@kensio/pinyinjs 1.10.2 → 1.10.4

package/data/standard.entries

@@ -73,6 +73,7 @@ she4
 tiao3
 lian2
 mian3				tian3
+weng1
 sa4
 tuo2
 hun2
@@ -7824,7 +7825,7 @@ juan4		zg
 zong1		zg
 ju4		ag
 		a
-qian4		zg
+qian4		zg		qing4
 		n
 ni2		nr	p
 lun2		ns
@@ -21325,10 +21326,10 @@ pi2		zg		pi4
 qing1		zg
 ju4		zg
 kong1		zg
-tang3	chong3
+tang3	chong3			shang3
 dong1		zg
 dai4		zg
-tan4		zg
+tan4		zg		tan2
 an3
 chu4		zg
 beng3
@@ -24907,7 +24908,7 @@ tuo3 dang5	tuo3 dang4	a
 		v
 		v
 feng1
-yuan2		zg
+yuan2		zg		wan4
 fang2		zg		fang1
 		v
 		v
@@ -25370,7 +25371,7 @@ wei2		zg
 di4
 rou2
 mei3
-ruan3
+ruan3				nen4
 wo4		zg
 chun1		zg
 miao2
@@ -25392,7 +25393,7 @@ mei4		zg
 		n
 yuan4		zg		yuan2
 zheng1		zg
-shi4
+shi4				ti2
 xie4
 tuo3		zg		duo4
 mao4
@@ -26417,7 +26418,7 @@ yi2		zg
 jing3		zg
 mi4		nr		fu2
 zhu3		zg
-dang4		g
+dang4		g		tan4
 zong1		nr		zong5
 		n
 		n
@@ -35163,7 +35164,7 @@ yi4 you2 wei4 jin4		i
 		n
 		n
 		n
-kai4		zg		he4
+kai4		zg		he4,qi4
 dang4
 yin1		zg
 e4		a
@@ -38663,7 +38664,7 @@ juan3 qu1		n
 juan4 lian2		n
 		v
 	juan3 fa3	v
-chen1
+chen1				tian3
 chui2		v
 		v
 		i
@@ -46333,7 +46334,7 @@ kuo4		zg
 jing4		zg
 po5		zg		bo2
 fu2		g
-rui2		zg
+rui2		zg		rui3
 tong3		zg
 jue2		g
 xi1		zg
@@ -53672,7 +53673,7 @@ lian2		ns
 		ns	p
 		n
 lan3		v
-chi2
+chi2				si1
 xuan2		n
 		n
 		n
@@ -57433,7 +57434,7 @@ ma3		nr
 		nrt
 wu3		zg
 fu1		zg
-wen2		zg
+wen2		zg		min2
 jie4		zg
 ya2		zg
 fen1		x		bin1
@@ -57555,7 +57556,7 @@ dai4		n
 dian4		zg
 		v
 	dian4 ru4	v
-xuan2		zg
+xuan2		zg		xian2
 xi3		ng
 bo1		n
 		ns	p
@@ -57872,6 +57873,7 @@ li3 xing4 ren4 shi5	li3 xing4 ren4 shi4	l
 	li3 fa3 shi1	n
 	li3 fa3 dian4	n
 xiu4		zg
+fu2
 liu2		ns
 		n
 		ns	p2
@@ -58038,6 +58040,7 @@ yao2		j
 		n
 ai4		x
 bin1
+shu1				tu1
 chang2		zg
 kun1
 cong1
@@ -65226,7 +65229,7 @@ jian3 piao2	jian3 pu2	a
 		v
 		a
 pou2		zg
-bi3				bei1
+bi3				bei1,pai2
 bi4		zg
 ge4		zg
 yu1
@@ -72777,7 +72780,7 @@ zi4		zg
 hui4
 sui1
 xiang1		ns
-bi3		x
+bi3		x		pi2
 fu2		nrt
 		n
 		ns
@@ -74470,7 +74473,8 @@ suan4		n
 li4		vg
 		v
 ju3		zg
-bang4
+xi1
+bang4				pang2
 chu2		zg
 huo4
 qian4		zg
@@ -75663,7 +75667,7 @@ rui4		zg
 ni2		zg
 qing1		n
 		n
-wei4
+wei4				wei3
 liang3		zg
 guo3		n
 wan1		v
@@ -78666,7 +78670,7 @@ an1		v
 		a
 xian2
 nan2
-chen2		nr
+chen2		nr		shen4
 feng3	feng4	v
 	feng4 ci4	v
 		v
@@ -79995,7 +79999,7 @@ mou2 hua4		n
 		v
 		i
 		n
-chen2		nr
+chen2		nr		shen4
 die2		n
 		n
 huang3		ng
@@ -80331,7 +80335,7 @@ xiu1		x
 huan2
 kun1		zg
 he2
-he2		ng		hao2
+he2		ng		hao2,mo4
 mo4		g
 mao4		ng
 		i
@@ -81981,7 +81985,7 @@ zhi3		n
 ta1		v
 yue4
 ba4
-qi2		zg		qi4
+qi2		zg		qi4,qi3
 yue4		v
 		v
 		v
@@ -87946,7 +87950,7 @@ sui4		zg
 lu3		zg
 juan1		zg
 		v
-hui4		zg
+hui4		zg		wei4
 yu4		zg
 lian2		n
 		n
@@ -97941,7 +97945,7 @@ nao2
 gong3
 geng4
 ying2
-run2
+run2				shun4
 cha2
 mu3
 xi4

