@kensio/pinyinjs 1.7.4 → 1.7.5

package/dist/numerals/time.js

@@ -0,0 +1,93 @@
+/**
+ * Saying a clock time out loud.
+ *
+ * A time is not read the way the same digits would be anywhere else: 6:30 is
+ * 六点三十分 rather than the digits in order, the 分 is written even where a
+ * speaker would drop it, and the hour takes 两 where a bare 2 takes 二.
+ */
+import { numeralHanzi, readNumeralHanzi, } from "./numerals.js";
+import { HOURS_IN_A_DAY, MINUTES_IN_AN_HOUR, TIME } from "./numeral-shapes.js";
+/**
+ * The minute below which the zero is said: 6:05 is 六点零五分.
+ */
+const SPOKEN_ZERO_BELOW = 10;
+/**
+ * The hour that is 两 rather than 二.
+ *
+ * Two o'clock is 两点 and never 二点, which is the same 两 that counts things —
+ * an hour is a quantity of hours. It is the only hour this applies to, since
+ * 12:00 is 十二点 with the 二 inside a larger number.
+ */
+const LIANG_HOUR = 2;
+/**
+ * A time written out in 汉字: 6:30 is 六点三十分.
+ *
+ * The 分 is written even though a speaker often drops it, because without it
+ * 六点三十 is the decimal 6.30 said aloud — the same 点 does both jobs, and the
+ * 分 is the only thing that separates them. On the hour it is left off
+ * instead, since 六点零零分 is not something anybody says.
+ */
+export function timeHanzi(hours, minutes, options) {
+    const counted = { ...options, style: "cardinal" };
+    const hour = hours === LIANG_HOUR ? "两" : numeralHanzi(hours, counted);
+    const minute = minutes === 0 ? "" : numeralHanzi(minutes, counted);
+    /* c8 ignore next 3 -- an hour is 0 to 23 and a minute 0 to 59, and the
+       cardinal reader counts every one of them */
+    if (hour === undefined || minute === undefined) {
+        return undefined;
+    }
+    if (minute === "") {
+        return `${hour}点`;
+    }
+    const zero = minutes < SPOKEN_ZERO_BELOW ? "零" : "";
+    return `${hour}点${zero}${minute}分`;
+}
+/**
+ * Where a written-out time breaks into words.
+ *
+ * The hour, the 点, the minutes and the 分, which is how the same time reads
+ * when it is written 6点30分 in the first place.
+ */
+export function timeWords(hanzi) {
+    const [hour = "", rest = ""] = hanzi.split("点");
+    const minutes = rest.replace("分", "");
+    return minutes === ""
+        ? [hour.length, 1]
+        : [hour.length, 1, minutes.length, 1];
+}
+/**
+ * Read a digit run as a time, where that is what it is.
+ *
+ * Tried before the identifier rule, which would otherwise leave the whole
+ * thing alone: a colon between digits is exactly the mark that makes 6:30 a
+ * time rather than two numbers.
+ */
+export function readTime(text, from, options) {
+    TIME.lastIndex = from;
+    const found = TIME.exec(text);
+    const written = found?.[1];
+    const minutes = found?.[2];
+    if (found === null || written === undefined || minutes === undefined) {
+        return undefined;
+    }
+    const hourValue = Number(written);
+    const minuteValue = Number(minutes);
+    if (hourValue >= HOURS_IN_A_DAY || minuteValue >= MINUTES_IN_AN_HOUR) {
+        return undefined;
+    }
+    const hanzi = timeHanzi(hourValue, minuteValue, options);
+    const reading = hanzi === undefined ? undefined : readNumeralHanzi(hanzi, options);
+    /* c8 ignore next 3 -- the hanzi is built out of the same table that reads it
+       back, so it reads whenever it was built at all */
+    if (hanzi === undefined || reading === undefined) {
+        return undefined;
+    }
+    return {
+        text: found[0],
+        reading,
+        hanzi,
+        style: "cardinal",
+        words: timeWords(hanzi),
+    };
+}
+//# sourceMappingURL=time.js.map

package/dist/numerals/time.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"time.js","sourceRoot":"","sources":["../../src/numerals/time.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,EACL,YAAY,EAEZ,gBAAgB,GACjB,MAAM,eAAe,CAAC;AACvB,OAAO,EAAE,cAAc,EAAE,kBAAkB,EAAE,IAAI,EAAE,MAAM,qBAAqB,CAAC;AAG/E;;GAEG;AACH,MAAM,iBAAiB,GAAG,EAAE,CAAC;AAE7B;;;;;;GAMG;AACH,MAAM,UAAU,GAAG,CAAC,CAAC;AAErB;;;;;;;GAOG;AACH,MAAM,UAAU,SAAS,CACvB,KAAa,EACb,OAAe,EACf,OAAuB;IAEvB,MAAM,OAAO,GAAmB,EAAE,GAAG,OAAO,EAAE,KAAK,EAAE,UAAU,EAAE,CAAC;IAClE,MAAM,IAAI,GAAG,KAAK,KAAK,UAAU,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,YAAY,CAAC,KAAK,EAAE,OAAO,CAAC,CAAC;IACvE,MAAM,MAAM,GAAG,OAAO,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,YAAY,CAAC,OAAO,EAAE,OAAO,CAAC,CAAC;IACnE;kDAC8C;IAC9C,IAAI,IAAI,KAAK,SAAS,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;QAC/C,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,IAAI,MAAM,KAAK,EAAE,EAAE,CAAC;QAClB,OAAO,GAAG,IAAI,GAAG,CAAC;IACpB,CAAC;IACD,MAAM,IAAI,GAAG,OAAO,GAAG,iBAAiB,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC;IACpD,OAAO,GAAG,IAAI,IAAI,IAAI,GAAG,MAAM,GAAG,CAAC;AACrC,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,SAAS,CAAC,KAAa;IACrC,MAAM,CAAC,IAAI,GAAG,EAAE,EAAE,IAAI,GAAG,EAAE,CAAC,GAAG,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;IAChD,MAAM,OAAO,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC;IACtC,OAAO,OAAO,KAAK,EAAE;QACnB,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,EAAE,CAAC,CAAC;QAClB,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,OAAO,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC;AAC1C,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,QAAQ,CACtB,IAAY,EACZ,IAAY,EACZ,OAAuB;IAEvB,IAAI,CAAC,SAAS,GAAG,IAAI,CAAC;IACtB,MAAM,KAAK,GAAG,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAC9B,MAAM,OAAO,GAAG,KAAK,EAAE,CAAC,CAAC,CAAC,CAAC;IAC3B,MAAM,OAAO,GAAG,KAAK,EAAE,CAAC,CAAC,CAAC,CAAC;IAC3B,IAAI,KAAK,KAAK,IAAI,IAAI,OAAO,KAAK,SAAS,IAAI,OAAO,KAAK,SAAS,EAAE,CAAC;QACrE,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,MAAM,SAAS,GAAG,MAAM,CAAC,OAAO,CAAC,CAAC;IAClC,MAAM,WAAW,GAAG,MAAM,CAAC,OAAO,CAAC,CAAC;IACpC,IAAI,SAAS,IAAI,cAAc,IAAI,WAAW,IAAI,kBAAkB,EAAE,CAAC;QACrE,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,MAAM,KAAK,GAAG,SAAS,CAAC,SAAS,EAAE,WAAW,EAAE,OAAO,CAAC,CAAC;IACzD,MAAM,OAAO,GACX,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,gBAAgB,CAAC,KAAK,EAAE,OAAO,CAAC,CAAC;IACrE;wDACoD;IACpD,IAAI,KAAK,KAAK,SAAS,IAAI,OAAO,KAAK,SAAS,EAAE,CAAC;QACjD,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,OAAO;QACL,IAAI,EAAE,KAAK,CAAC,CAAC,CAAC;QACd,OAAO;QACP,KAAK;QACL,KAAK,EAAE,UAAU;QACjB,KAAK,EAAE,SAAS,CAAC,KAAK,CAAC;KACxB,CAAC;AACJ,CAAC"}

