@kensio/pinyinjs 1.7.4 → 1.7.6

package/dist/decode/hint-lattice.js

@@ -0,0 +1,154 @@
+import { READING_CHARGE } from "./lattice.js";
+/**
+ * What an edge a hint brought with it costs, in both decodes.
+ *
+ * Zero, because a caller naming a reading is the strongest evidence the decode
+ * has. The other branch of {@link applyHints} says so already by overwriting a
+ * dictionary entry's reading where a hint names the same stretch, and an edge
+ * with no entry behind it has to say the same thing through its cost.
+ *
+ * It was {@link import("./lattice-types.js").UNKNOWN_COST} until
+ * {@link READING_CHARGE} came down, which worked only while one edge beat two
+ * whatever they cost: 玩儿 hinted as `wánr` went back to `wán ér` the moment
+ * frequency could speak.
+ */
+const HINT_COST = 0;
+/**
+ * The same hints with every position moved along by an offset.
+ *
+ * Word hints match on what is written and so are carried over untouched; only
+ * the positional ones are counted from somewhere. Used where a run is decoded
+ * with context in front of it, which moves the run's own characters up the
+ * lattice by the length of that context.
+ */
+export function shiftHints(hints, by) {
+    return {
+        words: hints.words,
+        at: new Map([...hints.at].map(([at, syllable]) => [at + by, syllable])),
+    };
+}
+/**
+ * The hints that fall inside a run, with their positions made relative to it.
+ *
+ * A caller counts positions from the start of the whole text, while a run is
+ * decoded on its own, so the positions have to be moved to where the run put
+ * them — and the ones naming a character in some other run dropped, rather than
+ * landing on whatever happens to sit at that index here.
+ */
+export function hintsWithin(hints, from, length) {
+    const at = new Map();
+    for (const [position, syllable] of hints.at) {
+        if (position >= from && position < from + length) {
+            at.set(position - from, syllable);
+        }
+    }
+    return { words: hints.words, at };
+}
+/**
+ * Whether a hint's characters are what stands at a position.
+ */
+function matchesAt(hint, characters, at) {
+    return hint.characters.every((character, index) => characters[at + index] === character);
+}
+/**
+ * A caller's readings, put into a lattice that was built without them.
+ *
+ * Two kinds of hint with deliberately different reach, which is the whole of
+ * the semantics:
+ *
+ * A **word** hint is an assertion about the text it names, so it rewrites the
+ * edge spanning exactly those characters and supplies the single-character
+ * edges underneath it. It says nothing about a *longer* word that happens to
+ * contain it — a bare 长 hint leaves 校长 as `xiàozhǎng`, because the dictionary
+ * knowing 校长 is better evidence about that stretch than a caller's remark
+ * about one of its characters. That is what makes a corrections table safe to
+ * accumulate: entries do not reach into words nobody was thinking about.
+ *
+ * A **positional** hint is an assertion about one character of one text, and it
+ * is the escape hatch precisely because nothing outranks it: it rewrites every
+ * edge covering that position, the enclosing word included. Only a reading of
+ * one syllable per character can be rewritten this way; a 儿化 edge is a single
+ * claim about its whole span and there is no one syllable in it to replace.
+ *
+ * Nothing is invented that a caller did not write, and nothing is dropped: an
+ * edge whose reading no hint touches comes through as it was.
+ */
+export function applyHints(lattice, hints) {
+    const { characters } = lattice;
+    /** The word hint covering a position, and how far into it that position is. */
+    const wordAt = (at) => {
+        for (const hint of hints.words) {
+            if (hint.reading.length !== hint.characters.length) {
+                continue;
+            }
+            for (let held = 0; held < hint.characters.length; held++) {
+                if (at - held >= 0 && matchesAt(hint, characters, at - held)) {
+                    return { hint, held };
+                }
+            }
+        }
+        return undefined;
+    };
+    const rewrite = (edge) => {
+        const span = edge.to - edge.from;
+        // A word hint naming exactly this stretch replaces its reading outright,
+        // which is the only way a hint can reach a word the dictionary already has
+        // an opinion about: 银行 asserted as `yín xíng` has to beat `yínháng`.
+        const exact = hints.words.find((hint) => hint.characters.length === span &&
+            matchesAt(hint, characters, edge.from));
+        let reading = exact === undefined ? edge.reading : exact.reading;
+        // Everything below needs one syllable per character to have a syllable to
+        // put anywhere.
+        if (reading.length === span) {
+            reading = reading.map((syllable, index) => {
+                const at = edge.from + index;
+                const positional = hints.at.get(at);
+                if (positional !== undefined) {
+                    return positional;
+                }
+                if (span > 1) {
+                    return syllable;
+                }
+                // Single-character edges also take what a longer word hint says about
+                // this position, so that the split path reads the same as the whole.
+                const found = wordAt(at);
+                return found?.hint.reading[found.held] ?? syllable;
+            });
+        }
+        return reading === edge.reading ? edge : { ...edge, reading };
+    };
+    return {
+        characters,
+        edges: lattice.edges.map((edges, at) => {
+            const rewritten = edges.map((edge) => rewrite(edge));
+            // A hint spanning characters no edge covers has nowhere to be written, so
+            // it brings its own edge: 玩儿 as `wánr` where the dictionary lacks it.
+            const spanning = hints.words.find((hint) => hint.reading.length !== hint.characters.length &&
+                matchesAt(hint, characters, at) &&
+                !rewritten.some((edge) => edge.to === at + hint.characters.length));
+            if (spanning === undefined) {
+                return rewritten;
+            }
+            const to = at + spanning.characters.length;
+            return [
+                ...rewritten,
+                {
+                    from: at,
+                    to,
+                    text: spanning.characters.join(""),
+                    reading: spanning.reading,
+                    cost: HINT_COST,
+                    readingCost: HINT_COST + READING_CHARGE,
+                    isProperNoun: false,
+                    partOfSpeech: "",
+                    // No dictionary entry backs this one — that is the whole reason it
+                    // had to be brought along — so it reports itself as the fallback it
+                    // is. A hint written over an edge that *was* backed keeps that edge's
+                    // own answer, since the entry still covers those characters.
+                    isKnown: false,
+                },
+            ];
+        }),
+    };
+}
+//# sourceMappingURL=hint-lattice.js.map

