kordoc 4.15.6 → 4.15.7

package/dist/{chunk-SOM7JIKG.cjs → chunk-4ANWGSUT.cjs}

@@ -2,11 +2,11 @@
 
 
 
-var _chunkCND4NRNFcjs = require('./chunk-CND4NRNF.cjs');
+var _chunkYNJLZVEOcjs = require('./chunk-YNJLZVEO.cjs');
 
 
 
-var _chunkFBZCWSPNcjs = require('./chunk-FBZCWSPN.cjs');
+var _chunkJ6ZU7AIRcjs = require('./chunk-J6ZU7AIR.cjs');
 
 // src/pdf/image-regions.ts
 var _pdfmjs = require('pdfjs-dist/legacy/build/pdf.mjs');
@@ -790,6 +790,10 @@ var CUT_INTERIOR_MATCH_TOL = 8;
 var CUT_MAX_INTERIOR_OVERLAP = 0.5;
 var CUT_VCHAIN_X_TOL = 1.5;
 var CUT_VCHAIN_GAP = 1;
+var NEST_STRICT_TOL = 0.5;
+var NEST_MIN_GAP = 0.8;
+var NEST_EDGE_COVER = 0.95;
+var NEST_DOUBLE_RULE_GAP = 6;
 var VERTEX_BUCKET_CELL = 100;
 function lowerBound(sorted, key) {
   let lo = 0, hi = sorted.length;
@@ -893,9 +897,12 @@ function buildTableGrids(horizontals, verticals) {
     ...verticals.map((l, i) => ({ ...l, type: "v", id: i + horizontals.length }))
   ];
   const groups = [];
-  for (const g of groupConnectedLines(allLines)) {
+  for (const g0 of groupConnectedLines(allLines)) {
+    const { outer: g, nested } = splitNestedBoxes(g0);
+    for (const n of nested) groups.push({ lines: n, fromSplit: true, nested: true });
+    if (g.length === 0) continue;
     const bands = splitStackedGroup(g);
-    for (const b of bands) groups.push({ lines: b, fromSplit: bands.length > 1 });
+    for (const b of bands) groups.push({ lines: b, fromSplit: bands.length > 1 || nested.length > 0 });
   }
   const grids = [];
   const byY = [...vertices].sort((a, b) => a.y - b.y);
@@ -909,7 +916,7 @@ function buildTableGrids(horizontals, verticals) {
     }
     return lo;
   };
-  for (const { lines: group, fromSplit } of groups) {
+  for (const { lines: group, fromSplit, nested } of groups) {
     const hLines = group.filter((l) => l.type === "h");
     const vLines = group.filter((l) => l.type === "v");
     if (hLines.length < 2 || vLines.length < 2) continue;
@@ -971,9 +978,9 @@ function buildTableGrids(horizontals, verticals) {
       x2: validColXs[validColXs.length - 1],
       y2: validRowYs[0]
     };
-    grids.push({ rowYs: validRowYs, colXs: validColXs, bbox, vertexRadius: groupRadius });
+    grids.push({ rowYs: validRowYs, colXs: validColXs, bbox, vertexRadius: groupRadius, ...nested ? { lineNested: true } : {} });
   }
-  return mergeAdjacentGrids(grids);
+  return [...mergeAdjacentGrids(grids.filter((g) => !g.lineNested)), ...grids.filter((g) => g.lineNested)];
 }
 var SHADE_CLIP_TOL = 1.5;
 var SHADE_CLIP_MAX_AREA = 0.5;
@@ -1152,8 +1159,52 @@ function splitStackedGroup(group) {
   }
   return bands.filter((b) => b.length > 0);
 }
+function splitNestedBoxes(group) {
+  const comps = groupConnectedLines(group, NEST_STRICT_TOL);
+  if (comps.length < 2) return { outer: group, nested: [] };
+  const boxes = comps.map((c) => {
+    const hs = c.filter((l) => l.type === "h"), vs = c.filter((l) => l.type === "v");
+    if (hs.length < 2 || vs.length < 2) return null;
+    const x1 = Math.min(...vs.map((v) => v.x1)), x2 = Math.max(...vs.map((v) => v.x1));
+    const y1 = Math.min(...hs.map((h) => h.y1)), y2 = Math.max(...hs.map((h) => h.y1));
+    if (x2 - x1 < 1 || y2 - y1 < 1) return null;
+    const cover = (segs, a, b) => segs.reduce((s, [p, q]) => s + Math.max(0, Math.min(q, b) - Math.max(p, a)), 0) >= (b - a) * NEST_EDGE_COVER;
+    const edgeH = (y) => cover(hs.filter((h) => Math.abs(h.y1 - y) <= NEST_STRICT_TOL).map((h) => [h.x1, h.x2]), x1, x2);
+    const edgeV = (x) => cover(vs.filter((v) => Math.abs(v.x1 - x) <= NEST_STRICT_TOL).map((v) => [v.y1, v.y2]), y1, y2);
+    return edgeH(y1) && edgeH(y2) && edgeV(x1) && edgeV(x2) ? { x1, y1, x2, y2 } : null;
+  });
+  const bboxOf = (c) => ({
+    x1: Math.min(...c.map((l) => Math.min(l.x1, l.x2))),
+    x2: Math.max(...c.map((l) => Math.max(l.x1, l.x2))),
+    y1: Math.min(...c.map((l) => Math.min(l.y1, l.y2))),
+    y2: Math.max(...c.map((l) => Math.max(l.y1, l.y2)))
+  });
+  const extents = comps.map(bboxOf);
+  const nestedIdx = /* @__PURE__ */ new Set();
+  comps.forEach((_, i) => {
+    const b = boxes[i];
+    if (!b) return;
+    const e = extents[i];
+    const uniq = (xs) => xs.filter((x, k) => xs.findIndex((y) => Math.abs(y - x) <= NEST_STRICT_TOL) === k).length;
+    const bare = uniq(comps[i].filter((l) => l.type === "h").map((l) => l.y1)) === 2 && uniq(comps[i].filter((l) => l.type === "v").map((l) => l.x1)) === 2;
+    const inside = comps.some((c, j) => {
+      if (j === i || c.filter((l) => l.type === "v").length < 2 || !c.some((l) => l.type === "h")) return false;
+      const p = extents[j];
+      const gaps = [e.x1 - p.x1, p.x2 - e.x2, e.y1 - p.y1, p.y2 - e.y2];
+      return gaps.every((g) => g >= NEST_MIN_GAP) && !(bare && gaps.every((g) => g <= NEST_DOUBLE_RULE_GAP));
+    });
+    if (!inside) return;
+    const crossed = comps.some((c, j) => j !== i && c.some((l) => l.type === "h" ? l.y1 > e.y1 + NEST_MIN_GAP && l.y1 < e.y2 - NEST_MIN_GAP && Math.min(l.x2, e.x2) - Math.max(l.x1, e.x1) > NEST_MIN_GAP : l.x1 > e.x1 + NEST_MIN_GAP && l.x1 < e.x2 - NEST_MIN_GAP && Math.min(l.y2, e.y2) - Math.max(l.y1, e.y1) > NEST_MIN_GAP) && !(extents[j].x1 >= e.x1 && extents[j].x2 <= e.x2 && extents[j].y1 >= e.y1 && extents[j].y2 <= e.y2));
+    if (!crossed) nestedIdx.add(i);
+  });
+  if (nestedIdx.size === 0) return { outer: group, nested: [] };
+  return {
+    outer: comps.filter((_, i) => !nestedIdx.has(i)).flat(),
+    nested: comps.filter((_, i) => nestedIdx.has(i))
+  };
+}
 var GROUP_BUCKET_CELL = 100;
-function groupConnectedLines(lines) {
+function groupConnectedLines(lines, tol = CONNECT_TOL) {
   const parent = lines.map((_, i) => i);
   function find(x) {
     while (parent[x] !== x) {
@@ -1195,7 +1246,7 @@ function groupConnectedLines(lines) {
         const key = i * lines.length + j;
         if (tested.has(key)) continue;
         tested.add(key);
-        if (linesIntersect(lines[i], lines[j])) {
+        if (linesIntersect(lines[i], lines[j], tol)) {
           union(i, j);
         }
       }
@@ -1209,19 +1260,18 @@ function groupConnectedLines(lines) {
   }
   return [...groups.values()];
 }
-function linesIntersect(a, b) {
+function linesIntersect(a, b, tol = CONNECT_TOL) {
   if (a.type === b.type) {
     if (a.type === "h") {
-      if (Math.abs(a.y1 - b.y1) > CONNECT_TOL) return false;
-      return Math.min(a.x2, b.x2) >= Math.max(a.x1, b.x1) - CONNECT_TOL;
+      if (Math.abs(a.y1 - b.y1) > tol) return false;
+      return Math.min(a.x2, b.x2) >= Math.max(a.x1, b.x1) - tol;
     } else {
-      if (Math.abs(a.x1 - b.x1) > CONNECT_TOL) return false;
-      return Math.min(a.y2, b.y2) >= Math.max(a.y1, b.y1) - CONNECT_TOL;
+      if (Math.abs(a.x1 - b.x1) > tol) return false;
+      return Math.min(a.y2, b.y2) >= Math.max(a.y1, b.y1) - tol;
     }
   }
   const h = a.type === "h" ? a : b;
   const v = a.type === "h" ? b : a;
-  const tol = CONNECT_TOL;
   return v.x1 >= h.x1 - tol && v.x1 <= h.x2 + tol && h.y1 >= v.y1 - tol && h.y1 <= v.y2 + tol;
 }
 
@@ -1265,6 +1315,13 @@ function clusterCoords(values) {
   if (run.length > 0) out.push(run.reduce((s, x) => s + x, 0) / run.length);
   return out;
 }
+function dropSliverGaps(coords) {
+  const out = [...coords];
+  if (out.length > 2 && out[1] - out[0] < 1) out.splice(0, 2, (out[0] + out[1]) / 2);
+  const n = out.length;
+  if (n > 2 && out[n - 1] - out[n - 2] < 1) out.splice(n - 2, 2, (out[n - 2] + out[n - 1]) / 2);
+  return out;
+}
 var nearestIndex = (coords, v) => {
   let best = 0;
   for (let i = 1; i < coords.length; i++) if (Math.abs(coords[i] - v) < Math.abs(coords[best] - v)) best = i;
@@ -1462,7 +1519,7 @@ function buildClipCellGrids(rects, strokedH, strokedV, pageWidth, pageHeight, te
       if (!edgeStroked(strokedV, "v", r.x2, r.y1, r.y2)) invisible++;
     }
     if (invisible / edges < CLIP_MIN_INVISIBLE) continue;
-    const colXs = clusterCoords(members.flatMap((r) => [r.x1, r.x2]));
+    const colXs = dropSliverGaps(clusterCoords(members.flatMap((r) => [r.x1, r.x2])));
     const rowYs = clusterCoords(members.flatMap((r) => [r.y1, r.y2])).reverse();
     const numRows = rowYs.length - 1, numCols = colXs.length - 1;
     if (numRows < 1 || numCols < 1) continue;
@@ -2224,7 +2281,7 @@ function mergeCellTextLines(textLines, wrap) {
       merged[merged.length - 1] = prev + curr.trim();
     } else if (!wrap && /[,(]$/.test(prev.trim()) && curr.trim().length <= 15) {
       merged[merged.length - 1] = prev + curr.trim();
-    } else if (/[\d,]$/.test(prev) && /^[\d,]+[)\]]?$/.test(curr.trim()) && curr.trim().length <= 10 && !(/\d,\d{3}$/.test(prev) && /^\d/.test(curr.trim())) && !(/(?:^|[^\d,])\d{1,3}$/.test(prev) && /^(\d{1,3}(,\d{3})+|\d{1,3})$/.test(curr.trim()))) {
+    } else if (/[\d,]$/.test(prev) && /^[\d,]+[)\]]?$/.test(curr.trim()) && curr.trim().length <= 10 && !(/\d,\d{3}$/.test(prev) && /^\d/.test(curr.trim())) && !(/(?:^|[^\d,])\d{1,3}$/.test(prev) && /^(\d{1,3}(,\d{3})+|\d{1,3})$/.test(curr.trim()) && !_optionalChain([wrap, 'optionalAccess', _38 => _38.wraps, 'access', _39 => _39[i - 1]])) && !(/(?:^|\s)\d{1,7}$/.test(prev) && /^\d{1,7}$/.test(curr.trim()) && !_optionalChain([wrap, 'optionalAccess', _40 => _40.wraps, 'access', _41 => _41[i - 1]]))) {
       merged[merged.length - 1] = prev + curr.trim();
