC;IACnB,CAAC;IAED;;;;;;;;OAQG;IACH,eAAe;QACb,MAAM,UAAU,GAAG,IAAI,GAAG,EAAU,CAAC;QACrC,KAAK,MAAM,OAAO,IAAI,IAAI,CAAC,SAAS,EAAE,EAAE,CAAC;YACvC,KAAK,MAAM,QAAQ,IAAI,OAAO,EAAE,CAAC;gBAC/B,IAAI,CAAC,cAAc,CAAC,QAAQ,CAAC,EAAE,CAAC;oBAC9B,UAAU,CAAC,GAAG,CAAC,aAAa,CAAC,EAAE,GAAG,QAAQ,EAAE,KAAK,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC;gBAC/D,CAAC;YACH,CAAC;QACH,CAAC;QACD,OAAO,UAAU,CAAC;IACpB,CAAC;CACF;AAED;;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;;GAEG;AACH,SAAS,KAAK,CAAC,IAAY,EAAE,QAAgB,EAAE,GAAW;IACxD,OAAO;QACL,WAAW,EAAE,GAAG,IAAI,MAAM,QAAQ,KAAK,GAAG,GAAG;QAC7C,KAAK,EAAE,CAAC,UAA2B,EAAsB,EAAE;YACzD,MAAM,MAAM,GAAG,UAAU,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC;YACxC,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;gBACzB,OAAO,GAAG,IAAI,iCAAiC,CAAC;YAClD,CAAC;YACD,OAAO,MAAM,KAAK,QAAQ;gBACxB,CAAC,CAAC,SAAS;gBACX,CAAC,CAAC,GAAG,IAAI,UAAU,MAAM,cAAc,QAAQ,EAAE,CAAC;QACtD,CAAC;KACF,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;GAYG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAA8B;IACzD,KAAK,CAAC,IAAI,EAAE,MAAM,EAAE,iCAAiC,CAAC;IACtD,KAAK,CAAC,IAAI,EAAE,OAAO,EAAE,+BAA+B,CAAC;IACrD,KAAK,CAAC,IAAI,EAAE,MAAM,EAAE,8CAA8C,CAAC;IACnE,KAAK,CAAC,IAAI,EAAE,OAAO,EAAE,6BAA6B,CAAC;IACnD,KAAK,CAAC,MAAM,EAAE,gBAAgB,EAAE,yBAAyB,CAAC;IAC1D,KAAK,CAAC,MAAM,EAAE,gBAAgB,EAAE,+BAA+B,CAAC;IAChE,KAAK,CAAC,IAAI,EAAE,QAAQ,EAAE,oCAAoC,CAAC;IAC3D,KAAK,CAAC,IAAI,EAAE,SAAS,EAAE,uCAAuC,CAAC;IAC/D,KAAK,CAAC,KAAK,EAAE,cAAc,EAAE,2CAA2C,CAAC;IACzE,KAAK,CAAC,IAAI,EAAE,QAAQ,EAAE,2CAA2C,CAAC;IAClE,KAAK,CAAC,IAAI,EAAE,UAAU,EAAE,oBAAoB,CAAC;IAC7C,KAAK,CAAC,IAAI,EAAE,YAAY,EAAE,oBAAoB,CAAC;IAC/C,KAAK,CAAC,IAAI,EAAE,QAAQ,EAAE,oCAAoC,CAAC;IAC3D,KAAK,CAAC,IAAI,EAAE,SAAS,EAAE,oCAAoC,CAAC;IAC5D,KAAK,CAAC,IAAI,EAAE,QAAQ,EAAE,kDAAkD,CAAC;IACzE,KAAK,CAAC,IAAI,EAAE,UAAU,EAAE,qCAAqC,CAAC;IAC9D,KAAK,CAAC,IAAI,EAAE,UAAU,EAAE,8BAA8B,CAAC;IACvD,KAAK,CAAC,IAAI,EAAE,UAAU,EAAE,iDAAiD,CAAC;IAC1E,oEAAoE;IACpE,wEAAwE;IACxE,0EAA0E;IAC1E,yEAAyE;IACzE,gEAAgE;IAChE,KAAK,CAAC,IAAI,EAAE,UAAU,EAAE,uCAAuC,CAAC;IAChE,KAAK,CAAC,IAAI,EAAE,SAAS,EAAE,6CAA6C,CAAC;IACrE,KAAK,CAAC,IAAI,EAAE,QAAQ,EAAE,gDAAgD,CAAC;IACvE,0EAA0E;IAC1E,wEAAwE;IACxE,0EAA0E;IAC1E,4EAA4E;IAC5E,uDAAuD;IACvD,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,yCAAyC,CAAC;IAC3D,KAAK,CAAC,GAAG,EAAE,KAAK,EAAE,+CAA+C,CAAC;IAClE,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,4CAA4C,CAAC;IAC9D,KAAK,CAAC,IAAI,EAAE,QAAQ,EAAE,mDAAmD,CAAC;IAC1E,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,