@kensio/pinyinjs 1.8.6 → 1.9.1

package/dist/decode/verbal-rule.js

@@ -0,0 +1,100 @@
+/**
+ * The reading rule for characters whose verb reads differently from the rest
+ * of them: 量, 数, 调, 系, 转, 钉, 结, 相 and 冲.
+ */
+import { isSameSyllable } from "../dictionary/entry.js";
+import { CHONG_FRAME } from "./chong-frames.js";
+import { DING_FRAME } from "./ding-frames.js";
+import { JI_FRAME } from "./ji-frames.js";
+import { JIE_FRAME } from "./jie-frames.js";
+import { LIANG_FRAME } from "./liang-frames.js";
+import { SHU_FRAME } from "./shu-frames.js";
+import { TIAO_FRAME } from "./tiao-frames.js";
+import { isBoundAt } from "./verbal-frame.js";
+import { isHeldAgainst } from "./verbal-held.js";
+import { XIANG_FRAME } from "./xiang-frames.js";
+import { ZHUAN_FRAME } from "./zhuan-frames.js";
+/**
+ * Every frame, one to a character.
+ */
+export const VERBAL_FRAMES = [
+    LIANG_FRAME,
+    SHU_FRAME,
+    TIAO_FRAME,
+    JI_FRAME,
+    ZHUAN_FRAME,
+    DING_FRAME,
+    JIE_FRAME,
+    XIANG_FRAME,
+    CHONG_FRAME,
+];
+/**
+ * The frames by the characters they read.
+ */
+const FRAMES_BY_CHARACTER = new Map(VERBAL_FRAMES.flatMap((frame) => [...frame.characters].map((character) => [character, frame])));
+/**
+ * The frame saying the character at a position is its verb, or undefined.
+ */
+function frameAt(context, at) {
+    /* c8 ignore next -- the position is inside the run */
+    const frame = FRAMES_BY_CHARACTER.get(context.characters[at] ?? "");
+    if (frame === undefined ||
+        isBoundAt(context, at, frame) ||
+        isHeldAgainst(context, at, frame.reading) ||
+        !frame.isVerbalAt(context, at)) {
+        return undefined;
+    }
+    return frame;
+}
+/**
+ * The verb's reading for nine characters where the sentence uses the verb.
+ *
+ * Each of them is stored with another reading first, and for the same reason:
+ * the commoner reading in a corpus is the one inside words. 量 is `liàng` in
+ * 数量, 质量 and 词汇量, 数 is `shù` in 数字 and 数十, 调 is `diào` in 调查, 系
+ * is `xì` in 关系, and so on. A word carries its own reading, so what the
+ * default decides is the character standing on its own, and in running text
+ * that is as often the verb as anything. 我妈妈量了我的体温 came out with
+ * `liàng`, 你能数到十吗 with `shù` and 请把音量调低一点 with `diào`.
+ *
+ * The verb is told apart by its neighbours, a different set for each
+ * character, so each has a frame of its own in a file of its own. They share
+ * how the verdict is given, which is the way {@link import("./zhi-rule.js").COUNTING_ZHI}
+ * gives it: where a frame holds, the edge of the character alone reading the
+ * verb is forced, and an edge of several characters reading it some other way
+ * is forbidden. A tagged word holding the character with another reading is
+ * left alone, since 词汇量 and 关系 decide their own.
+ *
+ * The reading the frame names has to be among the character's readings for
+ * the rule to do anything. Unihan gives 繫 and 釘 no `jì` and no `dìng`, so
+ * those two forms are left out until the data has them.
+ */
+export const VERBAL_READING = {
+    name: "verbal-reading",
+    verdictFor: (context) => {
+        const { edge } = context;
+        if (edge.reading.length !== edge.to - edge.from) {
+            return "keep";
+        }
+        let verdict = "keep";
+        for (let at = edge.from; at < edge.to; at++) {
+            const frame = frameAt(context, at);
+            if (frame === undefined) {
+                continue;
+            }
+            const here = edge.reading[at - edge.from];
+            /* c8 ignore next 3 -- the reading is as long as the span */
+            if (here === undefined) {
+                continue;
+            }
+            if (!isSameSyllable(here, frame.reading)) {
+                return edge.to - edge.from === 1 ? "keep" : "forbid";
+            }
+            if (edge.to - edge.from === 1) {
+                verdict = "force";
+            }
+        }
+        return verdict;
+    },
+};
+//# sourceMappingURL=verbal-rule.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"verbal-rule.js","sourceRoot":"","sources":["../../src/decode/verbal-rule.ts"],"names":[],"mappings":"AAAA;;;GAGG;AACH,OAAO,EAAE,cAAc,EAAE,MAAM,wBAAwB,CAAC;AACxD,OAAO,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAChD,OAAO,EAAE,UAAU,EAAE,MAAM,kBAAkB,CAAC;AAC9C,OAAO,EAAE,QAAQ,EAAE,MAAM,gBAAgB,CAAC;AAC1C,OAAO,EAAE,SAAS,EAAE,MAAM,iBAAiB,CAAC;AAC5C,OAAO,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAEhD,OAAO,EAAE,SAAS,EAAE,MAAM,iBAAiB,CAAC;AAC5C,OAAO,EAAE,UAAU,EAAE,MAAM,kBAAkB,CAAC;AAC9C,OAAO,EAAE,SAAS,EAAoB,MAAM,mBAAmB,CAAC;AAChE,OAAO,EAAE,aAAa,EAAE,MAAM,kBAAkB,CAAC;AACjD,OAAO,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAChD,OAAO,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAEhD;;GAEG;AACH,MAAM,CAAC,MAAM,aAAa,GAA2B;IACnD,WAAW;IACX,SAAS;IACT,UAAU;IACV,QAAQ;IACR,WAAW;IACX,UAAU;IACV,SAAS;IACT,WAAW;IACX,WAAW;CACZ,CAAC;AAEF;;GAEG;AACH,MAAM,mBAAmB,GAAqC,IAAI,GAAG,CACnE,aAAa,CAAC,OAAO,CAAC,CAAC,KAAK,EAAE,EAAE,CAC9B,CAAC,GAAG,KAAK,CAAC,UAAU,CAAC,CAAC,GAAG,CAAC,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC,SAAS,EAAE,KAAK,CAAU,CAAC,CACtE,CACF,CAAC;AAEF;;GAEG;AACH,SAAS,OAAO,CAAC,OAAoB,EAAE,EAAU;IAC/C,sDAAsD;IACtD,MAAM,KAAK,GAAG,mBAAmB,CAAC,GAAG,CAAC,OAAO,CAAC,UAAU,CAAC,EAAE,CAAC,IAAI,EAAE,CAAC,CAAC;IACpE,IACE,KAAK,KAAK,SAAS;QACnB,SAAS,CAAC,OAAO,EAAE,EAAE,EAAE,KAAK,CAAC;QAC7B,aAAa,CAAC,OAAO,EAAE,EAAE,EAAE,KAAK,CAAC,OAAO,CAAC;QACzC,CAAC,KAAK,CAAC,UAAU,CAAC,OAAO,EAAE,EAAE,CAAC,EAC9B,CAAC;QACD,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,OAAO,KAAK,CAAC;AACf,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;GAsBG;AACH,MAAM,CAAC,MAAM,cAAc,GAAa;IACtC,IAAI,EAAE,gBAAgB;IACtB,UAAU,EAAE,CAAC,OAAoB,EAAE,EAAE;QACnC,MAAM,EAAE,IAAI,EAAE,GAAG,OAAO,CAAC;QACzB,IAAI,IAAI,CAAC,OAAO,CAAC,MAAM,KAAK,IAAI,CAAC,EAAE,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;YAChD,OAAO,MAAM,CAAC;QAChB,CAAC;QACD,IAAI,OAAO,GAAgB,MAAM,CAAC;QAClC,KAAK,IAAI,EAAE,GAAG,IAAI,CAAC,IAAI,EAAE,EAAE,GAAG,IAAI,CAAC,EAAE,EAAE,EAAE,EAAE,EAAE,CAAC;YAC5C,MAAM,KAAK,GAAG,OAAO,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC;YACnC,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;gBACxB,SAAS;YACX,CAAC;YACD,MAAM,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,EAAE,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC;YAC1C,4DAA4D;YAC5D,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;gBACvB,SAAS;YACX,CAAC;YACD,IAAI,CAAC,cAAc,CAAC,IAAI,EAAE,KAAK,CAAC,OAAO,CAAC,EAAE,CAAC;gBACzC,OAAO,IAAI,CAAC,EAAE,GAAG,IAAI,CAAC,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,QAAQ,CAAC;YACvD,CAAC;YACD,IAAI,IAAI,CAAC,EAAE,GAAG,IAAI,CAAC,IAAI,KAAK,CAAC,EAAE,CAAC;gBAC9B,OAAO,GAAG,OAAO,CAAC;YACpB,CAAC;QACH,CAAC;QACD,OAAO,OAAO,CAAC;IACjB,CAAC;CACF,CAAC"}

