kordoc 4.16.2 → 4.17.0

package/dist/{chunk-WDQTFAXY.cjs → chunk-XEPKOC2O.cjs}

@@ -3,15 +3,16 @@
 
 
 
-var _chunk62W7QCEXcjs = require('./chunk-62W7QCEX.cjs');
 
+var _chunkAKB72TEHcjs = require('./chunk-AKB72TEH.cjs');
 
 
-var _chunkCEIATRDWcjs = require('./chunk-CEIATRDW.cjs');
 
+var _chunkWYON5ANUcjs = require('./chunk-WYON5ANU.cjs');
 
 
-var _chunkSTM3ORY7cjs = require('./chunk-STM3ORY7.cjs');
+
+var _chunkQNJLLF4Jcjs = require('./chunk-QNJLLF4J.cjs');
 
 // src/ocr/models.ts
 var _path = require('path');
@@ -39,10 +40,10 @@ var OCR_REC_DICT = {
 };
 var ALL_OCR_MODELS = [OCR_DET_MODEL, OCR_REC_MODEL, OCR_REC_DICT];
 function getOcrModelsDir() {
-  return _chunkCEIATRDWcjs.getModelsDir.call(void 0, "ppocr");
+  return _chunkWYON5ANUcjs.getModelsDir.call(void 0, "ppocr");
 }
 async function ensureOcrModels(onProgress) {
-  return _chunkCEIATRDWcjs.ensureModelsIn.call(void 0, getOcrModelsDir(), ALL_OCR_MODELS, onProgress);
+  return _chunkWYON5ANUcjs.ensureModelsIn.call(void 0, getOcrModelsDir(), ALL_OCR_MODELS, onProgress);
 }
 async function ocrModelsCached() {
   const dir = getOcrModelsDir();
@@ -103,6 +104,9 @@ function extractLines(fnArray, argsArray) {
   const colorStack = [];
   let fillAlpha = 1, strokeAlpha = 1;
   const alphaStack = [];
+  let strokeColor = 0;
+  const strokeColorStack = [];
+  const nonRules = /* @__PURE__ */ new Set();
   const hiddenBoxes = [];
   const paintedFills = /* @__PURE__ */ new Map();
   const thinFill = /* @__PURE__ */ new Set();
@@ -170,6 +174,7 @@ function extractLines(fnArray, argsArray) {
       for (const r of pathRects) painted.push(r);
       paintedFills.set(fillColor, painted);
     }
+    const areaFill = fromFill && (pathRects.length > 0 || thickPath(currentPath));
     pathRects = [];
     pathRectSegs = [];
     if (!isStroke) {
@@ -180,6 +185,9 @@ function extractLines(fnArray, argsArray) {
     }
     pendingClip = false;
     const effWidth = lineWidth * ctmScale();
+    const outs = [horizontals, verticals, shortH, shortV, thinShortH, thinShortV];
+    const white = (fromFill ? fillColor : strokeColor) === 16777215;
+    const from = white || areaFill ? outs.map((o) => o.length) : void 0;
     currentPath.forEach((seg, k) => {
       const role = segRole[k];
       if (role === "hidden") return;
@@ -191,6 +199,9 @@ function extractLines(fnArray, argsArray) {
       }
       classifyAndAdd(seg, effWidth, horizontals, verticals, fromFill, fromFill ? void 0 : { h: shortH, v: shortV });
     });
+    if (from) outs.forEach((o, k) => {
+      for (let n = from[k]; n < o.length; n++) if (white || k < 2 && !thinFill.has(o[n])) nonRules.add(o[n]);
+    });
     currentPath = [];
   }
   for (let i = 0; i < fnArray.length; i++) {
@@ -203,11 +214,13 @@ function extractLines(fnArray, argsArray) {
       case _pdfmjs.OPS.save:
         ctmStack.push(ctm.slice());
         colorStack.push(fillColor);
+        strokeColorStack.push(strokeColor);
         alphaStack.push([fillAlpha, strokeAlpha]);
         break;
       case _pdfmjs.OPS.restore:
         ctm = _nullishCoalesce(ctmStack.pop(), () => ( [1, 0, 0, 1, 0, 0]));
         fillColor = _nullishCoalesce(colorStack.pop(), () => ( 0));
+        strokeColor = _nullishCoalesce(strokeColorStack.pop(), () => ( 0));
         [fillAlpha, strokeAlpha] = _nullishCoalesce(alphaStack.pop(), () => ( [1, 1]));
         break;
       case _pdfmjs.OPS.setGState: {
@@ -225,12 +238,18 @@ function extractLines(fnArray, argsArray) {
         fillColor = c[0] << 16 | c[1] << 8 | c[2];
         break;
       }
+      case _pdfmjs.OPS.setStrokeRGBColor: {
+        const c = args;
+        strokeColor = c[0] << 16 | c[1] << 8 | c[2];
+        break;
+      }
       case _pdfmjs.OPS.transform:
       case _pdfmjs.OPS.paintFormXObjectBegin: {
         let t = args;
         if (op === _pdfmjs.OPS.paintFormXObjectBegin) {
           ctmStack.push(ctm.slice());
           colorStack.push(fillColor);
+          strokeColorStack.push(strokeColor);