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,8CAA8C,CAAC;IAChE,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,uDAAuD,CAAC;IACzE,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,gCAAgC,CAAC;IAClD,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,6CAA6C,CAAC;IAC/D,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,qCAAqC,CAAC;IACvD;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;IACD;QACE,WAAW,EAAE,iDAAiD;QAC9D,KAAK,EAAE,CAAC,UAA2B,EAAsB,EAAE,CACzD,UAAU,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,YAAY,KAAK,IAAI;YACzC,CAAC,CAAC,SAAS;YACX,CAAC,CAAC,mCAAmC;KAC1C;IACD;QACE,qEAAqE;QACrE,2EAA2E;QAC3E,WAAW,EAAE,wCAAwC;QACrD,KAAK,EAAE,CAAC,UAA2B,EAAsB,EAAE;YACzD,MAAM,KAAK,GAAG,CAAC,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,CAAC,MAAM,CAC3C,CAAC,IAAI,EAAE,EAAE,CAAC,UAAU,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,YAAY,KAAK,IAAI,CACtD,CAAC;YACF,OAAO,KAAK,CAAC,MAAM,KAAK,CAAC;gBACvB,CAAC,CAAC,SAAS;gBACX,CAAC,CAAC,2BAA2B,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;QACpD,CAAC;KACF;IACD;QACE,kEAAkE;QAClE,6DAA6D;QAC7D,WAAW,EAAE,mCAAmC;QAChD,KAAK,EAAE,CAAC,UAA2B,EAAsB,EAAE;YACzD,MAAM,KAAK,GAAG,CAAC,KAAK,EAAE,KAAK,EAAE,IAAI,CAAC,CAAC,MAAM,CACvC,CAAC,IAAI,EAAE,EAAE,CAAC,UAAU,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,YAAY,KAAK,IAAI,CACtD,CAAC;YACF,OAAO,KAAK,CAAC,MAAM,KAAK,CAAC;gBACvB,CAAC,CAAC,SAAS;gBACX,CAAC,CAAC,2BAA2B,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;QACpD,CAAC;KACF;IACD;QACE,yEAAyE;QACzE,gEAAgE;QAChE,sEAAsE;QACtE,WAAW,EACT,sDAAsD;QACxD,KAAK,EAAE,CAAC,UAA2B,EAAsB,EAAE;YACzD,MAAM,MAAM,GAAG,CAAC,IAAY,EAAsB,EAAE,CAClD,UAAU;iBACP,GAAG,CAAC,IAAI,CAAC;gBACV,EAAE,QAAQ,CAAC,EAAE,EAAE,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,aAAa,CAAC,QAAQ,CAAC,CAAC;iBACxD,IAAI,CAAC,GAAG,CAAC,CAAC;YACf,MAAM,QAAQ,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC;YAChC,IAAI,QAAQ,KAAK,eAAe,EAAE,CAAC;gBACjC,OAAO,cAAc,QAAQ,IAAI,SAAS,mCAAmC,CAAC;YAChF,CAAC;YACD,MAAM,KAAK,GAAG,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC,CAAC,MAAM,CAC3C,CAAC,IAAI,EAAE,EAAE,CAAC,MAAM,CAAC,IAAI,CAAC,KAAK,SAAS,CACrC,CAAC;YACF,OAAO,KAAK,CAAC,MAAM,KAAK,CAAC;gBACvB,CAAC,CAAC,SAAS;gBACX,CAAC,CAAC,iCAAiC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;QAC1D,CAAC;KACF;IACD;QACE,WAAW,EAAE,8BAA8B;QAC3C,KAAK,EAAE,CAAC,UAA2B,EAAsB,EAAE;YACzD,MAAM,IAAI,GAAG,UAAU,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,IAAI,CAAC;YACxC,OAAO,IAAI,KAAK,IAAI;gBAClB,CAAC,CAAC,SAAS;gBACX,CAAC,CAAC,cAAc,IAAI,IAAI,SAAS,eAAe,CAAC;QACrD,CAAC;KACF;IACD;QACE,4EAA4E;QAC5E,sEAAsE;QACtE,0EAA0E;QAC1E,0EAA0E;QAC1E,4EAA4E;QAC5E,kCAAkC;QAClC,WAAW,EAAE,gDAAgD;QAC7D,KAAK,EAAE,CAAC,UAA2B,EAAsB,EAAE;YACzD,MAAM,OAAO,GAAG,CAAC,GAAG,UAAU,CAAC,iBAAiB,EAAE,CAAC,CAAC,MAAM,CACxD,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC,oBAAoB,CAAC,GAAG,CAAC,QAAQ,CAAC,CAClD,CAAC;YACF,OAAO,OAAO,CAAC,MAAM,KAAK,CAAC;gBACzB,CAAC,CAAC,SAAS;gBACX,CAAC,CAAC,gBAAgB,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,uCAAuC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;QACxG,CAAC;KACF;IACD;QACE,yEAAyE;QACzE,uEAAuE;QACvE,0EAA0E;QAC1E,oEAAoE;QACpE,WAAW,EAAE,sDAAsD;QACnE,KAAK,EAAE,CAAC,UAA2B,EAAsB,EAAE;YACzD,MAAM,UAAU,GAAG,CAAC,GAAG,UAAU,CAAC,eAAe,EAAE,CAAC,CAAC;YACrD,OAAO,UAAU,CAAC,MAAM,KAAK,CAAC;gBAC5B,CAAC,CAAC,SAAS;gBACX,CAAC,CAAC,gBAAgB,MAAM,CAAC,UAAU,CAAC,MAAM,CAAC,uDAAuD,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;QAC9H,CAAC;KACF;IACD;QACE,WAAW,EAAE,0DAA0D;QACvE,KAAK,EAAE,CAAC,UAA2B,EAAsB,EAAE;YACzD,KAAK,MAAM,KAAK,IAAI,UAAU,CAAC,OAAO,EAAE,CAAC;gBACvC,MAAM,UAAU,GAAG,cAAc,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;gBAC9C,MAAM,KAAK,GAAG,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,MAAM,CACpC,CAAC,QAAQ,EAAE,EAAE,CAAC,QAAQ,CAAC,KAAK,KAAK,IAAI,CACtC,CAAC,MAAM,CAAC;gBACT,iEAAiE;gBACjE,iEAAiE;gBACjE,IAAI,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,MAAM,GAAG,KAAK,GAAG,UAAU,EAAE,CAAC;oBAClD,OAAO,GAAG,KAAK,CAAC,IAAI,QAAQ,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,MAAM,CAAC,kBAAkB,MAAM,CAAC,UAAU,CAAC,aAAa,CAAC;gBAChH,CAAC;YACH,CAAC;YACD,OAAO;QACT,CAAC;KACF;CACF,CAAC;AAEF;;GAEG;AACH,MAAM,UAAU,UAAU,CACxB,OAAmC;IAEnC,MAAM,UAAU,GAAG,IAAI,eAAe,CAAC,OAAO,CAAC,CAAC;IAChD,OAAO,gBAAgB,CAAC,GAAG,CAAC,CAAC,SAAS,EAAE,EAAE,CACxC,SAAS,CAAC,KAAK,CAAC,UAAU,CAAC,CAC5B,CAAC,MAAM,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,KAAK,SAAS,CAAC,CAAC;AAC/C,CAAC"}
+{"version":3,"file":"assertions.js","sourceRoot":"","sources":["../../src/dictionary/assertions.ts"],"names":[],"mappings":"AAIA,OAAO,EAAE,eAAe,EAAE,MAAM,uBAAuB,CAAC;AACxD,OAAO,EAAE,gBAAgB,EAAE,MAAM,uBAAuB,CAAC;AAGzD,OAAO,EAAE,eAAe,EAAE,MAAM,uBAAuB,CAAC;AACxD,OAAO,EAAE,gBAAgB,EAAE,MAAM,uBAAuB,CAAC;AAEzD;;GAEG;AACH,MAAM,UAAU,UAAU,CACxB,OAAmC;IAEnC,MAAM,UAAU,GAAG,IAAI,eAAe,CAAC,OAAO,CAAC,CAAC;IAChD,OAAO,gBAAgB,CAAC,GAAG,CAAC,CAAC,SAAS,EAAE,EAAE,CACxC,SAAS,CAAC,KAAK,CAAC,UAAU,CAAC,CAC5B,CAAC,MAAM,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,KAAK,SAAS,CAAC,CAAC;AAC/C,CAAC"}

