@ai-react-markdown/engine 2.4.1 → 2.4.2

package/dist/index.cjs

@@ -295,6 +295,10 @@ var TAG_OR_COMMENT_RE = /<(\/?)([A-Za-z][A-Za-z0-9-]*)(?=[\s/>])([^>]*)>|<!--|--
 var TRUNCATED_TAG_RE = /<(\/?)([A-Za-z][A-Za-z0-9-]*)([^<>]*)$/;
 var REF_RE = /!?\[((?:[^[\]\\]|\\.)*)\]/g;
 var BACKTICK_RUN_RE = /`+/g;
+var MD_BLANK_RE = /^[ \t\r]*$/;
+var isMdBlank = (text) => MD_BLANK_RE.test(text);
+var mdTrim = (text) => text.replace(/^[ \t\r]+|[ \t\r]+$/g, "");
+var mdTrimStart = (text) => text.replace(/^[ \t\r]+/, "");
 function computeIndent(text) {
   let indent = 0;
   for (const ch of text) {
@@ -304,8 +308,25 @@ function computeIndent(text) {
   }
   return indent;
 }
+function firstUnescaped(text, ch) {
+  for (let i = 0; i < text.length; i++) {
+    if (text[i] === "\\") i += 1;
+    else if (text[i] === ch) return i;
+  }
+  return -1;
+}
+function lastUnclosedBracket(text) {
+  let open = -1;
+  for (let i = 0; i < text.length; i++) {
+    const c = text[i];
+    if (c === "\\") i += 1;
+    else if (c === "[") open = i;
+    else if (c === "]") open = -1;
+  }
+  return open;
+}
 function normalizeLabel(label) {
-  const collapsed = label.trim().replace(/[ \t\r\n]+/g, " ");
+  const collapsed = label.replace(/[ \t\r\n]+/g, " ").replace(/^ | $/g, "");
   return collapsed ? (0, import_micromark_util_normalize_identifier.normalizeIdentifier)(collapsed) : "";
 }
 function canBecomeDdLine(text, confirmed) {
@@ -382,6 +403,7 @@ function freshCheckpoint(defListEnabled, mathFlow, referenceTaint) {
     prevLineWasText: false,
     prevLineWasValidDef: false,
     paragraphHasUnpairedRun: false,
+    openBracket: null,
     htmlFlowSinceBlank: false,
     htmlSeamPending: false,
     phasePoisonedAt: Infinity,
@@ -389,11 +411,11 @@ function freshCheckpoint(defListEnabled, mathFlow, referenceTaint) {
   };
 }
 function isPlausibleLinkDefRest(rest) {
-  const t = rest.trim();
+  const t = mdTrim(rest);
   if (t === "") return false;
   const destEnd = linkDestinationEnd(t);
   if (destEnd === -1) return false;
-  const after = t.slice(destEnd).trim();
+  const after = mdTrim(t.slice(destEnd));
   if (after === "") return true;
   const opener = after[0];
   if (opener !== '"' && opener !== "'" && opener !== "(") return false;
@@ -403,7 +425,7 @@ function isPlausibleLinkDefRest(rest) {
       i += 1;
       continue;
     }
-    if (after[i] === closer) return after.slice(i + 1).trim() === "";
+    if (after[i] === closer) return isMdBlank(after.slice(i + 1));
   }
   return false;
 }
@@ -440,6 +462,36 @@ function linkDestinationEnd(t) {
   if (balance !== 0 || i === 0) return -1;
   return i;
 }
+function inlineResourceEnd(text, openIdx) {
+  let i = openIdx + 1;
+  const skipWs = () => {
+    while (i < text.length && (text[i] === " " || text[i] === "	")) i += 1;
+  };
+  skipWs();
+  if (text[i] === ")") return i + 1;
+  const destEnd = linkDestinationEnd(text.slice(i));
+  if (destEnd === -1) return -1;
+  i += destEnd;
+  const beforeWs = i;
+  skipWs();
+  if (text[i] === ")") return i + 1;
+  if (i === beforeWs) return -1;
+  const opener = text[i];
+  if (opener !== '"' && opener !== "'" && opener !== "(") return -1;
+  const closer = opener === "(" ? ")" : opener;
+  for (i += 1; i < text.length; i++) {
+    if (text[i] === "\\") {
+      i += 1;
+      continue;
+    }
+    if (text[i] === closer) {
+      i += 1;
+      skipWs();
+      return text[i] === ")" ? i + 1 : -1;
+    }
+  }
+  return -1;
+}
 function classifyBlockStart(text, indent, defListEnabled) {
   if (indent >= 4) return true;
   if (LIST_MARKER_RE.test(text) || FOOTNOTE_DEF_RE.test(text)) return true;
@@ -457,12 +509,13 @@ function computeFreezeBoundary(text, options, resume) {
     let end = text.indexOf("\n", start);
     if (end === -1) end = text.length;
     const confirmed = end < text.length;
-    const lineText = text.slice(start, end);
+    const rawLine = text.slice(start, end);
+    const lineText = rawLine.endsWith("\r") ? rawLine.slice(0, -1) : rawLine;
     const ln = {
       start,
       end,
       text: lineText,
