buninu 0.1.1

package/apps/jsmdcui/src/cui/heading-ids.mjs

@@ -0,0 +1,1568 @@
+const EMPTY_BUN_HEADING_ID = /^-\d+$/;
+const HTML_ENTITY =
+  /&(?:#(\d+);?|#x([\da-f]+);?|([a-z][\da-z]*)(;?))/gi;
+const HTML_NUMERIC_REPLACEMENTS = new Map([
+  [0x80, 0x20ac],
+  [0x82, 0x201a],
+  [0x83, 0x0192],
+  [0x84, 0x201e],
+  [0x85, 0x2026],
+  [0x86, 0x2020],
+  [0x87, 0x2021],
+  [0x88, 0x02c6],
+  [0x89, 0x2030],
+  [0x8a, 0x0160],
+  [0x8b, 0x2039],
+  [0x8c, 0x0152],
+  [0x8e, 0x017d],
+  [0x91, 0x2018],
+  [0x92, 0x2019],
+  [0x93, 0x201c],
+  [0x94, 0x201d],
+  [0x95, 0x2022],
+  [0x96, 0x2013],
+  [0x97, 0x2014],
+  [0x98, 0x02dc],
+  [0x99, 0x2122],
+  [0x9a, 0x0161],
+  [0x9b, 0x203a],
+  [0x9c, 0x0153],
+  [0x9e, 0x017e],
+  [0x9f, 0x0178],
+]);
+const NAMED_HTML_ENTITIES = new Map([
+  ["amp", "&"],
+  ["apos", "'"],
+  ["gt", ">"],
+  ["lt", "<"],
+  ["nbsp", "\u00a0"],
+  ["quot", '"'],
+]);
+const LEGACY_HTML_ENTITIES = new Set([
+  "AElig", "AMP", "Aacute", "Acirc", "Agrave", "Aring", "Atilde",
+  "Auml", "COPY", "Ccedil", "ETH", "Eacute", "Ecirc", "Egrave",
+  "Euml", "GT", "Iacute", "Icirc", "Igrave", "Iuml", "LT",
+  "Ntilde", "Oacute", "Ocirc", "Ograve", "Oslash", "Otilde",
+  "Ouml", "QUOT", "REG", "THORN", "Uacute", "Ucirc", "Ugrave",
+  "Uuml", "Yacute", "aacute", "acirc", "acute", "aelig", "agrave",
+  "amp", "aring", "atilde", "auml", "brvbar", "ccedil", "cedil",
+  "cent", "copy", "curren", "deg", "divide", "eacute", "ecirc",
+  "egrave", "eth", "euml", "frac12", "frac14", "frac34", "gt",
+  "iacute", "icirc", "iexcl", "igrave", "iquest", "iuml", "laquo",
+  "lt", "macr", "micro", "middot", "nbsp", "not", "ntilde",
+  "oacute", "ocirc", "ograve", "ordf", "ordm", "oslash", "otilde",
+  "ouml", "para", "plusmn", "pound", "quot", "raquo", "reg", "sect",
+  "shy", "sup1", "sup2", "sup3", "szlig", "thorn", "times",
+  "uacute", "ucirc", "ugrave", "uml", "uuml", "yacute", "yen",
+  "yuml",
+]);
+const DECODED_HTML_ENTITIES = new Map();
+
+function decodeNamedHtmlEntity(whole, name) {
+  const known = NAMED_HTML_ENTITIES.get(String(name));
+  if (known != null) return known;
+  if (DECODED_HTML_ENTITIES.has(whole))
+    return DECODED_HTML_ENTITIES.get(whole);
+
+  let decoded = "";
+  try {
+    Bun.markdown.render(
+      whole,
+      {
+        text(value) {
+          decoded += String(value);
+          return "";
+        },
+        paragraph(children) {
+          return children;
+        },
+      },
+    );
+  } catch {
+    decoded = whole;
+  }
+  if (!decoded) decoded = whole;
+  DECODED_HTML_ENTITIES.set(whole, decoded);
+  return decoded;
+}
+
+function decodeHtmlEntities(value, { attribute = false } = {}) {
+  const input = String(value ?? "");
+  return input.replace(
+    HTML_ENTITY,
+    (whole, decimal, hexadecimal, name, semicolon, offset) => {
+      if (decimal != null || hexadecimal != null) {
+        const point = Number.parseInt(
+          decimal ?? hexadecimal,
+          hexadecimal == null ? 10 : 16,
+        );
+        if (
+          !Number.isInteger(point)
+          || point === 0
+          || point > 0x10ffff
+          || (point >= 0xd800 && point <= 0xdfff)
+        ) return "\ufffd";
+        return String.fromCodePoint(
+          HTML_NUMERIC_REPLACEMENTS.get(point) ?? point,
+        );
+      }
+      if (semicolon) return decodeNamedHtmlEntity(whole, name);
+
+      for (let end = name.length; end > 0; end--) {
+        const candidate = name.slice(0, end);
+        if (!LEGACY_HTML_ENTITIES.has(candidate)) continue;
+        const remainder = name.slice(end);
+        const following = remainder[0] ?? input[offset + whole.length] ?? "";
+        if (attribute && /[=A-Za-z0-9]/u.test(following)) return whole;
+        return decodeNamedHtmlEntity(`&${candidate};`, candidate) + remainder;
+      }
+      return whole;
+    },
+  );
+}
+
+export function readLeadingHtmlCharacterReference(value) {
+  const input = String(value ?? "");
+  if (!input.startsWith("&")) return null;
+
+  const numeric = input.match(/^&#(?:[xX][\da-fA-F]+|\d+);?/u);
+  if (numeric) {
+    return {
+      source: numeric[0],
+      decoded: decodeHtmlEntities(numeric[0]),
+    };
+  }
+
+  const terminated = input.match(/^&([A-Za-z][A-Za-z0-9]*);/u);
+  if (terminated) {
+    const decoded = decodeHtmlEntities(terminated[0]);
+    if (decoded !== terminated[0])
+      return { source: terminated[0], decoded };
+  }
+
+  const unterminated = input.match(/^&([A-Za-z][A-Za-z0-9]*)/u);
+  const name = unterminated?.[1] ?? "";
+  for (let end = name.length; end > 0; end--) {
+    const candidate = name.slice(0, end);
+    if (!LEGACY_HTML_ENTITIES.has(candidate)) continue;
+    const source = `&${candidate}`;
+    return {
+      source,
+      decoded: decodeNamedHtmlEntity(
+        `&${candidate};`,
+        candidate,
+      ),
+    };
+  }
+  return null;
+}
+
+function htmlTokenAt(input, start) {
+  if (input[start] !== "<") return null;
+  if (input.startsWith("<!--", start)) {
+    const close = input.indexOf("-->", start + 4);
+    return {
+      start,
+      end: close < 0 ? input.length : close + 3,
+      kind: "comment",
+      name: null,
+      closing: false,
+      selfClosing: false,
+      source: input.slice(start, close < 0 ? input.length : close + 3),
+    };
+  }
+  if (input.startsWith("<![CDATA[", start)) {
+    const close = input.indexOf("]]>", start + 9);
+    const end = close < 0 ? input.length : close + 3;
+    return {
+      start,
+      end,
+      kind: "comment",
+      name: null,
+      closing: false,
+      selfClosing: false,
+      source: input.slice(start, end),
+    };
+  }
+  if (input.startsWith("<?", start)) {
+    const close = input.indexOf("?>", start + 2);