package/dist/decode/hint-lattice.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"hint-lattice.js","sourceRoot":"","sources":["../../src/decode/hint-lattice.ts"],"names":[],"mappings":"AAQA,OAAO,EAAkC,cAAc,EAAE,MAAM,cAAc,CAAC;AAG9E;;;;;;;;;;;;GAYG;AACH,MAAM,SAAS,GAAG,CAAC,CAAC;AAEpB;;;;;;;GAOG;AACH,MAAM,UAAU,UAAU,CAAC,KAAoB,EAAE,EAAU;IACzD,OAAO;QACL,KAAK,EAAE,KAAK,CAAC,KAAK;QAClB,EAAE,EAAE,IAAI,GAAG,CAAC,CAAC,GAAG,KAAK,CAAC,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,QAAQ,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,GAAG,EAAE,EAAE,QAAQ,CAAC,CAAC,CAAC;KACxE,CAAC;AACJ,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,WAAW,CACzB,KAAoB,EACpB,IAAY,EACZ,MAAc;IAEd,MAAM,EAAE,GAAG,IAAI,GAAG,EAAoB,CAAC;IACvC,KAAK,MAAM,CAAC,QAAQ,EAAE,QAAQ,CAAC,IAAI,KAAK,CAAC,EAAE,EAAE,CAAC;QAC5C,IAAI,QAAQ,IAAI,IAAI,IAAI,QAAQ,GAAG,IAAI,GAAG,MAAM,EAAE,CAAC;YACjD,EAAE,CAAC,GAAG,CAAC,QAAQ,GAAG,IAAI,EAAE,QAAQ,CAAC,CAAC;QACpC,CAAC;IACH,CAAC;IACD,OAAO,EAAE,KAAK,EAAE,KAAK,CAAC,KAAK,EAAE,EAAE,EAAE,CAAC;AACpC,CAAC;AAED;;GAEG;AACH,SAAS,SAAS,CAChB,IAAc,EACd,UAA6B,EAC7B,EAAU;IAEV,OAAO,IAAI,CAAC,UAAU,CAAC,KAAK,CAC1B,CAAC,SAAS,EAAE,KAAK,EAAE,EAAE,CAAC,UAAU,CAAC,EAAE,GAAG,KAAK,CAAC,KAAK,SAAS,CAC3D,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;GAsBG;AACH,MAAM,UAAU,UAAU,CAAC,OAAgB,EAAE,KAAoB;IAC/D,MAAM,EAAE,UAAU,EAAE,GAAG,OAAO,CAAC;IAE/B,+EAA+E;IAC/E,MAAM,MAAM,GAAG,CAAC,EAAU,EAAgD,EAAE;QAC1E,KAAK,MAAM,IAAI,IAAI,KAAK,CAAC,KAAK,EAAE,CAAC;YAC/B,IAAI,IAAI,CAAC,OAAO,CAAC,MAAM,KAAK,IAAI,CAAC,UAAU,CAAC,MAAM,EAAE,CAAC;gBACnD,SAAS;YACX,CAAC;YACD,KAAK,IAAI,IAAI,GAAG,CAAC,EAAE,IAAI,GAAG,IAAI,CAAC,UAAU,CAAC,MAAM,EAAE,IAAI,EAAE,EAAE,CAAC;gBACzD,IAAI,EAAE,GAAG,IAAI,IAAI,CAAC,IAAI,SAAS,CAAC,IAAI,EAAE,UAAU,EAAE,EAAE,GAAG,IAAI,CAAC,EAAE,CAAC;oBAC7D,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC;gBACxB,CAAC;YACH,CAAC;QACH,CAAC;QACD,OAAO,SAAS,CAAC;IACnB,CAAC,CAAC;IAEF,MAAM,OAAO,GAAG,CAAC,IAAiB,EAAe,EAAE;QACjD,MAAM,IAAI,GAAG,IAAI,CAAC,EAAE,GAAG,IAAI,CAAC,IAAI,CAAC;QAEjC,yEAAyE;QACzE,2EAA2E;QAC3E,qEAAqE;QACrE,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC,IAAI,CAC5B,CAAC,IAAI,EAAE,EAAE,CACP,IAAI,CAAC,UAAU,CAAC,MAAM,KAAK,IAAI;YAC/B,SAAS,CAAC,IAAI,EAAE,UAAU,EAAE,IAAI,CAAC,IAAI,CAAC,CACzC,CAAC;QACF,IAAI,OAAO,GAAG,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC;QAEjE,0EAA0E;QAC1E,gBAAgB;QAChB,IAAI,OAAO,CAAC,MAAM,KAAK,IAAI,EAAE,CAAC;YAC5B,OAAO,GAAG,OAAO,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,KAAK,EAAE,EAAE;gBACxC,MAAM,EAAE,GAAG,IAAI,CAAC,IAAI,GAAG,KAAK,CAAC;gBAC7B,MAAM,UAAU,GAAG,KAAK,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;gBACpC,IAAI,UAAU,KAAK,SAAS,EAAE,CAAC;oBAC7B,OAAO,UAAU,CAAC;gBACpB,CAAC;gBACD,IAAI,IAAI,GAAG,CAAC,EAAE,CAAC;oBACb,OAAO,QAAQ,CAAC;gBAClB,CAAC;gBACD,sEAAsE;gBACtE,qEAAqE;gBACrE,MAAM,KAAK,GAAG,MAAM,CAAC,EAAE,CAAC,CAAC;gBACzB,OAAO,KAAK,EAAE,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,QAAQ,CAAC;YACrD,CAAC,CAAC,CAAC;QACL,CAAC;QAED,OAAO,OAAO,KAAK,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,GAAG,IAAI,EAAE,OAAO,EAAE,CAAC;IAChE,CAAC,CAAC;IAEF,OAAO;QACL,UAAU;QACV,KAAK,EAAE,OAAO,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,EAAE,EAAE;YACrC,MAAM,SAAS,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC;YACrD,0EAA0E;YAC1E,sEAAsE;YACtE,MAAM,QAAQ,GAAG,KAAK,CAAC,KAAK,CAAC,IAAI,CAC/B,CAAC,IAAI,EAAE,EAAE,CACP,IAAI,CAAC,OAAO,CAAC,MAAM,KAAK,IAAI,CAAC,UAAU,CAAC,MAAM;gBAC9C,SAAS,CAAC,IAAI,EAAE,UAAU,EAAE,EAAE,CAAC;gBAC/B,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,GAAG,IAAI,CAAC,UAAU,CAAC,MAAM,CAAC,CACrE,CAAC;YACF,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;gBAC3B,OAAO,SAAS,CAAC;YACnB,CAAC;YACD,MAAM,EAAE,GAAG,EAAE,GAAG,QAAQ,CAAC,UAAU,CAAC,MAAM,CAAC;YAC3C,OAAO;gBACL,GAAG,SAAS;gBACZ;oBACE,IAAI,EAAE,EAAE;oBACR,EAAE;oBACF,IAAI,EAAE,QAAQ,CAAC,UAAU,CAAC,IAAI,CAAC,EAAE,CAAC;oBAClC,OAAO,EAAE,QAAQ,CAAC,OAAO;oBACzB,IAAI,EAAE,SAAS;oBACf,WAAW,EAAE,SAAS,GAAG,cAAc;oBACvC,YAAY,EAAE,KAAK;oBACnB,YAAY,EAAE,EAAE;oBAChB,mEAAmE;oBACnE,oEAAoE;oBACpE,sEAAsE;oBACtE,6DAA6D;oBAC7D,OAAO,EAAE,KAAK;iBACf;aACF,CAAC;QACJ,CAAC,CAAC;KACH,CAAC;AACJ,CAAC"}