-      blank: confirmed && lineText.trim() === "",
+      blank: confirmed && isMdBlank(lineText),
       indent: computeIndent(lineText)
     };
     if (!confirmed) {
@@ -551,7 +604,7 @@ function processConfirmedLine(cp, ln, text) {
   const isBlockStart = cp.prevLineBlank;
   if (cp.htmlSeamPending && !ln.blank && !cp.htmlFlowSinceBlank && !(cp.commentOpen || cp.piOpen || cp.declOpen || cp.cdataOpen)) {
     const defShapedLine = DEF_RE.test(ln.text) || FOOTNOTE_DEF_RE.test(ln.text);
-    const commentOnly = ln.text.replace(/<!--[\s\S]*?-->/g, " ").replace(/<!--[\s\S]*$/, " ").trim() === "";
+    const commentOnly = ln.text.replace(/<!--[\s\S]*?-->/g, " ").replace(/<!--[\s\S]*$/, " ").replace(/[ \t\r]/g, "") === "";
     if (!defShapedLine && !commentOnly) {
       cp.htmlSeamPending = false;
     }
@@ -572,13 +625,14 @@ function processConfirmedLine(cp, ln, text) {
   const rawOpenAtLineStart = cp.commentOpen || cp.piOpen || cp.declOpen || cp.cdataOpen;
   if (cp.inFence) {
     const close = FENCE_RE.exec(ln.text);
-    if (close && close[1][0] === cp.fenceChar && close[1].length >= cp.fenceLen && ln.text.trim() === close[1]) {
+    if (close && close[1][0] === cp.fenceChar && close[1].length >= cp.fenceLen && isMdBlank(ln.text.slice(close[0].length))) {
       cp.inFence = false;
       cp.fenceChar = "";
       cp.fenceLen = 0;
     }
     cp.blankRun = 0;
     cp.paragraphHasUnpairedRun = false;
+    cp.openBracket = null;
     cp.prevLineBlank = false;
     cp.prevLineWasText = false;
     cp.prevLineWasValidDef = false;
@@ -600,6 +654,7 @@ function processConfirmedLine(cp, ln, text) {
       cp.openIndent = ln.indent;
       cp.blankRun = 0;
       cp.paragraphHasUnpairedRun = false;
+      cp.openBracket = null;
       cp.prevLineBlank = false;
       cp.prevLineWasText = false;
       cp.prevLineWasValidDef = false;
@@ -608,12 +663,13 @@ function processConfirmedLine(cp, ln, text) {
   }
   if (cp.inMath) {
     const close = MATH_RUN_RE.exec(ln.text);
-    if (close && close[1].length >= cp.mathFenceLen && ln.text.trim() === close[1]) {
+    if (close && close[1].length >= cp.mathFenceLen && isMdBlank(ln.text.slice(close[0].length))) {
       cp.inMath = false;
       cp.mathFenceLen = 0;
     }
     cp.blankRun = 0;
     cp.paragraphHasUnpairedRun = false;
+    cp.openBracket = null;
     cp.prevLineBlank = false;
     cp.prevLineWasText = false;
     cp.prevLineWasValidDef = false;
@@ -635,6 +691,7 @@ function processConfirmedLine(cp, ln, text) {
         cp.openIndent = ln.indent;
         cp.blankRun = 0;
         cp.paragraphHasUnpairedRun = false;
+        cp.openBracket = null;
         cp.prevLineBlank = false;
         cp.prevLineWasText = false;
         cp.prevLineWasValidDef = false;
@@ -658,6 +715,7 @@ function processConfirmedLine(cp, ln, text) {
       defListSettled: null
     });
     cp.paragraphHasUnpairedRun = false;
+    cp.openBracket = null;
     cp.htmlFlowSinceBlank = false;
     cp.prevLineBlank = true;
     cp.prevLineWasText = false;
@@ -670,13 +728,13 @@ function processConfirmedLine(cp, ln, text) {
   } else if (LIST_MARKER_RE.test(ln.text) || FOOTNOTE_DEF_RE.test(ln.text) || cp.defListEnabled && DEF_LIST_DD_RE.test(ln.text) || ln.indent >= 4) {
     cp.hazardVerdict = true;
   }
-  const tagStart = ln.indent <= 3 ? /^<\/?([A-Za-z][A-Za-z0-9-]*)/.exec(ln.text.trimStart()) : null;
+  const tagStart = ln.indent <= 3 ? /^<\/?([A-Za-z][A-Za-z0-9-]*)/.exec(mdTrimStart(ln.text)) : null;
   if (tagStart) {
     cp.htmlFlowSinceBlank = true;
     if (!TYPE6_NAMES.has(tagStart[1].toLowerCase())) cp.hazardVerdict = true;
   }
   const inRawText = cp.htmlFlowSinceBlank || rawOpenAtLineStart;
-  const rawFlowStart = ln.indent <= 3 && /^<(?:!--|\?|![A-Za-z]|!\[CDATA\[)/.test(ln.text.trimStart());
+  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);
   if (unpaired) cp.paragraphHasUnpairedRun = true;
   const scanText = masked ?? ln.text;
@@ -694,28 +752,50 @@ function processConfirmedLine(cp, ln, text) {
       if (key && !cp.defs.has(key)) cp.defs.set(key, ln.end);
     }
   }
