@kensio/pinyinjs 1.7.4 → 1.7.5

package/dist/dictionary/artifact-format.d.ts

@@ -0,0 +1,86 @@
+/**
+ * The artifact's on-disk format: its separators, its flags, and how a reading
+ * is written as one field.
+ *
+ * Both halves of `artifact.ts` — writing the file and reading it back — are
+ * held to the same answers here, which is what stops the two drifting apart.
+ */
+import { type Syllable } from "../syllable/syllable.js";
+import type { DictionaryEntry } from "./entry.js";
+/**
+ * Separator between a line's columns.
+ */
+export declare const COLUMN = "\t";
+/**
+ * Separator between lines, matching the key index's own separator.
+ */
+export declare const LINE = "\n";
+/**
+ * Separator between the alternate readings of a character.
+ */
+export declare const ALTERNATE = ",";
+/**
+ * The flag marking a proper noun.
+ */
+export declare 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.
+ */
+export declare const NAME_BOUNDARIES: RegExp;
+export declare const BOUNDARY_SEPARATOR = ".";
+/**
+ * The boundaries a flags column states, empty where it states none.
+ */
+export declare function nameBoundariesIn(flags: string): readonly number[];
+/**
+ * A compiled dictionary, in the form it is written to disk and read back.
+ *
+ * Three parallel pieces, all indexed by the same position: search `keys` for a
+ * word, and the line at that position in `entries` describes it. Nothing is
+ * parsed on load beyond scanning for separators — see BROWSER.md, where the
+ * measurement behind that choice is recorded.
+ */
+export interface DictionaryArtifact {
+    /** Sorted keys, newline-joined: the {@link KeyIndex} blob. */
+    readonly keys: string;
+    /** One line per key, in key order. */
+    readonly entries: string;
+    /** Quantised frequencies, packed two per byte. */
+    readonly frequencies: Uint8Array;
+}
+/**
+ * Write a reading in tone-numbered notation.
+ *
+ * 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.
+ */
+export declare function encodeReading(syllables: readonly Syllable[]): string;
+/**
+ * Read a reading back, or return undefined if any syllable is malformed.
+ */
+export declare function decodeReading(text: string): readonly Syllable[] | undefined;
+/**
+ * Order two keys the way {@link KeyIndex} requires them: by UTF-16 code unit.
+ */
+export declare function byCodeUnit(left: string, right: string): number;
+/**
+ * Every 繁體 spelling an entry claims a key for.
+ */
+export declare function traditionalForms(entry: DictionaryEntry): readonly string[];
+//# sourceMappingURL=artifact-format.d.ts.map

package/dist/dictionary/artifact-format.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"artifact-format.d.ts","sourceRoot":"","sources":["../../src/dictionary/artifact-format.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,EAEL,KAAK,QAAQ,EAEd,MAAM,yBAAyB,CAAC;AACjC,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,YAAY,CAAC;AAElD;;GAEG;AACH,eAAO,MAAM,MAAM,OAAO,CAAC;AAE3B;;GAEG;AACH,eAAO,MAAM,IAAI,OAAO,CAAC;AAEzB;;GAEG;AACH,eAAO,MAAM,SAAS,MAAM,CAAC;AAE7B;;GAEG;AACH,eAAO,MAAM,gBAAgB,MAAM,CAAC;AAEpC;;;;;;;;;;;;;;;;;GAiBG;AACH,eAAO,MAAM,eAAe,QAA6B,CAAC;AAE1D,eAAO,MAAM,kBAAkB,MAAM,CAAC;AAEtC;;GAEG;AACH,wBAAgB,gBAAgB,CAAC,KAAK,EAAE,MAAM,GAAG,SAAS,MAAM,EAAE,CAGjE;AAED;;;;;;;GAOG;AACH,MAAM,WAAW,kBAAkB;IACjC,8DAA8D;IAC9D,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,sCAAsC;IACtC,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;IACzB,kDAAkD;IAClD,QAAQ,CAAC,WAAW,EAAE,UAAU,CAAC;CAClC;AAED;;;;;;GAMG;AACH,wBAAgB,aAAa,CAAC,SAAS,EAAE,SAAS,QAAQ,EAAE,GAAG,MAAM,CAIpE;AAED;;GAEG;AACH,wBAAgB,aAAa,CAAC,IAAI,EAAE,MAAM,GAAG,SAAS,QAAQ,EAAE,GAAG,SAAS,CAQ3E;AAED;;GAEG;AACH,wBAAgB,UAAU,CAAC,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,GAAG,MAAM,CAK9D;AAED;;GAEG;AACH,wBAAgB,gBAAgB,CAAC,KAAK,EAAE,eAAe,GAAG,SAAS,MAAM,EAAE,CAE1E"}