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

@@ -0,0 +1,6 @@
+import { type VerbalFrame } from "./verbal-frame.js";
+/**
+ * 相 read `xiàng` where it is a photograph taken.
+ */
+export declare const XIANG_FRAME: VerbalFrame;
+//# sourceMappingURL=xiang-frames.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"xiang-frames.d.ts","sourceRoot":"","sources":["../../src/decode/xiang-frames.ts"],"names":[],"mappings":"AAQA,OAAO,EAGL,KAAK,WAAW,EACjB,MAAM,mBAAmB,CAAC;AAoD3B;;GAEG;AACH,eAAO,MAAM,WAAW,EAAE,WAMzB,CAAC"}

package/dist/decode/xiang-frames.js

@@ -0,0 +1,65 @@
+/**
+ * Whether a 相 is a photograph or is "each other".
+ *
+ * 照相 is `zhàoxiàng` and a word. Taken apart, as in 照了一张相 and 你照的相,
+ * the 相 stands on its own and took the commoner `xiāng` of 相信 and 互相.
+ */
+import { QUANTITY_CHARACTERS } from "../numerals/characters.js";
+import { closesTaggedWord, syllableOf, } from "./verbal-frame.js";
+/**
+ * The verb of the photograph, in both scripts.
+ */
+const TAKING = "照";
+/**
+ * What can stand between 照 and its 相: aspect, a count and 的.
+ *
+ * 照了一张相, 照张相, 照个相, 照过相, 照的相.
+ */
+const BETWEEN = new Set([
+    "了",
+    "过",
+    "過",
+    "的",
+    "张",
+    "張",
+    "个",
+    "個",
+    "几",
+    "幾",
+    ...QUANTITY_CHARACTERS,
+]);
+/**
+ * How far back the 照 may be.
+ */
+const REACH = 4;
+/**
+ * Whether the 相 at a position is the photograph of a 照 before it.
+ *
+ * A word ending at the 照 takes it away from the 相, which keeps 按照 and
+ * 护照 out.
+ */
+function isPhotographAt(context, at) {
+    const { characters } = context;
+    for (let back = at - 1; back >= Math.max(0, at - REACH); back--) {
+        /* c8 ignore next -- the position is inside the run */
+        const character = characters[back] ?? "";
+        if (character === TAKING) {
+            return !closesTaggedWord(context, back + 1);
+        }
+        if (!BETWEEN.has(character)) {
+            return false;
+        }
+    }
+    return false;
+}
+/**
+ * 相 read `xiàng` where it is a photograph taken.
+ */
+export const XIANG_FRAME = {
+    characters: new Set(["相"]),
+    reading: syllableOf("xiàng"),
+    boundBefore: new Set(),
+    boundAfter: new Set(),
+    isVerbalAt: isPhotographAt,
+};
+//# sourceMappingURL=xiang-frames.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"xiang-frames.js","sourceRoot":"","sources":["../../src/decode/xiang-frames.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,OAAO,EAAE,mBAAmB,EAAE,MAAM,2BAA2B,CAAC;AAEhE,OAAO,EACL,gBAAgB,EAChB,UAAU,GAEX,MAAM,mBAAmB,CAAC;AAE3B;;GAEG;AACH,MAAM,MAAM,GAAG,GAAG,CAAC;AAEnB;;;;GAIG;AACH,MAAM,OAAO,GAAG,IAAI,GAAG,CAAC;IACtB,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG;IACH,GAAG,mBAAmB;CACvB,CAAC,CAAC;AAEH;;GAEG;AACH,MAAM,KAAK,GAAG,CAAC,CAAC;AAEhB;;;;;GAKG;AACH,SAAS,cAAc,CAAC,OAAoB,EAAE,EAAU;IACtD,MAAM,EAAE,UAAU,EAAE,GAAG,OAAO,CAAC;IAC/B,KAAK,IAAI,IAAI,GAAG,EAAE,GAAG,CAAC,EAAE,IAAI,IAAI,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,EAAE,GAAG,KAAK,CAAC,EAAE,IAAI,EAAE,EAAE,CAAC;QAChE,sDAAsD;QACtD,MAAM,SAAS,GAAG,UAAU,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;QACzC,IAAI,SAAS,KAAK,MAAM,EAAE,CAAC;YACzB,OAAO,CAAC,gBAAgB,CAAC,OAAO,EAAE,IAAI,GAAG,CAAC,CAAC,CAAC;QAC9C,CAAC;QACD,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,SAAS,CAAC,EAAE,CAAC;YAC5B,OAAO,KAAK,CAAC;QACf,CAAC;IACH,CAAC;IACD,OAAO,KAAK,CAAC;AACf,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,WAAW,GAAgB;IACtC,UAAU,EAAE,IAAI,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC;IAC1B,OAAO,EAAE,UAAU,CAAC,OAAO,CAAC;IAC5B,WAAW,EAAE,IAAI,GAAG,EAAE;IACtB,UAAU,EAAE,IAAI,GAAG,EAAE;IACrB,UAAU,EAAE,cAAc;CAC3B,CAAC"}

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

@@ -0,0 +1,6 @@
+import { type VerbalFrame } from "./verbal-frame.js";
+/**
+ * 转 read `zhuàn` where it is turning round.
+ */
+export declare const ZHUAN_FRAME: VerbalFrame;
+//# sourceMappingURL=zhuan-frames.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"zhuan-frames.d.ts","sourceRoot":"","sources":["../../src/decode/zhuan-frames.ts"],"names":[],"mappings":"AASA,OAAO,EAKL,KAAK,WAAW,EAEjB,MAAM,mBAAmB,CAAC;AA2F3B;;GAEG;AACH,eAAO,MAAM,WAAW,EAAE,WAMzB,CAAC"}