package/dist/orthography/address-grouping.d.ts

@@ -0,0 +1,34 @@
+import type { GroupingRule } from "./rule.js";
+/**
+ * The 称呼语 GB/T 16159 writes in front of a surname.
+ *
+ * 老 and 小 only. 大 is written the same way in 大李, but it is also an ordinary
+ * adjective in front of anything at all, and it is the one that goes wrong:
+ * over 88,866 lines the three of them fire 49 times together and both clear
+ * mistakes are 大 — 泡大池 is a big pool and 那头大熊 is a big bear. Dropping it
+ * leaves 38 firings and every one visible in the sample is a real form of
+ * address.
+ */
+export declare const ADDRESS_PREFIXES: Set<string>;
+/**
+ * 老王 is `Lǎo Wáng`: the prefix takes a capital of its own.
+ *
+ * The words are already written apart — nothing here moves a boundary — and
+ * what was missing is only the capital, so this marks the prefix a proper noun
+ * and lets the writer do what it already does with one. That is why it is a
+ * grouping rule despite grouping nothing: `isProperNoun` is the flag the
+ * capital hangs off, and there is one place it is set.
+ *
+ * The surname is the evidence. A one-character word the dictionary marks a
+ * proper noun is what 老 and 小 attach to, and CC-CEDICT's capitalisation has
+ * already vetoed the tags that would otherwise let 小 in front of anything —
+ * see "jieba's 专名 tags need a second opinion" in ROADMAP.md.
+ */
+export declare const ADDRESS_PREFIX: GroupingRule;
+/**
+ * jieba's tags for the two kinds of proper noun 5.1 divides.
+ *
+ * A person and an organisation. Places are `PLACE_GENERICS`' business, and `nz`
+ * is left out — both are measured out in `docs/orthography/`.
+ */
+//# sourceMappingURL=address-grouping.d.ts.map

package/dist/orthography/address-grouping.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"address-grouping.d.ts","sourceRoot":"","sources":["../../src/orthography/address-grouping.ts"],"names":[],"mappings":"AAMA,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,WAAW,CAAC;AAE9C;;;;;;;;;GASG;AACH,eAAO,MAAM,gBAAgB,aAAsB,CAAC;AAEpD;;;;;;;;;;;;;GAaG;AACH,eAAO,MAAM,cAAc,EAAE,YAgB5B,CAAC;AAEF;;;;;GAKG"}

package/dist/orthography/address-grouping.js

@@ -0,0 +1,52 @@
+/**
+ * Joining 老 and 小 to the name they address.
+ *
+ * 老王 is one word and 老 王 is two, and only the prefix says which.
+ */
+import { characterCount } from "../script/characters.js";
+/**
+ * The 称呼语 GB/T 16159 writes in front of a surname.
+ *
+ * 老 and 小 only. 大 is written the same way in 大李, but it is also an ordinary
+ * adjective in front of anything at all, and it is the one that goes wrong:
+ * over 88,866 lines the three of them fire 49 times together and both clear
+ * mistakes are 大 — 泡大池 is a big pool and 那头大熊 is a big bear. Dropping it
+ * leaves 38 firings and every one visible in the sample is a real form of
+ * address.
+ */
+export const ADDRESS_PREFIXES = new Set(["老", "小"]);
+/**
+ * 老王 is `Lǎo Wáng`: the prefix takes a capital of its own.
+ *
+ * The words are already written apart — nothing here moves a boundary — and
+ * what was missing is only the capital, so this marks the prefix a proper noun
+ * and lets the writer do what it already does with one. That is why it is a
+ * grouping rule despite grouping nothing: `isProperNoun` is the flag the
+ * capital hangs off, and there is one place it is set.
+ *
+ * The surname is the evidence. A one-character word the dictionary marks a
+ * proper noun is what 老 and 小 attach to, and CC-CEDICT's capitalisation has
+ * already vetoed the tags that would otherwise let 小 in front of anything —
+ * see "jieba's 专名 tags need a second opinion" in ROADMAP.md.
+ */
+export const ADDRESS_PREFIX = {
+    name: "address-prefix",
+    apply: (words) => words.map((word, at) => {
+        const next = words[at + 1];
+        if (!ADDRESS_PREFIXES.has(word.text) ||
+            next === undefined ||
+            !next.isProperNoun ||
+            characterCount(next.text) !== 1 ||
+            next.separator !== undefined) {
+            return word;
+        }
+        return { ...word, isProperNoun: true };
+    }),
+};
+/**
+ * jieba's tags for the two kinds of proper noun 5.1 divides.
+ *
+ * A person and an organisation. Places are `PLACE_GENERICS`' business, and `nz`
+ * is left out — both are measured out in `docs/orthography/`.
+ */
+//# sourceMappingURL=address-grouping.js.map

