@ai-react-markdown/engine 2.4.0 → 2.4.1

package/dist/index.cjs

@@ -61,6 +61,7 @@ __export(src_exports, {
   extendSanitizeSchema: () => extendSanitizeSchema,
   extractContributions: () => extractContributions,
   extractDefBodiesFromHast: () => extractDefBodiesFromHast,
+  footnoteSafeId: () => footnoteSafeId,
   getEnginePluginInternals: () => getEnginePluginInternals,
   hasLoneSurrogate: () => hasLoneSurrogate,
   highlight: () => highlight,
@@ -717,7 +718,6 @@ function processConfirmedLine(cp, ln, text) {
       cp.unresolvedRefs.push({ offset: ln.start + m.index, label, footnote });
     }
   }
-  const rawOpenAtStart = cp.piOpen || cp.declOpen || cp.cdataOpen;
   const rawSpans = [];
   let pos = 0;
   const poisonRawDivergence = () => {
@@ -788,14 +788,17 @@ function processConfirmedLine(cp, ln, text) {
       pos = decl + 2;
     }
   }
-  const rawOpenAtEnd = cp.piOpen || cp.declOpen || cp.cdataOpen;
-  if (!rawOpenAtStart && !rawOpenAtEnd) {
+  let tagText = scanText;
+  for (const [from, to] of rawSpans) {
+    tagText = tagText.slice(0, from) + " ".repeat(to - from) + tagText.slice(to);
+  }
+  {
     TAG_OR_COMMENT_RE.lastIndex = 0;
     let m;
     let lastCommentOpenerIdx = -1;
-    while ((m = TAG_OR_COMMENT_RE.exec(scanText)) !== null) {
+    while ((m = TAG_OR_COMMENT_RE.exec(tagText)) !== null) {
       if (m[0] === "<!--") {
-        const next = scanText.slice(m.index + 4, m.index + 6);
+        const next = tagText.slice(m.index + 4, m.index + 6);
         if (cp.commentOpen) {
           if (next.startsWith(">") || next === "->") cp.commentOpen = false;
           else if (next === "!>" || next === "-!") poisonRawDivergence();
@@ -824,34 +827,56 @@ function processConfirmedLine(cp, ln, text) {
       applyTag(tag, closing);
     }
     if (cp.commentOpen && lastCommentOpenerIdx !== -1 && !inRawText) {
-      if (scanText.slice(0, lastCommentOpenerIdx).trim() !== "") {
+      if (tagText.slice(0, lastCommentOpenerIdx).trim() !== "") {
         cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start + lastCommentOpenerIdx);
       }
     }
-    if (cp.pendingTruncatedTags.length > 0 && scanText.includes(">")) {
+    if (masked !== null && masked !== ln.text) {
+      const inRaw = (i) => rawSpans.some(([from, to]) => i >= from && i < to);
+      TAG_OR_COMMENT_RE.lastIndex = 0;
+      let mr;
+      while ((mr = TAG_OR_COMMENT_RE.exec(ln.text)) !== null) {
+        if (mr[0] === "<!--" || mr[0] === "-->" || mr[0] === "--!>") continue;
+        const startMasked = masked[mr.index] !== ln.text[mr.index];
+        const wholeVisible = masked.slice(mr.index, mr.index + mr[0].length) === mr[0];
+        if (startMasked || wholeVisible || inRaw(mr.index) || cp.commentOpen) continue;
+        const closing = mr[1] === "/";
+        const tag = mr[2].toLowerCase();
+        const selfClosing = mr[3] !== void 0 && /\/\s*$/.test(mr[3]);
+        if (VOID_TAGS.has(tag) || selfClosing) continue;
+        applyTag(tag, closing);
+      }
+    }
+    if (cp.pendingTruncatedTags.length > 0 && ln.text.includes(">")) {
       cp.pendingTruncatedTags = [];
     }
     if (!cp.commentOpen) {
-      const lastLt = scanText.lastIndexOf("<");
-      if (lastLt !== -1 && !scanText.includes(">", lastLt)) {
-        const m2 = TRUNCATED_TAG_RE.exec(scanText.slice(lastLt));
+      const lastLt = tagText.lastIndexOf("<");
+      if (lastLt !== -1 && !tagText.includes(">", lastLt)) {
+        const m2 = TRUNCATED_TAG_RE.exec(tagText.slice(lastLt));
         if (m2) {
           const closing = m2[1] === "/";
           const tag = m2[2].toLowerCase();
           if (!VOID_TAGS.has(tag)) {
             applyTag(tag, closing);
-            if (!closing && !inRawText) cp.pendingTruncatedTags.push(tag);
+            const rawLastLt = ln.text.lastIndexOf("<");
+            const rawTruncated = rawLastLt !== -1 && !ln.text.includes(">", rawLastLt);
+            if (!closing && !inRawText && rawTruncated) cp.pendingTruncatedTags.push(tag);
           }
         }
       }
     }
   }
-  if ((inRawText || rawFlowStart) && cp.openTotal === 0) {
-    let masked2 = scanText;
+  const effectiveOpen = cp.openTotal - cp.pendingTruncatedTags.length;
+  if ((inRawText || rawFlowStart) && effectiveOpen <= 0) {
+    let masked2 = "";
+    let cursor = 0;
     for (const [from, to] of rawSpans) {
-      masked2 = masked2.slice(0, from) + " ".repeat(to - from) + masked2.slice(to);
+      masked2 += scanText.slice(cursor, from);
+      cursor = to;
     }
-    if (floatingResidue(masked2, commentOpenAtLineStart).trim() !== "") {
+    masked2 += scanText.slice(cursor);
+    if (floatingResidue(masked2, commentOpenAtLineStart).length > 0) {
       cp.htmlSeamPending = true;
     }
   }
@@ -914,7 +939,7 @@ function collectPrefixInjection(mdast, content, boundary, resume) {