     } else {
       merged.push(curr);
@@ -2294,7 +2351,7 @@ function normalizeUndersegmentedTable(originalCells, colXs, items, rowYs) {
   if (numRows > MAX_UNDERSEGMENTED_ROWS || numCols < MIN_UNDERSEGMENTED_COLUMNS) return null;
   if (items.length === 0) return null;
   if (numRows >= 3) {
-    if (_optionalChain([rowYs, 'optionalAccess', _38 => _38.length]) !== numRows + 1) return null;
+    if (_optionalChain([rowYs, 'optionalAccess', _42 => _42.length]) !== numRows + 1) return null;
     const rebuilt2 = [];
     let expanded = 0;
     for (let r = 0; r < numRows; r++) {
@@ -2435,7 +2492,7 @@ function dominantStyle(items) {
   let maxCount = 0, dominantSize = 0;
   for (const i of items) {
     if (i.fontSize <= 0) continue;
-    const count = (freq.get(i.fontSize) || 0) + Math.max(1, _nullishCoalesce(_optionalChain([i, 'access', _39 => _39.text, 'access', _40 => _40.match, 'call', _41 => _41(/[\p{L}\p{N}]/gu), 'optionalAccess', _42 => _42.length]), () => ( 0)));
+    const count = (freq.get(i.fontSize) || 0) + Math.max(1, _nullishCoalesce(_optionalChain([i, 'access', _43 => _43.text, 'access', _44 => _44.match, 'call', _45 => _45(/[\p{L}\p{N}]/gu), 'optionalAccess', _46 => _46.length]), () => ( 0)));
     freq.set(i.fontSize, count);
     if (count > maxCount) {
       maxCount = count;
@@ -2443,7 +2500,7 @@ function dominantStyle(items) {
     }
   }
   if (dominantSize === 0) return void 0;
-  const fontName = _optionalChain([items, 'access', _43 => _43.find, 'call', _44 => _44((i) => i.fontSize === dominantSize), 'optionalAccess', _45 => _45.fontName]) || void 0;
+  const fontName = _optionalChain([items, 'access', _47 => _47.find, 'call', _48 => _48((i) => i.fontSize === dominantSize), 'optionalAccess', _49 => _49.fontName]) || void 0;
   return fontName ? { fontSize: dominantSize, fontName } : { fontSize: dominantSize };
 }
 function normalizeItems(rawItems) {
@@ -2556,7 +2613,7 @@ function splitAroundOverlays(c, overlays) {
   const ov = overlays.filter((o) => {
     for (let at = c.text.indexOf(o.text); at >= 0; at = c.text.indexOf(o.text, at + 1)) {
       const prefix = c.text.slice(0, at);
-      const est = c.x + glyphUnits(prefix.replace(/\s+/g, "")) * a0 + (_nullishCoalesce(_optionalChain([prefix, 'access', _46 => _46.match, 'call', _47 => _47(/\s+/g), 'optionalAccess', _48 => _48.length]), () => ( 0))) * ws;
+      const est = c.x + glyphUnits(prefix.replace(/\s+/g, "")) * a0 + (_nullishCoalesce(_optionalChain([prefix, 'access', _50 => _50.match, 'call', _51 => _51(/\s+/g), 'optionalAccess', _52 => _52.length]), () => ( 0))) * ws;
       if (Math.abs(est - o.x) <= OVERLAY_DUP_EM * fs) return false;
     }
     return true;
@@ -2669,7 +2726,7 @@ function mergeSuperscriptLines(lines) {
   return result;
 }
 function mergeLineSimple(items) {
-  if (items.length <= 1) return _optionalChain([items, 'access', _49 => _49[0], 'optionalAccess', _50 => _50.text]) || "";
+  if (items.length <= 1) return _optionalChain([items, 'access', _53 => _53[0], 'optionalAccess', _54 => _54.text]) || "";
   const sorted = sortLineByX([...items]);
   const isEvenSpaced = detectEvenSpacedItems(sorted);
   let result = sorted[0].text;
@@ -2810,7 +2867,7 @@ function splitTrailingColumnRegion(items) {
 }
 function panelBlocks(items, panel, pageNum) {
   const lines = groupByY([...items].sort((a, b) => b.y - a.y || a.x - b.x));
-  const faceOf = (line) => _nullishCoalesce(_optionalChain([dominantStyle, 'call', _51 => _51(line), 'optionalAccess', _52 => _52.fontName]), () => ( ""));
+  const faceOf = (line) => _nullishCoalesce(_optionalChain([dominantStyle, 'call', _55 => _55(line), 'optionalAccess', _56 => _56.fontName]), () => ( ""));
   const runs = [];
   for (const line of lines) {
     const last = runs[runs.length - 1];
@@ -2840,9 +2897,12 @@ function pushLineParagraphs(out, yLines, pageNum, lex) {
   const lines = attachDropCaps(yLines).map((items) => ({ items, text: mergeLineSimple(items) })).filter((l) => l.text.trim());
   const geo = lines.map((l) => {
     const b = computeBBox(l.items, pageNum);
-    return { text: l.text, left: b.x, right: b.x + b.width, y: l.items.reduce((s, i) => s + i.y, 0) / l.items.length, fontSize: _nullishCoalesce(_optionalChain([dominantStyle, 'call', _53 => _53(l.items), 'optionalAccess', _54 => _54.fontSize]), () => ( 0)) };
+    return { text: l.text, left: b.x, right: b.x + b.width, y: l.items.reduce((s, i) => s + i.y, 0) / l.items.length, fontSize: _nullishCoalesce(_optionalChain([dominantStyle, 'call', _57 => _57(l.items), 'optionalAccess', _58 => _58.fontSize]), () => ( 0)) };
   });
   const joins = bodyLineJoins(geo, lex);
+  for (let i = 0; i + 1 < geo.length; i++) {
+    if (/^\d{1,2}$/.test(geo[i].text.trim()) && geo[i].fontSize >= geo[i + 1].fontSize * 1.2) joins[i] = "\n";
+  }
   if (hasNumberedStyledTitle(lines.map((line) => line.items))) {
     joins[0] = "\n";
     joins[1] = "\n";
@@ -2879,16 +2939,17 @@ function computeMedianFontSizeFromFreq(freq) {
   return sorted[sorted.length - 1][0];
 }
 function detectHeadings(blocks, medianFontSize) {
-  for (const block of blocks) {
-    if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _55 => _55.style, 'optionalAccess', _56 => _56.fontSize])) continue;
+  for (let bi = 0; bi < blocks.length; bi++) {
+    const block = blocks[bi];
+    if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _59 => _59.style, 'optionalAccess', _60 => _60.fontSize])) continue;
     const text = block.text.trim();
     if (text.length === 0 || text.length > 200) continue;
-    if (/^\d+$/.test(text)) continue;
+    if (/^\d+$/.test(text) && !isChapterNumber(block, blocks[bi + 1], medianFontSize)) continue;
     const ratio = block.style.fontSize / medianFontSize;
     let level = 0;
-    if (ratio >= _chunkCND4NRNFcjs.HEADING_RATIO_H1) level = 1;
-    else if (ratio >= _chunkCND4NRNFcjs.HEADING_RATIO_H2) level = 2;
-    else if (ratio >= _chunkCND4NRNFcjs.HEADING_RATIO_H3) level = 3;
+    if (ratio >= _chunkYNJLZVEOcjs.HEADING_RATIO_H1) level = 1;
+    else if (ratio >= _chunkYNJLZVEOcjs.HEADING_RATIO_H2) level = 2;
+    else if (ratio >= _chunkYNJLZVEOcjs.HEADING_RATIO_H3) level = 3;
     if (level > 0) {
       block.type = "heading";
       block.level = level;
@@ -2896,17 +2957,24 @@ function detectHeadings(blocks, medianFontSize) {
     }
   }
 }
+function isChapterNumber(block, next, medianFontSize) {
+  const size = _nullishCoalesce(_optionalChain([block, 'access', _61 => _61.style, 'optionalAccess', _62 => _62.fontSize]), () => ( 0));
+  const nextSize = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _63 => _63.style, 'optionalAccess', _64 => _64.fontSize]), () => ( 0));
+  const text = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _65 => _65.text, 'optionalAccess', _66 => _66.trim, 'call', _67 => _67()]), () => ( ""));
+  const a = block.bbox, b = _optionalChain([next, 'optionalAccess', _68 => _68.bbox]);
+  return /^\d{1,2}$/.test(_nullishCoalesce(_optionalChain([block, 'access', _69 => _69.text, 'optionalAccess', _70 => _70.trim, 'call', _71 => _71()]), () => ( ""))) && size >= medianFontSize * 2.5 && !!next && (next.type === "paragraph" || next.type === "heading") && next.pageNumber === block.pageNumber && nextSize >= medianFontSize * _chunkYNJLZVEOcjs.HEADING_RATIO_H1 && text.length > 0 && text.length <= 80 && !/^[\d\s.]+$/.test(text) && !!a && !!b && a.y > b.y && a.y - (b.y + b.height) <= size * 2;
+}
 function mergeStackedHeadingLines(blocks, medianFontSize) {
   for (let i = 0; i < blocks.length - 1; ) {
     const a = blocks[i], b = blocks[i + 1];
     const ab = a.bbox, bb = b.bbox;
-    const af = _nullishCoalesce(_optionalChain([a, 'access', _57 => _57.style, 'optionalAccess', _58 => _58.fontSize]), () => ( 0)), bf = _nullishCoalesce(_optionalChain([b, 'access', _59 => _59.style, 'optionalAccess', _60 => _60.fontSize]), () => ( 0));
+    const af = _nullishCoalesce(_optionalChain([a, 'access', _72 => _72.style, 'optionalAccess', _73 => _73.fontSize]), () => ( 0)), bf = _nullishCoalesce(_optionalChain([b, 'access', _74 => _74.style, 'optionalAccess', _75 => _75.fontSize]), () => ( 0));
     const gap = ab && bb ? ab.y - (bb.y + bb.height) : Infinity;
-    const centered = ab && bb && _optionalChain([a, 'access', _61 => _61.style, 'optionalAccess', _62 => _62.fontName]) === _optionalChain([b, 'access', _63 => _63.style, 'optionalAccess', _64 => _64.fontName]) && Math.abs(ab.x + ab.width / 2 - (bb.x + bb.width / 2)) <= Math.max(5, af * 0.5);
+    const centered = ab && bb && _optionalChain([a, 'access', _76 => _76.style, 'optionalAccess', _77 => _77.fontName]) === _optionalChain([b, 'access', _78 => _78.style, 'optionalAccess', _79 => _79.fontName]) && Math.abs(ab.x + ab.width / 2 - (bb.x + bb.width / 2)) <= Math.max(5, af * 0.5);
     const sameAnchor = ab && bb && Math.abs(ab.x - bb.x) <= 5;
-    const wrapped = sameAnchor && _optionalChain([a, 'access', _65 => _65.style, 'optionalAccess', _66 => _66.fontName]) === _optionalChain([b, 'access', _67 => _67.style, 'optionalAccess', _68 => _68.fontName]) && af === bf && ab.width >= bb.width * 0.95 && (_nullishCoalesce(_optionalChain([a, 'access', _69 => _69.text, 'optionalAccess', _70 => _70.trim, 'call', _71 => _71(), 'access', _72 => _72.split, 'call', _73 => _73(/\s+/), 'access', _74 => _74.length]), () => ( 0))) >= 3;
