@kensio/pinyinjs 0.1.0-beta.0 → 1.0.0

package/dist/numerals/text.js

@@ -0,0 +1,248 @@
+import { applySandhi } from "../decode/sandhi.js";
+import { toCharacters } from "../script/characters.js";
+import { numeralHanzi, percentHanzi, readNumeralHanzi, } from "./numerals.js";
+/**
+ * A number as it appears in running text: digits, with the separators a writer
+ * uses and an optional decimal part.
+ */
+const NUMBER = /\d[\d,]*(?:\.\d+)?/gu;
+/**
+ * The percent signs a number can carry, half-width and full-width.
+ */
+const PERCENT = new Set(["%", "％"]);
+/**
+ * Marks that make the digits beside them an identifier rather than a quantity.
+ *
+ * 6:30 is a time, 3202－5625 is a phone number and COVID-19 is a name; none of
+ * them is counted, and the mark between the parts has no reading at all. So a
+ * digit run touching one is left exactly as it was written — of 7,846 digit
+ * runs measured over Tatoeba and zh.wikipedia about 55 are this shape, and
+ * reading them produced `liù:sānshí` for 6:30, which is worse than leaving it.
+ */
+const IDENTIFIER = new Set([":", "：", "-", "－", "—", "–", "/", "／"]);
+/**
+ * A time as a writer punctuates it: 6:30, 07:05, 2：30.
+ *
+ * Sticky, because it is tried at the start of a digit run the number scanner
+ * has already found rather than searched for on its own.
+ *
+ * Two digits after the colon and no more, which is what separates a time from
+ * a ratio: 16:9 and 2:1 are proportions and scores and are read `shíliù bǐ
+ * jiǔ`, which this does not attempt. Measured over Tatoeba and zh.wikipedia,
+ * the shape catches 104 runs and every one of them is a time — none has an
+ * hour above 23 or a minute above 59, and the four colon runs that are not
+ * times all have a single digit after the colon.
+ */
+const TIME = /(\d{1,2})[:：](\d{2})(?!\d)/duy;
+/**
+ * The bounds a time is written within.
+ */
+const HOURS_IN_A_DAY = 24;
+const MINUTES_IN_AN_HOUR = 60;
+/**
+ * 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;
+/**
+ * How many digits a year is written with, when it is a year and not a count.
+ *
+ * 1997年 is a year and is spelled out; 30年 is thirty years and is counted.
+ * Four is the length that separates them, and a two-digit year — 98年 for 1998
+ * — is left counted, because 20年 meaning twenty years is far commoner than
+ * 20年 meaning 2020 and nothing in the text tells them apart.
+ */
+const YEAR_DIGITS = 4;
+/**
+ * Whether a number is read out digit by digit rather than counted.
+ *
+ * The one context where this is not a guess: a four-digit number in front of 年
+ * is a year, and a year is spelled out. Everything else is counted, which is
+ * what the overwhelming majority of digits in running text are — of 7,846 digit
+ * runs measured over Tatoeba and zh.wikipedia, the commonest things after them
+ * are 年, 月, 日, 人, 号, 点, 个 and 岁, and only the years want spelling out.
+ */
+function styleFor(digits, following) {
+    return following === "年" && digits.length === YEAR_DIGITS
+        ? "digits"
+        : "cardinal";
+}
+/**
+ * Read one number, with the sign that may follow it.
+ */
+function readNumber(digits, isPercent, style, options) {
+    // A writer's thousands separators are not part of the number; nothing is
+    // said for them.
+    const value = digits.replaceAll(",", "");
+    const settings = { ...options, style };
+    const hanzi = isPercent
+        ? percentHanzi(value, settings)
+        : numeralHanzi(value, settings);
+    const reading = hanzi === undefined ? undefined : readNumeralHanzi(hanzi, settings);
+    if (hanzi === undefined || reading === undefined) {
+        return undefined;
+    }
+    return { text: digits, reading, hanzi, style };
