mdast-util-to-markdown-cjk-friendly 1.0.0 → 1.0.1

package/README.md

@@ -2,6 +2,8 @@
 
 `mdast-util-to-markdown` extension for CJK-friendly emphasis and strong serialization.
 
+Requires `mdast-util-to-markdown` 2.1.3 or later.
+
 ## Usage
 
 ```js

package/dist/index.d.mts

@@ -1,10 +1,6 @@
 import { Options } from "mdast-util-to-markdown";
-
 //#region src/index.d.ts
-
-/**
- * Add CJK-friendly `toMarkdown` handlers for emphasis and strong.
- */
+/** Add CJK-friendly `toMarkdown` handlers for emphasis and strong. */
 declare function cjkFriendlyToMarkdown(): Options;
 //#endregion
 export { cjkFriendlyToMarkdown };

package/dist/index.mjs

@@ -1,248 +1,13 @@
-import { classifyCharacter, classifyPrecedingCharacter, isCjk, isCjkOrIvs, isNonCjkPunctuation, isNonEmojiGeneralUseVS, isSpaceOrPunctuation, isUnicodeWhitespace } from "micromark-extension-cjk-friendly-util";
-import { codes } from "micromark-util-symbol";
-
+import { n as cjkFriendlyPhrasingHandlers, t as cjkFriendlyAttention } from "./phrasing-CVe5_g2v.mjs";
+import { defaultHandlers } from "mdast-util-to-markdown";
 //#region src/index.ts
-const encodedOutsideBoundary = ";".codePointAt(0) ?? codes.eof;
-const encodedInsideBoundary = "&".codePointAt(0) ?? codes.eof;
-/**
-* Add CJK-friendly `toMarkdown` handlers for emphasis and strong.
-*/
+/** Add CJK-friendly `toMarkdown` handlers for emphasis and strong. */
 function cjkFriendlyToMarkdown() {
 	return { handlers: {
-		emphasis,
-		strong,
-		text
+		...cjkFriendlyPhrasingHandlers,
+		emphasis: cjkFriendlyAttention(defaultHandlers.emphasis),
+		strong: cjkFriendlyAttention(defaultHandlers.strong)
 	} };
 }
-emphasis.peek = emphasisPeek;
-strong.peek = strongPeek;
-function emphasis(node, parent, state, info) {
-	return serializeAttention(node, parent, state, info, emphasisPeek(node, parent, state), 1);
-}
-function strong(node, parent, state, info) {
-	return serializeAttention(node, parent, state, info, strongPeek(node, parent, state), 2);
-}
-function emphasisPeek(_, _parent, state) {
-	return state.options.emphasis || "*";
-}
-function strongPeek(_, _parent, state) {
-	return state.options.strong || "*";
-}
-function serializeAttention(node, parent, state, info, marker, size) {
-	const sequence = marker.repeat(size);
-	const exit = state.enter(size === 1 ? "emphasis" : "strong");
-	const tracker = state.createTracker(info);
-	const before = tracker.move(sequence);
-	let between = tracker.move(state.containerPhrasing(node, {
-		after: marker,
-		before,
-		...tracker.current()
-	}));
-	const beforeBoundary = resolveBeforeBoundary(node, parent, state, info.before);
-	const afterBoundary = resolveAfterBoundary(node, parent, state, info.after);
-	const open = encodeInfoCjk(beforeBoundary, firstCodePoint(between), marker, "open");
-	if (open.inside && between) between = encodeFirstCodePoint(between);
-	const close = encodeInfoCjk({
-		current: lastCodePoint(between),
-		previous: codePointBeforeLast(between)
-	}, afterBoundary, marker, "close");
-	if (close.inside && between) between = encodeLastCodePoint(between);
-	const after = tracker.move(sequence);
-	exit();
-	const encodeAfterSupplementary = close.outside && shouldEncodeAfterSupplementaryText(parent, state);
-	state.attentionEncodeSurroundingInfo = {
-		after: close.outside && !encodeAfterSupplementary,
-		before: open.outside
-	};
-	getCjkFriendlyState(state).cjkFriendlyEncodeAfterSupplementaryText = encodeAfterSupplementary;
-	return before + between + after;
-}
-function text(node, _parent, state, info) {
-	const cjkFriendlyState = getCjkFriendlyState(state);
-	if (!cjkFriendlyState.cjkFriendlyEncodeAfterSupplementaryText) return state.safe(node.value, info);