package/dist/dictionary/build-assertions.d.ts

@@ -0,0 +1,12 @@
+/**
+ * What must hold of every build, or no artifact is produced.
+ *
+ * These test the **stored** form, which is deliberately not what a user sees:
+ * sandhi is normalised out of the dictionary and reapplied at runtime, so
+ * 一丁不识 is stored as `yī dīng bù shí` and only becomes `yì dīng bù shí` after
+ * the runtime pass. The gold corpus asserts that output form. Asserting `yì`
+ * here would be permanently unsatisfiable — see MERGE.md.
+ */
+import type { BuildAssertion } from "./built-dictionary.js";
+export declare const BUILD_ASSERTIONS: readonly BuildAssertion[];
+//# sourceMappingURL=build-assertions.d.ts.map

package/dist/dictionary/build-assertions.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"build-assertions.d.ts","sourceRoot":"","sources":["../../src/dictionary/build-assertions.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AACH,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,uBAAuB,CAAC;AAO5D,eAAO,MAAM,gBAAgB,EAAE,SAAS,cAAc,EAMrD,CAAC"}

package/dist/dictionary/build-assertions.js

@@ -0,0 +1,13 @@
+import { CHARACTER_READINGS } from "./character-readings.js";
+import { ENTRY_INVARIANTS } from "./entry-invariants.js";
+import { INVENTORY_INVARIANTS } from "./inventory-invariants.js";
+import { TONE_INVARIANTS } from "./tone-invariants.js";
+import { WORD_READINGS } from "./word-readings.js";
+export const BUILD_ASSERTIONS = [
+    ...WORD_READINGS,
+    ...CHARACTER_READINGS,
+    ...TONE_INVARIANTS,
+    ...ENTRY_INVARIANTS,
+    ...INVENTORY_INVARIANTS,
+];
+//# sourceMappingURL=build-assertions.js.map

package/dist/dictionary/build-assertions.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"build-assertions.js","sourceRoot":"","sources":["../../src/dictionary/build-assertions.ts"],"names":[],"mappings":"AAUA,OAAO,EAAE,kBAAkB,EAAE,MAAM,yBAAyB,CAAC;AAC7D,OAAO,EAAE,gBAAgB,EAAE,MAAM,uBAAuB,CAAC;AACzD,OAAO,EAAE,oBAAoB,EAAE,MAAM,2BAA2B,CAAC;AACjE,OAAO,EAAE,eAAe,EAAE,MAAM,sBAAsB,CAAC;AACvD,OAAO,EAAE,aAAa,EAAE,MAAM,oBAAoB,CAAC;AAEnD,MAAM,CAAC,MAAM,gBAAgB,GAA8B;IACzD,GAAG,aAAa;IAChB,GAAG,kBAAkB;IACrB,GAAG,eAAe;IAClB,GAAG,gBAAgB;IACnB,GAAG,oBAAoB;CACxB,CAAC"}