           alphaStack.push([fillAlpha, strokeAlpha]);
           const m = args[0];
           if (!Array.isArray(m) || m.length < 6) break;
@@ -249,6 +268,7 @@ function extractLines(fnArray, argsArray) {
       case _pdfmjs.OPS.paintFormXObjectEnd:
         ctm = _nullishCoalesce(ctmStack.pop(), () => ( [1, 0, 0, 1, 0, 0]));
         fillColor = _nullishCoalesce(colorStack.pop(), () => ( 0));
+        strokeColor = _nullishCoalesce(strokeColorStack.pop(), () => ( 0));
         [fillAlpha, strokeAlpha] = _nullishCoalesce(alphaStack.pop(), () => ( [1, 1]));
         break;
       case _pdfmjs.OPS.constructPath: {
@@ -360,7 +380,8 @@ function extractLines(fnArray, argsArray) {
     fillRects,
     hiddenBoxes,
     shortH,
-    shortV
+    shortV,
+    nonRules
   };
 }
 var SHORT_CHAIN_POS_TOL = 0.5;
@@ -415,6 +436,17 @@ function chainShortSegments(longs, shorts, dir, chainable = (l) => !l.fromFill)
 }
 var CLIP_MIN_W = 4;
 var CLIP_MIN_H = 2;
+function thickPath(path) {
+  if (!path.length) return false;
+  let x1 = Infinity, y1 = Infinity, x2 = -Infinity, y2 = -Infinity;
+  for (const s of path) {
+    x1 = Math.min(x1, s.x1, s.x2);
+    x2 = Math.max(x2, s.x1, s.x2);
+    y1 = Math.min(y1, s.y1, s.y2);
+    y2 = Math.max(y2, s.y1, s.y2);
+  }
+  return Math.min(x2 - x1, y2 - y1) > THIN_FILL_MAX;
+}
 function captureClipRect(path, out, minW = CLIP_MIN_W, minH = CLIP_MIN_H) {
   if (path.length < 3 || path.length > 5) return;
   let x1 = Infinity, y1 = Infinity, x2 = -Infinity, y2 = -Infinity;
@@ -538,7 +570,7 @@ var CHAIN_GAP = 3;
 function chainCollinearRules(horizontals) {
   if (horizontals.length <= 1) return horizontals;
   const sorted = [...horizontals].sort((a, b) => a.y1 - b.y1 || a.x1 - b.x1);
-  const rules = [];
+  const rules2 = [];
   let bandStart = 0;
   const flushBand = (end) => {
     const band = sorted.slice(bandStart, end).sort((a, b) => a.x1 - b.x1);
@@ -549,11 +581,11 @@ function chainCollinearRules(horizontals) {
         if (seg.x2 > cur.x2) cur.x2 = seg.x2;
         if (seg.lineWidth > cur.lineWidth) cur.lineWidth = seg.lineWidth;
       } else {
-        rules.push(cur);
+        rules2.push(cur);
         cur = { ...seg };
       }
     }
-    rules.push(cur);
+    rules2.push(cur);
   };
   for (let i = 1; i <= sorted.length; i++) {
     if (i === sorted.length || sorted[i].y1 - sorted[bandStart].y1 > CHAIN_Y_TOL) {
@@ -561,7 +593,7 @@ function chainCollinearRules(horizontals) {
       bandStart = i;
     }
   }
-  return rules;
+  return rules2;
 }
 function closeOpenTableEdges(horizontals, verticals) {
   if (horizontals.length < EDGE_MIN_RULES) return verticals;
@@ -827,7 +859,7 @@ function bridgeSkippedRowVerticals(horizontals, verticals, items) {
       if (hi - lo > BRIDGE_MIN_GAP && hi - lo <= 40) gaps.push({ x, lo, hi, spanLo: s[i - 1].y1, spanHi: s[i].y2 });
     }
   }
-  const rules = chainCollinearRules(horizontals);
+  const rules2 = chainCollinearRules(horizontals);
   const added = [];
   const done = /* @__PURE__ */ new Set();
   for (const g of gaps) {
@@ -837,7 +869,7 @@ function bridgeSkippedRowVerticals(horizontals, verticals, items) {
     if (band.length < BRIDGE_MIN_COLUMNS) continue;
     const xs = band.map((o) => o.x);
     const x1 = Math.min(...xs), x2 = Math.max(...xs);
-    const ruled = (y) => rules.some((h) => Math.abs(h.y1 - y) <= BRIDGE_ENDPOINT_TOL && h.x1 <= x1 + BRIDGE_ENDPOINT_TOL && h.x2 >= x2 - BRIDGE_ENDPOINT_TOL);
+    const ruled = (y) => rules2.some((h) => Math.abs(h.y1 - y) <= BRIDGE_ENDPOINT_TOL && h.x1 <= x1 + BRIDGE_ENDPOINT_TOL && h.x2 >= x2 - BRIDGE_ENDPOINT_TOL);
     if (!ruled(g.lo) || !ruled(g.hi)) continue;
     const rowItems = items.filter((it) => {
       const cy = it.y + (it.h || it.fontSize) / 2;
@@ -1166,7 +1198,7 @@ function clusterCoordinates(values, tolerance) {
 function chainVerticals(vs) {
   if (vs.length <= 1) return vs.map((v) => ({ y1: v.y1, y2: v.y2 }));
   const sorted = [...vs].sort((a, b) => a.x1 - b.x1 || a.y1 - b.y1);
