kordoc 3.16.1 → 3.16.2

package/dist/{parser-3O2UF6VO.cjs → parser-YVRTKGKO.cjs}

@@ -6,7 +6,7 @@
 
 
 
-var _chunkVNIQK4ALcjs = require('./chunk-VNIQK4AL.cjs');
+var _chunk72LLOLFXcjs = require('./chunk-72LLOLFX.cjs');
 
 
 var _chunkDCZVOIEOcjs = require('./chunk-DCZVOIEO.cjs');
@@ -2147,14 +2147,14 @@ 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 = _chunkVNIQK4ALcjs.safeMax.call(void 0, gaps);
+  const maxGap = _chunk72LLOLFXcjs.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 = _chunkVNIQK4ALcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkVNIQK4ALcjs.safeMin.call(void 0, allItems.map((i) => i.x));
+  const pageWidth = _chunk72LLOLFXcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunk72LLOLFXcjs.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++) {
@@ -2376,9 +2376,9 @@ function detectHeadings(blocks, medianFontSize) {
     if (/^\d+$/.test(text)) continue;
     const ratio = block.style.fontSize / medianFontSize;
     let level = 0;
-    if (ratio >= _chunkVNIQK4ALcjs.HEADING_RATIO_H1) level = 1;
-    else if (ratio >= _chunkVNIQK4ALcjs.HEADING_RATIO_H2) level = 2;
-    else if (ratio >= _chunkVNIQK4ALcjs.HEADING_RATIO_H3) level = 3;
+    if (ratio >= _chunk72LLOLFXcjs.HEADING_RATIO_H1) level = 1;
+    else if (ratio >= _chunk72LLOLFXcjs.HEADING_RATIO_H2) level = 2;
+    else if (ratio >= _chunk72LLOLFXcjs.HEADING_RATIO_H3) level = 3;
     if (level > 0) {
       block.type = "heading";
       block.level = level;
@@ -2393,7 +2393,7 @@ function shouldDemoteTable(table) {
     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) {
+  if (table.rows <= 3 && table.cols <= 3 && allText.length <= 200) {
     const totalCells2 = table.rows * table.cols;
     const emptyCells2 = totalCells2 - allCells.length;
     if (emptyCells2 >= totalCells2 * 0.3) return true;
@@ -2954,7 +2954,7 @@ function extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals) {
     }
     if (remaining.length > 0) {
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkVNIQK4ALcjs.safeMax.call(void 0, allY) - _chunkVNIQK4ALcjs.safeMin.call(void 0, allY);
+      const pageH = _chunk72LLOLFXcjs.safeMax.call(void 0, allY) - _chunk72LLOLFXcjs.safeMin.call(void 0, allY);
       const groups = xyCutOrder(remaining, Math.max(15, pageH * 0.03));
       const textBlocks = [];
       for (const group of groups) {
@@ -3130,7 +3130,7 @@ 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 = _chunkVNIQK4ALcjs.safeMax.call(void 0, allY) - _chunkVNIQK4ALcjs.safeMin.call(void 0, allY);
+      const pageHeight = _chunk72LLOLFXcjs.safeMax.call(void 0, allY) - _chunk72LLOLFXcjs.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) {
@@ -3383,7 +3383,7 @@ async function loadPdfWithTimeout(buffer) {
       new Promise((_, reject) => {
         timer = setTimeout(() => {
           loadingTask.destroy();
-          reject(new (0, _chunkVNIQK4ALcjs.KordocError)("PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (30\uCD08 \uCD08\uACFC)"));
+          reject(new (0, _chunk72LLOLFXcjs.KordocError)("PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (30\uCD08 \uCD08\uACFC)"));
         }, PDF_LOAD_TIMEOUT_MS);
       })
     ]);
@@ -3396,7 +3396,7 @@ async function parsePdfDocument(buffer, options) {
   const doc = await loadPdfWithTimeout(buffer);
   try {
     const pageCount = doc.numPages;
-    if (pageCount === 0) throw new (0, _chunkVNIQK4ALcjs.KordocError)("PDF\uC5D0 \uD398\uC774\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
+    if (pageCount === 0) throw new (0, _chunk72LLOLFXcjs.KordocError)("PDF\uC5D0 \uD398\uC774\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
     const metadata = { pageCount };
     await extractPdfMetadata(doc, metadata);
     const blocks = [];
@@ -3456,11 +3456,11 @@ async function parsePdfDocument(buffer, options) {
           pageText += pageText ? "\n" + t : t;
         }
         pageQuality.push(computePageQuality(i, pageText));
-        if (totalTextBytes > MAX_TOTAL_TEXT) throw new (0, _chunkVNIQK4ALcjs.KordocError)("\uD14D\uC2A4\uD2B8 \uCD94\uCD9C \uD06C\uAE30 \uCD08\uACFC");
+        if (totalTextBytes > MAX_TOTAL_TEXT) throw new (0, _chunk72LLOLFXcjs.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 _chunkVNIQK4ALcjs.KordocError) throw pageErr;
+        if (pageErr instanceof _chunk72LLOLFXcjs.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" });
       }
     }
@@ -3528,7 +3528,7 @@ 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(_chunkVNIQK4ALcjs.blocksToMarkdown.call(void 0, blocks));
+    let markdown = cleanPdfText(_chunk72LLOLFXcjs.blocksToMarkdown.call(void 0, blocks));
     return {
       markdown,
       blocks,
@@ -3587,4 +3587,4 @@ 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-3O2UF6VO.cjs.map
+//# sourceMappingURL=parser-YVRTKGKO.cjs.map