     if (last && last.kind === "refs") last.tokens.push(token);
     else events.push({ kind: "refs", tokens: [token] });
   };
-  const visit6 = (node, nested) => {
+  const visit7 = (node, nested) => {
     const type = node.type;
     if (type === "definition") {
       const start = node.position?.start?.offset;
@@ -955,7 +980,7 @@ function collectPrefixInjection(mdast, content, boundary, resume) {
     }
     const children = node.children;
     if (children) {
-      for (const child of children) visit6(child, true);
+      for (const child of children) visit7(child, true);
     }
   };
   let lastStart = -1;
@@ -968,11 +993,11 @@ function collectPrefixInjection(mdast, content, boundary, resume) {
     if (start === void 0) {
       if (resumeAt > 0) return collectPrefixInjection(mdast, content, boundary, null);
       cacheable = false;
-      visit6(child, false);
+      visit7(child, false);
       continue;
     }
     if (start >= boundary || start < resumeAt) continue;
-    visit6(child, false);
+    visit7(child, false);
   }
   return { events, uninjectable, cacheable };
 }
@@ -985,7 +1010,7 @@ function cloneEventsForAppend(events) {
 var TERMINATOR_LABEL = "__aimd_injection_terminator__";
 var INJECTION_TERMINATOR = `[${TERMINATOR_LABEL}]: __aimd_sentinel_link__`;
 function tailMentionsTerminator(tailSource) {
-  return (0, import_micromark_util_normalize_identifier2.normalizeIdentifier)(tailSource).includes(`[${(0, import_micromark_util_normalize_identifier2.normalizeIdentifier)(TERMINATOR_LABEL)}`);
+  return (0, import_micromark_util_normalize_identifier2.normalizeIdentifier)(tailSource).replace(/\[ /g, "[").includes(`[${(0, import_micromark_util_normalize_identifier2.normalizeIdentifier)(TERMINATOR_LABEL)}`);
 }
 function buildInjectionPrefix(events) {
   if (events.length === 0) return { text: "", segments: [] };
@@ -1065,12 +1090,25 @@ function spliceTrees(input) {
       if (isTrailingLiteralText(node2)) {
         const prev = i > 0 ? prevHast.children[i - 1] : void 0;
         if (!prev || prev.type !== "element" || prev.position === void 0) return null;
+        if (!ownsTrailingLiteral(prev, node2, prefixMdast))
+          return null;
       }
       cutRegion.push(node2);
       continue;
     }
     const node = prevHast.children[i];
-    if (i > 0 && attrs[i - 1] < boundary && prevHast.children[i - 1].type === "element" && isTrailingLiteralText(node)) {
+    if (i > 0 && attrs[i - 1] < boundary && prevHast.children[i - 1].type === "element" && isTrailingLiteralText(node) && // …and only when that element really IS an html block's output: a
+    // raw literal can only trail an `html` mdast node. A position-less
+    // text after a `<p>` is the NEXT block's remnant — a stray end tag
+    // (`</t>\na`) parse5 dropped, whose text merged with the wrap
+    // separator — owned by the tail, which re-parses it; freezing it
+    // here duplicated it (v2.4.0 review P3). Falls through to the
+    // remnant look-ahead below, which bails to a full parse. And the
+    // literal must really be THAT block's trailing text: the block's raw
+    // source ends with it. A dropped-tag block right after a frozen html
+    // element (`</details>\n\n</t>\ntext`) puts its remnant in the same
+    // position, and freezing it duplicated it (release soak of the fix).
+    ownsTrailingLiteral(prevHast.children[i - 1], node, prefixMdast)) {
       cutRegion.push(node);
       break;
     }
@@ -1285,8 +1323,16 @@ function stripInjectedHast(tailMdast, tailHast, injectedLen, tailWrapVisible) {
 }
 function tailLeadingTextIsHoist(tailMdastChildren, tailHastChildren) {
   const firstText = tailHastChildren[0];
-  if (!firstText || !isSeparatorText(firstText)) return false;
+  if (!firstText) return false;
   const firstVisible = tailMdastChildren.find((c) => !isWrapInvisible(c));
+  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;
+      return null;
+    }
+    return false;
+  }
   if (!firstVisible) return false;
   const firstContent = tailHastChildren.find((c) => !isSeparatorText(c));
   if (firstContent) {
@@ -1308,6 +1354,14 @@ function isCompleteRawConstruct(value) {
 function isSeparatorText(node) {
   return node.type === "text" && node.position === void 0 && node.value.trim() === "";
 }
+function ownsTrailingLiteral(el, literal, prefixMdast) {
+  const start = el.position?.start?.offset;
+  if (start === void 0) return false;
+  const owner = prefixMdast.find((c) => c.type === "html" && c.position?.start?.offset === start);
+  if (!owner || owner.type !== "html") return false;
+  const text = literal.value.trim();
+  return text !== "" && owner.value.trimEnd().endsWith(text);
+}
 function isTrailingLiteralText(node) {
   return node.type === "text" && node.position === void 0 && node.value.trim() !== "";
 }
@@ -1542,7 +1596,11 @@ function createDefLabelScanner(parse = collectDefLabels) {
 