-  const rules = [];
+  const rules2 = [];
   let bandStart = 0;
   const flushBand = (end) => {
     const band = sorted.slice(bandStart, end).sort((a, b) => a.y1 - b.y1);
@@ -1176,11 +1208,11 @@ function chainVerticals(vs) {
       if (seg.y1 - cur.y2 <= CUT_VCHAIN_GAP) {
         if (seg.y2 > cur.y2) cur.y2 = seg.y2;
       } else {
-        rules.push(cur);
+        rules2.push(cur);
         cur = { y1: seg.y1, y2: seg.y2 };
       }
     }
-    rules.push(cur);
+    rules2.push(cur);
   };
   for (let i = 1; i <= sorted.length; i++) {
     if (i === sorted.length || sorted[i].x1 - sorted[bandStart].x1 > CUT_VCHAIN_X_TOL) {
@@ -1188,7 +1220,7 @@ function chainVerticals(vs) {
       bandStart = i;
     }
   }
-  return rules;
+  return rules2;
 }
 function splitStackedGroup(group) {
   const hs = group.filter((l) => l.type === "h");
@@ -1538,11 +1570,42 @@ function buildClipCellGrids(rects, strokedH, strokedV, pageWidth, pageHeight, te
       }
     }
   }
+  const stacked = /* @__PURE__ */ new Set();
+  if (ruledGaps.some((g) => g.axis === "y" && g.both && g.hi - g.lo > STACKED_TABLE_GAP)) {
+    for (let i = 0; i < cells.length; i++) {
+      if (tableClip[i]) continue;
+      const k = edgeBin(cells[i].y2);
+      for (let d = -1; d <= 1; d++) {
+        if ((_nullishCoalesce(startsY.get(k + d), () => ( []))).some((j) => j !== i && !tableClip[j] && parent[j] === parent[i] && adjacent(cells[i], cells[j]))) stacked.add(find(i));
+      }
+    }
+  }
+  const comp = cells.map((_, i) => find(i));
   for (const g of ruledGaps) {
     if (tableClip[g.i] || tableClip[g.j]) continue;
+    if (g.axis === "y" && g.both && g.hi - g.lo > STACKED_TABLE_GAP && comp[g.i] !== comp[g.j] && stacked.has(comp[g.i]) && stacked.has(comp[g.j])) continue;
     const ra = find(g.i), rb = find(g.j);
     if (ra !== rb) root[ra] = rb;
   }
+  if (prev.lastCells.length) {
+    const headTop = pageHeight * (1 - HEADER_BAND);
+    let top = -Infinity;
+    for (let i = 0; i < cells.length; i++) if (parent[i] < 0 && !tableClip[i]) top = Math.max(top, cells[i].y2);
+    const band = cells.flatMap((c, i) => parent[i] < 0 && !tableClip[i] && Math.abs(c.y2 - top) <= CLIP_COORD_TOL ? [i] : []);
+    const y1 = band.length ? cells[band[0]].y1 : 0;
+    if (band.length >= 2 && band.every((i) => Math.abs(cells[i].y1 - y1) <= CLIP_COORD_TOL) && !textPoints.some((p) => p.y > top + CLIP_EDGE_TOL && p.y < headTop)) {
+      for (let f = 0; f < cells.length; f++) {
+        const b = cells[f];
+        if (parent[f] >= 0 || tableClip[f] || Math.abs(b.y2 - y1) > CLIP_ADJ_GAP) continue;
+        const on = band.filter((i) => cells[i].x1 >= b.x1 - CLIP_COORD_TOL && cells[i].x2 <= b.x2 + CLIP_COORD_TOL);
+        if (on.length < 2 || !on.some((i) => adjacent(cells[i], b))) continue;
+        for (const i of on) {
+          const ra = find(i), rb = find(f);
+          if (ra !== rb) root[ra] = rb;
+        }
+      }
+    }
+  }
   const groups = /* @__PURE__ */ new Map();
   for (let i = 0; i < cells.length; i++) {
     const r = find(i);
@@ -1602,6 +1665,8 @@ function buildClipCellGrids(rects, strokedH, strokedV, pageWidth, pageHeight, te
     if (invisible / edges < CLIP_MIN_INVISIBLE) continue;
     const colXs = dropSliverGaps(clusterCoords(members.flatMap((r) => [r.x1, r.x2])));
     const rowYs = clusterCoords(members.flatMap((r) => [r.y1, r.y2])).reverse();
+    const band = parentRect ? void 0 : carriedBandTop(colXs, rowYs[0], strokedH, strokedV, textPoints, cells, prev.lastCells, headY);
+    if (band !== void 0) rowYs.unshift(band);
     const numRows = rowYs.length - 1, numCols = colXs.length - 1;
     if (numRows < 1 || numCols < 1) continue;
     const occupied = Array.from({ length: numRows }, () => new Array(numCols).fill(false));
@@ -1648,6 +1713,7 @@ function buildClipCellGrids(rects, strokedH, strokedV, pageWidth, pageHeight, te
   return { grids, containers, page: { lastCells, clips: cells } };
 }