-  if (cp.referenceTaint && scanText.includes("[")) {
-    const defBracket = validDef ? def.index + def[0].indexOf("[") : -1;
-    REF_RE.lastIndex = 0;
-    let m;
-    while ((m = REF_RE.exec(scanText)) !== null) {
-      const follow = scanText[m.index + m[0].length];
-      if (follow === "(") continue;
-      if (follow === ":" && m.index === defBracket) continue;
-      const inner = m[1];
+  if (cp.referenceTaint) {
+    const pushRef = (offset, inner, followAt) => {
+      const follow = scanText[followAt];
+      if (follow === "(" && inlineResourceEnd(scanText, followAt) !== -1) return;
       let label;
       let footnote = false;
       if (inner.startsWith("^")) {
         footnote = true;
         label = normalizeLabel(inner.slice(1));
       } else if (follow === "[") {
-        const explicit = /^\[((?:[^[\]\\]|\\.)*)\]/.exec(scanText.slice(m.index + m[0].length));
+        const explicit = /^\[((?:[^[\]\\]|\\.)*)\]/.exec(scanText.slice(followAt));
         label = normalizeLabel(explicit && explicit[1] ? explicit[1] : inner);
       } else {
         label = normalizeLabel(inner);
       }
-      if (!label) continue;
-      cp.unresolvedRefs.push({ offset: ln.start + m.index, label, footnote });
+      if (label) cp.unresolvedRefs.push({ offset, label, footnote });
+    };
+    const pending = cp.openBracket;
+    cp.openBracket = null;
+    if (pending) {
+      const close = firstUnescaped(scanText, "]");
+      const open = firstUnescaped(scanText, "[");
+      const cont = (t) => t.replace(/^ {0,3}>[ \t]?/, "");
+      if (close !== -1 && (open === -1 || close < open)) {
+        pushRef(pending.offset, `${pending.text}
+${cont(scanText.slice(0, close))}`, close + 1);
+      } else if (close === -1 && open === -1) {
+        cp.openBracket = { offset: pending.offset, text: `${pending.text}
+${cont(scanText)}` };
+      }
+    }
+    if (scanText.includes("[")) {
+      const defBracket = validDef ? def.index + def[0].indexOf("[") : -1;
+      REF_RE.lastIndex = 0;
+      let m;
+      while ((m = REF_RE.exec(scanText)) !== null) {
+        const followAt = m.index + m[0].length;
+        if (scanText[followAt] === ":" && m.index === defBracket) continue;
+        pushRef(ln.start + m.index, m[1], followAt);
+      }
+      const trailingOpen = lastUnclosedBracket(scanText);
+      if (trailingOpen !== -1) {
+        cp.openBracket = { offset: ln.start + trailingOpen, text: scanText.slice(trailingOpen + 1) };
+      }
     }
   }
   const rawSpans = [];
@@ -776,7 +856,7 @@ function processConfirmedLine(cp, ln, text) {
       if (scanText[pi + 2] === ">") {
         rawSpans.push([pi, pi + 3]);
         pos = pi + 3;
-        if (scanText.slice(0, pi).trim() !== "" || ln.indent > 3) poisonRawDivergence();
+        if (!isMdBlank(scanText.slice(0, pi)) || ln.indent > 3) poisonRawDivergence();
         continue;
       }
       rawSpans.push([pi, pi + 2]);
@@ -827,7 +907,7 @@ function processConfirmedLine(cp, ln, text) {
       applyTag(tag, closing);
     }
     if (cp.commentOpen && lastCommentOpenerIdx !== -1 && !inRawText) {
-      if (tagText.slice(0, lastCommentOpenerIdx).trim() !== "") {
+      if (!isMdBlank(tagText.slice(0, lastCommentOpenerIdx))) {
         cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start + lastCommentOpenerIdx);
       }
     }
@@ -1053,13 +1133,14 @@ function rebaseDualWalk(node, segments, maxEnd, offsetDelta, lineDelta) {
   const position = node.position;
   if (position) {
     for (const point of [position.start, position.end]) {
+      if (!point) continue;
       const seg = point.offset !== void 0 && point.offset <= maxEnd ? segments.find((s) => point.offset >= s.injStart && point.offset <= s.injEnd) : void 0;
       if (seg) {
         point.offset += seg.offsetDelta;
-        point.line += seg.lineDelta;
+        if (typeof point.line === "number") point.line += seg.lineDelta;
       } else {
         if (point.offset !== void 0) point.offset += offsetDelta;
-        point.line += lineDelta;
+        if (typeof point.line === "number") point.line += lineDelta;
