kordoc 4.15.6 → 4.16.0

package/dist/{chunk-FWT24USP.js → chunk-43FKGDDO.js}

@@ -3,76 +3,14 @@ import {
   HEADING_RATIO_H1,
   HEADING_RATIO_H2,
   HEADING_RATIO_H3
-} from "./chunk-3PDCVDSO.js";
+} from "./chunk-UQ6TZDW6.js";
 import {
   safeMax,
   safeMin
-} from "./chunk-JX7URXOL.js";
-
-// src/pdf/image-regions.ts
-import { OPS } from "pdfjs-dist/legacy/build/pdf.mjs";
-function multiplyTransform(m, t) {
-  return [
-    m[0] * t[0] + m[2] * t[1],
-    m[1] * t[0] + m[3] * t[1],
-    m[0] * t[2] + m[2] * t[3],
-    m[1] * t[2] + m[3] * t[3],
-    m[0] * t[4] + m[2] * t[5] + m[4],
-    m[1] * t[4] + m[3] * t[5] + m[5]
-  ];
-}
-function extractImageRegions(fnArray, argsArray) {
-  const regions = [];
-  let ctm = [1, 0, 0, 1, 0, 0];
-  const stack = [];
-  for (let i = 0; i < fnArray.length; i++) {
-    const op = fnArray[i];
-    switch (op) {
-      case OPS.save:
-        stack.push(ctm);
-        break;
-      case OPS.restore:
-        ctm = stack.pop() || [1, 0, 0, 1, 0, 0];
-        break;
-      case OPS.transform: {
-        const t = argsArray[i];
-        if (Array.isArray(t) && t.length >= 6) ctm = multiplyTransform(ctm, t);
-        break;
-      }
-      // Form XObject 는 /Matrix 공간에서 그린다 — pdfjs 인자 [matrix, bbox], End 에서 복원 (line-extract.ts 와 같음)
-      case OPS.paintFormXObjectBegin: {
-        stack.push(ctm);
-        const m = argsArray[i]?.[0];
-        if (Array.isArray(m) && m.length >= 6) ctm = multiplyTransform(ctm, m);
-        break;
-      }
-      case OPS.paintFormXObjectEnd:
-        ctm = stack.pop() || [1, 0, 0, 1, 0, 0];
-        break;
-      case OPS.paintImageXObject:
-      case OPS.paintInlineImageXObject:
-      case OPS.paintImageMaskXObject:
-      case OPS.paintImageXObjectRepeat: {
-        const corners = [[0, 0], [1, 0], [0, 1], [1, 1]];
-        let x1 = Infinity, y1 = Infinity, x2 = -Infinity, y2 = -Infinity;
-        for (const [u, v] of corners) {
-          const x = ctm[0] * u + ctm[2] * v + ctm[4];
-          const y = ctm[1] * u + ctm[3] * v + ctm[5];
-          if (x < x1) x1 = x;
-          if (x > x2) x2 = x;
-          if (y < y1) y1 = y;
-          if (y > y2) y2 = y;
-        }
-        if (x2 - x1 > 0 && y2 - y1 > 0) regions.push({ x1, y1, x2, y2 });
-        break;
-      }
-    }
-  }
-  return regions;
-}
+} from "./chunk-RC4C7Q33.js";
 
 // src/pdf/line-extract.ts
-import { OPS as OPS2 } from "pdfjs-dist/legacy/build/pdf.mjs";
+import { OPS } from "pdfjs-dist/legacy/build/pdf.mjs";
 var ORIENTATION_TOL = 2;
 var MIN_LINE_LENGTH = 15;
 var MAX_LINE_WIDTH = 5;
@@ -185,20 +123,20 @@ function extractLines(fnArray, argsArray) {
     const op = fnArray[i];
     const args = argsArray[i];
     switch (op) {
-      case OPS2.setLineWidth:
+      case OPS.setLineWidth:
         lineWidth = args[0] || 1;
         break;
-      case OPS2.save:
+      case OPS.save:
         ctmStack.push(ctm.slice());
         colorStack.push(fillColor);
         alphaStack.push([fillAlpha, strokeAlpha]);
         break;
-      case OPS2.restore:
+      case OPS.restore:
         ctm = ctmStack.pop() ?? [1, 0, 0, 1, 0, 0];
         fillColor = colorStack.pop() ?? 0;
         [fillAlpha, strokeAlpha] = alphaStack.pop() ?? [1, 1];
         break;
-      case OPS2.setGState: {
+      case OPS.setGState: {
         const entries = args[0];
         if (!Array.isArray(entries)) break;
         for (const e of entries) {
@@ -208,15 +146,15 @@ function extractLines(fnArray, argsArray) {
         }
         break;
       }
-      case OPS2.setFillRGBColor: {
+      case OPS.setFillRGBColor: {
         const c = args;
         fillColor = c[0] << 16 | c[1] << 8 | c[2];
         break;
       }
-      case OPS2.transform:
-      case OPS2.paintFormXObjectBegin: {
+      case OPS.transform:
+      case OPS.paintFormXObjectBegin: {
         let t = args;
-        if (op === OPS2.paintFormXObjectBegin) {
+        if (op === OPS.paintFormXObjectBegin) {
           ctmStack.push(ctm.slice());
           colorStack.push(fillColor);
           alphaStack.push([fillAlpha, strokeAlpha]);
@@ -234,41 +172,41 @@ function extractLines(fnArray, argsArray) {
         ];
         break;
       }
-      case OPS2.paintFormXObjectEnd:
+      case OPS.paintFormXObjectEnd:
         ctm = ctmStack.pop() ?? [1, 0, 0, 1, 0, 0];
         fillColor = colorStack.pop() ?? 0;
         [fillAlpha, strokeAlpha] = alphaStack.pop() ?? [1, 1];
         break;
-      case OPS2.constructPath: {
+      case OPS.constructPath: {
         const arg0 = args[0];
         if (Array.isArray(arg0)) {
           const subOps = arg0;
           const coords = args[1];
           let ci = 0;
           for (const subOp of subOps) {
-            if (subOp === OPS2.moveTo) {
+            if (subOp === OPS.moveTo) {
               curX = coords[ci++];
               curY = coords[ci++];
               pathStartX = curX;
               pathStartY = curY;
-            } else if (subOp === OPS2.lineTo) {
+            } else if (subOp === OPS.lineTo) {
               const x2 = coords[ci++], y2 = coords[ci++];
               pushSeg(curX, curY, x2, y2);
               curX = x2;
               curY = y2;
-            } else if (subOp === OPS2.rectangle) {
+            } else if (subOp === OPS.rectangle) {
               const rx = coords[ci++], ry = coords[ci++];
               const rw = coords[ci++], rh = coords[ci++];
               pushRectangle(rx, ry, rw, rh);
-            } else if (subOp === OPS2.closePath) {
+            } else if (subOp === OPS.closePath) {
               if (curX !== pathStartX || curY !== pathStartY) {
                 pushSeg(curX, curY, pathStartX, pathStartY);
               }
               curX = pathStartX;
               curY = pathStartY;
