@kensio/pinyinjs 1.7.4 → 1.7.5

package/dist/dictionary/artifact.js

@@ -1,135 +1,18 @@
-import { toCharacters } from "../script/characters.js";
-import { readSyllable, writeSyllable, } from "../syllable/syllable.js";
-import { isSameReading } from "./entry.js";
-import { FrequencyTable } from "./frequency-table.js";
-import { KeyIndex } from "./key-index.js";
-/**
- * Separator between a line's columns.
- */
-const COLUMN = "\t";
-/**
- * Separator between lines, matching the key index's own separator.
- */
-const LINE = "\n";
-/**
- * Separator between the alternate readings of a character.
- */
-const ALTERNATE = ",";
-/**
- * The flag marking a proper noun.
- */
-const PROPER_NOUN_FLAG = "p";
-/**
- * Where a proper name divides, written after the proper-noun flag.
- *
- * `p` alone is a proper noun whose parts CC-CEDICT does not divide; `p1` is one
- * that divides after the first character, `p2.4` after the second and fourth.
- * The positions are decimal and dot-separated rather than one digit each,
- * because the longest word carrying any is 19 characters.
- *
- * It rides in the flags column rather than taking a column of its own because
- * it is only ever present on a proper noun, and only 8,205 of 723,139 keys
- * carry one — a sixth column would cost a separator on every line to say
- * nothing on almost all of them. *
- * `security/detect-unsafe-regex` flags the nested quantifier, and there is no
- * ambiguity behind it: a literal `.` separates `\d+` from `(?:\.\d+)*`, so no
- * input can be split between them two ways. Timed on a string of digits and
- * dots that cannot match, it is linear — 8k to 128k characters doubles the time
- * for each doubling of the input, 39µs to 607µs.
- */
-const NAME_BOUNDARIES = /^p(?<at>\d+(?:\.\d+)*)$/u;
-const BOUNDARY_SEPARATOR = ".";
-/**
- * The boundaries a flags column states, empty where it states none.
- */
-function nameBoundariesIn(flags) {
-    const at = NAME_BOUNDARIES.exec(flags)?.groups?.["at"];
-    return at === undefined ? [] : at.split(BOUNDARY_SEPARATOR).map(Number);
-}
 /**
- * Write a reading in tone-numbered notation.
+ * Compiling merged entries into the artifact written to `data/`.
  *
- * Numbered rather than tone-marked because it is ASCII, so it survives any
- * encoding, and because the tone is then a character rather than a combining
- * mark that normalisation could move.
+ * Which entry owns each key is settled first, and separately, in
+ * `artifact-claims.ts`; what is here is turning the entries that survived that
+ * into lines.
  */
-export function encodeReading(syllables) {
-    return syllables
-        .map((syllable) => writeSyllable(syllable, "numbers"))
-        .join(" ");
-}
-/**
- * Read a reading back, or return undefined if any syllable is malformed.
- */
-export function decodeReading(text) {
-    if (text === "") {
-        return undefined;
-    }
-    const syllables = text.split(" ").map((token) => readSyllable(token));
-    return syllables.includes(undefined)
-        ? undefined
-        : syllables;
-}
-/**
- * Order two keys the way {@link KeyIndex} requires them: by UTF-16 code unit.
- */
-function byCodeUnit(left, right) {
-    if (left < right) {
-        return -1;
-    }
-    return left > right ? 1 : 0;
-}
-/**
- * Every 繁體 spelling an entry claims a key for.
- */
-export function traditionalForms(entry) {
-    return [entry.hant, ...(entry.hantVariants ?? [])];
-}
-/**
- * Let one entry claim each of its 繁體 spellings, where nothing outranks it.
- *
- * A spelling equal to the 简体 headword is already claimed, and a spelling that
- * is some entry's own headword belongs to that entry — 发 and 髮 disagree about
- * the reading, and the one the key names has to win.
