kordoc 3.10.1 → 3.11.0

package/dist/{parser-AT22HMF5.cjs → parser-X34HD2KQ.cjs}

@@ -6,7 +6,7 @@
 
 
 
-var _chunk7UXZTZBJcjs = require('./chunk-7UXZTZBJ.cjs');
+var _chunkRCSDSN5Ycjs = require('./chunk-RCSDSN5Y.cjs');
 
 
 var _chunkDCZVOIEOcjs = require('./chunk-DCZVOIEO.cjs');
@@ -201,10 +201,90 @@ function classifyAndAdd(seg, lineWidth, horizontals, verticals) {
 function preprocessLines(horizontals, verticals) {
   let h = horizontals.filter((l) => l.lineWidth <= MAX_LINE_WIDTH);
   let v = verticals.filter((l) => l.lineWidth <= MAX_LINE_WIDTH);
+  h = dropShadingStacks(h, "h");
+  v = dropShadingStacks(v, "v");
   h = mergeParallelLines(h, "h");
   v = mergeParallelLines(v, "v");
   return { horizontals: h, verticals: v };
 }
+var STACK_GAP = 2;
+var STACK_MIN_LINES = 6;
+function dropShadingStacks(lines, dir) {
+  if (lines.length < STACK_MIN_LINES) return lines;
+  const groups = /* @__PURE__ */ new Map();
+  for (const l of lines) {
+    const key = dir === "h" ? `${Math.round(l.x1)}:${Math.round(l.x2)}` : `${Math.round(l.y1)}:${Math.round(l.y2)}`;
+    const arr = groups.get(key);
+    if (arr) arr.push(l);
+    else groups.set(key, [l]);
+  }
+  const dropped = /* @__PURE__ */ new Set();
+  for (const group of groups.values()) {
+    if (group.length < STACK_MIN_LINES) continue;
+    group.sort((a, b) => dir === "h" ? a.y1 - b.y1 : a.x1 - b.x1);
+    let runStart = 0;
+    for (let i = 1; i <= group.length; i++) {
+      const gap = i < group.length ? dir === "h" ? group[i].y1 - group[i - 1].y1 : group[i].x1 - group[i - 1].x1 : Infinity;
+      if (gap < STACK_GAP) continue;
+      if (i - runStart >= STACK_MIN_LINES) {
+        for (let j = runStart; j < i; j++) dropped.add(group[j]);
+      }
+      runStart = i;
+    }
+  }
+  return dropped.size ? lines.filter((l) => !dropped.has(l)) : lines;
+}
+var EDGE_ALIGN_TOL = 3;
+var EDGE_MIN_RULES = 3;
+var EDGE_MIN_SPAN = 12;
+var EDGE_INSET = 15;
+var EDGE_NEAR = 10;
+var EDGE_CONNECT_TOL = 5;
+function closeOpenTableEdges(horizontals, verticals) {
+  if (horizontals.length < EDGE_MIN_RULES) return verticals;
+  const groups = [];
+  for (const hl of 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) {
+        g2.push(hl);
+        placed = true;
+        break;
+      }
+    }
+    if (!placed) groups.push([hl]);
+  }
+  const synthesized = [];
+  for (const g2 of groups) {
+    if (g2.length < EDGE_MIN_RULES) continue;
+    let yMin = Infinity, yMax = -Infinity, x1 = 0, x2 = 0;
+    for (const hl of g2) {
+      if (hl.y1 < yMin) yMin = hl.y1;
+      if (hl.y1 > yMax) yMax = hl.y1;
+      x1 += hl.x1;
+      x2 += hl.x2;
+    }
+    x1 /= g2.length;
+    x2 /= g2.length;
+    if (yMax - yMin < EDGE_MIN_SPAN) continue;
+    const crossCount = (v) => {
+      let n = 0;
+      for (const hl of g2) {
+        if (v.x1 >= hl.x1 - EDGE_CONNECT_TOL && v.x1 <= hl.x2 + EDGE_CONNECT_TOL && hl.y1 >= v.y1 - EDGE_CONNECT_TOL && hl.y1 <= v.y2 + EDGE_CONNECT_TOL) n++;
+      }
+      return n;
+    };
+    const hasInterior = verticals.some((v) => v.x1 > x1 + EDGE_INSET && v.x1 < x2 - EDGE_INSET && crossCount(v) >= 2);
+    if (!hasInterior) continue;
+    for (const edgeX of [x1, x2]) {
+      const closed = verticals.some((v) => Math.abs(v.x1 - edgeX) <= EDGE_NEAR && v.y1 <= yMax + EDGE_CONNECT_TOL && v.y2 >= yMin - EDGE_CONNECT_TOL);
+      if (!closed) {
+        synthesized.push({ x1: edgeX, y1: yMin, x2: edgeX, y2: yMax, lineWidth: 0.5 });
+      }
+    }
+  }
+  return synthesized.length ? [...verticals, ...synthesized] : verticals;
+}
 function mergeParallelLines(lines, dir) {
   if (lines.length <= 1) return lines;
   const sorted = [...lines].sort((a, b) => {
@@ -1976,14 +2056,14 @@ 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 = _chunk7UXZTZBJcjs.safeMax.call(void 0, gaps);
+  const maxGap = _chunkRCSDSN5Ycjs.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 = _chunk7UXZTZBJcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunk7UXZTZBJcjs.safeMin.call(void 0, allItems.map((i) => i.x));
+  const pageWidth = _chunkRCSDSN5Ycjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkRCSDSN5Ycjs.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++) {
