kordoc 4.16.0 → 4.16.2

package/dist/{chunk-MH6MAL64.cjs → chunk-WDQTFAXY.cjs}

@@ -2,15 +2,16 @@
 
 
 
-var _chunkZY5K3LPMcjs = require('./chunk-ZY5K3LPM.cjs');
 
+var _chunk62W7QCEXcjs = require('./chunk-62W7QCEX.cjs');
 
 
-var _chunkH6KMTYYZcjs = require('./chunk-H6KMTYYZ.cjs');
 
+var _chunkCEIATRDWcjs = require('./chunk-CEIATRDW.cjs');
 
 
-var _chunkNWODPYBDcjs = require('./chunk-NWODPYBD.cjs');
+
+var _chunkSTM3ORY7cjs = require('./chunk-STM3ORY7.cjs');
 
 // src/ocr/models.ts
 var _path = require('path');
@@ -38,10 +39,10 @@ var OCR_REC_DICT = {
 };
 var ALL_OCR_MODELS = [OCR_DET_MODEL, OCR_REC_MODEL, OCR_REC_DICT];
 function getOcrModelsDir() {
-  return _chunkH6KMTYYZcjs.getModelsDir.call(void 0, "ppocr");
+  return _chunkCEIATRDWcjs.getModelsDir.call(void 0, "ppocr");
 }
 async function ensureOcrModels(onProgress) {
-  return _chunkH6KMTYYZcjs.ensureModelsIn.call(void 0, getOcrModelsDir(), ALL_OCR_MODELS, onProgress);
+  return _chunkCEIATRDWcjs.ensureModelsIn.call(void 0, getOcrModelsDir(), ALL_OCR_MODELS, onProgress);
 }
 async function ocrModelsCached() {
   const dir = getOcrModelsDir();
@@ -1085,6 +1086,12 @@ function dropInsetClipGrids(clipGrids, lineGrids) {
   const near = (a, b) => Math.abs(a - b) <= SHADE_CLIP_TOL;
   return clipGrids.filter((c) => c.clipParent || c.rowYs.length !== 2 || !lineGrids.some((l) => l.rowYs.length >= 3 && near(c.bbox.x1, l.bbox.x1) && near(c.bbox.x2, l.bbox.x2) && l.rowYs.some((top, r) => r + 1 < l.rowYs.length && top - c.bbox.y2 >= INSET_CLIP_MIN && c.bbox.y1 - l.rowYs[r + 1] >= INSET_CLIP_MIN)));
 }
+function dropHeadBandClipGrids(clipGrids, lineGrids) {
+  if (clipGrids.length === 0 || lineGrids.length === 0) return clipGrids;
+  const near = (a, b) => Math.abs(a - b) <= SHADE_CLIP_TOL;
+  const overlaps = (a, b) => Math.min(a.x2, b.x2) > Math.max(a.x1, b.x1) && Math.min(a.y2, b.y2) > Math.max(a.y1, b.y1);
+  return clipGrids.filter((c) => c.clipParent || !_optionalChain([c, 'access', _15 => _15.cells, 'optionalAccess', _16 => _16.length]) || !lineGrids.some((l) => l.rowYs.length > c.rowYs.length && l.colXs.length === c.colXs.length && c.colXs.every((x, k) => near(x, l.colXs[k])) && c.rowYs.every((y, k) => near(y, l.rowYs[k])) && !clipGrids.some((o) => o !== c && overlaps(o.bbox, l.bbox))));
+}
 function enforceMinWidth(colXs, minWidth) {
   if (colXs.length <= 2) return colXs;
   const result = [colXs[0]];
@@ -1427,7 +1434,7 @@ function buildClipCellGrids(rects, strokedH, strokedV, pageWidth, pageHeight, te
     for (let dx = -1; dx <= 1 && !duplicate; dx++) {
       const ys2 = buckets.get(bx + dx);
       for (let dy = -1; dy <= 1 && !duplicate; dy++) {
-        duplicate = _nullishCoalesce(_optionalChain([ys2, 'optionalAccess', _15 => _15.get, 'call', _16 => _16(by + dy), 'optionalAccess', _17 => _17.some, 'call', _18 => _18((c) => sameRect(c, r))]), () => ( false));
+        duplicate = _nullishCoalesce(_optionalChain([ys2, 'optionalAccess', _17 => _17.get, 'call', _18 => _18(by + dy), 'optionalAccess', _19 => _19.some, 'call', _20 => _20((c) => sameRect(c, r))]), () => ( false));
       }
     }
     if (duplicate) continue;
@@ -1544,7 +1551,7 @@ function buildClipCellGrids(rects, strokedH, strokedV, pageWidth, pageHeight, te
     else groups.set(r, [i]);
   }
   const containers = cells.filter((_, i) => isContainer[i]);
-  const groupSize = (i) => _nullishCoalesce(_optionalChain([groups, 'access', _19 => _19.get, 'call', _20 => _20(find(i)), 'optionalAccess', _21 => _21.length]), () => ( 0));
+  const groupSize = (i) => _nullishCoalesce(_optionalChain([groups, 'access', _21 => _21.get, 'call', _22 => _22(find(i)), 'optionalAccess', _23 => _23.length]), () => ( 0));
   const isGridMember = (i) => groupSize(i) >= CLIP_MIN_GROUP;
   const framed = (r) => edgeStroked(strokedH, "h", r.y1, r.x1, r.x2) && edgeStroked(strokedH, "h", r.y2, r.x1, r.x2) && edgeStroked(strokedV, "v", r.x1, r.y1, r.y2) && edgeStroked(strokedV, "v", r.x2, r.y1, r.y2);
   const titledFrame = (r) => r.y1 >= pageHeight * TITLED_FRAME_MIN_TOP && r.x2 - r.x1 >= pageWidth * TITLED_FRAME_MIN_WIDTH && textPoints.some((p) => p.y < r.y1 && p.y > pageHeight * HEADER_BAND && p.x >= r.x1 - CLIP_EDGE_TOL && p.x <= r.x2 + CLIP_EDGE_TOL) && textPoints.some((p) => p.x > r.x1 && p.x < r.x2 && p.y > r.y1 && p.y < r.y2);
@@ -1921,6 +1928,60 @@ function hasHorizontalLine(horizontals, y, leftX, rightX, vertexRadius) {
   return false;
 }
 
+// src/pdf/script-items.ts
+function scriptKindsOfLine(sorted) {
+  const kinds = sorted.map(() => null);
+  if (sorted.length < 2) return kinds;
+  const weight = /* @__PURE__ */ new Map();
+  for (const it of sorted) {
+    const k = Math.round(it.fontSize * 2) / 2;
+    weight.set(k, (_nullishCoalesce(weight.get(k), () => ( 0))) + it.text.trim().length);
+  }
+  let bodyFs = 0, best = -1;
+  for (const [fs, n] of weight) if (n > best || n === best && fs > bodyFs) {
+    best = n;
+    bodyFs = fs;
+  }
+  if (bodyFs <= 0) return kinds;
+  const ys = sorted.filter((it) => Math.abs(it.fontSize - bodyFs) <= bodyFs * 0.12).map((it) => it.y).sort((a, b) => a - b);
+  const bodyY = ys[ys.length >> 1];
+  const isBody = (it) => it.fontSize > bodyFs * 0.85;
+  for (let i = 1; i < sorted.length; i++) {
+    const it = sorted[i], prev = sorted[i - 1];
+    const t = it.text.trim();
+    if (!t || t.length > 16 || isBody(it)) continue;
+    if (!isBody(prev) && !kinds[i - 1]) continue;
+    const gap = it.x - (prev.x + prev.w);
+    if (gap > bodyFs * 0.35 || gap < -bodyFs * 0.5) continue;
+    const dy = it.y - bodyY;
+    if (dy >= bodyFs * 0.15) kinds[i] = "sup";
+    else if (dy <= -bodyFs * 0.08) kinds[i] = "sub";
+  }
+  return kinds;
+}
+function tagScripts(text, lines) {
+  const marks = [];
+  let cursor = 0;
+  for (const line of lines) {
+    const kinds = scriptKindsOfLine(line);
+    for (let k = 0; k < line.length; k++) {
+      const t = line[k].text;
+      if (!t) continue;
+      const at = text.indexOf(t, cursor);
+      if (at < 0 || /\S/.test(text.slice(cursor, at))) return text;
+      if (kinds[k]) marks.push([at, at + t.length, kinds[k]]);
+      cursor = at + t.length;
+    }
+  }
+  if (!marks.length) return text;
+  let out = text;
+  for (let m = marks.length - 1; m >= 0; m--) {
+    const [s, e, kind] = marks[m];
+    out = out.slice(0, s) + `<${kind}>` + out.slice(s, e) + `</${kind}>` + out.slice(e);
+  }
+  return _chunk62W7QCEXcjs.tidyScriptTags.call(void 0, out);
+}
+
 // src/pdf/line-wrap.ts
 var MARKUP = /<\/?u>|~~/g;
 var bump = (m, k) => {
@@ -1970,12 +2031,19 @@ var WrapLexicon = (_class = class {constructor() { _class.prototype.__init.call(
   joinedWord(left, right) {
     return this.isWord(left + right) && !this.pairs.has(left + " " + right);
   }
+  /** 두 글자+두 글자 증거만 — 한 글자 쌍으로 물러서지 않는다 (표 칸 조각처럼 줄 꺾임이 아닌 자리에 쓸 때) */
+  evidence2(left, right) {
+    const n = left.length;
+    if (n < 2 || right.length < 2) return null;
+    const k1 = pairKey(left.charCodeAt(n - 2), left.charCodeAt(n - 1)), k2 = pairKey(right.charCodeAt(0), right.charCodeAt(1));
+    return decideCounts(_nullishCoalesce(_optionalChain([this, 'access', _24 => _24.joined2, 'access', _25 => _25.get, 'call', _26 => _26(k1), 'optionalAccess', _27 => _27.get, 'call', _28 => _28(k2)]), () => ( 0)), _nullishCoalesce(_optionalChain([this, 'access', _29 => _29.spaced2, 'access', _30 => _30.get, 'call', _31 => _31(k1), 'optionalAccess', _32 => _32.get, 'call', _33 => _33(k2)]), () => ( 0)));
+  }
   /** 꺾인 자리 증거: "" 붙음 · " " 띄움 · null 모름 (두 글자 증거가 갈리거나 없으면 한 글자) */
   evidence(left, right) {
     const n = left.length;
     if (n >= 2 && right.length >= 2) {
       const k1 = pairKey(left.charCodeAt(n - 2), left.charCodeAt(n - 1)), k2 = pairKey(right.charCodeAt(0), right.charCodeAt(1));
-      const v = decideCounts(_nullishCoalesce(_optionalChain([this, 'access', _22 => _22.joined2, 'access', _23 => _23.get, 'call', _24 => _24(k1), 'optionalAccess', _25 => _25.get, 'call', _26 => _26(k2)]), () => ( 0)), _nullishCoalesce(_optionalChain([this, 'access', _27 => _27.spaced2, 'access', _28 => _28.get, 'call', _29 => _29(k1), 'optionalAccess', _30 => _30.get, 'call', _31 => _31(k2)]), () => ( 0)));