package/data/standard.keys

@@ -73,6 +73,7 @@ B
 䠷
 䥥
 䩄
+䩺
 䬃
 䭾
 䰟
@@ -57872,6 +57873,7 @@ B
 理髮師
 理髮店
 琇
+琈
 琉
 琉球
 琉球群岛
@@ -58038,6 +58040,7 @@ B
 瑶池
 瑷
 瑸
+瑹
 瑺
 瑻
 瑽
@@ -74470,6 +74473,7 @@ B
 蒞
 蒞臨
 蒟
+蒠
 蒡
 蒢
 蒦

package/dist/decode/li-frames.d.ts

@@ -0,0 +1,29 @@
+/**
+ * Whether a 里 standing alone means "in, inside", which 繁體 writes 裡.
+ *
+ * 里 has two senses that read alike. `lǐ` is the unit of distance and the
+ * neighbourhood, written 里 in both scripts (公里, 邻里, 里程, 里长), and `lǐ`
+ * is also the locative after a noun, written 裡 (鏡子裡, 房間裡, 冰箱裡). The
+ * syllable cannot tell them apart. The character table takes 里 by default,
+ * because most of the words that hold 里 are the unit or the neighbourhood.
+ * The table converts the locative words it knows (这里, 心里, 家里, 夜里)
+ * whole. A 里 the decode leaves as a word of its own has no such word behind
+ * it, and is the locative unless a number comes before it: 屋子里 is inside the
+ * room, and 走了十里 is ten 里 walked.
+ */
+import type { Dictionary } from "../dictionary/dictionary.js";
+import type { ScriptChoice } from "./script-types.js";
+/**
+ * Write 裡 where a 里 standing alone means "in, inside".
+ *
+ * The character table takes 里 by default, which is right for 公里 and 邻里 and
+ * wrote 镜子里 as 鏡子里. A 里 the decode left as a word of its own after a
+ * noun is the locative. A number before it makes it the unit, and a
+ * transliteration syllable either side makes it part of a name. See
+ * {@link isInsideAt}.
+ *
+ * Only a choice that names 裡 as a rival is touched, which confines this to
+ * 简→繁: converting into 简体 has nothing to change 里 into.
+ */
+export declare function writeInside(choices: readonly ScriptChoice[], dictionary: Dictionary, characters: readonly string[], from: number): readonly ScriptChoice[];
+//# sourceMappingURL=li-frames.d.ts.map

package/dist/decode/li-frames.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"li-frames.d.ts","sourceRoot":"","sources":["../../src/decode/li-frames.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;GAYG;AACH,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,6BAA6B,CAAC;AAE9D,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,mBAAmB,CAAC;AAqFtD;;;;;;;;;;;GAWG;AACH,wBAAgB,WAAW,CACzB,OAAO,EAAE,SAAS,YAAY,EAAE,EAChC,UAAU,EAAE,UAAU,EACtB,UAAU,EAAE,SAAS,MAAM,EAAE,EAC7B,IAAI,EAAE,MAAM,GACX,SAAS,YAAY,EAAE,CAoBzB"}

package/dist/decode/li-frames.js

@@ -0,0 +1,84 @@
+import { toCharacters } from "../script/characters.js";
+/**
+ * The 繁體 form of the locative.
+ */
+const INSIDE = "裡";
+/**
+ * What can come before a 里 that counts distance, in both scripts.
+ *
+ * The numerals, and the words that stand in for one or round it off: 几里,
+ * 数里, 十多里, 十来里, 十余里, 半里.
+ */
+const COUNTING = new Set(toCharacters("〇零一二三四五六七八九十百千万萬两兩廿几幾数數多来來余餘半"));
+/**
+ * Transliteration syllables that come before a 里 in a foreign name: 克里斯,
+ * 加布里埃尔, 德米特里, 鲍里斯, 恩里科, 胡里奥, 齐里.
+ */
+const NAME_BEFORE = new Set(toCharacters("克布特鲍胡恩夏埃尤荷齐弗杰诺莫默奥艾迪阿伊佩塞贝哈帕茨维格"));
+/**
+ * Transliteration syllables that come after a 里 in a foreign name: 里玛,
+ * 鲍里斯, 胡里奥, 恩里科, 格里菲斯, 斯特里普, 叶里温.
+ */
+const NAME_AFTER = new Set(toCharacters("斯埃玛奥科希约马奇尔亚诺娜莎欧达雅菲普曼温"));
+/**
+ * Whether two characters are a word of their own, which the name syllables
+ * are not. 坦克里 and 表格里 end in 克 and 格, and 坦克 and 表格 are nouns.
+ */
+function isWord(dictionary, characters) {
+    return (characters.length === 2 &&
+        dictionary.lookup(characters.join("")) !== undefined);
+}
+/**
+ * Whether a 里 sits inside a foreign name, between syllables that make no
+ * word with what is either side of them.
+ */
+function isInName(dictionary, characters, at) {
+    return ((NAME_BEFORE.has(characters[at - 1]) &&
+        !isWord(dictionary, characters.slice(at - 2, at))) ||
+        (NAME_AFTER.has(characters[at + 1]) &&
+            !isWord(dictionary, characters.slice(at + 1, at + 3))));
+}
+/**
+ * Whether the 里 at a position means "in, inside".
+ *
+ * The caller has already settled that the character is a 里 standing alone.
+ * A 里 with nothing before it in the run is left alone, since there is no noun
+ * for it to be inside. A run starts after punctuation or any character that is
+ * not Han, digits among them, and 5里 is as likely as Facebook里.
+ */
+function isInsideAt(dictionary, characters, at) {
+    const before = characters[at - 1];
+    return (before !== undefined &&
+        !COUNTING.has(before) &&
+        !isInName(dictionary, characters, at));
+}
+/**
+ * Write 裡 where a 里 standing alone means "in, inside".
+ *
+ * The character table takes 里 by default, which is right for 公里 and 邻里 and
+ * wrote 镜子里 as 鏡子里. A 里 the decode left as a word of its own after a
+ * noun is the locative. A number before it makes it the unit, and a
+ * transliteration syllable either side makes it part of a name. See
+ * {@link isInsideAt}.
+ *
+ * Only a choice that names 裡 as a rival is touched, which confines this to
+ * 简→繁: converting into 简体 has nothing to change 里 into.
+ */
+export function writeInside(choices, dictionary, characters, from) {
+    const [choice] = choices;
+    if (choices.length !== 1 ||
+        choice === undefined ||
+        !choice.alternatives.includes(INSIDE) ||
+        !isInsideAt(dictionary, characters, from)) {
+        return choices;
+    }
+    return [
+        {
+            from: choice.from,
+            to: INSIDE,
+            evidence: "reading",
+            alternatives: [choice.to, ...choice.alternatives].filter((form) => form !== INSIDE),
+        },
+    ];
+}
+//# sourceMappingURL=li-frames.js.map