package/dist/decode/zhuan-frames.js

@@ -0,0 +1,97 @@
+/**
+ * Whether a 转 is turning round or changing course.
+ *
+ * `zhuàn` turns round and round, and `zhuǎn` changes: direction (转东南),
+ * state (天气转凉), post (转御史大夫) or vehicle (转火车). Both are verbs, so
+ * the tag says nothing, and what tells them apart is the motion around them.
+ */
+import { QUANTITY_CHARACTERS } from "../numerals/characters.js";
+import { wordEndingAt } from "./rules.js";
+import { isBareAt, isRunEnd, startsWithAny, syllableOf, charactersOf, } from "./verbal-frame.js";
+/**
+ * What going round something is said with, in both scripts: 绕着地球转, 围着你转,
+ * 围绕太阳转.
+ */
+const AROUND = ["绕", "繞", "围着", "圍著", "围绕", "圍繞"];
+/**
+ * What can close a clause after a 转 that is going round.
+ *
+ * The end of the run, 的 (世界不是围着你转的), aspect and the sentence
+ * particles. 个 is out, since 转个身 is turning round once.
+ */
+const CLOSING = new Set(["的", "了", "着", "呀", "啊", "吧", "呢"]);
+/**
+ * What only turning round is followed by.
+ *
+ * 转起来 (风车转起来), 转得 (转得晕头转向), and 转个不停. 转过身 is turning
+ * round once and reads `zhuǎn`, so 过 is out.
+ */
+const SPINNING = ["起来", "得", "个不停"];
+/**
+ * The laps a 转 goes round: 转一圈, 转了几圈, 转个圈.
+ */
+const LAP = "圈";
+/**
+ * Whether the text after a position is a number of laps, as in 一圈, 了三圈
+ * and 个圈.
+ */
+function isLapAt(context, at) {
+    const { characters } = context;
+    let from = at;
+    if (characters[from] === "了" || characters[from] === "个") {
+        from++;
+    }
+    while (QUANTITY_CHARACTERS.has(characters[from] ?? "")) {
+        from++;
+    }
+    return characters[from] === LAP;
+}
+/**
+ * What says a 转 is still at it: 这条流水线每天都在转.
+ */
+const ONGOING = new Set(["在", "正在"]);
+/**
+ * Whether the clause before a position says the 转 goes round something.
+ */
+function isAroundBefore(context, at) {
+    const before = context.characters.slice(0, at).join("");
+    return AROUND.some((word) => before.includes(word));
+}
+/**
+ * Whether the 转 at a position is turning round.
+ *
+ * Going round something and still at it: 地球绕着太阳转, 这条流水线每天都在转.
+ */
+function isTurningAt(context, at) {
+    if (startsWithAny(context, at + 1, SPINNING) || isLapAt(context, at + 1)) {
+        return true;
+    }
+    const closes = isRunEnd(context, at + 1) || isBareAt(context, at + 1, CLOSING);
+    if (!closes) {
+        return false;
+    }
+    return (isAroundBefore(context, at) || ONGOING.has(wordEndingAt(context, at) ?? ""));
+}
+/**
+ * The characters that bind it into a compound reading it otherwise.
+ *
+ * Bound before it: 运转, 旋转, 好转, 逆转, 周转 and the other compounds
+ * reading it `zhuǎn`, so 绕地球运转 is not read as spinning. 公转 and 自转 read
+ * `zhuàn` and are left out.
+ */
+const BOUND_BEFORE = charactersOf("运運旋扭流好逆周回反翻倒婉辗輾急掉");
+/**
+ * The characters after it that bind it into such a compound.
+ */
+const BOUND_AFTER = new Set();
+/**
+ * 转 read `zhuàn` where it is turning round.
+ */
+export const ZHUAN_FRAME = {
+    characters: new Set(["转", "轉"]),
+    reading: syllableOf("zhuàn"),
+    boundBefore: BOUND_BEFORE,
+    boundAfter: BOUND_AFTER,
+    isVerbalAt: isTurningAt,
+};
+//# sourceMappingURL=zhuan-frames.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"zhuan-frames.js","sourceRoot":"","sources":["../../src/decode/zhuan-frames.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,EAAE,mBAAmB,EAAE,MAAM,2BAA2B,CAAC;AAChE,OAAO,EAAoB,YAAY,EAAE,MAAM,YAAY,CAAC;AAC5D,OAAO,EACL,QAAQ,EACR,QAAQ,EACR,aAAa,EACb,UAAU,EAEV,YAAY,GACb,MAAM,mBAAmB,CAAC;AAE3B;;;GAGG;AACH,MAAM,MAAM,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,CAAC;AAElD;;;;;GAKG;AACH,MAAM,OAAO,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;AAE7D;;;;;GAKG;AACH,MAAM,QAAQ,GAAG,CAAC,IAAI,EAAE,GAAG,EAAE,KAAK,CAAC,CAAC;AAEpC;;GAEG;AACH,MAAM,GAAG,GAAG,GAAG,CAAC;AAEhB;;;GAGG;AACH,SAAS,OAAO,CAAC,OAAoB,EAAE,EAAU;IAC/C,MAAM,EAAE,UAAU,EAAE,GAAG,OAAO,CAAC;IAC/B,IAAI,IAAI,GAAG,EAAE,CAAC;IACd,IAAI,UAAU,CAAC,IAAI,CAAC,KAAK,GAAG,IAAI,UAAU,CAAC,IAAI,CAAC,KAAK,GAAG,EAAE,CAAC;QACzD,IAAI,EAAE,CAAC;IACT,CAAC;IACD,OAAO,mBAAmB,CAAC,GAAG,CAAC,UAAU,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC;QACvD,IAAI,EAAE,CAAC;IACT,CAAC;IACD,OAAO,UAAU,CAAC,IAAI,CAAC,KAAK,GAAG,CAAC;AAClC,CAAC;AAED;;GAEG;AACH,MAAM,OAAO,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC,CAAC;AAErC;;GAEG;AACH,SAAS,cAAc,CAAC,OAAoB,EAAE,EAAU;IACtD,MAAM,MAAM,GAAG,OAAO,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IACxD,OAAO,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,MAAM,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC;AACtD,CAAC;AAED;;;;GAIG;AACH,SAAS,WAAW,CAAC,OAAoB,EAAE,EAAU;IACnD,IAAI,aAAa,CAAC,OAAO,EAAE,EAAE,GAAG,CAAC,EAAE,QAAQ,CAAC,IAAI,OAAO,CAAC,OAAO,EAAE,EAAE,GAAG,CAAC,CAAC,EAAE,CAAC;QACzE,OAAO,IAAI,CAAC;IACd,CAAC;IACD,MAAM,MAAM,GACV,QAAQ,CAAC,OAAO,EAAE,EAAE,GAAG,CAAC,CAAC,IAAI,QAAQ,CAAC,OAAO,EAAE,EAAE,GAAG,CAAC,EAAE,OAAO,CAAC,CAAC;IAClE,IAAI,CAAC,MAAM,EAAE,CAAC;QACZ,OAAO,KAAK,CAAC;IACf,CAAC;IACD,OAAO,CACL,cAAc,CAAC,OAAO,EAAE,EAAE,CAAC,IAAI,OAAO,CAAC,GAAG,CAAC,YAAY,CAAC,OAAO,EAAE,EAAE,CAAC,IAAI,EAAE,CAAC,CAC5E,CAAC;AACJ,CAAC;AAED;;;;;;GAMG;AACH,MAAM,YAAY,GAAG,YAAY,CAAC,mBAAmB,CAAC,CAAC;AAEvD;;GAEG;AACH,MAAM,WAAW,GAAwB,IAAI,GAAG,EAAE,CAAC;AAEnD;;GAEG;AACH,MAAM,CAAC,MAAM,WAAW,GAAgB;IACtC,UAAU,EAAE,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC;IAC/B,OAAO,EAAE,UAAU,CAAC,OAAO,CAAC;IAC5B,WAAW,EAAE,YAAY;IACzB,UAAU,EAAE,WAAW;IACvB,UAAU,EAAE,WAAW;CACxB,CAAC"}