 var CLIP_SPACING_MAX = 3;
+var STACKED_TABLE_GAP = 2.6;
 function findRuledGaps(cells, parent, strokedH, strokedV) {
   const al = (u, v) => Math.abs(u - v) <= CLIP_COORD_TOL;
   const occupied = (i, j, x1, y1, x2, y2) => cells.some((k, n) => n !== i && n !== j && parent[n] === parent[i] && overlap(k.x1, k.x2, x1, x2) > CLIP_ADJ_GAP && overlap(k.y1, k.y2, y1, y2) > CLIP_ADJ_GAP);
@@ -1777,6 +1843,20 @@ function closeGaps(members, gaps) {
   }
   return out;
 }
+function carriedBandTop(colXs, top, strokedH, strokedV, textPoints, cells, prevLast, headY) {
+  const x1 = colXs[0], x2 = colXs[colXs.length - 1];
+  if (!prevLast.some((b) => overlap(b.x1, b.x2, x1, x2) > CLIP_EDGE_TOL)) return void 0;
+  let y;
+  for (const l of strokedH) {
+    if (l.y1 > top + CLIP_MIN_H2 && l.x1 <= x1 + STROKE_NEAR && l.x2 >= x2 - STROKE_NEAR && (y === void 0 || l.y1 < y)) y = l.y1;
+  }
+  if (y === void 0 || !edgeStroked(strokedV, "v", x1, top, y) || !edgeStroked(strokedV, "v", x2, top, y)) return void 0;
+  const above = Math.max(y, headY);
+  const inX = (x) => x > x1 && x < x2;
+  if (textPoints.some((p) => inX(p.x) && p.y > top && p.y < above)) return void 0;
+  if (cells.some((c) => overlap(c.x1, c.x2, x1, x2) > CLIP_EDGE_TOL && c.y1 >= top - CLIP_EDGE_TOL && c.y1 < above)) return void 0;
+  return y;
+}
 function addNarrowEdgeCols(colXs, rowYs, out, fills) {
   if (!fills.length) return;
   const rowIdx = (y) => {
@@ -1979,7 +2059,7 @@ function tagScripts(text, lines) {
     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);
+  return _chunkAKB72TEHcjs.tidyScriptTags.call(void 0, out);
 }
 
 // src/pdf/line-wrap.ts
@@ -3131,9 +3211,9 @@ function detectHeadings(blocks, medianFontSize) {
     if (/^\d+$/.test(text) && !isChapterNumber(block, blocks[bi + 1], medianFontSize)) continue;
     const ratio = block.style.fontSize / medianFontSize;
     let level = 0;
-    if (ratio >= _chunk62W7QCEXcjs.HEADING_RATIO_H1) level = 1;
-    else if (ratio >= _chunk62W7QCEXcjs.HEADING_RATIO_H2) level = 2;
-    else if (ratio >= _chunk62W7QCEXcjs.HEADING_RATIO_H3) level = 3;
+    if (ratio >= _chunkAKB72TEHcjs.HEADING_RATIO_H1) level = 1;
+    else if (ratio >= _chunkAKB72TEHcjs.HEADING_RATIO_H2) level = 2;
+    else if (ratio >= _chunkAKB72TEHcjs.HEADING_RATIO_H3) level = 3;
     if (level > 0) {
       block.type = "heading";
       block.level = level;
@@ -3146,7 +3226,7 @@ function isChapterNumber(block, next, medianFontSize) {
   const nextSize = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _76 => _76.style, 'optionalAccess', _77 => _77.fontSize]), () => ( 0));
   const text = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _78 => _78.text, 'optionalAccess', _79 => _79.trim, 'call', _80 => _80()]), () => ( ""));
   const a = block.bbox, b = _optionalChain([next, 'optionalAccess', _81 => _81.bbox]);
-  return /^\d{1,2}$/.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.]+$/.test(text) && !!a && !!b && a.y > b.y && a.y - (b.y + b.height) <= size * 2;
+  return /^\d{1,2}$/.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 * _chunkAKB72TEHcjs.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; ) {
@@ -3761,6 +3841,9 @@ function removeHeaderFooterBlocks(blocks, pageHeights, warnings, notes, tables =
 // src/pdf/table-meta.ts
 var CLIP_TABLES = /* @__PURE__ */ new WeakSet();
 var TABLE_COLXS = /* @__PURE__ */ new WeakMap();
+var TABLE_ROWYS = /* @__PURE__ */ new WeakMap();
+var TABLE_TAIL = /* @__PURE__ */ new WeakMap();
+var PART_COLXS = /* @__PURE__ */ new WeakMap();
 var EMPTY_PARTS = /* @__PURE__ */ new WeakSet();
 var IMAGE_CELLS = /* @__PURE__ */ new WeakSet();
 var FILLER_CELLS = /* @__PURE__ */ new WeakSet();
@@ -3779,6 +3862,34 @@ function recordCellLines(cell, items) {
   }
   CELL_LINES.set(cell, lines);
 }
+var ROW_RULES = /* @__PURE__ */ new WeakMap();
+function ruleCover(hs, y, x1, x2) {
+  const segs = hs.filter((l) => Math.abs(l.y1 - y) <= 1.5 && l.x2 > x1 && l.x1 < x2).map((l) => [Math.max(l.x1, x1), Math.min(l.x2, x2)]).sort((a, b) => a[0] - b[0]);