package/dist/orthography/address-grouping.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"address-grouping.js","sourceRoot":"","sources":["../../src/orthography/address-grouping.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AACH,OAAO,EAAE,cAAc,EAAE,MAAM,yBAAyB,CAAC;AAGzD;;;;;;;;;GASG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;AAEpD;;;;;;;;;;;;;GAaG;AACH,MAAM,CAAC,MAAM,cAAc,GAAiB;IAC1C,IAAI,EAAE,gBAAgB;IACtB,KAAK,EAAE,CAAC,KAAK,EAAE,EAAE,CACf,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,EAAE,EAAE;QACrB,MAAM,IAAI,GAAG,KAAK,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;QAC3B,IACE,CAAC,gBAAgB,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC;YAChC,IAAI,KAAK,SAAS;YAClB,CAAC,IAAI,CAAC,YAAY;YAClB,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC;YAC/B,IAAI,CAAC,SAAS,KAAK,SAAS,EAC5B,CAAC;YACD,OAAO,IAAI,CAAC;QACd,CAAC;QACD,OAAO,EAAE,GAAG,IAAI,EAAE,YAAY,EAAE,IAAI,EAAE,CAAC;IACzC,CAAC,CAAC;CACL,CAAC;AAEF;;;;;GAKG"}

package/dist/orthography/grouping.d.ts

@@ -1,14 +1,8 @@
 import type { Dictionary } from "../dictionary/dictionary.js";
 import type { DecodedWord } from "../decode/word.js";
-import { type GroupingRule } from "./rule.js";
-/**
- * 了, 着 and 过 attach to the verb in front of them.
- *
- * GB/T 16159: 他看了 is `tā kànle`, not `tā kàn le`. Safe as a rule because it
- * only ever joins what the decode had already separated — the dictionary has no
- * entry for 看了, so nothing the dictionary asserted is being overridden. The
- * readings do not change either way, so the worst case is untidy spacing.
- */
+import type { GroupingRule } from "./rule.js";
+export { ADDRESS_PREFIX, NAME_PARTS, SPACED_WORD_LIST, } from "./word-grouping.js";
+export { PLACE_GENERICS } from "./place-grouping.js";
 export declare const ASPECT_PARTICLES: GroupingRule;
 /**
  * A suffix attaches to the word in front of it.
@@ -18,84 +12,6 @@ export declare const ASPECT_PARTICLES: GroupingRule;
  * dictionary attached to anything.
  */
 export declare const SUFFIXES: GroupingRule;
-/**
- * A place name's generic half is written separately, and capitalised.
- *
- * GB/T 16159 专名和通名分写: 南京市 is `Nánjīng Shì`, 西湖区 is `Xīhú Qū`.
- *
- * Two conditions keep it from running away, both of them measured against the
- * whole dictionary rather than guessed. The word must be tagged a place name,
- * and **the part before the generic must itself be a word** — without the
- * second, 上山下乡 and 三街六市 are tagged `ns` and come apart as
- * `Shàngshānxià Xiāng` and `Sānjiēliù Shì`. Together they fire on 4,798 of the
- * dictionary's 8,781 candidates and hold back 427.
- *
- * The rule deliberately never touches a two-character name, so 上海 stays
- * `Shànghǎi` and 长城 stays `Chángchéng`: there, the generic character is part
- * of the proper name rather than a description of it, and no rule can tell the
- * two apart — 黄河 is `Huáng Hé` and 青海 is `Qīnghǎi`.
- */
-export declare const PLACE_GENERICS: GroupingRule;
-/**
- * Write the words GB/T 16159 spaces in a way no rule reaches.
- *
- * Matches longest-first at each word boundary, and only where the entry ends on
- * a word boundary too: a table entry is a claim about a whole word or several,
- * not about the middle of one.
- */
-export declare const SPACED_WORD_LIST: GroupingRule;
-/**
- * 老王 is `Lǎo Wáng`: the prefix takes a capital of its own.
- *
- * The words are already written apart — nothing here moves a boundary — and
- * what was missing is only the capital, so this marks the prefix a proper noun
- * and lets the writer do what it already does with one. That is why it is a
- * grouping rule despite grouping nothing: `isProperNoun` is the flag the
- * capital hangs off, and there is one place it is set.
- *
- * The surname is the evidence. A one-character word the dictionary marks a
- * proper noun is what 老 and 小 attach to, and CC-CEDICT's capitalisation has
- * already vetoed the tags that would otherwise let 小 in front of anything —
- * see "jieba's 专名 tags need a second opinion" in ROADMAP.md.
- */
-export declare const ADDRESS_PREFIX: GroupingRule;
-/**
- * The parts of a proper name are written apart: 齐白石 is `Qí Báishí`.
- *
- * GB/T 16159 5.1 writes 姓 apart from 名 and a proper noun apart from its
- * generic, each part capitalised. Both halves of that clause are this rule,
- * because the evidence for them is the same evidence.
- *
- * 齐白石 and 北京大学 are dictionary entries, so the decoder produces **one**
- * word each — and reading them as one word is what makes them read correctly at
- * all. That makes this a *split*, and splitting contradicts the dictionary's own
- * claim that the characters belong together. It needs a condition strong enough
- * to survive the whole dictionary, and the obvious ones are not: a surname list
- * takes 马克思, 高尔基, 巴赫 and 牛顿 apart, none of which is a Chinese name,
- * and a list of generics cannot say where 上海浦东发展银行 divides.
- *
- * **The condition is CC-CEDICT's own capitalisation**, which states the
- * boundaries outright: 齐白石 is `[Qi2 Bai2 shi2]`, 司马迁 `[Si1 ma3 Qian1]`,
- * 上海交通大学 `[Shang4 hai3 Jiao1 tong1 Da4 xue2]`, and 马克思 `[Ma3 ke4 si1]`
- * with no second capital at all. So a compound surname is recognised without a
- * list of compound surnames, a generic without a list of generics, and a
- * transliteration is excluded without a list of transliterations. It is the same
- * source and the same signal the proper-noun veto already trusts — see "jieba's
- * 专名 tags need a second opinion" in ROADMAP.md — extended from *whether* a
- * word is a proper noun to *where* its parts divide.
- *
- * **Every stated boundary is cut, not only the first**, which is what separates
- * an organisation from a person: 48% of `nt` entries carrying a boundary carry
- * more than one, against 1.6% of `nr`. One cut would leave 上海交通大学 as
- * `Shànghǎi Jiāotōngdàxué`.
- *
- * A tag is still required, because the mark is not confined to what 5.1 covers:
- * 5,341 `ns` entries carry one, where `PLACE_GENERICS` already applies a
- * measured condition, and `nz` carries 346 of which 第二次世界大战
- * `[Di4 er4 Ci4 Shi4 jie4 Da4 zhan4]` divides in the wrong place. Both are
- * measured out in `docs/orthography/`.
- */
-export declare const NAME_PARTS: GroupingRule;
 /**
  * The rules applied by default, in order.
  *