+      const v = decideCounts(_nullishCoalesce(_optionalChain([this, 'access', _34 => _34.joined2, 'access', _35 => _35.get, 'call', _36 => _36(k1), 'optionalAccess', _37 => _37.get, 'call', _38 => _38(k2)]), () => ( 0)), _nullishCoalesce(_optionalChain([this, 'access', _39 => _39.spaced2, 'access', _40 => _40.get, 'call', _41 => _41(k1), 'optionalAccess', _42 => _42.get, 'call', _43 => _43(k2)]), () => ( 0)));
       if (v !== null) return v;
     }
     if (n && right.length) {
@@ -2047,11 +2115,11 @@ function wrapJoiner(prevText, nextText, lex) {
   if (particleContinues(left, right)) return "";
   if (endsAsWord(left)) return " ";
   if (/[A-Za-z]-$/.test(left) && /^[A-Za-z]/.test(right) && !/[가-힣]/.test(prev + next)) {
-    return _optionalChain([lex, 'optionalAccess', _32 => _32.joinedWord, 'call', _33 => _33(left, right.replace(/[.,;:!?)]*$/, ""))]) ? "" : " ";
+    return _optionalChain([lex, 'optionalAccess', _44 => _44.joinedWord, 'call', _45 => _45(left, right.replace(/[.,;:!?)]*$/, ""))]) ? "" : " ";
   }
   if (/[·ㆍ‧-]/.test(a) && !/[·ㆍ‧…]{2}$/.test(prev) && /[가-힣A-Za-z0-9]/.test(b)) return "";
   if (/[A-Za-z]/.test(a) && /[A-Za-z]/.test(b) || /\d/.test(a) && /\d/.test(b)) return " ";
-  const ev = _optionalChain([lex, 'optionalAccess', _34 => _34.evidence, 'call', _35 => _35(left, right)]);
+  const ev = _optionalChain([lex, 'optionalAccess', _46 => _46.evidence, 'call', _47 => _47(left, right)]);
   if (ev != null) return ev;
   if (lex && /[가-힣A-Za-z0-9]$/.test(left) && /^[가-힣A-Za-z0-9]/.test(right) && lex.joinedWord(left, right.replace(/[.,)」』’”;:]+$/, ""))) return "";
   if (/\d$/.test(left) && COUNTER.test(right)) return "";
@@ -2073,7 +2141,7 @@ function joinPageBreakWraps(blocks, lex) {
     if (a.type !== "paragraph" && a.type !== "list" || b.type !== "paragraph") continue;
     const last = a.bbox && PARA_LAST_LINE.get(a.bbox), fs = last ? last.fontSize : 0;
     if (!last || fs <= 0 || (_nullishCoalesce(pageRight.get(a.pageNumber), () => ( Infinity))) - last.right >= BODY_FULL_TOL * fs) continue;
-    if (last.width < BODY_MIN_WIDTH_EM * fs || Math.abs((_nullishCoalesce(_optionalChain([b, 'access', _36 => _36.style, 'optionalAccess', _37 => _37.fontSize]), () => ( 0))) - fs) > 0.15 * fs) continue;
+    if (last.width < BODY_MIN_WIDTH_EM * fs || Math.abs((_nullishCoalesce(_optionalChain([b, 'access', _48 => _48.style, 'optionalAccess', _49 => _49.fontSize]), () => ( 0))) - fs) > 0.15 * fs) continue;
     if (startsNewItem(a.text, b.text)) continue;
     a.text += wrapJoiner(a.text, b.text, lex) + b.text;
     const bl = b.bbox && PARA_LAST_LINE.get(b.bbox);
@@ -2128,6 +2196,7 @@ function cellLineFills(box, contentLeft, cellRight, lineRight, fontSize) {
 }
 
 // src/pdf/cell-text.ts
+var REGION_LABELS = /* @__PURE__ */ new Set(["\uC11C\uC6B8", "\uBD80\uC0B0", "\uB300\uAD6C", "\uC778\uCC9C", "\uAD11\uC8FC", "\uB300\uC804", "\uC6B8\uC0B0", "\uC138\uC885", "\uACBD\uAE30", "\uAC15\uC6D0", "\uCDA9\uBD81", "\uCDA9\uB0A8", "\uC804\uBD81", "\uC804\uB0A8", "\uACBD\uBD81", "\uACBD\uB0A8", "\uC81C\uC8FC", "\uC804\uAD6D"]);
 var CELL_PADDING = 2;
 var SPACE_GAP_RATIO = 0.17;
 function spaceGapThreshold(fontSize) {
@@ -2222,8 +2291,8 @@ function cellTextToString(items, wrap) {
   const textLines = merged.map((line) => {
     const s = sortLineByX(line);
     if (s.length === 1) return s[0].text;
-    if (s.length === 2 && /^[가-힣]$/.test(s[0].text) && /^[가-힣]$/.test(s[1].text) && !!_optionalChain([wrap, 'optionalAccess', _38 => _38.lex]) && (wrap.lex.isWord(s[0].text + s[1].text) || wrap.lex.evidence(s[0].text, s[1].text) === "") && s[1].x - (s[0].x + s[0].w) >= Math.max(s[0].fontSize, s[1].fontSize) * 1.5 && s[1].x + s[1].w - s[0].x >= wrap.box.x2 - wrap.box.x1 - Math.max(s[0].fontSize, s[1].fontSize) * 2) return s[0].text + s[1].text;
-    const evenSpaced = detectEvenSpacedItems(s);
+    if (s.length === 2 && /^[가-힣]$/.test(s[0].text) && /^[가-힣]$/.test(s[1].text) && !!_optionalChain([wrap, 'optionalAccess', _50 => _50.lex]) && (wrap.lex.isWord(s[0].text + s[1].text) || wrap.lex.evidence(s[0].text, s[1].text) === "" || REGION_LABELS.has(s[0].text + s[1].text)) && s[1].x - (s[0].x + s[0].w) >= Math.max(s[0].fontSize, s[1].fontSize) * 1.5 && s[1].x + s[1].w - s[0].x >= wrap.box.x2 - wrap.box.x1 - Math.max(s[0].fontSize, s[1].fontSize) * 2) return s[0].text + s[1].text;
+    const evenSpaced = detectEvenSpacedItems(s, true);
     let result = s[0].text;
     for (let j = 1; j < s.length; j++) {
       if (evenSpaced[j]) {
@@ -2244,7 +2313,8 @@ function cellTextToString(items, wrap) {
     }
     return result;
   });
-  if (!wrap) return mergeCellTextLines(textLines);
+  const scripted = (s) => tagScripts(s, merged.map((l) => sortLineByX([...l])));
+  if (!wrap) return scripted(mergeCellTextLines(textLines));
   let contentLeft = Infinity;
   for (const it of items) if (it.x < contentLeft) contentLeft = it.x;
   const lineEnds = merged.map((line) => {
@@ -2257,7 +2327,7 @@ function cellTextToString(items, wrap) {
   let cellRight = -Infinity;
   for (const e of lineEnds) if (e.right > cellRight) cellRight = e.right;
   const wraps = lineEnds.slice(0, -1).map((a, i) => cellLineWraps(wrap.box, contentLeft, a.right, a.fontSize, lineEnds[i + 1].firstCharW) || lineEnds.length >= 3 && !textLines[i + 1].startsWith("(") && cellLineFills(wrap.box, contentLeft, cellRight, a.right, a.fontSize));
-  return mergeCellTextLines(textLines, { wraps, lex: wrap.lex });
+  return scripted(mergeCellTextLines(textLines, { wraps, lex: wrap.lex }));
 }
 function mergeSuperscriptRows(lines) {
   if (lines.length <= 1) return lines;
@@ -2293,9 +2363,13 @@ function mergeSuperscriptRows(lines) {
   }
   return result;
 }
-function detectEvenSpacedItems(items) {
+function detectEvenSpacedItems(items, cellLine = false) {
   const result = new Array(items.length).fill(false);
   if (items.length < 3) return result;
+  if (cellLine && items.every((it) => /^[가-힣]$/.test(it.text)) && items.slice(1).every((it) => !it.hasSpaceBefore || it.syntheticSpace) && items.slice(1).every((it, k) => it.x - (items[k].x + items[k].w) >= it.fontSize * 0.1)) {
+    markEvenRun(items, result, 0, items.length);
+    if (result.some(Boolean)) return result;
+  }
   let runStart = -1;
   for (let i = 0; i < items.length; i++) {
     const isShortKorean = /^[가-힣]{1}$/.test(items[i].text) || /^[\d]{1}$/.test(items[i].text);
@@ -2364,7 +2438,7 @@ function mergeCellTextLines(textLines, wrap) {
       merged[merged.length - 1] = prev + curr.trim();
     } else if (!wrap && /[,(]$/.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()) && !_optionalChain([wrap, 'optionalAccess', _39 => _39.wraps, 'access', _40 => _40[i - 1]])) && !(/(?:^|\s)\d{1,7}$/.test(prev) && /^\d{1,7}$/.test(curr.trim()) && !_optionalChain([wrap, 'optionalAccess', _41 => _41.wraps, 'access', _42 => _42[i - 1]]))) {
+    } 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', _51 => _51.wraps, 'access', _52 => _52[i - 1]])) && !(/(?:^|\s)\d{1,7}$/.test(prev) && /^\d{1,7}$/.test(curr.trim()) && !_optionalChain([wrap, 'optionalAccess', _53 => _53.wraps, 'access', _54 => _54[i - 1]]))) {
       merged[merged.length - 1] = prev + curr.trim();
     } else {
       merged.push(curr);
@@ -2434,7 +2508,7 @@ 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', _43 => _43.length]) !== numRows + 1) return null;
+    if (_optionalChain([rowYs, 'optionalAccess', _55 => _55.length]) !== numRows + 1) return null;
     const rebuilt2 = [];
     let expanded = 0;
     for (let r = 0; r < numRows; r++) {
@@ -2581,7 +2655,7 @@ 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', _44 => _44.text, 'access', _45 => _45.match, 'call', _46 => _46(/[\p{L}\p{N}]/gu), 'optionalAccess', _47 => _47.length]), () => ( 0)));
+    const count = (freq.get(i.fontSize) || 0) + Math.max(1, _nullishCoalesce(_optionalChain([i, 'access', _56 => _56.text, 'access', _57 => _57.match, 'call', _58 => _58(/[\p{L}\p{N}]/gu), 'optionalAccess', _59 => _59.length]), () => ( 0)));
     freq.set(i.fontSize, count);
