@furifont/dictionary 0.1.2 → 0.3.0

package/dist/index.d.mts

@@ -1,5 +1,5 @@
 import { Dictionary } from "@furifont/core";
 //#region src/index.d.ts
-/** Loads the compiled dictionary: JMdict and JMnedict place names, aligned by JmdictFurigana, with KANJIDIC2's on'yomi and JLPT levels. */
+/** Loads the compiled dictionary: JMdict and JMnedict place names, aligned by JmdictFurigana, with KANJIDIC2's on'yomi, JLPT levels and uses in Tatoeba's sentences. */
 export declare function loadDictionary(): Promise<Dictionary>;
 //#endregion

package/dist/index.mjs

@@ -1,18 +1,95 @@
 import * as v from "valibot";
-//#region src/defaults.ts
+//#region src/manual/mahjong-yaku.ts
 /**
-* Default readings the compiled data ranks wrong, in inline notation. JMdict's
+* The yaku (scoring hands) and tiles of Japanese riichi mahjong, as the Japan
+* Professional Mahjong League lists them: https://www.ma-jan.or.jp/activity/yaku.html
+* Each name reads in katakana there; here it is hiragana, in inline notation.
+* Their readings are jargon (立直 りーち, 七対子 ちーといつ), so they only fill runs
+* that no common word claims.
+*/
+const MAHJONG_YAKU = [
+	"門[めん]前[ぜん]清[ちん]自[つ]摸[も]和[ほー]",
+	"立[りー]直[ち]",
+	"平[ぴん]和[ふ]",
+	"断[たん]么九[やお]",
+	"役[やく]牌[はい]",
+	"三[さん]元[げん]牌[はい]",
+	"場[ば]風[かぜ]牌[はい]",
+	"自[じ]風[かぜ]牌[はい]",
+	"連[れん]風[ぷう]牌[はい]",
+	"客[きゃく]風[ふう]牌[はい]",
+	"一[いー]盃[ぺー]口[こー]",
+	"嶺[りん]上[しゃん]開[かい]花[ほー]",
+	"搶[ちゃん]槓[かん]",
+	"槍[ちゃん]槓[かん]",
+	"海[はい]底[てい]摸[もー]月[ゆえ]",
+	"河[ほー]底[てい]撈[らお]魚[ゆい]",
+	"一[いっ]発[ぱつ]",
+	"三[さん]色[しょく]同[どう]順[じゅん]",
+	"一[いっ]気[き]通[つう]貫[かん]",
+	"七[ちー]対[とい]子[つ]",
+	"対[とい]々[とい]和[ほー]",
+	"三[さん]暗[あん]刻[こう]",
+	"三[さん]色[しょく]同[どう]刻[こう]",
+	"混全帯么九[ちゃんた]",
+	"三[さん]槓[かん]子[つ]",
+	"小[しょう]三[さん]元[げん]",
+	"混[ほん]老[ろう]頭[とう]",
+	"混一色[ほんいつ]",
+	"純全帯么九[じゅんちゃん]",
+	"二[りゃん]盃[ぺー]口[こー]",
+	"清一色[ちんいつ]",
+	"国[こく]士[し]無[む]双[そう]",
+	"四[すー]暗[あん]刻[こう]",
+	"四[すー]暗[あん]刻[こう]単[たん]騎[き]",
+	"大[だい]三[さん]元[げん]",
+	"字[つー]一[いー]色[そー]",
+	"小[しょう]四[すー]喜[しー]",
+	"大[だい]四[すー]喜[しー]",
+	"清[ちん]老[ろう]頭[とう]",
+	"緑[りゅー]一[いー]色[そー]",
+	"九[ちゅー]蓮[れん]宝[ぽう]燈[とう]",
+	"純[じゅん]正[せい]九[ちゅー]蓮[れん]宝[ぽう]燈[とう]",
+	"四[すー]槓[かん]子[つ]",
+	"天[てん]和[ほー]",
+	"地[ち]和[ほー]",
+	"役[やく]満[まん]",
+	"赤[あか]牌[はい]"
+];
+//#endregion
+//#region src/manual/index.ts
+const NotationsSchema = v.array(v.string());
+/**
+* Word lists kept by hand, by source name, in inline notation. A source is a
+* list written in this directory, or a file under `data/manual/` that a script
+* compiled from a web page, which bundlers split into a lazy chunk. Add a
+* source here and a build includes it, unless its plan lists the sources it wants.
+*/
+const MANUAL_SOURCES = {
+	"mahjong-yaku": async () => MAHJONG_YAKU,
+	"yojijukugo": async () => v.parse(NotationsSchema, (await import("./yojijukugo-B8mguWRF.mjs")).default),
+	"jr-stations": async () => v.parse(NotationsSchema, (await import("./jr-stations-Rjpbmev2.mjs")).default)
+};
+async function loadManualSources() {