+    const end = close < 0 ? input.length : close + 2;
+    return {
+      start,
+      end,
+      kind: "comment",
+      name: null,
+      closing: false,
+      selfClosing: false,
+      source: input.slice(start, end),
+    };
+  }
+
+  const match = input.slice(start).match(
+    /^<\s*(\/?)\s*([A-Za-z][\w:-]*)(?=[\t\n\f\r />])/,
+  );
+  if (!match) {
+    const close = input.indexOf(">", start + 1);
+    if (close < 0) return null;
+    return {
+      start,
+      end: close + 1,
+      kind: "other",
+      name: null,
+      closing: false,
+      selfClosing: false,
+      source: input.slice(start, close + 1),
+    };
+  }
+
+  const isWhitespace = character =>
+    character === "\t"
+    || character === "\n"
+    || character === "\f"
+    || character === "\r"
+    || character === " ";
+  let index = start + match[0].length;
+  let state = "before-attribute";
+  let quote = null;
+  for (; index < input.length; index++) {
+    const character = input[index];
+    if (state === "quoted-value") {
+      if (character === quote) {
+        quote = null;
+        state = "before-attribute";
+      }
+      continue;
+    }
+    if (character === "<") break;
+    if (character === ">") {
+      index++;
+      break;
+    }
+
+    if (state === "before-attribute") {
+      if (isWhitespace(character) || character === "/") continue;
+      state = "attribute-name";
+      continue;
+    }
+    if (state === "attribute-name") {
+      if (character === "=") state = "before-value";
+      else if (isWhitespace(character)) state = "after-attribute-name";
+      continue;
+    }
+    if (state === "after-attribute-name") {
+      if (isWhitespace(character)) continue;
+      if (character === "=") state = "before-value";
+      else if (character === "/") state = "before-attribute";
+      else state = "attribute-name";
+      continue;
+    }
+    if (state === "before-value") {
+      if (isWhitespace(character)) continue;
+      if (character === '"' || character === "'") {
+        quote = character;
+        state = "quoted-value";
+      } else {
+        state = "unquoted-value";
+      }
+      continue;
+    }
+    if (state === "unquoted-value" && isWhitespace(character))
+      state = "before-attribute";
+  }
+  if (index >= input.length && input.at(-1) !== ">") return null;
+
+  const source = input.slice(start, index);
+  return {
+    start,
+    end: index,
+    kind: "tag",
+    name: match[2].toLowerCase(),
+    closing: Boolean(match[1]),
+    selfClosing: !match[1] && /\/[\t\n\f\r ]*>$/.test(source),
+    source,
+  };
+}
+
+function nextHtmlToken(input, from) {
+  let start = input.indexOf("<", from);
+  while (start >= 0) {
+    const token = htmlTokenAt(input, start);
+    if (token) return token;
+    start = input.indexOf("<", start + 1);
+  }
+  return null;
+}
+
+const HEADING_OPAQUE_TAGS = new Set([
+  "iframe",
+  "noembed",
+  "noframes",
+  "noscript",
+  "plaintext",
+  "script",
+  "style",
+  "template",
+  "textarea",
+  "title",
+  "xmp",
+]);
+const HEADING_RAW_TEXT_TAGS = new Set([
+  "iframe",
+  "noembed",
+  "noframes",
+  "noscript",
+  "script",
+  "style",
+  "textarea",
+  "title",
+  "xmp",
+]);
+
+function matchingHtmlClose(input, opening) {
+  if (
+    opening?.kind !== "tag"
+    || opening.closing
+    || opening.selfClosing
+  ) return null;
+
+  let depth = 1;
+  let scan = opening.end;
+  let token;
+  while ((token = nextHtmlToken(input, scan))) {
+    scan = token.end;
+    if (
+      token.kind === "tag"
+      && !token.closing
+      && !token.selfClosing
+      && token.name !== opening.name
+      && HEADING_OPAQUE_TAGS.has(token.name)
+    ) {
+      const closing = closingOpaqueTag(input, token);
+      scan = closing?.end ?? input.length;
+      continue;
+    }
+    if (token.kind !== "tag" || token.name !== opening.name) continue;
+    if (token.closing) {
+      depth--;
+      if (depth === 0) return token;
+    } else if (!token.selfClosing) {
+      depth++;
+    }
+  }
+  return null;
+}
+
+function closingOpaqueTag(input, opening) {
+  if (opening?.name === "plaintext") return null;
+  if (!HEADING_RAW_TEXT_TAGS.has(opening?.name))
+    return matchingHtmlClose(input, opening);
+
+  const closingPattern = new RegExp(
+    `</${opening.name}(?=[\\s>])`,
+    "ig",
+  );
+  closingPattern.lastIndex = opening.end;
+  let match;
+  while ((match = closingPattern.exec(input)) !== null) {
+    const token = htmlTokenAt(input, match.index);
+    if (token?.kind === "tag" && token.closing)
+      return token;
+    closingPattern.lastIndex = match.index + 2;
+  }
+  return null;
+}
+
+function htmlAttribute(source, expectedName) {
+  const input = String(source ?? "");
+  const opening = input.match(/^<\s*\/?\s*[^\s/>]+/);
+  if (!opening) return undefined;
+  const wanted = String(expectedName).toLowerCase();
+  let index = opening[0].length;
+
+  while (index < input.length) {
+    while (/\s/u.test(input[index] ?? "")) index++;
+    if (input[index] === ">" || input[index] === "/" || index >= input.length)
+      return undefined;
+
+    const nameStart = index;
+    while (index < input.length && !/[\s=/>]/u.test(input[index])) index++;
+    const name = input.slice(nameStart, index).toLowerCase();
+    while (/\s/u.test(input[index] ?? "")) index++;
+    if (input[index] !== "=") {
+      if (name === wanted) return "";
+      continue;
+    }
+
+    index++;
+    while (/\s/u.test(input[index] ?? "")) index++;
+    const quote = input[index] === '"' || input[index] === "'"
+      ? input[index++]
+      : null;
+    const valueStart = index;
+    if (quote) {
+      while (index < input.length && input[index] !== quote) index++;
+    } else {
