@ai-react-markdown/engine 2.7.0 → 2.8.1

package/dist/index.dev.cjs

@@ -48,7 +48,6 @@ __export(src_exports, {
   collectDefLabels: () => collectDefLabels,
   computeFreezeBoundary: () => computeFreezeBoundary,
   createDefLabelScanner: () => createDefLabelScanner,
-  createFile: () => createFile,
   createIncrementalLatexPreprocessor: () => createIncrementalLatexPreprocessor,
   createProcessor: () => createProcessor,
   createRegistry: () => createRegistry,
@@ -65,10 +64,8 @@ __export(src_exports, {
   hasLoneSurrogate: () => hasLoneSurrogate,
   highlight: () => highlight,
   isFootnoteSection: () => isFootnoteSection,
-  isWhitespaceText: () => isWhitespaceText,
   lastMeaningfulIdx: () => lastMeaningfulIdx,
   measureStage: () => measureStage,
-  mergeClassNameAllowlist: () => mergeClassNameAllowlist,
   normalizeForMatch: () => normalizeForMatch,
   normalizeId: () => normalizeId,
   pangu: () => pangu,
@@ -84,7 +81,6 @@ __export(src_exports, {
   shortenDocumentId: () => shortenDocumentId,
   smartypants: () => smartypants,
   sourceIdFromFootnoteLiId: () => sourceIdFromFootnoteLiId,
-  splitByProtectedRegions: () => splitByProtectedRegions,
   subscribeStageTimings: () => subscribeStageTimings,
   transformStage: () => transformStage,
   withDefs: () => withDefs
@@ -519,12 +515,82 @@ var RAW_TEXT_ELEMENTS = /* @__PURE__ */ new Set([
   // (oracle review of the r2 batch; regression caught before release).
   "plaintext"
 ]);
+var P5_MARKUP_RE = /<[!/?A-Za-z]/;
 var TYPE6_START_RE = /^<\/?([A-Za-z][A-Za-z0-9-]*)(?:[ \t\r]|\/?>|$)/;
 var TYPE1_START_RE = /^<(script|pre|style|textarea)(?:[ \t\r]|>|$)/i;
 var TYPE1_CLOSE_RE = /<\/(?:script|pre|style|textarea)>/i;
-var TYPE7_LINE_RE = /^(?:<[A-Za-z][A-Za-z0-9-]*(?:[ \t\r][^>]*|\/)?>|<\/[A-Za-z][A-Za-z0-9-]*[ \t\r]*>)[ \t\r]*$/;
-var t7Name = (line) => /^<\/?([A-Za-z][A-Za-z0-9-]*)/.exec(line)[1];
+var isSpaceTab = (c) => c === 32 || c === 9;
+var isAsciiAlpha = (c) => c >= 65 && c <= 90 || c >= 97 && c <= 122;
+var isAlnum = (c) => isAsciiAlpha(c) || c >= 48 && c <= 57;
+var isAttrNameRest = (c) => isAlnum(c) || c === 45 || c === 46 || c === 58 || c === 95;
+var isUnquotedExit = (c) => Number.isNaN(c) || c === 34 || c === 39 || c === 47 || c === 60 || c === 61 || c === 62 || c === 96 || isSpaceTab(c);
+var completeOpenTagRest = (t, from) => {
+  let i = from;
+  for (; ; ) {
+    const c = t.charCodeAt(i);
+    if (c === 47) {
+      i += 1;
+      break;
+    }
+    if (isSpaceTab(c)) {
+      i += 1;
+      continue;
+    }
+    if (!(c === 58 || c === 95 || isAsciiAlpha(c))) break;
+    i += 1;
+    while (isAttrNameRest(t.charCodeAt(i))) i += 1;
+    for (; ; ) {
+      while (isSpaceTab(t.charCodeAt(i))) i += 1;
+      if (t.charCodeAt(i) !== 61) break;
+      i += 1;
+      while (isSpaceTab(t.charCodeAt(i))) i += 1;
+      const v = t.charCodeAt(i);
+      if (Number.isNaN(v) || v === 60 || v === 61 || v === 62 || v === 96) return -1;
+      if (v === 34 || v === 39) {
+        i += 1;
+        while (t.charCodeAt(i) !== v) {
+          if (i >= t.length) return -1;
+          i += 1;
+        }
+        i += 1;
+        const a = t.charCodeAt(i);
+        if (!(a === 47 || a === 62 || isSpaceTab(a))) return -1;
+        break;
+      }
+      while (!isUnquotedExit(t.charCodeAt(i))) i += 1;
+    }
+  }
+  return t.charCodeAt(i) === 62 ? i + 1 : -1;
+};
+var isType7Line = (line) => {
+  const cr = line.indexOf("\r");
+  const t = cr === -1 ? line : line.slice(0, cr);
+  if (t.charCodeAt(0) !== 60) return false;
+  let i = 1;
+  const closing = t.charCodeAt(i) === 47;
+  if (closing) i += 1;
+  if (!isAsciiAlpha(t.charCodeAt(i))) return false;