     if (count > maxCount) {
       maxCount = count;
@@ -2589,7 +2663,7 @@ function dominantStyle(items) {
     }
   }
   if (dominantSize === 0) return void 0;
-  const fontName = _optionalChain([items, 'access', _48 => _48.find, 'call', _49 => _49((i) => i.fontSize === dominantSize), 'optionalAccess', _50 => _50.fontName]) || void 0;
+  const fontName = _optionalChain([items, 'access', _60 => _60.find, 'call', _61 => _61((i) => i.fontSize === dominantSize), 'optionalAccess', _62 => _62.fontName]) || void 0;
   return fontName ? { fontSize: dominantSize, fontName } : { fontSize: dominantSize };
 }
 function normalizeItems(rawItems) {
@@ -2602,7 +2676,7 @@ function normalizeItems(rawItems) {
     const x = Math.round(i.transform[4]);
     const y = Math.round(i.transform[5]);
     if (!i.str.trim()) {
-      spacePositions.push({ x, y });
+      spacePositions.push({ x, y, synthetic: i.synthetic });
       continue;
     }
     const scaleX = Math.hypot(i.transform[0], i.transform[1]);
@@ -2658,7 +2732,10 @@ function normalizeItems(rawItems) {
           nearest = item;
         }
       }
-      if (nearest) nearest.hasSpaceBefore = true;
+      if (nearest) {
+        nearest.syntheticSpace = (nearest.hasSpaceBefore ? nearest.syntheticSpace === true : true) && sp.synthetic === true;
+        nearest.hasSpaceBefore = true;
+      }
     }
   }
   return deduped;
@@ -2706,7 +2783,7 @@ 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', _51 => _51.match, 'call', _52 => _52(/\s+/g), 'optionalAccess', _53 => _53.length]), () => ( 0))) * ws;
+      const est = c.x + glyphUnits(prefix.replace(/\s+/g, "")) * a0 + (_nullishCoalesce(_optionalChain([prefix, 'access', _63 => _63.match, 'call', _64 => _64(/\s+/g), 'optionalAccess', _65 => _65.length]), () => ( 0))) * ws;
       if (Math.abs(est - o.x) <= OVERLAY_DUP_EM * fs) return false;
     }
     return true;
@@ -2801,6 +2878,13 @@ function mergeSuperscriptLines(lines) {
     for (const i of line) total += i.text.trim().length;
     return total > 0 && total <= 10;
   };
+  const isMarkers = (line, host) => line.length > 1 && line.length <= 16 && line.every((i) => {
+    if (i.text.trim().length > 3 || !i.text.trim()) return false;
+    return host.some((h) => {
+      const g = i.x - (h.x + h.w);
+      return g <= h.fontSize * 0.35 && g >= -h.fontSize * 0.1;
+    });
+  });
   const result = [lines[0]];
   for (let i = 1; i < lines.length; i++) {
     const prev = result[result.length - 1];
@@ -2808,8 +2892,8 @@ function mergeSuperscriptLines(lines) {
     const a = band(prev);
     const b = band(curr);
     const overlap2 = Math.min(a.top, b.top) - Math.max(a.bottom, b.bottom);
-    const prevIsFrag = isFrag(prev) && a.height <= b.height * 0.8 && overlap2 >= a.height * 0.5;
-    const currIsFrag = isFrag(curr) && b.height <= a.height * 0.8 && overlap2 >= b.height * 0.5;
+    const prevIsFrag = (isFrag(prev) || isMarkers(prev, curr)) && a.height <= b.height * 0.8 && overlap2 >= a.height * 0.5;
+    const currIsFrag = (isFrag(curr) || isMarkers(curr, prev)) && b.height <= a.height * 0.8 && overlap2 >= b.height * 0.5;
     if (prevIsFrag || currIsFrag) {
       result[result.length - 1] = [...prev, ...curr];
     } else {
@@ -2819,7 +2903,7 @@ function mergeSuperscriptLines(lines) {
   return result;
 }
 function mergeLineSimple(items) {
-  if (items.length <= 1) return _optionalChain([items, 'access', _54 => _54[0], 'optionalAccess', _55 => _55.text]) || "";
+  if (items.length <= 1) return _optionalChain([items, 'access', _66 => _66[0], 'optionalAccess', _67 => _67.text]) || "";
   const sorted = sortLineByX([...items]);
   const isEvenSpaced = detectEvenSpacedItems(sorted);
   let result = sorted[0].text;
@@ -2964,7 +3048,7 @@ 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', _56 => _56(line), 'optionalAccess', _57 => _57.fontName]), () => ( ""));
+  const faceOf = (line) => _nullishCoalesce(_optionalChain([dominantStyle, 'call', _68 => _68(line), 'optionalAccess', _69 => _69.fontName]), () => ( ""));
   const runs = [];
   for (const line of lines) {
     const last = runs[runs.length - 1];
@@ -2994,7 +3078,7 @@ 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', _58 => _58(l.items), 'optionalAccess', _59 => _59.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', _70 => _70(l.items), 'optionalAccess', _71 => _71.fontSize]), () => ( 0)) };
   });
   const joins = bodyLineJoins(geo, lex);
   for (let i = 0; i + 1 < geo.length; i++) {
@@ -3007,10 +3091,13 @@ function pushLineParagraphs(out, yLines, pageNum, lex) {
   for (let i = 0; i < lines.length; ) {
     let text = lines[i].text;
     const items = [...lines[i].items];
+    const srcLines = [lines[i].items];
     for (; i + 1 < lines.length && joins[i] !== "\n"; i++) {
       text += joins[i] + lines[i + 1].text;
       items.push(...lines[i + 1].items);
+      srcLines.push(lines[i + 1].items);
     }
+    text = tagScripts(text, srcLines.map((l) => sortLineByX([...l])));
     const block = { type: "paragraph", text, pageNumber: pageNum, bbox: computeBBox(items, pageNum), style: dominantStyle(items) };
     const faces = /* @__PURE__ */ new Map();
     for (const it of items) faces.set(it.fontName, (_nullishCoalesce(faces.get(it.fontName), () => ( 0))) + it.text.length);
@@ -3038,15 +3125,15 @@ function computeMedianFontSizeFromFreq(freq) {
 function detectHeadings(blocks, medianFontSize) {
   for (let bi = 0; bi < blocks.length; bi++) {
     const block = blocks[bi];
-    if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _60 => _60.style, 'optionalAccess', _61 => _61.fontSize])) continue;
+    if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _72 => _72.style, 'optionalAccess', _73 => _73.fontSize])) continue;
     const text = block.text.trim();
     if (text.length === 0 || text.length > 200) continue;
     if (/^\d+$/.test(text) && !isChapterNumber(block, blocks[bi + 1], medianFontSize)) continue;
     const ratio = block.style.fontSize / medianFontSize;
     let level = 0;
-    if (ratio >= _chunkZY5K3LPMcjs.HEADING_RATIO_H1) level = 1;
-    else if (ratio >= _chunkZY5K3LPMcjs.HEADING_RATIO_H2) level = 2;
-    else if (ratio >= _chunkZY5K3LPMcjs.HEADING_RATIO_H3) level = 3;
+    if (ratio >= _chunk62W7QCEXcjs.HEADING_RATIO_H1) level = 1;
+    else if (ratio >= _chunk62W7QCEXcjs.HEADING_RATIO_H2) level = 2;
+    else if (ratio >= _chunk62W7QCEXcjs.HEADING_RATIO_H3) level = 3;
     if (level > 0) {
       block.type = "heading";
       block.level = level;
@@ -3055,21 +3142,21 @@ function detectHeadings(blocks, medianFontSize) {
   }
 }
 function isChapterNumber(block, next, medianFontSize) {
-  const size = _nullishCoalesce(_optionalChain([block, 'access', _62 => _62.style, 'optionalAccess', _63 => _63.fontSize]), () => ( 0));
-  const nextSize = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _64 => _64.style, 'optionalAccess', _65 => _65.fontSize]), () => ( 0));
-  const text = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _66 => _66.text, 'optionalAccess', _67 => _67.trim, 'call', _68 => _68()]), () => ( ""));
-  const a = block.bbox, b = _optionalChain([next, 'optionalAccess', _69 => _69.bbox]);
-  return /^\d{1,2}$/.test(_nullishCoalesce(_optionalChain([block, 'access', _70 => _70.text, 'optionalAccess', _71 => _71.trim, 'call', _72 => _72()]), () => ( ""))) && size >= medianFontSize * 2.5 && !!next && (next.type === "paragraph" || next.type === "heading") && next.pageNumber === block.pageNumber && nextSize >= medianFontSize * _chunkZY5K3LPMcjs.HEADING_RATIO_H1 && text.length > 0 && text.length <= 80 && !/^[\d\s.]+$/.test(text) && !!a && !!b && a.y > b.y && a.y - (b.y + b.height) <= size * 2;
+  const size = _nullishCoalesce(_optionalChain([block, 'access', _74 => _74.style, 'optionalAccess', _75 => _75.fontSize]), () => ( 0));
+  const nextSize = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _76 => _76.style, 'optionalAccess', _77 => _77.fontSize]), () => ( 0));
+  const text = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _78 => _78.text, 'optionalAccess', _79 => _79.trim, 'call', _80 => _80()]), () => ( ""));