-	cjkFriendlyState.cjkFriendlyEncodeAfterSupplementaryText = false;
-	const [first = ""] = [...node.value];
-	const rest = node.value.slice(first.length);
-	return `${encodeCharacterReference(first.codePointAt(0) ?? codes.eof)}${state.safe(rest, {
-		...info,
-		before: ";"
-	})}`;
-}
-function encodeInfoCjk(before, after, marker, target) {
-	const beforeKind = classifyBoundaryBefore(before);
-	const afterKind = classifyCharacter(after);
-	if (!isCjkOrIvs(beforeKind) && !isCjkOrIvs(afterKind)) return target === "open" ? encodeInfoFallback(beforeKind, afterKind, marker) : encodeInfoFallback(afterKind, beforeKind, marker);
-	const raw = {
-		inside: false,
-		outside: false
-	};
-	const preserveOutside = {
-		inside: true,
-		outside: false
-	};
-	const preserveInside = {
-		inside: false,
-		outside: true
-	};
-	const encodeBoth = {
-		inside: true,
-		outside: true
-	};
-	for (const candidate of [
-		raw,
-		preserveOutside,
-		preserveInside,
-		encodeBoth
-	]) {
-		const candidateBefore = target === "open" ? candidate.outside ? encodedBoundaryBeforeContext : before : candidate.inside ? encodedBoundaryBeforeContext : before;
-		const candidateAfter = target === "open" ? candidate.inside ? encodedBoundaryAfter : after : candidate.outside ? encodedBoundaryAfter : after;
-		if (target === "open" ? canOpen(marker, candidateBefore, candidateAfter) : canClose(marker, candidateBefore, candidateAfter)) return candidate;
-	}
-	return encodeBoth;
-}
-function encodeInfoFallback(outsideKind, insideKind, marker) {
-	if (isLetterLike(outsideKind)) return isLetterLike(insideKind) ? marker === "_" ? {
-		inside: true,
-		outside: true
-	} : {
-		inside: false,
-		outside: false
-	} : isUnicodeWhitespace(insideKind) ? {
-		inside: true,
-		outside: true
-	} : {
-		inside: false,
-		outside: true
-	};
-	if (isUnicodeWhitespace(outsideKind)) return isLetterLike(insideKind) ? {
-		inside: false,
-		outside: false
-	} : isUnicodeWhitespace(insideKind) ? {
-		inside: true,
-		outside: true
-	} : {
-		inside: false,
-		outside: false
-	};
-	return isUnicodeWhitespace(insideKind) ? {
-		inside: true,
-		outside: false
-	} : {
-		inside: false,
-		outside: false
-	};
-}
-function canOpen(marker, before, afterCode) {
-	const beforeKind = classifyBoundaryBefore(before);
-	const { close, open } = getAttentionSides(beforeKind, classifyCharacter(afterCode));
-	return marker === "_" ? open && (isSpaceOrPunctuation(beforeKind) || !close) : open;
-}
-function canClose(marker, before, afterCode) {
-	const afterKind = classifyCharacter(afterCode);
-	const { close, open } = getAttentionSides(classifyBoundaryBefore(before), afterKind);
-	return marker === "_" ? close && (isSpaceOrPunctuation(afterKind) || !open) : close;
-}
-function getAttentionSides(beforeKind, afterKind) {
-	const beforeNonCjkPunctuation = isNonCjkPunctuation(beforeKind);
-	const beforeSpaceOrNonCjkPunctuation = beforeNonCjkPunctuation || isUnicodeWhitespace(beforeKind);
-	const afterNonCjkPunctuation = isNonCjkPunctuation(afterKind);
-	const afterSpaceOrNonCjkPunctuation = afterNonCjkPunctuation || isUnicodeWhitespace(afterKind);
-	return {
-		open: !afterSpaceOrNonCjkPunctuation || afterNonCjkPunctuation && (beforeSpaceOrNonCjkPunctuation || isCjkOrIvs(beforeKind)),
-		close: !beforeSpaceOrNonCjkPunctuation || beforeNonCjkPunctuation && (afterSpaceOrNonCjkPunctuation || isCjk(afterKind))
-	};
-}
-function isLetterLike(kind) {
-	return !isUnicodeWhitespace(kind) && !isNonCjkPunctuation(kind);
-}
-function classifyBoundaryBefore(before) {
-	const kind = classifyCharacter(before.current);