package/dist/dictionary/artifact-format.js

@@ -0,0 +1,91 @@
+/**
+ * The artifact's on-disk format: its separators, its flags, and how a reading
+ * is written as one field.
+ *
+ * Both halves of `artifact.ts` — writing the file and reading it back — are
+ * held to the same answers here, which is what stops the two drifting apart.
+ */
+import { readSyllable, writeSyllable, } from "../syllable/syllable.js";
+/**
+ * Separator between a line's columns.
+ */
+export const COLUMN = "\t";
+/**
+ * Separator between lines, matching the key index's own separator.
+ */
+export const LINE = "\n";
+/**
+ * Separator between the alternate readings of a character.
+ */
+export const ALTERNATE = ",";
+/**
+ * The flag marking a proper noun.
+ */
+export 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.
+ */
+export const NAME_BOUNDARIES = /^p(?<at>\d+(?:\.\d+)*)$/u;
+export const BOUNDARY_SEPARATOR = ".";
+/**
+ * The boundaries a flags column states, empty where it states none.
+ */
+export 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.
+ *
+ * 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.
+ */
+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.
+ */
+export 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 ?? [])];
+}
+//# sourceMappingURL=artifact-format.js.map

package/dist/dictionary/artifact-format.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"artifact-format.js","sourceRoot":"","sources":["../../src/dictionary/artifact-format.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,EACL,YAAY,EAEZ,aAAa,GACd,MAAM,yBAAyB,CAAC;AAGjC;;GAEG;AACH,MAAM,CAAC,MAAM,MAAM,GAAG,IAAI,CAAC;AAE3B;;GAEG;AACH,MAAM,CAAC,MAAM,IAAI,GAAG,IAAI,CAAC;AAEzB;;GAEG;AACH,MAAM,CAAC,MAAM,SAAS,GAAG,GAAG,CAAC;AAE7B;;GAEG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAG,GAAG,CAAC;AAEpC;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG,0BAA0B,CAAC;AAE1D,MAAM,CAAC,MAAM,kBAAkB,GAAG,GAAG,CAAC;AAEtC;;GAEG;AACH,MAAM,UAAU,gBAAgB,CAAC,KAAa;IAC5C,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,MAAM,UAAU,UAAU,CAAC,IAAY,EAAE,KAAa;IACpD,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"}

package/dist/dictionary/artifact-read.d.ts

@@ -0,0 +1,34 @@
+import type { DictionaryEntry } from "./entry.js";
+import { type DictionaryArtifact } from "./artifact-format.js";
+export { decodeReading, type DictionaryArtifact, encodeReading, traditionalForms, } from "./artifact-format.js";
+/**
+ * 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 declare function readArtifact(artifact: DictionaryArtifact): readonly DictionaryEntry[];
+/**
+ * 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 declare function readingsByKey(artifact: DictionaryArtifact): ReadonlyMap<string, string>;
+/**
+ * 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 declare function findRoundTripFailure(entries: readonly DictionaryEntry[], artifact: DictionaryArtifact): string | undefined;
+//# sourceMappingURL=artifact-read.d.ts.map