package/dist/dictionary/character-defaults.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"character-defaults.d.ts","sourceRoot":"","sources":["../../src/dictionary/character-defaults.ts"],"names":[],"mappings":"AAcA,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAC;AACxD,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAExD,OAAO,EAGL,gBAAgB,EACjB,MAAM,kBAAkB,CAAC;AAC1B,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,kBAAkB,CAAC;AAErD;;;;;;GAMG;AACH,wBAAgB,iBAAiB,CAC/B,SAAS,EAAE,MAAM,EACjB,OAAO,EAAE,MAAM,GACd,QAAQ,GAAG,SAAS,CAEtB;AAED;;;GAGG;AACH,MAAM,WAAW,iBAAiB;IAChC,QAAQ,CAAC,WAAW,EAAE,gBAAgB,CAAC;IACvC,QAAQ,CAAC,QAAQ,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,QAAQ,EAAE,CAAC,CAAC;IAC5D,QAAQ,CAAC,eAAe,EAAE,MAAM,CAAC;CAClC;AAED;;GAEG;AACH,wBAAgB,sBAAsB,CACpC,OAAO,EAAE,YAAY,EACrB,YAAY,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,WAAW,EAAE,CAAC,EACzD,YAAY,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,WAAW,EAAE,CAAC,GACxD,iBAAiB,CAiDnB"}
+{"version":3,"file":"character-defaults.d.ts","sourceRoot":"","sources":["../../src/dictionary/character-defaults.ts"],"names":[],"mappings":"AAeA,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAC;AACxD,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAExD,OAAO,EAGL,gBAAgB,EACjB,MAAM,kBAAkB,CAAC;AAC1B,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,kBAAkB,CAAC;AAErD;;;;;;GAMG;AACH,wBAAgB,iBAAiB,CAC/B,SAAS,EAAE,MAAM,EACjB,OAAO,EAAE,MAAM,GACd,QAAQ,GAAG,SAAS,CAEtB;AAED;;;GAGG;AACH,MAAM,WAAW,iBAAiB;IAChC,QAAQ,CAAC,WAAW,EAAE,gBAAgB,CAAC;IACvC,QAAQ,CAAC,QAAQ,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,QAAQ,EAAE,CAAC,CAAC;IAC5D,QAAQ,CAAC,eAAe,EAAE,MAAM,CAAC;CAClC;AAED;;GAEG;AACH,wBAAgB,sBAAsB,CACpC,OAAO,EAAE,YAAY,EACrB,YAAY,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,WAAW,EAAE,CAAC,EACzD,YAAY,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,WAAW,EAAE,CAAC,GACxD,iBAAiB,CAsDnB"}

package/dist/dictionary/character-defaults.js

@@ -8,6 +8,7 @@
  */
 import { countCorpusMass, rankByCorpusMass } from "./corpus-mass.js";
 import { demoteReducedNeutrals, isNeutralAlone, resolveFrequencyTones, } from "./frequency-tones.js";
+import { leadWithFreeReading } from "./free-readings.js";
 import { readAlignedReading, readDictionaryReading } from "./reading.js";
 import { pairScripts, TraditionalTable, } from "./traditional.js";
 /**
@@ -39,6 +40,8 @@ export function buildCharacterDefaults(sources, cedictByWord, cedictByHant) {
     // Unihan ranks the readings and the corpus re-ranks the ones it has seen: a
     // character's default is whichever of its readings the dictionary's own words
     // spend the most of jieba's corpus on, and Unihan's order decides the rest.
+    // A character listed in FREE_READINGS then leads with the reading it takes
+    // standing alone, which neither ranking can see.
     const mass = countCorpusMass({ phrase, cedict, jieba }, traditional);
     const defaults = new Map();
     let reducedNeutrals = 0;
@@ -59,7 +62,7 @@ export function buildCharacterDefaults(sources, cedictByWord, cedictByHant) {
             .map((reading) => characterSyllable(character, reading))
             .filter((syllable) => syllable !== undefined);
         if (parsed.length > 0) {
-            defaults.set(character, rankByCorpusMass(character, parsed, readings.fields, mass));
+            defaults.set(character, leadWithFreeReading(character, rankByCorpusMass(character, parsed, readings.fields, mass)));
         }
     }
     return { traditional, defaults, reducedNeutrals };