+      while (index < input.length && !/[\s>]/u.test(input[index])) index++;
+    }
+    const value = input.slice(valueStart, index);
+    if (quote && input[index] === quote) index++;
+    if (name === wanted)
+      return decodeHtmlEntities(value, { attribute: true });
+  }
+  return undefined;
+}
+
+function visibleHeadingText(value) {
+  const input = String(value ?? "");
+  let output = "";
+  let cursor = 0;
+  for (let index = 0; index < input.length;) {
+    const start = input.indexOf("<", index);
+    if (start < 0) break;
+    const token = htmlTokenAt(input, start);
+    if (!token) {
+      index = start + 1;
+      continue;
+    }
+    output += decodeHtmlEntities(input.slice(cursor, start));
+    if (!token.closing) {
+      if (token.name === "br") output += " ";
+      else if (token.name === "img")
+        output += htmlAttribute(token.source, "alt") ?? "";
+      else if (
+        !token.selfClosing
+        && HEADING_OPAQUE_TAGS.has(token.name)
+      ) {
+        const closing = closingOpaqueTag(input, token);
+        cursor = closing?.end ?? input.length;
+        index = cursor;
+        continue;
+      }
+    }
+    cursor = token.end;
+    index = token.end;
+  }
+  output += decodeHtmlEntities(input.slice(cursor));
+  return output;
+}
+
+function comparableHeadingText(value) {
+  return String(value ?? "")
+    .normalize("NFKC")
+    .replace(/\s+/gu, " ")
+    .trim();
+}
+
+function stableHeadingHash(value) {
+  const bytes = new TextEncoder().encode(
+    comparableHeadingText(value).toLowerCase(),
+  );
+  let hash = 0xcbf29ce484222325n;
+  for (const byte of bytes) {
+    hash ^= BigInt(byte);
+    hash = BigInt.asUintN(64, hash * 0x100000001b3n);
+  }
+  return hash.toString(16).padStart(16, "0");
+}
+
+export function normalizeUnicodeHeadingId(value) {
+  const normalized = comparableHeadingText(value).toLowerCase();
+  const tokens = normalized.match(
+    /[_\p{L}\p{N}][_\p{L}\p{M}\p{N}]*/gu,
+  );
+  return tokens?.join("-") || `mdcui-h-${stableHeadingHash(normalized)}`;
+}
+
+const TEXT_TOKEN_PATTERN = /\ue000([0-9a-z]+)\ue001/giu;
+const HEADING_STRUCTURE_PATTERN = /\ue100h([0-9a-z]+)\ue101/giu;
+
+function escapeHtmlText(value) {
+  return String(value)
+    .replaceAll("&", "&amp;")
+    .replaceAll("<", "&lt;")
+    .replaceAll(">", "&gt;");
+}
+
+function markdownHeadingEvents(markdown) {
+  const source = String(markdown ?? "");
+  const textValues = [];
+  const records = [];
+  const headings = [];
+  let containerOrdinal = 0;
+
+  const storedTextToken = (
+    value,
+    literal = false,
+    explicitSlugValue,
+  ) => {
+    const ordinal = textValues.length;
+    const text = String(value ?? "");
+    const token = literal ? null : htmlTokenAt(text, 0);
+    textValues.push({
+      value: text,
+      literal,
+      slugValue: explicitSlugValue ?? (
+        token && token.end === text.length ? "" : text
+      ),
+    });
+    return `\ue000${ordinal.toString(36)}\ue001`;
+  };
+  const textToken = value => storedTextToken(value);
+  const expandTextTokens = (children) =>
+    String(children ?? "").replace(
+      TEXT_TOKEN_PATTERN,
+      (whole, encoded) =>
+        textValues[Number.parseInt(encoded, 36)]?.value ?? whole,
+    );
+  const expandSlugTokens = (children) =>
+    String(children ?? "").replace(
+      TEXT_TOKEN_PATTERN,
+      (whole, encoded) =>
+        textValues[Number.parseInt(encoded, 36)]?.slugValue ?? whole,
+    ).replace(/\r\n?|\n/gu, " ");
+  const inlineHtmlFromTokens = (children) => {
+    const input = String(children ?? "");
+    let output = "";
+    let cursor = 0;
+    TEXT_TOKEN_PATTERN.lastIndex = 0;
+    let match;
+    while ((match = TEXT_TOKEN_PATTERN.exec(input)) !== null) {
+      output += escapeHtmlText(input.slice(cursor, match.index));
+      const stored = textValues[Number.parseInt(match[1], 36)];
+      const value = stored?.value ?? match[0];
+      const token = htmlTokenAt(value, 0);
+      output += !stored?.literal && token && token.end === value.length
+        ? value
+        : escapeHtmlText(value);
+      cursor = match.index + match[0].length;
+    }
+    return output + escapeHtmlText(input.slice(cursor));
+  };
+  const recordInlineHtml = (children) => {
+    const html = inlineHtmlFromTokens(children);
+    if (html.includes("<")) records.push({ type: "raw", html });
+    return "";
+  };
+  const headingStructure = (children) => {
+    const input = String(children ?? "");
+    HEADING_STRUCTURE_PATTERN.lastIndex = 0;
+    return Array.from(input.matchAll(HEADING_STRUCTURE_PATTERN), match =>
+      match[0]
+    ).join("");
+  };
+  const markHeadingContainer = (children, type) => {
+    const structure = headingStructure(children);
+    if (!structure) return "";
+    const container = `${type}:${containerOrdinal++}`;
+    HEADING_STRUCTURE_PATTERN.lastIndex = 0;
+    for (const match of structure.matchAll(HEADING_STRUCTURE_PATTERN)) {
+      const heading = headings[Number.parseInt(match[1], 36)];
+      if (heading) heading.containerPath.unshift(container);
+    }
+    return structure;
+  };
+  const passthrough = (children) => children;
+
+  Bun.markdown.render(
+    source,
+    {
+      text: textToken,
+      strong: passthrough,
+      emphasis: passthrough,
+      link: passthrough,
+      image(children) {
+        const slugValue = expandSlugTokens(children);
+        return storedTextToken(
+          visibleHeadingText(inlineHtmlFromTokens(children)),
+          true,
+          slugValue,
+        );
+      },
+      codespan(children) {
+        const value = expandTextTokens(children);
+        return storedTextToken(
+          value,
+          true,
+          value.replace(/\r\n?|\n/gu, ""),
+        );
+      },