package/dist/dictionary/artifact-read.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"artifact-read.d.ts","sourceRoot":"","sources":["../../src/dictionary/artifact-read.ts"],"names":[],"mappings":"AAQA,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,YAAY,CAAC;AAIlD,OAAO,EAIL,KAAK,kBAAkB,EAKxB,MAAM,sBAAsB,CAAC;AAE9B,OAAO,EACL,aAAa,EACb,KAAK,kBAAkB,EACvB,aAAa,EACb,gBAAgB,GACjB,MAAM,sBAAsB,CAAC;AAE9B;;;;;;;GAOG;AACH,wBAAgB,YAAY,CAC1B,QAAQ,EAAE,kBAAkB,GAC3B,SAAS,eAAe,EAAE,CA2D5B;AAED;;;;;;;;GAQG;AACH,wBAAgB,aAAa,CAC3B,QAAQ,EAAE,kBAAkB,GAC3B,WAAW,CAAC,MAAM,EAAE,MAAM,CAAC,CAO7B;AAED;;;;;;;;;GASG;AACH,wBAAgB,oBAAoB,CAClC,OAAO,EAAE,SAAS,eAAe,EAAE,EACnC,QAAQ,EAAE,kBAAkB,GAC3B,MAAM,GAAG,SAAS,CAkBpB"}

package/dist/dictionary/artifact-read.js

@@ -0,0 +1,114 @@
+/**
+ * Reading the dictionary artifact back.
+ *
+ * The mirror of buildArtifact, and held to it by the round-trip check at the
+ * end of the build.
+ */
+import { claimKeys } from "./artifact.js";
+import { toCharacters } from "../script/characters.js";
+import { isSameReading } from "./entry.js";
+import { FrequencyTable } from "./frequency-table.js";
+import { KeyIndex } from "./key-index.js";
+import { ALTERNATE, COLUMN, decodeReading, encodeReading, LINE, nameBoundariesIn, PROPER_NOUN_FLAG, } from "./artifact-format.js";
+export { decodeReading, encodeReading, traditionalForms, } from "./artifact-format.js";
+/**
+ * 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-read.js.map

package/dist/dictionary/artifact-read.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"artifact-read.js","sourceRoot":"","sources":["../../src/dictionary/artifact-read.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,EAAE,SAAS,EAAE,MAAM,eAAe,CAAC;AAC1C,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAEvD,OAAO,EAAE,aAAa,EAAE,MAAM,YAAY,CAAC;AAC3C,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AACtD,OAAO,EAAE,QAAQ,EAAE,MAAM,gBAAgB,CAAC;AAC1C,OAAO,EACL,SAAS,EACT,MAAM,EACN,aAAa,EAEb,aAAa,EACb,IAAI,EACJ,gBAAgB,EAChB,gBAAgB,GACjB,MAAM,sBAAsB,CAAC;AAE9B,OAAO,EACL,aAAa,EAEb,aAAa,EACb,gBAAgB,GACjB,MAAM,sBAAsB,CAAC;AAE9B;;;;;;;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"}

package/dist/dictionary/artifact.d.ts

@@ -1,37 +1,8 @@
-import { type Syllable } from "../syllable/syllable.js";
+export { claimKeys } from "./artifact-claims.js";
+export { findRoundTripFailure, readArtifact, readingsByKey, } from "./artifact-read.js";
 import type { DictionaryEntry } from "./entry.js";
-/**
- * A compiled dictionary, in the form it is written to disk and read back.
- *
- * Three parallel pieces, all indexed by the same position: search `keys` for a
- * word, and the line at that position in `entries` describes it. Nothing is
- * parsed on load beyond scanning for separators — see BROWSER.md, where the
- * measurement behind that choice is recorded.
- */
-export interface DictionaryArtifact {
-    /** Sorted keys, newline-joined: the {@link KeyIndex} blob. */
-    readonly keys: string;
-    /** One line per key, in key order. */
-    readonly entries: string;
-    /** Quantised frequencies, packed two per byte. */
-    readonly frequencies: Uint8Array;
-}
-/**
- * Write a reading in tone-numbered notation.
- *
- * 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.
- */
-export declare function encodeReading(syllables: readonly Syllable[]): string;
-/**
- * Read a reading back, or return undefined if any syllable is malformed.
- */
-export declare function decodeReading(text: string): readonly Syllable[] | undefined;
-/**
- * Every 繁體 spelling an entry claims a key for.
- */
-export declare function traditionalForms(entry: DictionaryEntry): readonly string[];
+import { type DictionaryArtifact } from "./artifact-format.js";
+export { decodeReading, type DictionaryArtifact, encodeReading, traditionalForms, } from "./artifact-format.js";
 /**
  * Compile merged entries into the artifact written to `data/`.
  *
@@ -46,34 +17,4 @@ export declare function traditionalForms(entry: DictionaryEntry): readonly strin
  * without the conversion path paying for it.
  */
 export declare function buildArtifact(entries: readonly DictionaryEntry[]): DictionaryArtifact;
-/**
- * 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 declare function readArtifact(artifact: DictionaryArtifact): readonly DictionaryEntry[];
-/**
- * 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 declare function readingsByKey(artifact: DictionaryArtifact): ReadonlyMap<string, string>;
-/**
- * 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 declare function findRoundTripFailure(entries: readonly DictionaryEntry[], artifact: DictionaryArtifact): string | undefined;
 //# sourceMappingURL=artifact.d.ts.map