+    const wrapped = sameAnchor && _optionalChain([a, 'access', _80 => _80.style, 'optionalAccess', _81 => _81.fontName]) === _optionalChain([b, 'access', _82 => _82.style, 'optionalAccess', _83 => _83.fontName]) && af === bf && ab.width >= bb.width * 0.95 && (_nullishCoalesce(_optionalChain([a, 'access', _84 => _84.text, 'optionalAccess', _85 => _85.trim, 'call', _86 => _86(), 'access', _87 => _87.split, 'call', _88 => _88(/\s+/), 'access', _89 => _89.length]), () => ( 0))) >= 3;
     const displayStack = centered && gap <= Math.min(af, bf) * 0.3 && Math.max(af, bf) <= Math.min(af, bf) * 1.45;
-    if (a.type !== "heading" || b.type !== "heading" || !a.text || !b.text || !ab || !bb || ab.page !== bb.page || !(sameAnchor && af >= medianFontSize * 2 && bf >= medianFontSize * 2) && !wrapped && !(centered && af >= medianFontSize * 1.25 && bf >= medianFontSize * 1.25) || Math.abs(af - bf) > Math.max(af, bf) * 0.15 && !displayStack || gap < -2 || gap > Math.max(af, bf) * 0.45 || a.text.length > (centered ? 120 : 50) || b.text.length > (centered ? 120 : 50)) {
+    if (a.type !== "heading" || b.type !== "heading" || !a.text || !b.text || /^\d+$/.test(a.text.trim()) || !ab || !bb || ab.page !== bb.page || !(sameAnchor && af >= medianFontSize * 2 && bf >= medianFontSize * 2) && !wrapped && !(centered && af >= medianFontSize * 1.25 && bf >= medianFontSize * 1.25) || Math.abs(af - bf) > Math.max(af, bf) * 0.15 && !displayStack || gap < -2 || gap > Math.max(af, bf) * 0.45 || a.text.length > (centered ? 120 : 50) || b.text.length > (centered ? 120 : 50)) {
       i++;
       continue;
     }
@@ -2933,22 +3001,22 @@ function detectTypographyHeadings(blocks) {
     if (page.some((block) => block.type === "heading")) continue;
     const faceWeight = /* @__PURE__ */ new Map();
     for (const block of page) {
-      if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _75 => _75.style, 'optionalAccess', _76 => _76.fontName])) continue;
+      if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _90 => _90.style, 'optionalAccess', _91 => _91.fontName])) continue;
       const face = block.style.fontName;
       faceWeight.set(face, (_nullishCoalesce(faceWeight.get(face), () => ( 0))) + block.text.length);
     }
-    const bodyFace = _optionalChain([[...faceWeight], 'access', _77 => _77.sort, 'call', _78 => _78((a, b) => b[1] - a[1]), 'access', _79 => _79[0], 'optionalAccess', _80 => _80[0]]);
+    const bodyFace = _optionalChain([[...faceWeight], 'access', _92 => _92.sort, 'call', _93 => _93((a, b) => b[1] - a[1]), 'access', _94 => _94[0], 'optionalAccess', _95 => _95[0]]);
     if (!bodyFace || (_nullishCoalesce(faceWeight.get(bodyFace), () => ( 0))) < 250) continue;
     const sizeWeight = /* @__PURE__ */ new Map();
     for (const block of page) {
-      if (block.type !== "paragraph" || _optionalChain([block, 'access', _81 => _81.style, 'optionalAccess', _82 => _82.fontName]) !== bodyFace || !block.style.fontSize) continue;
+      if (block.type !== "paragraph" || _optionalChain([block, 'access', _96 => _96.style, 'optionalAccess', _97 => _97.fontName]) !== bodyFace || !block.style.fontSize) continue;
       sizeWeight.set(block.style.fontSize, (_nullishCoalesce(sizeWeight.get(block.style.fontSize), () => ( 0))) + block.text.length);
     }
-    const bodySize = _nullishCoalesce(_optionalChain([[...sizeWeight], 'access', _83 => _83.sort, 'call', _84 => _84((a, b) => b[1] - a[1]), 'access', _85 => _85[0], 'optionalAccess', _86 => _86[0]]), () => ( 0));
+    const bodySize = _nullishCoalesce(_optionalChain([[...sizeWeight], 'access', _98 => _98.sort, 'call', _99 => _99((a, b) => b[1] - a[1]), 'access', _100 => _100[0], 'optionalAccess', _101 => _101[0]]), () => ( 0));
     for (let i = 0; i < page.length; i++) {
       const block = page[i];
       const { text, bbox, style } = block;
-      if (block.type !== "paragraph" || !text || !bbox || !_optionalChain([style, 'optionalAccess', _87 => _87.fontName]) || !style.fontSize) continue;
+      if (block.type !== "paragraph" || !text || !bbox || !_optionalChain([style, 'optionalAccess', _102 => _102.fontName]) || !style.fontSize) continue;
       const title = text.trim();
       const numbered = /^\d+(?:\.\d+)*\.?\s+[A-Z가-힣]/.test(title);
       const faces = FACE_CHARS.get(block);
@@ -2978,16 +3046,16 @@ function detectDocumentStyleHeadings(blocks) {
   for (const page of byPage.values()) {
     const faceChars = /* @__PURE__ */ new Map();
     for (const block of page) {
-      if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _88 => _88.style, 'optionalAccess', _89 => _89.fontName])) continue;
+      if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _103 => _103.style, 'optionalAccess', _104 => _104.fontName])) continue;
       faceChars.set(block.style.fontName, (_nullishCoalesce(faceChars.get(block.style.fontName), () => ( 0))) + block.text.length);
     }
-    const bodyFace = _optionalChain([[...faceChars], 'access', _90 => _90.sort, 'call', _91 => _91((a, b) => b[1] - a[1]), 'access', _92 => _92[0], 'optionalAccess', _93 => _93[0]]);
+    const bodyFace = _optionalChain([[...faceChars], 'access', _105 => _105.sort, 'call', _106 => _106((a, b) => b[1] - a[1]), 'access', _107 => _107[0], 'optionalAccess', _108 => _108[0]]);
     if (!bodyFace || (_nullishCoalesce(faceChars.get(bodyFace), () => ( 0))) < 100) continue;
-    const labels = page.filter((b) => b.type === "paragraph" && b.text && b.bbox && _optionalChain([b, 'access', _94 => _94.style, 'optionalAccess', _95 => _95.fontName]) && b.style.fontName !== bodyFace && b.text.trim().length >= 3 && b.text.trim().length <= 40);
+    const labels = page.filter((b) => b.type === "paragraph" && b.text && b.bbox && _optionalChain([b, 'access', _109 => _109.style, 'optionalAccess', _110 => _110.fontName]) && b.style.fontName !== bodyFace && b.text.trim().length >= 3 && b.text.trim().length <= 40);
     for (const label of labels) {
-      const peers = labels.filter((b) => _optionalChain([b, 'access', _96 => _96.style, 'optionalAccess', _97 => _97.fontName]) === _optionalChain([label, 'access', _98 => _98.style, 'optionalAccess', _99 => _99.fontName]) && Math.abs(b.bbox.y - label.bbox.y) <= 3 && Math.abs(b.bbox.x - label.bbox.x) >= 60);
+      const peers = labels.filter((b) => _optionalChain([b, 'access', _111 => _111.style, 'optionalAccess', _112 => _112.fontName]) === _optionalChain([label, 'access', _113 => _113.style, 'optionalAccess', _114 => _114.fontName]) && Math.abs(b.bbox.y - label.bbox.y) <= 3 && Math.abs(b.bbox.x - label.bbox.x) >= 60);
       if (peers.length < 2) continue;
-      const value = page.find((b) => b.type === "paragraph" && b.bbox && _optionalChain([b, 'access', _100 => _100.style, 'optionalAccess', _101 => _101.fontName]) !== _optionalChain([label, 'access', _102 => _102.style, 'optionalAccess', _103 => _103.fontName]) && b.bbox.x >= label.bbox.x - 5 && b.bbox.x <= label.bbox.x + 20 && b.bbox.y < label.bbox.y && label.bbox.y - (b.bbox.y + b.bbox.height) <= 60);
+      const value = page.find((b) => b.type === "paragraph" && b.bbox && _optionalChain([b, 'access', _115 => _115.style, 'optionalAccess', _116 => _116.fontName]) !== _optionalChain([label, 'access', _117 => _117.style, 'optionalAccess', _118 => _118.fontName]) && b.bbox.x >= label.bbox.x - 5 && b.bbox.x <= label.bbox.x + 20 && b.bbox.y < label.bbox.y && label.bbox.y - (b.bbox.y + b.bbox.height) <= 60);
       if (value) {
         label.type = "heading";
         label.level = 2;
@@ -2996,7 +3064,7 @@ function detectDocumentStyleHeadings(blocks) {
     const first = page.find((b) => b.type !== "image" && b.type !== "separator");
     const second = page[page.indexOf(first) + 1];
     const third = page[page.indexOf(first) + 2];
-    if (!first || !second || !third || first.type !== "list" && first.type !== "paragraph" || second.type !== "paragraph" || !first.text || !second.text || !first.bbox || !second.bbox || !_optionalChain([first, 'access', _104 => _104.style, 'optionalAccess', _105 => _105.fontName]) || !_optionalChain([second, 'access', _106 => _106.style, 'optionalAccess', _107 => _107.fontName]) || !/^\d+(?:\.\d+)*\.\s+/.test(first.text.trim()) || first.style.fontName === bodyFace || second.style.fontName === bodyFace || first.text.length + second.text.length > 140 || Math.abs(first.bbox.x - second.bbox.x) > 30 || first.bbox.y - (second.bbox.y + second.bbox.height) > (_nullishCoalesce(first.style.fontSize, () => ( 0))) * 1.5 || _optionalChain([third, 'access', _108 => _108.style, 'optionalAccess', _109 => _109.fontName]) !== bodyFace) continue;
+    if (!first || !second || !third || first.type !== "list" && first.type !== "paragraph" || second.type !== "paragraph" || !first.text || !second.text || !first.bbox || !second.bbox || !_optionalChain([first, 'access', _119 => _119.style, 'optionalAccess', _120 => _120.fontName]) || !_optionalChain([second, 'access', _121 => _121.style, 'optionalAccess', _122 => _122.fontName]) || !/^\d+(?:\.\d+)*\.\s+/.test(first.text.trim()) || first.style.fontName === bodyFace || second.style.fontName === bodyFace || first.text.length + second.text.length > 140 || Math.abs(first.bbox.x - second.bbox.x) > 30 || first.bbox.y - (second.bbox.y + second.bbox.height) > (_nullishCoalesce(first.style.fontSize, () => ( 0))) * 1.5 || _optionalChain([third, 'access', _123 => _123.style, 'optionalAccess', _124 => _124.fontName]) !== bodyFace) continue;