package/dist/decode/li-frames.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"li-frames.js","sourceRoot":"","sources":["../../src/decode/li-frames.ts"],"names":[],"mappings":"AAcA,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAGvD;;GAEG;AACH,MAAM,MAAM,GAAG,GAAG,CAAC;AAEnB;;;;;GAKG;AACH,MAAM,QAAQ,GAAG,IAAI,GAAG,CACtB,YAAY,CAAC,+BAA+B,CAAC,CAC9C,CAAC;AAEF;;;GAGG;AACH,MAAM,WAAW,GAAG,IAAI,GAAG,CACzB,YAAY,CAAC,+BAA+B,CAAC,CAC9C,CAAC;AAEF;;;GAGG;AACH,MAAM,UAAU,GAAG,IAAI,GAAG,CACxB,YAAY,CAAC,uBAAuB,CAAC,CACtC,CAAC;AAEF;;;GAGG;AACH,SAAS,MAAM,CACb,UAAsB,EACtB,UAA6B;IAE7B,OAAO,CACL,UAAU,CAAC,MAAM,KAAK,CAAC;QACvB,UAAU,CAAC,MAAM,CAAC,UAAU,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,KAAK,SAAS,CACrD,CAAC;AACJ,CAAC;AAED;;;GAGG;AACH,SAAS,QAAQ,CACf,UAAsB,EACtB,UAA6B,EAC7B,EAAU;IAEV,OAAO,CACL,CAAC,WAAW,CAAC,GAAG,CAAC,UAAU,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;QAClC,CAAC,MAAM,CAAC,UAAU,EAAE,UAAU,CAAC,KAAK,CAAC,EAAE,GAAG,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC;QACpD,CAAC,UAAU,CAAC,GAAG,CAAC,UAAU,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;YACjC,CAAC,MAAM,CAAC,UAAU,EAAE,UAAU,CAAC,KAAK,CAAC,EAAE,GAAG,CAAC,EAAE,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC,CACzD,CAAC;AACJ,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,UAAU,CACjB,UAAsB,EACtB,UAA6B,EAC7B,EAAU;IAEV,MAAM,MAAM,GAAG,UAAU,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;IAClC,OAAO,CACL,MAAM,KAAK,SAAS;QACpB,CAAC,QAAQ,CAAC,GAAG,CAAC,MAAM,CAAC;QACrB,CAAC,QAAQ,CAAC,UAAU,EAAE,UAAU,EAAE,EAAE,CAAC,CACtC,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;GAWG;AACH,MAAM,UAAU,WAAW,CACzB,OAAgC,EAChC,UAAsB,EACtB,UAA6B,EAC7B,IAAY;IAEZ,MAAM,CAAC,MAAM,CAAC,GAAG,OAAO,CAAC;IACzB,IACE,OAAO,CAAC,MAAM,KAAK,CAAC;QACpB,MAAM,KAAK,SAAS;QACpB,CAAC,MAAM,CAAC,YAAY,CAAC,QAAQ,CAAC,MAAM,CAAC;QACrC,CAAC,UAAU,CAAC,UAAU,EAAE,UAAU,EAAE,IAAI,CAAC,EACzC,CAAC;QACD,OAAO,OAAO,CAAC;IACjB,CAAC;IACD,OAAO;QACL;YACE,IAAI,EAAE,MAAM,CAAC,IAAI;YACjB,EAAE,EAAE,MAAM;YACV,QAAQ,EAAE,SAAS;YACnB,YAAY,EAAE,CAAC,MAAM,CAAC,EAAE,EAAE,GAAG,MAAM,CAAC,YAAY,CAAC,CAAC,MAAM,CACtD,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,KAAK,MAAM,CAC1B;SACF;KACF,CAAC;AACJ,CAAC"}