package/dist/dictionary/character-defaults.js.map

@@ -1 +1 @@
-{"version":3,"file":"character-defaults.js","sourceRoot":"","sources":["../../src/dictionary/character-defaults.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AACH,OAAO,EAAE,eAAe,EAAE,gBAAgB,EAAE,MAAM,kBAAkB,CAAC;AACrE,OAAO,EACL,qBAAqB,EACrB,cAAc,EACd,qBAAqB,GACtB,MAAM,sBAAsB,CAAC;AAG9B,OAAO,EAAE,kBAAkB,EAAE,qBAAqB,EAAE,MAAM,cAAc,CAAC;AACzE,OAAO,EACL,WAAW,EAEX,gBAAgB,GACjB,MAAM,kBAAkB,CAAC;AAG1B;;;;;;GAMG;AACH,MAAM,UAAU,iBAAiB,CAC/B,SAAiB,EACjB,OAAe;IAEf,OAAO,qBAAqB,CAAC,SAAS,EAAE,CAAC,OAAO,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;AAC1D,CAAC;AAYD;;GAEG;AACH,MAAM,UAAU,sBAAsB,CACpC,OAAqB,EACrB,YAAyD,EACzD,YAAyD;IAEzD,MAAM,EAAE,cAAc,EAAE,cAAc,EAAE,MAAM,EAAE,MAAM,EAAE,KAAK,EAAE,GAAG,OAAO,CAAC;IAC1E,6DAA6D;IAC7D,6EAA6E;IAC7E,wEAAwE;IACxE,uEAAuE;IACvE,MAAM,QAAQ,GAAoB,EAAE,CAAC;IACrC,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;QAC3B,MAAM,OAAO,GAAG,kBAAkB,CAAC,KAAK,CAAC,UAAU,EAAE,KAAK,CAAC,QAAQ,CAAC,CAAC;QACrE,QAAQ,CAAC,IAAI,CAAC,GAAG,WAAW,CAAC,KAAK,CAAC,UAAU,EAAE,KAAK,CAAC,WAAW,EAAE,OAAO,CAAC,CAAC,CAAC;IAC9E,CAAC;IACD,MAAM,WAAW,GAAG,gBAAgB,CAAC,KAAK,CACxC,QAAQ,EACR,cAAc,EACd,cAAc,CACf,CAAC;IAEF,8DAA8D;IAC9D,4EAA4E;IAC5E,8EAA8E;IAC9E,4EAA4E;IAC5E,MAAM,IAAI,GAAG,eAAe,CAAC,EAAE,MAAM,EAAE,MAAM,EAAE,KAAK,EAAE,EAAE,WAAW,CAAC,CAAC;IACrE,MAAM,QAAQ,GAAG,IAAI,GAAG,EAA+B,CAAC;IACxD,IAAI,eAAe,GAAG,CAAC,CAAC;IACxB,KAAK,MAAM,CAAC,SAAS,EAAE,QAAQ,CAAC,IAAI,cAAc,EAAE,CAAC;QACnD,MAAM,QAAQ,GAAG,qBAAqB,CAAC,QAAQ,CAAC,CAAC;QACjD,qEAAqE;QACrE,4EAA4E;QAC5E,MAAM,MAAM,GAAG,cAAc,CAAC,SAAS,EAAE;YACvC,GAAG,CAAC,YAAY,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC;YACtC,GAAG,CAAC,YAAY,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC;SACvC,CAAC;YACA,CAAC,CAAC,QAAQ;YACV,CAAC,CAAC,qBAAqB,CAAC,QAAQ,EAAE,QAAQ,CAAC,CAAC;QAC9C,IAAI,MAAM,KAAK,QAAQ,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,QAAQ,CAAC,CAAC,CAAC,EAAE,CAAC;YACrD,eAAe,EAAE,CAAC;QACpB,CAAC;QACD,MAAM,MAAM,GAAG,MAAM;aAClB,GAAG,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,iBAAiB,CAAC,SAAS,EAAE,OAAO,CAAC,CAAC;aACvD,MAAM,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,QAAQ,KAAK,SAAS,CAAC,CAAC;QAChD,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACtB,QAAQ,CAAC,GAAG,CACV,SAAS,EACT,gBAAgB,CAAC,SAAS,EAAE,MAAM,EAAE,QAAQ,CAAC,MAAM,EAAE,IAAI,CAAC,CAC3D,CAAC;QACJ,CAAC;IACH,CAAC;IAED,OAAO,EAAE,WAAW,EAAE,QAAQ,EAAE,eAAe,EAAE,CAAC;AACpD,CAAC"}
+{"version":3,"file":"character-defaults.js","sourceRoot":"","sources":["../../src/dictionary/character-defaults.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AACH,OAAO,EAAE,eAAe,EAAE,gBAAgB,EAAE,MAAM,kBAAkB,CAAC;AACrE,OAAO,EACL,qBAAqB,EACrB,cAAc,EACd,qBAAqB,GACtB,MAAM,sBAAsB,CAAC;AAC9B,OAAO,EAAE,mBAAmB,EAAE,MAAM,oBAAoB,CAAC;AAGzD,OAAO,EAAE,kBAAkB,EAAE,qBAAqB,EAAE,MAAM,cAAc,CAAC;AACzE,OAAO,EACL,WAAW,EAEX,gBAAgB,GACjB,MAAM,kBAAkB,CAAC;AAG1B;;;;;;GAMG;AACH,MAAM,UAAU,iBAAiB,CAC/B,SAAiB,EACjB,OAAe;IAEf,OAAO,qBAAqB,CAAC,SAAS,EAAE,CAAC,OAAO,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;AAC1D,CAAC;AAYD;;GAEG;AACH,MAAM,UAAU,sBAAsB,CACpC,OAAqB,EACrB,YAAyD,EACzD,YAAyD;IAEzD,MAAM,EAAE,cAAc,EAAE,cAAc,EAAE,MAAM,EAAE,MAAM,EAAE,KAAK,EAAE,GAAG,OAAO,CAAC;IAC1E,6DAA6D;IAC7D,6EAA6E;IAC7E,wEAAwE;IACxE,uEAAuE;IACvE,MAAM,QAAQ,GAAoB,EAAE,CAAC;IACrC,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;QAC3B,MAAM,OAAO,GAAG,kBAAkB,CAAC,KAAK,CAAC,UAAU,EAAE,KAAK,CAAC,QAAQ,CAAC,CAAC;QACrE,QAAQ,CAAC,IAAI,CAAC,GAAG,WAAW,CAAC,KAAK,CAAC,UAAU,EAAE,KAAK,CAAC,WAAW,EAAE,OAAO,CAAC,CAAC,CAAC;IAC9E,CAAC;IACD,MAAM,WAAW,GAAG,gBAAgB,CAAC,KAAK,CACxC,QAAQ,EACR,cAAc,EACd,cAAc,CACf,CAAC;IAEF,8DAA8D;IAC9D,4EAA4E;IAC5E,8EAA8E;IAC9E,4EAA4E;IAC5E,2EAA2E;IAC3E,iDAAiD;IACjD,MAAM,IAAI,GAAG,eAAe,CAAC,EAAE,MAAM,EAAE,MAAM,EAAE,KAAK,EAAE,EAAE,WAAW,CAAC,CAAC;IACrE,MAAM,QAAQ,GAAG,IAAI,GAAG,EAA+B,CAAC;IACxD,IAAI,eAAe,GAAG,CAAC,CAAC;IACxB,KAAK,MAAM,CAAC,SAAS,EAAE,QAAQ,CAAC,IAAI,cAAc,EAAE,CAAC;QACnD,MAAM,QAAQ,GAAG,qBAAqB,CAAC,QAAQ,CAAC,CAAC;QACjD,qEAAqE;QACrE,4EAA4E;QAC5E,MAAM,MAAM,GAAG,cAAc,CAAC,SAAS,EAAE;YACvC,GAAG,CAAC,YAAY,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC;YACtC,GAAG,CAAC,YAAY,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC;SACvC,CAAC;YACA,CAAC,CAAC,QAAQ;YACV,CAAC,CAAC,qBAAqB,CAAC,QAAQ,EAAE,QAAQ,CAAC,CAAC;QAC9C,IAAI,MAAM,KAAK,QAAQ,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,QAAQ,CAAC,CAAC,CAAC,EAAE,CAAC;YACrD,eAAe,EAAE,CAAC;QACpB,CAAC;QACD,MAAM,MAAM,GAAG,MAAM;aAClB,GAAG,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,iBAAiB,CAAC,SAAS,EAAE,OAAO,CAAC,CAAC;aACvD,MAAM,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,QAAQ,KAAK,SAAS,CAAC,CAAC;QAChD,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACtB,QAAQ,CAAC,GAAG,CACV,SAAS,EACT,mBAAmB,CACjB,SAAS,EACT,gBAAgB,CAAC,SAAS,EAAE,MAAM,EAAE,QAAQ,CAAC,MAAM,EAAE,IAAI,CAAC,CAC3D,CACF,CAAC;QACJ,CAAC;IACH,CAAC;IAED,OAAO,EAAE,WAAW,EAAE,QAAQ,EAAE,eAAe,EAAE,CAAC;AACpD,CAAC"}

package/dist/dictionary/character-readings.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"character-readings.d.ts","sourceRoot":"","sources":["../../src/dictionary/character-readings.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,uBAAuB,CAAC;AAG5D,eAAO,MAAM,kBAAkB,EAAE,SAAS,cAAc,EAsBvD,CAAC"}
+{"version":3,"file":"character-readings.d.ts","sourceRoot":"","sources":["../../src/dictionary/character-readings.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,uBAAuB,CAAC;AAG5D,eAAO,MAAM,kBAAkB,EAAE,SAAS,cAAc,EA+BvD,CAAC"}

package/dist/dictionary/character-readings.js

@@ -21,5 +21,14 @@ export const CHARACTER_READINGS = [
     reads("子", "zǐ", "and a suffix sense is not a claim about the character"),
     reads("夫", "fū", "nor is 大夫's and 丈夫's reduction"),
     reads("吗", "ma", "but a 语气词 CC-CEDICT reads 轻声 alone keeps it"),
+    // FREE_READINGS: the free verb or classifier leads, and the compounds that
+    // ranked the other reading first keep it.
+    reads("剥", "bāo", "the free verb, not 剥削's bō"),
+    reads("剥削", "bō xuē", "and the compounds keep the literary reading"),
+    reads("切", "qiē", "kHanyuPinlu's qiè is nearly all 一切"),
+    reads("一切", "yī qiè", "which keeps it"),
+    reads("幢", "zhuàng", "the classifier, not 经幢's banner"),
+    reads("脏", "zāng", "dirty, not the organ"),
+    reads("心脏", "xīn zàng", "which 心脏 keeps"),
 ];
 //# sourceMappingURL=character-readings.js.map

