kordoc 4.15.0 → 4.15.2

package/dist/{parser-Y5NX55GC.cjs → parser-XCXDSKQ3.cjs}

@@ -3,12 +3,12 @@
 
 
 
-var _chunk3CWXYYCLcjs = require('./chunk-3CWXYYCL.cjs');
 
+var _chunk3Y26ICMCcjs = require('./chunk-3Y26ICMC.cjs');
 
-var _chunkDCZVOIEOcjs = require('./chunk-DCZVOIEO.cjs');
 
 
+var _chunkJ55UQC5Hcjs = require('./chunk-J55UQC5H.cjs');
 
 
 
@@ -30,17 +30,20 @@ var _chunkDCZVOIEOcjs = require('./chunk-DCZVOIEO.cjs');
 
 
 
-var _chunkW33QEDLYcjs = require('./chunk-W33QEDLY.cjs');
-require('./chunk-VWUYVEFN.cjs');
 
 
+var _chunkVSI5SRP2cjs = require('./chunk-VSI5SRP2.cjs');
+require('./chunk-PY2M7YVN.cjs');
 
-var _chunkLBYKEJUJcjs = require('./chunk-LBYKEJUJ.cjs');
+
+
+var _chunkVGU4FLLDcjs = require('./chunk-VGU4FLLD.cjs');
 
 
 var _chunkGS7T56RPcjs = require('./chunk-GS7T56RP.cjs');
 
 // src/pdf/image-extract.ts
+var _crypto = require('crypto');
 var _pdfmjs = require('pdfjs-dist/legacy/build/pdf.mjs');
 var IMAGE_RESOLVE_TIMEOUT_MS = 5e3;
 function resolveImgData(page, objId) {
@@ -75,27 +78,26 @@ var MAX_PIXELS = 36e6;
 var MAX_IMAGES_PER_DOC = 200;
 var MAX_TOTAL_IMAGE_BYTES = 128 * 1024 * 1024;
 function createPdfImageState() {
-  return { imageIndex: 0, totalBytes: 0, seen: /* @__PURE__ */ new Map(), capWarned: false };
+  return { imageIndex: 0, totalBytes: 0, seen: /* @__PURE__ */ new Map(), globalKeys: /* @__PURE__ */ new Map(), capWarned: false };
 }
-function fnv1a(data) {
-  let h = 2166136261;
-  for (let i = 0; i < data.length; i++) {
-    h ^= data[i];
-    h = Math.imul(h, 16777619);
-  }
-  return h >>> 0;
+function contentHash(data) {
+  return _crypto.createHash.call(void 0, "sha1").update(data).digest("hex");
 }
-function toRgba(img) {
+var NO_BYTES = new Uint8Array(0);
+function toRgba(img, convert = true) {
   const { width: w, height: h, kind, data } = img;
   if (!data || !w || !h) return null;
-  const rgba = new Uint8Array(w * h * 4);
   if (kind === _pdfmjs.ImageKind.RGBA_32BPP) {
     if (data.length < w * h * 4) return null;
+    if (!convert) return NO_BYTES;
+    const rgba = new Uint8Array(w * h * 4);
     rgba.set(data.subarray(0, w * h * 4));
     return rgba;
   }
   if (kind === _pdfmjs.ImageKind.RGB_24BPP) {
     if (data.length < w * h * 3) return null;
+    if (!convert) return NO_BYTES;
+    const rgba = new Uint8Array(w * h * 4);
     for (let i = 0, s = 0, d = 0; i < w * h; i++, s += 3, d += 4) {
       rgba[d] = data[s];
       rgba[d + 1] = data[s + 1];
@@ -107,6 +109,8 @@ function toRgba(img) {
   if (kind === _pdfmjs.ImageKind.GRAYSCALE_1BPP) {
     const stride = w + 7 >> 3;
     if (data.length < stride * h) return null;
+    if (!convert) return NO_BYTES;
+    const rgba = new Uint8Array(w * h * 4);
     for (let y = 0; y < h; y++) {
       for (let x = 0; x < w; x++) {
         const bit = data[y * stride + (x >> 3)] >> 7 - (x & 7) & 1;
@@ -157,15 +161,20 @@ async function extractPageImages(page, fnArray, argsArray, pageNumber, state, wa
       warnings.push({ page: pageNumber, message: `\uC774\uBBF8\uC9C0 \uD06C\uAE30 \uCD08\uACFC\uB85C \uCD94\uCD9C \uC0DD\uB7B5 (${w}\xD7${h})`, code: "SKIPPED_IMAGE" });
       continue;
     }
-    const hash = `${w}x${h}k${_nullishCoalesce(imgData.kind, () => ( "?"))}h${fnv1a(imgData.data)}${dedupeId ? "" : "inline"}`;
+    const globalId = _optionalChain([dedupeId, 'optionalAccess', _7 => _7.startsWith, 'call', _8 => _8("g_")]) ? dedupeId : void 0;
+    let hash = globalId ? state.globalKeys.get(globalId) : void 0;
+    if (hash === void 0) {
+      hash = `${w}x${h}k${_nullishCoalesce(imgData.kind, () => ( "?"))}h${contentHash(imgData.data)}${dedupeId ? "" : "inline"}`;
+      if (globalId) state.globalKeys.set(globalId, hash);
+    }
     if (state.seen.has(hash)) continue;
-    const rgba = toRgba(imgData);
+    const rgba = toRgba(imgData, withBytes);
     if (!rgba) continue;
     state.imageIndex++;
     const filename = `image_${String(state.imageIndex).padStart(3, "0")}.png`;
     state.seen.set(hash, filename);
     if (withBytes) {
-      const png = _chunk3CWXYYCLcjs.encodePng.call(void 0, w, h, rgba);
+      const png = _chunk3Y26ICMCcjs.encodePng.call(void 0, w, h, rgba);
       state.totalBytes += png.length;
       images.push({ filename, data: png, mimeType: "image/png" });
     }
@@ -205,7 +214,7 @@ function injectPageImageBlocks(blocks, pageImages) {
     result.splice(at, 0, ...pageImages.get(p));
   }
   blocks.length = 0;
-  blocks.push(...result);
+  for (const b of result) blocks.push(b);
 }
 