+  const a = block.bbox, b = _optionalChain([next, 'optionalAccess', _81 => _81.bbox]);
+  return /^\d{1,2}$/.test(_nullishCoalesce(_optionalChain([block, 'access', _82 => _82.text, 'optionalAccess', _83 => _83.trim, 'call', _84 => _84()]), () => ( ""))) && size >= medianFontSize * 2.5 && !!next && (next.type === "paragraph" || next.type === "heading") && next.pageNumber === block.pageNumber && nextSize >= medianFontSize * _chunk62W7QCEXcjs.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', _73 => _73.style, 'optionalAccess', _74 => _74.fontSize]), () => ( 0)), bf = _nullishCoalesce(_optionalChain([b, 'access', _75 => _75.style, 'optionalAccess', _76 => _76.fontSize]), () => ( 0));
+    const af = _nullishCoalesce(_optionalChain([a, 'access', _85 => _85.style, 'optionalAccess', _86 => _86.fontSize]), () => ( 0)), bf = _nullishCoalesce(_optionalChain([b, 'access', _87 => _87.style, 'optionalAccess', _88 => _88.fontSize]), () => ( 0));
     const gap = ab && bb ? ab.y - (bb.y + bb.height) : Infinity;
-    const centered = ab && bb && _optionalChain([a, 'access', _77 => _77.style, 'optionalAccess', _78 => _78.fontName]) === _optionalChain([b, 'access', _79 => _79.style, 'optionalAccess', _80 => _80.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', _89 => _89.style, 'optionalAccess', _90 => _90.fontName]) === _optionalChain([b, 'access', _91 => _91.style, 'optionalAccess', _92 => _92.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', _81 => _81.style, 'optionalAccess', _82 => _82.fontName]) === _optionalChain([b, 'access', _83 => _83.style, 'optionalAccess', _84 => _84.fontName]) && af === bf && ab.width >= bb.width * 0.95 && (_nullishCoalesce(_optionalChain([a, 'access', _85 => _85.text, 'optionalAccess', _86 => _86.trim, 'call', _87 => _87(), 'access', _88 => _88.split, 'call', _89 => _89(/\s+/), 'access', _90 => _90.length]), () => ( 0))) >= 3;
+    const wrapped = sameAnchor && _optionalChain([a, 'access', _93 => _93.style, 'optionalAccess', _94 => _94.fontName]) === _optionalChain([b, 'access', _95 => _95.style, 'optionalAccess', _96 => _96.fontName]) && af === bf && ab.width >= bb.width * 0.95 && (_nullishCoalesce(_optionalChain([a, 'access', _97 => _97.text, 'optionalAccess', _98 => _98.trim, 'call', _99 => _99(), 'access', _100 => _100.split, 'call', _101 => _101(/\s+/), 'access', _102 => _102.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 || /^\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++;
@@ -3098,22 +3185,22 @@ 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', _91 => _91.style, 'optionalAccess', _92 => _92.fontName])) continue;
+      if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _103 => _103.style, 'optionalAccess', _104 => _104.fontName])) continue;
       const face = block.style.fontName;
       faceWeight.set(face, (_nullishCoalesce(faceWeight.get(face), () => ( 0))) + block.text.length);
     }
-    const bodyFace = _optionalChain([[...faceWeight], 'access', _93 => _93.sort, 'call', _94 => _94((a, b) => b[1] - a[1]), 'access', _95 => _95[0], 'optionalAccess', _96 => _96[0]]);
+    const bodyFace = _optionalChain([[...faceWeight], 'access', _105 => _105.sort, 'call', _106 => _106((a, b) => b[1] - a[1]), 'access', _107 => _107[0], 'optionalAccess', _108 => _108[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', _97 => _97.style, 'optionalAccess', _98 => _98.fontName]) !== bodyFace || !block.style.fontSize) continue;
+      if (block.type !== "paragraph" || _optionalChain([block, 'access', _109 => _109.style, 'optionalAccess', _110 => _110.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', _99 => _99.sort, 'call', _100 => _100((a, b) => b[1] - a[1]), 'access', _101 => _101[0], 'optionalAccess', _102 => _102[0]]), () => ( 0));
+    const bodySize = _nullishCoalesce(_optionalChain([[...sizeWeight], 'access', _111 => _111.sort, 'call', _112 => _112((a, b) => b[1] - a[1]), 'access', _113 => _113[0], 'optionalAccess', _114 => _114[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', _103 => _103.fontName]) || !style.fontSize) continue;
+      if (block.type !== "paragraph" || !text || !bbox || !_optionalChain([style, 'optionalAccess', _115 => _115.fontName]) || !style.fontSize) continue;
       const title = text.trim();
       const numbered = /^\d+(?:\.\d+)*\.?\s+[A-Z가-힣]/.test(title);
       const faces = FACE_CHARS.get(block);
@@ -3143,16 +3230,16 @@ 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', _104 => _104.style, 'optionalAccess', _105 => _105.fontName])) continue;
+      if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _116 => _116.style, 'optionalAccess', _117 => _117.fontName])) continue;
       faceChars.set(block.style.fontName, (_nullishCoalesce(faceChars.get(block.style.fontName), () => ( 0))) + block.text.length);
     }
-    const bodyFace = _optionalChain([[...faceChars], 'access', _106 => _106.sort, 'call', _107 => _107((a, b) => b[1] - a[1]), 'access', _108 => _108[0], 'optionalAccess', _109 => _109[0]]);
+    const bodyFace = _optionalChain([[...faceChars], 'access', _118 => _118.sort, 'call', _119 => _119((a, b) => b[1] - a[1]), 'access', _120 => _120[0], 'optionalAccess', _121 => _121[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', _110 => _110.style, 'optionalAccess', _111 => _111.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', _122 => _122.style, 'optionalAccess', _123 => _123.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', _112 => _112.style, 'optionalAccess', _113 => _113.fontName]) === _optionalChain([label, 'access', _114 => _114.style, 'optionalAccess', _115 => _115.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', _124 => _124.style, 'optionalAccess', _125 => _125.fontName]) === _optionalChain([label, 'access', _126 => _126.style, 'optionalAccess', _127 => _127.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', _116 => _116.style, 'optionalAccess', _117 => _117.fontName]) !== _optionalChain([label, 'access', _118 => _118.style, 'optionalAccess', _119 => _119.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', _128 => _128.style, 'optionalAccess', _129 => _129.fontName]) !== _optionalChain([label, 'access', _130 => _130.style, 'optionalAccess', _131 => _131.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;
@@ -3161,7 +3248,7 @@ 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', _120 => _120.style, 'optionalAccess', _121 => _121.fontName]) || !_optionalChain([second, 'access', _122 => _122.style, 'optionalAccess', _123 => _123.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', _124 => _124.style, 'optionalAccess', _125 => _125.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', _132 => _132.style, 'optionalAccess', _133 => _133.fontName]) || !_optionalChain([second, 'access', _134 => _134.style, 'optionalAccess', _135 => _135.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', _136 => _136.style, 'optionalAccess', _137 => _137.fontName]) !== bodyFace) continue;
     first.type = "heading";
     first.level = 1;
     first.text = `${first.text.trim()} ${second.text.trim()}`;
@@ -3185,18 +3272,18 @@ 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', _126 => _126.style, 'optionalAccess', _127 => _127.fontName])) continue;
+      if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _138 => _138.style, 'optionalAccess', _139 => _139.fontName])) continue;
       const face = block.style.fontName;
       bodyChars.set(face, (_nullishCoalesce(bodyChars.get(face), () => ( 0))) + block.text.length);
     }
-    const bodyFace = _optionalChain([[...bodyChars], 'access', _128 => _128.sort, 'call', _129 => _129((a, b) => b[1] - a[1]), 'access', _130 => _130[0], 'optionalAccess', _131 => _131[0]]);
-    const anchors = page.filter((b) => b.type === "heading" && b.text && _optionalChain([b, 'access', _132 => _132.style, 'optionalAccess', _133 => _133.fontName]) && b.style.fontName !== bodyFace);
+    const bodyFace = _optionalChain([[...bodyChars], 'access', _140 => _140.sort, 'call', _141 => _141((a, b) => b[1] - a[1]), 'access', _142 => _142[0], 'optionalAccess', _143 => _143[0]]);