package/dist/dictionary/character-readings.js.map

@@ -1 +1 @@
-{"version":3,"file":"character-readings.js","sourceRoot":"","sources":["../../src/dictionary/character-readings.ts"],"names":[],"mappings":"AAQA,OAAO,EAAE,KAAK,EAAE,MAAM,oBAAoB,CAAC;AAE3C,MAAM,CAAC,MAAM,kBAAkB,GAA8B;IAC3D,oEAAoE;IACpE,wEAAwE;IACxE,0EAA0E;IAC1E,yEAAyE;IACzE,gEAAgE;IAChE,KAAK,CAAC,IAAI,EAAE,UAAU,EAAE,uCAAuC,CAAC;IAChE,KAAK,CAAC,IAAI,EAAE,SAAS,EAAE,6CAA6C,CAAC;IACrE,KAAK,CAAC,IAAI,EAAE,QAAQ,EAAE,gDAAgD,CAAC;IACvE,0EAA0E;IAC1E,wEAAwE;IACxE,0EAA0E;IAC1E,4EAA4E;IAC5E,uDAAuD;IACvD,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,yCAAyC,CAAC;IAC3D,KAAK,CAAC,GAAG,EAAE,KAAK,EAAE,+CAA+C,CAAC;IAClE,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,4CAA4C,CAAC;IAC9D,KAAK,CAAC,IAAI,EAAE,QAAQ,EAAE,mDAAmD,CAAC;IAC1E,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,8CAA8C,CAAC;IAChE,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,uDAAuD,CAAC;IACzE,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,gCAAgC,CAAC;IAClD,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,6CAA6C,CAAC;CAChE,CAAC"}
+{"version":3,"file":"character-readings.js","sourceRoot":"","sources":["../../src/dictionary/character-readings.ts"],"names":[],"mappings":"AAQA,OAAO,EAAE,KAAK,EAAE,MAAM,oBAAoB,CAAC;AAE3C,MAAM,CAAC,MAAM,kBAAkB,GAA8B;IAC3D,oEAAoE;IACpE,wEAAwE;IACxE,0EAA0E;IAC1E,yEAAyE;IACzE,gEAAgE;IAChE,KAAK,CAAC,IAAI,EAAE,UAAU,EAAE,uCAAuC,CAAC;IAChE,KAAK,CAAC,IAAI,EAAE,SAAS,EAAE,6CAA6C,CAAC;IACrE,KAAK,CAAC,IAAI,EAAE,QAAQ,EAAE,gDAAgD,CAAC;IACvE,0EAA0E;IAC1E,wEAAwE;IACxE,0EAA0E;IAC1E,4EAA4E;IAC5E,uDAAuD;IACvD,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,yCAAyC,CAAC;IAC3D,KAAK,CAAC,GAAG,EAAE,KAAK,EAAE,+CAA+C,CAAC;IAClE,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,4CAA4C,CAAC;IAC9D,KAAK,CAAC,IAAI,EAAE,QAAQ,EAAE,mDAAmD,CAAC;IAC1E,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,8CAA8C,CAAC;IAChE,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,uDAAuD,CAAC;IACzE,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,gCAAgC,CAAC;IAClD,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,6CAA6C,CAAC;IAC/D,2EAA2E;IAC3E,0CAA0C;IAC1C,KAAK,CAAC,GAAG,EAAE,KAAK,EAAE,4BAA4B,CAAC;IAC/C,KAAK,CAAC,IAAI,EAAE,QAAQ,EAAE,6CAA6C,CAAC;IACpE,KAAK,CAAC,GAAG,EAAE,KAAK,EAAE,oCAAoC,CAAC;IACvD,KAAK,CAAC,IAAI,EAAE,QAAQ,EAAE,gBAAgB,CAAC;IACvC,KAAK,CAAC,GAAG,EAAE,QAAQ,EAAE,iCAAiC,CAAC;IACvD,KAAK,CAAC,GAAG,EAAE,MAAM,EAAE,sBAAsB,CAAC;IAC1C,KAAK,CAAC,IAAI,EAAE,UAAU,EAAE,gBAAgB,CAAC;CAC1C,CAAC"}

package/dist/dictionary/free-readings.d.ts

@@ -0,0 +1,70 @@
+import { type Syllable } from "../syllable/syllable.js";
+/**
+ * A character whose reading as a word of its own is not the one it leads with.
+ */
+export interface FreeReading {
+    /** The character, in either script. */
+    readonly character: string;
+    /** The reading it takes standing alone, as tone-marked pinyin. */
+    readonly reading: string;
+    /** The words that carry the other reading, which is why the sources rank it
+     * first. */
+    readonly displaces: string;
+}
+/**
+ * Characters that read one way inside words and another standing alone.
+ *
+ * A character's default is ranked by `kHanyuPinlu` where it lists several
+ * readings and by the corpus mass of the dictionary's words where it does not,
+ * and both count the character *inside words*. That is the right evidence for
+ * most polyphones and the wrong evidence for these, which have a literary
+ * reading in their compounds and an everyday one as a free verb or classifier.
+ * 剥削, 剥夺 and 剥落 give 剥 `bō`, and a reader peeling an orange says `bāo`.
+ * `kHanyuPinlu` counts 切 `qiè` 998 times against `qiē` 70, nearly all of it
+ * 一切, and 切土豆 is `qiē`.
+ *
+ * The compounds are dictionary words and keep their own readings. The default
+ * is what a character segmented on its own is read with, which is where these
+ * stand as free words, and what an unknown compound is derived from.
+ *
+ * **What is listed here.** A character whose 现代汉语词典 entry gives the listed
+ * reading to its free use and the other to its compounds or a literary sense,
+ * found by reading example sentences rather than by a measure. CPP labels
+ * news text, where these characters mostly sit inside words, so it is the check
+ * that nothing broke rather than the evidence for the list. What is *not*
+ * listed is a character whose free uses split, such as 卡 (a card, `kǎ`, and
+ * stuck, `qiǎ`) or 背 (the back and to recite, `bèi`, and to carry, `bēi`): a
+ * default cannot choose between two free words, and a context rule can. Nor
+ * is a character that names and transliterations use on its own: 勒 is `lēi`
+ * in 勒得难受 and `lè` in every 希特勒 and 达勒姆 the dictionary lacks, and 盛 is
+ * `chéng` in 盛汤 and the surname `Shèng` in 刘盛. Listing either cost CPP 25
+ * and 7 cases.
+ *
+ * **The reading must be one the sources already give the character.** It is
+ * moved to the front of the ranking rather than written over it, so a source
+ * refresh that dropped it would fail the build rather than leave a reading
+ * nothing attests. That is why 剝 and 掙 are missing beside 剥 and 挣: Unihan
+ * gives them `bō` and `zhēng` alone.
+ */
+export declare const FREE_READINGS: readonly FreeReading[];
+/**
+ * Read a free reading into a syllable, throwing if it is not one.
+ *
+ * Throwing for the reason {@link import("./overrides.js").readOverrideReading}
+ * does: an entry that failed to parse would leave the ranking it was written
+ * to correct, and nothing would notice.
+ */
+export declare function freeSyllable(free: FreeReading): Syllable;
+/**
+ * The table, keyed by character and parsed into syllables.
+ */
+export declare const FREE_SYLLABLES: ReadonlyMap<string, Syllable>;
+/**
+ * Lead a character's ranked readings with its free reading, where it has one.
+ *
+ * The rest keep the order the ranking gave them. A listed reading the ranking
+ * does not hold throws, which is the build failing on a source that stopped
+ * attesting it.
+ */
+export declare function leadWithFreeReading(character: string, ranked: readonly Syllable[]): readonly Syllable[];
+//# sourceMappingURL=free-readings.d.ts.map

package/dist/dictionary/free-readings.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"free-readings.d.ts","sourceRoot":"","sources":["../../src/dictionary/free-readings.ts"],"names":[],"mappings":"AAAA,OAAO,EAAgB,KAAK,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAEtE;;GAEG;AACH,MAAM,WAAW,WAAW;IAC1B,uCAAuC;IACvC,QAAQ,CAAC,SAAS,EAAE,MAAM,CAAC;IAC3B,kEAAkE;IAClE,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;IACzB;gBACY;IACZ,QAAQ,CAAC,SAAS,EAAE,MAAM,CAAC;CAC5B;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAkCG;AACH,eAAO,MAAM,aAAa,EAAE,SAAS,WAAW,EA0B/C,CAAC;AAEF;;;;;;GAMG;AACH,wBAAgB,YAAY,CAAC,IAAI,EAAE,WAAW,GAAG,QAAQ,CAQxD;AAED;;GAEG;AACH,eAAO,MAAM,cAAc,EAAE,WAAW,CAAC,MAAM,EAAE,QAAQ,CAExD,CAAC;AAaF;;;;;;GAMG;AACH,wBAAgB,mBAAmB,CACjC,SAAS,EAAE,MAAM,EACjB,MAAM,EAAE,SAAS,QAAQ,EAAE,GAC1B,SAAS,QAAQ,EAAE,CAYrB"}