 // src/pdf/quality.ts
@@ -567,7 +576,7 @@ function anchorsOf(table) {
   for (let r = 0; r < table.rows; r++) {
     for (let c = 0; c < table.cols; c++) {
       if (covered.has(r * 1e5 + c)) continue;
-      const cell = _optionalChain([table, 'access', _7 => _7.cells, 'access', _8 => _8[r], 'optionalAccess', _9 => _9[c]]);
+      const cell = _optionalChain([table, 'access', _9 => _9.cells, 'access', _10 => _10[r], 'optionalAccess', _11 => _11[c]]);
       if (!cell) continue;
       out.push({ r, c, rs: cell.rowSpan, cs: cell.colSpan, cell });
       for (let dr = 0; dr < cell.rowSpan; dr++) {
@@ -618,19 +627,19 @@ function joinSplitParts(prev, pcx, curr, ccx, dx = 0, prevBottom) {
       if (r + dr < rows) owner[r + dr][dc] = placed;
     }
     grid[r][c1] = { ...a.cell, colSpan: c2 - c1, rowSpan: a.rs };
-    const lines = _chunkW33QEDLYcjs.CELL_LINES.get(a.cell);
-    if (lines) _chunkW33QEDLYcjs.CELL_LINES.set(grid[r][c1], shift ? lines.map((l) => ({ ...l, l: l.l + shift, r: l.r + shift })) : lines);
-    if (_chunkW33QEDLYcjs.IMAGE_CELLS.has(a.cell)) _chunkW33QEDLYcjs.IMAGE_CELLS.add(grid[r][c1]);
+    const lines = _chunkVSI5SRP2cjs.CELL_LINES.get(a.cell);
+    if (lines) _chunkVSI5SRP2cjs.CELL_LINES.set(grid[r][c1], shift ? lines.map((l) => ({ ...l, l: l.l + shift, r: l.r + shift })) : lines);
+    if (_chunkVSI5SRP2cjs.IMAGE_CELLS.has(a.cell)) _chunkVSI5SRP2cjs.IMAGE_CELLS.add(grid[r][c1]);
     return true;
   };
-  for (const a of pa) if (!_chunkW33QEDLYcjs.FILLER_CELLS.has(a.cell) && !place(a, pcx, 0)) return null;
+  for (const a of pa) if (!_chunkVSI5SRP2cjs.FILLER_CELLS.has(a.cell) && !place(a, pcx, 0)) return null;
   for (const a of ca) {
-    if (a.r < skip || _chunkW33QEDLYcjs.FILLER_CELLS.has(a.cell)) continue;
+    if (a.r < skip || _chunkVSI5SRP2cjs.FILLER_CELLS.has(a.cell)) continue;
     if (!place(a, ccx, prev.rows - skip, dx)) return null;
   }
   const first = prev.rows;
   for (let c = 0; c < cols; c++) {
-    const above = _optionalChain([owner, 'access', _10 => _10[first - 1], 'optionalAccess', _11 => _11[c]]);
+    const above = _optionalChain([owner, 'access', _12 => _12[first - 1], 'optionalAccess', _13 => _13[c]]);
     if (!above) continue;
     let r = first;
     while (r < rows && owner[r][c] === null) r++;
@@ -649,7 +658,7 @@ function joinSplitParts(prev, pcx, curr, ccx, dx = 0, prevBottom) {
     const grown = { ...above, rs: newRs };
     for (let rr = above.r; rr < above.r + newRs; rr++) for (let dc = above.c; dc < above.c + above.cs; dc++) owner[rr][dc] = grown;
   }
-  for (let r = 0; r < rows; r++) for (let c = 0; c < cols; c++) if (owner[r][c] === null) _chunkW33QEDLYcjs.FILLER_CELLS.add(grid[r][c]);
+  for (let r = 0; r < rows; r++) for (let c = 0; c < cols; c++) if (owner[r][c] === null) _chunkVSI5SRP2cjs.FILLER_CELLS.add(grid[r][c]);
   const table = { rows, cols, cells: grid, hasHeader: prev.hasHeader, ...prev.caption ? { caption: prev.caption } : {} };