- */
-function claimTraditional(claimed, entry) {
-    for (const form of traditionalForms(entry)) {
-        const held = claimed.get(form);
-        const isTaken = form === entry.hans || held?.hans === form;
-        if (!isTaken && (held === undefined || entry.frequency > held.frequency)) {
-            claimed.set(form, entry);
-        }
-    }
-}
-/**
- * Decide which entry owns each key, since an entry claims both its scripts.
- *
- * Entries overlap: 头 claims 頭 as its 繁體 key, and 頭 is also a character
- * entry of its own. They agree, so the conflict is harmless there — but 发 and
- * 髮 both claim 发, and they emphatically do not agree, one reading `fā` and the
- * other `fà`. The entry whose own headword *is* the key wins, which is what
- * keeps 发 on its own reading rather than the one that reached it sideways.
- *
- * An entry claims every 繁體 spelling a source attests for it, not only the one
- * it stores as {@link DictionaryEntry.hant}: 重复 claims 重複 and 重覆 alike,
- * since a reader typing either expects `chóngfù` rather than the character by
- * character reading a missing key falls back to.
- */
-function claimKeys(entries) {
-    const claimed = new Map();
-    // Headwords first, so that the second pass can never displace one: a 繁體
-    // alias reaching a key sideways must not outrank the entry that key names.
-    for (const entry of entries) {
-        const held = claimed.get(entry.hans);
-        if (held === undefined || entry.frequency > held.frequency) {
-            claimed.set(entry.hans, entry);
-        }
-    }
-    for (const entry of entries) {
-        claimTraditional(claimed, entry);
-    }
-    return claimed;
-}
+import { toCharacters } from "../script/characters.js";
+export { claimKeys } from "./artifact-claims.js";
+export { findRoundTripFailure, readArtifact, readingsByKey, } from "./artifact-read.js";
+import { FrequencyTable } from "./frequency-table.js";
+import { KeyIndex } from "./key-index.js";
+import { claimKeys } from "./artifact-claims.js";
+import { ALTERNATE, BOUNDARY_SEPARATOR, byCodeUnit, COLUMN, encodeReading, LINE, PROPER_NOUN_FLAG, } from "./artifact-format.js";
+export { decodeReading, encodeReading, traditionalForms, } from "./artifact-format.js";
 /**
  * The default reading of each single character, as the artifact records it.
  */
@@ -219,104 +102,4 @@ export function buildArtifact(entries) {
         frequencies: FrequencyTable.build(frequencies).serialise(),
     };
 }
-/**
- * Read an artifact back into entries, for tests and for the loader.
- *
- * The reverse of {@link buildArtifact} but for the two things it does not
- * store: frequency comes back as a bucket rather than a count, since it is
- * quantised on the way out, and 简体/繁體 pairing is absent, so every entry
- * reports its key as both. Entries are returned in key order.
- */
-export function readArtifact(artifact) {
-    const index = KeyIndex.from(artifact.keys);
-    const lines = artifact.entries === "" ? [] : artifact.entries.split(LINE);
-    const frequencies = FrequencyTable.from(artifact.frequencies, index.size);
-    // Two passes: a derivable reading needs its characters' readings, which are
-    // themselves entries in the same artifact.
-    const defaults = new Map();
-    for (let at = 0; at < index.size; at++) {
-        const key = index.keyAt(at);
-        const reading = (lines[at] ?? "").split(COLUMN)[0] ?? "";
-        if (toCharacters(key).length === 1 && reading !== "") {
-            defaults.set(key, reading);
-        }
-    }
-    const entries = [];
-    for (let at = 0; at < index.size; at++) {
-        const key = index.keyAt(at);
-        const [reading = "", taiwan = "", partOfSpeech = "", flags = "", alternates = "",] = (lines[at] ?? "").split(COLUMN);
-        const nameBoundaries = nameBoundariesIn(flags);
-        const stored = reading === ""
-            ? toCharacters(key)
-                .map((character) => defaults.get(character) ?? "")
-                .join(" ")
-            : reading;
-        const cn = decodeReading(stored);
-        if (cn === undefined) {
-            throw new Error(`artifact entry for ${key} has no readable reading`);
-        }
-        const tw = decodeReading(taiwan);
-        // The artifact stores keys, not pairs, so a key read back is both scripts
-        // as far as it knows. See buildArtifact for what that leaves out and why.