     first.type = "heading";
     first.level = 1;
     first.text = `${first.text.trim()} ${second.text.trim()}`;
@@ -3020,18 +3088,18 @@ function detectSiblingStyleHeadings(blocks) {
   for (const page of byPage.values()) {
     const bodyChars = /* @__PURE__ */ new Map();
     for (const block of page) {
-      if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _110 => _110.style, 'optionalAccess', _111 => _111.fontName])) continue;
+      if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _125 => _125.style, 'optionalAccess', _126 => _126.fontName])) continue;
       const face = block.style.fontName;
       bodyChars.set(face, (_nullishCoalesce(bodyChars.get(face), () => ( 0))) + block.text.length);
     }
-    const bodyFace = _optionalChain([[...bodyChars], 'access', _112 => _112.sort, 'call', _113 => _113((a, b) => b[1] - a[1]), 'access', _114 => _114[0], 'optionalAccess', _115 => _115[0]]);
-    const anchors = page.filter((b) => b.type === "heading" && b.text && _optionalChain([b, 'access', _116 => _116.style, 'optionalAccess', _117 => _117.fontName]) && b.style.fontName !== bodyFace);
+    const bodyFace = _optionalChain([[...bodyChars], 'access', _127 => _127.sort, 'call', _128 => _128((a, b) => b[1] - a[1]), 'access', _129 => _129[0], 'optionalAccess', _130 => _130[0]]);
+    const anchors = page.filter((b) => b.type === "heading" && b.text && _optionalChain([b, 'access', _131 => _131.style, 'optionalAccess', _132 => _132.fontName]) && b.style.fontName !== bodyFace);
     for (const anchor of anchors) {
       const numbered = /^([A-Z])\.\d+\s+/.exec(anchor.text.trim());
       const lettered = /^([A-Z])\s+/.exec(anchor.text.trim());
-      const peers = anchors.filter((b) => _optionalChain([b, 'access', _118 => _118.style, 'optionalAccess', _119 => _119.fontName]) === _optionalChain([anchor, 'access', _120 => _120.style, 'optionalAccess', _121 => _121.fontName]) && Math.abs((_nullishCoalesce(_optionalChain([b, 'access', _122 => _122.style, 'optionalAccess', _123 => _123.fontSize]), () => ( 0))) - (_nullishCoalesce(_optionalChain([anchor, 'access', _124 => _124.style, 'optionalAccess', _125 => _125.fontSize]), () => ( 0)))) < 0.5);
+      const peers = anchors.filter((b) => _optionalChain([b, 'access', _133 => _133.style, 'optionalAccess', _134 => _134.fontName]) === _optionalChain([anchor, 'access', _135 => _135.style, 'optionalAccess', _136 => _136.fontName]) && Math.abs((_nullishCoalesce(_optionalChain([b, 'access', _137 => _137.style, 'optionalAccess', _138 => _138.fontSize]), () => ( 0))) - (_nullishCoalesce(_optionalChain([anchor, 'access', _139 => _139.style, 'optionalAccess', _140 => _140.fontSize]), () => ( 0)))) < 0.5);
       if (peers.length < 2 && !numbered && !lettered) continue;
-      const candidates = page.filter((b) => b.type === "paragraph" && b.text && b.bbox && _optionalChain([b, 'access', _126 => _126.style, 'optionalAccess', _127 => _127.fontName]) === _optionalChain([anchor, 'access', _128 => _128.style, 'optionalAccess', _129 => _129.fontName]) && Math.abs((_nullishCoalesce(_optionalChain([b, 'access', _130 => _130.style, 'optionalAccess', _131 => _131.fontSize]), () => ( 0))) - (_nullishCoalesce(_optionalChain([anchor, 'access', _132 => _132.style, 'optionalAccess', _133 => _133.fontSize]), () => ( 0)))) < 0.5 && b.text.trim().length >= 3 && b.text.trim().length <= 80 && b.bbox.height <= (_nullishCoalesce(_optionalChain([b, 'access', _134 => _134.style, 'optionalAccess', _135 => _135.fontSize]), () => ( 0))) * 1.6 && /^[A-Z]/.test(b.text.trim()) && !/[.!?:;,]$/.test(b.text.trim()) && !/^(?:Table|Figure|Fig\.?|Appendix)\s+\d/i.test(b.text.trim()));
+      const candidates = page.filter((b) => b.type === "paragraph" && b.text && b.bbox && _optionalChain([b, 'access', _141 => _141.style, 'optionalAccess', _142 => _142.fontName]) === _optionalChain([anchor, 'access', _143 => _143.style, 'optionalAccess', _144 => _144.fontName]) && Math.abs((_nullishCoalesce(_optionalChain([b, 'access', _145 => _145.style, 'optionalAccess', _146 => _146.fontSize]), () => ( 0))) - (_nullishCoalesce(_optionalChain([anchor, 'access', _147 => _147.style, 'optionalAccess', _148 => _148.fontSize]), () => ( 0)))) < 0.5 && b.text.trim().length >= 3 && b.text.trim().length <= 80 && b.bbox.height <= (_nullishCoalesce(_optionalChain([b, 'access', _149 => _149.style, 'optionalAccess', _150 => _150.fontSize]), () => ( 0))) * 1.6 && /^[A-Z]/.test(b.text.trim()) && !/[.!?:;,]$/.test(b.text.trim()) && !/^(?:Table|Figure|Fig\.?|Appendix)\s+\d/i.test(b.text.trim()));
       const siblings = candidates.filter((b) => numbered ? new RegExp(`^${numbered[1]}\\.\\d+\\s+`).test(b.text.trim()) : lettered ? /^[A-Z]\s+/.test(b.text.trim()) && Math.abs(anchor.bbox.x - b.bbox.x) < 8 : peers.some((h) => h.bbox && Math.abs(h.bbox.x - b.bbox.x) < 8));
       if (siblings.length < (numbered || lettered ? 1 : 2)) continue;
       for (const sibling of siblings) {
@@ -3050,11 +3118,11 @@ function detectRepeatedPageLabels(blocks) {
   }
   for (const page of byPage.values()) {
     const bodyChars = /* @__PURE__ */ new Map();
-    for (const b of page) if (b.type === "paragraph" && b.text && _optionalChain([b, 'access', _136 => _136.style, 'optionalAccess', _137 => _137.fontName])) {
+    for (const b of page) if (b.type === "paragraph" && b.text && _optionalChain([b, 'access', _151 => _151.style, 'optionalAccess', _152 => _152.fontName])) {
       bodyChars.set(b.style.fontName, (_nullishCoalesce(bodyChars.get(b.style.fontName), () => ( 0))) + b.text.length);
     }
-    const bodyFace = _optionalChain([[...bodyChars], 'access', _138 => _138.sort, 'call', _139 => _139((a, b) => b[1] - a[1]), 'access', _140 => _140[0], 'optionalAccess', _141 => _141[0]]);
-    const candidates = page.filter((b) => b.type === "paragraph" && b.text && b.bbox && _optionalChain([b, 'access', _142 => _142.style, 'optionalAccess', _143 => _143.fontName]) && b.style.fontName !== bodyFace && b.bbox.height <= (_nullishCoalesce(b.style.fontSize, () => ( 0))) * 1.6 && b.text.trim().length >= 5 && b.text.trim().length <= 80);
+    const bodyFace = _optionalChain([[...bodyChars], 'access', _153 => _153.sort, 'call', _154 => _154((a, b) => b[1] - a[1]), 'access', _155 => _155[0], 'optionalAccess', _156 => _156[0]]);
+    const candidates = page.filter((b) => b.type === "paragraph" && b.text && b.bbox && _optionalChain([b, 'access', _157 => _157.style, 'optionalAccess', _158 => _158.fontName]) && b.style.fontName !== bodyFace && b.bbox.height <= (_nullishCoalesce(b.style.fontSize, () => ( 0))) * 1.6 && b.text.trim().length >= 5 && b.text.trim().length <= 80);
     const groups = /* @__PURE__ */ new Map();
     for (const b of candidates) {
       const key = `${b.style.fontName}:${b.style.fontSize}`;
@@ -3088,21 +3156,21 @@ function detectPageLeadHeadings(blocks) {
     byPage.set(_nullishCoalesce(b.pageNumber, () => ( 0)), page);
   }
   for (const page of byPage.values()) {
-    const content = page.filter((b) => b.type !== "image" && b.type !== "separator");
+    const content = page.filter((b) => b.type !== "image" && b.type !== "separator" && !(b.type === "paragraph" && !b.style));
     const [first, second, third] = content;
-    if (!_optionalChain([first, 'optionalAccess', _144 => _144.bbox]) || !first.text || !_optionalChain([first, 'access', _145 => _145.style, 'optionalAccess', _146 => _146.fontSize]) || !second) continue;
+    if (!_optionalChain([first, 'optionalAccess', _159 => _159.bbox]) || !first.text || !_optionalChain([first, 'access', _160 => _160.style, 'optionalAccess', _161 => _161.fontSize]) || !second) continue;
     const firstText = first.text.trim();
     const captionLike = /^(?:Figure|Fig\.?|Table|표|그림)\s*\d/i;
     const hasImage = page.some((b) => b.type === "image");
-    const firstIsTitle = first.type === "paragraph" && firstText.length >= 5 && firstText.length <= 80 && first.bbox.height <= first.style.fontSize * 1.6 && (/^(?:CONTENTS|Table of Contents)$/i.test(firstText) || hasImage && /^[A-Z]/.test(firstText) && !captionLike.test(firstText) && second.type === "paragraph" && /^[A-Z]/.test(_nullishCoalesce(_optionalChain([second, 'access', _147 => _147.text, 'optionalAccess', _148 => _148.trim, 'call', _149 => _149()]), () => ( ""))) && (_nullishCoalesce(_optionalChain([second, 'access', _150 => _150.text, 'optionalAccess', _151 => _151.length]), () => ( 0))) >= 100 && second.bbox && first.bbox.y - (second.bbox.y + second.bbox.height) >= first.style.fontSize);
+    const firstIsTitle = first.type === "paragraph" && firstText.length >= 5 && firstText.length <= 80 && first.bbox.height <= first.style.fontSize * 1.6 && (/^(?:CONTENTS|Table of Contents)$/i.test(firstText) || hasImage && /^[A-Z]/.test(firstText) && !captionLike.test(firstText) && second.type === "paragraph" && /^[A-Z]/.test(_nullishCoalesce(_optionalChain([second, 'access', _162 => _162.text, 'optionalAccess', _163 => _163.trim, 'call', _164 => _164()]), () => ( ""))) && (_nullishCoalesce(_optionalChain([second, 'access', _165 => _165.text, 'optionalAccess', _166 => _166.length]), () => ( 0))) >= 100 && second.bbox && first.bbox.y - (second.bbox.y + second.bbox.height) >= first.style.fontSize);