   const split = rows > prev.rows && mergeSplitRow(table, owner, prev.rows, U);
   if (rows > prev.rows && !split && prevBottom !== void 0) mergeStraddlingCells(table, owner, prev.rows, prevBottom);
@@ -667,8 +676,8 @@ function mergeStraddlingCells(table, owner, first, prevBottom) {
     c = u.c + u.cs;
     if (!d || d === u || u.r >= last || u.r + u.rs - 1 !== last || d.r !== first || d.c !== u.c || d.cs !== u.cs) continue;
     const a = table.cells[u.r][u.c], b = table.cells[first][d.c];
-    const U = _chunkW33QEDLYcjs.CELL_LINES.get(a);
-    if (!_optionalChain([U, 'optionalAccess', _12 => _12.length]) || U.length > STRADDLE_MAX_LINES || !hasContent(b)) continue;
+    const U = _chunkVSI5SRP2cjs.CELL_LINES.get(a);
+    if (!_optionalChain([U, 'optionalAccess', _14 => _14.length]) || U.length > STRADDLE_MAX_LINES || !hasContent(b)) continue;
     const lu = U[U.length - 1];
     if (lu.y - prevBottom > STRADDLE_BOTTOM * (lu.h || 10)) continue;
     appendCell(a, b);
@@ -690,8 +699,8 @@ var LABEL_ROOM = 1.5;
 var HANGING_MIN = 0.5;
 var HANGING_TOL = 1;
 function continuesAcross(u, d, x1, x2) {
-  const U = _chunkW33QEDLYcjs.CELL_LINES.get(u), D = _chunkW33QEDLYcjs.CELL_LINES.get(d);
-  if (!_optionalChain([U, 'optionalAccess', _13 => _13.length]) || !_optionalChain([D, 'optionalAccess', _14 => _14.length])) return false;
+  const U = _chunkVSI5SRP2cjs.CELL_LINES.get(u), D = _chunkVSI5SRP2cjs.CELL_LINES.get(d);
+  if (!_optionalChain([U, 'optionalAccess', _15 => _15.length]) || !_optionalChain([D, 'optionalAccess', _16 => _16.length])) return false;
   const last = U[U.length - 1];
   const fs = last.h || 10;
   let minL = Infinity, maxR = -Infinity;
@@ -712,19 +721,19 @@ function continuesAcross(u, d, x1, x2) {
   return U.some(leftAligned) || D.some(leftAligned) || hanging;
 }
 function lineEnded(c, x1, x2) {
-  const L = _chunkW33QEDLYcjs.CELL_LINES.get(c);
+  const L = _chunkVSI5SRP2cjs.CELL_LINES.get(c);
   if (!L || L.length !== 1) return false;
   const l = L[0];
   return x2 - l.r - (l.l - x1) >= LABEL_ROOM * (l.h || 10);
 }
 function appendCell(a, b) {
-  if (_optionalChain([a, 'access', _15 => _15.blocks, 'optionalAccess', _16 => _16.length]) || _optionalChain([b, 'access', _17 => _17.blocks, 'optionalAccess', _18 => _18.length])) {
-    const asBlocks = (c) => _optionalChain([c, 'access', _19 => _19.blocks, 'optionalAccess', _20 => _20.length]) ? c.blocks : c.text.split("\n").map((l) => l.trim()).filter(Boolean).map((text) => ({ type: "paragraph", text }));
+  if (_optionalChain([a, 'access', _17 => _17.blocks, 'optionalAccess', _18 => _18.length]) || _optionalChain([b, 'access', _19 => _19.blocks, 'optionalAccess', _20 => _20.length])) {
+    const asBlocks = (c) => _optionalChain([c, 'access', _21 => _21.blocks, 'optionalAccess', _22 => _22.length]) ? c.blocks : c.text.split("\n").map((l) => l.trim()).filter(Boolean).map((text) => ({ type: "paragraph", text }));
     a.blocks = [...asBlocks(a), ...asBlocks(b)];
   }