-	return before.current === null || !isNonEmojiGeneralUseVS(kind) ? kind : classifyPrecedingCharacter(kind, () => before.previous, before.current);
-}
-function resolveBeforeBoundary(node, parent, state, fallback) {
-	let current = lastCodePoint(fallback);
-	let previous = codePointBeforeLast(fallback);
-	if (needsPreviousBoundaryRecovery(current) || needsPreviousContext(current, previous)) {
-		const siblingText = getAdjacentSiblingText(node, parent, state, -1);
-		if (siblingText) {
-			current = lastCodePoint(siblingText);
-			previous = codePointBeforeLast(siblingText);
-		}
-	}
-	return {
-		current,
-		previous
-	};
-}
-function resolveAfterBoundary(node, parent, state, fallback) {
-	const current = firstCodePoint(fallback);
-	if (!needsNextBoundaryRecovery(current)) return current;
-	const siblingText = getAdjacentSiblingText(node, parent, state, 1);
-	return siblingText ? firstCodePoint(siblingText) : current;
-}
-function needsPreviousBoundaryRecovery(codePoint) {
-	return codePoint !== null && 56320 <= codePoint && codePoint <= 57343;
-}
-function needsNextBoundaryRecovery(codePoint) {
-	return codePoint !== null && 55296 <= codePoint && codePoint <= 56319;
-}
-function needsPreviousContext(current, previous) {
-	return isNonEmojiGeneralUseVS(classifyCharacter(current)) && (previous === null || previous === codes.eof);
-}
-function getAdjacentSiblingText(node, parent, state, offset) {
-	if (!parent) return;
-	const stackIndex = state.indexStack.at(-1);
-	const siblings = parent.children;
-	const index = typeof stackIndex === "number" ? stackIndex : siblings.indexOf(node);
-	if (index < 0) return;
-	const sibling = siblings[index + offset];
-	return getNodeTextContent(sibling) || void 0;
-}
-function shouldEncodeAfterSupplementaryText(parent, state) {
-	if (!parent) return false;
-	const stackIndex = state.indexStack.at(-1);
-	if (typeof stackIndex !== "number") return false;
-	const sibling = parent.children[stackIndex + 1];
-	return sibling?.type === "text" && (firstCodePoint(sibling.value) ?? 0) > 65535;
-}
-function getCjkFriendlyState(state) {
-	return state;
-}
-function getNodeTextContent(node) {
-	if (!node || typeof node !== "object") return "";
-	if ("value" in node && typeof node.value === "string") return node.value;
-	if ("alt" in node && typeof node.alt === "string") return node.alt;
-	if ("children" in node && Array.isArray(node.children)) return node.children.map(getNodeTextContent).join("");
-	return "";
-}
-function encodeCharacterReference(codePoint) {
-	return `&#x${(codePoint ?? 0).toString(16).toUpperCase()};`;
-}
-function encodeFirstCodePoint(value) {
-	const [first = ""] = [...value];
-	return encodeCharacterReference(first.codePointAt(0) ?? codes.eof) + value.slice(first.length);
-}
-function encodeLastCodePoint(value) {
-	const characters = [...value];
-	const last = characters.pop();
-	return `${characters.join("")}${encodeCharacterReference(last?.codePointAt(0) ?? codes.eof)}`;
-}
-function codePointBeforeLast(value) {
-	const characters = [...value];
-	characters.pop();
-	return characters.at(-1)?.codePointAt(0) ?? codes.eof;
-}
-function firstCodePoint(value) {
-	return value.codePointAt(0) ?? codes.eof;
-}
-function lastCodePoint(value) {
-	return [...value].at(-1)?.codePointAt(0) ?? codes.eof;
-}
-const encodedBoundaryBeforeContext = {
-	current: encodedOutsideBoundary,
-	previous: codes.eof
-};
-const encodedBoundaryAfter = encodedInsideBoundary;
-
 //#endregion
-export { cjkFriendlyToMarkdown };
+export { cjkFriendlyToMarkdown };

package/dist/phrasing-CVe5_g2v.mjs

@@ -0,0 +1,430 @@
+import { defaultHandlers } from "mdast-util-to-markdown";