package/dist/decode/hints.d.ts

@@ -1,5 +1,4 @@
 import { type Syllable } from "../syllable/syllable.js";
-import { type Lattice } from "./lattice.js";
 /**
  * A reading a caller asserts for a word, wherever that word occurs.
  *
@@ -68,7 +67,7 @@ export interface ResolvedHints {
 /**
  * One word hint, with its characters and reading already taken apart.
  */
-interface WordHint {
+export interface WordHint {
     readonly characters: readonly string[];
     readonly reading: readonly Syllable[];
 }
@@ -92,47 +91,5 @@ export declare function readHintReading(reading: string, subject: string): reado
  * longest-wins rule the dictionary itself is read by.
  */
 export declare function resolveHints(hints: ReadingHints): ResolvedHints;
-/**
- * The same hints with every position moved along by an offset.
- *
- * Word hints match on what is written and so are carried over untouched; only
- * the positional ones are counted from somewhere. Used where a run is decoded
- * with context in front of it, which moves the run's own characters up the
- * lattice by the length of that context.
- */
-export declare function shiftHints(hints: ResolvedHints, by: number): ResolvedHints;
-/**
- * The hints that fall inside a run, with their positions made relative to it.
- *
- * A caller counts positions from the start of the whole text, while a run is
- * decoded on its own, so the positions have to be moved to where the run put
- * them — and the ones naming a character in some other run dropped, rather than
- * landing on whatever happens to sit at that index here.
- */
-export declare function hintsWithin(hints: ResolvedHints, from: number, length: number): ResolvedHints;
-/**
- * A caller's readings, put into a lattice that was built without them.
- *
- * Two kinds of hint with deliberately different reach, which is the whole of
- * the semantics:
- *
- * A **word** hint is an assertion about the text it names, so it rewrites the
- * edge spanning exactly those characters and supplies the single-character
- * edges underneath it. It says nothing about a *longer* word that happens to
- * contain it — a bare 长 hint leaves 校长 as `xiàozhǎng`, because the dictionary
- * knowing 校长 is better evidence about that stretch than a caller's remark
- * about one of its characters. That is what makes a corrections table safe to
- * accumulate: entries do not reach into words nobody was thinking about.
- *
- * A **positional** hint is an assertion about one character of one text, and it
- * is the escape hatch precisely because nothing outranks it: it rewrites every
- * edge covering that position, the enclosing word included. Only a reading of
- * one syllable per character can be rewritten this way; a 儿化 edge is a single
- * claim about its whole span and there is no one syllable in it to replace.
- *
- * Nothing is invented that a caller did not write, and nothing is dropped: an
- * edge whose reading no hint touches comes through as it was.
- */
-export declare function applyHints(lattice: Lattice, hints: ResolvedHints): Lattice;
-export {};
+export { applyHints, hintsWithin, shiftHints } from "./hint-lattice.js";
 //# sourceMappingURL=hints.d.ts.map

package/dist/decode/hints.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"hints.d.ts","sourceRoot":"","sources":["../../src/decode/hints.ts"],"names":[],"mappings":"AACA,OAAO,EAAgB,KAAK,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAEtE,OAAO,EACL,KAAK,OAAO,EAIb,MAAM,cAAc,CAAC;AAEtB;;;;;;;GAOG;AACH,MAAM,WAAW,WAAW;IAC1B,mDAAmD;IACnD,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,mDAAmD;IACnD,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;CAC1B;AAED;;;;;;;GAOG;AACH,MAAM,WAAW,iBAAiB;IAChC;;;;;OAKG;IACH,QAAQ,CAAC,EAAE,EAAE,MAAM,CAAC;IACpB,+DAA+D;IAC/D,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;CAC1B;AAED;;GAEG;AACH,MAAM,MAAM,WAAW,GAAG,WAAW,GAAG,iBAAiB,CAAC;AAE1D;;;;;;;;;;;GAWG;AACH,MAAM,MAAM,YAAY,GACpB,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,GAChC,SAAS,WAAW,EAAE,CAAC;AAE3B;;GAEG;AACH,MAAM,WAAW,aAAa;IAC5B,4EAA4E;IAC5E,QAAQ,CAAC,KAAK,EAAE,SAAS,QAAQ,EAAE,CAAC;IACpC;;;;;;OAMG;IACH,QAAQ,CAAC,EAAE,EAAE,WAAW,CAAC,MAAM,EAAE,QAAQ,CAAC,CAAC;CAC5C;AAED;;GAEG;AACH,UAAU,QAAQ;IAChB,QAAQ,CAAC,UAAU,EAAE,SAAS,MAAM,EAAE,CAAC;IACvC,QAAQ,CAAC,OAAO,EAAE,SAAS,QAAQ,EAAE,CAAC;CACvC;AASD;;;;;;;;;;GAUG;AACH,wBAAgB,eAAe,CAC7B,OAAO,EAAE,MAAM,EACf,OAAO,EAAE,MAAM,GACd,SAAS,QAAQ,EAAE,CAYrB;AAED;;;;;;GAMG;AACH,wBAAgB,YAAY,CAAC,KAAK,EAAE,YAAY,GAAG,aAAa,CAwC/D;AAED;;;;;;;GAOG;AACH,wBAAgB,UAAU,CAAC,KAAK,EAAE,aAAa,EAAE,EAAE,EAAE,MAAM,GAAG,aAAa,CAK1E;AAED;;;;;;;GAOG;AACH,wBAAgB,WAAW,CACzB,KAAK,EAAE,aAAa,EACpB,IAAI,EAAE,MAAM,EACZ,MAAM,EAAE,MAAM,GACb,aAAa,CAQf;AAeD;;;;;;;;;;;;;;;;;;;;;;GAsBG;AACH,wBAAgB,UAAU,CAAC,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,aAAa,GAAG,OAAO,CAyF1E"}
+{"version":3,"file":"hints.d.ts","sourceRoot":"","sources":["../../src/decode/hints.ts"],"names":[],"mappings":"AACA,OAAO,EAAgB,KAAK,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAGtE;;;;;;;GAOG;AACH,MAAM,WAAW,WAAW;IAC1B,mDAAmD;IACnD,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,mDAAmD;IACnD,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;CAC1B;AAED;;;;;;;GAOG;AACH,MAAM,WAAW,iBAAiB;IAChC;;;;;OAKG;IACH,QAAQ,CAAC,EAAE,EAAE,MAAM,CAAC;IACpB,+DAA+D;IAC/D,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;CAC1B;AAED;;GAEG;AACH,MAAM,MAAM,WAAW,GAAG,WAAW,GAAG,iBAAiB,CAAC;AAE1D;;;;;;;;;;;GAWG;AACH,MAAM,MAAM,YAAY,GACpB,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,GAChC,SAAS,WAAW,EAAE,CAAC;AAE3B;;GAEG;AACH,MAAM,WAAW,aAAa;IAC5B,4EAA4E;IAC5E,QAAQ,CAAC,KAAK,EAAE,SAAS,QAAQ,EAAE,CAAC;IACpC;;;;;;OAMG;IACH,QAAQ,CAAC,EAAE,EAAE,WAAW,CAAC,MAAM,EAAE,QAAQ,CAAC,CAAC;CAC5C;AAED;;GAEG;AACH,MAAM,WAAW,QAAQ;IACvB,QAAQ,CAAC,UAAU,EAAE,SAAS,MAAM,EAAE,CAAC;IACvC,QAAQ,CAAC,OAAO,EAAE,SAAS,QAAQ,EAAE,CAAC;CACvC;AASD;;;;;;;;;;GAUG;AACH,wBAAgB,eAAe,CAC7B,OAAO,EAAE,MAAM,EACf,OAAO,EAAE,MAAM,GACd,SAAS,QAAQ,EAAE,CAYrB;AAED;;;;;;GAMG;AACH,wBAAgB,YAAY,CAAC,KAAK,EAAE,YAAY,GAAG,aAAa,CAwC/D;AAED,OAAO,EAAE,UAAU,EAAE,WAAW,EAAE,UAAU,EAAE,MAAM,mBAAmB,CAAC"}