-            } else if (subOp === OPS2.curveTo) {
+            } else if (subOp === OPS.curveTo) {
               ci += 6;
-            } else if (subOp === OPS2.curveTo2 || subOp === OPS2.curveTo3) {
+            } else if (subOp === OPS.curveTo2 || subOp === OPS.curveTo3) {
               ci += 4;
             }
           }
@@ -306,37 +244,37 @@ function extractLines(fnArray, argsArray) {
               }
             }
           }
-          if (afterOp === OPS2.stroke || afterOp === OPS2.closeStroke) {
+          if (afterOp === OPS.stroke || afterOp === OPS.closeStroke) {
             flushPath(true);
-          } else if (afterOp === OPS2.fill || afterOp === OPS2.eoFill) {
+          } else if (afterOp === OPS.fill || afterOp === OPS.eoFill) {
             flushPath(true, true);
-          } else if (afterOp === OPS2.fillStroke || afterOp === OPS2.eoFillStroke || afterOp === OPS2.closeFillStroke || afterOp === OPS2.closeEOFillStroke) {
+          } else if (afterOp === OPS.fillStroke || afterOp === OPS.eoFillStroke || afterOp === OPS.closeFillStroke || afterOp === OPS.closeEOFillStroke) {
             flushPath(true, false, true);
-          } else if (afterOp === OPS2.endPath) {
+          } else if (afterOp === OPS.endPath) {
             flushPath(false);
           }
         }
         break;
       }
-      case OPS2.stroke:
-      case OPS2.closeStroke:
+      case OPS.stroke:
+      case OPS.closeStroke:
         flushPath(true);
         break;
-      case OPS2.fill:
-      case OPS2.eoFill:
+      case OPS.fill:
+      case OPS.eoFill:
         flushPath(true, true);
         break;
-      case OPS2.fillStroke:
-      case OPS2.eoFillStroke:
-      case OPS2.closeFillStroke:
-      case OPS2.closeEOFillStroke:
+      case OPS.fillStroke:
+      case OPS.eoFillStroke:
+      case OPS.closeFillStroke:
+      case OPS.closeEOFillStroke:
         flushPath(true, false, true);
         break;
-      case OPS2.endPath:
+      case OPS.endPath:
         flushPath(false);
         break;
-      case OPS2.clip:
-      case OPS2.eoClip:
+      case OPS.clip:
+      case OPS.eoClip:
         pendingClip = true;
         break;
     }
@@ -672,6 +610,68 @@ function filterPageBorderLines(horizontals, verticals, pageWidth, pageHeight) {
   };
 }
 
+// src/pdf/image-regions.ts
+import { OPS as OPS2 } from "pdfjs-dist/legacy/build/pdf.mjs";
+function multiplyTransform(m, t) {
+  return [
+    m[0] * t[0] + m[2] * t[1],
+    m[1] * t[0] + m[3] * t[1],
+    m[0] * t[2] + m[2] * t[3],
+    m[1] * t[2] + m[3] * t[3],
+    m[0] * t[4] + m[2] * t[5] + m[4],
+    m[1] * t[4] + m[3] * t[5] + m[5]
+  ];
+}
+function extractImageRegions(fnArray, argsArray) {
+  const regions = [];
+  let ctm = [1, 0, 0, 1, 0, 0];
+  const stack = [];
+  for (let i = 0; i < fnArray.length; i++) {
+    const op = fnArray[i];
+    switch (op) {
+      case OPS2.save:
+        stack.push(ctm);
+        break;
+      case OPS2.restore:
+        ctm = stack.pop() || [1, 0, 0, 1, 0, 0];
+        break;
+      case OPS2.transform: {
+        const t = argsArray[i];
+        if (Array.isArray(t) && t.length >= 6) ctm = multiplyTransform(ctm, t);
+        break;
+      }
+      // Form XObject 는 /Matrix 공간에서 그린다 — pdfjs 인자 [matrix, bbox], End 에서 복원 (line-extract.ts 와 같음)
+      case OPS2.paintFormXObjectBegin: {
+        stack.push(ctm);
+        const m = argsArray[i]?.[0];
+        if (Array.isArray(m) && m.length >= 6) ctm = multiplyTransform(ctm, m);
+        break;
+      }
+      case OPS2.paintFormXObjectEnd:
+        ctm = stack.pop() || [1, 0, 0, 1, 0, 0];
+        break;
+      case OPS2.paintImageXObject:
+      case OPS2.paintInlineImageXObject:
+      case OPS2.paintImageMaskXObject:
+      case OPS2.paintImageXObjectRepeat: {
+        const corners = [[0, 0], [1, 0], [0, 1], [1, 1]];
+        let x1 = Infinity, y1 = Infinity, x2 = -Infinity, y2 = -Infinity;
+        for (const [u, v] of corners) {
+          const x = ctm[0] * u + ctm[2] * v + ctm[4];
+          const y = ctm[1] * u + ctm[3] * v + ctm[5];
+          if (x < x1) x1 = x;
+          if (x > x2) x2 = x;
+          if (y < y1) y1 = y;
+          if (y > y2) y2 = y;
+        }
+        if (x2 - x1 > 0 && y2 - y1 > 0) regions.push({ x1, y1, x2, y2 });
+        break;
+      }
+    }
+  }
+  return regions;
+}
+
 // src/pdf/vertical-bridge.ts
 var BRIDGE_X_TOL = 1.5;
 var BRIDGE_MIN_GAP = 5;
@@ -791,6 +791,10 @@ var CUT_INTERIOR_MATCH_TOL = 8;
 var CUT_MAX_INTERIOR_OVERLAP = 0.5;
 var CUT_VCHAIN_X_TOL = 1.5;
 var CUT_VCHAIN_GAP = 1;
+var NEST_STRICT_TOL = 0.5;
+var NEST_MIN_GAP = 0.8;
+var NEST_EDGE_COVER = 0.95;
+var NEST_DOUBLE_RULE_GAP = 6;
 var VERTEX_BUCKET_CELL = 100;
 function lowerBound(sorted, key) {
   let lo = 0, hi = sorted.length;
@@ -894,9 +898,12 @@ function buildTableGrids(horizontals, verticals) {
     ...verticals.map((l, i) => ({ ...l, type: "v", id: i + horizontals.length }))
   ];
   const groups = [];
-  for (const g of groupConnectedLines(allLines)) {
+  for (const g0 of groupConnectedLines(allLines)) {
+    const { outer: g, nested } = splitNestedBoxes(g0);
+    for (const n of nested) groups.push({ lines: n, fromSplit: true, nested: true });
+    if (g.length === 0) continue;
     const bands = splitStackedGroup(g);
-    for (const b of bands) groups.push({ lines: b, fromSplit: bands.length > 1 });
+    for (const b of bands) groups.push({ lines: b, fromSplit: bands.length > 1 || nested.length > 0 });
   }
   const grids = [];
   const byY = [...vertices].sort((a, b) => a.y - b.y);
@@ -910,7 +917,7 @@ function buildTableGrids(horizontals, verticals) {
     }
     return lo;
   };
-  for (const { lines: group, fromSplit } of groups) {
+  for (const { lines: group, fromSplit, nested } of groups) {
     const hLines = group.filter((l) => l.type === "h");