   if (b.text.trim()) a.text = a.text.trim() ? a.text + "\n" + b.text : b.text;
 }
-var hasContent = (cell) => !!cell.text.trim() || !!_optionalChain([cell, 'access', _21 => _21.blocks, 'optionalAccess', _22 => _22.length]) || _chunkW33QEDLYcjs.IMAGE_CELLS.has(cell);
+var hasContent = (cell) => !!cell.text.trim() || !!_optionalChain([cell, 'access', _23 => _23.blocks, 'optionalAccess', _24 => _24.length]) || _chunkVSI5SRP2cjs.IMAGE_CELLS.has(cell);
 function mergeSplitRow(table, owner, first, colXs) {
   const last = first - 1;
   const pairs = [];
@@ -797,19 +806,19 @@ function mergeCrossPageTables(blocks, pageHeights) {
       placeJoined(blocks, i, j, joined);
       continue;
     }
-    if (_chunkW33QEDLYcjs.EMPTY_PARTS.has(curr.table)) {
+    if (_chunkVSI5SRP2cjs.EMPTY_PARTS.has(curr.table)) {
       blocks.splice(j, 1);
       i++;
       continue;
     }
     if (joined === null) continue;
-    if (prev.table.cols !== curr.table.cols || _chunkW33QEDLYcjs.EMPTY_PARTS.has(prev.table)) continue;
+    if (prev.table.cols !== curr.table.cols || _chunkVSI5SRP2cjs.EMPTY_PARTS.has(prev.table)) continue;
     const width = Math.max(prev.bbox.width, curr.bbox.width, 1);
     const leftDiff = Math.abs(prev.bbox.x - curr.bbox.x);
     const rightDiff = Math.abs(prev.bbox.x + prev.bbox.width - (curr.bbox.x + curr.bbox.width));
     if (leftDiff > width * NEIGHBOR_TABLE_EPSILON || rightDiff > width * NEIGHBOR_TABLE_EPSILON) continue;
-    const px = _chunkW33QEDLYcjs.TABLE_COLXS.get(prev.table), cx = _chunkW33QEDLYcjs.TABLE_COLXS.get(curr.table);
-    if (px && cx && !shiftedSame(px, cx, !_chunkW33QEDLYcjs.CLIP_TABLES.has(prev.table) && !_chunkW33QEDLYcjs.CLIP_TABLES.has(curr.table))) continue;
+    const px = _chunkVSI5SRP2cjs.TABLE_COLXS.get(prev.table), cx = _chunkVSI5SRP2cjs.TABLE_COLXS.get(curr.table);
+    if (px && cx && !shiftedSame(px, cx, !_chunkVSI5SRP2cjs.CLIP_TABLES.has(prev.table) && !_chunkVSI5SRP2cjs.CLIP_TABLES.has(curr.table))) continue;
     let currCells = curr.table.cells;
     if (currCells.length > 1 && prev.table.cells.length > 0 && rowTextsEqual(prev.table.cells[0], currCells[0])) {
       currCells = currCells.slice(1);
@@ -825,20 +834,20 @@ function mergeCrossPageTables(blocks, pageHeights) {
       hasHeader: prev.table.hasHeader,
       caption: prev.table.caption
     };
-    if (_nullishCoalesce(px, () => ( cx))) _chunkW33QEDLYcjs.TABLE_COLXS.set(merged, _nullishCoalesce(px, () => ( cx)));
-    if (_chunkW33QEDLYcjs.CLIP_TABLES.has(prev.table) || _chunkW33QEDLYcjs.CLIP_TABLES.has(curr.table)) _chunkW33QEDLYcjs.CLIP_TABLES.add(merged);
+    if (_nullishCoalesce(px, () => ( cx))) _chunkVSI5SRP2cjs.TABLE_COLXS.set(merged, _nullishCoalesce(px, () => ( cx)));
+    if (_chunkVSI5SRP2cjs.CLIP_TABLES.has(prev.table) || _chunkVSI5SRP2cjs.CLIP_TABLES.has(curr.table)) _chunkVSI5SRP2cjs.CLIP_TABLES.add(merged);
     placeJoined(blocks, i, j, merged);
   }
   for (let i = blocks.length - 1; i >= 0; i--) {
     const t = blocks[i].table;