     const firstAboveProse = first.type === "paragraph" && firstText.length >= 5 && firstText.length <= 80 && first.bbox.height <= first.style.fontSize * 2 && second.type === "paragraph" && second.bbox && second.bbox.height >= first.style.fontSize * 4 && Math.abs(first.bbox.x - second.bbox.x) <= first.style.fontSize * 2 && first.bbox.y - (second.bbox.y + second.bbox.height) >= first.style.fontSize * 2 && !captionLike.test(firstText) && !/^(?:doi:|https?:|www\.)/i.test(firstText);
-    const firstIsSection = first.type === "list" && /^\d+\.\s+[A-Z][A-Z\s]{12,}$/.test(firstText) && second.type === "paragraph" && (_nullishCoalesce(_optionalChain([second, 'access', _152 => _152.text, 'optionalAccess', _153 => _153.length]), () => ( 0))) >= 100;
+    const firstIsSection = first.type === "list" && /^\d+\.\s+[A-Z][A-Z\s]{12,}$/.test(firstText) && second.type === "paragraph" && (_nullishCoalesce(_optionalChain([second, 'access', _167 => _167.text, 'optionalAccess', _168 => _168.length]), () => ( 0))) >= 100;
     if (firstIsTitle || firstAboveProse || firstIsSection) {
       first.type = "heading";
       first.level = 1;
     }
-    if (first.type !== "heading" || captionLike.test(firstText) || second.type !== "paragraph" || !second.text || !second.bbox || !_optionalChain([second, 'access', _154 => _154.style, 'optionalAccess', _155 => _155.fontName]) || !second.style.fontSize || !_optionalChain([third, 'optionalAccess', _156 => _156.text]) || !third.bbox || second.text.trim().length < 5 || second.text.trim().length > 80 || captionLike.test(second.text.trim()) || /^(?:doi:|https?:)/i.test(second.text.trim()) || second.bbox.height > second.style.fontSize * 1.6 || Math.abs(second.bbox.x - third.bbox.x) > 30) continue;
-    const distinctFace = _optionalChain([third, 'access', _157 => _157.style, 'optionalAccess', _158 => _158.fontName]) !== second.style.fontName;
+    if (first.type !== "heading" || captionLike.test(firstText) || second.type !== "paragraph" || !second.text || !second.bbox || !_optionalChain([second, 'access', _169 => _169.style, 'optionalAccess', _170 => _170.fontName]) || !second.style.fontSize || !_optionalChain([third, 'optionalAccess', _171 => _171.text]) || !third.bbox || second.text.trim().length < 5 || second.text.trim().length > 80 || captionLike.test(second.text.trim()) || /^(?:doi:|https?:)/i.test(second.text.trim()) || second.bbox.height > second.style.fontSize * 1.6 || Math.abs(second.bbox.x - third.bbox.x) > 30) continue;
+    const distinctFace = _optionalChain([third, 'access', _172 => _172.style, 'optionalAccess', _173 => _173.fontName]) !== second.style.fontName;
     const namedContents = /^Table of Contents$/i.test(second.text.trim());
     if (distinctFace && third.text.length >= 60 || namedContents) {
       second.type = "heading";
@@ -3113,24 +3181,24 @@ function detectPageLeadHeadings(blocks) {
 function refineDocumentStyleHeadings(blocks) {
   const faceChars = /* @__PURE__ */ new Map();
   for (const block of blocks) {
-    if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _159 => _159.style, 'optionalAccess', _160 => _160.fontName])) continue;
+    if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _174 => _174.style, 'optionalAccess', _175 => _175.fontName])) continue;
     faceChars.set(block.style.fontName, (_nullishCoalesce(faceChars.get(block.style.fontName), () => ( 0))) + block.text.length);
   }
   const [bodyFace, bodyChars] = _nullishCoalesce([...faceChars].sort((a, b) => b[1] - a[1])[0], () => ( []));
   if (!bodyFace || !bodyChars || bodyChars < 300) return;
   for (let i = 0; i < blocks.length; i++) {
     const block = blocks[i];
-    const text = _nullishCoalesce(_optionalChain([block, 'access', _161 => _161.text, 'optionalAccess', _162 => _162.trim, 'call', _163 => _163()]), () => ( ""));
+    const text = _nullishCoalesce(_optionalChain([block, 'access', _176 => _176.text, 'optionalAccess', _177 => _177.trim, 'call', _178 => _178()]), () => ( ""));
     if (block.type === "heading" && /(?:https?:\/\/|www\.|^arxiv:|^doi:)/i.test(text)) {
       block.type = "paragraph";
       block.level = void 0;
       continue;
     }
     const box = block.bbox, style = block.style;
-    if (block.type !== "paragraph" || !box || !_optionalChain([style, 'optionalAccess', _164 => _164.fontName]) || !style.fontSize || style.fontName === bodyFace || (_nullishCoalesce(faceChars.get(style.fontName), () => ( 0))) > bodyChars * 0.25 || text.length < 3 || text.length > 70 || text.includes("\n") || box.height > style.fontSize * 1.6 || /^(?:table|figure|fig\.?|source|doi:|arxiv:|https?:|www\.|[•●○▪▫⮚*∗†])/i.test(text) || /(?:https?:\/\/|www\.|@|[=¼≪þ])/i.test(text) || /^definition\s+\d+[.:]?\s+.*\b(?:is|are|means)\b/i.test(text) || /^[\d\s.,:%+\-–]+$/.test(text)) continue;
+    if (block.type !== "paragraph" || !box || !_optionalChain([style, 'optionalAccess', _179 => _179.fontName]) || !style.fontSize || style.fontName === bodyFace || (_nullishCoalesce(faceChars.get(style.fontName), () => ( 0))) > bodyChars * 0.25 || text.length < 3 || text.length > 70 || text.includes("\n") || box.height > style.fontSize * 1.6 || /^(?:table|figure|fig\.?|source|doi:|arxiv:|https?:|www\.|[•●○▪▫⮚*∗†])/i.test(text) || /(?:https?:\/\/|www\.|@|[=¼≪þ])/i.test(text) || /^definition\s+\d+[.:]?\s+.*\b(?:is|are|means)\b/i.test(text) || /^[\d\s.,:%+\-–]+$/.test(text)) continue;
     const previous = blocks[i - 1], next = blocks[i + 1];
-    const pb = _optionalChain([previous, 'optionalAccess', _165 => _165.bbox]), nb = _optionalChain([next, 'optionalAccess', _166 => _166.bbox]);
-    if (!pb || !nb || previous.pageNumber !== block.pageNumber || next.pageNumber !== block.pageNumber || previous.type !== "paragraph" && previous.type !== "heading" || next.type !== "paragraph" || _optionalChain([next, 'access', _167 => _167.style, 'optionalAccess', _168 => _168.fontName]) !== bodyFace || !next.text || next.text.length < 40 || style.fontSize < (_nullishCoalesce(next.style.fontSize, () => ( 0))) * 0.95 || pb.y - (box.y + box.height) < style.fontSize * 0.5 || box.y - (nb.y + nb.height) < 0 || box.y - (nb.y + nb.height) > style.fontSize * 2 || box.x + box.width / 2 < nb.x - 10 || box.x + box.width / 2 > nb.x + nb.width + 10) continue;
+    const pb = _optionalChain([previous, 'optionalAccess', _180 => _180.bbox]), nb = _optionalChain([next, 'optionalAccess', _181 => _181.bbox]);
+    if (!pb || !nb || previous.pageNumber !== block.pageNumber || next.pageNumber !== block.pageNumber || previous.type !== "paragraph" && previous.type !== "heading" || next.type !== "paragraph" || _optionalChain([next, 'access', _182 => _182.style, 'optionalAccess', _183 => _183.fontName]) !== bodyFace || !next.text || next.text.length < 40 || style.fontSize < (_nullishCoalesce(next.style.fontSize, () => ( 0))) * 0.95 || pb.y - (box.y + box.height) < style.fontSize * 0.5 || box.y - (nb.y + nb.height) < 0 || box.y - (nb.y + nb.height) > style.fontSize * 2 || box.x + box.width / 2 < nb.x - 10 || box.x + box.width / 2 > nb.x + nb.width + 10) continue;
     block.type = "heading";
     block.level = 2;
   }
@@ -3180,7 +3248,7 @@ function detectMarkerHeadings(blocks) {
       block.level = 4;
       continue;
     }
-    if (/^[가-힣]{2,6}$/.test(text) && _optionalChain([block, 'access', _169 => _169.style, 'optionalAccess', _170 => _170.fontSize])) {
+    if (/^[가-힣]{2,6}$/.test(text) && _optionalChain([block, 'access', _184 => _184.style, 'optionalAccess', _185 => _185.fontSize])) {
       const prev = blocks[i - 1];
       const next = blocks[i + 1];
       const prevIsStructural = !prev || prev.type === "table" || prev.type === "heading" || prev.type === "separator";
@@ -3221,7 +3289,7 @@ function detectTableCaptions(blocks) {
       block.table.caption = blocks[i - 1].text.trim();
       blocks.splice(i - 1, 1);
       i--;
-    } else if (isCaptionCandidate(blocks[i + 1], block) && _optionalChain([blocks, 'access', _171 => _171[i + 2], 'optionalAccess', _172 => _172.type]) !== "table") {
+    } else if (isCaptionCandidate(blocks[i + 1], block) && _optionalChain([blocks, 'access', _186 => _186[i + 2], 'optionalAccess', _187 => _187.type]) !== "table") {
       block.table.caption = blocks[i + 1].text.trim();
       blocks.splice(i + 1, 1);
     }
@@ -3435,7 +3503,7 @@ function removeHeaderFooterBlocks(blocks, pageHeights, warnings) {
   const bottomEntries = [];
   for (let bi = 0; bi < blocks.length; bi++) {
     const b = blocks[bi];
-    if (!b.bbox || !b.pageNumber || !_optionalChain([b, 'access', _173 => _173.text, 'optionalAccess', _174 => _174.trim, 'call', _175 => _175()])) continue;
+    if (!b.bbox || !b.pageNumber || !_optionalChain([b, 'access', _188 => _188.text, 'optionalAccess', _189 => _189.trim, 'call', _190 => _190()])) continue;
     const ph = pageHeights.get(b.bbox.page) || pageHeights.get(b.pageNumber);
     if (!ph) continue;
     const blockTop = ph - (b.bbox.y + b.bbox.height);
@@ -3458,7 +3526,7 @@ function removeHeaderFooterBlocks(blocks, pageHeights, warnings) {
     }
     const repeatedPatterns = /* @__PURE__ */ new Set();
     for (const [p, count] of patternCount) {
-      if (count >= MIN_REPEAT && (_nullishCoalesce(_optionalChain([patternPages, 'access', _176 => _176.get, 'call', _177 => _177(p), 'optionalAccess', _178 => _178.size]), () => ( 0))) >= MIN_REPEAT) {