package/dist/orthography/grouping.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"grouping.d.ts","sourceRoot":"","sources":["../../src/orthography/grouping.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,6BAA6B,CAAC;AAC9D,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAIrD,OAAO,EAAY,KAAK,YAAY,EAAiB,MAAM,WAAW,CAAC;AAiEvE;;;;;;;GAOG;AACH,eAAO,MAAM,gBAAgB,EAAE,YAO9B,CAAC;AAEF;;;;;;GAMG;AACH,eAAO,MAAM,QAAQ,EAAE,YAItB,CAAC;AAEF;;;;;;;;;;;;;;;;GAgBG;AACH,eAAO,MAAM,cAAc,EAAE,YAsB5B,CAAC;AAgFF;;;;;;GAMG;AACH,eAAO,MAAM,gBAAgB,EAAE,YAmB9B,CAAC;AAcF;;;;;;;;;;;;;GAaG;AACH,eAAO,MAAM,cAAc,EAAE,YAgB5B,CAAC;AAUF;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAmCG;AACH,eAAO,MAAM,UAAU,EAAE,YAgBxB,CAAC;AAEF;;;;;GAKG;AACH,eAAO,MAAM,cAAc,EAAE,SAAS,YAAY,EAUjD,CAAC;AAEF;;GAEG;AACH,wBAAgB,aAAa,CAC3B,KAAK,EAAE,SAAS,WAAW,EAAE,EAC7B,UAAU,EAAE,UAAU,EACtB,KAAK,GAAE,SAAS,YAAY,EAAmB,GAC9C,SAAS,WAAW,EAAE,CAMxB"}
+{"version":3,"file":"grouping.d.ts","sourceRoot":"","sources":["../../src/orthography/grouping.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,6BAA6B,CAAC;AAC9D,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAGrD,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,WAAW,CAAC;AAO9C,OAAO,EACL,cAAc,EACd,UAAU,EACV,gBAAgB,GACjB,MAAM,oBAAoB,CAAC;AA0B5B,OAAO,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAC;AACrD,eAAO,MAAM,gBAAgB,EAAE,YAO9B,CAAC;AAEF;;;;;;GAMG;AACH,eAAO,MAAM,QAAQ,EAAE,YAItB,CAAC;AACF;;;;;GAKG;AACH,eAAO,MAAM,cAAc,EAAE,SAAS,YAAY,EAUjD,CAAC;AAEF;;GAEG;AACH,wBAAgB,aAAa,CAC3B,KAAK,EAAE,SAAS,WAAW,EAAE,EAC7B,UAAU,EAAE,UAAU,EACtB,KAAK,GAAE,SAAS,YAAY,EAAmB,GAC9C,SAAS,WAAW,EAAE,CAMxB"}