+      strikethrough: passthrough,
+      heading(children, meta) {
+        const inlineHtml = inlineHtmlFromTokens(children);
+        const text = visibleHeadingText(inlineHtml);
+        const bunId = String(meta?.id ?? "");
+        const ordinal = headings.length;
+        const heading = {
+          bunId,
+          containerPath: [],
+          fallback: bunId === "" || EMPTY_BUN_HEADING_ID.test(bunId),
+          inlineHtml,
+          level: Number(meta?.level),
+          ordinal,
+          slugText: expandSlugTokens(children),
+          text,
+        };
+        headings.push(heading);
+        records.push({ type: "markdown", heading });
+        return `\ue100h${ordinal.toString(36)}\ue101`;
+      },
+      html(children) {
+        records.push({
+          type: "raw",
+          html: expandTextTokens(children),
+        });
+        return "";
+      },
+      paragraph: recordInlineHtml,
+      listItem(children) {
+        recordInlineHtml(children);
+        return markHeadingContainer(children, "list-item");
+      },
+      th: recordInlineHtml,
+      td: recordInlineHtml,
+      code() {
+        return "";
+      },
+      blockquote(children) {
+        return markHeadingContainer(children, "blockquote");
+      },
+      list(children) {
+        return markHeadingContainer(children, "list");
+      },
+      hr() {
+        return "";
+      },
+      table() {
+        return "";
+      },
+      thead() {
+        return "";
+      },
+      tbody() {
+        return "";
+      },
+      tr() {
+        return "";
+      },
+    },
+    { headings: { ids: true } },
+  );
+
+  const events = [];
+  const rawHeadings = [];
+  for (const record of records) {
+    if (record.type === "markdown") {
+      events.push(record);
+      continue;
+    }
+    for (const heading of htmlHeadings(record.html)) {
+      const rawHeading = { ...heading, blockSource: record.html };
+      rawHeadings.push(rawHeading);
+      events.push({ type: "raw", heading: rawHeading });
+    }
+  }
+  return { events, headings, rawHeadings };
+}
+
+function htmlHeadings(html) {
+  const input = String(html ?? "");
+  const headings = [];
+  let scan = 0;
+  let opening;
+  while ((opening = nextHtmlToken(input, scan))) {
+    scan = opening.end;
+    if (
+      opening.kind !== "tag"
+      || opening.closing
+      || opening.selfClosing
+    ) continue;
+    if (HEADING_OPAQUE_TAGS.has(opening.name)) {
+      const closing = closingOpaqueTag(input, opening);
+      scan = closing?.end ?? input.length;
+      continue;
+    }
+    const headingMatch = opening.name?.match(/^h([1-6])$/);
+    if (!headingMatch) continue;
+
+    const closing = matchingHtmlClose(input, opening);
+    if (!closing) continue;
+    const innerHTML = input.slice(opening.end, closing.start);
+    headings.push({
+      start: opening.start,
+      end: closing.end,
+      source: input.slice(opening.start, closing.end),
+      openingSource: opening.source,
+      level: Number(headingMatch[1]),
+      innerHTML,
+      id: htmlAttribute(opening.source, "id"),
+      text: visibleHeadingText(innerHTML),
+    });
+    scan = closing.end;
+  }
+  return headings;
+}
+
+function legacyHtmlHeadings(html) {
+  return htmlHeadings(html).map((heading, ordinal) => ({
+    id: heading.id ?? null,
+    html: heading.innerHTML,
+    text: heading.text,
+    level: heading.level,
+    ordinal,
+  }));
+}
+
+function matchedMarkdownHeadings(
+  markdownHeadings,
+  renderedHeadings,
+  events,
+) {
+  if (events.length !== renderedHeadings.length) return null;
+
+  const matched = Array(markdownHeadings.length);
+  let markdownOrdinal = 0;
+  for (let index = 0; index < events.length; index++) {
+    const event = events[index];
+    const rendered = renderedHeadings[index];
+    if (event.type === "raw") {
+      const raw = event.heading;
+      if (
+        !raw
+        || rendered.level !== raw.level
+        || rendered.id !== raw.id
+        || comparableHeadingText(rendered.text)
+          !== comparableHeadingText(raw.text)
+      ) return null;
+      continue;
+    }
+
+    const heading = event.heading;
+    const ordinal = markdownOrdinal++;
+    if (
+      !heading
+      || heading !== markdownHeadings[ordinal]
+      || rendered.level !== heading.level
+      || rendered.id !== heading.bunId
+      || comparableHeadingText(rendered.text)
+        !== comparableHeadingText(heading.text)
+    ) return null;
+    matched[ordinal] = { ...heading, rendered };
+  }
+  return matched.every(Boolean) ? matched : null;
+}
+
+function escapeHtmlAttribute(value, quote) {
+  let output = String(value)
+    .replaceAll("&", "&amp;")
+    .replaceAll("<", "&lt;")
+    .replaceAll(">", "&gt;");
+  output = quote === "'"
+    ? output.replaceAll("'", "&#x27;")
+    : output.replaceAll('"', "&quot;");
+  return output;
+}
+
+function replaceHeadingId(source, id) {
+  const input = String(source);
+  const opening = htmlTokenAt(input, 0);
+  if (opening?.kind !== "tag" || opening.closing) return input;
+  const replacement = opening.source.replace(
+    /(\bid\s*=\s*)(["'])(.*?)\2/i,
+    (whole, prefix, quote) =>
+      `${prefix}${quote}${escapeHtmlAttribute(id, quote)}${quote}`,
+  );
+  return replacement + input.slice(opening.end);
+}
+
+function sourceLineMarkerName(source) {
+  let marker = "mdcui-heading-source-line";
+  let suffix = 0;
+  while (source.toLowerCase().includes(`<${marker}`))
+    marker = `mdcui-heading-source-line-${++suffix}`;
+  return marker;
+}
+
+function sourceLineContentOffset(line) {
+  const input = String(line ?? "");
+  let index = 0;
+  const skipWhitespace = () => {
+    while (input[index] === " " || input[index] === "\t") index++;