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

@@ -1,13 +1,3 @@
-/**
- * The rules about words that begin at an aspect particle or at the adverbial
- * 地.
- *
- * {@link import("./de-rule.js").PARTICLE_DE} and
- * {@link import("./complement-rule.js").COMPLEMENT_DE} settle the same fault for 的 and
- * 得: the full tier keeps keys no source tagged, many of them from the phrase
- * corpus, and one that starts at a particle reads that particle as whatever it
- * is at the head of a phrase.
- */
 import { type EdgeRule } from "./rules.js";
 /**
  * A word beginning with 着 or 了 does not start where the aspect particle does.
@@ -35,8 +25,16 @@ export declare const ASPECTUAL_PARTICLE: EdgeRule;
  * for a family's place of origin. After an adjective or an adverb 地 is the
  * particle that makes it modify a verb, and reads `de`.
  *
+ * Only an untagged key is forbidden there, since a tagged one such as 地方 can
+ * follow an adjective as a noun: 这么大地方. After a reduplication any key is
+ * forbidden, since the 地 after 一点儿一点儿 or 慢慢 is always the particle.
+ * 跟着丈夫一点儿一点儿地学 and 他慢慢地学 came out `dì xué` on the noun 地学.
+ *
  * Over the 88,612 sentences of Tatoeba and the 43,367 Chinese Boost 例句 it
  * changes 2 runs in the full tier, both corrections: 惆怅地望着 and 悠哉地坐.
+ * The reduplication changes 1 more of the 17,825 distinct runs holding a 地 in
+ * Tatoeba, zh.wikipedia and the 47,848 例句 there are now, also a correction:
+ * 斷斷續續地下了.
  */
 export declare const ADVERBIAL_DI: EdgeRule;
 //# sourceMappingURL=particle-rule.d.ts.map

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

@@ -1 +1 @@
-{"version":3,"file":"particle-rule.d.ts","sourceRoot":"","sources":["../../src/decode/particle-rule.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AACH,OAAO,EAEL,KAAK,QAAQ,EAGd,MAAM,YAAY,CAAC;AAgCpB;;;;;;;;;;;;;;;;;GAiBG;AACH,eAAO,MAAM,kBAAkB,EAAE,QAShC,CAAC;AAcF;;;;;;;;;GASG;AACH,eAAO,MAAM,YAAY,EAAE,QAS1B,CAAC"}
+{"version":3,"file":"particle-rule.d.ts","sourceRoot":"","sources":["../../src/decode/particle-rule.ts"],"names":[],"mappings":"AAWA,OAAO,EAEL,KAAK,QAAQ,EAGd,MAAM,YAAY,CAAC;AAgCpB;;;;;;;;;;;;;;;;;GAiBG;AACH,eAAO,MAAM,kBAAkB,EAAE,QAShC,CAAC;AAyEF;;;;;;;;;;;;;;;;;GAiBG;AACH,eAAO,MAAM,YAAY,EAAE,QAgB1B,CAAC"}

package/dist/decode/particle-rule.js

@@ -8,6 +8,7 @@
  * corpus, and one that starts at a particle reads that particle as whatever it
  * is at the head of a phrase.
  */
+import { toCharacters } from "../script/characters.js";
 import { tagOf, wordEndingAt, } from "./rules.js";
 /**
  * Whether an edge is a word of more than one character that no source tagged
@@ -69,6 +70,56 @@ const DI = new Set(["地"]);
  * 地球轨道.
  */
 const ADVERBIAL_TAGS = /^(?:a|z|d)/u;
+/** The numeral that starts a count said twice over: 一点儿一点儿, 一步一步. */
+const ONE = "一";
+/**
+ * The tags of a reduplicated word that says how something is done: 慢慢 and
+ * 渐渐 are `d`, 悄悄 and 匆匆 are `z`, and 断断续续 is `i`.
+ */
+const MANNER_TAGS = /^(?:d|z|i)/u;
+/**
+ * The reduplicated adverbs that say when or how often rather than how, and
+ * stand in front of a 地震 or a 地铁 as readily as in front of the particle.
+ */
+const TEMPORAL_REDUPLICATIONS = new Set([
+    "常常",
+    "刚刚",
+    "剛剛",
+    "每每",
+    "偏偏",
+    "仅仅",
+    "僅僅",
+    "恰恰",
+]);
+/**
+ * Whether the words ending at a position are a reduplication that 地 makes
+ * into an adverbial.
+ *
+ * Two shapes qualify. A count said twice, 一点儿一点儿 or 一步一步, is a
+ * manner of doing something and nothing else. A word that is itself
+ * reduplicated, 慢慢 or 渐渐, qualifies where its tag makes it an adverb of
+ * manner. Repeated characters alone are not enough: 学学 in 大学学地质 ends a
+ * 大学, and the 嚐嚐 of 嚐嚐地道的四川菜 is a verb whose object is 地道.
+ */
+function isReduplicatedBefore(context, at) {
+    const word = wordEndingAt(context, at);
+    if (word === undefined) {
+        return false;
+    }
+    const characters = toCharacters(word);
+    const length = characters.length;
+    if (characters[0] === ONE &&
+        length >= 2 &&
+        context.characters.slice(at - 2 * length, at - length).join("") === word) {
+        return true;
+    }
+    const isDoubled = (length === 2 || length === 4) &&
+        characters[0] === characters[1] &&
+        characters[2] === characters[3];
+    return (isDoubled &&
+        MANNER_TAGS.test(tagOf(context, word)) &&
+        !TEMPORAL_REDUPLICATIONS.has(word));
+}
 /**
  * A word beginning with 地 does not start where the adverbial particle does.
  *
@@ -76,16 +127,31 @@ const ADVERBIAL_TAGS = /^(?:a|z|d)/u;
  * for a family's place of origin. After an adjective or an adverb 地 is the
  * particle that makes it modify a verb, and reads `de`.
  *
+ * Only an untagged key is forbidden there, since a tagged one such as 地方 can
+ * follow an adjective as a noun: 这么大地方. After a reduplication any key is
+ * forbidden, since the 地 after 一点儿一点儿 or 慢慢 is always the particle.
+ * 跟着丈夫一点儿一点儿地学 and 他慢慢地学 came out `dì xué` on the noun 地学.
+ *
  * Over the 88,612 sentences of Tatoeba and the 43,367 Chinese Boost 例句 it
  * changes 2 runs in the full tier, both corrections: 惆怅地望着 and 悠哉地坐.
+ * The reduplication changes 1 more of the 17,825 distinct runs holding a 地 in
+ * Tatoeba, zh.wikipedia and the 47,848 例句 there are now, also a correction:
+ * 斷斷續續地下了.
  */
 export const ADVERBIAL_DI = {
     name: "adverbial-di",
     verdictFor: (context) => {
+        const { characters, edge } = context;
+        if (edge.to - edge.from < 2 || !DI.has(characters[edge.from] ?? "")) {
+            return "keep";
+        }
+        if (isReduplicatedBefore(context, edge.from)) {
+            return "forbid";
+        }
         if (!isUntaggedFrom(context, DI)) {
             return "keep";
         }
-        const before = wordEndingAt(context, context.edge.from);
+        const before = wordEndingAt(context, edge.from);
         return ADVERBIAL_TAGS.test(tagOf(context, before)) ? "forbid" : "keep";
     },
 };

