@ai-react-markdown/engine 2.5.3 → 2.5.5

package/dist/index.js

@@ -243,7 +243,28 @@ var HTML_BREAKOUT_TAGS = /* @__PURE__ */ new Set([
   "ul",
   "var"
 ]);
-var HTML_INTEGRATION_POINTS = ["foreignobject", "desc", "mi", "mo", "mn", "ms", "mtext", "annotation-xml"];
+var HTML_INTEGRATION_POINTS = ["foreignobject", "desc", "title", "mi", "mo", "mn", "ms", "mtext", "annotation-xml"];
+var SCOPE_BARRIER_NAMES = /* @__PURE__ */ new Set([
+  "applet",
+  "caption",
+  "html",
+  "table",
+  "td",
+  "th",
+  "marquee",
+  "object",
+  "template",
+  "mi",
+  "mo",
+  "mn",
+  "ms",
+  "mtext",
+  "annotation-xml",
+  "foreignobject",
+  "desc",
+  "title"
+]);
+var FOREIGN_ROOT_NAMES = ["svg", "math"];
 var TYPE1_NAMES = /* @__PURE__ */ new Set(["script", "pre", "style", "textarea"]);
 var RAW_TEXT_ELEMENTS = /* @__PURE__ */ new Set([
   "script",
@@ -413,6 +434,8 @@ function freshCheckpoint(defListEnabled, mathFlow, referenceTaint) {
     piOpen: false,
     bogusOpen: false,
     rawTextOpen: null,
+    openStack: [],
+    scriptDataEscaped: false,
     declOpen: false,
     cdataOpen: false,
     inFence: false,
@@ -653,7 +676,7 @@ function processConfirmedLine(cp, ln, text) {
       cp.htmlSeamPending = false;
     }
   }
-  const inForeignContent = () => (cp.tagBalance.get("svg") ?? 0) > 0 || (cp.tagBalance.get("math") ?? 0) > 0;
+  const inForeignContent = () => FOREIGN_ROOT_NAMES.some((name) => (cp.tagBalance.get(name) ?? 0) > 0);
   const honoursSelfClosing = (tag) => {
     if (tag === "svg" || tag === "math") return true;
     if (!inForeignContent() || HTML_BREAKOUT_TAGS.has(tag)) return false;
@@ -665,23 +688,57 @@ function processConfirmedLine(cp, ln, text) {
     for (const ip of HTML_INTEGRATION_POINTS) if ((cp.tagBalance.get(ip) ?? 0) > 0) return true;
     return false;
   };
+  const popForeignRoots = () => {
+    for (const name of FOREIGN_ROOT_NAMES) {
+      const count = cp.tagBalance.get(name) ?? 0;
+      if (count > 0) {
+        cp.tagBalance.set(name, 0);
+        cp.openTotal -= count;
+      }
+    }
+    if (cp.openStack.length > 0) cp.openStack = cp.openStack.filter((n) => !FOREIGN_ROOT_NAMES.includes(n));
+  };
+  const noteBreakout = (tag, closing) => {
+    if (closing || cp.rawTextOpen !== null) return;
+    if (HTML_BREAKOUT_TAGS.has(tag) && !htmlRulesApply()) popForeignRoots();
+  };
   const applyTag = (tag, closing) => {
     if (cp.rawTextOpen !== null) {
+      if (cp.scriptDataEscaped && !closing && tag === "script") {
+        cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start);
+      }
       if (!(closing && tag === cp.rawTextOpen)) return;
       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);
+      cp.scriptDataEscaped = false;
+    } else {
+      noteBreakout(tag, closing);
+      if (!closing && RAW_TEXT_ELEMENTS.has(tag) && htmlRulesApply()) {
+        cp.rawTextOpen = tag;
+        cp.scriptDataEscaped = false;
+        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 (tag === "template" && !closing) cp.phasePoisonedAt = 0;
     if (closing) {
+      let idx = -1;
+      for (let i = cp.openStack.length - 1; i >= 0; i--) {
+        if (cp.openStack[i] === tag) {