+  };
+  skipWhitespace();
+
+  while (index < input.length) {
+    if (input[index] === ">") {
+      index++;
+      skipWhitespace();
+      continue;
+    }
+    const listMarker = input.slice(index).match(
+      /^(?:[-+*]|\d{1,9}[.)])(?=[ \t])/u,
+    );
+    if (!listMarker) break;
+    index += listMarker[0].length;
+    skipWhitespace();
+  }
+  return index;
+}
+
+function sourceLineHasAtxCandidate(line) {
+  const input = String(line ?? "");
+  return /^#{1,6}(?=[ \t]|$)/u.test(
+    input.slice(sourceLineContentOffset(input)),
+  );
+}
+
+function sourceLineHasSetextCandidate(line) {
+  const input = String(line ?? "");
+  const content = input.slice(sourceLineContentOffset(input)).trim();
+  return /^(?:=+|-+)$/u.test(content);
+}
+
+const REFERENCE_DEFINITION_START =
+  /^\[(?:\\[^\r\n]|[^\[\]\\]){1,999}\]:/u;
+
+function setextHeadingStart(lines, contentIndex) {
+  let start = contentIndex;
+  while (start > 0) {
+    const previous = String(lines[start - 1] ?? "");
+    const content = previous.slice(
+      sourceLineContentOffset(previous),
+    ).trim();
+    if (!content) break;
+    if (sourceLineHasAtxCandidate(previous)) break;
+    if (/^(?:`{3,}|~{3,})/u.test(content)) break;
+    if (REFERENCE_DEFINITION_START.test(content)) break;
+    if (/^<\/?[A-Za-z][^>]*>$/u.test(content)) break;
+    if (sourceLineHasSetextCandidate(previous)) break;
+    start--;
+  }
+  return start;
+}
+
+const SETEXT_HTML_BLOCK_START = new RegExp(
+  "^(?:"
+    + "<!--|<\\?|<!\\[CDATA\\[|<![A-Z]"
+    + "|</?(?:script|pre|style|textarea)(?=[\\t\\n\\f\\r />])"
+    + "|</?(?:address|article|aside|base|basefont|blockquote|body|caption"
+    + "|center|col|colgroup|dd|details|dialog|dir|div|dl|dt|fieldset"
+    + "|figcaption|figure|footer|form|frame|frameset|h[1-6]|head|header"
+    + "|hr|html|iframe|legend|li|link|main|menu|menuitem|nav|noframes"
+    + "|ol|optgroup|option|p|param|search|section|summary|table|tbody"
+    + "|td|tfoot|th|thead|title|tr|track|ul)(?=[\\t\\n\\f\\r />])"
+    + ")",
+  "iu",
+);
+
+function setextMarkerDelimiters(source) {
+  let open = "\ue200MDCUI_SETEXT";
+  let close = "\ue201";
+  while (source.includes(open)) open += "X";
+  while (source.includes(close)) close += "\ue201";
+  return { open, close };
+}
+
+function sourceLineBelongsToReferenceDefinition(
+  lines,
+  lineIndex,
+  possibleStart,
+) {
+  if (!Number.isInteger(possibleStart)) return false;
+  let hasContent = false;
+  Bun.markdown.render(
+    lines.slice(possibleStart, lineIndex + 1).join("\n"),
+    {
+      text() {
+        hasContent = true;
+        return "";
+      },
+      html() {
+        hasContent = true;
+        return "";
+      },
+      code() {
+        hasContent = true;
+        return "";
+      },
+      image() {
+        hasContent = true;
+        return "";
+      },
+      heading() {
+        hasContent = true;
+        return "";
+      },
+      hr() {
+        hasContent = true;
+        return "";
+      },
+    },
+  );
+  return !hasContent;
+}
+
+function sameMarkdownHeadingContent(left, right) {
+  return (
+    left?.level === right?.level
+    && left?.inlineHtml === right?.inlineHtml
+    && left?.slugText === right?.slugText
+    && left?.text === right?.text
+  );
+}
+
+function locateSetextHeadingStart(lines, underline, expected) {
+  const content = underline - 1;
+  if (content < 0) return 0;
+  const probes = new Map();
+  const matches = (start) => {
+    if (!probes.has(start)) {
+      const heading = markdownHeadingEvents(
+        lines.slice(start, underline + 1).join("\n"),
+      ).headings.at(-1);
+      probes.set(start, sameMarkdownHeadingContent(heading, expected));
+    }
+    return probes.get(start);
+  };
+
+  if (matches(content)) return content;
+
+  let higher = content;
+  let lower = null;
+  for (let distance = 1; lower == null;) {
+    const candidate = Math.max(0, content - distance);
+    if (matches(candidate)) lower = candidate;
+    else higher = candidate;
+    if (candidate === 0) break;
+    distance *= 2;
+  }
+  if (lower == null)
+    return setextHeadingStart(lines, content);
+
+  while (higher - lower > 1) {
+    const middle = Math.floor((lower + higher) / 2);
+    if (matches(middle)) lower = middle;
+    else higher = middle;
+  }
+  return lower;
+}
+
+function baseMarkdownHeadingId(heading) {
+  if (!heading) return "";
+  if (heading.fallback) return normalizeUnicodeHeadingId(heading.text);
+
+  const encoded = Array.from(heading.slugText ?? "", character =>
+    `&#x${character.codePointAt(0).toString(16)};`
+  ).join("");
+  let bunId = "";
+  Bun.markdown.render(
+    `# ${encoded}\n`,
+    {
+      heading(children, meta) {
+        bunId = String(meta?.id ?? "");
+        return "";
+      },
+    },
+    { headings: { ids: true } },
+  );
+  return bunId || heading.bunId;
+}
+
+export function collectMarkdownHeadingDeclarations(
+  markdown,
+  { includeLevel = false, includeEndLine = false } = {},
+) {
+  const source = String(markdown ?? "");
+  const lines = source.split(/\r\n?|\n/);
+  const marker = sourceLineMarkerName(source);
+  const setextMarker = setextMarkerDelimiters(source);
+  const atxCandidates = new Map();
+  const setextCandidates = [];
+  const possibleReferenceStarts = [];
+  let possibleReferenceStart = null;
+
+  for (let index = 0; index < lines.length; index++) {
+    const content = lines[index].slice(
+      sourceLineContentOffset(lines[index]),
+    );
+    if (!content.trim()) possibleReferenceStart = null;
+    if (REFERENCE_DEFINITION_START.test(content))
+      possibleReferenceStart = index;
+    possibleReferenceStarts[index] = possibleReferenceStart;
+
+    if (sourceLineHasAtxCandidate(lines[index])) {