+}
+/**
+ * A stretch that stays exactly as it was written.
+ */
+function unread(text) {
+    return { text, reading: undefined, hanzi: undefined, style: undefined };
+}
+/**
+ * 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.
+ */
+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.
+ */
+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.
+ */
+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),
+    };
+}
+/**
+ * Split a stretch of non-Han text into what is read and what is not.
+ *
+ * `following` is the first character of the Han that comes after this stretch,
+ * which is the only context a number has: 年 makes 1997 a year and 个 makes 3 a
+ * count. Where there is nothing after it, or nothing that decides, the number
+ * is counted.
+ *
+ * Text that is not a number is returned unread rather than dropped, so the
+ * segments always concatenate back to the input.
+ */
+export function readNumbersIn(text, following, options = {}) {
+    const segments = [];
+    let at = 0;
+    for (const match of text.matchAll(NUMBER)) {
+        const digits = match[0];
+        const from = match.index;
+        // The minutes of a time the hour already took.
+        if (from < at) {
+            continue;
+        }
+        const time = readTime(text, from, options);
+        if (time !== undefined) {
+            if (from > at) {
+                segments.push(unread(text.slice(at, from)));
+            }
+            segments.push(time);
+            at = from + time.text.length;
+            continue;
+        }
+        const after = text[from + digits.length] ?? "";
+        if (IDENTIFIER.has(after) || IDENTIFIER.has(text[from - 1] ?? "")) {
+            continue;
+        }
+        const isPercent = PERCENT.has(after);
+        // What decides the style is whatever comes next: the sign if there is one,
+        // otherwise the first character of the following Han.
+        const style = styleFor(digits, isPercent ? "" : following);
+        const read = readNumber(digits, isPercent, style, options);
+        if (read === undefined) {
+            continue;
+        }
+        if (from > at) {
+            segments.push(unread(text.slice(at, from)));
+        }
+        segments.push(isPercent ? { ...read, text: digits + after } : read);
+        at = from + digits.length + (isPercent ? after.length : 0);
+    }
+    if (at < text.length) {
+        segments.push(unread(text.slice(at)));
+    }
+    return segments;
+}
+/**
+ * A number as it is said, with the sandhi a counted quantity takes.
+ *
+ * Two things decide how much of it assimilates. A number read out digit by
+ * digit does not at all — 110 is `yāo yāo líng` and never `yì yì líng` —
+ * and in a counted one the sandhi stops at the 点, since everything after the
+ * decimal point is read digit by digit whatever the style: 3.14 is
+ * `sān diǎn yī sì`.
+ *
+ * `following` is the syllable after the number, which is what a 一 at the end
+ * of it assimilates to: the 一 of 一个 is `yí` because 个 is a fourth tone, and
+ * that tone is in the next run rather than in the number.
+ */
+export function saidNumeral(segment, following, sandhi) {
+    const reading = segment.reading ?? [];
+    if (segment.style === "digits") {
+        return reading;
+    }
+    const point = toCharacters(segment.hanzi ?? "").indexOf("点");
+    const quantity = point === -1 ? reading.length : point;
+    const context = following === undefined ? [] : [following];
+    const said = applySandhi([...reading.slice(0, quantity), ...context], sandhi);
+    return [...said.slice(0, quantity), ...reading.slice(quantity)];
+}
+//# sourceMappingURL=text.js.map