+  const nameStart = i;
+  i += 1;
+  while (isAlnum(t.charCodeAt(i)) || t.charCodeAt(i) === 45) i += 1;
+  const c = t.charCodeAt(i);
+  if (!(Number.isNaN(c) || c === 47 || c === 62 || isSpaceTab(c))) return false;
+  const name = t.slice(nameStart, i).toLowerCase();
+  if (!closing && c !== 47 && TYPE1_NAMES.has(name)) return false;
+  if (TYPE6_NAMES.has(name)) return false;
+  if (closing) {
+    while (isSpaceTab(t.charCodeAt(i))) i += 1;
+    if (t.charCodeAt(i) !== 62) return false;
+    i += 1;
+  } else {
+    i = completeOpenTagRest(t, i);
+    if (i === -1) return false;
+  }
+  while (isSpaceTab(t.charCodeAt(i))) i += 1;
+  return i >= t.length;
+};
 var mdHtml = (b, type) => b.kind === "html" && b.type === type;
+var mdType1RawText = (b) => b.kind === "html" && b.type === 1 && b.raw;
 var mdHtml25 = (b) => b.kind === "html" && b.type >= 2 && b.type <= 5;
 var commentEitherOpen = (md, p5) => mdHtml(md, 2) || p5.kind === "comment";
 var inRawTextTok = (t) => t.kind === "rawText" || t.kind === "script";
@@ -546,7 +612,13 @@ var VOID_TAGS = /* @__PURE__ */ new Set([
   "wbr"
 ]);
 var LIST_MARKER_RE = /^ {0,3}(?:[-*+]|\d{1,9}[.)])(?:[ \t]|$)/;
+var CONTAINER_MARKER_RE = /^ {0,3}(?:>|(?:[-*+]|\d{1,9}[.)])[ \t]+\S)/;
+var ATX_HEADING_RE = /^#{1,6}(?:[ \t]|$)/;
+var THEMATIC_BREAK_RE = /^(?:(?:-[ \t]*){3,}|(?:\*[ \t]*){3,}|(?:_[ \t]*){3,})$/;
+var BARE_MARKER_RE = /^(?:[-*+]|\d{1,9}[.)])[ \t]*$/;
+var SETEXT_LEFTOVER_RE = /^(?:=+|--)[ \t]*$/;
 var DEF_LIST_DD_RE = /^ {0,3}:[ \t]/;
+var CLOSE_TAG_ONLY_RE = /^[ \t]*<\/[A-Za-z][A-Za-z0-9-]*[ \t]*>[ \t\r]*$/;
 var FENCE_RE = /^ {0,3}(`{3,}|~{3,})/;
 var MATH_RUN_RE = /^ {0,3}(\$\$+)/;
 var TAG_OR_COMMENT_RE = /<(\/?)([A-Za-z][A-Za-z0-9-]*)(?=[\s/>])([^>]*)>|<!--|-->|--!>/g;
@@ -655,10 +727,14 @@ function freshCheckpoint(defListEnabled, mathFlow, referenceTaint) {
     // doc start counts as a block start
     prevLineWasText: false,
     prevLineWasValidDef: false,
+    prevLineOpenContent: false,
+    tableMaybeOpen: false,
+    containerMaybeOpen: false,
     paragraphHasUnpairedRun: false,
     openBracket: null,
-    mayBeRawToMicromark: false,
     p5SealPending: false,
+    defBlockMaybeOpen: false,
+    fnDefResumable: false,
     phasePoisonedAt: Infinity,
     pendingTruncatedTags: [],
     pendingTruncatedCloses: [],
@@ -785,10 +861,11 @@ function processConfirmedLine(cp, ln, text) {
     newest.defListSettled = ln.blank ? true : !canBecomeDdLine(ln.text, true);
   }
   const isBlockStart = cp.prevLineBlank;
-  if (cp.p5SealPending && !ln.blank && !cp.mayBeRawToMicromark && !(mdHtml25(cp.mdBlock) || cp.p5Tok.kind === "comment" || cp.p5Tok.kind === "bogus")) {
-    const defShapedLine = DEF_RE.test(ln.text) || FOOTNOTE_DEF_RE.test(ln.text);
+  if (cp.p5SealPending && !ln.blank && cp.mdBlock.kind !== "html" && !(cp.p5Tok.kind === "comment" || cp.p5Tok.kind === "bogus")) {
+    const defShapedLine = DEF_RE.test(ln.text) || FOOTNOTE_DEF_RE.test(ln.text) || cp.defBlockMaybeOpen || cp.fnDefResumable && ln.indent >= 4;
     const commentOnly = ln.text.replace(/<!--[\s\S]*?-->/g, " ").replace(/<!--[\s\S]*$/, " ").replace(/[ \t\r]/g, "") === "";
-    if (!defShapedLine && !commentOnly) {
+    const closeTagOnly = CLOSE_TAG_ONLY_RE.test(ln.text);
+    if (!defShapedLine && !commentOnly && !closeTagOnly) {
       cp.p5SealPending = false;
     }
   }
@@ -798,9 +875,13 @@ function processConfirmedLine(cp, ln, text) {
   const applyTag = (tag, closing) => {
     if (inRawTextTok(cp.p5Tok)) {
       if (cp.p5Tok.kind === "script" && cp.p5Tok.escaped && !closing && tag === "script") {
-        cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start);
