@ai-matrx/content-ir 0.19.13 → 0.19.15

package/dist/source.d.cts

@@ -467,6 +467,33 @@ declare function parseFenceOpener(line: string): {
 } | null;
 declare function classifyInnerFenceLine(trimmed: string, openTicks: number, nests: boolean, nestedDepth: number): InnerFenceLine;
 
+/**
+ * THE link-reference rule (CommonMark 4.7 / 6.3) for surfaces that read one
+ * piece of a document at a time — a table cell, a split chat block, a printed
+ * row — and so cannot see the document's `[label]: url "title"` definitions.
+ * `collectLinkDefinitions` reads them from the WHOLE text (never inside fenced
+ * code); `resolveReferenceLinks` rewrites a fragment's full (`[text][label]`),
+ * collapsed (`[label][]`) and shortcut (`[label]`) reference links into inline
+ * links (`[text](url "title")`) so any renderer shows what GFM shows. Code spans,
+ * images, inline links, footnote references and undefined labels are left as
+ * they are. verify-RC-B4 round 9 (R9-4).
+ *
+ * Pure TS, no dependencies.
+ */
+interface LinkDefinition {
+    url: string;
+    title: string | null;
+}
+/** CommonMark label matching: case-folded, inner whitespace collapsed, trimmed. */
+declare function normalizeLinkLabel(label: string): string;
+/** Every link reference definition in `text` (first one wins, as in CommonMark); fenced code is skipped. */
+declare function collectLinkDefinitions(text: string): Map<string, LinkDefinition>;
+/**
+ * `fragment` with every reference link whose label is defined rewritten inline.
+ * An image (`![alt][ref]`) resolves too. Nothing else changes.
+ */
+declare function resolveReferenceLinks(fragment: string, defs: ReadonlyMap<string, LinkDefinition>): string;
+
 interface CodeRange {
     /** UTF-16 offset of the first character (the fence line start, or the opening backtick run). */
     start: number;
@@ -697,4 +724,4 @@ declare function findTableStart(lines: readonly string[], from?: number): number
 /** Index just past the table whose header is at `start` (header, delimiter, then continuation rows). */
 declare function findTableEnd(lines: readonly string[], start: number): number;
 
-export { AFTER_WINDOW, BacktickRuns, type BlockIslandType, CALLOUT_NAMES, type CodePointIndex, type CodeRange, type DisturbedIsland, FENCE_OPENER_WHITESPACE, FENCE_WHITESPACE, type FenceEnd, type FenceLineKind, type FenceLines, type FenceOpener, FenceReader, type InlineIslandType, type InnerFenceLine, type IslandMeta, type IslandType, type MapOptions, type MappedPosition, type MappedRange, NESTING_FENCE_LANGUAGES, type ReadXmlTag, type SourceBlock, type SourceBlockKind, type SourceChange, type SourceEdit, type SourceIsland, type SourceSpan, SourceSpliceError, type SourceSpliceErrorCode, type SpliceIntegrity, type SpliceOptions, type SpliceResult, type TokenizeOptions, XmlContainerTracker, blockAt, blockEdit, blockRangeEdit, buildCodePointIndex, classifyInnerFenceLine, closeFence, codeSpanText, continuesTable, fenceLineKinds, fenceNestsInnerFences, fenceOpenerOf, fenceParts, findCodeRanges, findInlineCodeSpans, findTableEnd, findTableStart, isEscapedBracketMath, isGfmDelimiterRow, isHtmlBlockTagName, isMathPlaceholder, isPipeLedRow, isSingleDollarMath, islandEdit, joinSource, lineIndent, listIslands, looksLikeDisplayMath, looksLikeOpenInlineMath, mapCodeRanges, mapInlineCodeSpans, mapPosition, mapRange, opensTable, pairDisplayMath, parseFenceOpener, pendingDisplayMathOpener, readXmlTag, rowCells, singleDollarAfter, singleDollarMathEnd, spliceSave, splitRowSegments, splitsSurrogatePair, startsHtmlBlock, startsPipelessTable, tableContainerIndent, tableStartsAt, toCodePointOffset, toUtf16Offset, tokenizeSource, trimFenceLine, unescapeCellPipes };
+export { AFTER_WINDOW, BacktickRuns, type BlockIslandType, CALLOUT_NAMES, type CodePointIndex, type CodeRange, type DisturbedIsland, FENCE_OPENER_WHITESPACE, FENCE_WHITESPACE, type FenceEnd, type FenceLineKind, type FenceLines, type FenceOpener, FenceReader, type InlineIslandType, type InnerFenceLine, type IslandMeta, type IslandType, type LinkDefinition, type MapOptions, type MappedPosition, type MappedRange, NESTING_FENCE_LANGUAGES, type ReadXmlTag, type SourceBlock, type SourceBlockKind, type SourceChange, type SourceEdit, type SourceIsland, type SourceSpan, SourceSpliceError, type SourceSpliceErrorCode, type SpliceIntegrity, type SpliceOptions, type SpliceResult, type TokenizeOptions, XmlContainerTracker, blockAt, blockEdit, blockRangeEdit, buildCodePointIndex, classifyInnerFenceLine, closeFence, codeSpanText, collectLinkDefinitions, continuesTable, fenceLineKinds, fenceNestsInnerFences, fenceOpenerOf, fenceParts, findCodeRanges, findInlineCodeSpans, findTableEnd, findTableStart, isEscapedBracketMath, isGfmDelimiterRow, isHtmlBlockTagName, isMathPlaceholder, isPipeLedRow, isSingleDollarMath, islandEdit, joinSource, lineIndent, listIslands, looksLikeDisplayMath, looksLikeOpenInlineMath, mapCodeRanges, mapInlineCodeSpans, mapPosition, mapRange, normalizeLinkLabel, opensTable, pairDisplayMath, parseFenceOpener, pendingDisplayMathOpener, readXmlTag, resolveReferenceLinks, rowCells, singleDollarAfter, singleDollarMathEnd, spliceSave, splitRowSegments, splitsSurrogatePair, startsHtmlBlock, startsPipelessTable, tableContainerIndent, tableStartsAt, toCodePointOffset, toUtf16Offset, tokenizeSource, trimFenceLine, unescapeCellPipes };