-    if (blocks[i].type === "table" && t && _chunkW33QEDLYcjs.EMPTY_PARTS.has(t)) blocks.splice(i, 1);
+    if (blocks[i].type === "table" && t && _chunkVSI5SRP2cjs.EMPTY_PARTS.has(t)) blocks.splice(i, 1);
   }
 }
 function mergeColumnFlow(blocks, pageHeights) {
   if (!pageHeights) return;
   for (let i = 0; i < blocks.length; i++) {
     const L = blocks[i];
-    if (L.type !== "table" || !L.table || !L.bbox || !L.pageNumber || !_chunkW33QEDLYcjs.CLIP_TABLES.has(L.table) || _chunkW33QEDLYcjs.EMPTY_PARTS.has(L.table)) continue;
+    if (L.type !== "table" || !L.table || !L.bbox || !L.pageNumber || !_chunkVSI5SRP2cjs.CLIP_TABLES.has(L.table) || _chunkVSI5SRP2cjs.EMPTY_PARTS.has(L.table)) continue;
     const ph = pageHeights.get(L.pageNumber);
     if (!ph || L.bbox.y > ph * PAGE_EDGE_BAND) continue;
     let s0 = i, s1 = i;
@@ -867,12 +876,12 @@ function mergeColumnFlow(blocks, pageHeights) {
 }
 function joinClipParts(prev, curr, pageHeights) {
   const pt = prev.table, ct = curr.table;
-  if (!_chunkW33QEDLYcjs.CLIP_TABLES.has(pt) || !_chunkW33QEDLYcjs.CLIP_TABLES.has(ct)) return null;
-  const px = _chunkW33QEDLYcjs.TABLE_COLXS.get(pt), cx = _chunkW33QEDLYcjs.TABLE_COLXS.get(ct);
+  if (!_chunkVSI5SRP2cjs.CLIP_TABLES.has(pt) || !_chunkVSI5SRP2cjs.CLIP_TABLES.has(ct)) return null;
+  const px = _chunkVSI5SRP2cjs.TABLE_COLXS.get(pt), cx = _chunkVSI5SRP2cjs.TABLE_COLXS.get(ct);
   if (!px || !cx) return null;
   let xs = cx, shifted = false, dx = 0;
   if (!shiftedSame(px, cx, false)) {
-    if (!_optionalChain([pageHeights, 'optionalAccess', _23 => _23.get, 'call', _24 => _24(prev.pageNumber)]) || !_optionalChain([pageHeights, 'optionalAccess', _25 => _25.get, 'call', _26 => _26(curr.pageNumber)])) return null;
+    if (!_optionalChain([pageHeights, 'optionalAccess', _25 => _25.get, 'call', _26 => _26(prev.pageNumber)]) || !_optionalChain([pageHeights, 'optionalAccess', _27 => _27.get, 'call', _28 => _28(curr.pageNumber)])) return null;
     dx = px[px.length - 1] - cx[cx.length - 1];
     if (Math.abs(dx) > CONTINUATION_COL_TOL) {
       xs = cx.map((x) => x + dx);
@@ -880,21 +889,21 @@ function joinClipParts(prev, curr, pageHeights) {
     } else dx = 0;
     if (!px.some((x) => Math.abs(x - xs[0]) <= CONTINUATION_COL_TOL)) return null;
   }
-  const res = joinSplitParts(pt, px, ct, xs, dx, _optionalChain([prev, 'access', _27 => _27.bbox, 'optionalAccess', _28 => _28.y]));
+  const res = joinSplitParts(pt, px, ct, xs, dx, _optionalChain([prev, 'access', _29 => _29.bbox, 'optionalAccess', _30 => _30.y]));
   if (!res || shifted && !res.split && !res.header) return null;
-  _chunkW33QEDLYcjs.TABLE_COLXS.set(res.table, res.colXs);
-  _chunkW33QEDLYcjs.CLIP_TABLES.add(res.table);
+  _chunkVSI5SRP2cjs.TABLE_COLXS.set(res.table, res.colXs);
+  _chunkVSI5SRP2cjs.CLIP_TABLES.add(res.table);
   return res.table;
 }
 var ANNEX_HEAD_RE = /^\s*[<\[(【]?\s*(?:붙\s*임|참\s*고|별\s*첨|별\s*지|별\s*표|첨\s*부|부\s*록)(?:\s*\d|\s*$|\s*[>\])】])|^\s*■/;