+      atxCandidates.set(index, {
+        line: index + 1,
+        source: lines[index].trim(),
+      });
+    }
+    if (
+      index > 0
+      && lines[index - 1].trim()
+      && sourceLineHasSetextCandidate(lines[index])
+    ) {
+      const start = setextHeadingStart(lines, index - 1);
+      setextCandidates.push({
+        ordinal: setextCandidates.length,
+        start,
+        content: index - 1,
+        underline: index,
+      });
+    }
+    if (
+      sourceLineHasAtxCandidate(lines[index])
+      || sourceLineHasSetextCandidate(lines[index])
+      || /^(?:`{3,}|~{3,})/u.test(content)
+      || SETEXT_HTML_BLOCK_START.test(content)
+    ) possibleReferenceStart = null;
+  }
+
+  const markerPattern = new RegExp(
+    `<${marker} data-mdcui-line="(\\d+)"></${marker}>`,
+    "g",
+  );
+  const setextMarkerPattern = new RegExp(
+    `${setextMarker.open}([0-9a-z]+)${setextMarker.close}[ ]?`,
+    "giu",
+  );
+
+  const markedLines = lines.slice();
+  for (const [index] of atxCandidates) {
+    const markerHtml =
+      `<${marker} data-mdcui-line="${index}"></${marker}>`;
+    const closingHashes =
+      markedLines[index].match(/[ \t]+#+[ \t]*$/u);
+    if (closingHashes?.index != null) {
+      markedLines[index] =
+        markedLines[index].slice(0, closingHashes.index)
+        + ` ${markerHtml}`
+        + markedLines[index].slice(closingHashes.index);
+    } else {
+      markedLines[index] += ` ${markerHtml}`;
+    }
+  }
+
+  const markedSetextLines = new Set();
+  const setextByMarker = new Map();
+  for (const candidate of setextCandidates) {
+    const content = lines[candidate.content].slice(
+      sourceLineContentOffset(lines[candidate.content]),
+    );
+    if (
+      markedSetextLines.has(candidate.content)
+      || sourceLineHasAtxCandidate(lines[candidate.content])
+      || sourceLineHasSetextCandidate(lines[candidate.content])
+      || /^(?:`{3,}|~{3,})/u.test(content)
+      || sourceLineBelongsToReferenceDefinition(
+        lines,
+        candidate.content,
+        possibleReferenceStarts[candidate.content],
+      )
+      || SETEXT_HTML_BLOCK_START.test(content)
+    ) continue;
+
+    const markerOrdinal = setextByMarker.size;
+    const token =
+      `${setextMarker.open}${markerOrdinal.toString(36)}`
+      + setextMarker.close;
+    const offset = sourceLineContentOffset(markedLines[candidate.content]);
+    markedLines[candidate.content] =
+      markedLines[candidate.content].slice(0, offset)
+      + `${token} `
+      + markedLines[candidate.content].slice(offset);
+    markedSetextLines.add(candidate.content);
+    setextByMarker.set(markerOrdinal, {
+      setextUnderline: candidate.underline,
+    });
+  }
+
+  const discovered = [];
+  const markedHeadings = markdownHeadingEvents(
+    markedLines.join("\n"),
+  ).headings;
+  for (const [headingOrdinal, heading] of markedHeadings.entries()) {
+    markerPattern.lastIndex = 0;
+    setextMarkerPattern.lastIndex = 0;
+    const atxMatch = markerPattern.exec(heading.inlineHtml);
+    const setextMatch = setextMarkerPattern.exec(heading.inlineHtml);
+    const candidate = atxMatch
+      ? atxCandidates.get(Number(atxMatch[1]))
+      : setextByMarker.get(
+        Number.parseInt(setextMatch?.[1] ?? "", 36),
+      );
+    if (!candidate) continue;
+    markerPattern.lastIndex = 0;
+    setextMarkerPattern.lastIndex = 0;
+    discovered.push({
+      ...candidate,
+      headingOrdinal,
+      level: heading.level,
+      text: visibleHeadingText(
+        heading.inlineHtml
+          .replace(markerPattern, "")
+          .replace(setextMarkerPattern, ""),
+      ),
+    });
+  }
+  discovered.sort((left, right) => left.line - right.line);
+
+  const actual = markdownHeadingEvents(source).headings;
+  const discoveredByOrdinal = new Map(
+    discovered.map(item => [item.headingOrdinal, item]),
+  );
+  const aligned = [];
+  for (const heading of actual) {
+    const found = discoveredByOrdinal.get(heading.ordinal);
+    if (!found || found.level !== heading.level) {
+      aligned.push({
+        line: 1,
+        source: heading.text || "(empty heading)",
+        heading,
+      });
+    } else if (Number.isInteger(found.setextUnderline)) {
+      const start = locateSetextHeadingStart(
+        lines,
+        found.setextUnderline,
+        heading,
+      );
+      aligned.push({
+        ...found,
+        line: start + 1,
+        source:
+          `${lines.slice(start, found.setextUnderline)
+            .map(line => line.trim()).join(" ")} / `
+          + lines[found.setextUnderline].trim(),
+        heading,
+      });
+    } else {
+      aligned.push({ ...found, heading });
+    }
+  }
+
+  return aligned.map((heading) => ({
+    id: baseMarkdownHeadingId(heading.heading),
+    kind: "heading",
+    ...(includeLevel ? { level: heading.heading.level } : {}),
+    line: heading.line,
+    ...(includeEndLine
+      ? {
+        endLine: Number.isInteger(heading.setextUnderline)
+          ? heading.setextUnderline + 1
+          : heading.line,
+      }
+      : {}),
+    source: heading.source,
+  })).filter(heading => heading.id);
+}
+
+function markRawHeadingCandidates(markdown) {
+  const source = String(markdown ?? "");
+  let marker = "data-mdcui-raw-heading-source";
+  let suffix = 0;
+  while (source.toLowerCase().includes(marker))
+    marker = `data-mdcui-raw-heading-source-${++suffix}`;
+
+  const lineStarts = [0];
+  const lineEnds = [];
+  const lineBreakPattern = /\r\n?|\n/g;
+  let lineBreak;
+  while ((lineBreak = lineBreakPattern.exec(source)) !== null) {
+    lineEnds.push(lineBreak.index);
+    lineStarts.push(lineBreak.index + lineBreak[0].length);
+  }
+  lineEnds.push(source.length);
+