-        entries.push({
-            hans: key,
-            hant: key,
-            readings: { cn, ...(tw !== undefined && { tw }) },
-            frequency: frequencies.bucketOf(at),
-            partOfSpeech,
-            isProperNoun: flags.includes(PROPER_NOUN_FLAG),
-            ...(nameBoundaries.length > 0 && { nameBoundaries }),
-            ...(alternates !== "" && {
-                alternates: alternates
-                    .split(ALTERNATE)
-                    .map((text) => decodeReading(text))
-                    .filter((decoded) => decoded !== undefined),
-            }),
-        });
-    }
-    return entries;
-}
-/**
- * The reading every key in an artifact resolves to, in tone-numbered notation.
- *
- * Resolved rather than read off the line, since 87.8% of multi-character lines
- * store no reading at all and mean "the characters' defaults, in order" — and
- * those defaults are whatever *this* artifact holds. That is exactly what makes
- * the comparison between tiers worth doing: a tier missing a character does not
- * only lose that character, it changes every derived reading that stands on it.
- */
-export function readingsByKey(artifact) {
-    const index = KeyIndex.from(artifact.keys);
-    const readings = new Map();
-    for (const [at, entry] of readArtifact(artifact).entries()) {
-        readings.set(index.keyAt(at), encodeReading(entry.readings.cn));
-    }
-    return readings;
-}
-/**
- * The first key an artifact does not read back the same, or undefined.
- *
- * Only the fields the artifact stores are compared: frequency is quantised on
- * the way out, and the 简体/繁體 labelling of a pair is not recoverable from a
- * key alone, so neither is checked here.
- *
- * Returns the offending key rather than a bare false so that a build failure
- * names the word to go and look at.
- */
-export function findRoundTripFailure(entries, artifact) {
-    const index = KeyIndex.from(artifact.keys);
-    const read = readArtifact(artifact);
-    const byKey = new Map();
-    for (const [at, entry] of read.entries()) {
-        byKey.set(index.keyAt(at), entry);
-    }
-    for (const [key, entry] of claimKeys(entries)) {
-        const other = byKey.get(key);
-        if (other === undefined ||
-            !isSameReading(entry.readings.cn, other.readings.cn)) {
-            return key;
-        }
-    }
-    return undefined;
-}
 //# sourceMappingURL=artifact.js.map

package/dist/dictionary/artifact.js.map

@@ -1 +1 @@