package/dist/decode/hints.js

@@ -1,7 +1,6 @@
 import { toCharacters } from "../script/characters.js";
 import { readSyllable } from "../syllable/syllable.js";
 import { NEUTRAL_TONE } from "../tone/tone.js";
-import { READING_CHARGE, UNKNOWN_COST, } from "./lattice.js";
 /**
  * Whether a hint names a position rather than a word.
  */
@@ -73,142 +72,5 @@ export function resolveHints(hints) {
     words.sort((left, right) => right.characters.length - left.characters.length);
     return { words, at };
 }
-/**
- * The same hints with every position moved along by an offset.
- *
- * Word hints match on what is written and so are carried over untouched; only
- * the positional ones are counted from somewhere. Used where a run is decoded
- * with context in front of it, which moves the run's own characters up the
- * lattice by the length of that context.
- */
-export function shiftHints(hints, by) {
-    return {
-        words: hints.words,
-        at: new Map([...hints.at].map(([at, syllable]) => [at + by, syllable])),
-    };
-}
-/**
- * The hints that fall inside a run, with their positions made relative to it.
- *
- * A caller counts positions from the start of the whole text, while a run is
- * decoded on its own, so the positions have to be moved to where the run put
- * them — and the ones naming a character in some other run dropped, rather than
- * landing on whatever happens to sit at that index here.
- */
-export function hintsWithin(hints, from, length) {
-    const at = new Map();
-    for (const [position, syllable] of hints.at) {
-        if (position >= from && position < from + length) {
-            at.set(position - from, syllable);
-        }
-    }
-    return { words: hints.words, at };
-}
-/**
- * Whether a hint's characters are what stands at a position.
- */
-function matchesAt(hint, characters, at) {
-    return hint.characters.every((character, index) => characters[at + index] === character);
-}
-/**
- * A caller's readings, put into a lattice that was built without them.
- *
- * Two kinds of hint with deliberately different reach, which is the whole of
- * the semantics:
- *
- * A **word** hint is an assertion about the text it names, so it rewrites the
- * edge spanning exactly those characters and supplies the single-character
- * edges underneath it. It says nothing about a *longer* word that happens to
- * contain it — a bare 长 hint leaves 校长 as `xiàozhǎng`, because the dictionary
- * knowing 校长 is better evidence about that stretch than a caller's remark
- * about one of its characters. That is what makes a corrections table safe to
- * accumulate: entries do not reach into words nobody was thinking about.
- *
- * A **positional** hint is an assertion about one character of one text, and it
- * is the escape hatch precisely because nothing outranks it: it rewrites every
- * edge covering that position, the enclosing word included. Only a reading of
- * one syllable per character can be rewritten this way; a 儿化 edge is a single
- * claim about its whole span and there is no one syllable in it to replace.
- *
- * Nothing is invented that a caller did not write, and nothing is dropped: an
- * edge whose reading no hint touches comes through as it was.
- */
-export function applyHints(lattice, hints) {
-    const { characters } = lattice;
-    /** The word hint covering a position, and how far into it that position is. */
-    const wordAt = (at) => {
-        for (const hint of hints.words) {
-            if (hint.reading.length !== hint.characters.length) {
-                continue;
-            }
-            for (let held = 0; held < hint.characters.length; held++) {
-                if (at - held >= 0 && matchesAt(hint, characters, at - held)) {
-                    return { hint, held };
-                }
-            }
-        }
-        return undefined;
-    };
-    const rewrite = (edge) => {
-        const span = edge.to - edge.from;
-        // A word hint naming exactly this stretch replaces its reading outright,
-        // which is the only way a hint can reach a word the dictionary already has
-        // an opinion about: 银行 asserted as `yín xíng` has to beat `yínháng`.
-        const exact = hints.words.find((hint) => hint.characters.length === span &&
-            matchesAt(hint, characters, edge.from));
-        let reading = exact === undefined ? edge.reading : exact.reading;
-        // Everything below needs one syllable per character to have a syllable to
-        // put anywhere.
-        if (reading.length === span) {
-            reading = reading.map((syllable, index) => {
-                const at = edge.from + index;
-                const positional = hints.at.get(at);
-                if (positional !== undefined) {
-                    return positional;
-                }
-                if (span > 1) {
-                    return syllable;
-                }
-                // Single-character edges also take what a longer word hint says about
-                // this position, so that the split path reads the same as the whole.
-                const found = wordAt(at);
-                return found?.hint.reading[found.held] ?? syllable;
-            });
-        }
-        return reading === edge.reading ? edge : { ...edge, reading };
-    };
-    return {
-        characters,
-        edges: lattice.edges.map((edges, at) => {
-            const rewritten = edges.map((edge) => rewrite(edge));
-            // A hint spanning characters no edge covers has nowhere to be written, so
-            // it brings its own edge: 玩儿 as `wánr` where the dictionary lacks it.
-            const spanning = hints.words.find((hint) => hint.reading.length !== hint.characters.length &&
-                matchesAt(hint, characters, at) &&
-                !rewritten.some((edge) => edge.to === at + hint.characters.length));
-            if (spanning === undefined) {
-                return rewritten;
-            }
-            const to = at + spanning.characters.length;
-            return [
-                ...rewritten,
-                {
-                    from: at,
-                    to,
-                    text: spanning.characters.join(""),
-                    reading: spanning.reading,
-                    cost: UNKNOWN_COST,
-                    readingCost: UNKNOWN_COST + READING_CHARGE,
-                    isProperNoun: false,
-                    partOfSpeech: "",
-                    // No dictionary entry backs this one — that is the whole reason it
-                    // had to be brought along — so it reports itself as the fallback it
-                    // is. A hint written over an edge that *was* backed keeps that edge's
-                    // own answer, since the entry still covers those characters.
-                    isKnown: false,
-                },
-            ];
-        }),
-    };
-}
+export { applyHints, hintsWithin, shiftHints } from "./hint-lattice.js";
 //# sourceMappingURL=hints.js.map