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

@@ -1 +1 @@
-{"version":3,"file":"particle-rule.js","sourceRoot":"","sources":["../../src/decode/particle-rule.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AACH,OAAO,EAGL,KAAK,EACL,YAAY,GACb,MAAM,YAAY,CAAC;AAEpB;;;GAGG;AACH,SAAS,cAAc,CACrB,OAAoB,EACpB,UAA+B;IAE/B,MAAM,EAAE,IAAI,EAAE,GAAG,OAAO,CAAC;IACzB,OAAO,CACL,IAAI,CAAC,EAAE,GAAG,IAAI,CAAC,IAAI,IAAI,CAAC;QACxB,IAAI,CAAC,YAAY,KAAK,EAAE;QACxB,UAAU,CAAC,GAAG,CAAC,OAAO,CAAC,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CACpD,CAAC;AACJ,CAAC;AAED;;;;;;GAMG;AACH,MAAM,gBAAgB,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;AAElD;;GAEG;AACH,MAAM,QAAQ,GAAG,GAAG,CAAC;AAErB;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAM,CAAC,MAAM,kBAAkB,GAAa;IAC1C,IAAI,EAAE,oBAAoB;IAC1B,UAAU,EAAE,CAAC,OAAoB,EAAE,EAAE;QACnC,IAAI,CAAC,cAAc,CAAC,OAAO,EAAE,gBAAgB,CAAC,EAAE,CAAC;YAC/C,OAAO,MAAM,CAAC;QAChB,CAAC;QACD,MAAM,MAAM,GAAG,YAAY,CAAC,OAAO,EAAE,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACxD,OAAO,KAAK,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC,UAAU,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,MAAM,CAAC;IACzE,CAAC;CACF,CAAC;AAEF,iEAAiE;AACjE,MAAM,EAAE,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;AAE1B;;;;;;GAMG;AACH,MAAM,cAAc,GAAG,aAAa,CAAC;AAErC;;;;;;;;;GASG;AACH,MAAM,CAAC,MAAM,YAAY,GAAa;IACpC,IAAI,EAAE,cAAc;IACpB,UAAU,EAAE,CAAC,OAAoB,EAAE,EAAE;QACnC,IAAI,CAAC,cAAc,CAAC,OAAO,EAAE,EAAE,CAAC,EAAE,CAAC;YACjC,OAAO,MAAM,CAAC;QAChB,CAAC;QACD,MAAM,MAAM,GAAG,YAAY,CAAC,OAAO,EAAE,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACxD,OAAO,cAAc,CAAC,IAAI,CAAC,KAAK,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,MAAM,CAAC;IACzE,CAAC;CACF,CAAC"}
+{"version":3,"file":"particle-rule.js","sourceRoot":"","sources":["../../src/decode/particle-rule.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AACH,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AACvD,OAAO,EAGL,KAAK,EACL,YAAY,GACb,MAAM,YAAY,CAAC;AAEpB;;;GAGG;AACH,SAAS,cAAc,CACrB,OAAoB,EACpB,UAA+B;IAE/B,MAAM,EAAE,IAAI,EAAE,GAAG,OAAO,CAAC;IACzB,OAAO,CACL,IAAI,CAAC,EAAE,GAAG,IAAI,CAAC,IAAI,IAAI,CAAC;QACxB,IAAI,CAAC,YAAY,KAAK,EAAE;QACxB,UAAU,CAAC,GAAG,CAAC,OAAO,CAAC,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CACpD,CAAC;AACJ,CAAC;AAED;;;;;;GAMG;AACH,MAAM,gBAAgB,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;AAElD;;GAEG;AACH,MAAM,QAAQ,GAAG,GAAG,CAAC;AAErB;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAM,CAAC,MAAM,kBAAkB,GAAa;IAC1C,IAAI,EAAE,oBAAoB;IAC1B,UAAU,EAAE,CAAC,OAAoB,EAAE,EAAE;QACnC,IAAI,CAAC,cAAc,CAAC,OAAO,EAAE,gBAAgB,CAAC,EAAE,CAAC;YAC/C,OAAO,MAAM,CAAC;QAChB,CAAC;QACD,MAAM,MAAM,GAAG,YAAY,CAAC,OAAO,EAAE,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACxD,OAAO,KAAK,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC,UAAU,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,MAAM,CAAC;IACzE,CAAC;CACF,CAAC;AAEF,iEAAiE;AACjE,MAAM,EAAE,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;AAE1B;;;;;;GAMG;AACH,MAAM,cAAc,GAAG,aAAa,CAAC;AAErC,qEAAqE;AACrE,MAAM,GAAG,GAAG,GAAG,CAAC;AAEhB;;;GAGG;AACH,MAAM,WAAW,GAAG,aAAa,CAAC;AAElC;;;GAGG;AACH,MAAM,uBAAuB,GAAG,IAAI,GAAG,CAAC;IACtC,IAAI;IACJ,IAAI;IACJ,IAAI;IACJ,IAAI;IACJ,IAAI;IACJ,IAAI;IACJ,IAAI;IACJ,IAAI;CACL,CAAC,CAAC;AAEH;;;;;;;;;GASG;AACH,SAAS,oBAAoB,CAAC,OAAoB,EAAE,EAAU;IAC5D,MAAM,IAAI,GAAG,YAAY,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC;IACvC,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;QACvB,OAAO,KAAK,CAAC;IACf,CAAC;IACD,MAAM,UAAU,GAAG,YAAY,CAAC,IAAI,CAAC,CAAC;IACtC,MAAM,MAAM,GAAG,UAAU,CAAC,MAAM,CAAC;IACjC,IACE,UAAU,CAAC,CAAC,CAAC,KAAK,GAAG;QACrB,MAAM,IAAI,CAAC;QACX,OAAO,CAAC,UAAU,CAAC,KAAK,CAAC,EAAE,GAAG,CAAC,GAAG,MAAM,EAAE,EAAE,GAAG,MAAM,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,KAAK,IAAI,EACxE,CAAC;QACD,OAAO,IAAI,CAAC;IACd,CAAC;IACD,MAAM,SAAS,GACb,CAAC,MAAM,KAAK,CAAC,IAAI,MAAM,KAAK,CAAC,CAAC;QAC9B,UAAU,CAAC,CAAC,CAAC,KAAK,UAAU,CAAC,CAAC,CAAC;QAC/B,UAAU,CAAC,CAAC,CAAC,KAAK,UAAU,CAAC,CAAC,CAAC,CAAC;IAClC,OAAO,CACL,SAAS;QACT,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC;QACtC,CAAC,uBAAuB,CAAC,GAAG,CAAC,IAAI,CAAC,CACnC,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAM,CAAC,MAAM,YAAY,GAAa;IACpC,IAAI,EAAE,cAAc;IACpB,UAAU,EAAE,CAAC,OAAoB,EAAE,EAAE;QACnC,MAAM,EAAE,UAAU,EAAE,IAAI,EAAE,GAAG,OAAO,CAAC;QACrC,IAAI,IAAI,CAAC,EAAE,GAAG,IAAI,CAAC,IAAI,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,GAAG,CAAC,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC;YACpE,OAAO,MAAM,CAAC;QAChB,CAAC;QACD,IAAI,oBAAoB,CAAC,OAAO,EAAE,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YAC7C,OAAO,QAAQ,CAAC;QAClB,CAAC;QACD,IAAI,CAAC,cAAc,CAAC,OAAO,EAAE,EAAE,CAAC,EAAE,CAAC;YACjC,OAAO,MAAM,CAAC;QAChB,CAAC;QACD,MAAM,MAAM,GAAG,YAAY,CAAC,OAAO,EAAE,IAAI,CAAC,IAAI,CAAC,CAAC;QAChD,OAAO,cAAc,CAAC,IAAI,CAAC,KAAK,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,MAAM,CAAC;IACzE,CAAC;CACF,CAAC"}

