@ai-react-markdown/engine 2.5.2 → 2.5.4

package/dist/index.cjs

@@ -270,6 +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",
@@ -316,7 +336,8 @@ 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 FOREIGN_ROOT_NAMES = ["svg", "math"];
 var TYPE1_NAMES = /* @__PURE__ */ new Set(["script", "pre", "style", "textarea"]);
 var RAW_TEXT_ELEMENTS = /* @__PURE__ */ new Set([
   "script",
@@ -335,6 +356,7 @@ 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([
@@ -485,6 +507,7 @@ function freshCheckpoint(defListEnabled, mathFlow, referenceTaint) {
     piOpen: false,
     bogusOpen: false,
     rawTextOpen: null,
+    scriptDataEscaped: false,
     declOpen: false,
     cdataOpen: false,
     inFence: false,
@@ -510,7 +533,9 @@ function freshCheckpoint(defListEnabled, mathFlow, referenceTaint) {
     tagAcrossLines: false,
     tagAcrossLinesIndent: 0,
     tagAcrossLinesState: "outside",
-    htmlFlowReal: false
+    htmlFlowReal: false,
+    type1FlowOpen: false,
+    rawTextInline: false
   };
 }
 function isPlausibleLinkDefRest(rest) {
@@ -655,8 +680,9 @@ 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;
@@ -722,7 +748,7 @@ 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;
@@ -734,14 +760,37 @@ 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;
+      }
+    }
+  };
+  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;
-      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 (closing) {
       const count = cp.tagBalance.get(tag) ?? 0;
       if (count > 0) {
@@ -852,15 +901,22 @@ 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;
@@ -874,7 +930,9 @@ 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);
@@ -1025,6 +1083,7 @@ ${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;
     }
@@ -1056,7 +1115,10 @@ ${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) {
@@ -1105,8 +1167,12 @@ ${cont(scanText)}` };
           TAG_OR_COMMENT_RE.lastIndex = gt + 1;
         }
       }
-      if (closing && !cp.htmlFlowReal && !/^\s*$/.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);
+      noteBreakout(tag, closing);
       if (VOID_TAGS.has(tag) || selfClosing && honoursSelfClosing(tag)) continue;
       applyTag(tag, closing);
     }
@@ -1129,6 +1195,7 @@ ${cont(scanText)}` };
         if (strayTablePart(tag)) cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start + mr.index);
         if (closing && mr[3] !== void 0 && !/^\s*$/.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);
       }
@@ -1186,6 +1253,14 @@ ${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;