 // src/components/extractDefBodiesFromHast.ts
 var import_unist_util_visit2 = require("unist-util-visit");
+var import_micromark_util_sanitize_uri = require("micromark-util-sanitize-uri");
 var FN_LI_ID_RE = /(?:^|-)user-content-fn-(.+)$/;
+function footnoteSafeId(identifier) {
+  return (0, import_micromark_util_sanitize_uri.normalizeUri)(identifier.toLowerCase());
+}
 function sourceIdFromFootnoteLiId(idProp, clobberPrefix) {
   let raw = null;
   if (clobberPrefix !== void 0) {
@@ -1699,10 +1757,10 @@ function phantomSuffixCloser(content) {
   const confirmed = endsWithNewline ? content : content + "\n";
   const { checkpoint } = computeFreezeBoundary(confirmed, { defListEnabled: false, referenceTaint: false });
   if (checkpoint.phasePoisonedAt !== Infinity) return "";
+  if (checkpoint.openIndent !== 0) return "";
   const nl = endsWithNewline ? "" : "\n";
-  const indent = " ".repeat(checkpoint.openIndent);
-  if (checkpoint.inFence) return `${nl}${indent}${checkpoint.fenceChar.repeat(checkpoint.fenceLen)}`;
-  if (checkpoint.inMath) return `${nl}${indent}${"$".repeat(checkpoint.mathFenceLen)}`;
+  if (checkpoint.inFence) return `${nl}${checkpoint.fenceChar.repeat(checkpoint.fenceLen)}`;
+  if (checkpoint.inMath) return `${nl}${"$".repeat(checkpoint.mathFenceLen)}`;
   return "";
 }
 