       }
     }
   }
@@ -1226,6 +1307,9 @@ function alignPrefixCut(prefixMdast, cutRegion, tailWrapVisible) {
       nextIdx = pairIdx + 1 + stripped;
       const candidate = visibles[nextIdx];
       if (!candidate) return null;
+      if (start === void 0 && sepBuffer.some((sep) => sep.type !== "text" || sep.value !== "\n")) {
+        return null;
+      }
       if (start !== void 0) {
         const cStart = candidate.position?.start?.offset;
         const cEnd = candidate.position?.end?.offset;
@@ -1242,6 +1326,9 @@ function alignPrefixCut(prefixMdast, cutRegion, tailWrapVisible) {
   const trailingGaps = sawContent ? trailingStripped : Math.max(0, visibles.length - 1);
   const seam = visibles.length > 0 && tailWrapVisible ? 1 : 0;
   const last = out[out.length - 1];
+  if (last !== void 0 && last.type === "text" && last.position !== void 0 && last.value.trim() === "") {
+    return null;
+  }
   const lastIsLiteral = last !== void 0 && last.type === "text" && last.value.trim() !== "";
   const litOwnerEnd = pairIdx >= 0 ? visibles[pairIdx].position?.end?.offset : void 0;
   const litEnd = lastIsLiteral ? last.position?.end?.offset : void 0;
@@ -1279,8 +1366,16 @@ function alignPrefixCut(prefixMdast, cutRegion, tailWrapVisible) {
     const v = lastPaired.value;
     const lastLt = v.lastIndexOf("<");
     if (lastLt !== -1 && /^<[!?]/.test(v.slice(lastLt))) return null;
+    const lastOut = out[out.length - 1];
+    const outEnd = lastOut?.type === "element" ? lastOut.position?.end?.offset : void 0;
+    const blockEnd = lastPaired.position?.end?.offset;
+    if (outEnd !== void 0 && blockEnd !== void 0 && outEnd < blockEnd) return null;
   }
   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;
+  }
   for (let i = 0; i < trailingGaps + seam; i++) {
     out.push({ type: "text", value: "\n" });
   }
@@ -1875,6 +1970,7 @@ function createRegistry(onEmpty) {
       }
     },
     contributeLabels(symbol, footnotes, links) {
+      if (!this.chunkOrder.includes(symbol)) return;
       const data = this.chunkData.get(symbol);
       if (data) {
         data.ownFootnoteLabels = footnotes;
@@ -1899,6 +1995,7 @@ function createRegistry(onEmpty) {
       this._notify();
     },
     contributeChunkData(symbol, data) {
+      if (!this.chunkOrder.includes(symbol)) return;
       this.chunkData.set(symbol, data);
       this.labelSet.footnoteLabels = /* @__PURE__ */ new Set();
       this.labelSet.linkLabels = /* @__PURE__ */ new Set();
@@ -2004,14 +2101,16 @@ var import_rehype_unwrap_images = __toESM(require("rehype-unwrap-images"), 1);
 var import_unist_util_visit4 = require("unist-util-visit");
 var IMAGE_TAGS = /* @__PURE__ */ new Set(["img", "cross-chunk-image"]);
 var LINK_TAGS = /* @__PURE__ */ new Set(["a", "cross-chunk-link"]);
-function applicable(node, inLink) {
+function applicable(node, inLink, seen) {
   let image = 0 /* Unknown */;
   for (const child of node.children) {
     if (child.type === "text" && /^\s*$/.test(child.value)) continue;
     if (child.type === "element" && IMAGE_TAGS.has(child.tagName)) {
+      if (child.tagName !== "img") seen.placeholder = true;
       image = 1 /* ContainsImage */;
     } else if (!inLink && child.type === "element" && LINK_TAGS.has(child.tagName)) {
-      const inner = applicable(child, true);
+      if (child.tagName !== "a") seen.placeholder = true;
+      const inner = applicable(child, true, seen);
       if (inner === 2 /* ContainsOther */) return 2 /* ContainsOther */;
       if (inner === 1 /* ContainsImage */) image = 1 /* ContainsImage */;
     } else {
@@ -2023,9 +2122,9 @@ function applicable(node, inLink) {
 function rehypeUnwrapCrossChunkImages() {
   return function transform(tree) {
     (0, import_unist_util_visit4.visit)(tree, "element", (node, index, parent) => {
-      if (node.tagName === "p" && parent && typeof index === "number" && applicable(node, false) === 1 /* ContainsImage */ && // Only paragraphs that actually hold a placeholder — plain <img>
-      // paragraphs were already unwrapped by rehype-unwrap-images.