+    const anchors = page.filter((b) => b.type === "heading" && b.text && _optionalChain([b, 'access', _144 => _144.style, 'optionalAccess', _145 => _145.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', _134 => _134.style, 'optionalAccess', _135 => _135.fontName]) === _optionalChain([anchor, 'access', _136 => _136.style, 'optionalAccess', _137 => _137.fontName]) && Math.abs((_nullishCoalesce(_optionalChain([b, 'access', _138 => _138.style, 'optionalAccess', _139 => _139.fontSize]), () => ( 0))) - (_nullishCoalesce(_optionalChain([anchor, 'access', _140 => _140.style, 'optionalAccess', _141 => _141.fontSize]), () => ( 0)))) < 0.5);
+      const peers = anchors.filter((b) => _optionalChain([b, 'access', _146 => _146.style, 'optionalAccess', _147 => _147.fontName]) === _optionalChain([anchor, 'access', _148 => _148.style, 'optionalAccess', _149 => _149.fontName]) && Math.abs((_nullishCoalesce(_optionalChain([b, 'access', _150 => _150.style, 'optionalAccess', _151 => _151.fontSize]), () => ( 0))) - (_nullishCoalesce(_optionalChain([anchor, 'access', _152 => _152.style, 'optionalAccess', _153 => _153.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', _142 => _142.style, 'optionalAccess', _143 => _143.fontName]) === _optionalChain([anchor, 'access', _144 => _144.style, 'optionalAccess', _145 => _145.fontName]) && Math.abs((_nullishCoalesce(_optionalChain([b, 'access', _146 => _146.style, 'optionalAccess', _147 => _147.fontSize]), () => ( 0))) - (_nullishCoalesce(_optionalChain([anchor, 'access', _148 => _148.style, 'optionalAccess', _149 => _149.fontSize]), () => ( 0)))) < 0.5 && b.text.trim().length >= 3 && b.text.trim().length <= 80 && b.bbox.height <= (_nullishCoalesce(_optionalChain([b, 'access', _150 => _150.style, 'optionalAccess', _151 => _151.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', _154 => _154.style, 'optionalAccess', _155 => _155.fontName]) === _optionalChain([anchor, 'access', _156 => _156.style, 'optionalAccess', _157 => _157.fontName]) && Math.abs((_nullishCoalesce(_optionalChain([b, 'access', _158 => _158.style, 'optionalAccess', _159 => _159.fontSize]), () => ( 0))) - (_nullishCoalesce(_optionalChain([anchor, 'access', _160 => _160.style, 'optionalAccess', _161 => _161.fontSize]), () => ( 0)))) < 0.5 && b.text.trim().length >= 3 && b.text.trim().length <= 80 && b.bbox.height <= (_nullishCoalesce(_optionalChain([b, 'access', _162 => _162.style, 'optionalAccess', _163 => _163.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) {
@@ -3215,11 +3302,11 @@ 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', _152 => _152.style, 'optionalAccess', _153 => _153.fontName])) {
+    for (const b of page) if (b.type === "paragraph" && b.text && _optionalChain([b, 'access', _164 => _164.style, 'optionalAccess', _165 => _165.fontName])) {
       bodyChars.set(b.style.fontName, (_nullishCoalesce(bodyChars.get(b.style.fontName), () => ( 0))) + b.text.length);
     }
-    const bodyFace = _optionalChain([[...bodyChars], 'access', _154 => _154.sort, 'call', _155 => _155((a, b) => b[1] - a[1]), 'access', _156 => _156[0], 'optionalAccess', _157 => _157[0]]);
-    const candidates = page.filter((b) => b.type === "paragraph" && b.text && b.bbox && _optionalChain([b, 'access', _158 => _158.style, 'optionalAccess', _159 => _159.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', _166 => _166.sort, 'call', _167 => _167((a, b) => b[1] - a[1]), 'access', _168 => _168[0], 'optionalAccess', _169 => _169[0]]);
+    const candidates = page.filter((b) => b.type === "paragraph" && b.text && b.bbox && _optionalChain([b, 'access', _170 => _170.style, 'optionalAccess', _171 => _171.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}`;
@@ -3255,19 +3342,19 @@ function detectPageLeadHeadings(blocks) {
   for (const page of byPage.values()) {
     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', _160 => _160.bbox]) || !first.text || !_optionalChain([first, 'access', _161 => _161.style, 'optionalAccess', _162 => _162.fontSize]) || !second) continue;
+    if (!_optionalChain([first, 'optionalAccess', _172 => _172.bbox]) || !first.text || !_optionalChain([first, 'access', _173 => _173.style, 'optionalAccess', _174 => _174.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)$/i.test(firstText) || hasImage && /^[A-Z]/.test(firstText) && !captionLike.test(firstText) && second.type === "paragraph" && /^[A-Z]/.test(_nullishCoalesce(_optionalChain([second, 'access', _163 => _163.text, 'optionalAccess', _164 => _164.trim, 'call', _165 => _165()]), () => ( ""))) && (_nullishCoalesce(_optionalChain([second, 'access', _166 => _166.text, 'optionalAccess', _167 => _167.length]), () => ( 0))) >= 100 && second.bbox && first.bbox.y - (second.bbox.y + second.bbox.height) >= first.style.fontSize);
+    const firstIsTitle = first.type === "paragraph" && firstText.length >= 5 && firstText.length <= 80 && first.bbox.height <= first.style.fontSize * 1.6 && (/^(?:CONTENTS|Table of Contents)$/i.test(firstText) || hasImage && /^[A-Z]/.test(firstText) && !captionLike.test(firstText) && second.type === "paragraph" && /^[A-Z]/.test(_nullishCoalesce(_optionalChain([second, 'access', _175 => _175.text, 'optionalAccess', _176 => _176.trim, 'call', _177 => _177()]), () => ( ""))) && (_nullishCoalesce(_optionalChain([second, 'access', _178 => _178.text, 'optionalAccess', _179 => _179.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,}$/.test(firstText) && second.type === "paragraph" && (_nullishCoalesce(_optionalChain([second, 'access', _168 => _168.text, 'optionalAccess', _169 => _169.length]), () => ( 0))) >= 100;
+    const firstIsSection = first.type === "list" && /^\d+\.\s+[A-Z][A-Z\s]{12,}$/.test(firstText) && second.type === "paragraph" && (_nullishCoalesce(_optionalChain([second, 'access', _180 => _180.text, 'optionalAccess', _181 => _181.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', _170 => _170.style, 'optionalAccess', _171 => _171.fontName]) || !second.style.fontSize || !_optionalChain([third, 'optionalAccess', _172 => _172.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', _173 => _173.style, 'optionalAccess', _174 => _174.fontName]) !== second.style.fontName;
+    if (first.type !== "heading" || captionLike.test(firstText) || second.type !== "paragraph" || !second.text || !second.bbox || !_optionalChain([second, 'access', _182 => _182.style, 'optionalAccess', _183 => _183.fontName]) || !second.style.fontSize || !_optionalChain([third, 'optionalAccess', _184 => _184.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', _185 => _185.style, 'optionalAccess', _186 => _186.fontName]) !== second.style.fontName;
     const namedContents = /^Table of Contents$/i.test(second.text.trim());
     if (distinctFace && third.text.length >= 60 || namedContents) {
       second.type = "heading";
@@ -3278,24 +3365,24 @@ 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', _175 => _175.style, 'optionalAccess', _176 => _176.fontName])) continue;
+    if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _187 => _187.style, 'optionalAccess', _188 => _188.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', _177 => _177.text, 'optionalAccess', _178 => _178.trim, 'call', _179 => _179()]), () => ( ""));
+    const text = _nullishCoalesce(_optionalChain([block, 'access', _189 => _189.text, 'optionalAccess', _190 => _190.trim, 'call', _191 => _191()]), () => ( ""));
     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', _180 => _180.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', _192 => _192.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.,:%+\-–]+$/.test(text)) continue;
     const previous = blocks[i - 1], next = blocks[i + 1];
-    const pb = _optionalChain([previous, 'optionalAccess', _181 => _181.bbox]), nb = _optionalChain([next, 'optionalAccess', _182 => _182.bbox]);