package/dist/decode/script-context.d.ts

@@ -0,0 +1,28 @@
+/**
+ * Revisiting a converted word in the light of the run around it.
+ *
+ * The character table decides each character alone, and for a few the words
+ * either side are what settle it. `script-words` converts a word on its own
+ * evidence, and this is what follows it.
+ */
+import type { Dictionary } from "../dictionary/dictionary.js";
+import type { CharacterConversion } from "../script/conversion.js";
+import type { Syllable } from "../syllable/syllable.js";
+import type { ScriptChoice } from "./script-types.js";
+/**
+ * What a word in a run is converted against.
+ */
+export interface RunTables {
+    readonly dictionary: Dictionary;
+    readonly table: ReadonlyMap<string, CharacterConversion>;
+    readonly words: ReadonlyMap<string, string>;
+}
+/**
+ * Convert a word, then revisit its choices in the light of the run around it.
+ *
+ * The character table decides 干 and 里 alone, and a frame around the word can
+ * overrule it: 与你无干 keeps 干 and 镜子里 writes 裡. `from` is where the
+ * word starts in `characters`, which are the whole run's.
+ */
+export declare function convertInRun(tables: RunTables, word: string, reading: readonly Syllable[], characters: readonly string[], from: number): readonly ScriptChoice[];
+//# sourceMappingURL=script-context.d.ts.map

package/dist/decode/script-context.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"script-context.d.ts","sourceRoot":"","sources":["../../src/decode/script-context.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,6BAA6B,CAAC;AAC9D,OAAO,KAAK,EAAE,mBAAmB,EAAE,MAAM,yBAAyB,CAAC;AACnE,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAIxD,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,mBAAmB,CAAC;AAEtD;;GAEG;AACH,MAAM,WAAW,SAAS;IACxB,QAAQ,CAAC,UAAU,EAAE,UAAU,CAAC;IAChC,QAAQ,CAAC,KAAK,EAAE,WAAW,CAAC,MAAM,EAAE,mBAAmB,CAAC,CAAC;IACzD,QAAQ,CAAC,KAAK,EAAE,WAAW,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;CAC7C;AAED;;;;;;GAMG;AACH,wBAAgB,YAAY,CAC1B,MAAM,EAAE,SAAS,EACjB,IAAI,EAAE,MAAM,EACZ,OAAO,EAAE,SAAS,QAAQ,EAAE,EAC5B,UAAU,EAAE,SAAS,MAAM,EAAE,EAC7B,IAAI,EAAE,MAAM,GACX,SAAS,YAAY,EAAE,CAUzB"}