package/dist/decode/hints.js.map

@@ -1 +1 @@
-{"version":3,"file":"hints.js","sourceRoot":"","sources":["../../src/decode/hints.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AACvD,OAAO,EAAE,YAAY,EAAiB,MAAM,yBAAyB,CAAC;AACtE,OAAO,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAC/C,OAAO,EAGL,cAAc,EACd,YAAY,GACb,MAAM,cAAc,CAAC;AAkFtB;;GAEG;AACH,SAAS,YAAY,CAAC,IAAiB;IACrC,OAAO,IAAI,IAAI,IAAI,CAAC;AACtB,CAAC;AAED;;;;;;;;;;GAUG;AACH,MAAM,UAAU,eAAe,CAC7B,OAAe,EACf,OAAe;IAEf,MAAM,SAAS,GAAG,OAAO,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,KAAK,EAAE,CAAC,CAAC;IACxE,IAAI,SAAS,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC3B,MAAM,IAAI,KAAK,CAAC,oBAAoB,OAAO,WAAW,CAAC,CAAC;IAC1D,CAAC;IACD,OAAO,SAAS,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE;QAC7B,MAAM,QAAQ,GAAG,YAAY,CAAC,KAAK,CAAC,CAAC;QACrC,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;YAC3B,MAAM,IAAI,KAAK,CAAC,oBAAoB,OAAO,mBAAmB,KAAK,EAAE,CAAC,CAAC;QACzE,CAAC;QACD,OAAO,EAAE,GAAG,QAAQ,EAAE,IAAI,EAAE,QAAQ,CAAC,IAAI,IAAI,YAAY,EAAE,CAAC;IAC9D,CAAC,CAAC,CAAC;AACL,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,YAAY,CAAC,KAAmB;IAC9C,MAAM,IAAI,GAA2B,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC;QACvD,CAAC,CAAC,KAAK;QACP,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,KAA+B,CAAC,CAAC,GAAG,CACjD,CAAC,CAAC,IAAI,EAAE,OAAO,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,CAAC,CACzC,CAAC;IAEN,MAAM,KAAK,GAAe,EAAE,CAAC;IAC7B,MAAM,EAAE,GAAG,IAAI,GAAG,EAAoB,CAAC;IAEvC,KAAK,MAAM,IAAI,IAAI,IAAI,EAAE,CAAC;QACxB,IAAI,YAAY,CAAC,IAAI,CAAC,EAAE,CAAC;YACvB,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,IAAI,CAAC,EAAE,CAAC,IAAI,IAAI,CAAC,EAAE,GAAG,CAAC,EAAE,CAAC;gBAC9C,MAAM,IAAI,KAAK,CAAC,0CAA0C,IAAI,CAAC,EAAE,EAAE,CAAC,CAAC;YACvE,CAAC;YACD,MAAM,OAAO,GAAG,YAAY,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,CAAC;YAC9C,MAAM,SAAS,GAAG,eAAe,CAAC,IAAI,CAAC,OAAO,EAAE,OAAO,CAAC,CAAC;YACzD,sEAAsE;YACtE,0EAA0E;YAC1E,8DAA8D;YAC9D,IAAI,SAAS,CAAC,MAAM,KAAK,CAAC,IAAI,SAAS,CAAC,CAAC,CAAC,KAAK,SAAS,EAAE,CAAC;gBACzD,MAAM,IAAI,KAAK,CACb,oBAAoB,OAAO,yBAAyB,IAAI,CAAC,OAAO,EAAE,CACnE,CAAC;YACJ,CAAC;YACD,EAAE,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,EAAE,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC;YAC9B,SAAS;QACX,CAAC;QACD,MAAM,UAAU,GAAG,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAC3C,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAC5B,MAAM,IAAI,KAAK,CAAC,0BAA0B,CAAC,CAAC;QAC9C,CAAC;QACD,KAAK,CAAC,IAAI,CAAC;YACT,UAAU;YACV,OAAO,EAAE,eAAe,CAAC,IAAI,CAAC,OAAO,EAAE,IAAI,CAAC,IAAI,CAAC;SAClD,CAAC,CAAC;IACL,CAAC;IAED,KAAK,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,UAAU,CAAC,MAAM,GAAG,IAAI,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC;IAC9E,OAAO,EAAE,KAAK,EAAE,EAAE,EAAE,CAAC;AACvB,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,UAAU,CAAC,KAAoB,EAAE,EAAU;IACzD,OAAO;QACL,KAAK,EAAE,KAAK,CAAC,KAAK;QAClB,EAAE,EAAE,IAAI,GAAG,CAAC,CAAC,GAAG,KAAK,CAAC,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,QAAQ,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,GAAG,EAAE,EAAE,QAAQ,CAAC,CAAC,CAAC;KACxE,CAAC;AACJ,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,WAAW,CACzB,KAAoB,EACpB,IAAY,EACZ,MAAc;IAEd,MAAM,EAAE,GAAG,IAAI,GAAG,EAAoB,CAAC;IACvC,KAAK,MAAM,CAAC,QAAQ,EAAE,QAAQ,CAAC,IAAI,KAAK,CAAC,EAAE,EAAE,CAAC;QAC5C,IAAI,QAAQ,IAAI,IAAI,IAAI,QAAQ,GAAG,IAAI,GAAG,MAAM,EAAE,CAAC;YACjD,EAAE,CAAC,GAAG,CAAC,QAAQ,GAAG,IAAI,EAAE,QAAQ,CAAC,CAAC;QACpC,CAAC;IACH,CAAC;IACD,OAAO,EAAE,KAAK,EAAE,KAAK,CAAC,KAAK,EAAE,EAAE,EAAE,CAAC;AACpC,CAAC;AAED;;GAEG;AACH,SAAS,SAAS,CAChB,IAAc,EACd,UAA6B,EAC7B,EAAU;IAEV,OAAO,IAAI,CAAC,UAAU,CAAC,KAAK,CAC1B,CAAC,SAAS,EAAE,KAAK,EAAE,EAAE,CAAC,UAAU,CAAC,EAAE,GAAG,KAAK,CAAC,KAAK,SAAS,CAC3D,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;GAsBG;AACH,MAAM,UAAU,UAAU,CAAC,OAAgB,EAAE,KAAoB;IAC/D,MAAM,EAAE,UAAU,EAAE,GAAG,OAAO,CAAC;IAE/B,+EAA+E;IAC/E,MAAM,MAAM,GAAG,CAAC,EAAU,EAAgD,EAAE;QAC1E,KAAK,MAAM,IAAI,IAAI,KAAK,CAAC,KAAK,EAAE,CAAC;YAC/B,IAAI,IAAI,CAAC,OAAO,CAAC,MAAM,KAAK,IAAI,CAAC,UAAU,CAAC,MAAM,EAAE,CAAC;gBACnD,SAAS;YACX,CAAC;YACD,KAAK,IAAI,IAAI,GAAG,CAAC,EAAE,IAAI,GAAG,IAAI,CAAC,UAAU,CAAC,MAAM,EAAE,IAAI,EAAE,EAAE,CAAC;gBACzD,IAAI,EAAE,GAAG,IAAI,IAAI,CAAC,IAAI,SAAS,CAAC,IAAI,EAAE,UAAU,EAAE,EAAE,GAAG,IAAI,CAAC,EAAE,CAAC;oBAC7D,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC;gBACxB,CAAC;YACH,CAAC;QACH,CAAC;QACD,OAAO,SAAS,CAAC;IACnB,CAAC,CAAC;IAEF,MAAM,OAAO,GAAG,CAAC,IAAiB,EAAe,EAAE;QACjD,MAAM,IAAI,GAAG,IAAI,CAAC,EAAE,GAAG,IAAI,CAAC,IAAI,CAAC;QAEjC,yEAAyE;QACzE,2EAA2E;QAC3E,qEAAqE;QACrE,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC,IAAI,CAC5B,CAAC,IAAI,EAAE,EAAE,CACP,IAAI,CAAC,UAAU,CAAC,MAAM,KAAK,IAAI;YAC/B,SAAS,CAAC,IAAI,EAAE,UAAU,EAAE,IAAI,CAAC,IAAI,CAAC,CACzC,CAAC;QACF,IAAI,OAAO,GAAG,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC;QAEjE,0EAA0E;QAC1E,gBAAgB;QAChB,IAAI,OAAO,CAAC,MAAM,KAAK,IAAI,EAAE,CAAC;YAC5B,OAAO,GAAG,OAAO,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,KAAK,EAAE,EAAE;gBACxC,MAAM,EAAE,GAAG,IAAI,CAAC,IAAI,GAAG,KAAK,CAAC;gBAC7B,MAAM,UAAU,GAAG,KAAK,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;gBACpC,IAAI,UAAU,KAAK,SAAS,EAAE,CAAC;oBAC7B,OAAO,UAAU,CAAC;gBACpB,CAAC;gBACD,IAAI,IAAI,GAAG,CAAC,EAAE,CAAC;oBACb,OAAO,QAAQ,CAAC;gBAClB,CAAC;gBACD,sEAAsE;gBACtE,qEAAqE;gBACrE,MAAM,KAAK,GAAG,MAAM,CAAC,EAAE,CAAC,CAAC;gBACzB,OAAO,KAAK,EAAE,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,QAAQ,CAAC;YACrD,CAAC,CAAC,CAAC;QACL,CAAC;QAED,OAAO,OAAO,KAAK,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,GAAG,IAAI,EAAE,OAAO,EAAE,CAAC;IAChE,CAAC,CAAC;IAEF,OAAO;QACL,UAAU;QACV,KAAK,EAAE,OAAO,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,EAAE,EAAE;YACrC,MAAM,SAAS,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC;YACrD,0EAA0E;YAC1E,sEAAsE;YACtE,MAAM,QAAQ,GAAG,KAAK,CAAC,KAAK,CAAC,IAAI,CAC/B,CAAC,IAAI,EAAE,EAAE,CACP,IAAI,CAAC,OAAO,CAAC,MAAM,KAAK,IAAI,CAAC,UAAU,CAAC,MAAM;gBAC9C,SAAS,CAAC,IAAI,EAAE,UAAU,EAAE,EAAE,CAAC;gBAC/B,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,GAAG,IAAI,CAAC,UAAU,CAAC,MAAM,CAAC,CACrE,CAAC;YACF,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;gBAC3B,OAAO,SAAS,CAAC;YACnB,CAAC;YACD,MAAM,EAAE,GAAG,EAAE,GAAG,QAAQ,CAAC,UAAU,CAAC,MAAM,CAAC;YAC3C,OAAO;gBACL,GAAG,SAAS;gBACZ;oBACE,IAAI,EAAE,EAAE;oBACR,EAAE;oBACF,IAAI,EAAE,QAAQ,CAAC,UAAU,CAAC,IAAI,CAAC,EAAE,CAAC;oBAClC,OAAO,EAAE,QAAQ,CAAC,OAAO;oBACzB,IAAI,EAAE,YAAY;oBAClB,WAAW,EAAE,YAAY,GAAG,cAAc;oBAC1C,YAAY,EAAE,KAAK;oBACnB,YAAY,EAAE,EAAE;oBAChB,mEAAmE;oBACnE,oEAAoE;oBACpE,sEAAsE;oBACtE,6DAA6D;oBAC7D,OAAO,EAAE,KAAK;iBACf;aACF,CAAC;QACJ,CAAC,CAAC;KACH,CAAC;AACJ,CAAC"}
+{"version":3,"file":"hints.js","sourceRoot":"","sources":["../../src/decode/hints.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AACvD,OAAO,EAAE,YAAY,EAAiB,MAAM,yBAAyB,CAAC;AACtE,OAAO,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAkF/C;;GAEG;AACH,SAAS,YAAY,CAAC,IAAiB;IACrC,OAAO,IAAI,IAAI,IAAI,CAAC;AACtB,CAAC;AAED;;;;;;;;;;GAUG;AACH,MAAM,UAAU,eAAe,CAC7B,OAAe,EACf,OAAe;IAEf,MAAM,SAAS,GAAG,OAAO,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,KAAK,EAAE,CAAC,CAAC;IACxE,IAAI,SAAS,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC3B,MAAM,IAAI,KAAK,CAAC,oBAAoB,OAAO,WAAW,CAAC,CAAC;IAC1D,CAAC;IACD,OAAO,SAAS,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE;QAC7B,MAAM,QAAQ,GAAG,YAAY,CAAC,KAAK,CAAC,CAAC;QACrC,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;YAC3B,MAAM,IAAI,KAAK,CAAC,oBAAoB,OAAO,mBAAmB,KAAK,EAAE,CAAC,CAAC;QACzE,CAAC;QACD,OAAO,EAAE,GAAG,QAAQ,EAAE,IAAI,EAAE,QAAQ,CAAC,IAAI,IAAI,YAAY,EAAE,CAAC;IAC9D,CAAC,CAAC,CAAC;AACL,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,YAAY,CAAC,KAAmB;IAC9C,MAAM,IAAI,GAA2B,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC;QACvD,CAAC,CAAC,KAAK;QACP,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,KAA+B,CAAC,CAAC,GAAG,CACjD,CAAC,CAAC,IAAI,EAAE,OAAO,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,CAAC,CACzC,CAAC;IAEN,MAAM,KAAK,GAAe,EAAE,CAAC;IAC7B,MAAM,EAAE,GAAG,IAAI,GAAG,EAAoB,CAAC;IAEvC,KAAK,MAAM,IAAI,IAAI,IAAI,EAAE,CAAC;QACxB,IAAI,YAAY,CAAC,IAAI,CAAC,EAAE,CAAC;YACvB,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,IAAI,CAAC,EAAE,CAAC,IAAI,IAAI,CAAC,EAAE,GAAG,CAAC,EAAE,CAAC;gBAC9C,MAAM,IAAI,KAAK,CAAC,0CAA0C,IAAI,CAAC,EAAE,EAAE,CAAC,CAAC;YACvE,CAAC;YACD,MAAM,OAAO,GAAG,YAAY,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,CAAC;YAC9C,MAAM,SAAS,GAAG,eAAe,CAAC,IAAI,CAAC,OAAO,EAAE,OAAO,CAAC,CAAC;YACzD,sEAAsE;YACtE,0EAA0E;YAC1E,8DAA8D;YAC9D,IAAI,SAAS,CAAC,MAAM,KAAK,CAAC,IAAI,SAAS,CAAC,CAAC,CAAC,KAAK,SAAS,EAAE,CAAC;gBACzD,MAAM,IAAI,KAAK,CACb,oBAAoB,OAAO,yBAAyB,IAAI,CAAC,OAAO,EAAE,CACnE,CAAC;YACJ,CAAC;YACD,EAAE,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,EAAE,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC;YAC9B,SAAS;QACX,CAAC;QACD,MAAM,UAAU,GAAG,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAC3C,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAC5B,MAAM,IAAI,KAAK,CAAC,0BAA0B,CAAC,CAAC;QAC9C,CAAC;QACD,KAAK,CAAC,IAAI,CAAC;YACT,UAAU;YACV,OAAO,EAAE,eAAe,CAAC,IAAI,CAAC,OAAO,EAAE,IAAI,CAAC,IAAI,CAAC;SAClD,CAAC,CAAC;IACL,CAAC;IAED,KAAK,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,UAAU,CAAC,MAAM,GAAG,IAAI,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC;IAC9E,OAAO,EAAE,KAAK,EAAE,EAAE,EAAE,CAAC;AACvB,CAAC;AAED,OAAO,EAAE,UAAU,EAAE,WAAW,EAAE,UAAU,EAAE,MAAM,mBAAmB,CAAC"}