+          idx = i;
+          break;
+        }
+        if (SCOPE_BARRIER_NAMES.has(cp.openStack[i])) break;
+      }
+      if (idx === -1) return;
+      cp.openStack.splice(idx, 1);
       const count = cp.tagBalance.get(tag) ?? 0;
       if (count > 0) {
         cp.tagBalance.set(tag, count - 1);
         cp.openTotal -= 1;
       }
     } else {
+      cp.openStack.push(tag);
       cp.tagBalance.set(tag, (cp.tagBalance.get(tag) ?? 0) + 1);
       cp.openTotal += 1;
     }
@@ -800,6 +857,9 @@ function processConfirmedLine(cp, ln, text) {
     if (!cp.type1FlowOpen) {
       cp.htmlFlowSinceBlank = false;
       cp.htmlFlowReal = false;
+      if (cp.rawTextOpen !== null && !cp.rawTextInline) {
+        cp.phasePoisonedAt = 0;
+      }
     }
     cp.prevLineBlank = true;
     cp.prevLineWasText = false;
@@ -899,6 +959,7 @@ ${cont(scanText)}` };
   const poisonRawDivergence = () => {
     cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start);
   };
+  let inlineRawOpenerIdx = -1;
   while (pos < scanText.length) {
     if (cp.piOpen) {
       const c = scanText.indexOf("?>", pos);
@@ -954,6 +1015,7 @@ ${cont(scanText)}` };
     } else if (first === cd) {
       rawSpans.push([cd, cd + 9]);
       cp.cdataOpen = true;
+      if (!isMdBlank(scanText.slice(0, cd)) || ln.indent > 3) inlineRawOpenerIdx = cd;
       pos = cd + 9;
     } else if (first === pi) {
       if (scanText[pi + 2] === ">") {
@@ -964,14 +1026,19 @@ ${cont(scanText)}` };
       }
       rawSpans.push([pi, pi + 2]);
       cp.piOpen = true;
+      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.declOpen = true;
+      if (!isMdBlank(scanText.slice(0, decl)) || ln.indent > 3) inlineRawOpenerIdx = decl;
       pos = decl + 2;
     }
   }
+  if (inlineRawOpenerIdx !== -1 && (cp.piOpen || cp.declOpen || cp.cdataOpen)) {
+    cp.phasePoisonedAt = 0;
+  }
   let tagText = scanText;
   for (const [from, to] of rawSpans) {
     tagText = tagText.slice(0, from) + " ".repeat(to - from) + tagText.slice(to);
@@ -999,7 +1066,10 @@ ${cont(scanText)}` };
     let m;
     let lastCommentOpenerIdx = -1;
     while ((m = TAG_OR_COMMENT_RE.exec(tagText)) !== null) {
-      if (cp.rawTextOpen !== null && (m[0] === "<!--" || m[0] === "-->" || m[0] === "--!>")) continue;
+      if (cp.rawTextOpen !== null && (m[0] === "<!--" || m[0] === "-->" || m[0] === "--!>")) {
+        if (cp.rawTextOpen === "script" && m[0] === "<!--") cp.scriptDataEscaped = true;
+        continue;
+      }
       if (m[0] === "<!--") {
         const next = tagText.slice(m.index + 4, m.index + 6);
         if (cp.commentOpen) {
@@ -1053,12 +1123,13 @@ ${cont(scanText)}` };
         continue;
       }
       const selfClosing = /\/\s*$/.test(attrs);
+      noteBreakout(tag, closing);
       if (VOID_TAGS.has(tag) || selfClosing && honoursSelfClosing(tag)) continue;
       applyTag(tag, closing);
     }
-    if (cp.commentOpen && lastCommentOpenerIdx !== -1 && !inRawText) {
-      if (!isMdBlank(tagText.slice(0, lastCommentOpenerIdx))) {
-        cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start + lastCommentOpenerIdx);
+    if (cp.commentOpen && lastCommentOpenerIdx !== -1) {