-    if (!pb || !nb || previous.pageNumber !== block.pageNumber || next.pageNumber !== block.pageNumber || previous.type !== "paragraph" && previous.type !== "heading" || next.type !== "paragraph" || _optionalChain([next, 'access', _183 => _183.style, 'optionalAccess', _184 => _184.fontName]) !== bodyFace || !next.text || next.text.length < 40 || style.fontSize < (_nullishCoalesce(next.style.fontSize, () => ( 0))) * 0.95 || pb.y - (box.y + box.height) < style.fontSize * 0.5 || box.y - (nb.y + nb.height) < 0 || box.y - (nb.y + nb.height) > style.fontSize * 2 || box.x + box.width / 2 < nb.x - 10 || box.x + box.width / 2 > nb.x + nb.width + 10) continue;
+    const pb = _optionalChain([previous, 'optionalAccess', _193 => _193.bbox]), nb = _optionalChain([next, 'optionalAccess', _194 => _194.bbox]);
+    if (!pb || !nb || previous.pageNumber !== block.pageNumber || next.pageNumber !== block.pageNumber || previous.type !== "paragraph" && previous.type !== "heading" || next.type !== "paragraph" || _optionalChain([next, 'access', _195 => _195.style, 'optionalAccess', _196 => _196.fontName]) !== bodyFace || !next.text || next.text.length < 40 || style.fontSize < (_nullishCoalesce(next.style.fontSize, () => ( 0))) * 0.95 || pb.y - (box.y + box.height) < style.fontSize * 0.5 || box.y - (nb.y + nb.height) < 0 || box.y - (nb.y + nb.height) > style.fontSize * 2 || box.x + box.width / 2 < nb.x - 10 || box.x + box.width / 2 > nb.x + nb.width + 10) continue;
     block.type = "heading";
     block.level = 2;
   }
@@ -3345,7 +3432,7 @@ function detectMarkerHeadings(blocks) {
       block.level = 4;
       continue;
     }
-    if (/^[가-힣]{2,6}$/.test(text) && _optionalChain([block, 'access', _185 => _185.style, 'optionalAccess', _186 => _186.fontSize])) {
+    if (/^[가-힣]{2,6}$/.test(text) && _optionalChain([block, 'access', _197 => _197.style, 'optionalAccess', _198 => _198.fontSize])) {
       const prev = blocks[i - 1];
       const next = blocks[i + 1];
       const prevIsStructural = !prev || prev.type === "table" || prev.type === "heading" || prev.type === "separator";
@@ -3386,7 +3473,7 @@ 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', _187 => _187[i + 2], 'optionalAccess', _188 => _188.type]) !== "table") {
+    } else if (isCaptionCandidate(blocks[i + 1], block) && _optionalChain([blocks, 'access', _199 => _199[i + 2], 'optionalAccess', _200 => _200.type]) !== "table") {
       block.table.caption = blocks[i + 1].text.trim();
       blocks.splice(i + 1, 1);
     }
@@ -3593,21 +3680,23 @@ function detectSpecialKoreanTables(blocks) {
   flushKvTable();
   return result;
 }
-function removeHeaderFooterBlocks(blocks, pageHeights, warnings) {
+function removeHeaderFooterBlocks(blocks, pageHeights, warnings, notes, tables = false) {
   const ZONE_RATIO = 0.12;
   const MIN_REPEAT = 3;
   const topEntries = [];
   const bottomEntries = [];
   for (let bi = 0; bi < blocks.length; bi++) {
     const b = blocks[bi];
-    if (!b.bbox || !b.pageNumber || !_optionalChain([b, 'access', _189 => _189.text, 'optionalAccess', _190 => _190.trim, 'call', _191 => _191()])) continue;
+    const text = !tables ? _optionalChain([b, 'access', _201 => _201.text, 'optionalAccess', _202 => _202.trim, 'call', _203 => _203()]) : b.type === "table" && b.table ? b.table.cells.flat().map((c) => c.text.trim()).filter(Boolean).join(" | ") : void 0;
+    if (!b.bbox || !b.pageNumber || !text) continue;
     const ph = pageHeights.get(b.bbox.page) || pageHeights.get(b.pageNumber);
     if (!ph) continue;
     const blockTop = ph - (b.bbox.y + b.bbox.height);
     const blockBottom = ph - b.bbox.y;
-    const entry = { blockIdx: bi, page: b.pageNumber, text: b.text.trim() };
-    if (blockBottom <= ph * ZONE_RATIO) topEntries.push(entry);
-    else if (blockTop >= ph * (1 - ZONE_RATIO)) bottomEntries.push(entry);
+    const entry = { blockIdx: bi, page: b.pageNumber, text };
+    const boxMid = b.type === "table" && b.bbox.height <= ph * 0.05 ? (blockTop + blockBottom) / 2 : void 0;
+    if ((_nullishCoalesce(boxMid, () => ( blockBottom))) <= ph * ZONE_RATIO) topEntries.push(entry);
+    else if ((_nullishCoalesce(boxMid, () => ( blockTop))) >= ph * (1 - ZONE_RATIO)) bottomEntries.push(entry);
   }
   const removeSet = /* @__PURE__ */ new Set();
   for (const entries of [topEntries, bottomEntries]) {
@@ -3629,7 +3718,7 @@ function removeHeaderFooterBlocks(blocks, pageHeights, warnings) {
     for (const [p, count] of patternCount) {
       const pages = [..._nullishCoalesce(patternPages.get(p), () => ( []))];
       const span = pages.length ? Math.max(...pages) - Math.min(...pages) + 1 : 0;
-      const pageNumbered = (_nullishCoalesce(_optionalChain([patternNumbers, 'access', _192 => _192.get, 'call', _193 => _193(p), 'optionalAccess', _194 => _194.size]), () => ( 0))) > 1;
+      const pageNumbered = !tables && (_nullishCoalesce(_optionalChain([patternNumbers, 'access', _204 => _204.get, 'call', _205 => _205(p), 'optionalAccess', _206 => _206.size]), () => ( 0))) > 1;
       if (count >= MIN_REPEAT && pages.length >= MIN_REPEAT && (pages.length >= span * 0.4 || pageNumbered)) {
         repeatedPatterns.add(p);
       }
@@ -3637,9 +3726,11 @@ function removeHeaderFooterBlocks(blocks, pageHeights, warnings) {
     for (const e of entries) {
       const norm = e.text.replace(/\d+/g, "#");
       if (!repeatedPatterns.has(norm)) continue;
+      const noteMark = _optionalChain([e, 'access', _207 => _207.text, 'access', _208 => _208.match, 'call', _209 => _209(/^\d{1,3}\)/), 'optionalAccess', _210 => _210[0]]);
+      if (noteMark && _optionalChain([notes, 'optionalAccess', _211 => _211.get, 'call', _212 => _212(e.page), 'optionalAccess', _213 => _213.marks, 'access', _214 => _214.some, 'call', _215 => _215((m) => m.mark === noteMark)])) continue;
       const cand = blocks[e.blockIdx];
       const cb = cand.bbox;
-      const tableNote = /^(?:주\s*(?:\d+\s*[).:]|[:：])|(?:자료|출처)\s*[:：])/.test(e.text) && blocks.some((o) => {
+      const tableNote = /^(?:주\s*(?:\d+\s*[).:]|[:：])|(?:자료|출처)\s*[:：])/.test(e.text.replace(/ \| /g, " ")) && blocks.some((o) => {
         if (o.type !== "table" || !o.bbox || o.bbox.page !== cb.page) return false;
         const gap = o.bbox.y - (cb.y + cb.height);
         return gap >= 0 && gap <= cb.height && cb.x >= o.bbox.x && cb.x + cb.width <= o.bbox.x + o.bbox.width;
@@ -3650,7 +3741,7 @@ 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', _195 => _195.text, 'optionalAccess', _196 => _196.trim, 'call', _197 => _197(), 'access', _198 => _198.replace, 'call', _199 => _199(/\d+/g, "#")]);
+        const oNorm = _optionalChain([o, 'access', _216 => _216.text, 'optionalAccess', _217 => _217.trim, 'call', _218 => _218(), 'access', _219 => _219.replace, 'call', _220 => _220(/\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) {
@@ -3693,7 +3784,7 @@ function recordCellLines(cell, items) {
 function markImageCell(cell) {
   IMAGE_CELLS.add(cell);
 }
-var emptyCell = (cell) => !cell || !_optionalChain([cell, 'access', _200 => _200.text, 'optionalAccess', _201 => _201.trim, 'call', _202 => _202()]) && !_optionalChain([cell, 'access', _203 => _203.blocks, 'optionalAccess', _204 => _204.length]) && !IMAGE_CELLS.has(cell);
+var emptyCell = (cell) => !cell || !_optionalChain([cell, 'access', _221 => _221.text, 'optionalAccess', _222 => _222.trim, 'call', _223 => _223()]) && !_optionalChain([cell, 'access', _224 => _224.blocks, 'optionalAccess', _225 => _225.length]) && !IMAGE_CELLS.has(cell);
 function trimTable(table) {
   let cols = table.cols;
   while (cols > 0 && table.cells.every((row) => emptyCell(row[cols - 1]))) cols--;
@@ -3709,7 +3800,7 @@ 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', _205 => _205.blocks, 'optionalAccess', _206 => _206.length])) trimTrailingEmptyTableCols(cell.blocks);
+    for (const row of b.table.cells) for (const cell of row) if (_optionalChain([cell, 'access', _226 => _226.blocks, 'optionalAccess', _227 => _227.length])) trimTrailingEmptyTableCols(cell.blocks);
     trimTable(b.table);
   }
 }
@@ -3729,7 +3820,7 @@ 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', _207 => _207.blocks, 'optionalAccess', _208 => _208.length]));
+      return !!o && (o.colSpan > 1 || !o.text.trim() && !_optionalChain([o, 'access', _228 => _228.blocks, 'optionalAccess', _229 => _229.length]));
     });
     if (!ok) continue;
