kordoc 3.11.0 → 3.12.0

package/dist/{parser-X34HD2KQ.cjs → parser-PCTX7LGN.cjs}

@@ -6,7 +6,7 @@
 
 
 
-var _chunkRCSDSN5Ycjs = require('./chunk-RCSDSN5Y.cjs');
+var _chunk2C5GCXPJcjs = require('./chunk-2C5GCXPJ.cjs');
 
 
 var _chunkDCZVOIEOcjs = require('./chunk-DCZVOIEO.cjs');
@@ -240,10 +240,40 @@ var EDGE_MIN_SPAN = 12;
 var EDGE_INSET = 15;
 var EDGE_NEAR = 10;
 var EDGE_CONNECT_TOL = 5;
+var CHAIN_Y_TOL = 1.5;
+var CHAIN_GAP = 3;
+function chainCollinearRules(horizontals) {
+  if (horizontals.length <= 1) return horizontals;
+  const sorted = [...horizontals].sort((a, b) => a.y1 - b.y1 || a.x1 - b.x1);
+  const rules = [];
+  let bandStart = 0;
+  const flushBand = (end) => {
+    const band = sorted.slice(bandStart, end).sort((a, b) => a.x1 - b.x1);
+    let cur = { ...band[0] };
+    for (let i = 1; i < band.length; i++) {
+      const seg = band[i];
+      if (seg.x1 - cur.x2 <= CHAIN_GAP) {
+        if (seg.x2 > cur.x2) cur.x2 = seg.x2;
+        if (seg.lineWidth > cur.lineWidth) cur.lineWidth = seg.lineWidth;
+      } else {
+        rules.push(cur);
+        cur = { ...seg };
+      }
+    }
+    rules.push(cur);
+  };
+  for (let i = 1; i <= sorted.length; i++) {
+    if (i === sorted.length || sorted[i].y1 - sorted[bandStart].y1 > CHAIN_Y_TOL) {
+      flushBand(i);
+      bandStart = i;
+    }
+  }
+  return rules;
+}
 function closeOpenTableEdges(horizontals, verticals) {
   if (horizontals.length < EDGE_MIN_RULES) return verticals;
   const groups = [];
-  for (const hl of horizontals) {
+  for (const hl of chainCollinearRules(horizontals)) {
     let placed = false;
     for (const g2 of groups) {
       if (Math.abs(g2[0].x1 - hl.x1) <= EDGE_ALIGN_TOL && Math.abs(g2[0].x2 - hl.x2) <= EDGE_ALIGN_TOL) {
@@ -2056,14 +2086,14 @@ function isProseSpread(items) {
   for (let i = 1; i < sorted.length; i++) {
     gaps.push(sorted[i].x - (sorted[i - 1].x + sorted[i - 1].w));
   }
-  const maxGap = _chunkRCSDSN5Ycjs.safeMax.call(void 0, gaps);
+  const maxGap = _chunk2C5GCXPJcjs.safeMax.call(void 0, gaps);
   const avgLen = items.reduce((s, i) => s + i.text.length, 0) / items.length;
   return maxGap < 40 && avgLen < 5;
 }
 function detectColumns(yLines) {
   const allItems = yLines.flat();
   if (allItems.length === 0) return null;
-  const pageWidth = _chunkRCSDSN5Ycjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkRCSDSN5Ycjs.safeMin.call(void 0, allItems.map((i) => i.x));
+  const pageWidth = _chunk2C5GCXPJcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunk2C5GCXPJcjs.safeMin.call(void 0, allItems.map((i) => i.x));
   if (pageWidth < 100) return null;
   let bigoLineIdx = -1;
   for (let i = 0; i < yLines.length; i++) {
@@ -2285,9 +2315,9 @@ function detectHeadings(blocks, medianFontSize) {
     if (/^\d+$/.test(text)) continue;
     const ratio = block.style.fontSize / medianFontSize;
     let level = 0;
-    if (ratio >= _chunkRCSDSN5Ycjs.HEADING_RATIO_H1) level = 1;
-    else if (ratio >= _chunkRCSDSN5Ycjs.HEADING_RATIO_H2) level = 2;
-    else if (ratio >= _chunkRCSDSN5Ycjs.HEADING_RATIO_H3) level = 3;
+    if (ratio >= _chunk2C5GCXPJcjs.HEADING_RATIO_H1) level = 1;
+    else if (ratio >= _chunk2C5GCXPJcjs.HEADING_RATIO_H2) level = 2;
+    else if (ratio >= _chunk2C5GCXPJcjs.HEADING_RATIO_H3) level = 3;
     if (level > 0) {
       block.type = "heading";
       block.level = level;
@@ -2298,6 +2328,10 @@ function detectHeadings(blocks, medianFontSize) {
 function shouldDemoteTable(table) {
   const allCells = table.cells.flatMap((row) => row.map((c) => c.text.trim())).filter(Boolean);
   const allText = allCells.join(" ");
+  if (table.rows >= 2 && table.cols >= 2 && table.cells[0].every((c) => {
+    const t = c.text.trim();
+    return t.length > 0 && t.length <= 12 && !/[□■◆○●▶ㅇ<>]/.test(t);
+  }) && table.cells.slice(1).some((row) => row.some((c) => c.text.trim() !== ""))) return false;
   if (table.rows <= 3 && table.cols <= 3) {
     const totalCells2 = table.rows * table.cols;
     const emptyCells2 = totalCells2 - allCells.length;
@@ -2807,7 +2841,7 @@ function extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals) {
     }
     if (remaining.length > 0) {
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkRCSDSN5Ycjs.safeMax.call(void 0, allY) - _chunkRCSDSN5Ycjs.safeMin.call(void 0, allY);