+      if (!isMdBlank(tagText.slice(0, lastCommentOpenerIdx)) || ln.indent > 3) {
+        cp.phasePoisonedAt = 0;
       }
     }
     if (masked !== null && masked !== ln.text) {
@@ -1075,6 +1146,7 @@ ${cont(scanText)}` };
         if (strayTablePart(tag)) cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start + mr.index);
         if (closing && mr[3] !== void 0 && !/^\s*$/.test(mr[3])) continue;
         const selfClosing = mr[3] !== void 0 && /\/\s*$/.test(mr[3]);
+        noteBreakout(tag, closing);
         if (VOID_TAGS.has(tag) || selfClosing && honoursSelfClosing(tag)) continue;
         applyTag(tag, closing);
       }
@@ -1357,6 +1429,40 @@ function hasStrayTablePart(values) {
   }
   return false;
 }
+var HEAD_ROUTED_NAMES = /* @__PURE__ */ new Set([
+  "base",
+  "basefont",
+  "bgsound",
+  "link",
+  "meta",
+  "noframes",
+  "script",
+  "style",
+  "template",
+  "title"
+]);
+var HEAD_ROUTED_RAW_TEXT_RE = /<(script|style|title|noframes)(?=[\s/>])/i;
+var ANY_START_TAG_RE = /<([a-z][a-z0-9-]*)(?=[\s/>])/gi;
+function headRoutedCaptureUnclosed(values) {
+  if (values.length === 0) return false;
+  const joined = values.join("\n");
+  const opener = HEAD_ROUTED_RAW_TEXT_RE.exec(joined);
+  if (opener === null) return false;
+  const before = joined.slice(0, opener.index);
+  if (/<[!?]/.test(before)) return true;
+  ANY_START_TAG_RE.lastIndex = 0;
+  let m;
+  while ((m = ANY_START_TAG_RE.exec(before)) !== null) {
+    if (!HEAD_ROUTED_NAMES.has(m[1].toLowerCase())) return false;
+  }
+  const name = opener[1].toLowerCase();
+  const after = joined.slice(opener.index + opener[0].length);
+  if (name === "script") {
+    const close = after.search(/<\/script(?=[\s/>])/i);
+    return close === -1 || /<!--/.test(after.slice(0, close));
+  }
+  return !new RegExp(`</${name}(?=[\\s/>])`, "i").test(after);
+}
 var STRAY_SYNTHESIZED_END_TAG_RE = /<\/(?:br|p)\b/i;
 function spliceTrees(input) {
   const { prevMdast, prevHast, tailMdast, tailHast, content, boundary, injectionPrefix, injectedSegments } = input;
@@ -1415,11 +1521,14 @@ function spliceTrees(input) {
   });
   const tailWrapVisible = tailMdastChildren.some((child) => !isWrapInvisible(child));
   if (hasStrayTablePart(prefixMdast.flatMap((c) => c.type === "html" ? [c.value] : []))) return null;
+  const leadingHtml = [];
   for (const child of tailMdastChildren) {
     if (isWrapInvisible(child)) continue;
     if (child.type !== "html") break;
     if (STRAY_SYNTHESIZED_END_TAG_RE.test(child.value) || TABLE_PART_TAG_RE.test(child.value)) return null;
+    leadingHtml.push(child.value);
   }
+  if (headRoutedCaptureUnclosed(leadingHtml)) return null;
   const aligned = alignPrefixCut(prefixMdast, cutRegion, tailWrapVisible);
   if (aligned === null) return null;
   const hastChildren = aligned.children;
@@ -2344,7 +2453,7 @@ function createRegistry(onEmpty) {
 
 // src/components/pluginChain.ts
 import rehypeKatex from "rehype-katex";
-import rehypeRaw from "rehype-raw";
+import rehypeRaw from "@ai-markdown/rehype-raw";
 import rehypeUnwrapImages from "rehype-unwrap-images";
 
 // src/components/rehypeUnwrapCrossChunkImages.ts