+  let len = 0, end = x1;
+  for (const [a, b] of segs) if (b > end) {
+    len += b - Math.max(a, end);
+    end = b;
+  }
+  return x2 > x1 ? len / (x2 - x1) : 0;
+}
+function recordRowRules(table, cells, bbox, hs) {
+  const real = cells.filter((c) => !c.filler);
+  const edge = (y, pick) => {
+    let w = 0, covered = 0;
+    for (const c of real) if (Math.abs(pick(c) - y) <= 1) {
+      w += c.bbox.x2 - c.bbox.x1;
+      covered += ruleCover(hs, y, c.bbox.x1, c.bbox.x2) * (c.bbox.x2 - c.bbox.x1);
+    }
+    return w > 0 && covered >= 0.9 * w;
+  };
+  let innerRuled = 0, innerOpen = 0;
+  for (const c of real) {
+    if (Math.abs(c.bbox.y1 - bbox.y1) <= 1) continue;
+    if (ruleCover(hs, c.bbox.y1, c.bbox.x1 + 0.5, c.bbox.x2 - 0.5) >= 0.9) innerRuled++;
+    else innerOpen++;
+  }
+  ROW_RULES.set(table, { top: edge(bbox.y2, (c) => c.bbox.y2), bottom: edge(bbox.y1, (c) => c.bbox.y1), innerRuled, innerOpen });
+}
 
 // src/pdf/table-trim.ts
 function markImageCell(cell) {
@@ -5227,16 +5338,16 @@ 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 = _chunkSTM3ORY7cjs.safeMax.call(void 0, gaps);
+  const maxGap = _chunkQNJLLF4Jcjs.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 = _chunkSTM3ORY7cjs.safeMin.call(void 0, gaps);
+  const minGap = _chunkQNJLLF4Jcjs.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 = _chunkSTM3ORY7cjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkSTM3ORY7cjs.safeMin.call(void 0, allItems.map((i) => i.x));
+  const pageWidth = _chunkQNJLLF4Jcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkQNJLLF4Jcjs.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++) {
@@ -5589,6 +5700,76 @@ function headerLineAbove(free, colXs, top) {
   return cols;
 }
 
+// src/pdf/cell-edges.ts
+var CLIP_CELL_EDGES = /* @__PURE__ */ new WeakMap();
+var NO_EDGES = Object.freeze({ t: false, b: false, l: false, r: false });
+var EDGE_NEAR2 = 1.6;
+var EDGE_COVER = 0.6;
+var EDGE_SLACK = 3;
+var DASH_GAP = 3;
+var TIP_MAX = 1.5;
+function rules(lines, dir) {
+  const sorted = [...lines].sort((a, b) => dir === "h" ? a.y1 - b.y1 : a.x1 - b.x1);
+  return { dir, pos: sorted.map((l) => dir === "h" ? l.y1 : l.x1), lines: sorted };
+}
+function seen(r, at, a1, a2) {
+  const len = a2 - a1;
+  if (len <= 0) return false;
+  const covered = coverage(r, at, a1, a2);
+  return covered >= len * EDGE_COVER && len - covered <= EDGE_SLACK;
+}
+function coverage(r, at, a1, a2) {
+  let i = 0, j = r.pos.length;
+  while (i < j) {
+    const m = i + j >> 1;
+    if (r.pos[m] < at - EDGE_NEAR2) i = m + 1;
+    else j = m;
+  }
+  const spans = [];
+  for (; i < r.pos.length && r.pos[i] <= at + EDGE_NEAR2; i++) {
+    const l = r.lines[i];
+    const l1 = r.dir === "h" ? Math.min(l.x1, l.x2) : Math.min(l.y1, l.y2), l2 = r.dir === "h" ? Math.max(l.x1, l.x2) : Math.max(l.y1, l.y2);
+    const lo = Math.max(a1, l1), hi = Math.min(a2, l2);
+    if (hi <= lo || l1 < a1 !== l2 > a2 && hi - lo < TIP_MAX) continue;
+    spans.push([lo, hi]);
+  }
+  if (!spans.length) return 0;
+  spans.sort((p, q) => p[0] - q[0]);
+  let covered = 0, s = spans[0][0], e = spans[0][1];
+  for (const [lo, hi] of spans.slice(1)) {
+    if (lo <= e + DASH_GAP) e = Math.max(e, hi);
+    else {
+      covered += e - s;
+      s = lo;
+      e = hi;
+    }
+  }
+  return covered + e - s;
+}
+function edgesOf(box, h, v) {
+  return {
+    t: seen(h, box.y2, box.x1, box.x2),
+    b: seen(h, box.y1, box.x1, box.x2),
+    l: seen(v, box.x1, box.y1, box.y2),
+    r: seen(v, box.x2, box.y1, box.y2)
+  };
+}
+function recordClipCellEdges(grids, horizontals, verticals, nonRules) {
+  if (!grids.some((g) => g.cells)) return;
+  const h = rules(_optionalChain([nonRules, 'optionalAccess', _257 => _257.size]) ? horizontals.filter((l) => !nonRules.has(l)) : horizontals, "h");
+  const v = rules(_optionalChain([nonRules, 'optionalAccess', _258 => _258.size]) ? verticals.filter((l) => !nonRules.has(l)) : verticals, "v");