@@ -1941,6 +1999,41 @@ function createRegistry(onEmpty) {
 var import_rehype_katex = __toESM(require("rehype-katex"), 1);
 var import_rehype_raw = __toESM(require("rehype-raw"), 1);
 var import_rehype_unwrap_images = __toESM(require("rehype-unwrap-images"), 1);
+
+// src/components/rehypeUnwrapCrossChunkImages.ts
+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) {
+  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)) {
+      image = 1 /* ContainsImage */;
+    } else if (!inLink && child.type === "element" && LINK_TAGS.has(child.tagName)) {
+      const inner = applicable(child, true);
+      if (inner === 2 /* ContainsOther */) return 2 /* ContainsOther */;
+      if (inner === 1 /* ContainsImage */) image = 1 /* ContainsImage */;
+    } else {
+      return 2 /* ContainsOther */;
+    }
+  }
+  return image;
+}
+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"')) {
+        parent.children.splice(index, 1, ...node.children);
+        return [import_unist_util_visit4.SKIP, index];
+      }
+    });
+  };
+}
+
+// src/components/pluginChain.ts
 var import_rehype_sanitize = __toESM(require("rehype-sanitize"), 1);
 var import_remark_breaks = __toESM(require("remark-breaks"), 1);
 var import_remark_cjk_friendly = __toESM(require("remark-cjk-friendly"), 1);
@@ -1956,13 +2049,13 @@ var import_remark_pangu = __toESM(require("remark-pangu"), 1);
 var import_remark_remove_comments = __toESM(require("remark-remove-comments"), 1);
 
 // src/components/rehypeRebaseHashLinks.ts
-var import_unist_util_visit4 = require("unist-util-visit");
+var import_unist_util_visit5 = require("unist-util-visit");
 var DEFAULT_PREFIX = "user-content-";
 var rehypeRebaseHashLinks = (options) => {
   const prefix = options?.prefix ?? DEFAULT_PREFIX;
   const hashPrefix = "#" + prefix;
   return (tree) => {
-    (0, import_unist_util_visit4.visit)(tree, "element", (node) => {
+    (0, import_unist_util_visit5.visit)(tree, "element", (node) => {
       if (node.tagName !== "a") return;
       const href = node.properties?.href;
       if (typeof href !== "string" || !href.startsWith("#")) return;
@@ -1974,7 +2067,7 @@ var rehypeRebaseHashLinks = (options) => {
 var rehypeRebaseHashLinks_default = rehypeRebaseHashLinks;
 
 // src/components/rehypeFooterAdorn.ts
-var import_unist_util_visit5 = require("unist-util-visit");
+var import_unist_util_visit6 = require("unist-util-visit");
 var FOOTNOTE_LABEL_ID_RE = /(?:^|-)footnote-label$/;
 function isFootnoteLabelH2(node) {
   if (node.type !== "element") return false;
@@ -1989,7 +2082,7 @@ function isHr(node) {
 }
 function rehypeFooterAdorn() {
   return (tree) => {
-    (0, import_unist_util_visit5.visit)(tree, "element", (n) => {
+    (0, import_unist_util_visit6.visit)(tree, "element", (n) => {
       const el = n;
       if (el.tagName !== "section") return;
       if (!(el.properties && "dataFootnotes" in el.properties)) return;
@@ -2067,7 +2160,10 @@ function buildCoreRehypePlugins(sanitizeSchema2, clobberPrefix) {
     // above.
     [rehypeRebaseHashLinks_default, { prefix: clobberPrefix }],
     import_rehype_katex.default,
-    import_rehype_unwrap_images.default
+    import_rehype_unwrap_images.default,
+    // Same unwrap for `<cross-chunk-image>` placeholders (coordinated mode);
+    // no-op on standalone documents.
+    rehypeUnwrapCrossChunkImages
   ];
 }
 function buildCoreRemarkRehypeOptions(enableDefinitionList) {
@@ -2186,6 +2282,9 @@ function buildCrossChunkHandlers() {
   };
 }
 
+// src/components/crossChunkUrlSanitize.ts
+var import_micromark_util_sanitize_uri2 = require("micromark-util-sanitize-uri");
+
 // ../../node_modules/.pnpm/lodash-es@4.18.1/node_modules/lodash-es/_listCacheClear.js
 function listCacheClear() {
   this.__data__ = [];
@@ -3413,6 +3512,7 @@ function isProtocolAllowed(url, allowed) {
   return allowed.some((p) => p === protocol);
 }
 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);
@@ -4275,6 +4375,7 @@ function createRemendPreprocessor(options) {
   extendSanitizeSchema,
   extractContributions,
   extractDefBodiesFromHast,
+  footnoteSafeId,
   getEnginePluginInternals,
   hasLoneSurrogate,
   highlight,