package/dist/numerals/text.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"text.js","sourceRoot":"","sources":["../../src/numerals/text.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAAsB,MAAM,qBAAqB,CAAC;AACtE,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAEvD,OAAO,EACL,YAAY,EAGZ,YAAY,EACZ,gBAAgB,GACjB,MAAM,eAAe,CAAC;AAyBvB;;;GAGG;AACH,MAAM,MAAM,GAAG,sBAAsB,CAAC;AAEtC;;GAEG;AACH,MAAM,OAAO,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;AAEpC;;;;;;;;GAQG;AACH,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;AAErE;;;;;;;;;;;;GAYG;AACH,MAAM,IAAI,GAAG,+BAA+B,CAAC;AAE7C;;GAEG;AACH,MAAM,cAAc,GAAG,EAAE,CAAC;AAC1B,MAAM,kBAAkB,GAAG,EAAE,CAAC;AAE9B;;GAEG;AACH,MAAM,iBAAiB,GAAG,EAAE,CAAC;AAE7B;;;;;;GAMG;AACH,MAAM,UAAU,GAAG,CAAC,CAAC;AAErB;;;;;;;GAOG;AACH,MAAM,WAAW,GAAG,CAAC,CAAC;AAEtB;;;;;;;;GAQG;AACH,SAAS,QAAQ,CAAC,MAAc,EAAE,SAAiB;IACjD,OAAO,SAAS,KAAK,GAAG,IAAI,MAAM,CAAC,MAAM,KAAK,WAAW;QACvD,CAAC,CAAC,QAAQ;QACV,CAAC,CAAC,UAAU,CAAC;AACjB,CAAC;AAED;;GAEG;AACH,SAAS,UAAU,CACjB,MAAc,EACd,SAAkB,EAClB,KAAmB,EACnB,OAAuB;IAEvB,yEAAyE;IACzE,iBAAiB;IACjB,MAAM,KAAK,GAAG,MAAM,CAAC,UAAU,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC;IACzC,MAAM,QAAQ,GAAmB,EAAE,GAAG,OAAO,EAAE,KAAK,EAAE,CAAC;IACvD,MAAM,KAAK,GAAG,SAAS;QACrB,CAAC,CAAC,YAAY,CAAC,KAAK,EAAE,QAAQ,CAAC;QAC/B,CAAC,CAAC,YAAY,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC;IAClC,MAAM,OAAO,GACX,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,gBAAgB,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC;IACtE,IAAI,KAAK,KAAK,SAAS,IAAI,OAAO,KAAK,SAAS,EAAE,CAAC;QACjD,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,KAAK,EAAE,KAAK,EAAE,CAAC;AACjD,CAAC;AAED;;GAEG;AACH,SAAS,MAAM,CAAC,IAAY;IAC1B,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,SAAS,EAAE,KAAK,EAAE,SAAS,EAAE,KAAK,EAAE,SAAS,EAAE,CAAC;AAC1E,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,SAAS,CAChB,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,SAAS,SAAS,CAAC,KAAa;IAC9B,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,SAAS,QAAQ,CACf,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;AAED;;;;;;;;;;GAUG;AACH,MAAM,UAAU,aAAa,CAC3B,IAAY,EACZ,SAAiB,EACjB,OAAO,GAAmB,EAAE;IAE5B,MAAM,QAAQ,GAAqB,EAAE,CAAC;IACtC,IAAI,EAAE,GAAG,CAAC,CAAC;IAEX,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;QAC1C,MAAM,MAAM,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;QACxB,MAAM,IAAI,GAAG,KAAK,CAAC,KAAK,CAAC;QACzB,+CAA+C;QAC/C,IAAI,IAAI,GAAG,EAAE,EAAE,CAAC;YACd,SAAS;QACX,CAAC;QACD,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI,EAAE,IAAI,EAAE,OAAO,CAAC,CAAC;QAC3C,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;YACvB,IAAI,IAAI,GAAG,EAAE,EAAE,CAAC;gBACd,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC;YAC9C,CAAC;YACD,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACpB,EAAE,GAAG,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC;YAC7B,SAAS;QACX,CAAC;QACD,MAAM,KAAK,GAAG,IAAI,CAAC,IAAI,GAAG,MAAM,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC;QAC/C,IAAI,UAAU,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,UAAU,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC;YAClE,SAAS;QACX,CAAC;QACD,MAAM,SAAS,GAAG,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;QACrC,2EAA2E;QAC3E,sDAAsD;QACtD,MAAM,KAAK,GAAG,QAAQ,CAAC,MAAM,EAAE,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC;QAC3D,MAAM,IAAI,GAAG,UAAU,CAAC,MAAM,EAAE,SAAS,EAAE,KAAK,EAAE,OAAO,CAAC,CAAC;QAC3D,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;YACvB,SAAS;QACX,CAAC;QACD,IAAI,IAAI,GAAG,EAAE,EAAE,CAAC;YACd,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC;QAC9C,CAAC;QACD,QAAQ,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE,GAAG,IAAI,EAAE,IAAI,EAAE,MAAM,GAAG,KAAK,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;QACpE,EAAE,GAAG,IAAI,GAAG,MAAM,CAAC,MAAM,GAAG,CAAC,SAAS,CAAC,CAAC,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC7D,CAAC;IAED,IAAI,EAAE,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC;QACrB,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;IACxC,CAAC;IACD,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED;;;;;;;;;;;;GAYG;AACH,MAAM,UAAU,WAAW,CACzB,OAAuB,EACvB,SAA+B,EAC/B,MAAiC;IAEjC,MAAM,OAAO,GAAG,OAAO,CAAC,OAAO,IAAI,EAAE,CAAC;IACtC,IAAI,OAAO,CAAC,KAAK,KAAK,QAAQ,EAAE,CAAC;QAC/B,OAAO,OAAO,CAAC;IACjB,CAAC;IACD,MAAM,KAAK,GAAG,YAAY,CAAC,OAAO,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;IAC7D,MAAM,QAAQ,GAAG,KAAK,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,KAAK,CAAC;IACvD,MAAM,OAAO,GAAG,SAAS,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC;IAC3D,MAAM,IAAI,GAAG,WAAW,CAAC,CAAC,GAAG,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,QAAQ,CAAC,EAAE,GAAG,OAAO,CAAC,EAAE,MAAM,CAAC,CAAC;IAC9E,OAAO,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,QAAQ,CAAC,EAAE,GAAG,OAAO,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,CAAC;AAClE,CAAC"}