package/dist/source.d.ts

@@ -467,6 +467,33 @@ declare function parseFenceOpener(line: string): {
 } | null;
 declare function classifyInnerFenceLine(trimmed: string, openTicks: number, nests: boolean, nestedDepth: number): InnerFenceLine;
 
+/**
+ * THE link-reference rule (CommonMark 4.7 / 6.3) for surfaces that read one
+ * piece of a document at a time — a table cell, a split chat block, a printed
+ * row — and so cannot see the document's `[label]: url "title"` definitions.
+ * `collectLinkDefinitions` reads them from the WHOLE text (never inside fenced
+ * code); `resolveReferenceLinks` rewrites a fragment's full (`[text][label]`),
+ * collapsed (`[label][]`) and shortcut (`[label]`) reference links into inline
+ * links (`[text](url "title")`) so any renderer shows what GFM shows. Code spans,
+ * images, inline links, footnote references and undefined labels are left as
+ * they are. verify-RC-B4 round 9 (R9-4).
+ *
+ * Pure TS, no dependencies.
+ */
+interface LinkDefinition {
+    url: string;
+    title: string | null;
+}
+/** CommonMark label matching: case-folded, inner whitespace collapsed, trimmed. */
+declare function normalizeLinkLabel(label: string): string;
+/** Every link reference definition in `text` (first one wins, as in CommonMark); fenced code is skipped. */
+declare function collectLinkDefinitions(text: string): Map<string, LinkDefinition>;
+/**
+ * `fragment` with every reference link whose label is defined rewritten inline.
+ * An image (`![alt][ref]`) resolves too. Nothing else changes.
+ */
+declare function resolveReferenceLinks(fragment: string, defs: ReadonlyMap<string, LinkDefinition>): string;
+
 interface CodeRange {
     /** UTF-16 offset of the first character (the fence line start, or the opening backtick run). */
     start: number;
@@ -697,4 +724,4 @@ declare function findTableStart(lines: readonly string[], from?: number): number
 /** Index just past the table whose header is at `start` (header, delimiter, then continuation rows). */
 declare function findTableEnd(lines: readonly string[], start: number): number;
 