+	return Object.fromEntries(await Promise.all(Object.entries(MANUAL_SOURCES).map(async ([name, load]) => [name, await load()])));
+}
+//#endregion
+//#region src/overrides.ts
+/**
+* Readings that override the compiled data where it ranks them wrong, in inline notation. JMdict's
 * frequency tags belong to a written form, so across entries they cannot tell
 * two readings apart; and a form that drops its okurigana usually meets it in
 * running text, where the font cannot see it. Users' pins override these.
 */
-const DEFAULT_READINGS = [
+const OVERRIDE_READINGS = [
 	"出[で]来[き]",
 	"何[なん]人[にん]",
 	"何[なん]時[じ]",
 	"三[さん]十[じゅう]日[にち]",
 	"外[そと]付[づ]",
-	"支[し]払[はら]"
+	"支[し]払[はら]",
+	"字[じ]",
+	"明[みん]朝[ちょう]"
 ];
 /**
 * Kanji left bare when alone in a run, though one reading dominates the
@@ -50,6 +127,11 @@ const JlptSchema = v.array(v.tuple([
 		5
 	])
 ]));
+const TatoebaSchema = v.array(v.tuple([
+	v.string(),
+	v.string(),
+	v.pipe(v.number(), v.integer(), v.minValue(1))
+]));
 /**
 * A data file as a JSON module, which Node loads and bundlers split into a
 * lazy chunk. The path is a template so that TypeScript leaves the 11 MB file
@@ -59,25 +141,31 @@ async function readData(name, schema) {
 	const { default: data } = await import(`../data/${name}.json`, { with: { type: "json" } });
 	return v.parse(schema, data);
 }
-/** Loads the compiled dictionary: JMdict and JMnedict place names, aligned by JmdictFurigana, with KANJIDIC2's on'yomi and JLPT levels. */
+/** Loads the compiled dictionary: JMdict and JMnedict place names, aligned by JmdictFurigana, with KANJIDIC2's on'yomi, JLPT levels and uses in Tatoeba's sentences. */
 async function loadDictionary() {
-	const [data, places, jlpt] = await Promise.all([
+	const [data, places, jlpt, tatoeba, manual] = await Promise.all([
 		readData("dictionary", DataSchema),
 		readData("places", PlacesSchema),
-		readData("jlpt", JlptSchema)
+		readData("jlpt", JlptSchema),
+		readData("tatoeba", TatoebaSchema),
+		loadManualSources()
 	]);
 	const levels = new Map(jlpt.map(([form, furigana, level]) => [`${form}|${furigana}`, level]));
+	const uses = new Map(tatoeba.map(([form, furigana, n]) => [`${form}|${furigana}`, n]));
 	return {
 		sources: data.sources,
 		onyomi: data.onyomi,
-		defaults: DEFAULT_READINGS,
+		defaults: OVERRIDE_READINGS,
 		bare: BARE_FORMS,
+		manual,
 		places: places.map(([form, furigana]) => ({
 			form,
 			furigana
 		})),
 		entries: data.entries.map(([form, furigana, nf, priority, order, conjugates]) => {
 			const level = levels.get(`${form}|${furigana}`);
+			const n = uses.get(`${form}|${furigana}`);
+			uses.delete(`${form}|${furigana}`);
 			return {
 				form,
 				furigana,
@@ -85,7 +173,8 @@ async function loadDictionary() {
 				priority,
 				order,
 				...conjugates && { conjugatesReading: true },
-				...level && { jlpt: level }
+				...level && { jlpt: level },
+				...n && { tatoeba: n }
 			};
 		})
 	};