+import { asciiAlpha, asciiAlphanumeric, asciiPunctuation } from "micromark-util-character";
+import { htmlBlockNames, htmlRawNames } from "micromark-util-html-tag-name";
+import { classifyCharacter, classifyPrecedingCharacter, isCjk, isCjkOrIvs, isNonCjkPunctuation, isNonEmojiGeneralUseVS, isSpaceOrPunctuation, isUnicodeWhitespace } from "micromark-extension-cjk-friendly-util";
+import { classifyCharacter as classifyCharacter$1 } from "micromark-util-classify-character";
+//#region src/encode-info.ts
+function firstCharacter(value) {
+	return value[Symbol.iterator]().next().value ?? "";
+}
+function lastCharacter(value, end = value.length) {
+	if (end === 0) return "";
+	const lastCodeUnit = value.charCodeAt(end - 1);
+	if (lastCodeUnit < 56320 || lastCodeUnit >= 57344 || end === 1 || (value.codePointAt(end - 2) ?? 0) < 65536) return value[end - 1];
+	return value.slice(end - 2, end);
+}
+function classifyBefore(value) {
+	const character = lastCharacter(value);
+	const current = character.codePointAt(0) ?? null;
+	const kind = classifyCharacter(current);
+	return current !== null && isNonEmojiGeneralUseVS(kind) ? classifyPrecedingCharacter(kind, () => lastCharacter(value, value.length - character.length).codePointAt(0) ?? null, current) : kind;
+}
+function sides(before, after, marker) {
+	const beforePunctuation = isNonCjkPunctuation(before);
+	const afterPunctuation = isNonCjkPunctuation(after);
+	const beforeSpaceOrPunctuation = beforePunctuation || isUnicodeWhitespace(before);
+	const afterSpaceOrPunctuation = afterPunctuation || isUnicodeWhitespace(after);
+	const open = !afterSpaceOrPunctuation || afterPunctuation && (beforeSpaceOrPunctuation || isCjkOrIvs(before));
+	const close = !beforeSpaceOrPunctuation || beforePunctuation && (afterSpaceOrPunctuation || isCjk(after));
+	return {
+		open: marker === "_" ? open && (isSpaceOrPunctuation(before) || !close) : open,
+		close: marker === "_" ? close && (isSpaceOrPunctuation(after) || !open) : close
+	};
+}
+function attentionSides(before, after, marker, cjk) {
+	if (cjk) return sides(classifyBefore(before), classifyCharacter(after.codePointAt(0) ?? null), marker);
+	const previous = classifyCharacter$1(lastCharacter(before).codePointAt(0) ?? null);
+	const next = classifyCharacter$1(after.codePointAt(0) ?? null);
+	const open = !next || next === 2 && Boolean(previous);
+	const close = !previous || previous === 2 && Boolean(next);
+	return {
+		open: marker === "_" ? open && (Boolean(previous) || !close) : open,
+		close: marker === "_" ? close && (Boolean(next) || !open) : close
+	};
+}
+function fallback(outside, inside, marker) {
+	if (outside === void 0) return inside === void 0 ? {
+		inside: marker === "_",
+		outside: marker === "_"
+	} : inside === 1 ? {
+		inside: true,
+		outside: true
+	} : {
+		inside: false,
+		outside: true
+	};
+	if (outside === 1) return {
+		inside: inside === 1,
+		outside: inside === 1
+	};
+	return {
+		inside: inside === 1,
+		outside: false
+	};
+}
+/** Choose boundary encodings without changing punctuation used by constructs. */
+function encodeInfo(before, after, marker, opening, cjk) {
+	const beforeKind = classifyBefore(before);
+	const afterKind = classifyCharacter(after.codePointAt(0) ?? null);
+	if (!cjk || !isCjkOrIvs(beforeKind) && !isCjkOrIvs(afterKind)) {
+		const previous = cjk ? commonMarkKind(beforeKind) : classifyCharacter$1(lastCharacter(before).codePointAt(0) ?? null);
+		const next = cjk ? commonMarkKind(afterKind) : classifyCharacter$1(after.codePointAt(0) ?? null);
+		return opening ? fallback(previous, next, marker) : fallback(next, previous, marker);
+	}
+	for (const candidate of [
+		{
+			inside: false,
