@ai-react-markdown/engine 2.5.1 → 2.5.3

package/dist/index.cjs

@@ -65,6 +65,8 @@ __export(src_exports, {
   hasLoneSurrogate: () => hasLoneSurrogate,
   highlight: () => highlight,
   isFootnoteSection: () => isFootnoteSection,
+  isWhitespaceText: () => isWhitespaceText,
+  lastMeaningfulIdx: () => lastMeaningfulIdx,
   measureStage: () => measureStage,
   mergeClassNameAllowlist: () => mergeClassNameAllowlist,
   normalizeForMatch: () => normalizeForMatch,
@@ -268,6 +270,26 @@ var import_micromark_util_html_tag_name = require("micromark-util-html-tag-name"
 var import_micromark_util_normalize_identifier = require("micromark-util-normalize-identifier");
 var TYPE6_NAMES = new Set(import_micromark_util_html_tag_name.htmlBlockNames);
 var TABLE_PART_NAMES = /* @__PURE__ */ new Set(["td", "th", "tr", "tbody", "thead", "tfoot", "caption", "col", "colgroup"]);
+var DOCUMENT_STRUCTURE_NAMES = /* @__PURE__ */ new Set(["html", "head", "body", "frameset"]);
+var RETROACTIVE_OPENERS = [
+  "<!doctype",
+  "<html",
+  "<head",
+  "<body",
+  "<frameset",
+  "</html",
+  "</head",
+  "</body",
+  "</frameset"
+];
+function tailCarriesRetroactive(text) {
+  const lower = text.toLowerCase();
+  for (let i = lower.indexOf("<"); i !== -1; i = lower.indexOf("<", i + 1)) {
+    const rest = lower.slice(i);
+    for (const op of RETROACTIVE_OPENERS) if (rest.startsWith(op)) return true;
+  }
+  return false;
+}
 var HTML_BREAKOUT_TAGS = /* @__PURE__ */ new Set([
   "b",
   "big",
@@ -333,6 +355,7 @@ var RAW_TEXT_ELEMENTS = /* @__PURE__ */ new Set([
 ]);
 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 VOID_TAGS = /* @__PURE__ */ new Set([
@@ -508,7 +531,9 @@ function freshCheckpoint(defListEnabled, mathFlow, referenceTaint) {
     tagAcrossLines: false,
     tagAcrossLinesIndent: 0,
     tagAcrossLinesState: "outside",
-    htmlFlowReal: false
+    htmlFlowReal: false,
+    type1FlowOpen: false,
+    rawTextInline: false
   };
 }
 function isPlausibleLinkDefRest(rest) {
@@ -653,8 +678,9 @@ function computeFreezeBoundary(text, options, resume) {
     if (tailLine) return !canBecomeDdLine(tailLine.text, false);
     return false;
   };
+  const tailRetroactive = tailLine !== null && tailCarriesRetroactive(tailLine.text);
   let boundary = 0;
-  for (let i = cp.candidates.length - 1; i >= 0; i--) {
+  for (let i = cp.candidates.length - 1; i >= 0 && !tailRetroactive; i--) {
     const c = cp.candidates[i];
     if (!c.htmlBalanced || c.hazard || c.seamRisk) continue;
     if (c.offset > cp.phasePoisonedAt) continue;
@@ -738,8 +764,10 @@ function processConfirmedLine(cp, ln, text) {
       cp.rawTextOpen = null;
     } else if (!closing && RAW_TEXT_ELEMENTS.has(tag) && htmlRulesApply()) {
       cp.rawTextOpen = tag;
+      cp.rawTextInline = !cp.htmlFlowReal;
       if (tag === "plaintext") cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start);
     }
+    if (DOCUMENT_STRUCTURE_NAMES.has(tag)) cp.phasePoisonedAt = 0;
     if (closing) {
       const count = cp.tagBalance.get(tag) ?? 0;
       if (count > 0) {
@@ -850,15 +878,22 @@ function processConfirmedLine(cp, ln, text) {
     cp.candidates.push({
       offset: Math.min(ln.end + 1, text.length),
       blankRun: cp.blankRun,
-      htmlBalanced: cp.openTotal === 0 && !cp.commentOpen && !cp.piOpen && !cp.declOpen && !cp.cdataOpen && !cp.bogusOpen,
+      // `type1FlowOpen`: an unterminated type-1 block swallows this blank
+      // and everything after it as RAW content, so nothing here is a block
+      // boundary at all. Its tags are invisible to the balance scan
+      // (`rawTextOpen` suppresses them), which is exactly why `openTotal`
+      // reads 0 and the candidate looked safe.
+      htmlBalanced: cp.openTotal === 0 && !cp.commentOpen && !cp.piOpen && !cp.declOpen && !cp.cdataOpen && !cp.bogusOpen && !cp.type1FlowOpen,
       hazard: cp.hazardVerdict,
       seamRisk: cp.htmlSeamPending,
       defListSettled: null
     });
     cp.paragraphHasUnpairedRun = false;
     cp.openBracket = null;
-    cp.htmlFlowSinceBlank = false;
-    cp.htmlFlowReal = false;
+    if (!cp.type1FlowOpen) {
+      cp.htmlFlowSinceBlank = false;
+      cp.htmlFlowReal = false;
+    }
     cp.prevLineBlank = true;
     cp.prevLineWasText = false;
     cp.prevLineWasValidDef = false;
@@ -872,7 +907,9 @@ 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.htmlFlowReal;
     cp.htmlFlowSinceBlank = true;
+    if (noRealBlockOpen && TYPE1_START_RE.test(mdTrimStart(ln.text))) cp.type1FlowOpen = true;
     if (!TYPE6_NAMES.has(tagStart[1].toLowerCase())) cp.hazardVerdict = true;
     if (!cp.htmlFlowReal) {
       const t = mdTrimStart(ln.text);
@@ -1023,6 +1060,7 @@ ${cont(scanText)}` };
       pos = pi + 2;
     } else {
       rawSpans.push([decl, decl + 2]);
+      if (ln.indent <= 3 && /^doctype/i.test(scanText.slice(decl + 2))) cp.phasePoisonedAt = 0;
       cp.declOpen = true;
       pos = decl + 2;
     }
@@ -1103,7 +1141,10 @@ ${cont(scanText)}` };
           TAG_OR_COMMENT_RE.lastIndex = gt + 1;
         }
       }
-      if (closing && !cp.htmlFlowReal && !/^\s*$/.test(attrs)) continue;
+      if (closing && !cp.htmlFlowReal && !/^\s*$/.test(attrs)) {
+        TAG_OR_COMMENT_RE.lastIndex = m.index + 2 + m[2].length;
+        continue;
+      }
       const selfClosing = /\/\s*$/.test(attrs);
       if (VOID_TAGS.has(tag) || selfClosing && honoursSelfClosing(tag)) continue;
       applyTag(tag, closing);
@@ -1184,6 +1225,14 @@ ${cont(scanText)}` };
       cp.htmlSeamPending = true;
     }
   }
+  if (cp.rawTextOpen !== null && cp.rawTextInline) {
+    cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start);
+  }
+  if (cp.type1FlowOpen && TYPE1_CLOSE_RE.test(ln.text)) {
+    cp.type1FlowOpen = false;
+    cp.htmlFlowSinceBlank = false;
+    cp.htmlFlowReal = false;
+  }
   cp.blankRun = 0;
   cp.prevLineBlank = false;
   cp.prevLineWasText = true;
@@ -1199,6 +1248,15 @@ function isFootnoteSection(node) {
   const props = node.properties;
   return props?.dataFootnotes !== void 0;
 }
+function isWhitespaceText(c) {
+  return c.type === "text" && /^\s*$/.test(c.value);
+}
+function lastMeaningfulIdx(children) {
+  for (let i = children.length - 1; i >= 0; i--) {
+    if (!isWhitespaceText(children[i])) return i;
+  }
+  return -1;
+}
 