+      if (count >= MIN_REPEAT && (_nullishCoalesce(_optionalChain([patternPages, 'access', _191 => _191.get, 'call', _192 => _192(p), 'optionalAccess', _193 => _193.size]), () => ( 0))) >= MIN_REPEAT) {
         repeatedPatterns.add(p);
       }
     }
@@ -3478,7 +3546,7 @@ function removeHeaderFooterBlocks(blocks, pageHeights, warnings) {
         if (bi === e.blockIdx) continue;
         const o = blocks[bi];
         if (!o.bbox || o.bbox.page !== cb.page) continue;
-        const oNorm = _optionalChain([o, 'access', _179 => _179.text, 'optionalAccess', _180 => _180.trim, 'call', _181 => _181(), 'access', _182 => _182.replace, 'call', _183 => _183(/\d+/g, "#")]);
+        const oNorm = _optionalChain([o, 'access', _194 => _194.text, 'optionalAccess', _195 => _195.trim, 'call', _196 => _196(), 'access', _197 => _197.replace, 'call', _198 => _198(/\d+/g, "#")]);
         if (oNorm && repeatedPatterns.has(oNorm)) continue;
         const overlap2 = Math.min(cb.y + cb.height, o.bbox.y + o.bbox.height) - Math.max(cb.y, o.bbox.y);
         if (overlap2 >= cb.height * 0.5) {
@@ -3521,7 +3589,7 @@ function recordCellLines(cell, items) {
 function markImageCell(cell) {
   IMAGE_CELLS.add(cell);
 }
-var emptyCell = (cell) => !cell || !_optionalChain([cell, 'access', _184 => _184.text, 'optionalAccess', _185 => _185.trim, 'call', _186 => _186()]) && !_optionalChain([cell, 'access', _187 => _187.blocks, 'optionalAccess', _188 => _188.length]) && !IMAGE_CELLS.has(cell);
+var emptyCell = (cell) => !cell || !_optionalChain([cell, 'access', _199 => _199.text, 'optionalAccess', _200 => _200.trim, 'call', _201 => _201()]) && !_optionalChain([cell, 'access', _202 => _202.blocks, 'optionalAccess', _203 => _203.length]) && !IMAGE_CELLS.has(cell);
 function trimTable(table) {
   let cols = table.cols;
   while (cols > 0 && table.cells.every((row) => emptyCell(row[cols - 1]))) cols--;
@@ -3537,7 +3605,7 @@ function trimTable(table) {
 function trimTrailingEmptyTableCols(blocks) {
   for (const b of blocks) {
     if (b.type !== "table" || !b.table) continue;
-    for (const row of b.table.cells) for (const cell of row) if (_optionalChain([cell, 'access', _189 => _189.blocks, 'optionalAccess', _190 => _190.length])) trimTrailingEmptyTableCols(cell.blocks);
+    for (const row of b.table.cells) for (const cell of row) if (_optionalChain([cell, 'access', _204 => _204.blocks, 'optionalAccess', _205 => _205.length])) trimTrailingEmptyTableCols(cell.blocks);
     trimTable(b.table);
   }
 }
@@ -3557,7 +3625,7 @@ function mergeSliverColumns(grid, colXs, maxWidth) {
     }
     const ok = owner.every((row) => {
       const o = row[c];
-      return !!o && (o.colSpan > 1 || !o.text.trim() && !_optionalChain([o, 'access', _191 => _191.blocks, 'optionalAccess', _192 => _192.length]));
+      return !!o && (o.colSpan > 1 || !o.text.trim() && !_optionalChain([o, 'access', _206 => _206.blocks, 'optionalAccess', _207 => _207.length]));
     });
     if (!ok) continue;
     const spanning = /* @__PURE__ */ new Set();
@@ -3623,7 +3691,7 @@ function isProseTable(table) {
   if (!(long > 0 && long >= total * 0.5 && longRows >= rows * 0.4)) return false;
   for (let c = 0; c < table.cols; c++) {
     const filled = table.cells.map((row) => row[c]).filter((cell) => cell && cell.colSpan < table.cols && cell.text.trim()).map((cell) => cell.text.trim());
-    if (filled.length >= 3 && filled.length >= rows * 0.6 && filled.every((text) => text.length <= 30 && (_nullishCoalesce(_optionalChain([text, 'access', _193 => _193.match, 'call', _194 => _194(/\p{L}/gu), 'optionalAccess', _195 => _195.length]), () => ( 0))) >= 2)) return false;
+    if (filled.length >= 3 && filled.length >= rows * 0.6 && filled.every((text) => text.length <= 30 && (_nullishCoalesce(_optionalChain([text, 'access', _208 => _208.match, 'call', _209 => _209(/\p{L}/gu), 'optionalAccess', _210 => _210.length]), () => ( 0))) >= 2)) return false;
   }
   return true;
 }
@@ -3650,7 +3718,7 @@ function isChartTable(table) {
   const numeric = texts.filter((text) => NUMERIC_CELL.test(text)).length;
   if (numeric < texts.length * 0.5) return false;
   for (let c = 0; c < table.cols; c++) {
-    const ticks = table.cells.flatMap((row) => (_nullishCoalesce(_optionalChain([row, 'access', _196 => _196[c], 'optionalAccess', _197 => _197.text]), () => ( ""))).split(/\s+/)).filter((token) => /^-?[\d,]+(?:\.\d+)?%?$/.test(token)).map((token) => Number(token.replace(/[,%]/g, "")));
+    const ticks = table.cells.flatMap((row) => (_nullishCoalesce(_optionalChain([row, 'access', _211 => _211[c], 'optionalAccess', _212 => _212.text]), () => ( ""))).split(/\s+/)).filter((token) => /^-?[\d,]+(?:\.\d+)?%?$/.test(token)).map((token) => Number(token.replace(/[,%]/g, "")));
     if (hasValueAxis(ticks)) return true;
   }
   const values = texts.filter((text) => text.split(/\s+/).every((token) => /^[-−–]?[\d,]*\.?\d+%?$/.test(token))).length;
@@ -3798,7 +3866,7 @@ function detectAlignedTwoColumnTables(rows, pageNum) {
       continue;
     }
     const top = rows[i - 1];
-    const title = _optionalChain([top, 'optionalAccess', _198 => _198.items, 'access', _199 => _199.length]) === 1 && Math.abs(top.items[0].x - first[0].x) <= 5 && top.items[0].text.length <= 65 && top.y - first[0].y <= Math.max(24, first[0].fontSize * 2.8) ? top : void 0;
+    const title = _optionalChain([top, 'optionalAccess', _213 => _213.items, 'access', _214 => _214.length]) === 1 && Math.abs(top.items[0].x - first[0].x) <= 5 && top.items[0].text.length <= 65 && top.y - first[0].y <= Math.max(24, first[0].fontSize * 2.8) ? top : void 0;
     const tableRows = title ? [title, ...body] : body;
     const cells = tableRows.map((row, idx) => {
       if (title && idx === 0) return [{ text: row.items[0].text, colSpan: 2, rowSpan: 1 }, { text: "", colSpan: 1, rowSpan: 1 }];
@@ -3854,12 +3922,12 @@ function isTwoColumnProse(r) {
   if (t.rows < 8) return false;
   const dense = [];
   for (let c = 0; c < t.cols; c++) {
-    const filled = t.cells.filter((row) => _optionalChain([row, 'access', _200 => _200[c], 'optionalAccess', _201 => _201.text, 'access', _202 => _202.trim, 'call', _203 => _203()])).length;
+    const filled = t.cells.filter((row) => _optionalChain([row, 'access', _215 => _215[c], 'optionalAccess', _216 => _216.text, 'access', _217 => _217.trim, 'call', _218 => _218()])).length;
     if (filled / t.rows >= 0.3) dense.push(c);
   }
   if (dense.length !== 2) return false;
   for (const c of dense) {
-    const lens = t.cells.map((row) => _nullishCoalesce(_optionalChain([row, 'access', _204 => _204[c], 'optionalAccess', _205 => _205.text, 'access', _206 => _206.replace, 'call', _207 => _207(/\s+/g, ""), 'access', _208 => _208.length]), () => ( 0))).filter((n) => n > 0);
+    const lens = t.cells.map((row) => _nullishCoalesce(_optionalChain([row, 'access', _219 => _219[c], 'optionalAccess', _220 => _220.text, 'access', _221 => _221.replace, 'call', _222 => _222(/\s+/g, ""), 'access', _223 => _223.length]), () => ( 0))).filter((n) => n > 0);
     const avg = lens.reduce((s, v) => s + v, 0) / (lens.length || 1);
     if (avg < 12) return false;
   }
@@ -4255,7 +4323,7 @@ function assignRowItems(items, columns, numCols, headerFace) {
   const runs = [];
   for (const it of sorted) {
     const last = runs[runs.length - 1];
-    const prev = _optionalChain([last, 'optionalAccess', _209 => _209[last.length - 1]]);
+    const prev = _optionalChain([last, 'optionalAccess', _224 => _224[last.length - 1]]);
     if (prev && it.x - (prev.x + prev.w) <= Math.max(0.5, it.fontSize * 0.05) && !it.hasSpaceBefore) last.push(it);
     else runs.push([it]);
   }
@@ -4329,7 +4397,7 @@ function buildClusterTable(rows, columns, pageNum) {
     () => Array.from({ length: numCols }, () => ({ text: "", colSpan: 1, rowSpan: 1 }))
   );
   const usedItems = /* @__PURE__ */ new Set();
-  const headFaces = new Set(_optionalChain([rows, 'access', _210 => _210[0], 'optionalAccess', _211 => _211.items, 'access', _212 => _212.map, 'call', _213 => _213((i) => i.fontName)]));
+  const headFaces = new Set(_optionalChain([rows, 'access', _225 => _225[0], 'optionalAccess', _226 => _226.items, 'access', _227 => _227.map, 'call', _228 => _228((i) => i.fontName)]));
   const headerFace = headFaces.size === 1 && rows.slice(1).every((row) => row.items.every((i) => !headFaces.has(i.fontName))) ? [...headFaces][0] : void 0;
   for (let r = 0; r < numRows; r++) {
     const row = rows[r];
@@ -4359,7 +4427,7 @@ function buildClusterTable(rows, columns, pageNum) {
     const nonEmptyCols = cells[r].filter((c) => c.text.trim()).length;
     if (nonEmptyCols !== 1) continue;
     if (cells[r][0].text.trim() !== "") continue;
-    const contentText = _optionalChain([cells, 'access', _214 => _214[r], 'access', _215 => _215.find, 'call', _216 => _216((c) => c.text.trim()), 'optionalAccess', _217 => _217.text, 'access', _218 => _218.trim, 'call', _219 => _219()]) || "";
+    const contentText = _optionalChain([cells, 'access', _229 => _229[r], 'access', _230 => _230.find, 'call', _231 => _231((c) => c.text.trim()), 'optionalAccess', _232 => _232.text, 'access', _233 => _233.trim, 'call', _234 => _234()]) || "";
     if (/^[○●▶\-·]/.test(contentText)) continue;