 function looksContinued(prev, curr, pageHeights) {
-  const ph = _optionalChain([pageHeights, 'optionalAccess', _29 => _29.get, 'call', _30 => _30(prev.pageNumber)]), ch = _optionalChain([pageHeights, 'optionalAccess', _31 => _31.get, 'call', _32 => _32(curr.pageNumber)]);
+  const ph = _optionalChain([pageHeights, 'optionalAccess', _31 => _31.get, 'call', _32 => _32(prev.pageNumber)]), ch = _optionalChain([pageHeights, 'optionalAccess', _33 => _33.get, 'call', _34 => _34(curr.pageNumber)]);
   if (ph && ch && (prev.bbox.y > ph * PAGE_EDGE_BAND || curr.bbox.y + curr.bbox.height < ch * (1 - PAGE_EDGE_BAND))) return false;
-  const firstText = _nullishCoalesce(_optionalChain([curr, 'access', _33 => _33.table, 'access', _34 => _34.cells, 'access', _35 => _35[0], 'optionalAccess', _36 => _36.find, 'call', _37 => _37((c) => c.text.trim()), 'optionalAccess', _38 => _38.text]), () => ( ""));
+  const firstText = _nullishCoalesce(_optionalChain([curr, 'access', _35 => _35.table, 'access', _36 => _36.cells, 'access', _37 => _37[0], 'optionalAccess', _38 => _38.find, 'call', _39 => _39((c) => c.text.trim()), 'optionalAccess', _40 => _40.text]), () => ( ""));
   return !ANNEX_HEAD_RE.test(firstText);
 }
 function firstRowSig(t) {
-  const x = _chunkW33QEDLYcjs.TABLE_COLXS.get(t);
+  const x = _chunkVSI5SRP2cjs.TABLE_COLXS.get(t);
   if (!x || x.length !== t.cols + 1) return null;
   return anchorsOf(t).filter((a) => a.r === 0).map((a) => ({ x1: x[a.c], x2: x[a.c + a.cs], t: a.cell.text.replace(/\s+/g, "") }));
 }
@@ -930,7 +939,7 @@ function chainHead(blocks, i, pageHeights) {
     if (pb.pageNumber !== (_nullishCoalesce(cur.pageNumber, () => ( 0))) - 1) break;
     if (!blocks.slice(p + 1, k).every((b) => besideOwn(b, pb, cur) || insideTable(b, pb))) break;
     if (!looksContinued(pb, cur, pageHeights)) break;
-    const px = _chunkW33QEDLYcjs.TABLE_COLXS.get(pb.table), cx = _chunkW33QEDLYcjs.TABLE_COLXS.get(cur.table);
+    const px = _chunkVSI5SRP2cjs.TABLE_COLXS.get(pb.table), cx = _chunkVSI5SRP2cjs.TABLE_COLXS.get(cur.table);
     if (!px || !cx) break;
     const dx = px[px.length - 1] - cx[cx.length - 1];
     if (!shiftedSame(px, cx) && !px.some((x) => Math.abs(x - (cx[0] + dx)) <= CONTINUATION_COL_TOL)) break;
@@ -959,12 +968,12 @@ var RESTART_LOOKBACK_PAGES = 5;
 // src/pdf/cell-continuation.ts
 var COL_MATCH_TOL = 0.5;
 function lastRowCell(t, x1, x2) {
-  const xs = _chunkW33QEDLYcjs.TABLE_COLXS.get(t);
+  const xs = _chunkVSI5SRP2cjs.TABLE_COLXS.get(t);
   const covered = /* @__PURE__ */ new Set();
   for (let r = 0; r < t.rows; r++) {
     for (let c = 0; c < t.cols; c++) {
       if (covered.has(r * 1e5 + c)) continue;
-      const cell = _optionalChain([t, 'access', _39 => _39.cells, 'access', _40 => _40[r], 'optionalAccess', _41 => _41[c]]);
+      const cell = _optionalChain([t, 'access', _41 => _41.cells, 'access', _42 => _42[r], 'optionalAccess', _43 => _43[c]]);
       if (!cell) continue;