+        cp.p5Tok = { ...cp.p5Tok, double: true };
       }
       if (!(closing && tag === rawTextElement(cp.p5Tok))) return;
+      if (cp.p5Tok.kind === "script" && cp.p5Tok.double) {
+        cp.p5Tok = { ...cp.p5Tok, double: false };
+        return;
+      }
       cp.p5Tok = { kind: "data" };
     } else {
       if (!closing && RAW_TEXT_ELEMENTS.has(tag) && foreignRawTextSwitchUnknowable()) {
@@ -808,7 +889,7 @@ function processConfirmedLine(cp, ln, text) {
       } else if (!closing && RAW_TEXT_ELEMENTS.has(tag)) {
         if (cp.p5Tok.kind !== "data") cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start);
         const openedInline = cp.mdBlock.kind !== "html";
-        cp.p5Tok = tag === "script" ? { kind: "script", escaped: false, openedInline } : { kind: "rawText", element: tag, openedInline };
+        cp.p5Tok = tag === "script" ? { kind: "script", escaped: false, double: false, openedInline } : { kind: "rawText", element: tag, openedInline };
         if (tag === "plaintext") cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start);
       }
     }
@@ -836,9 +917,14 @@ function processConfirmedLine(cp, ln, text) {
       cp.openTotal += 1;
     }
   };
-  const definitelyInsideTable = () => (cp.tagBalance.get("table") ?? 0) > 0;
+  const definitelyInsideTable = () => (cp.tagBalance.get("table") ?? 0) > cp.pendingTruncatedTags.filter((t) => t === "table").length;
   const strayTablePart = (tag) => TABLE_PART_NAMES.has(tag) && !definitelyInsideTable();
   const commentOpenAtLineStart = commentEitherOpen(cp.mdBlock, cp.p5Tok);
+  const bothCommentsOpenAtLineStart = mdHtml(cp.mdBlock, 2) && cp.p5Tok.kind === "comment";
+  const inDivergenceWindow = mdHtml(cp.mdBlock, 2) && cp.p5Tok.kind !== "comment" || (mdHtml(cp.mdBlock, 3) || mdHtml(cp.mdBlock, 5)) && cp.p5Tok.kind !== "bogus";
+  if (inDivergenceWindow && P5_MARKUP_RE.test(ln.text)) {
+    cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start);
+  }
   const rawOpenAtLineStart = mdHtml25(cp.mdBlock) || cp.p5Tok.kind === "comment" || cp.p5Tok.kind === "bogus";
   if (cp.mdBlock.kind === "fence") {
     const close = FENCE_RE.exec(ln.text);
@@ -851,12 +937,15 @@ function processConfirmedLine(cp, ln, text) {
     cp.prevLineBlank = false;
     cp.prevLineWasText = false;
     cp.prevLineWasValidDef = false;
+    cp.prevLineOpenContent = false;
+    cp.tableMaybeOpen = false;
+    cp.containerMaybeOpen = false;
     return;
   }
   if (cp.mdBlock.kind !== "math" && !rawOpenAtLineStart) {
     const open = FENCE_RE.exec(ln.text);
     const bogusInfo = open !== null && open[1][0] === "`" && ln.text.slice(ln.text.indexOf(open[1]) + open[1].length).includes("`");
-    if (open && !bogusInfo && cp.mayBeRawToMicromark) {
+    if (open && !bogusInfo && cp.mdBlock.kind === "html") {
       cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start);
     } else if (open && !bogusInfo) {
       if (isBlockStart) {
@@ -870,6 +959,9 @@ function processConfirmedLine(cp, ln, text) {
       cp.prevLineBlank = false;
       cp.prevLineWasText = false;
       cp.prevLineWasValidDef = false;
+      cp.prevLineOpenContent = false;
+      cp.tableMaybeOpen = false;
+      cp.containerMaybeOpen = false;
       return;
     }
   }
@@ -884,13 +976,16 @@ function processConfirmedLine(cp, ln, text) {
     cp.prevLineBlank = false;
     cp.prevLineWasText = false;
     cp.prevLineWasValidDef = false;
+    cp.prevLineOpenContent = false;
+    cp.tableMaybeOpen = false;
+    cp.containerMaybeOpen = false;
     return;
   }
   const mathRun = cp.mathFlow && !rawOpenAtLineStart ? MATH_RUN_RE.exec(ln.text) : null;