package/dist/decode/script-context.js

@@ -0,0 +1,45 @@
+import { CONCERNING, isConcerningAt, isGan } from "./gan-frames.js";
+import { writeInside } from "./li-frames.js";
+import { convertWord } from "./script-words.js";
+/**
+ * Convert a word, then revisit its choices in the light of the run around it.
+ *
+ * The character table decides 干 and 里 alone, and a frame around the word can
+ * overrule it: 与你无干 keeps 干 and 镜子里 writes 裡. `from` is where the
+ * word starts in `characters`, which are the whole run's.
+ */
+export function convertInRun(tables, word, reading, characters, from) {
+    const converted = convertWord(tables.table, tables.words, word, reading);
+    const concerning = keepConcerning(converted, tables.dictionary, characters, from, reading);
+    return writeInside(concerning, tables.dictionary, characters, from);
+}
+/**
+ * Keep 干 where it means "to concern".
+ *
+ * 干 read `gān` is 乾 when it means dry (干燥, 干货) and stays 干 when it means
+ * to concern (干扰, 相干), and the syllable cannot tell the two apart. A word
+ * settles most of them. A 干 standing alone at `gān` fell back on 乾, which
+ * wrote 与你无干 as 與你無乾 and 干卿底事 as 乾卿底事. The frames that read
+ * it `gān` in the first place are the ones that decide it here. See
+ * {@link isConcerningAt}.
+ *
+ * `from` is where the word starts in `characters`, which are the whole run's,
+ * since the frames reach past the word.
+ */
+function keepConcerning(choices, dictionary, characters, from, reading) {
+    if (reading.length !== choices.length) {
+        return choices;
+    }
+    return choices.map((choice, at) => choice.from === CONCERNING &&
+        choice.to !== CONCERNING &&
+        isGan(reading[at]) &&
+        isConcerningAt(dictionary, characters, from + at)
+        ? {
+            from: choice.from,
+            to: CONCERNING,
+            evidence: "reading",
+            alternatives: [choice.to, ...choice.alternatives].filter((form) => form !== CONCERNING),
+        }
+        : choice);
+}
+//# sourceMappingURL=script-context.js.map

package/dist/decode/script-context.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"script-context.js","sourceRoot":"","sources":["../../src/decode/script-context.ts"],"names":[],"mappings":"AAUA,OAAO,EAAE,UAAU,EAAE,cAAc,EAAE,KAAK,EAAE,MAAM,iBAAiB,CAAC;AACpE,OAAO,EAAE,WAAW,EAAE,MAAM,gBAAgB,CAAC;AAC7C,OAAO,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAYhD;;;;;;GAMG;AACH,MAAM,UAAU,YAAY,CAC1B,MAAiB,EACjB,IAAY,EACZ,OAA4B,EAC5B,UAA6B,EAC7B,IAAY;IAEZ,MAAM,SAAS,GAAG,WAAW,CAAC,MAAM,CAAC,KAAK,EAAE,MAAM,CAAC,KAAK,EAAE,IAAI,EAAE,OAAO,CAAC,CAAC;IACzE,MAAM,UAAU,GAAG,cAAc,CAC/B,SAAS,EACT,MAAM,CAAC,UAAU,EACjB,UAAU,EACV,IAAI,EACJ,OAAO,CACR,CAAC;IACF,OAAO,WAAW,CAAC,UAAU,EAAE,MAAM,CAAC,UAAU,EAAE,UAAU,EAAE,IAAI,CAAC,CAAC;AACtE,CAAC;AAED;;;;;;;;;;;;GAYG;AACH,SAAS,cAAc,CACrB,OAAgC,EAChC,UAAsB,EACtB,UAA6B,EAC7B,IAAY,EACZ,OAA4B;IAE5B,IAAI,OAAO,CAAC,MAAM,KAAK,OAAO,CAAC,MAAM,EAAE,CAAC;QACtC,OAAO,OAAO,CAAC;IACjB,CAAC;IACD,OAAO,OAAO,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,EAAE,EAAE,EAAE,CAChC,MAAM,CAAC,IAAI,KAAK,UAAU;QAC1B,MAAM,CAAC,EAAE,KAAK,UAAU;QACxB,KAAK,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC;QAClB,cAAc,CAAC,UAAU,EAAE,UAAU,EAAE,IAAI,GAAG,EAAE,CAAC;QAC/C,CAAC,CAAC;YACE,IAAI,EAAE,MAAM,CAAC,IAAI;YACjB,EAAE,EAAE,UAAU;YACd,QAAQ,EAAE,SAAS;YACnB,YAAY,EAAE,CAAC,MAAM,CAAC,EAAE,EAAE,GAAG,MAAM,CAAC,YAAY,CAAC,CAAC,MAAM,CACtD,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,KAAK,UAAU,CAC9B;SACF;QACH,CAAC,CAAC,MAAM,CACX,CAAC;AACJ,CAAC"}