+			outside: false
+		},
+		{
+			inside: true,
+			outside: false
+		},
+		{
+			inside: false,
+			outside: true
+		},
+		{
+			inside: true,
+			outside: true
+		}
+	]) {
+		const encodedBefore = opening ? candidate.outside : candidate.inside;
+		const encodedAfter = opening ? candidate.inside : candidate.outside;
+		const forms = sides(encodedBefore ? classifyCharacter(";".codePointAt(0) ?? null) : beforeKind, encodedAfter ? classifyCharacter("&".codePointAt(0) ?? null) : afterKind, marker);
+		if (opening ? forms.open : forms.close) return candidate;
+	}
+	return {
+		inside: true,
+		outside: true
+	};
+}
+function commonMarkKind(kind) {
+	return isUnicodeWhitespace(kind) ? 1 : isNonCjkPunctuation(kind) ? 2 : void 0;
+}
+//#endregion
+//#region src/phrasing.ts
+/*!
+* Phrasing algorithm adapted from mdast-util-to-markdown (commit 388140e).
+* Copyright (c) Titus Wormer <tituswormer@gmail.com>
+*
+* Permission is hereby granted, free of charge, to any person obtaining a copy
+* of this software and associated documentation files (the "Software"), to deal
+* in the Software without restriction, including without limitation the rights
+* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+* copies of the Software, and to permit persons to whom the Software is
+* furnished to do so, subject to the following conditions:
+*
+* The above copyright notice and this permission notice shall be included in
+* all copies or substantial portions of the Software.
+*
+* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+* SOFTWARE.
+*/
+/** Mark an attention handler for CJK-aware encoding and pairing. */
+function cjkFriendlyAttention(handle) {
+	const attention = handle.attention;
+	if (!attention) throw new Error("Expected a handler with the attention API");
+	const result = (node, parent, state, info) => {
+		install(state);
+		return handle(node, parent, state, info);
+	};
+	result.peek = handle.peek;
+	result.attention = (node, state) => ({
+		...attention(node, state),
+		cjkFriendly: true
+	});
+	return result;
+}
+function install(state) {
+	state.containerPhrasing = (parent, info) => containerPhrasing(parent, state, info);
+}
+function withCjkPhrasing(handle) {
+	return (node, parent, state, info) => {
+		install(state);
+		return handle(node, parent, state, info);
+	};
+}
+/** Install the container before serializing a document or a phrasing block. */
+const cjkFriendlyPhrasingHandlers = {
+	root: withCjkPhrasing(defaultHandlers.root),
+	paragraph: withCjkPhrasing(defaultHandlers.paragraph),
+	heading: withCjkPhrasing(defaultHandlers.heading)
+};
+function candidates(tokens, involved, runOf) {
+	const related = /* @__PURE__ */ new Set();
+	for (const token of tokens) {
+		const siblings = runOf.get(token);
+		if (siblings && token.attention && involved.some((d) => d.attention === token.attention)) for (const sibling of siblings.tokens) related.add(sibling.attention);
+	}
+	const ordered = [];
+	for (const token of tokens) if (token.attention && related.delete(token.attention)) ordered.push(token.attention);
+	return ordered.reverse();
+}
+function check(tokens, before, after) {
+	const runs = [];
+	const wrong = /* @__PURE__ */ new Map();
+	const runOf = /* @__PURE__ */ new Map();
+	let index = 0;
+	while (index < tokens.length) {
+		if (!tokens[index].attention) {
+			index++;
+			continue;
+		}
+		const marker = tokens[index].value.charAt(0);
+		let end = index + 1;
+		while (end < tokens.length && tokens[end].attention && tokens[end].value.charAt(0) === marker) end++;
+		const group = tokens.slice(index, end);