+  for (const g of grids) for (const c of _nullishCoalesce(g.cells, () => ( []))) CLIP_CELL_EDGES.set(c, edgesOf(c.bbox, h, v));
+}
+function takeClipCellEdges(from, to) {
+  const e = CLIP_CELL_EDGES.get(from);
+  if (e) _chunkAKB72TEHcjs.CELL_EDGES.set(to, e);
+}
+function joinCellEdges(a, b) {
+  const u = _chunkAKB72TEHcjs.CELL_EDGES.get(a), d = _chunkAKB72TEHcjs.CELL_EDGES.get(b);
+  if (!u || !d) return;
+  _chunkAKB72TEHcjs.CELL_EDGES.set(a, { t: u.t, b: d.b, l: u.l || d.l, r: u.r || d.r });
+}
+
 // src/pdf/page-frame.ts
 function isPageFrameGrid(grid, cells, pageWidth, pageHeight, items = []) {
   const b = grid.bbox;
@@ -5843,8 +6024,8 @@ function detectRuledBandTables(horizontals, verticals, items, pageNum) {
   };
   for (const g of groups) {
     if (g.length < 2) continue;
-    const rules = [...g].sort((a, b) => b.y1 - a.y1);
-    const x1 = Math.min(...rules.map((r) => r.x1)), x2 = Math.max(...rules.map((r) => r.x2));
+    const rules2 = [...g].sort((a, b) => b.y1 - a.y1);
+    const x1 = Math.min(...rules2.map((r) => r.x1)), x2 = Math.max(...rules2.map((r) => r.x2));
     let run = [];
     const flush = () => {
       if (run.length >= 2) emit(run, x1, x2);
@@ -5858,8 +6039,8 @@ function detectRuledBandTables(horizontals, verticals, items, pageNum) {
       run = [];
     };
     let title = null;
-    for (let i = 0; i + 1 < rules.length; i++) {
-      const top = rules[i].y1, bottom = rules[i + 1].y1;
+    for (let i = 0; i + 1 < rules2.length; i++) {
+      const top = rules2[i].y1, bottom = rules2[i + 1].y1;
       const band = lines.filter((l) => l[0].y < top && l[0].y > bottom);
       const inside = !band.some((l) => l.some((it) => it.x < x1 - EDGE_TOL3 || it.x + it.w > x2 + EDGE_TOL3));
       const fs = median(band.flat().map((it) => it.fontSize).filter((s) => s > 0)) || 10;
@@ -6117,7 +6298,7 @@ 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', _257 => _257[i + 1], 'optionalAccess', _258 => _258.start]), () => ( lines.length))))) return null;
+  if (runs.some((run, i) => !caption(run.end, _nullishCoalesce(_optionalChain([runs, 'access', _259 => _259[i + 1], 'optionalAccess', _260 => _260.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);
@@ -6277,16 +6458,18 @@ 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 rawRules = extracted.horizontals.concat(extracted.shortH);
   const filled = fillBlanks(items, horizontals, verticals);
   items = filled.items;
   horizontals = filled.horizontals;
-  const prevPage = _optionalChain([carry, 'optionalAccess', _259 => _259.page]) === pageNum - 1 ? carry.clip : void 0;
+  const prevPage = _optionalChain([carry, 'optionalAccess', _261 => _261.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;
     carry.clip = clipResult.page;
   }
   const clipGrids = clipResult.grids;
+  recordClipCellEdges(clipGrids, horizontals.concat(extracted.shortH), verticals.concat(extracted.shortV), extracted.nonRules);
   if (clipGrids.length === 0) {
     horizontals = chainShortSegments(horizontals, extracted.shortH, "h");
     verticals = chainShortSegments(verticals, extracted.shortV, "v");
@@ -6318,7 +6501,7 @@ function extractPageBlocksWithLines(items, pageNum, opList, pageWidth, pageHeigh
   if (detectTables && clipGrids.length === 0 && ruled.length === 0 && lineGrids.length === 0) ruled.push(...detectTextBoxTables(extracted.hiddenBoxes, items, pageNum));
   if (ruled.length > 0) {
     const imageRegions = extractImageRegions(opList.fnArray, opList.argsArray).filter((r) => r.x2 - r.x1 >= 8 && r.y2 - r.y1 >= 8);
-    return extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, horizontals, verticals, imageRegions, lex, ruled);
+    return extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, horizontals, verticals, imageRegions, lex, ruled, rawRules);
   }
   if (detectTables && stackedTableBands(items)) {
     return extractPageBlocksFallback(items, pageNum, true, detectTables, lex);