   if (mathRun) {
     const rest = ln.text.slice(ln.text.indexOf(mathRun[1]) + mathRun[1].length);
     if (!rest.includes("$")) {
-      if (cp.mayBeRawToMicromark) {
+      if (cp.mdBlock.kind === "html") {
         cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start);
       } else {
         if (isBlockStart) {
@@ -904,6 +999,9 @@ function processConfirmedLine(cp, ln, text) {
         cp.prevLineBlank = false;
         cp.prevLineWasText = false;
         cp.prevLineWasValidDef = false;
+        cp.prevLineOpenContent = false;
+        cp.tableMaybeOpen = false;
+        cp.containerMaybeOpen = false;
         return;
       }
     }
@@ -919,8 +1017,11 @@ function processConfirmedLine(cp, ln, text) {
     }
     cp.pendingTag = null;
     if (cp.p5Tok.kind === "bogus") {
-      cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start);
-      cp.p5Tok = { kind: "data" };
+      if (mdHtml25(cp.mdBlock)) {
+      } else {
+        cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start);
+        cp.p5Tok = { kind: "data" };
+      }
     }
     if (cp.mdBlock.kind === "html" && cp.mdBlock.type >= 6) cp.mdBlock = { kind: "none" };
     cp.blankRun += 1;
@@ -942,7 +1043,6 @@ function processConfirmedLine(cp, ln, text) {
     cp.paragraphHasUnpairedRun = false;
     cp.openBracket = null;
     if (!mdHtml(cp.mdBlock, 1)) {
-      cp.mayBeRawToMicromark = false;
       if (inRawTextTok(cp.p5Tok) && !cp.p5Tok.openedInline) {
         cp.phasePoisonedAt = 0;
       }
@@ -950,6 +1050,10 @@ function processConfirmedLine(cp, ln, text) {
     cp.prevLineBlank = true;
     cp.prevLineWasText = false;
     cp.prevLineWasValidDef = false;
+    cp.defBlockMaybeOpen = false;
+    cp.prevLineOpenContent = false;
+    cp.tableMaybeOpen = false;
+    cp.containerMaybeOpen = false;
     return;
   }
   if (isBlockStart) {
@@ -961,24 +1065,35 @@ function processConfirmedLine(cp, ln, text) {
   const tagStart = ln.indent <= 3 ? /^<\/?([A-Za-z][A-Za-z0-9-]*)/.exec(mdTrimStart(ln.text)) : null;
   if (tagStart) {
     const noRealBlockOpen = cp.mdBlock.kind !== "html";
-    cp.mayBeRawToMicromark = true;
-    if (noRealBlockOpen && TYPE1_START_RE.test(mdTrimStart(ln.text))) cp.mdBlock = { kind: "html", type: 1 };
-    if (!TYPE6_NAMES.has(tagStart[1].toLowerCase())) cp.hazardVerdict = true;
+    const t1 = noRealBlockOpen ? TYPE1_START_RE.exec(mdTrimStart(ln.text)) : null;
+    if (t1) cp.mdBlock = { kind: "html", type: 1, raw: RAW_TEXT_ELEMENTS.has(t1[1].toLowerCase()) };
     if (cp.mdBlock.kind !== "html") {
       const t = mdTrimStart(ln.text);
       const t6 = TYPE6_START_RE.exec(t);
-      const t7 = TYPE7_LINE_RE.exec(t);
       const realT6 = t6 !== null && TYPE6_NAMES.has(t6[1].toLowerCase());
-      if (realT6 || TYPE1_START_RE.test(t) || // Type 7 cannot interrupt a paragraph, and excludes the raw-text
-      // names (those are type 1 as start tags, paragraph as end tags).
-      t7 !== null && !cp.prevLineWasText && !TYPE1_NAMES.has(t7Name(t).toLowerCase())) {
+      const t1Line = TYPE1_START_RE.test(t);
+      const type7Shaped = !realT6 && !t1Line && isType7Line(t);
+      if (realT6 || t1Line || // Type 7 cannot interrupt CONTENT (micromark's paragraph/definition
+      // construct — `prevLineOpenContent`, the exact interrupt input; the
+      // old `prevLineWasText` gate refused after headings, terminator
+      // lines and fence closes, where micromark measurably opens). The
+      // classifier itself is exact too (isType7Line) — including closing
+      // raw-text names (`</style>` alone is type 7, measured) and
+      // quoted-`>` attribute values.