-{"version":3,"file":"artifact.js","sourceRoot":"","sources":["../../src/dictionary/artifact.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AACvD,OAAO,EACL,YAAY,EAEZ,aAAa,GACd,MAAM,yBAAyB,CAAC;AAEjC,OAAO,EAAE,aAAa,EAAE,MAAM,YAAY,CAAC;AAC3C,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AACtD,OAAO,EAAE,QAAQ,EAAE,MAAM,gBAAgB,CAAC;AAE1C;;GAEG;AACH,MAAM,MAAM,GAAG,IAAI,CAAC;AAEpB;;GAEG;AACH,MAAM,IAAI,GAAG,IAAI,CAAC;AAElB;;GAEG;AACH,MAAM,SAAS,GAAG,GAAG,CAAC;AAEtB;;GAEG;AACH,MAAM,gBAAgB,GAAG,GAAG,CAAC;AAE7B;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAM,eAAe,GAAG,0BAA0B,CAAC;AAEnD,MAAM,kBAAkB,GAAG,GAAG,CAAC;AAE/B;;GAEG;AACH,SAAS,gBAAgB,CAAC,KAAa;IACrC,MAAM,EAAE,GAAG,eAAe,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,MAAM,EAAE,CAAC,IAAI,CAAC,CAAC;IACvD,OAAO,EAAE,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,KAAK,CAAC,kBAAkB,CAAC,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;AAC1E,CAAC;AAmBD;;;;;;GAMG;AACH,MAAM,UAAU,aAAa,CAAC,SAA8B;IAC1D,OAAO,SAAS;SACb,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,aAAa,CAAC,QAAQ,EAAE,SAAS,CAAC,CAAC;SACrD,IAAI,CAAC,GAAG,CAAC,CAAC;AACf,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,aAAa,CAAC,IAAY;IACxC,IAAI,IAAI,KAAK,EAAE,EAAE,CAAC;QAChB,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,YAAY,CAAC,KAAK,CAAC,CAAC,CAAC;IACtE,OAAO,SAAS,CAAC,QAAQ,CAAC,SAAS,CAAC;QAClC,CAAC,CAAC,SAAS;QACX,CAAC,CAAE,SAAiC,CAAC;AACzC,CAAC;AAED;;GAEG;AACH,SAAS,UAAU,CAAC,IAAY,EAAE,KAAa;IAC7C,IAAI,IAAI,GAAG,KAAK,EAAE,CAAC;QACjB,OAAO,CAAC,CAAC,CAAC;IACZ,CAAC;IACD,OAAO,IAAI,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;AAC9B,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,gBAAgB,CAAC,KAAsB;IACrD,OAAO,CAAC,KAAK,CAAC,IAAI,EAAE,GAAG,CAAC,KAAK,CAAC,YAAY,IAAI,EAAE,CAAC,CAAC,CAAC;AACrD,CAAC;AAED;;;;;;GAMG;AACH,SAAS,gBAAgB,CACvB,OAAqC,EACrC,KAAsB;IAEtB,KAAK,MAAM,IAAI,IAAI,gBAAgB,CAAC,KAAK,CAAC,EAAE,CAAC;QAC3C,MAAM,IAAI,GAAG,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QAC/B,MAAM,OAAO,GAAG,IAAI,KAAK,KAAK,CAAC,IAAI,IAAI,IAAI,EAAE,IAAI,KAAK,IAAI,CAAC;QAC3D,IAAI,CAAC,OAAO,IAAI,CAAC,IAAI,KAAK,SAAS,IAAI,KAAK,CAAC,SAAS,GAAG,IAAI,CAAC,SAAS,CAAC,EAAE,CAAC;YACzE,OAAO,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;QAC3B,CAAC;IACH,CAAC;AACH,CAAC;AAED;;;;;;;;;;;;;GAaG;AACH,SAAS,SAAS,CAChB,OAAmC;IAEnC,MAAM,OAAO,GAAG,IAAI,GAAG,EAA2B,CAAC;IAEnD,wEAAwE;IACxE,2EAA2E;IAC3E,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,MAAM,IAAI,GAAG,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QACrC,IAAI,IAAI,KAAK,SAAS,IAAI,KAAK,CAAC,SAAS,GAAG,IAAI,CAAC,SAAS,EAAE,CAAC;YAC3D,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;QACjC,CAAC;IACH,CAAC;IAED,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,gBAAgB,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC;IACnC,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC;AAED;;GAEG;AACH,SAAS,iBAAiB,CACxB,OAA6C;IAE7C,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAkB,CAAC;IAC3C,KAAK,MAAM,CAAC,GAAG,EAAE,KAAK,CAAC,IAAI,OAAO,EAAE,CAAC;QACnC,IAAI,YAAY,CAAC,GAAG,CAAC,