@@ -2205,9 +2285,9 @@ function detectHeadings(blocks, medianFontSize) {
     if (/^\d+$/.test(text)) continue;
     const ratio = block.style.fontSize / medianFontSize;
     let level = 0;
-    if (ratio >= _chunk7UXZTZBJcjs.HEADING_RATIO_H1) level = 1;
-    else if (ratio >= _chunk7UXZTZBJcjs.HEADING_RATIO_H2) level = 2;
-    else if (ratio >= _chunk7UXZTZBJcjs.HEADING_RATIO_H3) level = 3;
+    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 (level > 0) {
       block.type = "heading";
       block.level = level;
@@ -2555,6 +2635,7 @@ function extractPageBlocksWithLines(items, pageNum, opList, pageWidth, pageHeigh
   let { horizontals, verticals } = extractLines(opList.fnArray, opList.argsArray);
   ({ horizontals, verticals } = filterPageBorderLines(horizontals, verticals, pageWidth, pageHeight));
   ({ horizontals, verticals } = preprocessLines(horizontals, verticals));
+  verticals = closeOpenTableEdges(horizontals, verticals);
   markStrikethroughItems(items, horizontals);
   wrapStrikethroughRuns(items);
   const grids = buildTableGrids(horizontals, verticals);
@@ -2726,7 +2807,7 @@ function extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals) {
     }
     if (remaining.length > 0) {
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunk7UXZTZBJcjs.safeMax.call(void 0, allY) - _chunk7UXZTZBJcjs.safeMin.call(void 0, allY);
+      const pageH = _chunkRCSDSN5Ycjs.safeMax.call(void 0, allY) - _chunkRCSDSN5Ycjs.safeMin.call(void 0, allY);
       const groups = xyCutOrder(remaining, Math.max(15, pageH * 0.03));
       const textBlocks = [];
       for (const group of groups) {
@@ -2902,7 +2983,7 @@ 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 = _chunk7UXZTZBJcjs.safeMax.call(void 0, allY) - _chunk7UXZTZBJcjs.safeMin.call(void 0, allY);
+      const pageHeight = _chunkRCSDSN5Ycjs.safeMax.call(void 0, allY) - _chunkRCSDSN5Ycjs.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) {
@@ -3155,7 +3236,7 @@ async function loadPdfWithTimeout(buffer) {
       new Promise((_, reject) => {
         timer = setTimeout(() => {
           loadingTask.destroy();
-          reject(new (0, _chunk7UXZTZBJcjs.KordocError)("PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (30\uCD08 \uCD08\uACFC)"));
+          reject(new (0, _chunkRCSDSN5Ycjs.KordocError)("PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (30\uCD08 \uCD08\uACFC)"));
         }, PDF_LOAD_TIMEOUT_MS);
       })
     ]);
@@ -3168,7 +3249,7 @@ async function parsePdfDocument(buffer, options) {
   const doc = await loadPdfWithTimeout(buffer);
   try {
     const pageCount = doc.numPages;
-    if (pageCount === 0) throw new (0, _chunk7UXZTZBJcjs.KordocError)("PDF\uC5D0 \uD398\uC774\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
+    if (pageCount === 0) throw new (0, _chunkRCSDSN5Ycjs.KordocError)("PDF\uC5D0 \uD398\uC774\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
     const metadata = { pageCount };
     await extractPdfMetadata(doc, metadata);
     const blocks = [];
@@ -3228,11 +3309,11 @@ async function parsePdfDocument(buffer, options) {
           pageText += pageText ? "\n" + t : t;
         }
         pageQuality.push(computePageQuality(i, pageText));
-        if (totalTextBytes > MAX_TOTAL_TEXT) throw new (0, _chunk7UXZTZBJcjs.KordocError)("\uD14D\uC2A4\uD2B8 \uCD94\uCD9C \uD06C\uAE30 \uCD08\uACFC");
+        if (totalTextBytes > MAX_TOTAL_TEXT) throw new (0, _chunkRCSDSN5Ycjs.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 _chunk7UXZTZBJcjs.KordocError) throw pageErr;
+        if (pageErr instanceof _chunkRCSDSN5Ycjs.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" });
       }
     }
@@ -3300,7 +3381,7 @@ 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(_chunk7UXZTZBJcjs.blocksToMarkdown.call(void 0, blocks));
+    let markdown = cleanPdfText(_chunkRCSDSN5Ycjs.blocksToMarkdown.call(void 0, blocks));
     return {
       markdown,
       blocks,
@@ -3359,4 +3440,4 @@ 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-AT22HMF5.cjs.map
+//# sourceMappingURL=parser-X34HD2KQ.cjs.map