+      !cp.prevLineOpenContent && type7Shaped) {
         if (cp.mdBlock.kind === "none") cp.mdBlock = { kind: "html", type: realT6 ? 6 : 7 };
+      } else if (cp.prevLineOpenContent && cp.tableMaybeOpen && type7Shaped) {
+        cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start);
+      }
+      if (type7Shaped && cp.containerMaybeOpen) {
+        cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start);
       }
     }
   }
-  const inRawText = cp.mayBeRawToMicromark || rawOpenAtLineStart;
   const rawFlowStart = ln.indent <= 3 && /^<(?:!--|\?|![A-Za-z]|!\[CDATA\[)/.test(mdTrimStart(ln.text));
-  const { masked, unpaired } = inRawText ? { masked: null, unpaired: false } : maskIntraLineCodeSpans(ln.text, cp.paragraphHasUnpairedRun);
+  const htmlOwnedLine = cp.mdBlock.kind === "html" || rawOpenAtLineStart || rawFlowStart;
+  const maskingSuppressed = htmlOwnedLine || inRawTextTok(cp.p5Tok);
+  const { masked, unpaired } = maskingSuppressed ? { masked: null, unpaired: false } : maskIntraLineCodeSpans(ln.text, cp.paragraphHasUnpairedRun);
   if (unpaired) cp.paragraphHasUnpairedRun = true;
   const scanText = masked ?? ln.text;
   const defRawToMicromark = cp.mdBlock.kind === "html" || rawOpenAtLineStart || inRawTextTok(cp.p5Tok);
@@ -990,30 +1105,27 @@ function processConfirmedLine(cp, ln, text) {
   };
   let inlineRawOpenerIdx = -1;
   while (pos < scanText.length) {
-    if (mdHtml(cp.mdBlock, 3)) {
-      const c = scanText.indexOf("?>", pos);
-      const gt = scanText.indexOf(">", pos);
-      if (gt !== -1 && (c === -1 || gt !== c + 1)) poisonRawDivergence();
-      if (c === -1) {
-        rawSpans.push([pos, scanText.length]);
-        break;
-      }
-      rawSpans.push([pos, c + 2]);
-      cp.mdBlock = { kind: "none" };
-      pos = c + 2;
-      continue;
-    }
-    if (mdHtml(cp.mdBlock, 5)) {
-      const c = scanText.indexOf("]]>", pos);
-      const gt = scanText.indexOf(">", pos);
-      if (gt !== -1 && (c === -1 || gt !== c + 2)) poisonRawDivergence();
-      if (c === -1) {
-        rawSpans.push([pos, scanText.length]);
-        break;
+    if (mdHtml(cp.mdBlock, 3) || mdHtml(cp.mdBlock, 5)) {
+      const isPi = mdHtml(cp.mdBlock, 3);
+      const term = isPi ? "?>" : "]]>";
+      const c = scanText.indexOf(term, pos);
+      const mdEnd = c === -1 ? scanText.length : c + term.length;
+      if (cp.p5Tok.kind === "bogus") {
+        const gt = scanText.indexOf(">", pos);
+        if (gt !== -1 && (c === -1 || gt !== c + term.length - 1)) {
+          cp.p5Tok = { kind: "data" };
+          rawSpans.push([pos, gt + 1]);
+          if (P5_MARKUP_RE.test(scanText.slice(gt + 1, c === -1 ? scanText.length : c))) {
+            poisonRawDivergence();
+          }
+        } else {
+          rawSpans.push([pos, mdEnd]);
+          if (c !== -1) cp.p5Tok = { kind: "data" };
+        }
       }
-      rawSpans.push([pos, c + 3]);
+      if (c === -1) break;
       cp.mdBlock = { kind: "none" };
-      pos = c + 3;
+      pos = mdEnd;
       continue;
     }
     if (mdHtml(cp.mdBlock, 4)) {
@@ -1039,7 +1151,12 @@ function processConfirmedLine(cp, ln, text) {
       pos = c + 1;
       continue;
     }
-    if (commentOpenAtLineStart || inRawTextTok(cp.p5Tok) || mdHtml(cp.mdBlock, 1)) break;
+    if (commentOpenAtLineStart || inRawTextTok(cp.p5Tok) || mdType1RawText(cp.mdBlock)) {
+      if (inRawTextTok(cp.p5Tok) && cp.p5Tok.openedInline && (tailCarriesRetroactive(ln.text) || /<template(?![a-z0-9-])/i.test(ln.text))) {
+        cp.phasePoisonedAt = 0;
+      }
+      break;
+    }
     const pi = scanText.indexOf("<?", pos);
     const cd = scanText.indexOf("<![CDATA[", pos);