package/dist/dictionary/free-readings.js

@@ -0,0 +1,108 @@
+import { readSyllable } from "../syllable/syllable.js";
+/**
+ * Characters that read one way inside words and another standing alone.
+ *
+ * A character's default is ranked by `kHanyuPinlu` where it lists several
+ * readings and by the corpus mass of the dictionary's words where it does not,
+ * and both count the character *inside words*. That is the right evidence for
+ * most polyphones and the wrong evidence for these, which have a literary
+ * reading in their compounds and an everyday one as a free verb or classifier.
+ * 剥削, 剥夺 and 剥落 give 剥 `bō`, and a reader peeling an orange says `bāo`.
+ * `kHanyuPinlu` counts 切 `qiè` 998 times against `qiē` 70, nearly all of it
+ * 一切, and 切土豆 is `qiē`.
+ *
+ * The compounds are dictionary words and keep their own readings. The default
+ * is what a character segmented on its own is read with, which is where these
+ * stand as free words, and what an unknown compound is derived from.
+ *
+ * **What is listed here.** A character whose 现代汉语词典 entry gives the listed
+ * reading to its free use and the other to its compounds or a literary sense,
+ * found by reading example sentences rather than by a measure. CPP labels
+ * news text, where these characters mostly sit inside words, so it is the check
+ * that nothing broke rather than the evidence for the list. What is *not*
+ * listed is a character whose free uses split, such as 卡 (a card, `kǎ`, and
+ * stuck, `qiǎ`) or 背 (the back and to recite, `bèi`, and to carry, `bēi`): a
+ * default cannot choose between two free words, and a context rule can. Nor
+ * is a character that names and transliterations use on its own: 勒 is `lēi`
+ * in 勒得难受 and `lè` in every 希特勒 and 达勒姆 the dictionary lacks, and 盛 is
+ * `chéng` in 盛汤 and the surname `Shèng` in 刘盛. Listing either cost CPP 25
+ * and 7 cases.
+ *
+ * **The reading must be one the sources already give the character.** It is
+ * moved to the front of the ranking rather than written over it, so a source
+ * refresh that dropped it would fail the build rather than leave a reading
+ * nothing attests. That is why 剝 and 掙 are missing beside 剥 and 挣: Unihan
+ * gives them `bō` and `zhēng` alone.
+ */
+export const FREE_READINGS = [
+    { character: "幢", reading: "zhuàng", displaces: "幢幢 and 经幢, a banner" },
+    { character: "抡", reading: "lūn", displaces: "抡材 and 抡元, to select" },
+    { character: "掄", reading: "lūn", displaces: "掄材 and 掄元, to select" },
+    { character: "剥", reading: "bāo", displaces: "剥削, 剥夺 and 剥落" },
+    { character: "削", reading: "xiāo", displaces: "削弱, 削减 and 剥削" },
+    { character: "铺", reading: "pū", displaces: "店铺 and 床铺, the noun" },
+    { character: "鋪", reading: "pū", displaces: "店鋪 and 床鋪, the noun" },
+    { character: "咳", reading: "ké", displaces: "the interjection hāi" },
+    { character: "哄", reading: "hǒng", displaces: "哄堂 and 哄抢, an uproar" },
+    { character: "刨", reading: "páo", displaces: "刨床 and 刨子, a plane" },
+    { character: "攒", reading: "zǎn", displaces: "攒聚 and 攒眉, to gather" },
+    { character: "攢", reading: "zǎn", displaces: "攢聚 and 攢眉, to gather" },
+    { character: "绷", reading: "bēng", displaces: "绷着脸, a stiff face" },
+    { character: "繃", reading: "bēng", displaces: "繃著臉, a stiff face" },
+    { character: "翘", reading: "qiào", displaces: "翘首 and 翘楚" },
+    { character: "翹", reading: "qiào", displaces: "翹首 and 翹楚" },
+    { character: "划", reading: "huá", displaces: "计划, 划分 and 规划" },
+    { character: "劃", reading: "huá", displaces: "計劃, 劃分 and 規劃" },
+    { character: "曲", reading: "qǔ", displaces: "曲线, 弯曲 and 曲折" },
+    { character: "切", reading: "qiē", displaces: "一切, 亲切 and 切实" },
+    { character: "嚼", reading: "jiáo", displaces: "咀嚼, the literary word" },
+    { character: "倔", reading: "juè", displaces: "倔强 and 倔犟" },
+    { character: "脏", reading: "zāng", displaces: "心脏 and 内脏, an organ" },
+    { character: "挣", reading: "zhèng", displaces: "挣扎, to struggle" },
+    { character: "漂", reading: "piāo", displaces: "漂亮 and 漂白" },
+];
+/**
+ * Read a free reading into a syllable, throwing if it is not one.
+ *
+ * Throwing for the reason {@link import("./overrides.js").readOverrideReading}
+ * does: an entry that failed to parse would leave the ranking it was written
+ * to correct, and nothing would notice.
+ */
+export function freeSyllable(free) {
+    const syllable = readSyllable(free.reading);
+    if (syllable?.tone === undefined) {
+        throw new Error(`free reading for ${free.character} is not a reading: ${free.reading}`);
+    }
+    return syllable;
+}
+/**
+ * The table, keyed by character and parsed into syllables.
+ */
+export const FREE_SYLLABLES = new Map(FREE_READINGS.map((free) => [free.character, freeSyllable(free)]));
+/**
+ * Whether two syllables are the same reading, 儿化 aside.
+ */
+function sameReading(left, right) {
+    return (left.initial === right.initial &&
+        left.final === right.final &&
+        left.tone === right.tone);
+}
+/**
+ * Lead a character's ranked readings with its free reading, where it has one.
+ *
+ * The rest keep the order the ranking gave them. A listed reading the ranking
+ * does not hold throws, which is the build failing on a source that stopped
+ * attesting it.
+ */
+export function leadWithFreeReading(character, ranked) {
+    const free = FREE_SYLLABLES.get(character);
+    if (free === undefined) {
+        return ranked;
+    }
+    const lead = ranked.find((syllable) => sameReading(syllable, free));
+    if (lead === undefined) {
+        throw new Error(`free reading for ${character} is not one the sources give it`);
+    }
+    return [lead, ...ranked.filter((syllable) => syllable !== lead)];
+}
+//# sourceMappingURL=free-readings.js.map