     const vLines = group.filter((l) => l.type === "v");
     if (hLines.length < 2 || vLines.length < 2) continue;
@@ -972,9 +979,9 @@ function buildTableGrids(horizontals, verticals) {
       x2: validColXs[validColXs.length - 1],
       y2: validRowYs[0]
     };
-    grids.push({ rowYs: validRowYs, colXs: validColXs, bbox, vertexRadius: groupRadius });
+    grids.push({ rowYs: validRowYs, colXs: validColXs, bbox, vertexRadius: groupRadius, ...nested ? { lineNested: true } : {} });
   }
-  return mergeAdjacentGrids(grids);
+  return [...mergeAdjacentGrids(grids.filter((g) => !g.lineNested)), ...grids.filter((g) => g.lineNested)];
 }
 var SHADE_CLIP_TOL = 1.5;
 var SHADE_CLIP_MAX_AREA = 0.5;
@@ -1153,8 +1160,52 @@ function splitStackedGroup(group) {
   }
   return bands.filter((b) => b.length > 0);
 }
+function splitNestedBoxes(group) {
+  const comps = groupConnectedLines(group, NEST_STRICT_TOL);
+  if (comps.length < 2) return { outer: group, nested: [] };
+  const boxes = comps.map((c) => {
+    const hs = c.filter((l) => l.type === "h"), vs = c.filter((l) => l.type === "v");
+    if (hs.length < 2 || vs.length < 2) return null;
+    const x1 = Math.min(...vs.map((v) => v.x1)), x2 = Math.max(...vs.map((v) => v.x1));
+    const y1 = Math.min(...hs.map((h) => h.y1)), y2 = Math.max(...hs.map((h) => h.y1));
+    if (x2 - x1 < 1 || y2 - y1 < 1) return null;
+    const cover = (segs, a, b) => segs.reduce((s, [p, q]) => s + Math.max(0, Math.min(q, b) - Math.max(p, a)), 0) >= (b - a) * NEST_EDGE_COVER;
+    const edgeH = (y) => cover(hs.filter((h) => Math.abs(h.y1 - y) <= NEST_STRICT_TOL).map((h) => [h.x1, h.x2]), x1, x2);
+    const edgeV = (x) => cover(vs.filter((v) => Math.abs(v.x1 - x) <= NEST_STRICT_TOL).map((v) => [v.y1, v.y2]), y1, y2);
+    return edgeH(y1) && edgeH(y2) && edgeV(x1) && edgeV(x2) ? { x1, y1, x2, y2 } : null;
+  });
+  const bboxOf = (c) => ({
+    x1: Math.min(...c.map((l) => Math.min(l.x1, l.x2))),
+    x2: Math.max(...c.map((l) => Math.max(l.x1, l.x2))),
+    y1: Math.min(...c.map((l) => Math.min(l.y1, l.y2))),
+    y2: Math.max(...c.map((l) => Math.max(l.y1, l.y2)))
+  });
+  const extents = comps.map(bboxOf);
+  const nestedIdx = /* @__PURE__ */ new Set();
+  comps.forEach((_, i) => {
+    const b = boxes[i];
+    if (!b) return;
+    const e = extents[i];
+    const uniq = (xs) => xs.filter((x, k) => xs.findIndex((y) => Math.abs(y - x) <= NEST_STRICT_TOL) === k).length;
+    const bare = uniq(comps[i].filter((l) => l.type === "h").map((l) => l.y1)) === 2 && uniq(comps[i].filter((l) => l.type === "v").map((l) => l.x1)) === 2;
+    const inside = comps.some((c, j) => {
+      if (j === i || c.filter((l) => l.type === "v").length < 2 || !c.some((l) => l.type === "h")) return false;
+      const p = extents[j];
+      const gaps = [e.x1 - p.x1, p.x2 - e.x2, e.y1 - p.y1, p.y2 - e.y2];
+      return gaps.every((g) => g >= NEST_MIN_GAP) && !(bare && gaps.every((g) => g <= NEST_DOUBLE_RULE_GAP));
+    });
+    if (!inside) return;
+    const crossed = comps.some((c, j) => j !== i && c.some((l) => l.type === "h" ? l.y1 > e.y1 + NEST_MIN_GAP && l.y1 < e.y2 - NEST_MIN_GAP && Math.min(l.x2, e.x2) - Math.max(l.x1, e.x1) > NEST_MIN_GAP : l.x1 > e.x1 + NEST_MIN_GAP && l.x1 < e.x2 - NEST_MIN_GAP && Math.min(l.y2, e.y2) - Math.max(l.y1, e.y1) > NEST_MIN_GAP) && !(extents[j].x1 >= e.x1 && extents[j].x2 <= e.x2 && extents[j].y1 >= e.y1 && extents[j].y2 <= e.y2));
+    if (!crossed) nestedIdx.add(i);
+  });
+  if (nestedIdx.size === 0) return { outer: group, nested: [] };
+  return {
+    outer: comps.filter((_, i) => !nestedIdx.has(i)).flat(),
+    nested: comps.filter((_, i) => nestedIdx.has(i))
+  };
+}
 var GROUP_BUCKET_CELL = 100;
-function groupConnectedLines(lines) {
+function groupConnectedLines(lines, tol = CONNECT_TOL) {
   const parent = lines.map((_, i) => i);
   function find(x) {
     while (parent[x] !== x) {
@@ -1196,7 +1247,7 @@ function groupConnectedLines(lines) {
         const key = i * lines.length + j;
         if (tested.has(key)) continue;
         tested.add(key);
-        if (linesIntersect(lines[i], lines[j])) {
+        if (linesIntersect(lines[i], lines[j], tol)) {
           union(i, j);
         }
       }
@@ -1210,19 +1261,18 @@ function groupConnectedLines(lines) {
   }
   return [...groups.values()];
 }
-function linesIntersect(a, b) {
+function linesIntersect(a, b, tol = CONNECT_TOL) {
   if (a.type === b.type) {
     if (a.type === "h") {
-      if (Math.abs(a.y1 - b.y1) > CONNECT_TOL) return false;
-      return Math.min(a.x2, b.x2) >= Math.max(a.x1, b.x1) - CONNECT_TOL;
+      if (Math.abs(a.y1 - b.y1) > tol) return false;
+      return Math.min(a.x2, b.x2) >= Math.max(a.x1, b.x1) - tol;
     } else {
-      if (Math.abs(a.x1 - b.x1) > CONNECT_TOL) return false;
-      return Math.min(a.y2, b.y2) >= Math.max(a.y1, b.y1) - CONNECT_TOL;
+      if (Math.abs(a.x1 - b.x1) > tol) return false;
+      return Math.min(a.y2, b.y2) >= Math.max(a.y1, b.y1) - tol;
     }
   }
   const h = a.type === "h" ? a : b;
   const v = a.type === "h" ? b : a;
-  const tol = CONNECT_TOL;
   return v.x1 >= h.x1 - tol && v.x1 <= h.x2 + tol && h.y1 >= v.y1 - tol && h.y1 <= v.y2 + tol;
 }
 
@@ -1266,6 +1316,13 @@ function clusterCoords(values) {
   if (run.length > 0) out.push(run.reduce((s, x) => s + x, 0) / run.length);
   return out;
 }
+function dropSliverGaps(coords) {
+  const out = [...coords];
+  if (out.length > 2 && out[1] - out[0] < 1) out.splice(0, 2, (out[0] + out[1]) / 2);
+  const n = out.length;
+  if (n > 2 && out[n - 1] - out[n - 2] < 1) out.splice(n - 2, 2, (out[n - 2] + out[n - 1]) / 2);
+  return out;
+}
 var nearestIndex = (coords, v) => {
   let best = 0;
   for (let i = 1; i < coords.length; i++) if (Math.abs(coords[i] - v) < Math.abs(coords[best] - v)) best = i;