     const dm = scanText.slice(pos).search(/<![A-Za-z]/);
@@ -1056,7 +1173,8 @@ function processConfirmedLine(cp, ln, text) {
       pos = bogus + 2;
     } else if (first === cd) {
       rawSpans.push([cd, cd + 9]);
-      cp.mdBlock = { kind: "html", type: 5 };
+      if (cp.mdBlock.kind === "none") cp.mdBlock = { kind: "html", type: 5 };
+      if (cp.p5Tok.kind === "data") cp.p5Tok = { kind: "bogus" };
       if (!isMdBlank(scanText.slice(0, cd)) || ln.indent > 3) inlineRawOpenerIdx = cd;
       pos = cd + 9;
     } else if (first === pi) {
@@ -1067,13 +1185,15 @@ function processConfirmedLine(cp, ln, text) {
         continue;
       }
       rawSpans.push([pi, pi + 2]);
-      cp.mdBlock = { kind: "html", type: 3 };
+      if (cp.mdBlock.kind === "none") cp.mdBlock = { kind: "html", type: 3 };
+      if (cp.p5Tok.kind === "data") cp.p5Tok = { kind: "bogus" };
       if (!isMdBlank(scanText.slice(0, pi)) || ln.indent > 3) inlineRawOpenerIdx = pi;
       pos = pi + 2;
     } else {
       rawSpans.push([decl, decl + 2]);
       if (ln.indent <= 3 && /^doctype/i.test(scanText.slice(decl + 2))) cp.phasePoisonedAt = 0;
-      cp.mdBlock = { kind: "html", type: 4 };
+      if (cp.mdBlock.kind === "none") cp.mdBlock = { kind: "html", type: 4 };
+      if (cp.p5Tok.kind === "data") cp.p5Tok = { kind: "bogus" };
       if (!isMdBlank(scanText.slice(0, decl)) || ln.indent > 3) inlineRawOpenerIdx = decl;
       pos = decl + 2;
     }
@@ -1110,7 +1230,7 @@ function processConfirmedLine(cp, ln, text) {
       if (inRawTextTok(cp.p5Tok) && (m[0] === "<!--" || m[0] === "-->" || m[0] === "--!>")) {
         if (cp.p5Tok.kind === "script") {
           if (m[0] === "<!--") cp.p5Tok = { ...cp.p5Tok, escaped: true };
-          if (m[0] === "-->") cp.p5Tok = { ...cp.p5Tok, escaped: false };
+          if (m[0] === "-->") cp.p5Tok = { ...cp.p5Tok, escaped: false, double: false };
         }
         continue;
       }
@@ -1122,7 +1242,9 @@ function processConfirmedLine(cp, ln, text) {
             if (cp.p5Tok.kind === "comment") cp.p5Tok = { kind: "data" };
           } else if (next === "!>" || next === "-!") {
             if (cp.p5Tok.kind === "comment") cp.p5Tok = { kind: "data" };
-            poisonRawDivergence();
+            if (mdHtml(cp.mdBlock, 2) && P5_MARKUP_RE.test(tagText.slice(m.index + m[0].length))) {
+              poisonRawDivergence();
+            }
           }
           continue;
         }
@@ -1140,7 +1262,9 @@ function processConfirmedLine(cp, ln, text) {
         continue;
       }
       if (m[0] === "--!>") {
-        if (commentEitherOpen(cp.mdBlock, cp.p5Tok)) poisonRawDivergence();
+        if (mdHtml(cp.mdBlock, 2) && P5_MARKUP_RE.test(tagText.slice(m.index + m[0].length))) {
+          poisonRawDivergence();
+        }
         if (cp.p5Tok.kind === "comment") cp.p5Tok = { kind: "data" };
         continue;
       }
@@ -1231,14 +1355,16 @@ function processConfirmedLine(cp, ln, text) {
             applyTag(tag, closing);
             const rawLastLt = ln.text.lastIndexOf("<");
             const rawTruncated = rawLastLt !== -1 && !ln.text.includes(">", rawLastLt);
-            if (!closing && !inRawText && rawTruncated) cp.pendingTruncatedTags.push(tag);
+            if (!closing && !(htmlOwnedLine || inRawTextTok(cp.p5Tok)) && rawTruncated) {
+              cp.pendingTruncatedTags.push(tag);
+            }
           }
         }
       }
     }
   }
   const effectiveOpen = cp.openTotal - cp.pendingTruncatedTags.length;
-  if ((inRawText || rawFlowStart) && effectiveOpen <= 0) {
+  if ((htmlOwnedLine || inRawTextTok(cp.p5Tok)) && effectiveOpen <= 0) {
     let masked2 = "";
     let cursor = 0;
     for (const [from, to] of rawSpans) {