package/dist/orthography/grouping.js

@@ -1,8 +1,7 @@
-import { characterCount, toCharacters } from "../script/characters.js";
 import { IDIOM_HYPHENS } from "./idioms.js";
 import { AABB_REDUPLICATION, ABAB_REDUPLICATION } from "./reduplication.js";
-import { divideAt, join, splitAt } from "./rule.js";
-import { LONGEST_SPACED_WORD, SPACED_WORD_FORMS } from "./word-list.js";
+import { ADDRESS_PREFIX, NAME_PARTS, SPACED_WORD_LIST, } from "./word-grouping.js";
+export { ADDRESS_PREFIX, NAME_PARTS, SPACED_WORD_LIST, } from "./word-grouping.js";
 /**
  * jieba's tags for the aspect particles 了, 着 and 过.
  */
@@ -22,49 +21,8 @@ function isAspectHost(word) {
  * jieba's tag for a suffix, which is what 们 and 者 are.
  */
 const SUFFIX_TAG = "k";
-/**
- * The generic half of an administrative place name.
- *
- * Both scripts, since a 繁體 name is a key in its own right and never converted
- * before lookup.
- */
-const ADMINISTRATIVE_GENERICS = new Set([
-    "市",
-    "省",
-    "县",
-    "縣",
-    "区",
-    "區",
-    "镇",
-    "鎮",
-    "乡",
-    "鄉",
-    "村",
-    "州",
-]);
-/**
- * Fold every word a predicate accepts into the word before it.
- */
-function joinBackwards(words, isAttached) {
-    const grouped = [];
-    for (const word of words) {
-        const head = grouped.at(-1);
-        if (head !== undefined && isAttached(word, head)) {
-            grouped[grouped.length - 1] = join(head, word);
-            continue;
-        }
-        grouped.push(word);
-    }
-    return grouped;
-}
-/**
- * 了, 着 and 过 attach to the verb in front of them.
- *
- * GB/T 16159: 他看了 is `tā kànle`, not `tā kàn le`. Safe as a rule because it
- * only ever joins what the decode had already separated — the dictionary has no
- * entry for 看了, so nothing the dictionary asserted is being overridden. The
- * readings do not change either way, so the worst case is untidy spacing.
- */
+import { joinBackwards, PLACE_GENERICS } from "./place-grouping.js";
+export { PLACE_GENERICS } from "./place-grouping.js";
 export const ASPECT_PARTICLES = {
     name: "aspect-particles",
     apply: (words) => joinBackwards(words, (word, head) => ASPECT_TAGS.has(word.partOfSpeech) && isAspectHost(head)),
@@ -80,230 +38,6 @@ export const SUFFIXES = {
     name: "suffixes",
     apply: (words) => joinBackwards(words, (word) => word.partOfSpeech === SUFFIX_TAG),
 };
-/**
- * A place name's generic half is written separately, and capitalised.
- *
- * GB/T 16159 专名和通名分写: 南京市 is `Nánjīng Shì`, 西湖区 is `Xīhú Qū`.
- *
- * Two conditions keep it from running away, both of them measured against the
- * whole dictionary rather than guessed. The word must be tagged a place name,
- * and **the part before the generic must itself be a word** — without the
- * second, 上山下乡 and 三街六市 are tagged `ns` and come apart as
- * `Shàngshānxià Xiāng` and `Sānjiēliù Shì`. Together they fire on 4,798 of the
- * dictionary's 8,781 candidates and hold back 427.
- *
- * The rule deliberately never touches a two-character name, so 上海 stays
- * `Shànghǎi` and 长城 stays `Chángchéng`: there, the generic character is part
- * of the proper name rather than a description of it, and no rule can tell the
- * two apart — 黄河 is `Huáng Hé` and 青海 is `Qīnghǎi`.
- */
-export const PLACE_GENERICS = {
-    name: "place-generics",
-    apply: (words, dictionary) => words.flatMap((word) => {
-        const characters = toCharacters(word.text);
-        const generic = characters.at(-1) ?? "";
-        if (word.partOfSpeech !== "ns" ||
-            characters.length < 3 ||
-            !ADMINISTRATIVE_GENERICS.has(generic) ||
-            dictionary.lookup(characters.slice(0, -1).join("")) === undefined) {
-            return [word];
-        }
-        const split = splitAt(word, characters.length - 1);
-        if (split === undefined) {
-            return [word];
-        }
-        // Both halves are capitalised: 通名 is part of the name, not a common
-        // noun trailing it.
-        return split.map((part) => ({ ...part, isProperNoun: true }));
-    }),
-};
-/**
- * The character span each decoded word covers, in order.
- */
-function spansOf(words) {
-    let at = 0;
-    return words.map((word) => {
-        const span = { at, word };
-        at += toCharacters(word.text).length;
-        return span;
-    });
-}
-/**
- * Rewrite the words covering a span into the parts the standard writes.
- *
- * The readings are redistributed by character count, so this only applies where
- * every covered word reads one syllable per character. 儿化 cannot be
- * redistributed and is left alone.
- */
-function respace(covered, parts) {
-    const reading = covered.flatMap((word) => [...word.reading]);
-    const characters = covered.flatMap((word) => toCharacters(word.text));
-    if (reading.length !== characters.length) {
-        return undefined;
-    }
-    const head = covered[0];
-    /* c8 ignore next 3 -- the caller only passes a span it matched */
-    if (head === undefined) {
-        return undefined;
-    }
-    let at = 0;
-    return parts.map((part) => {
-        const length = toCharacters(part).length;
-        const written = {
-            text: part,
-            reading: reading.slice(at, at + length),
-            isProperNoun: head.isProperNoun,
-            partOfSpeech: head.partOfSpeech,
-            isKnown: head.isKnown,
-        };
-        at += length;
-        return written;
-    });
-}
-/**
- * The longest listed form starting at a word, with how many words it consumed.
- *
- * Longest-first, so 一个人 would beat 一个 if both were listed. A listed entry
- * has to end on a word boundary too: it is a claim about whole words, not
- * about the middle of one.
- */
-function longestListedFrom(spans, index) {
-    for (let end = spans.length; end > index; end--) {
-        const covered = spans.slice(index, end).map((span) => span.word);
-        const text = covered.map((word) => word.text).join("");
-        if (characterCount(text) <= LONGEST_SPACED_WORD) {
-            const parts = SPACED_WORD_FORMS.get(text);
-            const written = parts === undefined ? undefined : respace(covered, parts);
-            if (written !== undefined) {
-                return { written, consumed: end - index - 1 };
-            }
-        }
-    }
-    return undefined;
-}
-/**
- * Write the words GB/T 16159 spaces in a way no rule reaches.
- *
- * Matches longest-first at each word boundary, and only where the entry ends on
- * a word boundary too: a table entry is a claim about a whole word or several,
- * not about the middle of one.
- */
-export const SPACED_WORD_LIST = {
-    name: "spaced-word-list",
-    apply: (words) => {
-        const spans = spansOf(words);
-        const grouped = [];
-        for (let index = 0; index < spans.length; index++) {
-            const start = spans[index];
-            /* c8 ignore next 3 -- index stays inside the array */
-            if (start === undefined) {
-                continue;
-            }
-            const matched = longestListedFrom(spans, index);
-            grouped.push(...(matched?.written ?? [start.word]));
-            index += matched?.consumed ?? 0;
-        }
-        return grouped;
-    },
-};
-/**
- * The 称呼语 GB/T 16159 writes in front of a surname.
- *
- * 老 and 小 only. 大 is written the same way in 大李, but it is also an ordinary
- * adjective in front of anything at all, and it is the one that goes wrong:
- * over 88,866 lines the three of them fire 49 times together and both clear
- * mistakes are 大 — 泡大池 is a big pool and 那头大熊 is a big bear. Dropping it
- * leaves 38 firings and every one visible in the sample is a real form of
- * address.
- */
-const ADDRESS_PREFIXES = new Set(["老", "小"]);
-/**
- * 老王 is `Lǎo Wáng`: the prefix takes a capital of its own.
- *
- * The words are already written apart — nothing here moves a boundary — and
- * what was missing is only the capital, so this marks the prefix a proper noun
- * and lets the writer do what it already does with one. That is why it is a
- * grouping rule despite grouping nothing: `isProperNoun` is the flag the
- * capital hangs off, and there is one place it is set.
- *
- * The surname is the evidence. A one-character word the dictionary marks a
- * proper noun is what 老 and 小 attach to, and CC-CEDICT's capitalisation has
- * already vetoed the tags that would otherwise let 小 in front of anything —
- * see "jieba's 专名 tags need a second opinion" in ROADMAP.md.
- */
-export const ADDRESS_PREFIX = {
-    name: "address-prefix",
-    apply: (words) => words.map((word, at) => {
-        const next = words[at + 1];
-        if (!ADDRESS_PREFIXES.has(word.text) ||
-            next === undefined ||
-            !next.isProperNoun ||
-            characterCount(next.text) !== 1 ||
-            next.separator !== undefined) {
-            return word;
-        }
-        return { ...word, isProperNoun: true };
-    }),
-};
-/**
- * jieba's tags for the two kinds of proper noun 5.1 divides.
- *
- * A person and an organisation. Places are `PLACE_GENERICS`' business, and `nz`
- * is left out — both are measured out in `docs/orthography/`.
- */
-const DIVIDED_TAGS = new Set(["nr", "nt"]);
-/**
- * The parts of a proper name are written apart: 齐白石 is `Qí Báishí`.
- *
- * GB/T 16159 5.1 writes 姓 apart from 名 and a proper noun apart from its
- * generic, each part capitalised. Both halves of that clause are this rule,
- * because the evidence for them is the same evidence.
- *
- * 齐白石 and 北京大学 are dictionary entries, so the decoder produces **one**
- * word each — and reading them as one word is what makes them read correctly at
- * all. That makes this a *split*, and splitting contradicts the dictionary's own
- * claim that the characters belong together. It needs a condition strong enough
- * to survive the whole dictionary, and the obvious ones are not: a surname list
- * takes 马克思, 高尔基, 巴赫 and 牛顿 apart, none of which is a Chinese name,
- * and a list of generics cannot say where 上海浦东发展银行 divides.
- *
- * **The condition is CC-CEDICT's own capitalisation**, which states the
- * boundaries outright: 齐白石 is `[Qi2 Bai2 shi2]`, 司马迁 `[Si1 ma3 Qian1]`,
- * 上海交通大学 `[Shang4 hai3 Jiao1 tong1 Da4 xue2]`, and 马克思 `[Ma3 ke4 si1]`
- * with no second capital at all. So a compound surname is recognised without a
- * list of compound surnames, a generic without a list of generics, and a
- * transliteration is excluded without a list of transliterations. It is the same
- * source and the same signal the proper-noun veto already trusts — see "jieba's
- * 专名 tags need a second opinion" in ROADMAP.md — extended from *whether* a
- * word is a proper noun to *where* its parts divide.
- *
- * **Every stated boundary is cut, not only the first**, which is what separates
- * an organisation from a person: 48% of `nt` entries carrying a boundary carry
- * more than one, against 1.6% of `nr`. One cut would leave 上海交通大学 as
- * `Shànghǎi Jiāotōngdàxué`.
- *
- * A tag is still required, because the mark is not confined to what 5.1 covers:
- * 5,341 `ns` entries carry one, where `PLACE_GENERICS` already applies a
- * measured condition, and `nz` carries 346 of which 第二次世界大战
- * `[Di4 er4 Ci4 Shi4 jie4 Da4 zhan4]` divides in the wrong place. Both are
- * measured out in `docs/orthography/`.
- */
-export const NAME_PARTS = {
-    name: "name-parts",
-    apply: (words, dictionary) => words.flatMap((word) => {
-        const boundaries = dictionary.lookup(word.text)?.nameBoundaries ?? [];
-        if (boundaries.length === 0 || !word.isProperNoun) {
-            return [word];
-        }
-        if (!DIVIDED_TAGS.has(word.partOfSpeech)) {
-            return [word];
-        }
-        const parts = divideAt(word, boundaries);
-        // Every part is a proper noun in its own right, so each takes a capital:
-        // `Qí Báishí`, `Běijīng Dàxué`.
-        return parts.map((part) => ({ ...part, isProperNoun: true }));
-    }),
-};
 /**
  * The rules applied by default, in order.
  *

