@ai-react-markdown/engine 2.8.2 → 2.9.1

package/dist/index.js

@@ -289,9 +289,9 @@ function inlineResourceEnd(text, openIdx) {
   }
   return -1;
 }
-function collectRefLine(cp, lnStart, lnEnd, scanText, inRawText, isBlockStart) {
-  const defShaped = inRawText ? null : DEF_RE.exec(scanText);
-  const def = defShaped !== null && (defShaped[1].startsWith("^") || isPlausibleLinkDefRest(scanText.slice(defShaped.index + defShaped[0].length))) ? defShaped : null;
+function collectRefLine(cp, lnStart, lnEnd, scanText, rawText, inRawText, isBlockStart) {
+  const defShaped = inRawText ? null : DEF_RE.exec(rawText);
+  const def = defShaped !== null && (defShaped[1].startsWith("^") || isPlausibleLinkDefRest(rawText.slice(defShaped.index + defShaped[0].length))) ? defShaped : null;
   const defLineStart = isBlockStart || !cp.prevLineWasText || cp.prevLineWasValidDef;
   const validDef = def !== null && defLineStart;
   if (validDef) {
@@ -370,16 +370,15 @@ function settleRefsAndEarliestUnresolved(cp) {
 var TYPE6_NAMES = new Set(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 NO_ELEMENT_NAMES = /* @__PURE__ */ new Set([...DOCUMENT_STRUCTURE_NAMES, "frame", "image"]);
 var RETROACTIVE_OPENERS = [
   "<!doctype",
-  "<html",
-  "<head",
-  "<body",
-  "<frameset",
-  "</html",
-  "</head",
-  "</body",
-  "</frameset"
+  // DERIVED, so the trailing-partial guard cannot drift from the poison the
+  // confirmed lines get. It was transcribed until F28, and the transcription
+  // was already two names behind by then. `<frame` subsumes `<frameset`,
+  // which is why the derived list is SHORTER than the one it replaces and
+  // still covers strictly more.
+  ...[...NO_ELEMENT_NAMES].flatMap((name) => [`<${name}`, `</${name}`])
 ];
 function tailCarriesRetroactive(text) {
   const lower = text.toLowerCase();
@@ -509,12 +508,15 @@ var rawTextElement = (t) => t.kind === "rawText" ? t.element : t.kind === "scrip
 var VOID_TAGS = /* @__PURE__ */ new Set([
   "area",
   "base",
+  "basefont",
+  "bgsound",
   "br",
   "col",
   "embed",
   "hr",
   "img",
   "input",
+  "keygen",
   "link",
   "meta",
   "param",
@@ -529,7 +531,6 @@ var THEMATIC_BREAK_RE = /^(?:(?:-[ \t]*){3,}|(?:\*[ \t]*){3,}|(?:_[ \t]*){3,})$/
 var BARE_MARKER_RE = /^(?:[-*+]|\d{1,9}[.)])[ \t]*$/;
 var SETEXT_LEFTOVER_RE = /^(?:=+|--)[ \t]*$/;
 var DEF_LIST_DD_RE = /^ {0,3}:[ \t]/;
-var CLOSE_TAG_ONLY_RE = /^[ \t]*<\/[A-Za-z][A-Za-z0-9-]*[ \t]*>[ \t\r]*$/;
 var FENCE_RE = /^ {0,3}(`{3,}|~{3,})/;
 var MATH_RUN_RE = /^ {0,3}(\$\$+)/;
 var TAG_OR_COMMENT_RE = /<(\/?)([A-Za-z][A-Za-z0-9-]*)(?=[\s/>])([^>]*)>|<!--|-->|--!>/g;
@@ -629,6 +630,7 @@ function freshCheckpoint(defListEnabled, mathFlow, referenceTaint) {
     tagBalance: /* @__PURE__ */ new Map(),
     openTotal: 0,
     p5Tok: { kind: "data" },
+    formPointerMaybeSet: false,
     openStack: [],
     mdBlock: { kind: "none" },
     blankRun: 0,
@@ -766,6 +768,19 @@ function floatingResidue(text, commentOpenAtStart) {
   if (!open) out += text.slice(last);
   return out;
 }
+var RAW_CONSTRUCT_START_RE = /^<(?:!--|\?|![A-Za-z]|!\[CDATA\[)/;
+var sealResumableAbove = (cp) => cp.defBlockMaybeOpen || cp.fnDefResumable || cp.containerMaybeOpen;
+function sealReleaseDerived(cp, ln, isBlockStart) {
+  if (sealResumableAbove(cp) && !(isBlockStart && ln.indent <= 3)) return false;
+  if (cp.prevLineOpenContent) return false;
+  if (cp.tableMaybeOpen) return false;
+  const head = mdTrimStart(ln.text);
+  if (RAW_CONSTRUCT_START_RE.test(head) || TYPE1_START_RE.test(head) || isType7Line(head)) return false;
+  const t6 = TYPE6_START_RE.exec(head);
+  if (t6 !== null && TYPE6_NAMES.has(t6[1].toLowerCase())) return false;