@@ -1246,7 +1372,7 @@ function processConfirmedLine(cp, ln, text) {
       cursor = to;
     }
     masked2 += scanText.slice(cursor);
-    if (floatingResidue(masked2, commentOpenAtLineStart).length > 0) {
+    if (floatingResidue(masked2, bothCommentsOpenAtLineStart).length > 0) {
       cp.p5SealPending = true;
     }
   }
@@ -1255,12 +1381,36 @@ function processConfirmedLine(cp, ln, text) {
   }
   if (mdHtml(cp.mdBlock, 1) && TYPE1_CLOSE_RE.test(ln.text)) {
     cp.mdBlock = { kind: "none" };
-    cp.mayBeRawToMicromark = false;
+    if (inRawTextTok(cp.p5Tok)) cp.phasePoisonedAt = 0;
   }
   cp.blankRun = 0;
   cp.prevLineBlank = false;
   cp.prevLineWasText = true;
+  {
+    const tt = mdTrimStart(ln.text);
+    let openContent;
+    if (htmlOwnedLine) {
+      openContent = false;
+    } else if (ln.indent >= 4) {
+      openContent = cp.prevLineOpenContent;
+    } else if (ATX_HEADING_RE.test(tt) || THEMATIC_BREAK_RE.test(tt)) {
+      openContent = false;
+    } else if (BARE_MARKER_RE.test(tt)) {
+      openContent = tt[0] === "-" ? false : cp.prevLineOpenContent;
+    } else if (SETEXT_LEFTOVER_RE.test(tt)) {
+      openContent = cp.tableMaybeOpen && cp.prevLineOpenContent ? true : !cp.prevLineOpenContent;
+    } else {
+      openContent = true;
+    }
+    cp.prevLineOpenContent = openContent;
+    cp.tableMaybeOpen = !htmlOwnedLine && ln.text.includes("|") || cp.tableMaybeOpen && openContent;
+    const containerMarker = CONTAINER_MARKER_RE.test(ln.text) || FOOTNOTE_DEF_RE.test(ln.text) || cp.defListEnabled && DEF_LIST_DD_RE.test(ln.text);
+    cp.containerMaybeOpen = openContent && (containerMarker || cp.containerMaybeOpen);
+  }
   cp.prevLineWasValidDef = validLinkDef;
+  cp.defBlockMaybeOpen = cp.defBlockMaybeOpen || DEF_RE.test(ln.text) || FOOTNOTE_DEF_RE.test(ln.text);
+  if (FOOTNOTE_DEF_RE.test(ln.text)) cp.fnDefResumable = true;
+  else if (isBlockStart && ln.indent <= 3) cp.fnDefResumable = false;
 }
 
 // src/components/incrementalParse/spliceParse.ts
@@ -1509,6 +1659,33 @@ function headRoutedCaptureUnclosed(values) {
   return !new RegExp(`</${name}(?=[\\s/>])`, "i").test(after);
 }
 var STRAY_SYNTHESIZED_END_TAG_RE = /<\/(?:br|p)\b/i;
+var RAW_TEXT_OPEN_RE = /<(script|style|textarea|title|xmp|iframe|noembed|noframes|plaintext)(?=[\s/>])/gi;
+function rawTextRegionCrossesOut(values) {
+  for (const value of values) {
+    let pos = 0;
+    for (; ; ) {
+      RAW_TEXT_OPEN_RE.lastIndex = pos;
+      const open = RAW_TEXT_OPEN_RE.exec(value);
+      if (open === null) break;
+      const name = open[1].toLowerCase();
+      const bodyStart = open.index + open[0].length;
+      const closeRe = new RegExp(`</${name}(?=[\\s/>])`, "ig");
+      closeRe.lastIndex = bodyStart;
+      let close = closeRe.exec(value);
+      if (name === "script") {
+        while (close !== null) {
+          const body = value.slice(bodyStart, close.index);
+          const lastOpen = body.lastIndexOf("<!--");
+          if (lastOpen === -1 || body.indexOf("-->", lastOpen + 4) !== -1) break;
+          close = closeRe.exec(value);
+        }
+      }
+      if (close === null) return true;
+      pos = close.index + close[0].length;
+    }
+  }
+  return false;
+}
 function spliceTrees(input) {
   const { prevMdast, prevHast, tailMdast, tailHast, content, boundary, injectionPrefix, injectedSegments } = input;
   const injectedLen = injectionPrefix.length;
@@ -1565,7 +1742,9 @@ function spliceTrees(input) {
     return !(start !== void 0 && start < injectedLen);
   });
   const tailWrapVisible = tailMdastChildren.some((child) => !isWrapInvisible(child));
-  if (hasStrayTablePart(prefixMdast.flatMap((c) => c.type === "html" ? [c.value] : []))) return null;
