@furifont/core 0.2.0 → 0.3.0

package/dist/index.d.mts

@@ -18,6 +18,7 @@ export declare const DEFAULTS: {
     /** Uses in Tatoeba's example sentences that select a word whatever its bucket: about 1,000 more words, and 50 glyphs. */
     readonly minTatoeba: 3;
     readonly places: true;
+    readonly untagged: true;
     readonly isolated: true;
     readonly isolatedMinShare: 0.8;
     /** Sizes in em, relative to the base font. */
@@ -119,6 +120,8 @@ interface Dictionary {
     form: string;
     furigana: string;
   }[];
+  /** Word lists kept by hand, by source name, in inline notation; like place names, they fill runs no common word claims. */
+  manual?: Record<string, string[]>;
 }
 /** A ruby annotation over characters `[start, end)` of a run. */
 interface RubySpan {
@@ -154,6 +157,8 @@ export declare function formatNotation(text: string, spans: readonly RubySpan[])
 export declare const RUBY_SCRIPTS: readonly ["hiragana", "katakanaOn", "hepburn"];
 export declare const PRIORITY_TAGS: readonly ["ichi1", "ichi2", "news1", "news2", "spec1", "spec2", "gai1", "gai2"];
 /** Easiest first. */
+/** The dictionary's manually maintained sources, which a plan may include by name. */
+export declare const MANUAL_SOURCES: readonly ["mahjong-yaku", "jr-stations", "yojijukugo"];
 export declare const JLPT_LEVELS: readonly ["N5", "N4", "N3", "N2", "N1"];
 type RubyScript = typeof RUBY_SCRIPTS[number];
 /** Sizes in em, relative to the base font. */
@@ -200,6 +205,10 @@ export declare const PlanOptionsSchema: v.SchemaWithPipe<readonly [v.ObjectSchem
   readonly skipJlpt: v.OptionalSchema<v.PicklistSchema<readonly ["N5", "N4", "N3", "N2", "N1"], undefined>, undefined>;
   /** Annotate place names from the dictionary where no common word claims the run. */
   readonly places: v.OptionalSchema<v.BooleanSchema<undefined>, true>;
+  /** Annotate the words of JMdict that no list ranks or tags (such as 四字熟語), where no common word, manual list or place name claims the run; each gets its first reading. */
+  readonly untagged: v.OptionalSchema<v.BooleanSchema<undefined>, true>;
+  /** The dictionary's manually maintained sources (such as `mahjong-yaku`) to annotate; all of them when left out. */
+  readonly manual: v.OptionalSchema<v.ArraySchema<v.PicklistSchema<readonly ["mahjong-yaku", "jr-stations", "yojijukugo"], undefined>, undefined>, undefined>;
   /** Annotate a kanji alone in its run, or one no word claims after another kanji (a suffix); when off, only selectors give them ruby. */
   readonly isolated: v.OptionalSchema<v.BooleanSchema<undefined>, true>;
   /** A kanji alone in its run is annotated only when one reading has this share of the evidence. */
@@ -231,6 +240,8 @@ export declare const PlanOptionsSchema: v.SchemaWithPipe<readonly [v.ObjectSchem
   minTatoeba: number;
   skipJlpt?: "N5" | "N4" | "N3" | "N2" | "N1" | undefined;
   places: boolean;
+  untagged: boolean;
+  manual?: ("mahjong-yaku" | "jr-stations" | "yojijukugo")[] | undefined;
   isolated: boolean;
   isolatedMinShare: number;
   words: {