+		const { open, close } = attentionSides(index ? tokens[index - 1].value : before, end < tokens.length ? tokens[end].value : after, marker, group.some((token) => token.attention?.cjk));
+		const run = {
+			tokens: group,
+			markers: [],
+			start: 0,
+			end: 0,
+			open,
+			close,
+			split: marker === "*" || marker === "_"
+		};
+		for (const token of group) {
+			wrong.set(token, [...group]);
+			runOf.set(token, run);
+			for (let size = token.value.length; size > 0; size--) run.markers.push(token);
+		}
+		run.end = run.markers.length;
+		runs.push(run);
+		index = end;
+	}
+	pair(runs, wrong);
+	for (let index = 0; index < tokens.length; index++) {
+		const involved = wrong.get(tokens[index]);
+		if (involved) return {
+			attention: candidates(tokens, involved, runOf),
+			index
+		};
+	}
+}
+function pair(runs, wrong) {
+	let index = 0;
+	while (index < runs.length) {
+		const closer = runs[index];
+		const marker = closer.tokens[0].value.charAt(0);
+		const closerSize = closer.end - closer.start;
+		let open = closer.close && closerSize ? index : 0;
+		while (open--) {
+			const opener = runs[open];
+			const openerSize = opener.end - opener.start;
+			if (opener.tokens[0].value.charAt(0) === marker && opener.open && openerSize && (closer.split ? !((opener.close || closer.open) && closerSize % 3) || (openerSize + closerSize) % 3 : openerSize === opener.markers.length && openerSize === closerSize)) {
+				const size = closer.split ? openerSize > 1 && closerSize > 1 ? 2 : 1 : closerSize;
+				const opening = opener.markers.slice(opener.end - size, opener.end);
+				const closing = closer.markers.slice(closer.start, closer.start + size);
+				const openToken = opening[0];
+				const closeToken = closing[0];
+				if (openToken === opening[size - 1] && openToken.value.length === size && closeToken === closing[size - 1] && closeToken.value.length === size && openToken.attention === closeToken.attention) {
+					wrong.delete(openToken);
+					wrong.delete(closeToken);
+				} else {
+					const involved = [.../* @__PURE__ */ new Set([...opening, ...closing])];
+					for (const token of involved) wrong.get(token)?.push(...involved);
+				}
+				opener.end -= size;
+				closer.start += size;
+				break;
+			}
+		}
+		if (open === -1 || closer.start === closer.end) index++;
+	}
+}
+function encode(tokens, start) {
+	const token = tokens[start ? 0 : tokens.length - 1];
+	if (!token || token.attention || !token.value) return;
+	const character = start ? firstCharacter(token.value) : lastCharacter(token.value);
+	const reference = `&#x${character.codePointAt(0)?.toString(16).toUpperCase()};`;
+	token.value = start ? reference + token.value.slice(character.length) : token.value.slice(0, -character.length) + reference;
+}
+function attentionSequences(type, markers, sizes) {
+	if (!markers.length || !sizes.length) throw new Error(`Cannot serialize ${type} as attention without markers and sizes`);
+	for (const size of sizes) if (!Number.isSafeInteger(size) || size < 1) throw new Error(`Invalid attention size for ${type}: ${size}`);
+	const sequences = [];
+	for (const marker of markers) {
+		if (marker.length !== 1 || !asciiPunctuation(marker.charCodeAt(0))) throw new Error(`Invalid attention marker for ${type}: ${marker}`);
+		for (const size of sizes) sequences.push(marker.repeat(size));
+	}
+	return sequences;
+}
+function htmlKind(value) {
+	if (value.charCodeAt(0) !== 60) return;
+	const next = value.charCodeAt(1);
+	if (next === 33) {
+		const code = value.charCodeAt(2);
+		if (code === 45) return 2;
+		if (code === 91) return 5;
+		if (asciiAlpha(code)) return 4;
+		return;
+	}
+	if (next === 63) return 3;
+	const closing = next === 47;
+	const start = closing ? 2 : 1;