+      const pageH = _chunk2C5GCXPJcjs.safeMax.call(void 0, allY) - _chunk2C5GCXPJcjs.safeMin.call(void 0, allY);
       const groups = xyCutOrder(remaining, Math.max(15, pageH * 0.03));
       const textBlocks = [];
       for (const group of groups) {
@@ -2983,7 +3017,7 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false) {
       blocks.push({ type: "paragraph", text: tableText, pageNumber: pageNum, bbox, style: dominantStyle(items) });
     } else {
       const allY = items.map((i) => i.y);
-      const pageHeight = _chunkRCSDSN5Ycjs.safeMax.call(void 0, allY) - _chunkRCSDSN5Ycjs.safeMin.call(void 0, allY);
+      const pageHeight = _chunk2C5GCXPJcjs.safeMax.call(void 0, allY) - _chunk2C5GCXPJcjs.safeMin.call(void 0, allY);
       const gapThreshold = Math.max(15, pageHeight * 0.03);
       const orderedGroups = proseCutX !== null ? splitTwoColumnProse(items, proseCutX) : xyCutOrder(items, gapThreshold);
       for (const group of orderedGroups) {
@@ -3236,7 +3270,7 @@ async function loadPdfWithTimeout(buffer) {
       new Promise((_, reject) => {
         timer = setTimeout(() => {
           loadingTask.destroy();
-          reject(new (0, _chunkRCSDSN5Ycjs.KordocError)("PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (30\uCD08 \uCD08\uACFC)"));
+          reject(new (0, _chunk2C5GCXPJcjs.KordocError)("PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (30\uCD08 \uCD08\uACFC)"));
         }, PDF_LOAD_TIMEOUT_MS);
       })
     ]);
@@ -3249,7 +3283,7 @@ async function parsePdfDocument(buffer, options) {
   const doc = await loadPdfWithTimeout(buffer);
   try {
     const pageCount = doc.numPages;
-    if (pageCount === 0) throw new (0, _chunkRCSDSN5Ycjs.KordocError)("PDF\uC5D0 \uD398\uC774\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
+    if (pageCount === 0) throw new (0, _chunk2C5GCXPJcjs.KordocError)("PDF\uC5D0 \uD398\uC774\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
     const metadata = { pageCount };
     await extractPdfMetadata(doc, metadata);
     const blocks = [];
@@ -3309,11 +3343,11 @@ async function parsePdfDocument(buffer, options) {
           pageText += pageText ? "\n" + t : t;
         }
         pageQuality.push(computePageQuality(i, pageText));
-        if (totalTextBytes > MAX_TOTAL_TEXT) throw new (0, _chunkRCSDSN5Ycjs.KordocError)("\uD14D\uC2A4\uD2B8 \uCD94\uCD9C \uD06C\uAE30 \uCD08\uACFC");
+        if (totalTextBytes > MAX_TOTAL_TEXT) throw new (0, _chunk2C5GCXPJcjs.KordocError)("\uD14D\uC2A4\uD2B8 \uCD94\uCD9C \uD06C\uAE30 \uCD08\uACFC");
         parsedPages++;
         _optionalChain([options, 'optionalAccess', _36 => _36.onProgress, 'optionalCall', _37 => _37(parsedPages, totalTarget)]);
       } catch (pageErr) {
-        if (pageErr instanceof _chunkRCSDSN5Ycjs.KordocError) throw pageErr;
+        if (pageErr instanceof _chunk2C5GCXPJcjs.KordocError) throw pageErr;
         warnings.push({ page: i, message: `\uD398\uC774\uC9C0 ${i} \uD30C\uC2F1 \uC2E4\uD328: ${pageErr instanceof Error ? pageErr.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958"}`, code: "PARTIAL_PARSE" });
       }
     }
@@ -3381,7 +3415,7 @@ async function parsePdfDocument(buffer, options) {
     detectKoreanListBlocks(blocks);
     const outline = blocks.filter((b) => b.type === "heading" && b.level && b.text).map((b) => ({ level: b.level, text: b.text, pageNumber: b.pageNumber }));
     sanitizeBlockControlChars(blocks);
-    let markdown = cleanPdfText(_chunkRCSDSN5Ycjs.blocksToMarkdown.call(void 0, blocks));
+    let markdown = cleanPdfText(_chunk2C5GCXPJcjs.blocksToMarkdown.call(void 0, blocks));
     return {
       markdown,
       blocks,
@@ -3440,4 +3474,4 @@ async function extractPdfMetadataOnly(buffer) {
 
 
 exports.cleanPdfText = cleanPdfText; exports.detectKoreanListBlocks = detectKoreanListBlocks; exports.detectTableCaptions = detectTableCaptions; exports.extractPdfMetadataOnly = extractPdfMetadataOnly; exports.mergeCrossPageTables = mergeCrossPageTables; exports.parsePdfDocument = parsePdfDocument; exports.removeHeaderFooterBlocks = removeHeaderFooterBlocks;
-//# sourceMappingURL=parser-X34HD2KQ.cjs.map
+//# sourceMappingURL=parser-PCTX7LGN.cjs.map