+  const prefixHtmlValues = prefixMdast.flatMap((c) => c.type === "html" ? [c.value] : []);
+  if (hasStrayTablePart(prefixHtmlValues)) return null;
+  if (rawTextRegionCrossesOut(prefixHtmlValues)) return null;
   const leadingHtml = [];
   for (const child of tailMdastChildren) {
     if (isWrapInvisible(child)) continue;
@@ -1746,7 +1925,7 @@ function alignPrefixCut(prefixMdast, cutRegion, tailWrapVisible) {
   if (sepBuffer.length !== trailingGaps && sepBuffer.length !== trailingGaps + 1) return null;
   for (let j = pairIdx + 1; j < visibles.length; j++) {
     const v = visibles[j];
-    if (v.type === "html" && !/^\s*<[!?]/.test(v.value)) return null;
+    if (v.type === "html" && !isExactSanitizeStrippedConstruct(v.value)) return null;
   }
   for (let i = 0; i < trailingGaps + seam; i++) {
     out.push({ type: "text", value: "\n" });
@@ -1796,7 +1975,7 @@ function tailLeadingTextIsHoist(tailMdastChildren, tailHastChildren) {
   if (!isSeparatorText(firstText)) {
     if (firstText.type === "text" && firstText.position === void 0 && firstVisible?.type === "html") {
       if (/^\s*<\/[A-Za-z][A-Za-z0-9-]*\s*>/.test(firstVisible.value)) return true;
-      if (isCompleteRawConstruct(firstVisible.value)) return false;
+      if (isSanitizeStrippedConstruct(firstVisible.value)) return false;
       return null;
     }
     return false;
@@ -1812,12 +1991,26 @@ function tailLeadingTextIsHoist(tailMdastChildren, tailHastChildren) {
     }
   }
   if (firstVisible.type === "math") return false;
-  if (firstVisible.type === "html" && isCompleteRawConstruct(firstVisible.value)) return false;
+  if (firstVisible.type === "html" && isSanitizeStrippedConstruct(firstVisible.value)) return false;
   return null;
 }
-function isCompleteRawConstruct(value) {
+function isSanitizeStrippedConstruct(value) {
   const v = value.trim();
-  return v.startsWith("<!--") && v.endsWith("-->") || v.startsWith("<?") && v.endsWith("?>") || v.startsWith("<![CDATA[") && v.endsWith("]]>") || /^<![A-Za-z]/.test(v) && v.endsWith(">");
+  return v.startsWith("<!--") && v.endsWith("-->") || v.startsWith("<?") && v.endsWith("?>") || v.startsWith("<![CDATA[") && v.endsWith("]]>") || /^<![A-Za-z]/.test(v) && !/^<!doctype/i.test(v) && v.endsWith(">");
+}
+function isExactSanitizeStrippedConstruct(value) {
+  if (value.startsWith("<!--")) {
+    if (value.length < 7 || !value.endsWith("-->")) return false;
+    const close = value.indexOf("-->", 4);
+    const bangClose = value.indexOf("--!>", 4);
+    if (close !== value.length - 3) return false;
+    return bangClose === -1 || bangClose > close;
+  }
+  if (/^<!doctype/i.test(value)) return false;
+  if (value.startsWith("<?") || /^<!(?:[A-Za-z]|\[CDATA\[)/.test(value)) {
+    return value.indexOf(">") === value.length - 1;
+  }
+  return false;
 }
 function isSeparatorText(node) {
   return node.type === "text" && node.position === void 0 && node.value.trim() === "";
@@ -4983,7 +5176,6 @@ function createRemendPreprocessor(options) {
   collectDefLabels,
   computeFreezeBoundary,
   createDefLabelScanner,
-  createFile,
   createIncrementalLatexPreprocessor,
   createProcessor,
   createRegistry,
@@ -5000,10 +5192,8 @@ function createRemendPreprocessor(options) {
   hasLoneSurrogate,
   highlight,
   isFootnoteSection,
-  isWhitespaceText,
   lastMeaningfulIdx,
   measureStage,
-  mergeClassNameAllowlist,
   normalizeForMatch,
   normalizeId,
   pangu,
@@ -5019,7 +5209,6 @@ function createRemendPreprocessor(options) {
   shortenDocumentId,
   smartypants,
   sourceIdFromFootnoteLiId,
-  splitByProtectedRegions,
   subscribeStageTimings,
   transformStage,
   withDefs