package/dist/decode/jiao-rule.d.ts

@@ -0,0 +1,64 @@
+/**
+ * Whether a 教 is teaching or is the noun 教.
+ *
+ * Teaching takes an object or is followed by aspect, and only a verb takes
+ * either, so the tags around it are the evidence.
+ */
+import type { Syllable } from "../syllable/syllable.js";
+import { type EdgeContext, type EdgeRule } from "./rules.js";
+/**
+ * Whether a syllable is `jiāo`, the verbal reading of 教.
+ */
+export declare function isJiao(syllable: Syllable | undefined): boolean;
+/**
+ * What a 教 that is teaching governs: the thing or the person taught.
+ *
+ * A pronoun, a common noun, a name or one of the tags jieba gives a subject —
+ * 教我, 教英语, 教历史, 教孩子, 教玛丽. `ns` is deliberately absent, since the
+ * only 教 with a place tag after it over the corpus is 教青局, an office.
+ */
+export declare const TAUGHT_TAGS: Set<string>;
+/**
+ * Whether the 教 at a position is teaching something to somebody.
+ *
+ * The object is what says so, which is the same shape the modal 得 rule takes:
+ * a 教 with a pronoun, a noun or a name after it is governing it, and a 教 with
+ * 了, 过, 着 or 得 after it is a verb whatever follows that. A particle in front
+ * of it rules it out, since a verb does not follow one.
+ *
+ * {@link GOVERNING} is the case where nothing follows to look at.
+ */
+export declare function isTeachingAt(context: EdgeContext, at: number): boolean;
+/**
+ * 教 read as `jiāo` where it is teaching rather than a religion.
+ *
+ * The dictionary stores 教 as `jiào` with `jiāo` as an alternate, and `jiào` is
+ * right for the compounds — 教育, 教师, 宗教, 主教 — which are words and reach
+ * their reading through the word. What is left is the 教 that stands as a word
+ * of its own, and that one is the verb: 他在北京大学教了三年书 came out
+ * `jiàole sān nián shū`, and 我教英语, 谁教你法语 and 她教我如何游泳 with it.
+ * 教书 is a word and was always `jiāo shū`; nothing carried that across a 了.
+ *
+ * See {@link isTeachingAt} for what decides it. Forcing the single-character
+ * edge is not enough on its own, because a reading spanning two characters
+ * carries its own 教 into the position: 王老师教我们汉语 read `jiào` off 师教,
+ * and 来教我 off 来教. Those are pairs the dictionary holds with no part of
+ * speech and at the cost of a word nothing has ever counted — a reading rather
+ * than a word — so a *tagged* word ending in 教 is left alone, which is every
+ * one that matters: 任教, 宗教, 主教, 佛教, 传教, 执教, 请教, 家教.
+ *
+ * Measured over 88,866 lines of Tatoeba and zh.wikipedia, 181 教 decode as a
+ * word of their own and every one of them read `jiào`. This moves 162 to
+ * `jiāo` and is wrong on three: 统一教创始人, 方法教深思 and 做到了教政分离,
+ * where a nominal compound takes an object's shape, the last of them because
+ * 做到了 is a word and so the 了 in front is not a particle to look at. Of the
+ * 17 it leaves, 8 are right — 诸教中, 教外别传, 董教总, 风教, 教青局 twice, a
+ * 教室 reached through a bad split, and a 教 used as a bare noun — and 9 are
+ * misses, where jieba tags the object something this does not name.
+ *
+ * {@link GOVERNING} is four of that 162 and was added later: 能教教我,
+ * 要教一節課, 也會教像拉丁語, 這在學校是不教的. 有 is pointedly outside it,
+ * since 有教无类 is `yǒujiào wúlèi`.
+ */
+export declare const TEACHING_JIAO: EdgeRule;
+//# sourceMappingURL=jiao-rule.d.ts.map

package/dist/decode/jiao-rule.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"jiao-rule.d.ts","sourceRoot":"","sources":["../../src/decode/jiao-rule.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AACxD,OAAO,EAGL,KAAK,WAAW,EAChB,KAAK,QAAQ,EAId,MAAM,YAAY,CAAC;AAEpB;;GAEG;AACH,wBAAgB,MAAM,CAAC,QAAQ,EAAE,QAAQ,GAAG,SAAS,GAAG,OAAO,CAI9D;AAED;;;;;;GAMG;AACH,eAAO,MAAM,WAAW,aAA8C,CAAC;AAkCvE;;;;;;;;;GASG;AACH,wBAAgB,YAAY,CAAC,OAAO,EAAE,WAAW,EAAE,EAAE,EAAE,MAAM,GAAG,OAAO,CAUtE;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA8BG;AACH,eAAO,MAAM,aAAa,EAAE,QAiB3B,CAAC"}

package/dist/decode/jiao-rule.js

@@ -0,0 +1,115 @@
+import { ASPECT, PARTICLE_TAG, tagOf, wordEndingAt, wordStartingAt, } from "./rules.js";
+/**
+ * Whether a syllable is `jiāo`, the verbal reading of 教.
+ */
+export function isJiao(syllable) {
+    return (syllable?.initial === "j" && syllable.final === "iao" && syllable.tone === 1);
+}
+/**
+ * What a 教 that is teaching governs: the thing or the person taught.
+ *
+ * A pronoun, a common noun, a name or one of the tags jieba gives a subject —
+ * 教我, 教英语, 教历史, 教孩子, 教玛丽. `ns` is deliberately absent, since the
+ * only 教 with a place tag after it over the corpus is 教青局, an office.
+ */
+export const TAUGHT_TAGS = new Set(["r", "n", "nz", "nr", "nt", "ng"]);
+/**
+ * The words that can only be governing a verb, whatever comes after them.
+ *
+ * The far side of the context, for the 教 that teaches nothing named:
+ * 他怎么教，我都学不会, 你没能力教, 这在学校是不教的, 有时也会教. A modal or a
+ * negator has to be followed by a verb, so the 教 after one is the verb even
+ * where nothing at all follows it.
+ *
+ * Written out rather than taken from a tag, for the reason the 长 rule gives:
+ * jieba calls 会 `v` along with every other verb and 怎么 `r` along with every
+ * pronoun, so neither tag names this set.
+ */
+const GOVERNING = new Set([
+    "不",
+    "能",
+    "会",
+    "會",
+    "要",
+    "想",
+    "该",
+    "該",
+    "也",
+    "别",
+    "別",
+    "没",
+    "沒",
+    "怎么",
+    "怎麼",
+    "怎样",
+    "怎樣",
+]);
+/**
+ * Whether the 教 at a position is teaching something to somebody.
+ *
+ * The object is what says so, which is the same shape the modal 得 rule takes:
+ * a 教 with a pronoun, a noun or a name after it is governing it, and a 教 with
+ * 了, 过, 着 or 得 after it is a verb whatever follows that. A particle in front
+ * of it rules it out, since a verb does not follow one.
+ *
+ * {@link GOVERNING} is the case where nothing follows to look at.
+ */
+export function isTeachingAt(context, at) {
+    const before = wordEndingAt(context, at);
+    if (tagOf(context, before).startsWith(PARTICLE_TAG)) {
+        return false;
+    }
+    return (GOVERNING.has(before ?? "") ||
+        ASPECT.has(context.characters[at + 1] ?? "") ||
+        TAUGHT_TAGS.has(tagOf(context, wordStartingAt(context, at + 1))));
+}
+/**
+ * 教 read as `jiāo` where it is teaching rather than a religion.
+ *
+ * The dictionary stores 教 as `jiào` with `jiāo` as an alternate, and `jiào` is
+ * right for the compounds — 教育, 教师, 宗教, 主教 — which are words and reach
+ * their reading through the word. What is left is the 教 that stands as a word
+ * of its own, and that one is the verb: 他在北京大学教了三年书 came out
+ * `jiàole sān nián shū`, and 我教英语, 谁教你法语 and 她教我如何游泳 with it.
+ * 教书 is a word and was always `jiāo shū`; nothing carried that across a 了.
+ *
+ * See {@link isTeachingAt} for what decides it. Forcing the single-character
+ * edge is not enough on its own, because a reading spanning two characters
+ * carries its own 教 into the position: 王老师教我们汉语 read `jiào` off 师教,
+ * and 来教我 off 来教. Those are pairs the dictionary holds with no part of
+ * speech and at the cost of a word nothing has ever counted — a reading rather
+ * than a word — so a *tagged* word ending in 教 is left alone, which is every
+ * one that matters: 任教, 宗教, 主教, 佛教, 传教, 执教, 请教, 家教.
+ *
+ * Measured over 88,866 lines of Tatoeba and zh.wikipedia, 181 教 decode as a
+ * word of their own and every one of them read `jiào`. This moves 162 to
+ * `jiāo` and is wrong on three: 统一教创始人, 方法教深思 and 做到了教政分离,
+ * where a nominal compound takes an object's shape, the last of them because
+ * 做到了 is a word and so the 了 in front is not a particle to look at. Of the
+ * 17 it leaves, 8 are right — 诸教中, 教外别传, 董教总, 风教, 教青局 twice, a
+ * 教室 reached through a bad split, and a 教 used as a bare noun — and 9 are
+ * misses, where jieba tags the object something this does not name.
+ *
+ * {@link GOVERNING} is four of that 162 and was added later: 能教教我,
+ * 要教一節課, 也會教像拉丁語, 這在學校是不教的. 有 is pointedly outside it,
+ * since 有教无类 is `yǒujiào wúlèi`.
+ */
+export const TEACHING_JIAO = {
+    name: "teaching-jiao",
+    verdictFor: (context) => {
+        const { edge } = context;
+        if (edge.text === "教") {
+            return isJiao(edge.reading[0]) && isTeachingAt(context, edge.from)
+                ? "force"
+                : "keep";
+        }
+        return edge.partOfSpeech === "" &&
+            edge.text.endsWith("教") &&
+            edge.reading.length === edge.to - edge.from &&
+            !isJiao(edge.reading.at(-1)) &&
+            isTeachingAt(context, edge.to - 1)
+            ? "forbid"
+            : "keep";
+    },
+};
+//# sourceMappingURL=jiao-rule.js.map