     for (let pr = r - 1; pr >= 0; pr--) {
       if (cells[pr].some((c) => c.text.trim())) {
@@ -4485,9 +4553,9 @@ function xyCutOrder(items, gapThreshold, depth = 0) {
   const hCut = findHorizontalCut(items);
   const vCut = findVerticalCutWithOutlierFilter(items, minGap);
   const hValid = hCut.gap >= minGap;
-  const vValid = vCut.gap >= minGap || vCut.gap >= PROSE_GUTTER_MIN_GAP && isProseGutter(items, vCut.position);
+  const vValid = (vCut.gap >= minGap || vCut.gap >= PROSE_GUTTER_MIN_GAP && isProseGutter(items, vCut.position)) && !splitsSpacedLabel(items, vCut.position);
   let useHorizontal;
-  if (hValid && vValid) useHorizontal = vCut.gap <= hCut.gap * 1.5;
+  if (hValid && vValid) useHorizontal = vCut.gap <= hCut.gap * 1.5 || staggeredSides(items, vCut.position);
   else if (hValid) useHorizontal = true;
   else if (vValid) useHorizontal = false;
   else return _nullishCoalesce(splitEdgeSpannedColumns(items, gapThreshold, depth), () => ( [items]));
@@ -4506,6 +4574,33 @@ function xyCutOrder(items, gapThreshold, depth = 0) {
   }
   return [items];
 }
+function splitsSpacedLabel(items, cutX) {
+  const lone = (i) => /^[가-힣]$/.test(i.text.trim());
+  const sameRow = (a, b) => Math.min(a.y, b.y) > Math.max(a.y - a.h, b.y - b.h);
+  const left = items.filter((i) => i.x + i.w / 2 < cutX);
+  const right = items.filter((i) => i.x + i.w / 2 >= cutX);
+  let spaced = 0;
+  for (const r of right) {
+    if (right.some((o) => o !== r && o.x < r.x && sameRow(o, r))) continue;
+    const row = left.filter((l2) => sameRow(l2, r));
+    if (row.length === 0) continue;
+    const l = row.reduce((a, b) => b.x + b.w > a.x + a.w ? b : a);
+    if (!lone(l) || !lone(r)) return false;
+    spaced++;
+  }
+  return spaced > 0;
+}
+function staggeredSides(items, cutX) {
+  const left = items.filter((i) => i.x + i.w / 2 < cutX);
+  const right = items.filter((i) => i.x + i.w / 2 >= cutX);
+  if (left.length * right.length > 2e5) return false;
+  const shareRow = (a, b) => Math.min(a.y, b.y) > Math.max(a.y - a.h, b.y - b.h);
+  if (left.some((a) => right.some((b) => shareRow(a, b)))) return false;
+  const sides = [...items].sort((a, b) => b.y - a.y).map((i) => i.x + i.w / 2 < cutX);
+  let switches = 0;
+  for (let k = 1; k < sides.length; k++) if (sides[k] !== sides[k - 1]) switches++;
+  return switches >= 3;
+}
 function identifyCrossLayoutItems(items) {
   const cross = /* @__PURE__ */ new Set();
   if (items.length < 3) return cross;
@@ -4910,16 +5005,16 @@ 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 = _chunkFBZCWSPNcjs.safeMax.call(void 0, gaps);
+  const maxGap = _chunkJ6ZU7AIRcjs.safeMax.call(void 0, gaps);
   const avgLen = items.reduce((s, i) => s + i.text.length, 0) / items.length;
   if (items.length >= 4 && maxGap < 40 && avgLen < 5) return true;
-  const minGap = _chunkFBZCWSPNcjs.safeMin.call(void 0, gaps);
+  const minGap = _chunkJ6ZU7AIRcjs.safeMin.call(void 0, gaps);
   return minGap > 0 && maxGap - minGap <= Math.max(2, maxGap * 0.15) && items.filter((i) => /^\P{L}*\p{Ll}/u.test(i.text)).length * 2 >= items.length;
 }
 function detectColumns(yLines) {
   const allItems = yLines.flat();
   if (allItems.length === 0) return null;
-  const pageWidth = _chunkFBZCWSPNcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkFBZCWSPNcjs.safeMin.call(void 0, allItems.map((i) => i.x));
+  const pageWidth = _chunkJ6ZU7AIRcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkJ6ZU7AIRcjs.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++) {
@@ -5305,6 +5400,19 @@ function commonEnd(ends) {
   }
   return bestN >= MIN_REACHING ? best : null;
 }
+function splitBodies(g, chained) {
+  if (g.length < MIN_RULES * 2) return [g];
+  const x1 = Math.min(...g.map((r) => r.x1)), x2 = Math.max(...g.map((r) => r.x2));
+  const sorted = [...g].sort((a, b) => b.y1 - a.y1);
+  const parts = [[sorted[0]]];
+  for (let i = 1; i < sorted.length; i++) {
+    const upper = sorted[i - 1].y1, lower = sorted[i].y1;
+    const joined = chained.some((v) => v.x1 >= x1 - ALIGN_TOL && v.x1 <= x2 + ALIGN_TOL && v.y2 >= upper - TOUCH_TOL && v.y1 <= lower + TOUCH_TOL);
+    if (joined) parts[parts.length - 1].push(sorted[i]);
+    else parts.push([sorted[i]]);
+  }
+  return parts.filter((p) => p.length >= MIN_RULES).length >= 2 ? parts : [g];
+}
 function closeOpenTableEnds(horizontals, verticals, textLayer = false) {
   if (horizontals.length < MIN_RULES || verticals.length < MIN_REACHING) return horizontals;
   const groups = [];
@@ -5315,7 +5423,7 @@ function closeOpenTableEnds(horizontals, verticals, textLayer = false) {
   }
   const chained = chainVerticals2(verticals);
   const added = [];
-  for (const g of groups) {
+  for (const g of groups.flatMap((gr) => splitBodies(gr, chained))) {
     if (g.length < MIN_RULES) continue;
     const x1 = Math.min(...g.map((r) => r.x1)), x2 = Math.max(...g.map((r) => r.x2));
     const yLo = Math.min(...g.map((r) => r.y1)), yHi = Math.max(...g.map((r) => r.y1));
@@ -5763,7 +5871,7 @@ function stackedTableBands(items) {
   const caption = (start, end) => lines.slice(start, end).some((line) => /^Table\s+\d+\s*[:.]/i.test(mergeLineSimple(line).trim()));
   while (runs.length > 0 && !caption(runs[runs.length - 1].end, lines.length)) runs.pop();
   if (runs.length < 2) return null;
-  if (runs.some((run, i) => !caption(run.end, _nullishCoalesce(_optionalChain([runs, 'access', _220 => _220[i + 1], 'optionalAccess', _221 => _221.start]), () => ( lines.length))))) return null;
+  if (runs.some((run, i) => !caption(run.end, _nullishCoalesce(_optionalChain([runs, 'access', _235 => _235[i + 1], 'optionalAccess', _236 => _236.start]), () => ( lines.length))))) return null;
   const tables = runs.map((run) => lines.slice(run.start, run.end).flat());
   const between = runs.slice(0, -1).map((run, i) => lines.slice(run.end, runs[i + 1].start).flat());
   const tail = lines.slice(runs[runs.length - 1].end);
@@ -5923,7 +6031,7 @@ function extractPageBlocksWithLines(items, pageNum, opList, pageWidth, pageHeigh
   for (const line of streamLines(items)) lex.addLine(mergeLineSimple(line));
   const extracted = extractLines(opList.fnArray, opList.argsArray);
   let { horizontals, verticals } = extracted;
-  const prevPage = _optionalChain([carry, 'optionalAccess', _222 => _222.page]) === pageNum - 1 ? carry.clip : void 0;
+  const prevPage = _optionalChain([carry, 'optionalAccess', _237 => _237.page]) === pageNum - 1 ? carry.clip : void 0;
   const clipResult = detectTables ? buildClipCellGrids(extracted.clipRects, horizontals, verticals, pageWidth, pageHeight, items.map((it) => ({ x: it.x + it.w / 2, y: it.y + it.h / 2 })), extracted.fillRects, prevPage) : { grids: [], containers: [], page: void 0 };
   if (carry) {
     carry.page = pageNum;
@@ -5988,6 +6096,7 @@ function markStrikethroughItems(items, horizontals) {
     if (line.lineWidth > STRIKE_MAX_THICKNESS) continue;
     const matches = [];
     for (const item of items) {
+      if (item.fontName === "ocr") continue;
       const h = item.h > 0 ? item.h : item.fontSize;
       if (h <= 0 || item.w <= 0) continue;
       if (line.lineWidth > h * STRIKE_MAX_THICKNESS_RATIO) continue;
@@ -6068,16 +6177,25 @@ function isProseBoxGrid(grid, verticals, table) {
   if (longCells < PROSEBOX_LONG_CELL_MIN || longCells < texts.length * PROSEBOX_LONG_CELL_RATIO) return false;
   return true;
 }
+function isSparseProseGrid(table) {
+  if (table.rows < 4) return false;
+  const texts = table.cells.flat().map((c) => c.text.trim());
+  const filled = texts.filter(Boolean);
+  if (filled.length === 0 || filled.length > texts.length * 0.25) return false;
+  const chars = filled.reduce((n, t) => n + t.length, 0);
+  const prose = filled.filter((t) => t.length >= 60 && /[.!?。]/.test(t)).reduce((n, t) => n + t.length, 0);
+  return prose >= chars * 0.6;
+}
 function cleanCellText(text) {
   const stripped = text.replace(/^[\s]*[-–—]\s*\d+\s*[-–—][\s]*$/gm, "").trim();
   return stripped.split("\n").map((line) => collapseEvenSpacing(line)).join("\n");
 }
 var FRAME_RECT_TOL = 1.5;
-function takePendingNested(pending, cellBox) {
+function takePendingNested(pending, cellBox, clipCell) {
   const out = [];
   for (let i = pending.length - 1; i >= 0; i--) {
     const p = pending[i].parent;
-    if (Math.abs(p.x1 - cellBox.x1) <= FRAME_RECT_TOL && Math.abs(p.x2 - cellBox.x2) <= FRAME_RECT_TOL && Math.abs(p.y1 - cellBox.y1) <= FRAME_RECT_TOL && Math.abs(p.y2 - cellBox.y2) <= FRAME_RECT_TOL) {
+    if (pending[i].contained ? p.x1 >= cellBox.x1 - FRAME_RECT_TOL && p.x2 <= cellBox.x2 + FRAME_RECT_TOL && p.y1 >= cellBox.y1 - FRAME_RECT_TOL && p.y2 <= cellBox.y2 + FRAME_RECT_TOL : clipCell && Math.abs(p.x1 - cellBox.x1) <= FRAME_RECT_TOL && Math.abs(p.x2 - cellBox.x2) <= FRAME_RECT_TOL && Math.abs(p.y1 - cellBox.y1) <= FRAME_RECT_TOL && Math.abs(p.y2 - cellBox.y2) <= FRAME_RECT_TOL) {
       out.push(pending[i].block);
       pending.splice(i, 1);
     }
@@ -6106,6 +6224,7 @@ function buildFrameCellBlocks(cellItems, nested, pageNum, wrap) {
   return { blocks, text };
 }
 function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, horizontals, verticals, imageRegions = [], lex, ruled = []) {
+  const ocrPage = items.length > 0 && items.every((i) => i.fontName === "ocr");
   const blocks = [];