+  if (DEF_RE.test(ln.text) || FOOTNOTE_DEF_RE.test(ln.text)) return false;
+  return true;
+}
 function processConfirmedLine(cp, ln, text) {
   const newest = cp.candidates[cp.candidates.length - 1];
   if (newest && newest.defListSettled === null) {
@@ -773,16 +788,22 @@ function processConfirmedLine(cp, ln, text) {
   }
   const isBlockStart = cp.prevLineBlank;
   if (cp.p5SealPending && !ln.blank && cp.mdBlock.kind !== "html" && !(cp.p5Tok.kind === "comment" || cp.p5Tok.kind === "bogus")) {
-    const defShapedLine = DEF_RE.test(ln.text) || FOOTNOTE_DEF_RE.test(ln.text) || cp.defBlockMaybeOpen || cp.fnDefResumable && !(isBlockStart && ln.indent <= 3);
-    const commentOnly = ln.text.replace(/<!--[\s\S]*?-->/g, " ").replace(/<!--[\s\S]*$/, " ").replace(/[ \t\r]/g, "") === "";
-    const closeTagOnly = CLOSE_TAG_ONLY_RE.test(ln.text);
-    if (!defShapedLine && !commentOnly && !closeTagOnly) {
+    if (sealReleaseDerived(cp, ln, isBlockStart)) {
       cp.p5SealPending = false;
+      if (false) {
+        sealReleaseEvaluations += 1;
+        if (!sealReleaseEnumerated(cp, ln, isBlockStart)) {
+          console.error(
+            `[ai-react-markdown] seal-release containment broken at offset ${ln.start}: the derived predicate released a line the retired enumeration withheld (${JSON.stringify(ln.text.slice(0, 80))}).`
+          );
+        }
+      }
     }
   }
   const possiblyInsideForeign = () => FOREIGN_ROOT_NAMES.some((name) => (cp.tagBalance.get(name) ?? 0) > 0);
   const honoursSelfClosing = (tag) => tag === "svg" || tag === "math";
   const foreignRawTextSwitchUnknowable = () => possiblyInsideForeign();
+  let discardedEndTag = false;
   const applyTag = (tag, closing) => {
     if (inRawTextTok(cp.p5Tok)) {
       if (cp.p5Tok.kind === "script" && cp.p5Tok.escaped && !closing && tag === "script") {
@@ -804,7 +825,7 @@ function processConfirmedLine(cp, ln, text) {
         if (tag === "plaintext") cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start);
       }
     }
-    if (DOCUMENT_STRUCTURE_NAMES.has(tag)) cp.phasePoisonedAt = 0;
+    if (NO_ELEMENT_NAMES.has(tag)) cp.phasePoisonedAt = 0;
     if (tag === "template" && !closing) cp.phasePoisonedAt = 0;
     if (closing) {
       let idx = -1;
@@ -815,14 +836,19 @@ function processConfirmedLine(cp, ln, text) {
         }
         if (SCOPE_BARRIER_NAMES.has(cp.openStack[i])) break;
       }
-      if (idx === -1) return;
+      if (idx === -1) {
+        discardedEndTag = true;
+        return;
+      }
       cp.openStack.splice(idx, 1);
+      if (tag === "form") cp.formPointerMaybeSet = false;
       const count = cp.tagBalance.get(tag) ?? 0;
       if (count > 0) {
         cp.tagBalance.set(tag, count - 1);
         cp.openTotal -= 1;
       }
     } else {
+      if (tag === "form") cp.formPointerMaybeSet = true;
       cp.openStack.push(tag);
       cp.tagBalance.set(tag, (cp.tagBalance.get(tag) ?? 0) + 1);
       cp.openTotal += 1;
@@ -934,6 +960,9 @@ function processConfirmedLine(cp, ln, text) {
         cp.p5Tok = { kind: "data" };
       }
     }
+    if (cp.p5Tok.kind === "comment" && !mdHtml(cp.mdBlock, 2)) {
+      cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start);
+    }
     if (cp.mdBlock.kind === "html" && cp.mdBlock.type >= 6) cp.mdBlock = { kind: "none" };
     cp.blankRun += 1;
     cp.lastBlankStart = ln.start;
@@ -946,7 +975,11 @@ function processConfirmedLine(cp, ln, text) {
       // mask suppresses them — which is exactly why `openTotal` reads 0
       // and the candidate looked safe), and the 2-5 interiors are the
       // same construct to both grammars.