package/dist/decode/jiao-rule.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"jiao-rule.js","sourceRoot":"","sources":["../../src/decode/jiao-rule.ts"],"names":[],"mappings":"AAOA,OAAO,EACL,MAAM,EACN,YAAY,EAGZ,KAAK,EACL,YAAY,EACZ,cAAc,GACf,MAAM,YAAY,CAAC;AAEpB;;GAEG;AACH,MAAM,UAAU,MAAM,CAAC,QAA8B;IACnD,OAAO,CACL,QAAQ,EAAE,OAAO,KAAK,GAAG,IAAI,QAAQ,CAAC,KAAK,KAAK,KAAK,IAAI,QAAQ,CAAC,IAAI,KAAK,CAAC,CAC7E,CAAC;AACJ,CAAC;AAED;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,WAAW,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,CAAC,CAAC;AAEvE;;;;;;;;;;;GAWG;AACH,MAAM,SAAS,GAAG,IAAI,GAAG,CAAC;IACxB,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,IAAI;IACJ,IAAI;IACJ,IAAI;IACJ,IAAI;CACL,CAAC,CAAC;AAEH;;;;;;;;;GASG;AACH,MAAM,UAAU,YAAY,CAAC,OAAoB,EAAE,EAAU;IAC3D,MAAM,MAAM,GAAG,YAAY,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC;IACzC,IAAI,KAAK,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC,UAAU,CAAC,YAAY,CAAC,EAAE,CAAC;QACpD,OAAO,KAAK,CAAC;IACf,CAAC;IACD,OAAO,CACL,SAAS,CAAC,GAAG,CAAC,MAAM,IAAI,EAAE,CAAC;QAC3B,MAAM,CAAC,GAAG,CAAC,OAAO,CAAC,UAAU,CAAC,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;QAC5C,WAAW,CAAC,GAAG,CAAC,KAAK,CAAC,OAAO,EAAE,cAAc,CAAC,OAAO,EAAE,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC,CACjE,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA8BG;AACH,MAAM,CAAC,MAAM,aAAa,GAAa;IACrC,IAAI,EAAE,eAAe;IACrB,UAAU,EAAE,CAAC,OAAoB,EAAE,EAAE;QACnC,MAAM,EAAE,IAAI,EAAE,GAAG,OAAO,CAAC;QACzB,IAAI,IAAI,CAAC,IAAI,KAAK,GAAG,EAAE,CAAC;YACtB,OAAO,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,IAAI,YAAY,CAAC,OAAO,EAAE,IAAI,CAAC,IAAI,CAAC;gBAChE,CAAC,CAAC,OAAO;gBACT,CAAC,CAAC,MAAM,CAAC;QACb,CAAC;QACD,OAAO,IAAI,CAAC,YAAY,KAAK,EAAE;YAC7B,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC;YACvB,IAAI,CAAC,OAAO,CAAC,MAAM,KAAK,IAAI,CAAC,EAAE,GAAG,IAAI,CAAC,IAAI;YAC3C,CAAC,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;YAC5B,YAAY,CAAC,OAAO,EAAE,IAAI,CAAC,EAAE,GAAG,CAAC,CAAC;YAClC,CAAC,CAAC,QAAQ;YACV,CAAC,CAAC,MAAM,CAAC;IACb,CAAC;CACF,CAAC"}

package/dist/decode/lattice-costs.d.ts

@@ -0,0 +1,55 @@
+/**
+ * What the lattice cost, forwards and backwards.
+ *
+ * The two sweeps a confidence score is read off: the cheapest way into every
+ * position and the cheapest way out of it. An edge is uncertain exactly where
+ * some rival edge over the same characters costs nearly as little.
+ */
+import type { ReadingUnit } from "./locking.js";
+import type { Lattice, LatticeEdge } from "./lattice.js";
+import type { CostOf } from "./viterbi.js";
+/**
+ * One distinct reading claimed in a stretch, and what claiming it costs.
+ */
+export interface ScoredClaim {
+    readonly unit: ReadingUnit;
+    readonly cost: number;
+}
+/**
+ * Every distinct reading claimed in a stretch, and the stretch's own best cost.
+ */
+export interface ScoredStretch {
+    readonly claims: ReadonlyMap<string, ScoredClaim>;
+    readonly best: number;
+}
+/**
+ * The edges leaving one position.
+ */
+export declare function edgesAt(lattice: Lattice, at: number): readonly LatticeEdge[];
+/**
+ * Every edge lying wholly inside a stretch.
+ */
+export declare function edgesInside(lattice: Lattice, from: number, to: number): readonly LatticeEdge[];
+/**
+ * Cheapest cost of reaching each position in a stretch from its start.
+ */
+export declare function forwardCosts(lattice: Lattice, from: number, to: number, costOf: CostOf): readonly number[];
+/**
+ * Cheapest cost of reaching the end of a stretch from each position in it.
+ */
+export declare function backwardCosts(lattice: Lattice, from: number, to: number, costOf: CostOf): readonly number[];
+/**
+ * Price every distinct reading a stretch offers.
+ *
+ * A claim is priced by the cheapest path through the whole stretch that makes
+ * it, which is what makes the figures comparable: a reading can be expensive
+ * because it forces expensive readings around it as much as because it is
+ * itself unlikely. One forward sweep and one backward sweep give that for every
+ * edge at once, which is the same work the decode already does in one
+ * direction.
+ */
+export declare function scoreClaims(lattice: Lattice, from: number, to: number, costOf: CostOf): ScoredStretch;
+/**
+ * What a unit beat: the claims overlapping its span, other than its own.
+ */
+//# sourceMappingURL=lattice-costs.d.ts.map