+	let end = start;
+	let code = value.charCodeAt(end);
+	if (!asciiAlpha(code)) return;
+	while (code === 45 || asciiAlphanumeric(code)) code = value.charCodeAt(++end);
+	const name = value.slice(start, end).toLowerCase();
+	if (!closing && code !== 47 && htmlRawNames.includes(name)) return 1;
+	return htmlBlockNames.includes(name) ? 6 : 7;
+}
+function phrasing(parent, state, info) {
+	const children = parent.children;
+	const results = [];
+	let before = info.before;
+	state.indexStack.push(-1);
+	let tracker = state.createTracker(info);
+	try {
+		for (let index = 0; index < children.length; index++) {
+			const child = children[index];
+			state.indexStack[state.indexStack.length - 1] = index;
+			const next = children[index + 1];
+			let after = info.after;
+			if (next) {
+				const handle = state.handlers[next.type];
+				const peek = handle?.peek ?? handle;
+				after = peek ? firstCharacter(peek(next, parent, state, {
+					before: "",
+					after: "",
+					...tracker.current()
+				})) : "";
+			}
+			const previous = results[results.length - 1];
+			if (typeof previous === "string" && (before === "\r" || before === "\n") && child.type === "html" && htmlKind(child.value) !== 7) {
+				results[results.length - 1] = previous.replace(/(\r?\n|\r)$/, " ");
+				before = " ";
+				tracker = state.createTracker(info);
+				tracker.move(serialize(results, info.before, info.after));
+			}
+			const handle = state.handlers[child.type];
+			if (handle?.attention) {
+				const { construct, markers, sizes, cjkFriendly } = handle.attention(child, state);
+				const sequences = attentionSequences(child.type, markers, sizes);
+				const sequence = sequences[0];
+				const exit = state.enter(construct);
+				tracker.move(sequence);
+				let inside;
+				try {
+					inside = phrasing(child, state, {
+						...tracker.current(),
+						before: sequence,
+						after: sequence
+					});
+					tracker.move(serialize(inside, sequence, sequence));
+					tracker.move(sequence);
+				} finally {
+					exit();
+				}
+				results.push({
+					children: inside,
+					sequences,
+					cjk: Boolean(cjkFriendly)
+				});
+				before = lastCharacter(sequence);
+			} else {
+				const value = state.handle(child, parent, state, {
+					...tracker.current(),
+					after,
+					before
+				});
+				if (!value) continue;
+				tracker.move(value);
+				results.push(value);
+				before = value;
+			}
+		}
+		return results;
+	} finally {
+		state.indexStack.pop();
+	}
+}
+function render(items, chosen, before, after) {
+	const tokens = [];
+	let encodeAfter;
+	let previous = before;
+	for (let index = 0; index < items.length; index++) {
+		const item = items[index];
+		if (typeof item === "string") {
+			const token = {
+				value: item,
+				attention: void 0
+			};
+			if (encodeAfter && encodeAfter === firstCharacter(item)) encode([token], true);
+			encodeAfter = void 0;
+			tokens.push(token);
+			if (token.value) previous = token.value;
+			continue;
+		}
+		const sequence = chosen.get(item) || item.sequences[0];
+		const marker = sequence.charAt(0);
+		const next = items[index + 1];
+		const outsideAfter = next === void 0 ? after : typeof next === "string" ? next : chosen.get(next) || next.sequences[0];
+		const inside = render(item.children, chosen, sequence, sequence);
+		const head = inside[0]?.value ?? "";
+		const open = encodeInfo(previous, head, marker, true, item.cjk);
+		if (open.inside) encode(inside, true);
+		const close = encodeInfo(inside[inside.length - 1]?.value ?? "", outsideAfter, marker, false, item.cjk);
+		if (close.inside) encode(inside, false);