@@ -6329,7 +6512,7 @@ function extractPageBlocksWithLines(items, pageNum, opList, pageWidth, pageHeigh
   }
   if (grids.length > 0) {
     const imageRegions = extractImageRegions(opList.fnArray, opList.argsArray).filter((r) => r.x2 - r.x1 >= 8 && r.y2 - r.y1 >= 8);
-    return extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, horizontals, verticals, imageRegions, lex);
+    return extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, horizontals, verticals, imageRegions, lex, [], rawRules);
   }
   const figures = extractImageRegions(opList.fnArray, opList.argsArray).filter((r) => r.x2 - r.x1 >= 40 && r.y2 - r.y1 >= 40).map((r) => ({ x: r.x1, y: r.y1, w: r.x2 - r.x1, h: r.y2 - r.y1 }));
   return extractPageBlocksFallback(items, pageNum, true, detectTables, lex, figures);
@@ -6472,7 +6655,7 @@ function buildFrameCellBlocks(cellItems, nested, pageNum, wrap) {
   const text = blocks.map((b) => b.type === "table" && b.table ? b.table.cells.flat().map((c) => c.text).filter(Boolean).join("\n") : _nullishCoalesce(b.text, () => ( ""))).filter(Boolean).join("\n");
   return { blocks, text };
 }
-function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, horizontals, verticals, imageRegions = [], lex, ruled = []) {
+function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, horizontals, verticals, imageRegions = [], lex, ruled = [], rawRules = horizontals) {
   const ocrPage = items.length > 0 && items.every((i) => i.fontName === "ocr");
   const blocks = [];
   const usedItems = /* @__PURE__ */ new Set();
@@ -6534,6 +6717,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
         const built = buildFrameCellBlocks(cellItems, nested, pageNum, { box: cell.bbox, lex });
         irGrid[cell.row][cell.col] = { text: built.text, colSpan: cell.colSpan, rowSpan: cell.rowSpan, blocks: built.blocks };
         if (cellItems.length) recordCellLines(irGrid[cell.row][cell.col], cellItems);
+        takeClipCellEdges(cell, irGrid[cell.row][cell.col]);
         continue;
       }
       irGrid[cell.row][cell.col] = {
@@ -6551,6 +6735,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       }
       if (cell.filler && !cellItems.length) FILLER_CELLS.add(irGrid[cell.row][cell.col]);
       if (grid.cells && cellItems.length) recordCellLines(irGrid[cell.row][cell.col], cellItems);
+      takeClipCellEdges(cell, irGrid[cell.row][cell.col]);
     }
     let finalGrid = irGrid;
     let finalRows = numRows;
@@ -6581,26 +6766,27 @@ 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', _260 => _260[0], 'optionalAccess', _261 => _261[0], 'optionalAccess', _262 => _262.text]), () => ( "")))) {
+    if (unitLine.length > 0 && rebuiltUsed && !/^\s*\(\s*단위\s*[:：]/.test(_nullishCoalesce(_optionalChain([finalGrid, 'access', _262 => _262[0], 'optionalAccess', _263 => _263[0], 'optionalAccess', _264 => _264.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', _263 => _263[0], 'optionalAccess', _264 => _264[0], 'optionalAccess', _265 => _265.text]), () => ( "")))) {
+    if (!rebuiltUsed && filledCells * 2 >= finalGrid.flat().length && !/^\s*\(\s*단위\s*[:：]/.test(_nullishCoalesce(_optionalChain([finalGrid, 'access', _265 => _265[0], 'optionalAccess', _266 => _266[0], 'optionalAccess', _267 => _267.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("");
       const inset = line.length ? grid.bbox.x2 - (line[line.length - 1].x + line[line.length - 1].w) : 0;
       if (line.length && /^\s*\(\s*단위\s*[:：]/.test(text) && inset >= 4.6 && inset <= 5.6) {
         finalGrid.unshift(Array.from({ length: numCols }, (_, c) => ({ text: c === 0 ? cleanCellText(text) : "", colSpan: c === 0 ? numCols : 1, rowSpan: 1 })));
+        if (grid.cells) for (const u of finalGrid[0]) _chunkAKB72TEHcjs.CELL_EDGES.set(u, NO_EDGES);
         finalRows++;
         for (const item of line) usedItems.add(item);
       }
     }
     let semanticOneColumn = false;
     if (!grid.cells && numCols === 1 && numGridRows >= 2) {