package/dist/orthography/grouping.js.map

@@ -1 +1 @@
-{"version":3,"file":"grouping.js","sourceRoot":"","sources":["../../src/orthography/grouping.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,cAAc,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AACvE,OAAO,EAAE,aAAa,EAAE,MAAM,aAAa,CAAC;AAC5C,OAAO,EAAE,kBAAkB,EAAE,kBAAkB,EAAE,MAAM,oBAAoB,CAAC;AAC5E,OAAO,EAAE,QAAQ,EAAqB,IAAI,EAAE,OAAO,EAAE,MAAM,WAAW,CAAC;AACvE,OAAO,EAAE,mBAAmB,EAAE,iBAAiB,EAAE,MAAM,gBAAgB,CAAC;AAExE;;GAEG;AACH,MAAM,WAAW,GAAG,IAAI,GAAG,CAAC,CAAC,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,CAAC,CAAC;AAEhD;;;;;;;GAOG;AACH,SAAS,YAAY,CAAC,IAAiB;IACrC,OAAO,IAAI,CAAC,YAAY,CAAC,UAAU,CAAC,GAAG,CAAC,IAAI,IAAI,CAAC,YAAY,KAAK,GAAG,CAAC;AACxE,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,GAAG,GAAG,CAAC;AAEvB;;;;;GAKG;AACH,MAAM,uBAAuB,GAAG,IAAI,GAAG,CAAC;IACtC,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;CACJ,CAAC,CAAC;AAEH;;GAEG;AACH,SAAS,aAAa,CACpB,KAA6B,EAC7B,UAA6D;IAE7D,MAAM,OAAO,GAAkB,EAAE,CAAC;IAClC,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,MAAM,IAAI,GAAG,OAAO,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;QAC5B,IAAI,IAAI,KAAK,SAAS,IAAI,UAAU,CAAC,IAAI,EAAE,IAAI,CAAC,EAAE,CAAC;YACjD,OAAO,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,GAAG,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC;YAC/C,SAAS;QACX,CAAC;QACD,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IACrB,CAAC;IACD,OAAO,OAAO,CAAC;AACjB,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAiB;IAC5C,IAAI,EAAE,kBAAkB;IACxB,KAAK,EAAE,CAAC,KAAK,EAAE,EAAE,CACf,aAAa,CACX,KAAK,EACL,CAAC,IAAI,EAAE,IAAI,EAAE,EAAE,CAAC,WAAW,CAAC,GAAG,CAAC,IAAI,CAAC,YAAY,CAAC,IAAI,YAAY,CAAC,IAAI,CAAC,CACzE;CACJ,CAAC;AAEF;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,QAAQ,GAAiB;IACpC,IAAI,EAAE,UAAU;IAChB,KAAK,EAAE,CAAC,KAAK,EAAE,EAAE,CACf,aAAa,CAAC,KAAK,EAAE,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,YAAY,KAAK,UAAU,CAAC;CACnE,CAAC;AAEF;;;;;;;;;;;;;;;;GAgBG;AACH,MAAM,CAAC,MAAM,cAAc,GAAiB;IAC1C,IAAI,EAAE,gBAAgB;IACtB,KAAK,EAAE,CAAC,KAAK,EAAE,UAAU,EAAE,EAAE,CAC3B,KAAK,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,EAAE;QACrB,MAAM,UAAU,GAAG,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAC3C,MAAM,OAAO,GAAG,UAAU,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;QACxC,IACE,IAAI,CAAC,YAAY,KAAK,IAAI;YAC1B,UAAU,CAAC,MAAM,GAAG,CAAC;YACrB,CAAC,uBAAuB,CAAC,GAAG,CAAC,OAAO,CAAC;YACrC,UAAU,CAAC,MAAM,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,KAAK,SAAS,EACjE,CAAC;YACD,OAAO,CAAC,IAAI,CAAC,CAAC;QAChB,CAAC;QACD,MAAM,KAAK,GAAG,OAAO,CAAC,IAAI,EAAE,UAAU,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QACnD,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;YACxB,OAAO,CAAC,IAAI,CAAC,CAAC;QAChB,CAAC;QACD,oEAAoE;QACpE,oBAAoB;QACpB,OAAO,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,EAAE,GAAG,IAAI,EAAE,YAAY,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC;IAChE,CAAC,CAAC;CACL,CAAC;AAEF;;GAEG;AACH,SAAS,OAAO,CACd,KAA6B;IAE7B,IAAI,EAAE,GAAG,CAAC,CAAC;IACX,OAAO,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE;QACxB,MAAM,IAAI,GAAG,EAAE,EAAE,EAAE,IAAI,EAAE,CAAC;QAC1B,EAAE,IAAI,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,MAAM,CAAC;QACrC,OAAO,IAAI,CAAC;IACd,CAAC,CAAC,CAAC;AACL,CAAC;AAED;;;;;;GAMG;AACH,SAAS,OAAO,CACd,OAA+B,EAC/B,KAAwB;IAExB,MAAM,OAAO,GAAG,OAAO,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,GAAG,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC;IAC7D,MAAM,UAAU,GAAG,OAAO,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;IACtE,IAAI,OAAO,CAAC,MAAM,KAAK,UAAU,CAAC,MAAM,EAAE,CAAC;QACzC,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,MAAM,IAAI,GAAG,OAAO,CAAC,CAAC,CAAC,CAAC;IACxB,kEAAkE;IAClE,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;QACvB,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,IAAI,EAAE,GAAG,CAAC,CAAC;IACX,OAAO,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE;QACxB,MAAM,MAAM,GAAG,YAAY,CAAC,IAAI,CAAC,CAAC,MAAM,CAAC;QACzC,MAAM,OAAO,GAAG;YACd,IAAI,EAAE,IAAI;YACV,OAAO,EAAE,OAAO,CAAC,KAAK,CAAC,EAAE,EAAE,EAAE,GAAG,MAAM,CAAC;YACvC,YAAY,EAAE,IAAI,CAAC,YAAY;YAC/B,YAAY,EAAE,IAAI,CAAC,YAAY;YAC/B,OAAO,EAAE,IAAI,CAAC,OAAO;SACtB,CAAC;QACF,EAAE,IAAI,MAAM,CAAC;QACb,OAAO,OAAO,CAAC;IACjB,CAAC,CAAC,CAAC;AACL,CAAC;AAED;;;;;;GAMG;AACH,SAAS,iBAAiB,CACxB,KAAqE,EACrE,KAAa;IAIb,KAAK,IAAI,GAAG,GAAG,KAAK,CAAC,MAAM,EAAE,GAAG,GAAG,KAAK,EAAE,GAAG,EAAE,EAAE,CAAC;QAChD,MAAM,OAAO,GAAG,KAAK,CAAC,KAAK,CAAC,KAAK,EAAE,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACjE,MAAM,IAAI,GAAG,OAAO,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QACvD,IAAI,cAAc,CAAC,IAAI,CAAC,IAAI,mBAAmB,EAAE,CAAC;YAChD,MAAM,KAAK,GAAG,iBAAiB,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;YAC1C,MAAM,OAAO,GAAG,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,OAAO,