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACnC,QAAQ,CAAC,GAAG,CAAC,GAAG,EAAE,aAAa,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAC,CAAC;QACtD,CAAC;IACH,CAAC;IACD,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,WAAW,CAClB,GAAW,EACX,OAAe,EACf,QAAqC;IAErC,MAAM,KAAK,GAAG,YAAY,CAAC,GAAG,CAAC,CAAC;IAChC,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACrB,OAAO,KAAK,CAAC;IACf,CAAC;IACD,MAAM,OAAO,GAAa,EAAE,CAAC;IAC7B,KAAK,MAAM,SAAS,IAAI,KAAK,EAAE,CAAC;QAC9B,MAAM,KAAK,GAAG,QAAQ,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QACtC,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;YACxB,OAAO,KAAK,CAAC;QACf,CAAC;QACD,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;IACtB,CAAC;IACD,OAAO,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,OAAO,CAAC;AACvC,CAAC;AAED;;GAEG;AACH,SAAS,WAAW,CAAC,OAA0B;IAC7C,IAAI,GAAG,GAAG,OAAO,CAAC,MAAM,CAAC;IACzB,OAAO,GAAG,GAAG,CAAC,IAAI,OAAO,CAAC,GAAG,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC;QAC1C,GAAG,EAAE,CAAC;IACR,CAAC;IACD,OAAO,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;AAC5C,CAAC;AAED;;;;;;;;;;;;GAYG;AACH,MAAM,UAAU,aAAa,CAC3B,OAAmC;IAEnC,MAAM,OAAO,GAAG,SAAS,CAAC,OAAO,CAAC,CAAC;IACnC,MAAM,QAAQ,GAAG,iBAAiB,CAAC,OAAO,CAAC,CAAC;IAC5C,0EAA0E;IAC1E,wEAAwE;IACxE,MAAM,OAAO,GAAG,CAAC,GAAG,OAAO,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,KAAK,CAAC,EAAE,EAAE,CACxD,UAAU,CAAC,IAAI,EAAE,KAAK,CAAC,CACxB,CAAC;IACF,MAAM,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;IAEpE,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,MAAM,WAAW,GAAa,EAAE,CAAC;IACjC,KAAK,MAAM,CAAC,GAAG,EAAE,KAAK,CAAC,IAAI,OAAO,EAAE,CAAC;QACnC,MAAM,OAAO,GAAG,aAAa,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAC;QAEjD,KAAK,CAAC,IAAI,CACR,WAAW,CAAC;YACV,WAAW,CAAC,GAAG,EAAE,OAAO,EAAE,QAAQ,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,OAAO;YAClD,KAAK,CAAC,QAAQ,CAAC,EAAE,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,aAAa,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC;YACvE,KAAK,CAAC,YAAY;YAClB,KAAK,CAAC,YAAY;gBAChB,CAAC,CAAC,gBAAgB;oBAChB,CAAC,KAAK,CAAC,cAAc,IAAI,EAAE,CAAC,CAAC,IAAI,CAAC,kBAAkB,CAAC;gBACvD,CAAC,CAAC,EAAE;YACN,CAAC,KAAK,CAAC,UAAU,IAAI,EAAE,CAAC;iBACrB,GAAG,CAAC,CAAC,SAAS,EAAE,EAAE,CAAC,aAAa,CAAC,SAAS,CAAC,CAAC;iBAC5C,IAAI,CAAC,SAAS,CAAC;SACnB,CAAC,CACH,CAAC;QACF,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC;IACpC,CAAC;IAED,OAAO;QACL,IAAI,EAAE,KAAK,CAAC,SAAS,EAAE;QACvB,OAAO,EAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC;QACzB,WAAW,EAAE,cAAc,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC,SAAS,EAAE;KAC3D,CAAC;AACJ,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,YAAY,CAC1B,QAA4B;IAE5B,MAAM,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC;IAC3C,MAAM,KAAK,GAAG,QAAQ,CAAC,OAAO,