package/dist/dictionary/built-dictionary.d.ts

@@ -0,0 +1,46 @@
+import type { DictionaryEntry } from "./entry.js";
+/**
+ * One thing that must be true of a freshly built dictionary.
+ */
+export interface BuildAssertion {
+    readonly description: string;
+    /** The failure message, or undefined when the assertion holds. */
+    readonly check: (dictionary: BuiltDictionary) => string | undefined;
+}
+/**
+ * A merged dictionary, indexed for the assertions to interrogate.
+ */
+export declare class BuiltDictionary {
+    #private;
+    readonly entries: readonly DictionaryEntry[];
+    /**
+     * Index entries by every key they claim, headwords first.
+     *
+     * The order matters for the same reason it does in the artifact: 发 and 髮
+     * both claim 发, and the entry the key names has to win.
+     */
+    constructor(entries: readonly DictionaryEntry[]);
+    /**
+     * The entry for a word under either script, or undefined.
+     */
+    get(word: string): DictionaryEntry | undefined;
+    /**
+     * A word's zh-CN reading in tone-marked notation, or undefined.
+     */
+    reading(word: string): string | undefined;
+    /**
+     * Every distinct toneless syllable the dictionary uses.
+     */
+    syllableInventory(): ReadonlySet<string>;
+    /**
+     * Every syllable the dictionary uses that is not in the tone the inventory
+     * says that syllable is written in.
+     *
+     * The tones are read off the dictionary in the first place, so this is empty
+     * by construction — until a source refresh brings in a reading nobody has
+     * seen, which is exactly when a reader would start handing that syllable
+     * back and the table would have to be regenerated.
+     */
+    unattestedTones(): ReadonlySet<string>;
+}
+//# sourceMappingURL=built-dictionary.d.ts.map

package/dist/dictionary/built-dictionary.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"built-dictionary.d.ts","sourceRoot":"","sources":["../../src/dictionary/built-dictionary.ts"],"names":[],"mappings":"AASA,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,YAAY,CAAC;AAElD;;GAEG;AACH,MAAM,WAAW,cAAc;IAC7B,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,kEAAkE;IAClE,QAAQ,CAAC,KAAK,EAAE,CAAC,UAAU,EAAE,eAAe,KAAK,MAAM,GAAG,SAAS,CAAC;CACrE;AAED;;GAEG;AACH,qBAAa,eAAe;;IAG1B,QAAQ,CAAC,OAAO,EAAE,SAAS,eAAe,EAAE,CAAC;IAE7C;;;;;OAKG;IACH,YAAY,OAAO,EAAE,SAAS,eAAe,EAAE,EAc9C;IAeD;;OAEG;IACH,GAAG,CAAC,IAAI,EAAE,MAAM,GAAG,eAAe,GAAG,SAAS,CAE7C;IAED;;OAEG;IACH,OAAO,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,SAAS,CAKxC;IAED;;OAEG;IACH,iBAAiB,IAAI,WAAW,CAAC,MAAM,CAAC,CAUvC;IAED;;;;;;;;OAQG;IACH,eAAe,IAAI,WAAW,CAAC,MAAM,CAAC,CAUrC;CACF"}