+  const candidates = [];
+  let lineIndex = 0;
+  for (let index = 0; index < source.length;) {
+    const openingName = source.slice(index).match(
+      /^<([A-Za-z][\w:-]*)([\t\n\f\r />])/u,
+    );
+    if (
+      !openingName
+      || !/^h[1-6]$/iu.test(openingName[1])
+    ) {
+      index++;
+      continue;
+    }
+
+    const boundary = openingName[2];
+    const afterBoundary = index + openingName[0].length;
+    const selfClosing = boundary === "/"
+      && /^[\t\n\f\r ]*>/u.test(source.slice(afterBoundary));
+    const insertAt = afterBoundary
+      - (boundary === ">" || selfClosing ? 1 : 0);
+    while (
+      lineIndex + 1 < lineStarts.length
+      && lineStarts[lineIndex + 1] <= index
+    ) lineIndex++;
+    candidates.push({
+      index,
+      insertAt,
+      line: lineIndex + 1,
+      source: source.slice(
+        lineStarts[lineIndex],
+        lineEnds[lineIndex],
+      ).trim(),
+    });
+    index = insertAt;
+  }
+
+  const pieces = [];
+  let cursor = 0;
+  for (let ordinal = 0; ordinal < candidates.length; ordinal++) {
+    const candidate = candidates[ordinal];
+    pieces.push(
+      source.slice(cursor, candidate.insertAt),
+      ` ${marker}=${ordinal} `,
+    );
+    cursor = candidate.insertAt;
+  }
+  pieces.push(source.slice(cursor));
+  return {
+    candidates,
+    marker,
+    source: pieces.join(""),
+  };
+}
+
+function transformHtmlSynchronously(html, configure) {
+  let rewriter = new HTMLRewriter();
+  rewriter = configure(rewriter);
+  const promise = rewriter.transform(new Response(String(html))).text();
+  const result = Bun.peek(promise);
+  if (typeof result !== "string")
+    throw new Error("HTMLRewriter did not complete synchronously");
+  return result;
+}
+
+function browserHeadingRecords(html, rawMarker) {
+  const records = [];
+  let templateDepth = 0;
+  const templateHandler = {
+    element(element) {
+      if (
+        element.namespaceURI
+          !== "http://www.w3.org/1999/xhtml"
+      ) return;
+      templateDepth++;
+      element.onEndTag(() => {
+        templateDepth = Math.max(0, templateDepth - 1);
+      });
+    },
+  };
+  const headingHandler = {
+    element(element) {
+      const rawOrdinal = element.getAttribute(rawMarker);
+      const hasId = element.hasAttribute("id");
+      records.push({
+        id: hasId
+          ? decodeHtmlEntities(
+            element.getAttribute("id") ?? "",
+            { attribute: true },
+          )
+          : undefined,
+        level: Number(String(element.tagName).slice(1)),
+        namespace: element.namespaceURI,
+        rawOrdinal: /^\d+$/u.test(rawOrdinal ?? "")
+          ? Number(rawOrdinal)
+          : null,
+        template: templateDepth > 0,
+      });
+    },
+  };
+  transformHtmlSynchronously(html, (rewriter) => {
+    rewriter = rewriter.on("template", templateHandler);
+    for (let level = 1; level <= 6; level++)
+      rewriter = rewriter.on(`h${level}`, headingHandler);
+    return rewriter;
+  });
+  return records;
+}
+
+function equivalentHeadingRecords(original, marked) {
+  return (
+    original.length === marked.length
+    && original.every((record, ordinal) => {
+      const candidate = marked[ordinal];
+      return (
+        record.level === candidate.level
+        && record.namespace === candidate.namespace
+        && record.template === candidate.template
+        && (
+          candidate.rawOrdinal == null
+          || String(record.id ?? "") === String(candidate.id ?? "")
+        )
+      );
+    })
+  );
+}
+
+export function collectRawHtmlHeadingDeclarations(
+  markdown,
+  { strict = false, includeLevel = false } = {},
+) {
+  const marked = markRawHeadingCandidates(markdown);
+  if (marked.candidates.length === 0) return [];
+
+  const original = Bun.markdown.html(
+    String(markdown ?? ""),
+    { headings: { ids: true } },
+  );
+  const rendered = Bun.markdown.html(
+    marked.source,
+    { headings: { ids: true } },
+  );
+  const originalRecords = browserHeadingRecords(original, marked.marker);
+  const markedRecords = browserHeadingRecords(rendered, marked.marker);
+  if (!equivalentHeadingRecords(originalRecords, markedRecords)) {
+    if (strict)
+      throw new Error("Raw HTML headings could not be classified safely");
+    return [];
+  }
+
+  const declarations = [];
+  for (const heading of markedRecords) {
+    if (
+      heading.template
+      || heading.rawOrdinal == null
+      || !heading.id
+    ) continue;
+    const candidate = marked.candidates[heading.rawOrdinal];
+    if (!candidate) continue;
+    declarations.push({
+      ordinal: heading.rawOrdinal,
+      id: heading.id,
+      kind: "raw HTML heading",
+      ...(includeLevel ? { level: heading.level } : {}),
+      line: candidate.line,
+      source: candidate.source,
+    });
+  }
+  declarations.sort((left, right) => left.ordinal - right.ordinal);
+  return declarations.map(({ ordinal, ...declaration }) => declaration);
+}
+
+function rewriteBrowserMarkdownHeadings(
+  originalHtml,
+  markedHtml,
+  rawMarker,
+  markdownHeadings,
+) {
+  const records = browserHeadingRecords(markedHtml, rawMarker);
+  const originalRecords = browserHeadingRecords(originalHtml, rawMarker);
+  if (!equivalentHeadingRecords(originalRecords, records))
+    throw new Error("Raw heading markers changed the rendered structure");
+
+  const mainRecords = records
+    .map((record, ordinal) => ({
+      marked: record,
+      original: originalRecords[ordinal],
+    }))
+    .filter(({ marked }) => !marked.template);
+  const renderedMarkdown = mainRecords.filter(
+    ({ marked }) => marked.rawOrdinal == null,
+  );
+  const normalized = (
+    renderedMarkdown.length === markdownHeadings.length