 // src/components/incrementalParse/attributeHastChildren.ts
 function attributeHastChildren(mdast, hast, stopAt = Infinity) {
@@ -1981,16 +2039,6 @@ function isBackrefAnchor(c) {
   if (el.tagName !== "a") return false;
   return Boolean(el.properties && "dataFootnoteBackref" in el.properties);
 }
-function isWhitespaceText(c) {
-  if (c.type !== "text") return false;
-  return /^\s*$/.test(c.value);
-}
-function lastMeaningfulIdx(children) {
-  for (let i = children.length - 1; i >= 0; i--) {
-    if (!isWhitespaceText(children[i])) return i;
-  }
-  return -1;
-}
 function dropTrailingBackrefs(children) {
   let trailingWsStart = children.length;
   while (trailingWsStart > 0 && isWhitespaceText(children[trailingWsStart - 1])) {
@@ -4657,21 +4705,7 @@ function convertSingleToDoubleDollar(text) {
 }
 function preprocessLaTeX(str) {
   if (!hasLatexTrigger(str)) return str;
-  const segments = splitByProtectedRegions(str);
-  const result = segments.map((segment) => {
-    if (segment.isCode) return segment.text;
-    let text = segment.text;
-    text = escapeMhchemCommands(text);
-    text = escapeCurrencyDollarSigns(text);
-    text = convertLatexDelimiters(text);
-    text = escapeLatexPipes(text);
-    text = escapeLatexPipesInUnclosed(text);
-    text = escapeTextUnderscores(text);
-    text = convertSingleToDoubleDollar(text);
-    text = truncateUnclosedLatexBlock(text);
-    return text;
-  });
-  return result.join("");
+  return processSlice(str, false).out;
 }
 function hasLatexTrigger(str) {
   return str.includes("$") || str.includes("\\[") || str.includes("\\(");
@@ -4700,7 +4734,7 @@ function hasUnclosedTextCommand(text) {
 }
 var RESIDUAL_OPEN_BRACKET_RE = /(?<!!)\\\[/;
 var LEADING_DOUBLE_DOLLAR_RE = /^\s*\$\$/;
-function processSliceInstrumented(slice, probe = true) {
+function processSlice(slice, probe = true) {
   const segments = splitByProtectedRegions(slice);
   const parts = [];
   let quiescent = true;
@@ -4824,13 +4858,13 @@ function createIncrementalLatexPreprocessor(options) {
       };
       const cut = findRawSafeCut(active);
       if (cut > 0) {
-        const candidate = processSliceInstrumented(active.slice(0, cut));
+        const candidate = processSlice(active.slice(0, cut));
         if (candidate.quiescent) freeze(cut, candidate);
       }
       nextAttemptLen = advanced || !backoff ? 0 : active.length * 2;
       onAttempt?.({ activeLength, frozenBytes });
     }
-    const tail = processSliceInstrumented(active, false);
+    const tail = processSlice(active, false);
     const head = tail.truncatedAtSeamStart ? frozenOut.replace(/\s+$/, "") : frozenOut;
     const out = head + tail.out;
     prevSource = source;
@@ -4895,6 +4929,8 @@ function createRemendPreprocessor(options) {
   hasLoneSurrogate,
   highlight,
   isFootnoteSection,
+  isWhitespaceText,
+  lastMeaningfulIdx,
   measureStage,
   mergeClassNameAllowlist,
   normalizeForMatch,