package/dist/dictionary/built-dictionary.js

@@ -0,0 +1,97 @@
+/**
+ * The dictionary a build assertion is asked about.
+ *
+ * Indexed the way the artifact is, so that an assertion interrogates the same
+ * lookups a reader would rather than the entry list behind them.
+ */
+import { isAttestedTone } from "../syllable/inventory.js";
+import { writeSyllable } from "../syllable/syllable.js";
+import { traditionalForms } from "./artifact.js";
+/**
+ * A merged dictionary, indexed for the assertions to interrogate.
+ */
+export class BuiltDictionary {
+    #byKey;
+    entries;
+    /**
+     * Index entries by every key they claim, headwords first.
+     *
+     * The order matters for the same reason it does in the artifact: 发 and 髮
+     * both claim 发, and the entry the key names has to win.
+     */
+    constructor(entries) {
+        const byKey = new Map();
+        for (const entry of entries) {
+            byKey.set(entry.hans, entry);
+        }
+        for (const entry of entries) {
+            for (const form of traditionalForms(entry)) {
+                if (!byKey.has(form)) {
+                    byKey.set(form, entry);
+                }
+            }
+        }
+        this.#byKey = byKey;
+        this.entries = entries;
+    }
+    /**
+     * Every reading the dictionary holds, under every locale it holds one for.
+     */
+    *#readings() {
+        for (const entry of this.entries) {
+            yield* [
+                entry.readings.cn,
+                entry.readings.tw ?? [],
+                ...(entry.alternates ?? []),
+            ];
+        }
+    }
+    /**
+     * The entry for a word under either script, or undefined.
+     */
+    get(word) {
+        return this.#byKey.get(word);
+    }
+    /**
+     * A word's zh-CN reading in tone-marked notation, or undefined.
+     */
+    reading(word) {
+        const entry = this.#byKey.get(word);
+        return entry === undefined
+            ? undefined
+            : entry.readings.cn.map((syllable) => writeSyllable(syllable)).join(" ");
+    }
+    /**
+     * Every distinct toneless syllable the dictionary uses.
+     */
+    syllableInventory() {
+        const inventory = new Set();
+        for (const reading of this.#readings()) {
+            for (const syllable of reading) {
+                inventory.add(writeSyllable({ ...syllable, erhua: false, tone: undefined }));
+            }
+        }
+        return inventory;
+    }
+    /**
+     * Every syllable the dictionary uses that is not in the tone the inventory
+     * says that syllable is written in.
+     *
+     * The tones are read off the dictionary in the first place, so this is empty
+     * by construction — until a source refresh brings in a reading nobody has
+     * seen, which is exactly when a reader would start handing that syllable
+     * back and the table would have to be regenerated.
+     */
+    unattestedTones() {
+        const unattested = new Set();
+        for (const reading of this.#readings()) {
+            for (const syllable of reading) {
+                if (!isAttestedTone(syllable)) {
+                    unattested.add(writeSyllable({ ...syllable, erhua: false }));
+                }
+            }
+        }
+        return unattested;
+    }
+}
+//# sourceMappingURL=built-dictionary.js.map

package/dist/dictionary/built-dictionary.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"built-dictionary.js","sourceRoot":"","sources":["../../src/dictionary/built-dictionary.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,EAAE,cAAc,EAAE,MAAM,0BAA0B,CAAC;AAC1D,OAAO,EAAiB,aAAa,EAAE,MAAM,yBAAyB,CAAC;AACvE,OAAO,EAAE,gBAAgB,EAAE,MAAM,eAAe,CAAC;AAYjD;;GAEG;AACH,MAAM,OAAO,eAAe;IACjB,MAAM,CAAuC;IAE7C,OAAO,CAA6B;IAE7C;;;;;OAKG;IACH,YAAY,OAAmC;QAC7C,MAAM,KAAK,GAAG,IAAI,GAAG,EAA2B,CAAC;QACjD,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;YAC5B,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;QAC/B,CAAC;QACD,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;YAC5B,KAAK,MAAM,IAAI,IAAI,gBAAgB,CAAC,KAAK,CAAC,EAAE,CAAC;gBAC3C,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;oBACrB,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;gBACzB,CAAC;YACH,CAAC;QACH,CAAC;QACD,IAAI,CAAC,MAAM,GAAG,KAAK,CAAC;QACpB,IAAI,CAAC,OAAO,GAAG,OAAO,CAAC;IACzB,CAAC;IAED;;OAEG;IACH,CAAC,SAAS;QACR,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,OAAO,EAAE,CAAC;YACjC,KAAK,CAAC,CAAC;gBACL,KAAK,CAAC,QAAQ,CAAC,EAAE;gBACjB,KAAK,CAAC,QAAQ,CAAC,EAAE,IAAI,EAAE;gBACvB,GAAG,CAAC,KAAK,CAAC,UAAU,IAAI,EAAE,CAAC;aAC5B,CAAC;QACJ,CAAC;IACH,CAAC;IAED;;OAEG;IACH,GAAG,CAAC,IAAY;QACd,OAAO,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IAC/B,CAAC;IAED;;OAEG;IACH,OAAO,CAAC,IAAY;QAClB,MAAM,KAAK,GAAG,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QACpC,OAAO,KAAK,KAAK,SAAS;YACxB,CAAC,CAAC,SAAS;YACX,CAAC,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,aAAa,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;IAC7E,CAAC;IAED;;OAEG;IACH,iBAAiB;QACf,MAAM,SAAS,GAAG,IAAI,GAAG,EAAU,CAAC;QACpC,KAAK,MAAM,OAAO,IAAI,IAAI,CAAC,SAAS,EAAE,EAAE,CAAC;YACvC,KAAK,MAAM,QAAQ,IAAI,OAAO,EAAE,CAAC;gBAC/B,SAAS,CAAC,GAAG,CACX,aAAa,CAAC,EAAE,GAAG,QAAQ,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,SAAS,EAAE,CAAC,CAC9D,CAAC;YACJ,CAAC;QACH,CAAC;QACD,OAAO,SAAS,CAAC;IACnB,CAAC;IAED;;;;;;;;OAQG;IACH,eAAe;QACb,MAAM,UAAU,GAAG,IAAI,GAAG,EAAU,CAAC;QACrC,KAAK,MAAM,OAAO,IAAI,IAAI,CAAC,SAAS,EAAE,EAAE,CAAC;YACvC,KAAK,MAAM,QAAQ,IAAI,OAAO,EAAE,CAAC;gBAC/B,IAAI,CAAC,cAAc,CAAC,QAAQ,CAAC,EAAE,CAAC;oBAC9B,UAAU,CAAC,GAAG,CAAC,aAAa,CAAC,EAAE,GAAG,QAAQ,EAAE,KAAK,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC;gBAC/D,CAAC;YACH,CAAC;QACH,CAAC;QACD,OAAO,UAAU,CAAC;IACpB,CAAC;CACF"}