-      const populatedRows = finalGrid.filter((row) => _optionalChain([row, 'access', _266 => _266[0], 'optionalAccess', _267 => _267.text, 'access', _268 => _268.trim, 'call', _269 => _269()]));
+      const populatedRows = finalGrid.filter((row) => _optionalChain([row, 'access', _268 => _268[0], 'optionalAccess', _269 => _269.text, 'access', _270 => _270.trim, 'call', _271 => _271()]));
       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));
@@ -6647,8 +6833,12 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       hasHeader: finalRows > 1,
       ...semanticOneColumn ? { renderAsTable: true } : {}
     };
-    if (grid.cells) CLIP_TABLES.add(irTable);
+    if (grid.cells) {
+      CLIP_TABLES.add(irTable);
+      recordRowRules(irTable, grid.cells, grid.bbox, rawRules);
+    }
     TABLE_COLXS.set(irTable, grid.colXs);
+    if (grid.cells && finalRows === numRows) TABLE_ROWYS.set(irTable, grid.rowYs);
     if (grid.continues) CONT_PARTS.set(irTable, grid.continues);
     const hasContent = finalGrid.some((row) => row.some((cell) => cell.text.trim() !== ""));
     const emptyPart = !hasContent && !!grid.cells && !grid.clipParent && !nestedAttached;
@@ -6675,7 +6865,7 @@ 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', _270 => _270[0], 'optionalAccess', _271 => _271[0], 'optionalAccess', _272 => _272.text]), () => ( ""));
+      const prose = _nullishCoalesce(_optionalChain([finalGrid, 'access', _272 => _272[0], 'optionalAccess', _273 => _273[0], 'optionalAccess', _274 => _274.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);
@@ -6729,7 +6919,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
     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', _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);
+      return !blocks.some((x) => x.type === "paragraph" && x.bbox && (_nullishCoalesce(_optionalChain([x, 'access', _275 => _275.text, 'optionalAccess', _276 => _276.match, 'call', _277 => _277(/\p{L}/gu), 'optionalAccess', _278 => _278.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();
@@ -6763,7 +6953,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
     if (gutterX !== null) {
       const gx = gutterX;
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkSTM3ORY7cjs.safeMax.call(void 0, allY) - _chunkSTM3ORY7cjs.safeMin.call(void 0, allY);
+      const pageH = _chunkQNJLLF4Jcjs.safeMax.call(void 0, allY) - _chunkQNJLLF4Jcjs.safeMin.call(void 0, allY);
       const gapThreshold = Math.max(15, pageH * 0.03);
       const sides = [
         remaining.filter((i) => i.x + i.w <= gx),
@@ -6783,7 +6973,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       finalTextBlocks = detectListBlocks(textBlocks);
     } else {
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkSTM3ORY7cjs.safeMax.call(void 0, allY) - _chunkSTM3ORY7cjs.safeMin.call(void 0, allY);
+      const pageH = _chunkQNJLLF4Jcjs.safeMax.call(void 0, allY) - _chunkQNJLLF4Jcjs.safeMin.call(void 0, allY);
       const groups = xyCutOrder(remaining, Math.max(15, pageH * 0.03));
       const textBlocks = [];
       for (const group of groups) {
@@ -7097,7 +7287,7 @@ 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 = _chunkSTM3ORY7cjs.safeMax.call(void 0, allY) - _chunkSTM3ORY7cjs.safeMin.call(void 0, allY);
+      const pageHeight = _chunkQNJLLF4Jcjs.safeMax.call(void 0, allY) - _chunkQNJLLF4Jcjs.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) {
@@ -7158,5 +7348,11 @@ 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.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
+
+
+
+
+
+
+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.TABLE_ROWYS = TABLE_ROWYS; exports.TABLE_TAIL = TABLE_TAIL; exports.PART_COLXS = PART_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.ROW_RULES = ROW_RULES; exports.trimTrailingEmptyTableCols = trimTrailingEmptyTableCols; exports.NO_EDGES = NO_EDGES; exports.joinCellEdges = joinCellEdges; exports.extractPageBlocksWithLines = extractPageBlocksWithLines;
+//# sourceMappingURL=chunk-XEPKOC2O.cjs.map