package/dist/index.mjs

@@ -23,6 +23,7 @@ const DEFAULTS = {
 		/** Uses in Tatoeba's example sentences that select a word whatever its bucket: about 1,000 more words, and 50 glyphs. */
 		minTatoeba: 3,
 		places: true,
+		untagged: true,
 		isolated: true,
 		isolatedMinShare: .8,
 		/** Sizes in em, relative to the base font. */
@@ -680,6 +681,40 @@ function addPlaces(patterns, places, skipped, ctx) {
 		}]);
 	}
 }
+/** Words of the manual sources fill the runs no common word claims, nor a skipped one; they come before place names. */
+function addManual(patterns, manual, skipped, ctx) {
+	const names = ctx.options.manual ?? Object.keys(manual);
+	for (const notation of names.flatMap((name) => manual[name] ?? [])) {
+		const { form, spans } = parseNotation(notation);
+		for (const run of hanRuns(form, spans).filter(ctx.renderable)) {
+			if (length(run.text) === 1 || patterns.has(run.text) || skipped.has(run.text) || ctx.removed.has(run.text)) continue;
+			patterns.set(run.text, [{
+				spans: run.spans,
+				weight: 0,
+				tagged: false,
+				order: 0,
+				nf: 0,
+				jlpt: 0,
+				tags: /* @__PURE__ */ new Set(),
+				place: true
+			}]);
+		}
+	}
+}
+/** Words no list ranks fill the runs nothing else claims, each with its first reading: they are no evidence for the kanji's own. */
+function addUntagged(patterns, index, skipped, ctx) {
+	for (const [text, readings] of index.runs) {
+		if (patterns.has(text) || skipped.has(text) || ctx.removed.has(text)) continue;
+		const first = [...readings.values()].filter((r) => ctx.renderable({
+			text,
+			spans: r.spans
+		})).sort((a, b) => a.order - b.order)[0];
+		if (first) patterns.set(text, [{
+			...first,
+			place: true
+		}]);
+	}
+}
 /**
 * Runs of common words, then place names and custom words; pinned readings of
 * lone kanji are returned apart. A word is skipped whole: were only its known
@@ -700,7 +735,9 @@ function selectWords(index, dictionary, ctx) {
 		}));
 		if (common.length) patterns.set(text, common.map((r) => ({ ...r })));
 	}
+	if (dictionary.manual) addManual(patterns, dictionary.manual, skipped, ctx);
 	if (ctx.options.places) addPlaces(patterns, dictionary.places, skipped, ctx);
+	if (ctx.options.untagged) addUntagged(patterns, index, skipped, ctx);
 	const lonePins = /* @__PURE__ */ new Map();
 	for (const notation of ctx.options.words.add) addCustomWord(patterns, lonePins, notation, 0, ctx);
 	for (const notation of dictionary.defaults ?? []) addCustomWord(patterns, lonePins, notation, 1, ctx);
@@ -1226,6 +1263,12 @@ const PRIORITY_TAGS = [
 	"gai2"
 ];
 /** Easiest first. */
+/** The dictionary's manually maintained sources, which a plan may include by name. */
+const MANUAL_SOURCES = [
+	"mahjong-yaku",
+	"jr-stations",
+	"yojijukugo"
+];
 const JLPT_LEVELS = [
 	"N5",
 	"N4",
@@ -1276,6 +1319,10 @@ const PlanOptionsSchema = withNfWindow(v.object({
 	skipJlpt: v.optional(v.picklist(JLPT_LEVELS)),
 	/** Annotate place names from the dictionary where no common word claims the run. */
 	places: v.optional(v.boolean(), DEFAULTS.plan.places),
+	/** Annotate the words of JMdict that no list ranks or tags (such as 四字熟語), where no common word, manual list or place name claims the run; each gets its first reading. */
+	untagged: v.optional(v.boolean(), DEFAULTS.plan.untagged),
+	/** The dictionary's manually maintained sources (such as `mahjong-yaku`) to annotate; all of them when left out. */
+	manual: v.optional(v.array(v.picklist(MANUAL_SOURCES))),
 	/** Annotate a kanji alone in its run, or one no word claims after another kanji (a suffix); when off, only selectors give them ruby. */
 	isolated: v.optional(v.boolean(), DEFAULTS.plan.isolated),
 	/** A kanji alone in its run is annotated only when one reading has this share of the evidence. */
@@ -1343,4 +1390,4 @@ const PlanSchema = v.object({
 	}))
 });
 //#endregion
-export { DEFAULTS, JLPT_LEVELS, MAX_READINGS, PREBUILT, PRIORITY_TAGS, PlanOptionsSchema, PlanSchema, RUBY_SCRIPTS, SELECTOR_FIRST_READING, SELECTOR_SUPPRESS, VERSION_PROBE, createFontPlanner, formatNotation, glyphName, hanRuns, isHan, isOnyomi, parseFurigana, parseNotation, resolveOptions, toHepburn, toHiragana, toKatakana, withNfWindow };
+export { DEFAULTS, JLPT_LEVELS, MANUAL_SOURCES, MAX_READINGS, PREBUILT, PRIORITY_TAGS, PlanOptionsSchema, PlanSchema, RUBY_SCRIPTS, SELECTOR_FIRST_READING, SELECTOR_SUPPRESS, VERSION_PROBE, createFontPlanner, formatNotation, glyphName, hanRuns, isHan, isOnyomi, parseFurigana, parseNotation, resolveOptions, toHepburn, toHiragana, toKatakana, withNfWindow };

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@furifont/core",
   "type": "module",
-  "version": "0.2.0",
+  "version": "0.3.0",
   "description": "Plan furigana glyphs and OpenType rules from a Japanese dictionary",
   "author": "Anthony Fu <anthonyfu117@hotmail.com>",
   "license": "MIT",