package/dist/dictionary/cedict-senses.d.ts

@@ -0,0 +1,80 @@
+import type { CedictEntry } from "../sources/cedict.js";
+import type { Syllable } from "../syllable/syllable.js";
+/**
+ * Whether a phrase-corpus headword is really the 繁體 spelling of another word.
+ *
+ * `large_pinyin.txt` is a 简体 corpus, and where it carries a 繁體 headword
+ * anyway it has read it as though the characters were 简体: 特徵 is `tè zhǐ`
+ * there because 徵 alone is `zhǐ`, 沈溺 is `shěn nì` because 沈 alone is the
+ * surname, 蝨子 is `shī zǐ`, 纔然 is `shān rán`. Every one of those is a word
+ * whose 简体 form the corpus also carries, correctly.
+ *
+ * Left in, each becomes an entry of its own that outranks the 繁體 key the
+ * derivation would have hung on the 简体 entry, so the word reads one way in a
+ * tier holding the phrase tail and another way in a tier without it.
+ *
+ * So the corpus is held to its own contract: a headword CC-CEDICT knows only as
+ * the 繁體 spelling of a different 简体 word contributes no entry, and the word
+ * keeps the reading CC-CEDICT pairs it with. It is 72 headwords, 13 of which
+ * the corpus reads wrongly; the other 59 arrive at the same reading either way.
+ *
+ * Narrow on purpose. 2,854 corpus headwords contain a 繁體-only character and
+ * only these 72 are ones CC-CEDICT pairs — the rest are rare or mixed-script
+ * words whose characters do read them, and dropping those would lose coverage
+ * to fix nothing.
+ */
+export declare function isSpeltTraditionally(word: string, byWord: ReadonlyMap<string, readonly CedictEntry[]>, byHant: ReadonlyMap<string, readonly CedictEntry[]>): boolean;
+/**
+ * CC-CEDICT's entries for a word, keyed by one of its two written forms.
+ */
+export declare function indexCedict(cedict: readonly CedictEntry[], formOf: (entry: CedictEntry) => string): ReadonlyMap<string, readonly CedictEntry[]>;
+/**
+ * A word's readings as CC-CEDICT gives them, one per sense.
+ */
+export declare function cedictReadingsOf(word: string, entries: readonly CedictEntry[]): readonly (readonly Syllable[])[];
+/**
+ * Whether a proposed 國語 reading is really another 普通话 sense of the word.
+ *
+ * The test that separates a locale shift from a sense selection, which the
+ * sources write the same way and mean differently. 地 is offered `dì` against a
+ * 普通话 `de` — but CC-CEDICT lists 地[de5] and 地[di4] as two entries, so `dì`
+ * is what 地 reads in 普通话 when it means the ground, not what 國語 does to the
+ * particle. The adverbial 地 is `de` in Taipei too, and 4,240 entries end in it.
+ * 都 is the same shape, and so are 着, 应, 差, 称, 斗, 舍, 薄 and 万.
+ *
+ * A genuine delta leaves no such trace: nothing in 普通话 reads 和 as `hàn`, 期
+ * as `qí` or 垃 as `lè`, which is why those survive this and 71 characters and
+ * 3 words do not.
+ *
+ * Only CC-CEDICT is consulted. Unihan's reading fields carry rare and historical
+ * pronunciations alongside current ones — 驯 is listed `xún` somewhere in them —
+ * and a reading no one uses today is not evidence that a source meant a sense
+ * rather than a locale.
+ */
+export declare function isOwnSense(taiwan: readonly Syllable[], senseReadings: readonly (readonly Syllable[])[]): boolean;
+/**
+ * Whether a character's `Taiwan pr.` note describes one sense rather than the
+ * character.
+ *
+ * The other half of the sense test, and the one {@link isOwnSense} cannot do.
+ * That test asks whether the offered reading is a 普通话 sense of the word; this
+ * asks whether the note was ever about the whole word to begin with. 從's
+ * `Taiwan pr. [zong4]` sits on three of its eight senses — 侍從, 從兄弟, 從犯 —
+ * and 教育部's dictionary agrees with CC-CEDICT about which: 跟隨, 依順, 參與 and
+ * the preposition are `cóng` in Taipei exactly as they are in Beijing. Applying
+ * it to the headword made 我从北京来 read `wǒ zòng Běijīng lái`. 會 (`huǐ`, only
+ * 一會兒), 勞 (`lào`, only 慰勞) and 燥 (`sào`, only 肉燥) are the same shape.
+ *
+ * A note the entry leads with is kept, because the leading sense is what the
+ * character means with nothing to narrow it: 和's `hàn` is on the conjunction,
+ * which is what a bare 和 almost always is.
+ *
+ * **Only a character is tested this way.** Its entry is what every occurrence
+ * no longer word covers falls back to, so it has to carry the reading that
+ * survives out of context; a multi-character headword is reached only where
+ * that exact word is written. Four of them carry an inline note — 相親, 載具,
+ * 高挑 and 樂色 — and only 相親 has senses that differ, its dominant one being
+ * the matchmaking sense that really is `xiàngqīn`.
+ */
+export declare function isSenseScopedNote(word: string, sense: CedictEntry | undefined): boolean;
+//# sourceMappingURL=cedict-senses.d.ts.map