package/dist/decode/script-words.d.ts

@@ -6,7 +6,6 @@
  * can show where the conversion was guessing.
  */
 import type { ScriptChoice, ScriptEvidence, ScriptTarget } from "./script-types.js";
-import type { Dictionary } from "../dictionary/dictionary.js";
 import { type CharacterConversion } from "../script/conversion.js";
 import { type Region } from "../script/glyphs.js";
 import type { Script } from "../script/script.js";
@@ -23,19 +22,10 @@ export declare function targetOf(target: ScriptTarget): {
  */
 export declare function convertWord(table: ReadonlyMap<string, CharacterConversion>, words: ReadonlyMap<string, string>, word: string, reading: readonly Syllable[]): readonly ScriptChoice[];
 /**
- * Keep 干 where it means "to concern".
- *
- * 干 read `gān` is 乾 when it means dry (干燥, 干货) and stays 干 when it means
- * to concern (干扰, 相干), and the syllable cannot tell the two apart. A word
- * settles most of them. A 干 standing alone at `gān` fell back on 乾, which
- * wrote 与你无干 as 與你無乾 and 干卿底事 as 乾卿底事. The frames that read
- * it `gān` in the first place are the ones that decide it here. See
- * {@link isConcerningAt}.
- *
- * `from` is where the word starts in `characters`, which are the whole run's,
- * since the frames reach past the word.
+ * Keep a word as it stands, which is what a text already in the target script
+ * needs. Every character is locked, because none of them was in question.
  */
-export declare function keepConcerning(choices: readonly ScriptChoice[], dictionary: Dictionary, characters: readonly string[], from: number, reading: readonly Syllable[]): readonly ScriptChoice[];
+export declare function keepWord(word: string): readonly ScriptChoice[];
 /**
  * Rank what settled a character, strongest evidence first.
  *

package/dist/decode/script-words.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"script-words.d.ts","sourceRoot":"","sources":["../../src/decode/script-words.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,KAAK,EACV,YAAY,EACZ,cAAc,EACd,YAAY,EACb,MAAM,mBAAmB,CAAC;AAC3B,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,6BAA6B,CAAC;AAE9D,OAAO,EACL,KAAK,mBAAmB,EAIzB,MAAM,yBAAyB,CAAC;AACjC,OAAO,EAAkB,KAAK,MAAM,EAAE,MAAM,qBAAqB,CAAC;AAClE,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,qBAAqB,CAAC;AAClD,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAGxD;;GAEG;AACH,wBAAgB,QAAQ,CAAC,MAAM,EAAE,YAAY,GAAG;IAC9C,MAAM,EAAE,MAAM,CAAC;IACf,MAAM,EAAE,MAAM,CAAC;CAChB,CAQA;AAED;;GAEG;AACH,wBAAgB,WAAW,CACzB,KAAK,EAAE,WAAW,CAAC,MAAM,EAAE,mBAAmB,CAAC,EAC/C,KAAK,EAAE,WAAW,CAAC,MAAM,EAAE,MAAM,CAAC,EAClC,IAAI,EAAE,MAAM,EACZ,OAAO,EAAE,SAAS,QAAQ,EAAE,GAC3B,SAAS,YAAY,EAAE,CAuCzB;AAED;;;;;;;;;;;;GAYG;AACH,wBAAgB,cAAc,CAC5B,OAAO,EAAE,SAAS,YAAY,EAAE,EAChC,UAAU,EAAE,UAAU,EACtB,UAAU,EAAE,SAAS,MAAM,EAAE,EAC7B,IAAI,EAAE,MAAM,EACZ,OAAO,EAAE,SAAS,QAAQ,EAAE,GAC3B,SAAS,YAAY,EAAE,CAmBzB;AAED;;;;;;;;GAQG;AACH,wBAAgB,WAAW,CACzB,UAAU,EAAE,OAAO,EACnB,WAAW,EAAE,OAAO,EACpB,QAAQ,EAAE,OAAO,GAChB,cAAc,CAQhB"}
+{"version":3,"file":"script-words.d.ts","sourceRoot":"","sources":["../../src/decode/script-words.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,KAAK,EACV,YAAY,EACZ,cAAc,EACd,YAAY,EACb,MAAM,mBAAmB,CAAC;AAE3B,OAAO,EACL,KAAK,mBAAmB,EAIzB,MAAM,yBAAyB,CAAC;AACjC,OAAO,EAAkB,KAAK,MAAM,EAAE,MAAM,qBAAqB,CAAC;AAClE,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,qBAAqB,CAAC;AAClD,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAExD;;GAEG;AACH,wBAAgB,QAAQ,CAAC,MAAM,EAAE,YAAY,GAAG;IAC9C,MAAM,EAAE,MAAM,CAAC;IACf,MAAM,EAAE,MAAM,CAAC;CAChB,CAQA;AAED;;GAEG;AACH,wBAAgB,WAAW,CACzB,KAAK,EAAE,WAAW,CAAC,MAAM,EAAE,mBAAmB,CAAC,EAC/C,KAAK,EAAE,WAAW,CAAC,MAAM,EAAE,MAAM,CAAC,EAClC,IAAI,EAAE,MAAM,EACZ,OAAO,EAAE,SAAS,QAAQ,EAAE,GAC3B,SAAS,YAAY,EAAE,CAuCzB;AAED;;;GAGG;AACH,wBAAgB,QAAQ,CAAC,IAAI,EAAE,MAAM,GAAG,SAAS,YAAY,EAAE,CAO9D;AAED;;;;;;;;GAQG;AACH,wBAAgB,WAAW,CACzB,UAAU,EAAE,OAAO,EACnB,WAAW,EAAE,OAAO,EACpB,QAAQ,EAAE,OAAO,GAChB,cAAc,CAQhB"}