@@ -1463,7 +1520,7 @@ function buildClipCellGrids(rects, strokedH, strokedV, pageWidth, pageHeight, te
       if (!edgeStroked(strokedV, "v", r.x2, r.y1, r.y2)) invisible++;
     }
     if (invisible / edges < CLIP_MIN_INVISIBLE) continue;
-    const colXs = clusterCoords(members.flatMap((r) => [r.x1, r.x2]));
+    const colXs = dropSliverGaps(clusterCoords(members.flatMap((r) => [r.x1, r.x2])));
     const rowYs = clusterCoords(members.flatMap((r) => [r.y1, r.y2])).reverse();
     const numRows = rowYs.length - 1, numCols = colXs.length - 1;
     if (numRows < 1 || numCols < 1) continue;
@@ -1860,6 +1917,7 @@ var hasBatchim = (c) => {
   const k = c.charCodeAt(0) - 44032;
   return k >= 0 && k < 11172 && k % 28 !== 0;
 };
+var CJK = /[\u3000-\u303F\u3040-\u30FF\u3400-\u4DBF\u4E00-\u9FFF\uF900-\uFAFF\uFF00-\uFFEF]/;
 var CLOSE_TAIL = /[’”」』)\]〉》>]+$/;
 var TAIL = "[.,)\u300D\u300F\u2019\u201D]*";
 var PARTICLE = new RegExp(`^(?:\uB2E4\\.|\uC758|\uC5D0|\uC5D0\uC11C|\uC5D0\uB294|\uC5D0\uAC8C|\uC5D0\uB3C4|\uC5D0\uC11C\uB294|\uAE4C\uC9C0|\uBD80\uD130|\uCC98\uB7FC|\uB9C8\uB2E4|\uB9CC|\uB3C4|\uB4E4|\uB4E4\uC774|\uB4E4\uC758|\uB4E4\uC740|\uB4E4\uC744|\uB4E4\uC5D0\uAC8C|\uC11C\uB294|\uC11C\uB3C4|\uB77C\uACE0|\uB77C\uBA70|\uC740|\uB294|\uC744|\uB97C|\uC73C\uB85C|\uB85C|\uC73C\uB85C\uC11C|\uB85C\uC11C|\uC73C\uB85C\uC368|\uB85C\uC368)${TAIL}$`);
@@ -1909,6 +1967,7 @@ function wrapJoiner(prevText, nextText, lex) {
   const prev = prevText.replace(MARKUP, "").trimEnd(), next = nextText.replace(MARKUP, "").trimStart();
   const a = prev[prev.length - 1], b = next[0];
   if (!a || !b || /[,;:!?]/.test(a) || DATE_DAY_END.test(prev) && b !== "(") return " ";
+  if (CJK.test(a) && CJK.test(b) && !/[가-힣ㄱ-ㅎㅏ-ㅣ]/.test(prev + next)) return "";
   let s = prev.length;
   while (s > 0 && !/\s/.test(prev[s - 1])) s--;
   const left = prev.slice(s), right = next.match(/^\S+/)[0];
@@ -1950,6 +2009,7 @@ function joinPageBreakWraps(blocks, lex) {
     blocks.splice(i, 1);
   }
 }
+var HANGING = /[，。、：；！？）」』】〕》〉．]\s*$/;
 var BODY_FULL_TOL = 0.25;
 var BODY_MIN_WIDTH_EM = 12;
 var BODY_MAX_PITCH_EM = 2;
@@ -1958,12 +2018,15 @@ var BODY_MAX_PITCH_ABS_EM = 3.5;
 function bodyLineJoins(lines, lex) {
   let right = -Infinity;
   for (const l of lines) if (l.right > right) right = l.right;
+  let inner = -Infinity;
+  for (const l of lines) if (!HANGING.test(l.text) && l.right > inner) inner = l.right;
+  const full = (l) => right - l.right < BODY_FULL_TOL * l.fontSize || right - inner <= 1.2 * l.fontSize && inner - l.right < BODY_FULL_TOL * l.fontSize;
   const pitch = (k) => lines[k].y - lines[k + 1].y;
   const sameSize = (k) => Math.abs(lines[k + 1].fontSize - lines[k].fontSize) <= 0.15 * lines[k].fontSize;
   let p1 = Infinity, p2 = Infinity, i1 = -1;
   for (let k = 0; k + 1 < lines.length; k++) {
     const p = pitch(k);
-    if (p <= 0 || !sameSize(k)) continue;
+    if (p <= 0 || !sameSize(k) || Math.min(lines[k].right - lines[k].left, lines[k + 1].right - lines[k + 1].left) < 3 * lines[k].fontSize) continue;
     if (p < p1) {
       p2 = p1;
       p1 = p;
@@ -1976,7 +2039,7 @@ function bodyLineJoins(lines, lex) {
     const fs = a.fontSize;
     const others = i === i1 ? p2 : p1;
     const maxPitch = Number.isFinite(others) ? Math.min(BODY_MAX_PITCH_ABS_EM * fs, Math.max(BODY_MAX_PITCH_EM * fs, others * BODY_PITCH_REL)) : BODY_MAX_PITCH_EM * fs;
-    const wraps = fs > 0 && right - a.right < BODY_FULL_TOL * fs && a.right - a.left >= BODY_MIN_WIDTH_EM * fs && a.y - b.y > 0 && a.y - b.y < maxPitch && Math.abs(b.fontSize - fs) <= 0.15 * fs && !startsNewItem(a.text, b.text);
+    const wraps = fs > 0 && full(a) && a.right - a.left >= BODY_MIN_WIDTH_EM * fs && a.y - b.y > 0 && a.y - b.y < maxPitch && Math.abs(b.fontSize - fs) <= 0.15 * fs && !startsNewItem(a.text, b.text);
     out.push(wraps ? wrapJoiner(a.text, b.text, lex) : "\n");
   }
   return out;
@@ -2086,6 +2149,7 @@ 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) && !!wrap?.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);
     let result = s[0].text;
     for (let j = 1; j < s.length; j++) {
@@ -2095,7 +2159,9 @@ function cellTextToString(items, wrap) {
       }
       const gap = s[j].x - (s[j - 1].x + s[j - 1].w);
       const avgFs = (s[j].fontSize + s[j - 1].fontSize) / 2;
-      if (s[j].hasSpaceBefore && gap >= avgFs * 0.05) {
+      if (isCjkLatinAutospace(s[j - 1].text, s[j].text, gap, avgFs)) {
+        result += s[j].text;
+      } else if (s[j].hasSpaceBefore && gap >= avgFs * 0.05) {
         result += " " + s[j].text;
       } else if (gap > spaceGapThreshold(avgFs)) {
         result += " " + s[j].text;
@@ -2225,7 +2291,7 @@ function mergeCellTextLines(textLines, wrap) {
       merged[merged.length - 1] = prev + curr.trim();
     } else if (!wrap && /[,(]$/.test(prev.trim()) && curr.trim().length <= 15) {
       merged[merged.length - 1] = prev + curr.trim();
-    } else if (/[\d,]$/.test(prev) && /^[\d,]+[)\]]?$/.test(curr.trim()) && curr.trim().length <= 10 && !(/\d,\d{3}$/.test(prev) && /^\d/.test(curr.trim())) && !(/(?:^|[^\d,])\d{1,3}$/.test(prev) && /^(\d{1,3}(,\d{3})+|\d{1,3})$/.test(curr.trim()))) {
+    } else if (/[\d,]$/.test(prev) && /^[\d,]+[)\]]?$/.test(curr.trim()) && curr.trim().length <= 10 && !(/\d,\d{3}$/.test(prev) && /^\d/.test(curr.trim())) && !(/(?:^|[^\d,])\d{1,3}$/.test(prev) && /^(\d{1,3}(,\d{3})+|\d{1,3})$/.test(curr.trim()) && !wrap?.wraps[i - 1]) && !(/(?:^|\s)\d{1,7}$/.test(prev) && /^\d{1,7}$/.test(curr.trim()) && !wrap?.wraps[i - 1])) {
       merged[merged.length - 1] = prev + curr.trim();
     } else {
       merged.push(curr);
@@ -2388,6 +2454,12 @@ function sortLineByX(items) {
   }
   return items;
 }
+function isCjkLatinAutospace(prevText, nextText, gap, fontSize) {