package/dist/dictionary/cedict-senses.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"cedict-senses.d.ts","sourceRoot":"","sources":["../../src/dictionary/cedict-senses.ts"],"names":[],"mappings":"AAQA,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAC;AACxD,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAIxD;;;;;;;;;;;;;;;;;;;;;;GAsBG;AACH,wBAAgB,oBAAoB,CAClC,IAAI,EAAE,MAAM,EACZ,MAAM,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,WAAW,EAAE,CAAC,EACnD,MAAM,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,WAAW,EAAE,CAAC,GAClD,OAAO,CAOT;AAED;;GAEG;AACH,wBAAgB,WAAW,CACzB,MAAM,EAAE,SAAS,WAAW,EAAE,EAC9B,MAAM,EAAE,CAAC,KAAK,EAAE,WAAW,KAAK,MAAM,GACrC,WAAW,CAAC,MAAM,EAAE,SAAS,WAAW,EAAE,CAAC,CAY7C;AAED;;GAEG;AACH,wBAAgB,gBAAgB,CAC9B,IAAI,EAAE,MAAM,EACZ,OAAO,EAAE,SAAS,WAAW,EAAE,GAC9B,SAAS,CAAC,SAAS,QAAQ,EAAE,CAAC,EAAE,CAIlC;AAED;;;;;;;;;;;;;;;;;;GAkBG;AACH,wBAAgB,UAAU,CACxB,MAAM,EAAE,SAAS,QAAQ,EAAE,EAC3B,aAAa,EAAE,SAAS,CAAC,SAAS,QAAQ,EAAE,CAAC,EAAE,GAC9C,OAAO,CAET;AAED;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AACH,wBAAgB,iBAAiB,CAC/B,IAAI,EAAE,MAAM,EACZ,KAAK,EAAE,WAAW,GAAG,SAAS,GAC7B,OAAO,CAET"}