package/dist/orthography/grouping.d.ts

@@ -1,16 +1,6 @@
 import type { Dictionary } from "../dictionary/dictionary.js";
 import type { DecodedWord } from "../decode/word.js";
-/**
- * One 分词连写 rule, rewriting the words a decode produced.
- *
- * Typed and individually testable, which is the point ALGORITHM.md makes about
- * rules: the old project patched the output string with regexes, so its rules
- * were order-dependent and could not be tested apart from each other.
- */
-export interface GroupingRule {
-    readonly name: string;
-    readonly apply: (words: readonly DecodedWord[], dictionary: Dictionary) => readonly DecodedWord[];
-}
+import { type GroupingRule } from "./rule.js";
 /**
  * 了, 着 and 过 attach to the verb in front of them.
  *
@@ -54,8 +44,63 @@ export declare const PLACE_GENERICS: GroupingRule;
  * 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: 毛泽东 is `Máo Zédōng`.
+ *
+ * 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.
+ *
+ * 毛泽东 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: 毛泽东 is `[Mao2 Ze2 dong1]`, 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.
+ *
+ * The hyphens come last, because every rule before them moves word boundaries
+ * and a hyphen is a statement about where a boundary ended up.
  */
 export declare const GROUPING_RULES: readonly GroupingRule[];
 /**

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;;;;;;GAMG;AACH,MAAM,WAAW,YAAY;IAC3B,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,QAAQ,CAAC,KAAK,EAAE,CACd,KAAK,EAAE,SAAS,WAAW,EAAE,EAC7B,UAAU,EAAE,UAAU,KACnB,SAAS,WAAW,EAAE,CAAC;CAC7B;AAgFD;;;;;;;GAOG;AACH,eAAO,MAAM,gBAAgB,EAAE,YAO9B,CAAC;AAEF;;;;;;GAMG;AACH,eAAO,MAAM,QAAQ,EAAE,YAItB,CAAC;AA0BF;;;;;;;;;;;;;;;;GAgBG;AACH,eAAO,MAAM,cAAc,EAAE,YAsB5B,CAAC;AAgFF;;;;;;GAMG;AACH,eAAO,MAAM,gBAAgB,EAAE,YAmB9B,CAAC;AAEF;;GAEG;AACH,eAAO,MAAM,cAAc,EAAE,SAAS,YAAY,EAKjD,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;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"}