+    && renderedMarkdown.every(({ original }, ordinal) => {
+      const heading = markdownHeadings[ordinal];
+      return (
+        original.level === heading.level
+        && String(original.id ?? "") === heading.bunId
+      );
+    })
+  );
+  if (!normalized)
+    throw new Error("Rendered Markdown headings could not be aligned");
+
+  const usedIds = new Set(
+    mainRecords
+      .filter(({ marked }) => marked.rawOrdinal != null)
+      .map(({ original }) => original.id)
+      .filter(Boolean),
+  );
+  for (const heading of markdownHeadings) {
+    if (!heading.fallback && heading.bunId) usedIds.add(heading.bunId);
+  }
+  for (const heading of markdownHeadings) {
+    if (!heading.fallback) {
+      heading.id = heading.bunId;
+      continue;
+    }
+    const base = normalizeUnicodeHeadingId(heading.text);
+    let id = base;
+    let suffix = 1;
+    while (usedIds.has(id)) id = `${base}-${suffix++}`;
+    heading.id = id;
+    usedIds.add(id);
+  }
+
+  let boundary = "MDCUI_HEADING_CONTENT";
+  let boundarySuffix = 0;
+  while (String(originalHtml).includes(boundary))
+    boundary = `MDCUI_HEADING_CONTENT_${++boundarySuffix}`;
+  let markdownOrdinal = 0;
+  let recordOrdinal = 0;
+  const headingHandler = {
+    element(element) {
+      const record = records[recordOrdinal++];
+      if (record?.template || record?.rawOrdinal != null) return;
+
+      const ordinal = markdownOrdinal++;
+      const heading = markdownHeadings[ordinal];
+      if (!heading) return;
+      element.setAttribute("id", heading.id);
+      element.prepend(
+        `<!--${boundary}:s:${ordinal.toString(36)}-->`,
+        { html: true },
+      );
+      element.append(
+        `<!--${boundary}:e:${ordinal.toString(36)}-->`,
+        { html: true },
+      );
+    },
+  };
+  const rewritten = transformHtmlSynchronously(originalHtml, (rewriter) => {
+    for (let level = 1; level <= 6; level++)
+      rewriter = rewriter.on(`h${level}`, headingHandler);
+    return rewriter;
+  });
+
+  const innerHtml = new Map();
+  const starts = new Map();
+  const boundaryPattern = new RegExp(
+    `<!--${boundary}:([se]):([0-9a-z]+)-->`,
+    "g",
+  );
+  let match;
+  while ((match = boundaryPattern.exec(rewritten)) !== null) {
+    const ordinal = Number.parseInt(match[2], 36);
+    if (match[1] === "s") {
+      starts.set(ordinal, match.index + match[0].length);
+    } else if (starts.has(ordinal)) {
+      innerHtml.set(
+        ordinal,
+        rewritten.slice(starts.get(ordinal), match.index),
+      );
+    }
+  }
+  boundaryPattern.lastIndex = 0;
+  const cleanHtml = rewritten.replace(boundaryPattern, "");
+
+  return {
+    html: cleanHtml,
+    normalized,
+    headings: markdownHeadings.map((heading, ordinal) => ({
+      id: heading.id,
+      html: innerHtml.get(ordinal) ?? heading.inlineHtml,
+      text: heading.text,
+      level: heading.level,
+      containerPath: heading.containerPath,
+      ordinal,
+    })),
+  };
+}
+
+export function renderMarkdownWithHeadingIds(markdown) {
+  const source = String(markdown ?? "");
+  const headingEvents = markdownHeadingEvents(source);
+  const markdownHeadings = headingEvents.headings;
+  const html = String(Bun.markdown.html(
+    source,
+    { headings: { ids: true } },
+  ));
+  if (markdownHeadings.length === 0)
+    return { html, headings: [], normalized: true };
+
+  const marked = markRawHeadingCandidates(source);
+  if (
+    typeof HTMLRewriter === "function"
+    && typeof Bun.peek === "function"
+  ) {
+    try {
+      const markedHtml = String(Bun.markdown.html(
+        marked.source,
+        { headings: { ids: true } },
+      ));
+      return rewriteBrowserMarkdownHeadings(
+        html,
+        markedHtml,
+        marked.marker,
+        markdownHeadings,
+      );
+    } catch {
+      return {
+        html,
+        headings: [],
+        normalized: false,
+      };
+    }
+  }
+
+  const renderedHeadings = htmlHeadings(html);
+  const matched = matchedMarkdownHeadings(
+    markdownHeadings,
+    renderedHeadings,
+    headingEvents.events,
+  );
+  if (!matched) {
+    return {
+      html,
+      headings: legacyHtmlHeadings(html),
+      normalized: false,
+    };
+  }
+
+  const matchedRendered = new Set(matched.map(item => item.rendered));
+  const usedIds = new Set(
+    renderedHeadings
+      .filter(heading => !matchedRendered.has(heading))
+      .map(heading => heading.id)
+      .filter(Boolean),
+  );
+  for (const heading of matched) {
+    if (!heading.fallback && heading.bunId) usedIds.add(heading.bunId);
+  }
+
+  for (const heading of matched) {
+    if (!heading.fallback) {
+      heading.id = heading.bunId;
+      continue;
+    }
+    const base = normalizeUnicodeHeadingId(heading.text);
+    let id = base;
+    let suffix = 1;
+    while (usedIds.has(id)) id = `${base}-${suffix++}`;
+    heading.id = id;
+    usedIds.add(id);
+  }
+
+  const replacementByStart = new Map(
+    matched.map(heading => [
+      heading.rendered.start,
+      replaceHeadingId(heading.rendered.source, heading.id),
+    ]),
+  );
+  let rewritten = "";
+  let cursor = 0;
+  for (const rendered of renderedHeadings) {
+    const replacement = replacementByStart.get(rendered.start);
+    if (replacement == null) continue;
+    rewritten += html.slice(cursor, rendered.start) + replacement;
+    cursor = rendered.end;
+  }
+  rewritten += html.slice(cursor);
+
+  return {
+    html: rewritten,
+    normalized: true,
+    headings: matched.map((heading, ordinal) => ({
+      id: heading.id,
+      html: heading.rendered.innerHTML,
+      text: heading.text,
+      level: heading.level,
+      containerPath: heading.containerPath,
+      ordinal,
+    })),
+  };
+}