       for (let dr = 0; dr < cell.rowSpan; dr++) for (let dc = 0; dc < cell.colSpan; dc++) covered.add((r + dr) * 1e5 + c + dc);
       if (r + cell.rowSpan !== t.rows) continue;
@@ -977,22 +986,36 @@ var cellBlocks = (cell, pageNumber) => _nullishCoalesce(cell.blocks, () => ( cel
 function mergeContinuedCells(blocks, pageHeights) {
   for (let j = blocks.length - 1; j > 0; j--) {
     const part = blocks[j];
-    const from = part.table ? _chunkW33QEDLYcjs.CONT_PARTS.get(part.table) : void 0;
+    const from = part.table ? _chunkVSI5SRP2cjs.CONT_PARTS.get(part.table) : void 0;
     if (!from || !part.table) continue;
     let i = j - 1;
     while (i >= 0 && blocks[i].type !== "table") i--;
     const prev = blocks[i];
     if (i < 0 || !prev.table || prev.pageNumber !== (_nullishCoalesce(part.pageNumber, () => ( 0))) - 1) continue;
     const cell = lastRowCell(prev.table, from.x1, from.x2);
-    const add = _optionalChain([part, 'access', _42 => _42.table, 'access', _43 => _43.cells, 'access', _44 => _44[0], 'optionalAccess', _45 => _45[0]]);
+    const add = _optionalChain([part, 'access', _44 => _44.table, 'access', _45 => _45.cells, 'access', _46 => _46[0], 'optionalAccess', _47 => _47[0]]);
     if (!cell || !add) continue;
+    const at = cell.blocks || add.blocks ? cellBlocks(cell, prev.pageNumber).length : -1;
     if (cell.blocks || add.blocks) cell.blocks = [...cellBlocks(cell, prev.pageNumber), ...cellBlocks(add, part.pageNumber)];
     cell.text = [cell.text, add.text].filter((s) => s.trim()).join("\n");
-    if (cell.text.trim() || _optionalChain([cell, 'access', _46 => _46.blocks, 'optionalAccess', _47 => _47.length])) _chunkW33QEDLYcjs.EMPTY_PARTS.delete(prev.table);
+    if (cell.text.trim() || _optionalChain([cell, 'access', _48 => _48.blocks, 'optionalAccess', _49 => _49.length])) _chunkVSI5SRP2cjs.EMPTY_PARTS.delete(prev.table);
     blocks.splice(j, 1);
+    if (cell.blocks && at > 0) fillNestedContinuation(cell.blocks, at);
     if (cell.blocks) mergeCrossPageTables(cell.blocks, pageHeights);
   }
 }
+function fillNestedContinuation(blocks, at) {
+  const prev = blocks[at - 1], next = blocks[at];
+  const t = _optionalChain([prev, 'optionalAccess', _50 => _50.table]), u = _optionalChain([next, 'optionalAccess', _51 => _51.table]);
+  if (!t || !u || u.rows !== 1 || u.cols !== 1 || !_chunkVSI5SRP2cjs.CLIP_TABLES.has(t) || !_chunkVSI5SRP2cjs.CLIP_TABLES.has(u) || next.pageNumber !== (_nullishCoalesce(prev.pageNumber, () => ( 0))) + 1) return;
+  const xs = _chunkVSI5SRP2cjs.TABLE_COLXS.get(u);
+  const cell = xs ? lastRowCell(t, xs[0], xs[xs.length - 1]) : void 0;
+  const add = _optionalChain([u, 'access', _52 => _52.cells, 'access', _53 => _53[0], 'optionalAccess', _54 => _54[0]]);
+  if (!cell || !add || cell.text.trim() || _optionalChain([cell, 'access', _55 => _55.blocks, 'optionalAccess', _56 => _56.length])) return;
+  cell.text = add.text;
+  if (add.blocks) cell.blocks = add.blocks;
+  blocks.splice(at, 1);
+}
 