-export { AFTER_WINDOW, BacktickRuns, type BlockIslandType, CALLOUT_NAMES, type CodePointIndex, type CodeRange, type DisturbedIsland, FENCE_OPENER_WHITESPACE, FENCE_WHITESPACE, type FenceEnd, type FenceLineKind, type FenceLines, type FenceOpener, FenceReader, type InlineIslandType, type InnerFenceLine, type IslandMeta, type IslandType, type MapOptions, type MappedPosition, type MappedRange, NESTING_FENCE_LANGUAGES, type ReadXmlTag, type SourceBlock, type SourceBlockKind, type SourceChange, type SourceEdit, type SourceIsland, type SourceSpan, SourceSpliceError, type SourceSpliceErrorCode, type SpliceIntegrity, type SpliceOptions, type SpliceResult, type TokenizeOptions, XmlContainerTracker, blockAt, blockEdit, blockRangeEdit, buildCodePointIndex, classifyInnerFenceLine, closeFence, codeSpanText, continuesTable, fenceLineKinds, fenceNestsInnerFences, fenceOpenerOf, fenceParts, findCodeRanges, findInlineCodeSpans, findTableEnd, findTableStart, isEscapedBracketMath, isGfmDelimiterRow, isHtmlBlockTagName, isMathPlaceholder, isPipeLedRow, isSingleDollarMath, islandEdit, joinSource, lineIndent, listIslands, looksLikeDisplayMath, looksLikeOpenInlineMath, mapCodeRanges, mapInlineCodeSpans, mapPosition, mapRange, opensTable, pairDisplayMath, parseFenceOpener, pendingDisplayMathOpener, readXmlTag, rowCells, singleDollarAfter, singleDollarMathEnd, spliceSave, splitRowSegments, splitsSurrogatePair, startsHtmlBlock, startsPipelessTable, tableContainerIndent, tableStartsAt, toCodePointOffset, toUtf16Offset, tokenizeSource, trimFenceLine, unescapeCellPipes };
+export { AFTER_WINDOW, BacktickRuns, type BlockIslandType, CALLOUT_NAMES, type CodePointIndex, type CodeRange, type DisturbedIsland, FENCE_OPENER_WHITESPACE, FENCE_WHITESPACE, type FenceEnd, type FenceLineKind, type FenceLines, type FenceOpener, FenceReader, type InlineIslandType, type InnerFenceLine, type IslandMeta, type IslandType, type LinkDefinition, type MapOptions, type MappedPosition, type MappedRange, NESTING_FENCE_LANGUAGES, type ReadXmlTag, type SourceBlock, type SourceBlockKind, type SourceChange, type SourceEdit, type SourceIsland, type SourceSpan, SourceSpliceError, type SourceSpliceErrorCode, type SpliceIntegrity, type SpliceOptions, type SpliceResult, type TokenizeOptions, XmlContainerTracker, blockAt, blockEdit, blockRangeEdit, buildCodePointIndex, classifyInnerFenceLine, closeFence, codeSpanText, collectLinkDefinitions, continuesTable, fenceLineKinds, fenceNestsInnerFences, fenceOpenerOf, fenceParts, findCodeRanges, findInlineCodeSpans, findTableEnd, findTableStart, isEscapedBracketMath, isGfmDelimiterRow, isHtmlBlockTagName, isMathPlaceholder, isPipeLedRow, isSingleDollarMath, islandEdit, joinSource, lineIndent, listIslands, looksLikeDisplayMath, looksLikeOpenInlineMath, mapCodeRanges, mapInlineCodeSpans, mapPosition, mapRange, normalizeLinkLabel, opensTable, pairDisplayMath, parseFenceOpener, pendingDisplayMathOpener, readXmlTag, resolveReferenceLinks, rowCells, singleDollarAfter, singleDollarMathEnd, spliceSave, splitRowSegments, splitsSurrogatePair, startsHtmlBlock, startsPipelessTable, tableContainerIndent, tableStartsAt, toCodePointOffset, toUtf16Offset, tokenizeSource, trimFenceLine, unescapeCellPipes };

package/dist/source.js

@@ -2603,6 +2603,51 @@ function mapRange(changes, start, end, unit = "utf16") {
   return { start: newStart, end: newEnd, touched, collapsed: newEnd === newStart && end > start };
 }
 