+  if (!(gap < fontSize * 0.3)) return false;
+  const a = prevText.slice(-1), b = nextText[0] ?? "";
+  const cjk = /[\u3040-\u30ff\u3400-\u9fff\uf900-\ufaff]/, latin = /[A-Za-z0-9]/;
+  return cjk.test(a) && latin.test(b) || latin.test(a) && cjk.test(b);
+}
 function filterHiddenText(items, pageWidth, pageHeight, originX = 0, originY = 0) {
   let hiddenCount = 0;
   const visible = [];
@@ -2463,10 +2535,14 @@ function normalizeItems(rawItems) {
     const scaleX = Math.hypot(i.transform[0], i.transform[1]);
     const scaleY = Math.hypot(i.transform[2], i.transform[3]);
     const fontSize = Math.round(Math.max(scaleY, scaleX));
-    const w = Math.round(i.width);
+    let w = Math.round(i.width);
     const h = Math.round(i.height);
+    if (/[^\u0000]\u0000+$/.test(i.str)) {
+      w = Math.max(1, Math.round(i.width - Math.round(Math.max(Math.hypot(i.transform[2], i.transform[3]), Math.hypot(i.transform[0], i.transform[1]))) * 0.3));
+      spacePositions.push({ x: x + w, y });
+    }
     const isHidden = fontSize === 0 || i.width === 0 && i.str.trim().length > 0;
-    let text = i.str.trim();
+    let text = i.str.trim().replace(/[\u2F00-\u2FD5]/g, (c) => c.normalize("NFKC"));
     if (/^[\d\s\-().·,☎]+$/.test(text) && /\d/.test(text) && / /.test(text)) {
       text = text.replace(/ /g, "");
     }
@@ -2687,6 +2763,10 @@ function mergeLineSimple(items) {
       result += sorted[i].text;
       continue;
     }
+    if (isCjkLatinAutospace(sorted[i - 1].text, sorted[i].text, gap, avgFs)) {
+      result += sorted[i].text;
+      continue;
+    }
     if (sorted[i].hasSpaceBefore && gap >= avgFs * 0.05) {
       result += " ";
       result += sorted[i].text;
@@ -2844,6 +2924,9 @@ function pushLineParagraphs(out, yLines, pageNum, lex) {
     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: dominantStyle(l.items)?.fontSize ?? 0 };
   });
   const joins = bodyLineJoins(geo, lex);
+  for (let i = 0; i + 1 < geo.length; i++) {
+    if (/^\d{1,2}$/.test(geo[i].text.trim()) && geo[i].fontSize >= geo[i + 1].fontSize * 1.2) joins[i] = "\n";
+  }
   if (hasNumberedStyledTitle(lines.map((line) => line.items))) {
     joins[0] = "\n";
     joins[1] = "\n";
@@ -2880,11 +2963,12 @@ function computeMedianFontSizeFromFreq(freq) {
   return sorted[sorted.length - 1][0];
 }
 function detectHeadings(blocks, medianFontSize) {
-  for (const block of blocks) {
+  for (let bi = 0; bi < blocks.length; bi++) {
+    const block = blocks[bi];
     if (block.type !== "paragraph" || !block.text || !block.style?.fontSize) continue;
     const text = block.text.trim();
     if (text.length === 0 || text.length > 200) continue;
-    if (/^\d+$/.test(text)) continue;
+    if (/^\d+$/.test(text) && !isChapterNumber(block, blocks[bi + 1], medianFontSize)) continue;
     const ratio = block.style.fontSize / medianFontSize;
     let level = 0;
     if (ratio >= HEADING_RATIO_H1) level = 1;
@@ -2897,6 +2981,13 @@ function detectHeadings(blocks, medianFontSize) {
     }
   }
 }
+function isChapterNumber(block, next, medianFontSize) {
+  const size = block.style?.fontSize ?? 0;
+  const nextSize = next?.style?.fontSize ?? 0;
+  const text = next?.text?.trim() ?? "";
+  const a = block.bbox, b = next?.bbox;
+  return /^\d{1,2}$/.test(block.text?.trim() ?? "") && size >= medianFontSize * 2.5 && !!next && (next.type === "paragraph" || next.type === "heading") && next.pageNumber === block.pageNumber && nextSize >= medianFontSize * HEADING_RATIO_H1 && text.length > 0 && text.length <= 80 && !/^[\d\s.]+$/.test(text) && !!a && !!b && a.y > b.y && a.y - (b.y + b.height) <= size * 2;
+}
 function mergeStackedHeadingLines(blocks, medianFontSize) {
   for (let i = 0; i < blocks.length - 1; ) {
     const a = blocks[i], b = blocks[i + 1];
@@ -2907,7 +2998,7 @@ function mergeStackedHeadingLines(blocks, medianFontSize) {
     const sameAnchor = ab && bb && Math.abs(ab.x - bb.x) <= 5;
     const wrapped = sameAnchor && a.style?.fontName === b.style?.fontName && af === bf && ab.width >= bb.width * 0.95 && (a.text?.trim().split(/\s+/).length ?? 0) >= 3;
     const displayStack = centered && gap <= Math.min(af, bf) * 0.3 && Math.max(af, bf) <= Math.min(af, bf) * 1.45;
-    if (a.type !== "heading" || b.type !== "heading" || !a.text || !b.text || !ab || !bb || ab.page !== bb.page || !(sameAnchor && af >= medianFontSize * 2 && bf >= medianFontSize * 2) && !wrapped && !(centered && af >= medianFontSize * 1.25 && bf >= medianFontSize * 1.25) || Math.abs(af - bf) > Math.max(af, bf) * 0.15 && !displayStack || gap < -2 || gap > Math.max(af, bf) * 0.45 || a.text.length > (centered ? 120 : 50) || b.text.length > (centered ? 120 : 50)) {
+    if (a.type !== "heading" || b.type !== "heading" || !a.text || !b.text || /^\d+$/.test(a.text.trim()) || !ab || !bb || ab.page !== bb.page || !(sameAnchor && af >= medianFontSize * 2 && bf >= medianFontSize * 2) && !wrapped && !(centered && af >= medianFontSize * 1.25 && bf >= medianFontSize * 1.25) || Math.abs(af - bf) > Math.max(af, bf) * 0.15 && !displayStack || gap < -2 || gap > Math.max(af, bf) * 0.45 || a.text.length > (centered ? 120 : 50) || b.text.length > (centered ? 120 : 50)) {
       i++;
       continue;
     }
@@ -3089,7 +3180,7 @@ function detectPageLeadHeadings(blocks) {
     byPage.set(b.pageNumber ?? 0, page);
   }
   for (const page of byPage.values()) {
-    const content = page.filter((b) => b.type !== "image" && b.type !== "separator");
+    const content = page.filter((b) => b.type !== "image" && b.type !== "separator" && !(b.type === "paragraph" && !b.style));
     const [first, second, third] = content;
     if (!first?.bbox || !first.text || !first.style?.fontSize || !second) continue;
     const firstText = first.text.trim();