     const spanning = /* @__PURE__ */ new Set();
@@ -3795,7 +3886,7 @@ 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', _209 => _209.match, 'call', _210 => _210(/\p{L}/gu), 'optionalAccess', _211 => _211.length]), () => ( 0))) >= 2)) return false;
+    if (filled.length >= 3 && filled.length >= rows * 0.6 && filled.every((text) => text.length <= 30 && (_nullishCoalesce(_optionalChain([text, 'access', _230 => _230.match, 'call', _231 => _231(/\p{L}/gu), 'optionalAccess', _232 => _232.length]), () => ( 0))) >= 2)) return false;
   }
   return true;
 }
@@ -3822,7 +3913,7 @@ 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', _212 => _212[c], 'optionalAccess', _213 => _213.text]), () => ( ""))).split(/\s+/)).filter((token) => /^-?[\d,]+(?:\.\d+)?%?$/.test(token)).map((token) => Number(token.replace(/[,%]/g, "")));
+    const ticks = table.cells.flatMap((row) => (_nullishCoalesce(_optionalChain([row, 'access', _233 => _233[c], 'optionalAccess', _234 => _234.text]), () => ( ""))).split(/\s+/)).filter((token) => /^-?[\d,]+(?:\.\d+)?%?$/.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;
@@ -3976,7 +4067,7 @@ function detectAlignedTwoColumnTables(rows, pageNum) {
       continue;
     }
     const top = rows[i - 1];
-    const title = _optionalChain([top, 'optionalAccess', _214 => _214.items, 'access', _215 => _215.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', _235 => _235.items, 'access', _236 => _236.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 }];
@@ -4032,12 +4123,12 @@ 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', _216 => _216[c], 'optionalAccess', _217 => _217.text, 'access', _218 => _218.trim, 'call', _219 => _219()])).length;
+    const filled = t.cells.filter((row) => _optionalChain([row, 'access', _237 => _237[c], 'optionalAccess', _238 => _238.text, 'access', _239 => _239.trim, 'call', _240 => _240()])).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', _220 => _220[c], 'optionalAccess', _221 => _221.text, 'access', _222 => _222.replace, 'call', _223 => _223(/\s+/g, ""), 'access', _224 => _224.length]), () => ( 0))).filter((n) => n > 0);
+    const lens = t.cells.map((row) => _nullishCoalesce(_optionalChain([row, 'access', _241 => _241[c], 'optionalAccess', _242 => _242.text, 'access', _243 => _243.replace, 'call', _244 => _244(/\s+/g, ""), 'access', _245 => _245.length]), () => ( 0))).filter((n) => n > 0);
     const avg = lens.reduce((s, v) => s + v, 0) / (lens.length || 1);
     if (avg < 12) return false;
   }
@@ -4433,7 +4524,7 @@ function assignRowItems(items, columns, numCols, headerFace) {
   const runs = [];
   for (const it of sorted) {
     const last = runs[runs.length - 1];
-    const prev = _optionalChain([last, 'optionalAccess', _225 => _225[last.length - 1]]);
+    const prev = _optionalChain([last, 'optionalAccess', _246 => _246[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]);
   }
@@ -4507,7 +4598,7 @@ 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', _226 => _226[0], 'optionalAccess', _227 => _227.items, 'access', _228 => _228.map, 'call', _229 => _229((i) => i.fontName)]));
+  const headFaces = new Set(_optionalChain([rows, 'access', _247 => _247[0], 'optionalAccess', _248 => _248.items, 'access', _249 => _249.map, 'call', _250 => _250((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];
@@ -4537,7 +4628,7 @@ 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', _230 => _230[r], 'access', _231 => _231.find, 'call', _232 => _232((c) => c.text.trim()), 'optionalAccess', _233 => _233.text, 'access', _234 => _234.trim, 'call', _235 => _235()]) || "";
+    const contentText = _optionalChain([cells, 'access', _251 => _251[r], 'access', _252 => _252.find, 'call', _253 => _253((c) => c.text.trim()), 'optionalAccess', _254 => _254.text, 'access', _255 => _255.trim, 'call', _256 => _256()]) || "";
     if (/^[○●▶\-·]/.test(contentText)) continue;
     for (let pr = r - 1; pr >= 0; pr--) {
       if (cells[pr].some((c) => c.text.trim())) {
@@ -5136,16 +5227,16 @@ 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 = _chunkNWODPYBDcjs.safeMax.call(void 0, gaps);
+  const maxGap = _chunkSTM3ORY7cjs.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 = _chunkNWODPYBDcjs.safeMin.call(void 0, gaps);
+  const minGap = _chunkSTM3ORY7cjs.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 = _chunkNWODPYBDcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkNWODPYBDcjs.safeMin.call(void 0, allItems.map((i) => i.x));
+  const pageWidth = _chunkSTM3ORY7cjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkSTM3ORY7cjs.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++) {
@@ -5868,17 +5959,23 @@ function splitTwoColumnProse(items, cutX) {
     if (last && Math.abs(last[0].y - c.y) <= 3) last.push(c);
     else crossLines.push([c]);
   }
-  const bandItem = (arr) => arr.filter((i) => {
+  const scriptOf = (i, cl) => {
+    const fs = cl[0].fontSize, dy = i.y - cl[0].y;
+    if (i.fontSize > fs * 0.85 || dy > fs * 0.6 || dy < -fs * 0.35) return false;
+    return cl.some((c) => c !== i && c.fontSize > fs * 0.85 && i.x - (c.x + c.w) <= fs * 0.35 && i.x - (c.x + c.w) >= -fs * 0.1);
+  };
+  const bandItem = (arr, fits) => arr.filter((i) => {
     for (const cl of crossLines) {
-      if (Math.abs(cl[0].y - i.y) <= 3) {
+      if (fits(i, cl)) {
         cl.push(i);
         return false;
       }
     }
     return true;
   });
-  const leftRest = bandItem(left);
-  const rightRest = bandItem(right);
+  const sameY = (i, cl) => Math.abs(cl[0].y - i.y) <= 3;
+  const leftRest = bandItem(bandItem(left, sameY), scriptOf);
+  const rightRest = bandItem(bandItem(right, sameY), scriptOf);
   const boundYs = crossLines.map((cl) => cl[0].y);
   const bandOf = (y) => {
     let k = 0;
@@ -5890,7 +5987,7 @@ function splitTwoColumnProse(items, cutX) {
     const L = leftRest.filter((i) => bandOf(i.y) === k);
     const R = rightRest.filter((i) => bandOf(i.y) === k);
     groups.push(...columnPair(L, R));
-    if (k < crossLines.length) groups.push(crossLines[k]);
+    if (k < crossLines.length) groups.push(crossLines[k].sort((a, b) => b.y - a.y || a.x - b.x));
   }
   return groups;
 }
@@ -5965,7 +6062,8 @@ function topTableBand(items) {
       h: i.h,
       fontSize: i.fontSize,
       fontName: i.fontName,
-      hasSpaceBefore: i.hasSpaceBefore
+      hasSpaceBefore: i.hasSpaceBefore,
+      syntheticSpace: i.syntheticSpace
     })), 1);
     if (!candidate.some((t) => t.table.cols >= 3 && t.table.rows >= 3 && t.usedItems.size >= top.length * 0.75)) continue;
     return { top, rest: lines.slice(n).flat() };
@@ -6019,7 +6117,7 @@ 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', _236 => _236[i + 1], 'optionalAccess', _237 => _237.start]), () => ( lines.length))))) return null;
+  if (runs.some((run, i) => !caption(run.end, _nullishCoalesce(_optionalChain([runs, 'access', _257 => _257[i + 1], 'optionalAccess', _258 => _258.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);
@@ -6182,7 +6280,7 @@ function extractPageBlocksWithLines(items, pageNum, opList, pageWidth, pageHeigh
   const filled = fillBlanks(items, horizontals, verticals);
   items = filled.items;
   horizontals = filled.horizontals;
-  const prevPage = _optionalChain([carry, 'optionalAccess', _238 => _238.page]) === pageNum - 1 ? carry.clip : void 0;
+  const prevPage = _optionalChain([carry, 'optionalAccess', _259 => _259.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;
@@ -6211,7 +6309,7 @@ function extractPageBlocksWithLines(items, pageNum, opList, pageWidth, pageHeigh
   if (underlines.size) horizontals = horizontals.filter((l) => !underlines.has(l));
   wrapUnderlineRuns(items);
   const lineGrids = detectTables ? buildTableGrids(horizontals, verticals) : [];
-  const tableClipGrids = dropInsetClipGrids(dropShadingClipGrids(clipGrids, lineGrids, extracted.fillRects, verticals), lineGrids);
+  const tableClipGrids = dropHeadBandClipGrids(dropInsetClipGrids(dropShadingClipGrids(clipGrids, lineGrids, extracted.fillRects, verticals), lineGrids), lineGrids);
   const grids = [...tableClipGrids, ...dropGridsInside(lineGrids, tableClipGrids, clipResult.containers)];