package/dist/orthography/grouping.js

@@ -1,4 +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";
 /**
  * jieba's tags for the aspect particles 了, 着 and 过.
@@ -39,21 +42,6 @@ const ADMINISTRATIVE_GENERICS = new Set([
     "村",
     "州",
 ]);
-/**
- * Join two decoded words into one.
- *
- * The first word's flags carry, because it is the head: 看了 is the verb 看
- * with an aspect particle stuck to it, not a new word of its own.
- */
-function join(head, tail) {
-    return {
-        text: head.text + tail.text,
-        reading: [...head.reading, ...tail.reading],
-        isProperNoun: head.isProperNoun,
-        partOfSpeech: head.partOfSpeech,
-        isKnown: head.isKnown && tail.isKnown,
-    };
-}
 /**
  * Fold every word a predicate accepts into the word before it.
  */
@@ -92,26 +80,6 @@ export const SUFFIXES = {
     name: "suffixes",
     apply: (words) => joinBackwards(words, (word) => word.partOfSpeech === SUFFIX_TAG),
 };
-/**
- * Split a decoded word into two at a character offset.
- *
- * Only possible where the reading has one syllable per character, which 儿化
- * does not: there is no way to cut `wánr` in half.
- */
-function splitAt(word, at) {
-    const characters = toCharacters(word.text);
-    if (word.reading.length !== characters.length) {
-        return undefined;
-    }
-    const part = (from, to) => ({
-        text: characters.slice(from, to).join(""),
-        reading: word.reading.slice(from, to),
-        isProperNoun: word.isProperNoun,
-        partOfSpeech: word.partOfSpeech,
-        isKnown: word.isKnown,
-    });
-    return [part(0, at), part(at, characters.length)];
-}
 /**
  * A place name's generic half is written separately, and capitalised.
  *
@@ -238,14 +206,120 @@ export const SPACED_WORD_LIST = {
         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: 毛泽东 is `Máo Zédōng`.
+ *
+ * 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.
+ *
+ * 毛泽东 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: 毛泽东 is `[Mao2 Ze2 dong1]`, 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:
+        // `Máo Zédōng`, `Běijīng Dàxué`.
+        return parts.map((part) => ({ ...part, isProperNoun: true }));
+    }),
+};
 /**
  * The rules applied by default, in order.
+ *
+ * The hyphens come last, because every rule before them moves word boundaries
+ * and a hyphen is a statement about where a boundary ended up.
  */
 export const GROUPING_RULES = [
     ASPECT_PARTICLES,
     SUFFIXES,
     PLACE_GENERICS,
+    NAME_PARTS,
     SPACED_WORD_LIST,
+    ADDRESS_PREFIX,
+    AABB_REDUPLICATION,
+    ABAB_REDUPLICATION,
+    IDIOM_HYPHENS,
 ];
 /**
  * Run the 分词连写 rules over a decoded run.

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,mBAAmB,EAAE,iBAAiB,EAAE,MAAM,gBAAgB,CAAC;AAiBxE;;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;;;;;GAKG;AACH,SAAS,IAAI,CAAC,IAAiB,EAAE,IAAiB;IAChD,OAAO;QACL,IAAI,EAAE,IAAI,CAAC,IAAI,GAAG,IAAI,CAAC,IAAI;QAC3B,OAAO,EAAE,CAAC,GAAG,IAAI,CAAC,OAAO,EAAE,GAAG,IAAI,CAAC,OAAO,CAAC;QAC3C,YAAY,EAAE,IAAI,CAAC,YAAY;QAC/B,YAAY,EAAE,IAAI,CAAC,YAAY;QAC/B,OAAO,EAAE,IAAI,CAAC,OAAO,IAAI,IAAI,CAAC,OAAO;KACtC,CAAC;AACJ,CAAC;AAED;;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;;;;;GAKG;AACH,SAAS,OAAO,CACd,IAAiB,EACjB,EAAU;IAEV,MAAM,UAAU,GAAG,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAC3C,IAAI,IAAI,CAAC,OAAO,CAAC,MAAM,KAAK,UAAU,CAAC,MAAM,EAAE,CAAC;QAC9C,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,MAAM,IAAI,GAAG,CAAC,IAAY,EAAE,EAAU,EAAe,EAAE,CAAC,CAAC;QACvD,IAAI,EAAE,UAAU,CAAC,KAAK,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC;QACzC,OAAO,EAAE,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,IAAI,EAAE,EAAE,CAAC;QACrC,YAAY,EAAE,IAAI,CAAC,YAAY;QAC/B,YAAY,EAAE,IAAI,CAAC,YAAY;QAC/B,OAAO,EAAE,IAAI,CAAC,OAAO;KACtB,CAAC,CAAC;IACH,OAAO,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,CAAC,EAAE,IAAI,CAAC,EAAE,EAAE,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC;AACpD,CAAC;AAED;;;;;;;;;;;;;;;;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;;GAEG;AACH,MAAM,CAAC,MAAM,cAAc,GAA4B;IACrD,gBAAgB;IAChB,QAAQ;IACR,cAAc;IACd,gBAAgB;CACjB,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,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,iCAAiC;QACjC,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"}