@@ -3450,16 +3541,23 @@ function removeHeaderFooterBlocks(blocks, pageHeights, warnings) {
     if (entries.length === 0) continue;
     const patternCount = /* @__PURE__ */ new Map();
     const patternPages = /* @__PURE__ */ new Map();
+    const patternNumbers = /* @__PURE__ */ new Map();
     for (const e of entries) {
       const norm = e.text.replace(/\d+/g, "#");
       patternCount.set(norm, (patternCount.get(norm) || 0) + 1);
+      const nums = patternNumbers.get(norm) || /* @__PURE__ */ new Set();
+      nums.add((e.text.match(/\d+/g) ?? []).join(","));
+      patternNumbers.set(norm, nums);
       const pages = patternPages.get(norm) || /* @__PURE__ */ new Set();
       pages.add(e.page);
       patternPages.set(norm, pages);
     }
     const repeatedPatterns = /* @__PURE__ */ new Set();
     for (const [p, count] of patternCount) {
-      if (count >= MIN_REPEAT && (patternPages.get(p)?.size ?? 0) >= MIN_REPEAT) {
+      const pages = [...patternPages.get(p) ?? []];
+      const span = pages.length ? Math.max(...pages) - Math.min(...pages) + 1 : 0;
+      const pageNumbered = (patternNumbers.get(p)?.size ?? 0) > 1;
+      if (count >= MIN_REPEAT && pages.length >= MIN_REPEAT && (pages.length >= span * 0.4 || pageNumbered)) {
         repeatedPatterns.add(p);
       }
     }
@@ -3664,6 +3762,12 @@ function isFormulaTable(table) {
   const math = texts.filter((text) => /^[a-z]{1,2}$/.test(text) || /[=∂∑∫∣Γ∇√∞]/.test(text)).length;
   return math >= texts.length * 0.6;
 }
+function isExamLayoutTable(table) {
+  const texts = table.cells.flat().map((cell) => cell.text.trim()).filter(Boolean);
+  if (texts.length < 3 || !texts.some((text) => /[①-⑤]|문\d+）/.test(text))) return false;
+  const layout = texts.filter((text) => /\$|[①-⑤]|문\d+）/.test(text) || !/[\p{L}\p{N}]/u.test(text)).length;
+  return layout >= texts.length * 0.6;
+}
 
 // src/pdf/cluster-detector.ts
 var Y_TOL = 3;
@@ -4433,7 +4537,7 @@ function joinCellItems(items) {
     for (let i = 1; i < line.length; i++) {
       const gap = line[i].x - (line[i - 1].x + line[i - 1].w);
       const fs = (line[i].fontSize + line[i - 1].fontSize) / 2;
-      s += (line[i].hasSpaceBefore && gap >= fs * 0.05 || gap > spaceGapThreshold(fs) ? " " : "") + line[i].text;
+      s += (!isCjkLatinAutospace(line[i - 1].text, line[i].text, gap, fs) && (line[i].hasSpaceBefore && gap >= fs * 0.05 || gap > spaceGapThreshold(fs)) ? " " : "") + line[i].text;
     }
     return s;
   }).join(" ");
@@ -4486,9 +4590,9 @@ function xyCutOrder(items, gapThreshold, depth = 0) {
   const hCut = findHorizontalCut(items);
   const vCut = findVerticalCutWithOutlierFilter(items, minGap);
   const hValid = hCut.gap >= minGap;
-  const vValid = vCut.gap >= minGap || vCut.gap >= PROSE_GUTTER_MIN_GAP && isProseGutter(items, vCut.position);
+  const vValid = (vCut.gap >= minGap || vCut.gap >= PROSE_GUTTER_MIN_GAP && isProseGutter(items, vCut.position)) && !splitsSpacedLabel(items, vCut.position);
   let useHorizontal;
-  if (hValid && vValid) useHorizontal = vCut.gap <= hCut.gap * 1.5;
+  if (hValid && vValid) useHorizontal = vCut.gap <= hCut.gap * 1.5 || staggeredSides(items, vCut.position);
   else if (hValid) useHorizontal = true;
   else if (vValid) useHorizontal = false;
   else return splitEdgeSpannedColumns(items, gapThreshold, depth) ?? [items];
@@ -4507,6 +4611,33 @@ function xyCutOrder(items, gapThreshold, depth = 0) {
   }
   return [items];
 }
+function splitsSpacedLabel(items, cutX) {
+  const lone = (i) => /^[가-힣]$/.test(i.text.trim());
+  const sameRow = (a, b) => Math.min(a.y, b.y) > Math.max(a.y - a.h, b.y - b.h);
+  const left = items.filter((i) => i.x + i.w / 2 < cutX);
+  const right = items.filter((i) => i.x + i.w / 2 >= cutX);
+  let spaced = 0;
+  for (const r of right) {
+    if (right.some((o) => o !== r && o.x < r.x && sameRow(o, r))) continue;
+    const row = left.filter((l2) => sameRow(l2, r));
+    if (row.length === 0) continue;
+    const l = row.reduce((a, b) => b.x + b.w > a.x + a.w ? b : a);
+    if (!lone(l) || !lone(r)) return false;
+    spaced++;
+  }
+  return spaced > 0;
+}
+function staggeredSides(items, cutX) {
+  const left = items.filter((i) => i.x + i.w / 2 < cutX);
+  const right = items.filter((i) => i.x + i.w / 2 >= cutX);
+  if (left.length * right.length > 2e5) return false;
+  const shareRow = (a, b) => Math.min(a.y, b.y) > Math.max(a.y - a.h, b.y - b.h);
+  if (left.some((a) => right.some((b) => shareRow(a, b)))) return false;
+  const sides = [...items].sort((a, b) => b.y - a.y).map((i) => i.x + i.w / 2 < cutX);
+  let switches = 0;
+  for (let k = 1; k < sides.length; k++) if (sides[k] !== sides[k - 1]) switches++;
+  return switches >= 3;
+}
 function identifyCrossLayoutItems(items) {
   const cross = /* @__PURE__ */ new Set();
   if (items.length < 3) return cross;
@@ -4669,6 +4800,27 @@ function splitEdgeSpannedColumns(items, gapThreshold, depth) {
   ];
 }
 
+// src/pdf/blank-fills.ts
+var FILL_EDGE_GAP = 6;
+var FILL_MAX_THICKNESS = 2;
+function fillBlanks(items, horizontals, verticals = []) {
+  const widened = /* @__PURE__ */ new Map();
+  const used = /* @__PURE__ */ new Set();
+  for (const h of horizontals) {
+    if (h.lineWidth > FILL_MAX_THICKNESS || Math.abs(h.y1 - h.y2) > 1) continue;
+    const x1 = Math.min(h.x1, h.x2), x2 = Math.max(h.x1, h.x2);
+    const row = items.filter((i) => i.y >= h.y1 && i.y - h.y1 <= Math.max(3, i.fontSize * 0.5));
+    const left = row.find((i) => i.x + i.w <= x1 + 1 && x1 - (i.x + i.w) <= FILL_EDGE_GAP);
+    const right = row.find((i) => i.x >= x2 - 1 && i.x - x2 <= FILL_EDGE_GAP);
+    if (!left || !right || widened.has(left) || row.some((i) => i.x < x2 && i.x + i.w > x1) || x2 - x1 < 2 * left.fontSize) continue;