   const usedItems = /* @__PURE__ */ new Set();
   for (const r of ruled) {
@@ -6115,13 +6234,14 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
   const proseSidebars = /* @__PURE__ */ new Set();
   const pendingNested = [];
   const gridArea = (g) => (g.bbox.x2 - g.bbox.x1) * (g.bbox.y2 - g.bbox.y1);
-  const sortedGrids = [...grids].sort((a, b) => (b.cells ? 1 : 0) - (a.cells ? 1 : 0) || (a.cells && b.cells ? gridArea(a) - gridArea(b) : 0) || b.bbox.y2 - a.bbox.y2);
+  const sortedGrids = [...grids].sort((a, b) => (b.cells ? 1 : 0) - (a.cells ? 1 : 0) || (a.cells && b.cells ? gridArea(a) - gridArea(b) : 0) || (b.lineNested ? 1 : 0) - (a.lineNested ? 1 : 0) || (a.lineNested && b.lineNested ? gridArea(a) - gridArea(b) : 0) || b.bbox.y2 - a.bbox.y2);
   for (const grid of sortedGrids) {
     const numGridRows = grid.rowYs.length - 1;
     const numGridCols = grid.colXs.length - 1;
     const gridW = grid.bbox.x2 - grid.bbox.x1;
-    if (!grid.cells && numGridRows === 1 && numGridCols >= 2) continue;
-    if (!grid.cells && numGridCols === 1 && numGridRows >= 2 && (numGridRows < 5 || gridW > pageWidth * 0.7)) continue;
+    const holdsNested = !grid.cells && pendingNested.some((p) => p.contained && p.parent.x1 >= grid.bbox.x1 - FRAME_RECT_TOL && p.parent.x2 <= grid.bbox.x2 + FRAME_RECT_TOL && p.parent.y1 >= grid.bbox.y1 - FRAME_RECT_TOL && p.parent.y2 <= grid.bbox.y2 + FRAME_RECT_TOL);
+    if (!grid.cells && !holdsNested && numGridRows === 1 && numGridCols >= 2) continue;
+    if (!grid.cells && !holdsNested && numGridCols === 1 && numGridRows >= 2 && (numGridRows < 5 || gridW > pageWidth * 0.7)) continue;
     const tableItems = [];
     const pad = 3;
     for (const item of items) {
@@ -6157,7 +6277,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
     let nestedAttached = false;
     for (const cell of cells) {
       const cellItems = cellTextMap.get(cell) || [];
-      const nested = grid.cells ? takePendingNested(pendingNested, cell.bbox) : [];
+      const nested = pendingNested.length ? takePendingNested(pendingNested, cell.bbox, !!grid.cells) : [];
       if (nested.length > 0) {
         nestedAttached = true;
         const built = buildFrameCellBlocks(cellItems, nested, pageNum, { box: cell.bbox, lex });
@@ -6210,14 +6330,14 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
         if (!assignedItems.has(textItems[ti])) usedItems.delete(tableItems[ti]);
       }
     }
-    if (unitLine.length > 0 && rebuiltUsed && !/^\s*\(\s*단위\s*[:：]/.test(_nullishCoalesce(_optionalChain([finalGrid, 'access', _223 => _223[0], 'optionalAccess', _224 => _224[0], 'optionalAccess', _225 => _225.text]), () => ( "")))) {
+    if (unitLine.length > 0 && rebuiltUsed && !/^\s*\(\s*단위\s*[:：]/.test(_nullishCoalesce(_optionalChain([finalGrid, 'access', _238 => _238[0], 'optionalAccess', _239 => _239[0], 'optionalAccess', _240 => _240.text]), () => ( "")))) {
       finalGrid.unshift(Array.from({ length: numCols }, (_, c) => ({ text: c === 0 ? cleanCellText(unitText) : "", colSpan: c === 0 ? numCols : 1, rowSpan: 1 })));
       finalRows++;
       for (const item of unitLine) usedItems.add(item);
     }
     let semanticOneColumn = false;
     if (!grid.cells && numCols === 1 && numGridRows >= 2) {
-      const populatedRows = finalGrid.filter((row) => _optionalChain([row, 'access', _226 => _226[0], 'optionalAccess', _227 => _227.text, 'access', _228 => _228.trim, 'call', _229 => _229()]));
+      const populatedRows = finalGrid.filter((row) => _optionalChain([row, 'access', _241 => _241[0], 'optionalAccess', _242 => _242.text, 'access', _243 => _243.trim, 'call', _244 => _244()]));
       if (populatedRows.length >= 4) {
         const fontSizes = tableItems.map((item) => item.fontSize).filter((size) => size > 0).sort((a, b) => a - b);
         const medianFont = _nullishCoalesce(fontSizes[Math.floor(fontSizes.length / 2)], () => ( 0));
@@ -6275,11 +6395,11 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       pendingNested.push({ parent: grid.clipParent, block: { type: "table", table: irTable, pageNumber: pageNum, bbox: nb } });
       continue;
     }
-    if (!grid.cells && isProseBoxGrid(grid, verticals, irTable)) {
+    if (!grid.cells && !nestedAttached && (isProseBoxGrid(grid, verticals, irTable) || ocrPage && isSparseProseGrid(irTable))) {
       for (const it of tableItems) usedItems.delete(it);
       continue;
     }
-    if (!grid.cells && isChartTable(irTable)) {
+    if (!grid.cells && !nestedAttached && isChartTable(irTable)) {
       blocks.push(chartBlock(tableItems, pageNum, { page: pageNum, x: grid.bbox.x1, y: grid.bbox.y1, width: gridW, height: grid.bbox.y2 - grid.bbox.y1 }));
       continue;
     }
@@ -6291,7 +6411,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       height: grid.bbox.y2 - grid.bbox.y1
     };
     if (numRows === 1 && numCols === 1 && gridW < pageWidth * 0.35) {
-      const prose = _nullishCoalesce(_optionalChain([finalGrid, 'access', _230 => _230[0], 'optionalAccess', _231 => _231[0], 'optionalAccess', _232 => _232.text]), () => ( ""));
+      const prose = _nullishCoalesce(_optionalChain([finalGrid, 'access', _245 => _245[0], 'optionalAccess', _246 => _246[0], 'optionalAccess', _247 => _247.text]), () => ( ""));
       const besideLines = (side) => new Set(side.map((it) => Math.round(it.y / 3))).size;
       const rightLines = items.filter((it) => it.x >= grid.bbox.x2 + 3 && it.y >= grid.bbox.y1 && it.y <= grid.bbox.y2);
       const leftLines = items.filter((it) => it.x + it.w <= grid.bbox.x1 - 3 && it.y >= grid.bbox.y1 && it.y <= grid.bbox.y2);
@@ -6317,6 +6437,10 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       }
       continue;
     }
+    if (grid.lineNested) {
+      pendingNested.push({ parent: grid.bbox, block: { type: "table", table: irTable, pageNumber: pageNum, bbox: tableBbox }, contained: true });
+      continue;
+    }
     blocks.push({ type: "table", table: irTable, pageNumber: pageNum, bbox: tableBbox });
   }
   for (const p of pendingNested) blocks.push(p.block);
@@ -6371,7 +6495,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
     if (gutterX !== null) {
       const gx = gutterX;
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkFBZCWSPNcjs.safeMax.call(void 0, allY) - _chunkFBZCWSPNcjs.safeMin.call(void 0, allY);
+      const pageH = _chunkJ6ZU7AIRcjs.safeMax.call(void 0, allY) - _chunkJ6ZU7AIRcjs.safeMin.call(void 0, allY);
       const gapThreshold = Math.max(15, pageH * 0.03);
       const sides = [
         remaining.filter((i) => i.x + i.w <= gx),
@@ -6391,7 +6515,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       finalTextBlocks = detectListBlocks(textBlocks);
     } else {
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkFBZCWSPNcjs.safeMax.call(void 0, allY) - _chunkFBZCWSPNcjs.safeMin.call(void 0, allY);
+      const pageH = _chunkJ6ZU7AIRcjs.safeMax.call(void 0, allY) - _chunkJ6ZU7AIRcjs.safeMin.call(void 0, allY);
       const groups = xyCutOrder(remaining, Math.max(15, pageH * 0.03));
       const textBlocks = [];
       for (const group of groups) {
@@ -6703,7 +6827,7 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false, detectTable
       blocks.push(...columnTextToBlocks(tableText, pageNum, bbox, dominantStyle(items)));
     } else {
       const allY = items.map((i) => i.y);
-      const pageHeight = _chunkFBZCWSPNcjs.safeMax.call(void 0, allY) - _chunkFBZCWSPNcjs.safeMin.call(void 0, allY);
+      const pageHeight = _chunkJ6ZU7AIRcjs.safeMax.call(void 0, allY) - _chunkJ6ZU7AIRcjs.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) {
@@ -6756,4 +6880,4 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false, detectTable
 
 
 exports.extractImageRegions = extractImageRegions; exports.filterHiddenText = filterHiddenText; exports.collapseEvenSpacing = collapseEvenSpacing; exports.normalizeItems = normalizeItems; exports.WrapLexicon = WrapLexicon; exports.wrapJoiner = wrapJoiner; exports.joinPageBreakWraps = joinPageBreakWraps; exports.latinSoftWrap = latinSoftWrap; exports.TOC_BLOCKS = TOC_BLOCKS; exports.computeMedianFontSizeFromFreq = computeMedianFontSizeFromFreq; exports.detectHeadings = detectHeadings; exports.mergeStackedHeadingLines = mergeStackedHeadingLines; exports.detectTypographyHeadings = detectTypographyHeadings; exports.detectDocumentStyleHeadings = detectDocumentStyleHeadings; exports.detectSiblingStyleHeadings = detectSiblingStyleHeadings; exports.detectRepeatedPageLabels = detectRepeatedPageLabels; exports.detectPageLeadHeadings = detectPageLeadHeadings; exports.refineDocumentStyleHeadings = refineDocumentStyleHeadings; exports.detectMarkerHeadings = detectMarkerHeadings; exports.detectTableCaptions = detectTableCaptions; exports.detectKoreanListBlocks = detectKoreanListBlocks; exports.removeHeaderFooterBlocks = removeHeaderFooterBlocks; exports.CLIP_TABLES = CLIP_TABLES; exports.TABLE_COLXS = TABLE_COLXS; exports.EMPTY_PARTS = EMPTY_PARTS; exports.IMAGE_CELLS = IMAGE_CELLS; exports.FILLER_CELLS = FILLER_CELLS; exports.CONT_PARTS = CONT_PARTS; exports.CELL_LINES = CELL_LINES; exports.trimTrailingEmptyTableCols = trimTrailingEmptyTableCols; exports.extractPageBlocksWithLines = extractPageBlocksWithLines;
-//# sourceMappingURL=chunk-SOM7JIKG.cjs.map
+//# sourceMappingURL=chunk-4ANWGSUT.cjs.map