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IAC1E,MAAM,WAAW,GAAG,cAAc,CAAC,IAAI,CAAC,QAAQ,CAAC,WAAW,EAAE,KAAK,CAAC,IAAI,CAAC,CAAC;IAE1E,4EAA4E;IAC5E,2CAA2C;IAC3C,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAkB,CAAC;IAC3C,KAAK,IAAI,EAAE,GAAG,CAAC,EAAE,EAAE,GAAG,KAAK,CAAC,IAAI,EAAE,EAAE,EAAE,EAAE,CAAC;QACvC,MAAM,GAAG,GAAG,KAAK,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC;QAC5B,MAAM,OAAO,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC,IAAI,EAAE,CAAC,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;QACzD,IAAI,YAAY,CAAC,GAAG,CAAC,CAAC,MAAM,KAAK,CAAC,IAAI,OAAO,KAAK,EAAE,EAAE,CAAC;YACrD,QAAQ,CAAC,GAAG,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC;QAC7B,CAAC;IACH,CAAC;IAED,MAAM,OAAO,GAAsB,EAAE,CAAC;IACtC,KAAK,IAAI,EAAE,GAAG,CAAC,EAAE,EAAE,GAAG,KAAK,CAAC,IAAI,EAAE,EAAE,EAAE,EAAE,CAAC;QACvC,MAAM,GAAG,GAAG,KAAK,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC;QAC5B,MAAM,CACJ,OAAO,GAAG,EAAE,EACZ,MAAM,GAAG,EAAE,EACX,YAAY,GAAG,EAAE,EACjB,KAAK,GAAG,EAAE,EACV,UAAU,GAAG,EAAE,EAChB,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC,IAAI,EAAE,CAAC,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;QACpC,MAAM,cAAc,GAAG,gBAAgB,CAAC,KAAK,CAAC,CAAC;QAE/C,MAAM,MAAM,GACV,OAAO,KAAK,EAAE;YACZ,CAAC,CAAC,YAAY,CAAC,GAAG,CAAC;iBACd,GAAG,CAAC,CAAC,SAAS,EAAE,EAAE,CAAC,QAAQ,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC;iBACjD,IAAI,CAAC,GAAG,CAAC;YACd,CAAC,CAAC,OAAO,CAAC;QACd,MAAM,EAAE,GAAG,aAAa,CAAC,MAAM,CAAC,CAAC;QACjC,IAAI,EAAE,KAAK,SAAS,EAAE,CAAC;YACrB,MAAM,IAAI,KAAK,CAAC,sBAAsB,GAAG,0BAA0B,CAAC,CAAC;QACvE,CAAC;QACD,MAAM,EAAE,GAAG,aAAa,CAAC,MAAM,CAAC,CAAC;QAEjC,0EAA0E;QAC1E,0EAA0E;QAC1E,OAAO,CAAC,IAAI,CAAC;YACX,IAAI,EAAE,GAAG;YACT,IAAI,EAAE,GAAG;YACT,QAAQ,EAAE,EAAE,EAAE,EAAE,GAAG,CAAC,EAAE,KAAK,SAAS,IAAI,EAAE,EAAE,EAAE,CAAC,EAAE;YACjD,SAAS,EAAE,WAAW,CAAC,QAAQ,CAAC,EAAE,CAAC;YACnC,YAAY;YACZ,YAAY,EAAE,KAAK,CAAC,QAAQ,CAAC,gBAAgB,CAAC;YAC9C,GAAG,CAAC,cAAc,CAAC,MAAM,GAAG,CAAC,IAAI,EAAE,cAAc,EAAE,CAAC;YACpD,GAAG,CAAC,UAAU,KAAK,EAAE,IAAI;gBACvB,UAAU,EAAE,UAAU;qBACnB,KAAK,CAAC,SAAS,CAAC;qBAChB,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC;qBAClC,MAAM,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,KAAK,SAAS,CAAC;aAC9C,CAAC;SACH,CAAC,CAAC;IACL,CAAC;IACD,OAAO,OAAO,CAAC;AACjB,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,aAAa,CAC3B,QAA4B;IAE5B,MAAM,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC;IAC3C,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAkB,CAAC;IAC3C,KAAK,MAAM,CAAC,EAAE,EAAE,KAAK,CAAC,IAAI,YAAY,CAAC,QAAQ,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC;QAC3D,QAAQ,CAAC,GAAG,CAAC,KAAK,CAAC,KAAK,CAAC,EAAE,CAAC,EAAE,aAAa,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAC,CAAC;IAClE,CAAC;IACD,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