   if (grids.length === 1 && grids[0].rowYs.length === 2 && grids[0].colXs.length === 2 && grids[0].bbox.x2 - grids[0].bbox.x1 > pageWidth * 1.2 && grids[0].bbox.y2 - grids[0].bbox.y1 > pageHeight * 1.2) {
     return extractPageBlocksFallback(items, pageNum, true, detectTables, lex);
@@ -6415,6 +6513,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       fontSize: i.fontSize,
       fontName: i.fontName,
       hasSpaceBefore: i.hasSpaceBefore,
+      syntheticSpace: i.syntheticSpace,
       seq: i.seq
     }));
     const cellTextMap = mapTextToCells(textItems, cells);
@@ -6482,13 +6581,13 @@ 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', _239 => _239[0], 'optionalAccess', _240 => _240[0], 'optionalAccess', _241 => _241.text]), () => ( "")))) {
+    if (unitLine.length > 0 && rebuiltUsed && !/^\s*\(\s*단위\s*[:：]/.test(_nullishCoalesce(_optionalChain([finalGrid, 'access', _260 => _260[0], 'optionalAccess', _261 => _261[0], 'optionalAccess', _262 => _262.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);
     }
     const filledCells = finalGrid.flat().filter((c) => c.text.trim()).length;
-    if (!rebuiltUsed && filledCells * 2 >= finalGrid.flat().length && !/^\s*\(\s*단위\s*[:：]/.test(_nullishCoalesce(_optionalChain([finalGrid, 'access', _242 => _242[0], 'optionalAccess', _243 => _243[0], 'optionalAccess', _244 => _244.text]), () => ( "")))) {
+    if (!rebuiltUsed && filledCells * 2 >= finalGrid.flat().length && !/^\s*\(\s*단위\s*[:：]/.test(_nullishCoalesce(_optionalChain([finalGrid, 'access', _263 => _263[0], 'optionalAccess', _264 => _264[0], 'optionalAccess', _265 => _265.text]), () => ( "")))) {
       const above = items.filter((it) => !usedItems.has(it) && it.y >= grid.bbox.y2 && it.y - grid.bbox.y2 <= 8 && it.x >= grid.bbox.x1 - 3 && it.x + it.w <= grid.bbox.x2 + 3);
       const line = above.filter((it) => Math.abs(it.y - Math.min(...above.map((a) => a.y))) <= 1).sort((a, b) => a.x - b.x);
       const text = line.map((it) => it.text).join("");
@@ -6501,7 +6600,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
     }
     let semanticOneColumn = false;
     if (!grid.cells && numCols === 1 && numGridRows >= 2) {
-      const populatedRows = finalGrid.filter((row) => _optionalChain([row, 'access', _245 => _245[0], 'optionalAccess', _246 => _246.text, 'access', _247 => _247.trim, 'call', _248 => _248()]));
+      const populatedRows = finalGrid.filter((row) => _optionalChain([row, 'access', _266 => _266[0], 'optionalAccess', _267 => _267.text, 'access', _268 => _268.trim, 'call', _269 => _269()]));
       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));
@@ -6527,6 +6626,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
           fontSize: i.fontSize,
           fontName: i.fontName,
           hasSpaceBefore: i.hasSpaceBefore,
+          syntheticSpace: i.syntheticSpace,
           seq: i.seq
         })))), colSpan: 1, rowSpan: 1 })));
         finalRows++;
@@ -6575,7 +6675,7 @@ 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', _249 => _249[0], 'optionalAccess', _250 => _250[0], 'optionalAccess', _251 => _251.text]), () => ( ""));
+      const prose = _nullishCoalesce(_optionalChain([finalGrid, 'access', _270 => _270[0], 'optionalAccess', _271 => _271[0], 'optionalAccess', _272 => _272.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);
@@ -6623,12 +6723,13 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       h: i.h,
       fontSize: i.fontSize,
       fontName: i.fontName,
-      hasSpaceBefore: i.hasSpaceBefore
+      hasSpaceBefore: i.hasSpaceBefore,
+      syntheticSpace: i.syntheticSpace
     }));
     const proseColumns = findTwoColumnProseCutX(clusterItems) !== null || detectPersistentColumnGutter(remaining.map((i) => ({ x: i.x, y: i.y, w: i.w, h: i.h > 0 ? i.h : i.fontSize }))) !== null;
     const clusterResults = (proseColumns ? [] : detectClusterTables(clusterItems, pageNum)).filter((cr) => {
       const b = cr.bbox;
-      return !blocks.some((x) => x.type === "paragraph" && x.bbox && (_nullishCoalesce(_optionalChain([x, 'access', _252 => _252.text, 'optionalAccess', _253 => _253.match, 'call', _254 => _254(/\p{L}/gu), 'optionalAccess', _255 => _255.length]), () => ( 0))) >= 10 && Math.min(x.bbox.x + x.bbox.width, b.x + b.width) - Math.max(x.bbox.x, b.x) >= 0.8 * b.width && x.bbox.y >= b.y - 2 && x.bbox.y + x.bbox.height <= b.y + b.height + 2);
+      return !blocks.some((x) => x.type === "paragraph" && x.bbox && (_nullishCoalesce(_optionalChain([x, 'access', _273 => _273.text, 'optionalAccess', _274 => _274.match, 'call', _275 => _275(/\p{L}/gu), 'optionalAccess', _276 => _276.length]), () => ( 0))) >= 10 && Math.min(x.bbox.x + x.bbox.width, b.x + b.width) - Math.max(x.bbox.x, b.x) >= 0.8 * b.width && x.bbox.y >= b.y - 2 && x.bbox.y + x.bbox.height <= b.y + b.height + 2);
     });
     if (clusterResults.length > 0) {
       const ciToIdx = /* @__PURE__ */ new Map();
@@ -6662,7 +6763,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
     if (gutterX !== null) {
       const gx = gutterX;
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkNWODPYBDcjs.safeMax.call(void 0, allY) - _chunkNWODPYBDcjs.safeMin.call(void 0, allY);
+      const pageH = _chunkSTM3ORY7cjs.safeMax.call(void 0, allY) - _chunkSTM3ORY7cjs.safeMin.call(void 0, allY);
       const gapThreshold = Math.max(15, pageH * 0.03);
       const sides = [
         remaining.filter((i) => i.x + i.w <= gx),
@@ -6682,7 +6783,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       finalTextBlocks = detectListBlocks(textBlocks);
     } else {
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkNWODPYBDcjs.safeMax.call(void 0, allY) - _chunkNWODPYBDcjs.safeMin.call(void 0, allY);
+      const pageH = _chunkSTM3ORY7cjs.safeMax.call(void 0, allY) - _chunkSTM3ORY7cjs.safeMin.call(void 0, allY);
       const groups = xyCutOrder(remaining, Math.max(15, pageH * 0.03));
       const textBlocks = [];
       for (const group of groups) {
@@ -6937,7 +7038,8 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false, detectTable
     h: i.h,
     fontSize: i.fontSize,
     fontName: i.fontName,
-    hasSpaceBefore: i.hasSpaceBefore
+    hasSpaceBefore: i.hasSpaceBefore,
+    syntheticSpace: i.syntheticSpace
   }));
   const textRects = items.map((i) => ({ x: i.x, y: i.y, w: i.w, h: i.h > 0 ? i.h : i.fontSize }));
   const earlyProseCut = fullPage && detectTables ? _nullishCoalesce(_nullishCoalesce(findTwoColumnProseCutX(clusterItems), () => ( persistentGutter(textRects))), () => ( (figures.length > 0 ? detectColumnGutter([...textRects, ...figures]) : null))) : null;
@@ -6995,7 +7097,7 @@ 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 = _chunkNWODPYBDcjs.safeMax.call(void 0, allY) - _chunkNWODPYBDcjs.safeMin.call(void 0, allY);
+      const pageHeight = _chunkSTM3ORY7cjs.safeMax.call(void 0, allY) - _chunkSTM3ORY7cjs.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) {
@@ -7055,5 +7157,6 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false, detectTable
 
 
 
-exports.OCR_DET_MODEL = OCR_DET_MODEL; exports.OCR_REC_MODEL = OCR_REC_MODEL; exports.OCR_REC_DICT = OCR_REC_DICT; exports.getOcrModelsDir = getOcrModelsDir; exports.ensureOcrModels = ensureOcrModels; exports.ocrModelsCached = ocrModelsCached; exports.parseCharacterDict = parseCharacterDict; exports.extractLines = extractLines; 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-MH6MAL64.cjs.map
+
+exports.OCR_DET_MODEL = OCR_DET_MODEL; exports.OCR_REC_MODEL = OCR_REC_MODEL; exports.OCR_REC_DICT = OCR_REC_DICT; exports.getOcrModelsDir = getOcrModelsDir; exports.ensureOcrModels = ensureOcrModels; exports.ocrModelsCached = ocrModelsCached; exports.parseCharacterDict = parseCharacterDict; exports.extractLines = extractLines; exports.extractImageRegions = extractImageRegions; exports.filterHiddenText = filterHiddenText; exports.collapseEvenSpacing = collapseEvenSpacing; exports.normalizeItems = normalizeItems; exports.WrapLexicon = WrapLexicon; exports.startsNewItem = startsNewItem; 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-WDQTFAXY.cjs.map