package/dist/orthography/idiom-list.d.ts

@@ -0,0 +1,56 @@
+/**
+ * The 成语 GB/T 16159 writes with a hyphen down the middle.
+ *
+ * 6.3.2: a four-syllable 成语 *that can be read as two disyllables* takes a
+ * connecting hyphen — `fēngpíng-làngjìng` — and one that cannot is written
+ * solid: `bùyìlèhū`, `chūqíbùyì`. The condition is prosodic and structural, and
+ * **nothing in the data tracks it.** Conditioning on both halves being
+ * dictionary words fires on 10,202 of the 22,192 four-character words jieba tags
+ * `i` and is uncorrelated with the standard: it fires on 层出不穷 and not on
+ * 风平浪静, which are the standard's own two examples of the same rule, and it
+ * fires on 精神文明 and 凯旋归来, which want a space rather than a hyphen. No
+ * source carries hyphenated pinyin either, so unlike every other claim this
+ * package makes, a 成语 rule could not be scored even if one were written.
+ *
+ * So: a list, on the same argument as [the spaced word list](./word-list.ts).
+ *
+ * **What is in it, and what is not.** Only 成语 whose two halves are each a
+ * self-contained disyllable — parallel halves (风平｜浪静, 千军｜万马), a
+ * disyllable and its predicate (声势｜浩大), or a disyllabic stem with a
+ * reduplicated tail (小心｜翼翼). Those are the ones where 2+2 is beyond doubt.
+ * Anything with a monosyllabic word breaking the rhythm is left out and stays
+ * solid, which is what the standard does with it anyway: 目不转睛 is 目 ｜
+ * 不转睛, 迫不及待 is 迫 ｜ 不及待, 情不自禁 and 显而易见 and 名副其实 the same.
+ * Left out too is anything this file is not sure of, because a missing entry
+ * keeps today's output and a wrong one breaks it.
+ *
+ * It is not a complete 成语 list and is not meant to become one — 22,192 of them
+ * exist. Measured over 88,866 lines of Tatoeba and zh.wikipedia, the 117 here
+ * cover 264 of the 1,737 four-character `i`-tagged words that turn up, or
+ * 15.2%. Most of what is left is either genuinely not 2+2 (据我所知, 一无所知,
+ * 心不在焉, 难以置信) or not a 成语 at all (非常感谢, 重新启动, 可口可乐), which
+ * is the shape a curated list should have: the cases it declines are mostly
+ * cases it should decline.
+ */
+/**
+ * One 成语, spelled both ways.
+ *
+ * The two spellings are often the same string — 山清水秀 and 三心二意 are
+ * written identically in both scripts — and that is why the pair is kept rather
+ * than a flat list of forms: it is what makes "every 简体 entry has the 繁體 one
+ * that reads the same beside it" a checkable claim.
+ */
+export interface IdiomSpelling {
+    readonly hans: string;
+    /** Empty where the list is missing its counterpart, which a test rejects. */
+    readonly hant: string;
+}
+/**
+ * Every 成语 written with the 2+2 hyphen, paired across the two scripts.
+ */
+export declare const HYPHENATED_IDIOMS: readonly IdiomSpelling[];
+/**
+ * Every spelling that takes the hyphen, which is how the rule asks.
+ */
+export declare const HYPHENATED_IDIOM_FORMS: ReadonlySet<string>;
+//# sourceMappingURL=idiom-list.d.ts.map

package/dist/orthography/idiom-list.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"idiom-list.d.ts","sourceRoot":"","sources":["../../src/orthography/idiom-list.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAiCG;AAmQH;;;;;;;GAOG;AACH,MAAM,WAAW,aAAa;IAC5B,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,6EAA6E;IAC7E,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;CACvB;AAED;;GAEG;AACH,eAAO,MAAM,iBAAiB,EAAE,SAAS,aAAa,EAErD,CAAC;AAEF;;GAEG;AACH,eAAO,MAAM,sBAAsB,EAAE,WAAW,CAAC,MAAM,CAEtD,CAAC"}