-      JSON.stringify(node.children).includes('"cross-chunk-image"')) {
+      if (node.tagName !== "p" || !parent || typeof index !== "number") return;
+      const seen = { placeholder: false };
+      if (applicable(node, false, seen) === 1 /* ContainsImage */ && seen.placeholder) {
         parent.children.splice(index, 1, ...node.children);
         return [import_unist_util_visit4.SKIP, index];
       }
@@ -3513,14 +3612,12 @@ function isProtocolAllowed(url, allowed) {
 }
 function sanitizeCrossChunkUrl(rawUrl, key, tagName, urlTransform, schema) {
   rawUrl = (0, import_micromark_util_sanitize_uri2.normalizeUri)(rawUrl);
-  const transformed = urlTransform(rawUrl, key, fakeElement(tagName, key, rawUrl));
-  if (transformed == null) return "";
-  const stringUrl = String(transformed);
-  if (stringUrl === "") return "";
   const callerProtocols = schema.protocols;
   const allowed = callerProtocols === void 0 || callerProtocols === null ? sanitizeSchema.protocols?.[key] : callerProtocols[key];
-  if (!allowed || allowed.length === 0) return stringUrl;
-  return isProtocolAllowed(stringUrl, allowed) ? stringUrl : "";
+  if (allowed && allowed.length > 0 && !isProtocolAllowed(rawUrl, allowed)) return null;
+  const transformed = urlTransform(rawUrl, key, fakeElement(tagName, key, rawUrl));
+  if (transformed == null) return null;
+  return String(transformed);
 }
 
 // src/plugins/defs.ts
@@ -3900,6 +3997,20 @@ var LITERAL_CONTENT_CLOSE_REGEX = {
   math: /<\/math\s*>/gi,
   svg: /<\/svg\s*>/gi
 };
+function restOfLineIsBlank(content, pos) {
+  for (let i = pos; i < content.length; i++) {
+    const c = content[i];
+    if (c === "\n") return true;
+    if (c !== " " && c !== "	" && c !== "\r") return false;
+  }
+  return true;
+}
+function lineHasBacktick(content, pos) {
+  for (let i = pos; i < content.length && content[i] !== "\n"; i++) {
+    if (content[i] === "`") return true;
+  }
+  return false;
+}
 function isAtLineStart(content, pos) {
   let i = pos - 1;
   let spaces = 0;
@@ -3942,7 +4053,7 @@ function splitByProtectedRegions(content) {
     if (multilineStart !== -1) {
       if (char === multilineFenceMarker) {
         const runLen = getRepeatedMarkerLength(content, i, multilineFenceMarker);
-        if (runLen >= multilineFenceLength && isAtLineStart(content, i)) {
+        if (runLen >= multilineFenceLength && isAtLineStart(content, i) && restOfLineIsBlank(content, i + runLen)) {
           pushProtected(multilineStart, i + runLen);
           multilineStart = -1;
           multilineFenceMarker = null;
@@ -3958,7 +4069,7 @@ function splitByProtectedRegions(content) {
     }
     if (char === "`" || char === "~") {
       const runLen = getRepeatedMarkerLength(content, i, char);
-      if (runLen >= 3 && isAtLineStart(content, i)) {
+      if (runLen >= 3 && isAtLineStart(content, i) && !(char === "`" && lineHasBacktick(content, i + runLen))) {
         multilineStart = i;
         multilineFenceMarker = char;
         multilineFenceLength = runLen;