package/dist/dictionary/free-readings.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"free-readings.js","sourceRoot":"","sources":["../../src/dictionary/free-readings.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAiB,MAAM,yBAAyB,CAAC;AAetE;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAkCG;AACH,MAAM,CAAC,MAAM,aAAa,GAA2B;IACnD,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,QAAQ,EAAE,SAAS,EAAE,qBAAqB,EAAE;IACvE,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,KAAK,EAAE,SAAS,EAAE,sBAAsB,EAAE;IACrE,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,KAAK,EAAE,SAAS,EAAE,sBAAsB,EAAE;IACrE,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,KAAK,EAAE,SAAS,EAAE,eAAe,EAAE;IAC9D,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,eAAe,EAAE;IAC/D,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,qBAAqB,EAAE;IACnE,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,qBAAqB,EAAE;IACnE,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,sBAAsB,EAAE;IACpE,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,sBAAsB,EAAE;IACtE,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,KAAK,EAAE,SAAS,EAAE,oBAAoB,EAAE;IACnE,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,KAAK,EAAE,SAAS,EAAE,sBAAsB,EAAE;IACrE,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,KAAK,EAAE,SAAS,EAAE,sBAAsB,EAAE;IACrE,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,mBAAmB,EAAE;IACnE,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,mBAAmB,EAAE;IACnE,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,WAAW,EAAE;IAC3D,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,WAAW,EAAE;IAC3D,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,KAAK,EAAE,SAAS,EAAE,eAAe,EAAE;IAC9D,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,KAAK,EAAE,SAAS,EAAE,eAAe,EAAE;IAC9D,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,eAAe,EAAE;IAC7D,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,KAAK,EAAE,SAAS,EAAE,eAAe,EAAE;IAC9D,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,uBAAuB,EAAE;IACvE,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,KAAK,EAAE,SAAS,EAAE,WAAW,EAAE;IAC1D,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,qBAAqB,EAAE;IACrE,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,OAAO,EAAE,SAAS,EAAE,iBAAiB,EAAE;IAClE,EAAE,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,WAAW,EAAE;CAC5D,CAAC;AAEF;;;;;;GAMG;AACH,MAAM,UAAU,YAAY,CAAC,IAAiB;IAC5C,MAAM,QAAQ,GAAG,YAAY,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAC5C,IAAI,QAAQ,EAAE,IAAI,KAAK,SAAS,EAAE,CAAC;QACjC,MAAM,IAAI,KAAK,CACb,oBAAoB,IAAI,CAAC,SAAS,sBAAsB,IAAI,CAAC,OAAO,EAAE,CACvE,CAAC;IACJ,CAAC;IACD,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,cAAc,GAAkC,IAAI,GAAG,CAClE,aAAa,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,SAAS,EAAE,YAAY,CAAC,IAAI,CAAC,CAAC,CAAC,CAClE,CAAC;AAEF;;GAEG;AACH,SAAS,WAAW,CAAC,IAAc,EAAE,KAAe;IAClD,OAAO,CACL,IAAI,CAAC,OAAO,KAAK,KAAK,CAAC,OAAO;QAC9B,IAAI,CAAC,KAAK,KAAK,KAAK,CAAC,KAAK;QAC1B,IAAI,CAAC,IAAI,KAAK,KAAK,CAAC,IAAI,CACzB,CAAC;AACJ,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,mBAAmB,CACjC,SAAiB,EACjB,MAA2B;IAE3B,MAAM,IAAI,GAAG,cAAc,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;IAC3C,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;QACvB,OAAO,MAAM,CAAC;IAChB,CAAC;IACD,MAAM,IAAI,GAAG,MAAM,CAAC,IAAI,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,WAAW,CAAC,QAAQ,EAAE,IAAI,CAAC,CAAC,CAAC;IACpE,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;QACvB,MAAM,IAAI,KAAK,CACb,oBAAoB,SAAS,iCAAiC,CAC/D,CAAC;IACJ,CAAC;IACD,OAAO,CAAC,IAAI,EAAE,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,QAAQ,KAAK,IAAI,CAAC,CAAC,CAAC;AACnE,CAAC"}

package/dist/dictionary/frequency-table.d.ts

@@ -108,6 +108,14 @@ export declare function countForQuantising(frequency: number, isProperNoun: bool
  * have to be read off the same corpus for the result to mean anything.
  */
 export declare function derivedWordCharge(total: number, highest: number): number;
+/**
+ * The decoding cost of a frequency bucket.
+ *
+ * What {@link FrequencyTable.costOf} answers for an entry in that bucket, for a
+ * caller that wants to ask about a bucket rather than an entry: a rule asking
+ * whether an edge is one nobody counted compares its cost against this.
+ */
+export declare function costOfBucket(bucket: number): number;
 /**
  * Quantised word frequencies, packed two to a byte.
  *

package/dist/dictionary/frequency-table.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"frequency-table.d.ts","sourceRoot":"","sources":["../../src/dictionary/frequency-table.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AACH,eAAO,MAAM,iBAAiB,KAAK,CAAC;AAIpC;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAoCG;AACH,eAAO,MAAM,WAAW,OAAO,CAAC;AAEhC;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA0CG;AACH,eAAO,MAAM,cAAc,IAAI,CAAC;AAEhC;;;;;;;;GAQG;AACH,wBAAgB,kBAAkB,CAChC,SAAS,EAAE,MAAM,EACjB,YAAY,EAAE,OAAO,GACpB,MAAM,CAER;AAED;;;;;;GAMG;AACH,wBAAgB,iBAAiB,CAAC,KAAK,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,GAAG,MAAM,CAExE;AAED;;;;;;GAMG;AACH,qBAAa,cAAc;;IACzB;;;;;;;;;;;OAWG;IACH,MAAM,CAAC,KAAK,CAAC,WAAW,EAAE,SAAS,MAAM,EAAE,GAAG,cAAc,CA4B3D;IAED;;OAEG;IACH,MAAM,CAAC,IAAI,CAAC,MAAM,EAAE,UAAU,EAAE,IAAI,EAAE,MAAM,GAAG,cAAc,CAE5D;IAKD;;;OAGG;IACH,OAAO,eAGN;IAED;;OAEG;IACH,IAAI,IAAI,IAAI,MAAM,CAEjB;IAED;;;;;OAKG;IACH,QAAQ,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAM9B;IAED;;;;;;;;OAQG;IACH,MAAM,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAE5B;IAED;;OAEG;IACH,SAAS,IAAI,UAAU,CAEtB;CACF"}
+{"version":3,"file":"frequency-table.d.ts","sourceRoot":"","sources":["../../src/dictionary/frequency-table.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AACH,eAAO,MAAM,iBAAiB,KAAK,CAAC;AAIpC;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAoCG;AACH,eAAO,MAAM,WAAW,OAAO,CAAC;AAEhC;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA0CG;AACH,eAAO,MAAM,cAAc,IAAI,CAAC;AAEhC;;;;;;;;GAQG;AACH,wBAAgB,kBAAkB,CAChC,SAAS,EAAE,MAAM,EACjB,YAAY,EAAE,OAAO,GACpB,MAAM,CAER;AAED;;;;;;GAMG;AACH,wBAAgB,iBAAiB,CAAC,KAAK,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,GAAG,MAAM,CAExE;AAED;;;;;;GAMG;AACH,wBAAgB,YAAY,CAAC,MAAM,EAAE,MAAM,GAAG,MAAM,CAEnD;AAED;;;;;;GAMG;AACH,qBAAa,cAAc;;IACzB;;;;;;;;;;;OAWG;IACH,MAAM,CAAC,KAAK,CAAC,WAAW,EAAE,SAAS,MAAM,EAAE,GAAG,cAAc,CA4B3D;IAED;;OAEG;IACH,MAAM,CAAC,IAAI,CAAC,MAAM,EAAE,UAAU,EAAE,IAAI,EAAE,MAAM,GAAG,cAAc,CAE5D;IAKD;;;OAGG;IACH,OAAO,eAGN;IAED;;OAEG;IACH,IAAI,IAAI,IAAI,MAAM,CAEjB;IAED;;;;;OAKG;IACH,QAAQ,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAM9B;IAED;;;;;;;;OAQG;IACH,MAAM,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAE5B;IAED;;OAEG;IACH,SAAS,IAAI,UAAU,CAEtB;CACF"}

package/dist/dictionary/frequency-table.js

@@ -113,6 +113,16 @@ export function countForQuantising(frequency, isProperNoun) {
 export function derivedWordCharge(total, highest) {
     return MAX_BUCKET * (Math.log(total) / Math.log1p(highest) - 1);
 }
+/**
+ * The decoding cost of a frequency bucket.
+ *
+ * What {@link FrequencyTable.costOf} answers for an entry in that bucket, for a
+ * caller that wants to ask about a bucket rather than an entry: a rule asking
+ * whether an edge is one nobody counted compares its cost against this.
+ */
+export function costOfBucket(bucket) {
+    return MAX_BUCKET - bucket + WORD_CHARGE;
+}
 /**
  * Quantised word frequencies, packed two to a byte.
  *
@@ -202,7 +212,7 @@ export class FrequencyTable {
      * every extra word in a path pays the charge again.
      */
     costOf(index) {
-        return MAX_BUCKET - this.bucketOf(index) + WORD_CHARGE;
+        return costOfBucket(this.bucketOf(index));
     }
     /**
      * The packed bytes, ready to be written out.