kordoc 3.16.0 → 3.16.2

package/dist/{parser-6V7F3GET.cjs → parser-YVRTKGKO.cjs}

@@ -6,7 +6,7 @@
 
 
 
-var _chunkBIP4JAVGcjs = require('./chunk-BIP4JAVG.cjs');
+var _chunk72LLOLFXcjs = require('./chunk-72LLOLFX.cjs');
 
 
 var _chunkDCZVOIEOcjs = require('./chunk-DCZVOIEO.cjs');
@@ -49,14 +49,14 @@ function extractLines(fnArray, argsArray) {
       pushSeg(rx, ry + rh, rx, ry);
     }
   }
-  function flushPath(isStroke) {
+  function flushPath(isStroke, fromFill = false) {
     if (!isStroke) {
       currentPath = [];
       return;
     }
     const effWidth = lineWidth * ctmScale();
     for (const seg of currentPath) {
-      classifyAndAdd(seg, effWidth, horizontals, verticals);
+      classifyAndAdd(seg, effWidth, horizontals, verticals, fromFill);
     }
     currentPath = [];
   }
@@ -154,7 +154,9 @@ function extractLines(fnArray, argsArray) {
           }
           if (afterOp === _pdfmjs.OPS.stroke || afterOp === _pdfmjs.OPS.closeStroke) {
             flushPath(true);
-          } else if (afterOp === _pdfmjs.OPS.fill || afterOp === _pdfmjs.OPS.eoFill || afterOp === _pdfmjs.OPS.fillStroke || afterOp === _pdfmjs.OPS.eoFillStroke || afterOp === _pdfmjs.OPS.closeFillStroke || afterOp === _pdfmjs.OPS.closeEOFillStroke) {
+          } else if (afterOp === _pdfmjs.OPS.fill || afterOp === _pdfmjs.OPS.eoFill) {
+            flushPath(true, true);
+          } else if (afterOp === _pdfmjs.OPS.fillStroke || afterOp === _pdfmjs.OPS.eoFillStroke || afterOp === _pdfmjs.OPS.closeFillStroke || afterOp === _pdfmjs.OPS.closeEOFillStroke) {
             flushPath(true);
           } else if (afterOp === _pdfmjs.OPS.endPath) {
             flushPath(false);
@@ -168,6 +170,8 @@ function extractLines(fnArray, argsArray) {
         break;
       case _pdfmjs.OPS.fill:
       case _pdfmjs.OPS.eoFill:
+        flushPath(true, true);
+        break;
       case _pdfmjs.OPS.fillStroke:
       case _pdfmjs.OPS.eoFillStroke:
       case _pdfmjs.OPS.closeFillStroke:
@@ -181,7 +185,7 @@ function extractLines(fnArray, argsArray) {
   }
   return { horizontals, verticals };
 }
-function classifyAndAdd(seg, lineWidth, horizontals, verticals) {
+function classifyAndAdd(seg, lineWidth, horizontals, verticals, fromFill = false) {
   const dx = Math.abs(seg.x2 - seg.x1);
   const dy = Math.abs(seg.y2 - seg.y1);
   const length = Math.sqrt(dx * dx + dy * dy);
@@ -190,12 +194,12 @@ function classifyAndAdd(seg, lineWidth, horizontals, verticals) {
     const y = (seg.y1 + seg.y2) / 2;
     const x1 = Math.min(seg.x1, seg.x2);
     const x2 = Math.max(seg.x1, seg.x2);
-    horizontals.push({ x1, y1: y, x2, y2: y, lineWidth });
+    horizontals.push({ x1, y1: y, x2, y2: y, lineWidth, fromFill });
   } else if (dx <= ORIENTATION_TOL) {
     const x = (seg.x1 + seg.x2) / 2;
     const y1 = Math.min(seg.y1, seg.y2);
     const y2 = Math.max(seg.y1, seg.y2);
-    verticals.push({ x1: x, y1, x2: x, y2, lineWidth });
+    verticals.push({ x1: x, y1, x2: x, y2, lineWidth, fromFill });
   }
 }
 function preprocessLines(horizontals, verticals) {
@@ -243,7 +247,10 @@ function dropShadingStacks(lines, dir) {
         for (let j = s + 1; j <= e; j++) pitches.push(coord(group[j]) - coord(group[j - 1]));
         pitches.sort((a, b) => a - b);
         const medPitch = _nullishCoalesce(pitches[Math.floor(pitches.length / 2)], () => ( 0));
-        const edgeAlien = (j, inwardGap) => wKey(group[j]) !== domW || medPitch > 0 && inwardGap > medPitch * 1.8;
+        let fillN = 0;
+        for (let j = s; j <= e; j++) if (group[j].fromFill) fillN++;
+        const stackIsFill = fillN * 2 > e - s + 1;
+        const edgeAlien = (j, inwardGap) => stackIsFill && !group[j].fromFill || wKey(group[j]) !== domW || medPitch > 0 && inwardGap > medPitch * 1.8;
