@ai-react-markdown/engine 2.8.1 → 2.9.0

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) {
@@ -509,12 +509,15 @@ 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 +532,6 @@ 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;
@@ -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 +788,19 @@ 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 && ln.indent >= 4;
-    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) {
+        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") {
@@ -815,7 +833,10 @@ 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);
       const count = cp.tagBalance.get(tag) ?? 0;
       if (count > 0) {
@@ -934,6 +955,9 @@ 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;
@@ -977,7 +1001,7 @@ function processConfirmedLine(cp, ln, text) {
   if (tagStart) {
     const noRealBlockOpen = cp.mdBlock.kind !== "html";
     const t1 = noRealBlockOpen ? TYPE1_START_RE.exec(mdTrimStart(ln.text)) : null;
-    if (t1) cp.mdBlock = { kind: "html", type: 1, raw: RAW_TEXT_ELEMENTS.has(t1[1].toLowerCase()) };
+    if (t1) cp.mdBlock = { kind: "html", type: 1, raw: RAW_TEXT_ELEMENTS.has(t1[1].toLowerCase()), indent: ln.indent };
     if (cp.mdBlock.kind !== "html") {
       const t = mdTrimStart(ln.text);
       const t6 = TYPE6_START_RE.exec(t);
@@ -992,7 +1016,7 @@ function processConfirmedLine(cp, ln, text) {
       // raw-text names (`</style>` alone is type 7, measured) and
       // quoted-`>` attribute values.
       !cp.prevLineOpenContent && type7Shaped) {
-        if (cp.mdBlock.kind === "none") cp.mdBlock = { kind: "html", type: realT6 ? 6 : 7 };
+        if (cp.mdBlock.kind === "none") cp.mdBlock = { kind: "html", type: realT6 ? 6 : 7, indent: ln.indent };
       } else if (cp.prevLineOpenContent && cp.tableMaybeOpen && type7Shaped) {
         cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start);
       }
@@ -1001,14 +1025,14 @@ 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 = () => {
@@ -1084,7 +1108,7 @@ function processConfirmedLine(cp, ln, text) {
       pos = bogus + 2;
     } else if (first === cd) {
       rawSpans.push([cd, cd + 9]);
-      if (cp.mdBlock.kind === "none") cp.mdBlock = { kind: "html", type: 5 };
+      if (cp.mdBlock.kind === "none") cp.mdBlock = { kind: "html", type: 5, indent: ln.indent };
       if (cp.p5Tok.kind === "data") cp.p5Tok = { kind: "bogus" };
       if (!isMdBlank(scanText.slice(0, cd)) || ln.indent > 3) inlineRawOpenerIdx = cd;
       pos = cd + 9;
@@ -1096,14 +1120,14 @@ function processConfirmedLine(cp, ln, text) {
         continue;
       }
       rawSpans.push([pi, pi + 2]);
-      if (cp.mdBlock.kind === "none") cp.mdBlock = { kind: "html", type: 3 };
+      if (cp.mdBlock.kind === "none") cp.mdBlock = { kind: "html", type: 3, indent: ln.indent };
       if (cp.p5Tok.kind === "data") cp.p5Tok = { kind: "bogus" };
       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;
-      if (cp.mdBlock.kind === "none") cp.mdBlock = { kind: "html", type: 4 };
+      if (cp.mdBlock.kind === "none") cp.mdBlock = { kind: "html", type: 4, indent: ln.indent };
       if (cp.p5Tok.kind === "data") cp.p5Tok = { kind: "bogus" };
       if (!isMdBlank(scanText.slice(0, decl)) || ln.indent > 3) inlineRawOpenerIdx = decl;
       pos = decl + 2;
@@ -1162,7 +1186,7 @@ function processConfirmedLine(cp, ln, text) {
         if (next.startsWith(">") || next === "->") {
           continue;
         }
-        if (cp.mdBlock.kind === "none") cp.mdBlock = { kind: "html", type: 2 };
+        if (cp.mdBlock.kind === "none") cp.mdBlock = { kind: "html", type: 2, indent: ln.indent };
         if (cp.p5Tok.kind === "data") cp.p5Tok = { kind: "comment" };
         lastCommentOpenerIdx = m.index;
         continue;
@@ -1283,7 +1307,7 @@ 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;
     }
   }
@@ -1322,7 +1346,30 @@ function processConfirmedLine(cp, ln, text) {
   cp.defBlockMaybeOpen = cp.defBlockMaybeOpen || DEF_RE.test(ln.text) || FOOTNOTE_DEF_RE.test(ln.text);
   if (FOOTNOTE_DEF_RE.test(ln.text)) cp.fnDefResumable = true;
   else if (isBlockStart && ln.indent <= 3) cp.fnDefResumable = false;
+  const isBarrier = (name) => SCOPE_BARRIER_NAMES.has(name);
+  const topLevelHtmlBlock = cp.mdBlock.kind === "html" && cp.mdBlock.indent === 0;
+  const confirmedBarriers = cp.openStack.filter(isBarrier).length - cp.pendingTruncatedTags.filter(isBarrier).length;
+  if (!topLevelHtmlBlock && confirmedBarriers > 0) {
+    cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start);
+  }
+  const maskUnbacked = mdType1RawText(cp.mdBlock) && !inRawTextTok(cp.p5Tok);
+  if (maskUnbacked) {
+    cp.phasePoisonedAt = Math.min(cp.phasePoisonedAt, ln.start);
+    if (tailCarriesRetroactive(ln.text) || /<template(?![a-z0-9-])/i.test(ln.text)) {
+      cp.phasePoisonedAt = 0;
+    }
+  }
 }
+var SCANNER_NAME_LISTS = [
+  ["type1", TYPE1_NAMES],
+  ["rawText", RAW_TEXT_ELEMENTS],
+  ["type6", TYPE6_NAMES],
+  ["void", VOID_TAGS],
+  ["documentStructure", DOCUMENT_STRUCTURE_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";