+// source/link-references.ts
+function normalizeLinkLabel(label) {
+  return label.trim().replace(/\s+/g, " ").toLowerCase();
+}
+var DEFINITION = /^ {0,3}\[((?:[^\]\\]|\\.){1,999})\]:[ \t]*(?:<([^<>\n]*)>|(\S+))(?:[ \t]+(?:"((?:[^"\\]|\\.)*)"|'((?:[^'\\]|\\.)*)'|\(((?:[^()\\]|\\.)*)\)))?[ \t]*$/;
+function collectLinkDefinitions(text) {
+  const defs = /* @__PURE__ */ new Map();
+  if (!text.includes("]:")) return defs;
+  const lines = text.split("\n");
+  const kinds = fenceLineKinds(text);
+  lines.forEach((line, i) => {
+    if (kinds[i] !== "prose") return;
+    const m = DEFINITION.exec(line);
+    if (!m) return;
+    const label = m[1] ?? "";
+    if (label.startsWith("^")) return;
+    const key = normalizeLinkLabel(label);
+    if (!key || defs.has(key)) return;
+    defs.set(key, { url: m[2] ?? m[3] ?? "", title: m[4] ?? m[5] ?? m[6] ?? null });
+  });
+  return defs;
+}
+var REFERENCE = /(!?)\[((?:[^\]\\]|\\.)*)\](?:\[((?:[^\]\\]|\\.)*)\])?(?![(:[])/g;
+var escapeTitle = (title) => title.replace(/["\\]/g, "\\$&");
+function resolveReferenceLinks(fragment, defs) {
+  if (defs.size === 0 || !fragment.includes("[")) return fragment;
+  const PLACEHOLDER = "\uE011";
+  const spans = [];
+  const masked = mapInlineCodeSpans(fragment, (_range, raw) => {
+    spans.push(raw);
+    return `${PLACEHOLDER}${spans.length - 1}${PLACEHOLDER}`;
+  });
+  const resolved = masked.replace(REFERENCE, (whole, bang, text, ref, at) => {
+    if (at > 0 && masked[at - 1] === "\\") return whole;
+    if (text.startsWith("^")) return whole;
+    const key = normalizeLinkLabel(ref === void 0 || ref === "" ? text : ref);
+    const def = defs.get(key);
+    if (!def) return whole;
+    const title = def.title === null ? "" : ` "${escapeTitle(def.title)}"`;
+    const url = /[\s()<>]/.test(def.url) ? `<${def.url}>` : def.url;
+    return `${bang}[${text}](${url}${title})`;
+  });
+  return resolved.replace(new RegExp(`${PLACEHOLDER}(\\d+)${PLACEHOLDER}`, "g"), (_m, i) => spans[Number(i)] ?? "");
+}
+
 // source/gfm-table.ts
 function splitRowSegments(line) {
   const segments = [];
@@ -2631,7 +2676,7 @@ function rowCells(line) {
 function unescapeCellPipes(cell) {
   return cell.replace(/\\\|/g, "|");
 }
-var BLOCK_START = /^(?:[-+*](?:\s|$)|\d{1,9}[.)](?:\s|$)|>|#{1,6}(?:\s|$)|(?:-\s*){3,}$|(?:\*\s*){3,}$|(?:_\s*){3,}$)/;
+var BLOCK_START = /^(?:[-+*](?:\s|$)|\d{1,9}[.)](?:\s|$)|>|#{1,6}(?:\s|$)|(?:-\s*){3,}$|(?:\*\s*){3,}$|(?:_\s*){3,}$|\[\^[^\]\s]+\]:)/;
 var startsBlock = (trimmed) => BLOCK_START.test(trimmed) || opensStrictFence(trimmed) || startsHtmlBlock(trimmed);
 var HTML_RAW = /^<(?:script|pre|style|textarea)(?:[\s>]|$)/i;
 var HTML_COMMENT_ETC = /^(?:<!--|<\?|<![A-Za-z]|<!\[CDATA\[)/;
@@ -2721,11 +2766,45 @@ function continuesTable(line, containerIndent = 0) {
 function opensTable(lines, index) {
   const line = lines[index] ?? "";
   if (!isPipeLedRow(line) && !startsPipelessTable(line, lines[index + 1])) return false;
-  return !isLazyContinuation(lines, index);
+  return !isLazyContinuation(lines, index) && !insideHtmlBlock(lines, index);
 }
 var QUOTE_LINE = /^ {0,3}>/;
 var THEMATIC_BREAK = /^(?:(?:-[ \t]*){3,}|(?:\*[ \t]*){3,}|(?:_[ \t]*){3,})$/;
 var breaksParagraph = (trimmed) => /^#{1,6}(?:\s|$)/.test(trimmed) || THEMATIC_BREAK.test(trimmed) || opensStrictFence(trimmed) || startsHtmlBlock(trimmed);
+function htmlBlockEnd(trimmed) {
+  if (HTML_RAW.test(trimmed)) return /<\/(?:script|pre|style|textarea)>/i;
+  if (trimmed.startsWith("<!--")) return /-->/;
+  if (trimmed.startsWith("<?")) return /\?>/;
+  if (trimmed.startsWith("<![CDATA[")) return /\]\]>/;
+  if (/^<![A-Za-z]/.test(trimmed)) return />/;
+  if (HTML_BLOCK_TAG.test(trimmed)) return null;
+  if (HTML_COMPLETE_TAG.test(trimmed)) return null;
+  return void 0;
+}
+function insideHtmlBlock(lines, index) {
+  let start = index;
+  while (start > 0 && (lines[start - 1] ?? "").trim()) start -= 1;
+  let open;
+  let paragraph = false;
+  for (let k = start; k < index; k += 1) {
+    const line = lines[k] ?? "";
+    if (open !== void 0) {
+      if (open && open.test(line)) open = void 0;
+      continue;
+    }
+    const trimmed = line.trim();
+    const end = lineIndent(line) < 4 ? htmlBlockEnd(trimmed) : void 0;
+    const type7 = end === null && !HTML_BLOCK_TAG.test(trimmed);
+    const interrupts = end !== void 0 && !(paragraph && type7);
+    if (!interrupts) {
+      paragraph = true;
+      continue;
+    }
+    paragraph = false;
+    open = end && end.test(trimmed.slice(2)) ? void 0 : end;
+  }
+  return open !== void 0;
+}
 function isLazyContinuation(lines, index) {
   const header = lines[index] ?? "";
   let first = -1;
@@ -2767,6 +2846,6 @@ function findTableEnd(lines, start) {
   return end;
 }
 