         while (e - s + 1 >= STACK_MIN_LINES) {
           if (edgeAlien(s, coord(group[s + 1]) - coord(group[s]))) {
             s++;
@@ -271,6 +278,7 @@ var EDGE_INSET = 15;
 var EDGE_NEAR = 10;
 var EDGE_CONNECT_TOL = 5;
 var EDGE_YGAP_SPLIT_K = 2.5;
+var EDGE_YGAP_ABS_MIN = 30;
 var CHAIN_Y_TOL = 1.5;
 var CHAIN_GAP = 3;
 function chainCollinearRules(horizontals) {
@@ -324,7 +332,12 @@ function closeOpenTableEdges(horizontals, verticals) {
     const threshold = median * EDGE_YGAP_SPLIT_K;
     let cur = sorted.length ? [sorted[0]] : [];
     for (let i = 1; i < sorted.length; i++) {
-      if (median > 0 && sorted[i].y1 - sorted[i - 1].y1 > threshold && cur.length >= EDGE_MIN_RULES && sorted.length - i >= EDGE_MIN_RULES) {
+      const yLo = sorted[i - 1].y1, yHi = sorted[i].y1;
+      const gap = yHi - yLo;
+      const gx1 = Math.min(sorted[i - 1].x1, sorted[i].x1);
+      const gx2 = Math.max(sorted[i - 1].x2, sorted[i].x2);
+      const bridged = verticals.some((vl) => vl.y1 <= yLo + EDGE_NEAR && vl.y2 >= yHi - EDGE_NEAR && vl.x1 >= gx1 - EDGE_CONNECT_TOL && vl.x1 <= gx2 + EDGE_CONNECT_TOL);
+      if (median > 0 && gap > threshold && gap > EDGE_YGAP_ABS_MIN && !bridged && cur.length >= EDGE_MIN_RULES && sorted.length - i >= EDGE_MIN_RULES) {
         splitGroups.push(cur);
         cur = [];
       }
@@ -2134,14 +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 = _chunkBIP4JAVGcjs.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 = _chunkBIP4JAVGcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkBIP4JAVGcjs.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++) {
@@ -2363,9 +2376,9 @@ function detectHeadings(blocks, medianFontSize) {
     if (/^\d+$/.test(text)) continue;
     const ratio = block.style.fontSize / medianFontSize;
     let level = 0;
-    if (ratio >= _chunkBIP4JAVGcjs.HEADING_RATIO_H1) level = 1;
-    else if (ratio >= _chunkBIP4JAVGcjs.HEADING_RATIO_H2) level = 2;
-    else if (ratio >= _chunkBIP4JAVGcjs.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;
@@ -2380,7 +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;
@@ -2941,7 +2954,7 @@ function extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals) {
     }
     if (remaining.length > 0) {
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkBIP4JAVGcjs.safeMax.call(void 0, allY) - _chunkBIP4JAVGcjs.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) {
@@ -3117,7 +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 = _chunkBIP4JAVGcjs.safeMax.call(void 0, allY) - _chunkBIP4JAVGcjs.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) {
@@ -3370,7 +3383,7 @@ async function loadPdfWithTimeout(buffer) {
       new Promise((_, reject) => {
         timer = setTimeout(() => {
           loadingTask.destroy();
-          reject(new (0, _chunkBIP4JAVGcjs.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);
       })
     ]);
@@ -3383,7 +3396,7 @@ async function parsePdfDocument(buffer, options) {
   const doc = await loadPdfWithTimeout(buffer);
   try {
     const pageCount = doc.numPages;
-    if (pageCount === 0) throw new (0, _chunkBIP4JAVGcjs.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 = [];
@@ -3443,11 +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, _chunkBIP4JAVGcjs.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 _chunkBIP4JAVGcjs.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" });
       }
     }
@@ -3515,7 +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(_chunkBIP4JAVGcjs.blocksToMarkdown.call(void 0, blocks));
+    let markdown = cleanPdfText(_chunk72LLOLFXcjs.blocksToMarkdown.call(void 0, blocks));
     return {
       markdown,
       blocks,
@@ -3574,4 +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-6V7F3GET.cjs.map
+//# sourceMappingURL=parser-YVRTKGKO.cjs.map