+    if (verticals.some((v) => (Math.abs(v.x1 - x1) <= 3 || Math.abs(v.x1 - x2) <= 3) && Math.min(v.y1, v.y2) - 3 <= h.y1 && Math.max(v.y1, v.y2) + 3 >= h.y1)) continue;
+    widened.set(left, { ...left, text: left.text.trimEnd() + " ", w: x2 - left.x });
+    used.add(h);
+  }
+  if (!used.size) return { items, horizontals };
+  return { items: items.map((i) => widened.get(i) ?? i), horizontals: horizontals.filter((h) => !used.has(h)) };
+}
+
 // src/pdf/two-column.ts
 var GUTTER_SCAN_LO = 0.3;
 var GUTTER_SCAN_HI = 0.7;
@@ -5273,6 +5425,23 @@ function headerLineAbove(free, colXs, top) {
   return cols;
 }
 
+// src/pdf/page-frame.ts
+function isPageFrameGrid(grid, cells, pageWidth, pageHeight, items = []) {
+  const b = grid.bbox;
+  if (grid.cells || b.x2 - b.x1 < pageWidth * 0.85 || b.y2 - b.y1 < pageHeight * 0.75) return false;
+  const pageArea = pageWidth * pageHeight;
+  const area = (c) => (c.bbox.x2 - c.bbox.x1) * (c.bbox.y2 - c.bbox.y1);
+  const big = cells.filter((c) => area(c) >= pageArea * 0.15);
+  if (big.reduce((s, c) => s + area(c), 0) < pageArea * 0.5) return false;
+  const inner = grid.colXs.slice(1, -1);
+  const inBig = items.filter((it) => big.some((c) => {
+    const cx = it.x + it.w / 2, cy = it.y + it.h / 2;
+    return cx > c.bbox.x1 && cx < c.bbox.x2 && cy > c.bbox.y1 && cy < c.bbox.y2;
+  }));
+  const crossing = inBig.filter((it) => inner.some((x) => x > it.x + 1 && x < it.x + it.w - 1)).length;
+  return inner.length === 0 || crossing >= 1 && crossing >= inBig.length * 0.01;
+}
+
 // src/pdf/open-table-ends.ts
 var ALIGN_TOL = 3;
 var MIN_RULES = 3;
@@ -5306,6 +5475,19 @@ function commonEnd(ends) {
   }
   return bestN >= MIN_REACHING ? best : null;
 }
+function splitBodies(g, chained) {
+  if (g.length < MIN_RULES * 2) return [g];
+  const x1 = Math.min(...g.map((r) => r.x1)), x2 = Math.max(...g.map((r) => r.x2));
+  const sorted = [...g].sort((a, b) => b.y1 - a.y1);
+  const parts = [[sorted[0]]];
+  for (let i = 1; i < sorted.length; i++) {
+    const upper = sorted[i - 1].y1, lower = sorted[i].y1;
+    const joined = chained.some((v) => v.x1 >= x1 - ALIGN_TOL && v.x1 <= x2 + ALIGN_TOL && v.y2 >= upper - TOUCH_TOL && v.y1 <= lower + TOUCH_TOL);
+    if (joined) parts[parts.length - 1].push(sorted[i]);
+    else parts.push([sorted[i]]);
+  }
+  return parts.filter((p) => p.length >= MIN_RULES).length >= 2 ? parts : [g];
+}
 function closeOpenTableEnds(horizontals, verticals, textLayer = false) {
   if (horizontals.length < MIN_RULES || verticals.length < MIN_REACHING) return horizontals;
   const groups = [];
@@ -5316,7 +5498,7 @@ function closeOpenTableEnds(horizontals, verticals, textLayer = false) {
   }
   const chained = chainVerticals2(verticals);
   const added = [];
-  for (const g of groups) {
+  for (const g of groups.flatMap((gr) => splitBodies(gr, chained))) {
     if (g.length < MIN_RULES) continue;
     const x1 = Math.min(...g.map((r) => r.x1)), x2 = Math.max(...g.map((r) => r.x2));
     const yLo = Math.min(...g.map((r) => r.y1)), yHi = Math.max(...g.map((r) => r.y1));
@@ -5924,6 +6106,9 @@ 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 filled = fillBlanks(items, horizontals, verticals);
+  items = filled.items;
+  horizontals = filled.horizontals;
   const prevPage = carry?.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) {
@@ -5989,6 +6174,7 @@ function markStrikethroughItems(items, horizontals) {
     if (line.lineWidth > STRIKE_MAX_THICKNESS) continue;
     const matches = [];
     for (const item of items) {
+      if (item.fontName === "ocr") continue;
       const h = item.h > 0 ? item.h : item.fontSize;
       if (h <= 0 || item.w <= 0) continue;
       if (line.lineWidth > h * STRIKE_MAX_THICKNESS_RATIO) continue;
@@ -6069,16 +6255,25 @@ function isProseBoxGrid(grid, verticals, table) {
   if (longCells < PROSEBOX_LONG_CELL_MIN || longCells < texts.length * PROSEBOX_LONG_CELL_RATIO) return false;
   return true;
 }
+function isSparseProseGrid(table) {
+  if (table.rows < 4) return false;
+  const texts = table.cells.flat().map((c) => c.text.trim());
+  const filled = texts.filter(Boolean);
+  if (filled.length === 0 || filled.length > texts.length * 0.25) return false;
+  const chars = filled.reduce((n, t) => n + t.length, 0);
+  const prose = filled.filter((t) => t.length >= 60 && /[.!?。]/.test(t)).reduce((n, t) => n + t.length, 0);
+  return prose >= chars * 0.6;
+}
 function cleanCellText(text) {
   const stripped = text.replace(/^[\s]*[-–—]\s*\d+\s*[-–—][\s]*$/gm, "").trim();
   return stripped.split("\n").map((line) => collapseEvenSpacing(line)).join("\n");
 }
 var FRAME_RECT_TOL = 1.5;
-function takePendingNested(pending, cellBox) {
+function takePendingNested(pending, cellBox, clipCell) {
   const out = [];
   for (let i = pending.length - 1; i >= 0; i--) {
     const p = pending[i].parent;
-    if (Math.abs(p.x1 - cellBox.x1) <= FRAME_RECT_TOL && Math.abs(p.x2 - cellBox.x2) <= FRAME_RECT_TOL && Math.abs(p.y1 - cellBox.y1) <= FRAME_RECT_TOL && Math.abs(p.y2 - cellBox.y2) <= FRAME_RECT_TOL) {
+    if (pending[i].contained ? p.x1 >= cellBox.x1 - FRAME_RECT_TOL && p.x2 <= cellBox.x2 + FRAME_RECT_TOL && p.y1 >= cellBox.y1 - FRAME_RECT_TOL && p.y2 <= cellBox.y2 + FRAME_RECT_TOL : clipCell && Math.abs(p.x1 - cellBox.x1) <= FRAME_RECT_TOL && Math.abs(p.x2 - cellBox.x2) <= FRAME_RECT_TOL && Math.abs(p.y1 - cellBox.y1) <= FRAME_RECT_TOL && Math.abs(p.y2 - cellBox.y2) <= FRAME_RECT_TOL) {
       out.push(pending[i].block);
       pending.splice(i, 1);
     }
@@ -6107,6 +6302,7 @@ function buildFrameCellBlocks(cellItems, nested, pageNum, wrap) {
   return { blocks, text };
 }
 function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, horizontals, verticals, imageRegions = [], lex, ruled = []) {
+  const ocrPage = items.length > 0 && items.every((i) => i.fontName === "ocr");