-      htmlBalanced: cp.openTotal === 0 && cp.mdBlock.kind !== "html" && cp.p5Tok.kind !== "bogus",
+      htmlBalanced: cp.openTotal === 0 && cp.mdBlock.kind !== "html" && cp.p5Tok.kind !== "bogus" && // A form pointer parse5 may still be holding makes a LATER `<form>`
+      // in the tail vanish from the full parse while a split parse opens
+      // it — forward independence fails with every other condition clean.
+      // See the field doc.
+      !cp.formPointerMaybeSet,
       hazard: cp.hazardVerdict,
       seamRisk: cp.p5SealPending,
       defListSettled: null
@@ -1001,14 +1034,14 @@ function processConfirmedLine(cp, ln, text) {
       }
     }
   }
-  const rawFlowStart = ln.indent <= 3 && /^<(?:!--|\?|![A-Za-z]|!\[CDATA\[)/.test(mdTrimStart(ln.text));
+  const rawFlowStart = ln.indent <= 3 && RAW_CONSTRUCT_START_RE.test(mdTrimStart(ln.text));
   const htmlOwnedLine = cp.mdBlock.kind === "html" || rawOpenAtLineStart || rawFlowStart;
   const maskingSuppressed = htmlOwnedLine || inRawTextTok(cp.p5Tok);
   const { masked, unpaired } = maskingSuppressed ? { masked: null, unpaired: false } : maskIntraLineCodeSpans(ln.text, cp.paragraphHasUnpairedRun);
   if (unpaired) cp.paragraphHasUnpairedRun = true;
   const scanText = masked ?? ln.text;
   const defRawToMicromark = cp.mdBlock.kind === "html" || rawOpenAtLineStart || inRawTextTok(cp.p5Tok);
-  const { validLinkDef } = collectRefLine(cp, ln.start, ln.end, scanText, defRawToMicromark, isBlockStart);
+  const { validLinkDef } = collectRefLine(cp, ln.start, ln.end, scanText, ln.text, defRawToMicromark, isBlockStart);
   const rawSpans = [];
   let pos = 0;
   const poisonRawDivergence = () => {
@@ -1283,7 +1316,7 @@ function processConfirmedLine(cp, ln, text) {
       cursor = to;
     }
     masked2 += scanText.slice(cursor);
-    if (floatingResidue(masked2, bothCommentsOpenAtLineStart).length > 0) {
+    if (floatingResidue(masked2, bothCommentsOpenAtLineStart).length > 0 || discardedEndTag) {
       cp.p5SealPending = true;
     }
   }
@@ -1336,6 +1369,21 @@ function processConfirmedLine(cp, ln, text) {
     }
   }
 }
+var SCANNER_NAME_LISTS = [
+  ["type1", TYPE1_NAMES],
+  ["rawText", RAW_TEXT_ELEMENTS],
+  ["type6", TYPE6_NAMES],
+  ["void", VOID_TAGS],
+  ["documentStructure", DOCUMENT_STRUCTURE_NAMES],
+  // Added with F28, and it is the list that would have made F28 visible: the
+  // derived census alphabet partitions names by their membership across THIS
+  // table, so before it existed `frame` shared one 39-name class with `div`
+  // and could never be sampled apart from it.
+  ["noElement", NO_ELEMENT_NAMES],
+  ["tablePart", TABLE_PART_NAMES],
+  ["scopeBarrier", SCOPE_BARRIER_NAMES],
+  ["foreignRoot", new Set(FOREIGN_ROOT_NAMES)]
+];
 
 // src/components/incrementalParse/spliceParse.ts
 import { normalizeIdentifier as normalizeIdentifier2 } from "micromark-util-normalize-identifier";