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC;YAC1E,IAAI,OAAO,KAAK,SAAS,EAAE,CAAC;gBAC1B,OAAO,EAAE,OAAO,EAAE,QAAQ,EAAE,GAAG,GAAG,KAAK,GAAG,CAAC,EAAE,CAAC;YAChD,CAAC;QACH,CAAC;IACH,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC;AAED;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAiB;IAC5C,IAAI,EAAE,kBAAkB;IACxB,KAAK,EAAE,CAAC,KAAK,EAAE,EAAE;QACf,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,CAAC;QAC7B,MAAM,OAAO,GAAkB,EAAE,CAAC;QAElC,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,KAAK,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC;YAClD,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC;YAC3B,sDAAsD;YACtD,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;gBACxB,SAAS;YACX,CAAC;YACD,MAAM,OAAO,GAAG,iBAAiB,CAAC,KAAK,EAAE,KAAK,CAAC,CAAC;YAChD,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,OAAO,EAAE,OAAO,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;YACpD,KAAK,IAAI,OAAO,EAAE,QAAQ,IAAI,CAAC,CAAC;QAClC,CAAC;QAED,OAAO,OAAO,CAAC;IACjB,CAAC;CACF,CAAC;AAEF;;;;;;;;;GASG;AACH,MAAM,gBAAgB,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;AAE7C;;;;;;;;;;;;;GAaG;AACH,MAAM,CAAC,MAAM,cAAc,GAAiB;IAC1C,IAAI,EAAE,gBAAgB;IACtB,KAAK,EAAE,CAAC,KAAK,EAAE,EAAE,CACf,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,EAAE,EAAE;QACrB,MAAM,IAAI,GAAG,KAAK,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;QAC3B,IACE,CAAC,gBAAgB,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC;YAChC,IAAI,KAAK,SAAS;YAClB,CAAC,IAAI,CAAC,YAAY;YAClB,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC;YAC/B,IAAI,CAAC,SAAS,KAAK,SAAS,EAC5B,CAAC;YACD,OAAO,IAAI,CAAC;QACd,CAAC;QACD,OAAO,EAAE,GAAG,IAAI,EAAE,YAAY,EAAE,IAAI,EAAE,CAAC;IACzC,CAAC,CAAC;CACL,CAAC;AAEF;;;;;GAKG;AACH,MAAM,YAAY,GAAG,IAAI,GAAG,CAAC,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC,CAAC;AAE3C;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAmCG;AACH,MAAM,CAAC,MAAM,UAAU,GAAiB;IACtC,IAAI,EAAE,YAAY;IAClB,KAAK,EAAE,CAAC,KAAK,EAAE,UAAU,EAAE,EAAE,CAC3B,KAAK,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,EAAE;QACrB,MAAM,UAAU,GAAG,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,cAAc,IAAI,EAAE,CAAC;QACtE,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,YAAY,EAAE,CAAC;YAClD,OAAO,CAAC,IAAI,CAAC,CAAC;QAChB,CAAC;QACD,IAAI,CAAC,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,YAAY,CAAC,EAAE,CAAC;YACzC,OAAO,CAAC,IAAI,CAAC,CAAC;QAChB,CAAC;QACD,MAAM,KAAK,GAAG,QAAQ,CAAC,IAAI,EAAE,UAAU,CAAC,CAAC;QACzC,yEAAyE;QACzE,gCAAgC;QAChC,OAAO,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,EAAE,GAAG,IAAI,EAAE,YAAY,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC;IAChE,CAAC,CAAC;CACL,CAAC;AAEF;;;;;GAKG;AACH,MAAM,CAAC,MAAM,cAAc,GAA4B;IACrD,gBAAgB;IAChB,QAAQ;IACR,cAAc;IACd,UAAU;IACV,gBAAgB;IAChB,cAAc;IACd,kBAAkB;IAClB,kBAAkB;IAClB,aAAa;CACd,CAAC;AAEF;;GAEG;AACH,MAAM,UAAU,aAAa,CAC3B,KAA6B,EAC7B,UAAsB,EACtB,KAAK,GAA4B,cAAc;IAE/C,IAAI,OAAO,GAAG,KAAK,CAAC;IACpB,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,OAAO,GAAG,IAAI,CAAC,KAAK,CAAC,OAAO,EAAE,UAAU,CAAC,CAAC;IAC5C,CAAC;IACD,OAAO,OAAO,CAAC;AACjB,CAAC"}
+{"version":3,"file":"grouping.js","sourceRoot":"","sources":["../../src/orthography/grouping.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,aAAa,EAAE,MAAM,aAAa,CAAC;AAC5C,OAAO,EAAE,kBAAkB,EAAE,kBAAkB,EAAE,MAAM,oBAAoB,CAAC;AAE5E,OAAO,EACL,cAAc,EACd,UAAU,EACV,gBAAgB,GACjB,MAAM,oBAAoB,CAAC;AAE5B,OAAO,EACL,cAAc,EACd,UAAU,EACV,gBAAgB,GACjB,MAAM,oBAAoB,CAAC;AAE5B;;GAEG;AACH,MAAM,WAAW,GAAG,IAAI,GAAG,CAAC,CAAC,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,CAAC,CAAC;AAEhD;;;;;;;GAOG;AACH,SAAS,YAAY,CAAC,IAAiB;IACrC,OAAO,IAAI,CAAC,YAAY,CAAC,UAAU,CAAC,GAAG,CAAC,IAAI,IAAI,CAAC,YAAY,KAAK,GAAG,CAAC;AACxE,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,GAAG,GAAG,CAAC;AAEvB,OAAO,EAAE,aAAa,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAC;AAEpE,OAAO,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAC;AACrD,MAAM,CAAC,MAAM,gBAAgB,GAAiB;IAC5C,IAAI,EAAE,kBAAkB;IACxB,KAAK,EAAE,CAAC,KAAK,EAAE,EAAE,CACf,aAAa,CACX,KAAK,EACL,CAAC,IAAI,EAAE,IAAI,EAAE,EAAE,CAAC,WAAW,CAAC,GAAG,CAAC,IAAI,CAAC,YAAY,CAAC,IAAI,YAAY,CAAC,IAAI,CAAC,CACzE;CACJ,CAAC;AAEF;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,QAAQ,GAAiB;IACpC,IAAI,EAAE,UAAU;IAChB,KAAK,EAAE,CAAC,KAAK,EAAE,EAAE,CACf,aAAa,CAAC,KAAK,EAAE,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,YAAY,KAAK,UAAU,CAAC;CACnE,CAAC;AACF;;;;;GAKG;AACH,MAAM,CAAC,MAAM,cAAc,GAA4B;IACrD,gBAAgB;IAChB,QAAQ;IACR,cAAc;IACd,UAAU;IACV,gBAAgB;IAChB,cAAc;IACd,kBAAkB;IAClB,kBAAkB;IAClB,aAAa;CACd,CAAC;AAEF;;GAEG;AACH,MAAM,UAAU,aAAa,CAC3B,KAA6B,EAC7B,UAAsB,EACtB,KAAK,GAA4B,cAAc;IAE/C,IAAI,OAAO,GAAG,KAAK,CAAC;IACpB,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,OAAO,GAAG,IAAI,CAAC,KAAK,CAAC,OAAO,EAAE,UAAU,CAAC,CAAC;IAC5C,CAAC;IACD,OAAO,OAAO,CAAC;AACjB,CAAC"}