   const blocks = [];
   const usedItems = /* @__PURE__ */ new Set();
   for (const r of ruled) {
@@ -6116,13 +6312,15 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
   const proseSidebars = /* @__PURE__ */ new Set();
   const pendingNested = [];
   const gridArea = (g) => (g.bbox.x2 - g.bbox.x1) * (g.bbox.y2 - g.bbox.y1);
-  const sortedGrids = [...grids].sort((a, b) => (b.cells ? 1 : 0) - (a.cells ? 1 : 0) || (a.cells && b.cells ? gridArea(a) - gridArea(b) : 0) || b.bbox.y2 - a.bbox.y2);
+  const sortedGrids = [...grids].sort((a, b) => (b.cells ? 1 : 0) - (a.cells ? 1 : 0) || (a.cells && b.cells ? gridArea(a) - gridArea(b) : 0) || (b.lineNested ? 1 : 0) - (a.lineNested ? 1 : 0) || (a.lineNested && b.lineNested ? gridArea(a) - gridArea(b) : 0) || b.bbox.y2 - a.bbox.y2);
   for (const grid of sortedGrids) {
     const numGridRows = grid.rowYs.length - 1;
     const numGridCols = grid.colXs.length - 1;
     const gridW = grid.bbox.x2 - grid.bbox.x1;
-    if (!grid.cells && numGridRows === 1 && numGridCols >= 2) continue;
-    if (!grid.cells && numGridCols === 1 && numGridRows >= 2 && (numGridRows < 5 || gridW > pageWidth * 0.7)) continue;
+    const holdsNested = !grid.cells && pendingNested.some((p) => p.contained && p.parent.x1 >= grid.bbox.x1 - FRAME_RECT_TOL && p.parent.x2 <= grid.bbox.x2 + FRAME_RECT_TOL && p.parent.y1 >= grid.bbox.y1 - FRAME_RECT_TOL && p.parent.y2 <= grid.bbox.y2 + FRAME_RECT_TOL);
+    if (!grid.cells && !holdsNested && numGridRows === 1 && numGridCols >= 2) continue;
+    if (!grid.cells && !holdsNested && numGridCols === 1 && numGridRows >= 2 && (numGridRows < 5 || gridW > pageWidth * 0.7)) continue;
+    if (!holdsNested && isPageFrameGrid(grid, extractCells(grid, horizontals, verticals), pageWidth, pageHeight, items)) continue;
     const tableItems = [];
     const pad = 3;
     for (const item of items) {
@@ -6158,7 +6356,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
     let nestedAttached = false;
     for (const cell of cells) {
       const cellItems = cellTextMap.get(cell) || [];
-      const nested = grid.cells ? takePendingNested(pendingNested, cell.bbox) : [];
+      const nested = pendingNested.length ? takePendingNested(pendingNested, cell.bbox, !!grid.cells) : [];
       if (nested.length > 0) {
         nestedAttached = true;
         const built = buildFrameCellBlocks(cellItems, nested, pageNum, { box: cell.bbox, lex });
@@ -6216,6 +6414,18 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       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(finalGrid[0]?.[0]?.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 })));
+        finalRows++;
+        for (const item of line) usedItems.add(item);
+      }
+    }
     let semanticOneColumn = false;
     if (!grid.cells && numCols === 1 && numGridRows >= 2) {
       const populatedRows = finalGrid.filter((row) => row[0]?.text.trim());
@@ -6276,11 +6486,11 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       pendingNested.push({ parent: grid.clipParent, block: { type: "table", table: irTable, pageNumber: pageNum, bbox: nb } });
       continue;
     }
-    if (!grid.cells && isProseBoxGrid(grid, verticals, irTable)) {
+    if (!grid.cells && !nestedAttached && (isProseBoxGrid(grid, verticals, irTable) || ocrPage && isSparseProseGrid(irTable))) {
       for (const it of tableItems) usedItems.delete(it);
       continue;
     }
-    if (!grid.cells && isChartTable(irTable)) {
+    if (!grid.cells && !nestedAttached && isChartTable(irTable)) {
       blocks.push(chartBlock(tableItems, pageNum, { page: pageNum, x: grid.bbox.x1, y: grid.bbox.y1, width: gridW, height: grid.bbox.y2 - grid.bbox.y1 }));
       continue;
     }
@@ -6318,6 +6528,10 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       }
       continue;
     }
+    if (grid.lineNested) {
+      pendingNested.push({ parent: grid.bbox, block: { type: "table", table: irTable, pageNumber: pageNum, bbox: tableBbox }, contained: true });
+      continue;
+    }
     blocks.push({ type: "table", table: irTable, pageNumber: pageNum, bbox: tableBbox });
   }
   for (const p of pendingNested) blocks.push(p.block);
@@ -6339,7 +6553,10 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       hasSpaceBefore: i.hasSpaceBefore
     }));
     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);
+    const clusterResults = (proseColumns ? [] : detectClusterTables(clusterItems, pageNum)).filter((cr) => {
+      const b = cr.bbox;
+      return !blocks.some((x) => x.type === "paragraph" && x.bbox && (x.text?.match(/\p{L}/gu)?.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();
       for (let ci = 0; ci < clusterItems.length; ci++) ciToIdx.set(clusterItems[ci], ci);
@@ -6503,6 +6720,7 @@ function clusterTableBlock(cr, source, pageNum, horizontals = []) {
   if (!ruled && isTableOfContents(cr.table)) return tocBlock(cr.table, pageNum, cr.bbox, dominantStyle(source));
   if (isChartTable(cr.table)) return chartBlock(source, pageNum, cr.bbox);
   if (isFormulaTable(cr.table)) return chartBlock(source, pageNum, cr.bbox);
+  if (isExamLayoutTable(cr.table)) return { ...chartBlock(source, pageNum, cr.bbox), text: groupByY(source).map((line) => mergeLineSimple(line).trim()).filter(Boolean).join("\n") };
   return { type: "table", table: cr.table, pageNumber: pageNum, bbox: cr.bbox };
 }
 function chartBlock(source, pageNum, bbox) {
@@ -6725,6 +6943,7 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false, detectTable
 }
 
 export {
+  extractLines,
   extractImageRegions,
   filterHiddenText,
   collapseEvenSpacing,
@@ -6757,4 +6976,4 @@ export {
   trimTrailingEmptyTableCols,
   extractPageBlocksWithLines
 };
-//# sourceMappingURL=chunk-FWT24USP.js.map
+//# sourceMappingURL=chunk-43FKGDDO.js.map