,UAAU,oBAAoB,CAClC,OAAmC,EACnC,QAA4B;IAE5B,MAAM,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC;IAC3C,MAAM,IAAI,GAAG,YAAY,CAAC,QAAQ,CAAC,CAAC;IACpC,MAAM,KAAK,GAAG,IAAI,GAAG,EAA2B,CAAC;IACjD,KAAK,MAAM,CAAC,EAAE,EAAE,KAAK,CAAC,IAAI,IAAI,CAAC,OAAO,EAAE,EAAE,CAAC;QACzC,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,KAAK,CAAC,EAAE,CAAC,EAAE,KAAK,CAAC,CAAC;IACpC,CAAC;IAED,KAAK,MAAM,CAAC,GAAG,EAAE,KAAK,CAAC,IAAI,SAAS,CAAC,OAAO,CAAC,EAAE,CAAC;QAC9C,MAAM,KAAK,GAAG,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;QAC7B,IACE,KAAK,KAAK,SAAS;YACnB,CAAC,aAAa,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,EAAE,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,EACpD,CAAC;YACD,OAAO,GAAG,CAAC;QACb,CAAC;IACH,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC"}
+{"version":3,"file":"artifact.js","sourceRoot":"","sources":["../../src/dictionary/artifact.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AACvD,OAAO,EAAE,SAAS,EAAE,MAAM,sBAAsB,CAAC;AACjD,OAAO,EACL,oBAAoB,EACpB,YAAY,EACZ,aAAa,GACd,MAAM,oBAAoB,CAAC;AAE5B,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AACtD,OAAO,EAAE,QAAQ,EAAE,MAAM,gBAAgB,CAAC;AAC1C,OAAO,EAAE,SAAS,EAAE,MAAM,sBAAsB,CAAC;AACjD,OAAO,EACL,SAAS,EACT,kBAAkB,EAClB,UAAU,EACV,MAAM,EAEN,aAAa,EACb,IAAI,EACJ,gBAAgB,GACjB,MAAM,sBAAsB,CAAC;AAE9B,OAAO,EACL,aAAa,EAEb,aAAa,EACb,gBAAgB,GACjB,MAAM,sBAAsB,CAAC;AAE9B;;GAEG;AACH,SAAS,iBAAiB,CACxB,OAA6C;IAE7C,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAkB,CAAC;IAC3C,KAAK,MAAM,CAAC,GAAG,EAAE,KAAK,CAAC,IAAI,OAAO,EAAE,CAAC;QACnC,IAAI,YAAY,CAAC,GAAG,CAAC,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACnC,QAAQ,CAAC,GAAG,CAAC,GAAG,EAAE,aAAa,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAC,CAAC;QACtD,CAAC;IACH,CAAC;IACD,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,WAAW,CAClB,GAAW,EACX,OAAe,EACf,QAAqC;IAErC,MAAM,KAAK,GAAG,YAAY,CAAC,GAAG,CAAC,CAAC;IAChC,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACrB,OAAO,KAAK,CAAC;IACf,CAAC;IACD,MAAM,OAAO,GAAa,EAAE,CAAC;IAC7B,KAAK,MAAM,SAAS,IAAI,KAAK,EAAE,CAAC;QAC9B,MAAM,KAAK,GAAG,QAAQ,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QACtC,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;YACxB,OAAO,KAAK,CAAC;QACf,CAAC;QACD,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;IACtB,CAAC;IACD,OAAO,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,OAAO,CAAC;AACvC,CAAC;AAED;;GAEG;AACH,SAAS,WAAW,CAAC,OAA0B;IAC7C,IAAI,GAAG,GAAG,OAAO,CAAC,MAAM,CAAC;IACzB,OAAO,GAAG,GAAG,CAAC,IAAI,OAAO,CAAC,GAAG,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC;QAC1C,GAAG,EAAE,CAAC;IACR,CAAC;IACD,OAAO,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;AAC5C,CAAC;AAED;;;;;;;;;;;;GAYG;AACH,MAAM,UAAU,aAAa,CAC3B,OAAmC;IAEnC,MAAM,OAAO,GAAG,SAAS,CAAC,OAAO,CAAC,CAAC;IACnC,MAAM,QAAQ,GAAG,iBAAiB,CAAC,OAAO,CAAC,CAAC;IAC5C,0EAA0E;IAC1E,wEAAwE;IACxE,MAAM,OAAO,GAAG,CAAC,GAAG,OAAO,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,KAAK,CAAC,EAAE,EAAE,CACxD,UAAU,CAAC,IAAI,EAAE,KAAK,CAAC,CACxB,CAAC;IACF,MAAM,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;IAEpE,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,MAAM,WAAW,GAAa,EAAE,CAAC;IACjC,KAAK,MAAM,CAAC,GAAG,EAAE,KAAK,CAAC,IAAI,OAAO,EAAE,CAAC;QACnC,MAAM,OAAO,GAAG,aAAa,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAC;QAEjD,KAAK,CAAC,IAAI,CACR,WAAW,CAAC;YACV,WAAW,CAAC,GAAG,EAAE,OAAO,EAAE,QAAQ,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,OAAO;YAClD,KAAK,CAAC,QAAQ,CAAC,EAAE,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,aAAa,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC;YACvE,KAAK,CAAC,YAAY;YAClB,KAAK,CAAC,YAAY;gBAChB,CAAC,CAAC,gBAAgB;oBAChB,CAAC,KAAK,CAAC,cAAc,IAAI,EAAE,CAAC,CAAC,IAAI,CAAC,kBAAkB,CAAC;gBACvD,CAAC,CAAC,EAAE;YACN,CAAC,KAAK,CAAC,UAAU,IAAI,EAAE,CAAC;iBACrB,GAAG,CAAC,CAAC,SAAS,EAAE,EAAE,CAAC,aAAa,CAAC,SAAS,CAAC,CAAC;iBAC5C,IAAI,CAAC,SAAS,CAAC;SACnB,CAAC,CACH,CAAC;QACF,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC;IACpC,CAAC;IAED,OAAO;QACL,IAAI,EAAE,KAAK,CAAC,SAAS,EAAE;QACvB,OAAO,EAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC;QACzB,WAAW,EAAE,cAAc,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC,SAAS,EAAE;KAC3D,CAAC;AACJ,CAAC"}

package/dist/dictionary/assertions.d.ts

@@ -1,62 +1,10 @@
-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.
+ * Running the build assertions.
  */
-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>;
-}
-/**
- * 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 declare const BUILD_ASSERTIONS: readonly BuildAssertion[];
+import type { DictionaryEntry } from "./entry.js";
+export type { BuildAssertion } from "./built-dictionary.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.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"assertions.d.ts","sourceRoot":"","sources":["../../src/dictionary/assertions.ts"],"names":[],"mappings":"AAKA,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,YAAY,CAAC;AAGlD;;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;AAoCD;;;;;;;;;;;;GAYG;AACH,eAAO,MAAM,gBAAgB,EAAE,SAAS,cAAc,EA+MrD,CAAC;AAEF;;GAEG;AACH,wBAAgB,UAAU,CACxB,OAAO,EAAE,SAAS,eAAe,EAAE,GAClC,SAAS,MAAM,EAAE,CAKnB"}
+{"version":3,"file":"assertions.d.ts","sourceRoot":"","sources":["../../src/dictionary/assertions.ts"],"names":[],"mappings":"AAAA;;GAEG;AACH,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,YAAY,CAAC;AAIlD,YAAY,EAAE,cAAc,EAAE,MAAM,uBAAuB,CAAC;AAC5D,OAAO,EAAE,eAAe,EAAE,MAAM,uBAAuB,CAAC;AACxD,OAAO,EAAE,gBAAgB,EAAE,MAAM,uBAAuB,CAAC;AAEzD;;GAEG;AACH,wBAAgB,UAAU,CACxB,OAAO,EAAE,SAAS,eAAe,EAAE,GAClC,SAAS,MAAM,EAAE,CAKnB"}