package/dist/orthography/place-grouping.d.ts

@@ -0,0 +1,39 @@
+/**
+ * Joining a place name to the generic that names its kind.
+ *
+ * 北京市 is one orthographic word and 北京 市 is two, and which it is depends
+ * on the generic rather than on the tag jieba gave the whole.
+ */
+import type { DecodedWord } from "../decode/word.js";
+import { type GroupingRule } from "./rule.js";
+/**
+ * Fold every word a predicate accepts into the word before it.
+ */
+export declare function joinBackwards(words: readonly DecodedWord[], isAttached: (word: DecodedWord, head: DecodedWord) => boolean): readonly DecodedWord[];
+/**
+ * 了, 着 and 过 attach to the verb in front of them.
+ *
+ * GB/T 16159: 他看了 is `tā kànle`, not `tā kàn le`. Safe as a rule because it
+ * only ever joins what the decode had already separated — the dictionary has no
+ * entry for 看了, so nothing the dictionary asserted is being overridden. The
+ * readings do not change either way, so the worst case is untidy spacing.
+ */
+/**
+ * A place name's generic half is written separately, and capitalised.
+ *
+ * GB/T 16159 专名和通名分写: 南京市 is `Nánjīng Shì`, 西湖区 is `Xīhú Qū`.
+ *
+ * Two conditions keep it from running away, both of them measured against the
+ * whole dictionary rather than guessed. The word must be tagged a place name,
+ * and **the part before the generic must itself be a word** — without the
+ * second, 上山下乡 and 三街六市 are tagged `ns` and come apart as
+ * `Shàngshānxià Xiāng` and `Sānjiēliù Shì`. Together they fire on 4,798 of the
+ * dictionary's 8,781 candidates and hold back 427.
+ *
+ * The rule deliberately never touches a two-character name, so 上海 stays
+ * `Shànghǎi` and 长城 stays `Chángchéng`: there, the generic character is part
+ * of the proper name rather than a description of it, and no rule can tell the
+ * two apart — 黄河 is `Huáng Hé` and 青海 is `Qīnghǎi`.
+ */
+export declare const PLACE_GENERICS: GroupingRule;
+//# sourceMappingURL=place-grouping.d.ts.map

package/dist/orthography/place-grouping.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"place-grouping.d.ts","sourceRoot":"","sources":["../../src/orthography/place-grouping.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAErD,OAAO,EAAE,KAAK,YAAY,EAAiB,MAAM,WAAW,CAAC;AAuB7D;;GAEG;AACH,wBAAgB,aAAa,CAC3B,KAAK,EAAE,SAAS,WAAW,EAAE,EAC7B,UAAU,EAAE,CAAC,IAAI,EAAE,WAAW,EAAE,IAAI,EAAE,WAAW,KAAK,OAAO,GAC5D,SAAS,WAAW,EAAE,CAWxB;AAED;;;;;;;GAOG;AAEH;;;;;;;;;;;;;;;;GAgBG;AACH,eAAO,MAAM,cAAc,EAAE,YAsB5B,CAAC"}