-export { AFTER_WINDOW, BacktickRuns, CALLOUT_NAMES, FENCE_OPENER_WHITESPACE, FENCE_WHITESPACE, FenceReader, NESTING_FENCE_LANGUAGES, SourceSpliceError, XmlContainerTracker, blockAt, blockEdit, blockRangeEdit, buildCodePointIndex, classifyInnerFenceLine, closeFence, codeSpanText, continuesTable, fenceLineKinds, fenceNestsInnerFences, fenceOpenerOf, fenceParts, findCodeRanges, findInlineCodeSpans, findTableEnd, findTableStart, isEscapedBracketMath, isGfmDelimiterRow, isHtmlBlockTagName, isMathPlaceholder, isPipeLedRow, isSingleDollarMath, islandEdit, joinSource, lineIndent, listIslands, looksLikeDisplayMath, looksLikeOpenInlineMath, mapCodeRanges, mapInlineCodeSpans, mapPosition, mapRange, opensTable, pairDisplayMath, parseFenceOpener, pendingDisplayMathOpener, readXmlTag, rowCells, singleDollarAfter, singleDollarMathEnd, spliceSave, splitRowSegments, splitsSurrogatePair, startsHtmlBlock, startsPipelessTable, tableContainerIndent, tableStartsAt, toCodePointOffset, toUtf16Offset, tokenizeSource, trimFenceLine, unescapeCellPipes };
+export { AFTER_WINDOW, BacktickRuns, CALLOUT_NAMES, FENCE_OPENER_WHITESPACE, FENCE_WHITESPACE, FenceReader, NESTING_FENCE_LANGUAGES, SourceSpliceError, XmlContainerTracker, blockAt, blockEdit, blockRangeEdit, buildCodePointIndex, classifyInnerFenceLine, closeFence, codeSpanText, collectLinkDefinitions, continuesTable, fenceLineKinds, fenceNestsInnerFences, fenceOpenerOf, fenceParts, findCodeRanges, findInlineCodeSpans, findTableEnd, findTableStart, isEscapedBracketMath, isGfmDelimiterRow, isHtmlBlockTagName, isMathPlaceholder, isPipeLedRow, isSingleDollarMath, islandEdit, joinSource, lineIndent, listIslands, looksLikeDisplayMath, looksLikeOpenInlineMath, mapCodeRanges, mapInlineCodeSpans, mapPosition, mapRange, normalizeLinkLabel, opensTable, pairDisplayMath, parseFenceOpener, pendingDisplayMathOpener, readXmlTag, resolveReferenceLinks, rowCells, singleDollarAfter, singleDollarMathEnd, spliceSave, splitRowSegments, splitsSurrogatePair, startsHtmlBlock, startsPipelessTable, tableContainerIndent, tableStartsAt, toCodePointOffset, toUtf16Offset, tokenizeSource, trimFenceLine, unescapeCellPipes };
 //# sourceMappingURL=source.js.map
 //# sourceMappingURL=source.js.map