+		if (open.outside && lastCharacter(previous) !== "\n" && lastCharacter(previous) !== "\r") encode(tokens, false);
+		if (close.outside) encodeAfter = firstCharacter(outsideAfter);
+		tokens.push({
+			value: sequence,
+			attention: item
+		}, ...inside, {
+			value: sequence,
+			attention: item
+		});
+		previous = sequence;
+	}
+	return tokens;
+}
+function serialize(items, before, after) {
+	let chosen = /* @__PURE__ */ new Map();
+	const initial = render(items, chosen, before, after);
+	let tokens = initial;
+	let mistake = check(tokens, before, after);
+	while (mistake) {
+		let improvement;
+		for (const attention of mistake.attention) {
+			const current = chosen.get(attention) || attention.sequences[0];
+			for (const sequence of attention.sequences) {
+				if (sequence === current) continue;
+				const trial = new Map(chosen);
+				trial.set(attention, sequence);
+				const rendered = render(items, trial, before, after);
+				const next = check(rendered, before, after);
+				if (!next || next.index > mistake.index) {
+					improvement = {
+						chosen: trial,
+						tokens: rendered,
+						mistake: next
+					};
+					break;
+				}
+			}
+			if (improvement) break;
+		}
+		if (!improvement) {
+			tokens = initial;
+			break;
+		}
+		chosen = improvement.chosen;
+		tokens = improvement.tokens;
+		mistake = improvement.mistake;
+	}
+	return tokens.map((token) => token.value).join("");
+}
+function containerPhrasing(parent, state, info) {
+	return serialize(phrasing(parent, state, info), info.before, info.after);
+}
+//#endregion
+export { cjkFriendlyPhrasingHandlers as n, cjkFriendlyAttention as t };

package/dist/phrasing.d.mts

@@ -0,0 +1,18 @@
+import { Attention, AttentionInfo, Handle, State } from "mdast-util-to-markdown";
+//#region src/phrasing.d.ts
+type AttentionHandle = Handle & {
+  peek?: Handle;
+  attention?: (node: Parameters<Attention>[0], state: State) => AttentionInfo & {
+    cjkFriendly?: boolean;
+  };
+};
+/** Mark an attention handler for CJK-aware encoding and pairing. */
+declare function cjkFriendlyAttention(handle: Handle): AttentionHandle;
+/** Install the container before serializing a document or a phrasing block. */
+declare const cjkFriendlyPhrasingHandlers: {
+  root: Handle;
+  paragraph: Handle;
+  heading: Handle;
+};
+//#endregion
+export { cjkFriendlyAttention, cjkFriendlyPhrasingHandlers };

package/dist/phrasing.mjs

@@ -0,0 +1,2 @@
+import { n as cjkFriendlyPhrasingHandlers, t as cjkFriendlyAttention } from "./phrasing-CVe5_g2v.mjs";
+export { cjkFriendlyAttention, cjkFriendlyPhrasingHandlers };

package/package.json

@@ -1,12 +1,16 @@
 {
   "name": "mdast-util-to-markdown-cjk-friendly",
-  "version": "1.0.0",
+  "version": "1.0.1",
   "type": "module",
   "exports": {
     ".": {
       "types": "./dist/index.d.mts",
       "default": "./dist/index.mjs"
     },
+    "./phrasing": {
+      "types": "./dist/phrasing.d.mts",
+      "default": "./dist/phrasing.mjs"
+    },
     "./package.json": "./package.json"
   },
   "module": "./dist/index.mjs",
@@ -35,9 +39,11 @@
   "description": "mdast-util-to-markdown extension to make Markdown emphasis output more friendly with Chinese, Japanese, and Korean (CJK)",
   "sideEffects": false,
   "dependencies": {
-    "mdast-util-to-markdown": "^2.1.2",
-    "micromark-util-symbol": "^2.0.1",
-    "micromark-extension-cjk-friendly-util": "3.0.1"
+    "mdast-util-to-markdown": "^2.1.3",
+    "micromark-extension-cjk-friendly-util": "3.0.1",
+    "micromark-util-character": "^2.1.1",
+    "micromark-util-classify-character": "^2.0.1",
+    "micromark-util-html-tag-name": "^2.0.1"
   },
   "peerDependencies": {
     "@types/mdast": "*"