package/dist/dictionary/artifact.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"artifact.d.ts","sourceRoot":"","sources":["../../src/dictionary/artifact.ts"],"names":[],"mappings":"AACA,OAAO,EAEL,KAAK,QAAQ,EAEd,MAAM,yBAAyB,CAAC;AACjC,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,YAAY,CAAC;AAuDlD;;;;;;;GAOG;AACH,MAAM,WAAW,kBAAkB;IACjC,8DAA8D;IAC9D,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,sCAAsC;IACtC,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;IACzB,kDAAkD;IAClD,QAAQ,CAAC,WAAW,EAAE,UAAU,CAAC;CAClC;AAED;;;;;;GAMG;AACH,wBAAgB,aAAa,CAAC,SAAS,EAAE,SAAS,QAAQ,EAAE,GAAG,MAAM,CAIpE;AAED;;GAEG;AACH,wBAAgB,aAAa,CAAC,IAAI,EAAE,MAAM,GAAG,SAAS,QAAQ,EAAE,GAAG,SAAS,CAQ3E;AAYD;;GAEG;AACH,wBAAgB,gBAAgB,CAAC,KAAK,EAAE,eAAe,GAAG,SAAS,MAAM,EAAE,CAE1E;AA+GD;;;;;;;;;;;;GAYG;AACH,wBAAgB,aAAa,CAC3B,OAAO,EAAE,SAAS,eAAe,EAAE,GAClC,kBAAkB,CAqCpB;AAED;;;;;;;GAOG;AACH,wBAAgB,YAAY,CAC1B,QAAQ,EAAE,kBAAkB,GAC3B,SAAS,eAAe,EAAE,CA2D5B;AAED;;;;;;;;GAQG;AACH,wBAAgB,aAAa,CAC3B,QAAQ,EAAE,kBAAkB,GAC3B,WAAW,CAAC,MAAM,EAAE,MAAM,CAAC,CAO7B;AAED;;;;;;;;;GASG;AACH,wBAAgB,oBAAoB,CAClC,OAAO,EAAE,SAAS,eAAe,EAAE,EACnC,QAAQ,EAAE,kBAAkB,GAC3B,MAAM,GAAG,SAAS,CAkBpB"}
+{"version":3,"file":"artifact.d.ts","sourceRoot":"","sources":["../../src/dictionary/artifact.ts"],"names":[],"mappings":"AAQA,OAAO,EAAE,SAAS,EAAE,MAAM,sBAAsB,CAAC;AACjD,OAAO,EACL,oBAAoB,EACpB,YAAY,EACZ,aAAa,GACd,MAAM,oBAAoB,CAAC;AAC5B,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,YAAY,CAAC;AAIlD,OAAO,EAKL,KAAK,kBAAkB,EAIxB,MAAM,sBAAsB,CAAC;AAE9B,OAAO,EACL,aAAa,EACb,KAAK,kBAAkB,EACvB,aAAa,EACb,gBAAgB,GACjB,MAAM,sBAAsB,CAAC;AAwD9B;;;;;;;;;;;;GAYG;AACH,wBAAgB,aAAa,CAC3B,OAAO,EAAE,SAAS,eAAe,EAAE,GAClC,kBAAkB,CAqCpB"}