 // src/pdf/symbol-fonts.ts
 var WINANSI_REVERSE = {
@@ -1026,7 +1049,7 @@ var WINANSI_REVERSE = {
 };
 function symbolFontTable(fontName) {
   if (!fontName) return void 0;
-  if (/wingdings(?![\s-]*[23])/i.test(fontName)) return _chunk3CWXYYCLcjs.WINGDINGS;
+  if (/wingdings(?![\s-]*[23])/i.test(fontName)) return _chunk3Y26ICMCcjs.WINGDINGS;
   return void 0;
 }
 function remapSymbolText(text, table) {
@@ -1087,8 +1110,8 @@ function splitSingleCellTables(blocks) {
   const out = [];
   for (const b of blocks) {
     const t = b.type === "table" ? b.table : void 0;
-    const cell = t && t.rows === 1 && t.cols === 1 ? _optionalChain([t, 'access', _48 => _48.cells, 'access', _49 => _49[0], 'optionalAccess', _50 => _50[0]]) : void 0;
-    if (!cell || _optionalChain([cell, 'access', _51 => _51.blocks, 'optionalAccess', _52 => _52.some, 'call', _53 => _53((x) => x.type === "table")])) {
+    const cell = t && t.rows === 1 && t.cols === 1 ? _optionalChain([t, 'access', _57 => _57.cells, 'access', _58 => _58[0], 'optionalAccess', _59 => _59[0]]) : void 0;
+    if (!cell || _optionalChain([cell, 'access', _60 => _60.blocks, 'optionalAccess', _61 => _61.some, 'call', _62 => _62((x) => x.type === "table")])) {
       out.push(b);
       continue;
     }
@@ -1104,10 +1127,10 @@ function cleanPdfText(text) {
   ).replace(/^(?!\| ---).*$/gm, (line) => {
     if (/^\s*\${1,2}.+\${1,2}\s*$/.test(line)) return line;
     if (line.startsWith("|")) {
-      return line.replace(/(?:\\.|[^|\\])+/g, (cell) => cell.split("<br>").map((seg) => seg.replace(/\S(?:.*\S)?/, (t) => _chunkW33QEDLYcjs.collapseEvenSpacing.call(void 0, t, false))).join("<br>"));
+      return line.replace(/(?:\\.|[^|\\])+/g, (cell) => cell.split("<br>").map((seg) => seg.replace(/\S(?:.*\S)?/, (t) => _chunkVSI5SRP2cjs.collapseEvenSpacing.call(void 0, t, false))).join("<br>"));
     }
-    const mark = _nullishCoalesce(_optionalChain([/^(?:#{1,6}|-) /, 'access', _54 => _54.exec, 'call', _55 => _55(line), 'optionalAccess', _56 => _56[0]]), () => ( ""));
-    return mark + _chunkW33QEDLYcjs.collapseEvenSpacing.call(void 0, line.slice(mark.length), false);
+    const mark = _nullishCoalesce(_optionalChain([/^(?:#{1,6}|-) /, 'access', _63 => _63.exec, 'call', _64 => _64(line), 'optionalAccess', _65 => _65[0]]), () => ( ""));
+    return mark + _chunkVSI5SRP2cjs.collapseEvenSpacing.call(void 0, line.slice(mark.length), false);
   }).replace(/\\~\\~/g, "~~").replace(/~~~~/g, "").replace(/<u>\s*<\/u>/g, "").replace(/\n{3,}/g, "\n\n").trim();
 }
 function startsWithMarker(line) {
@@ -1139,7 +1162,7 @@ function mergeKoreanLines(text) {
       continue;
     }
     if (/[가-힣·,\-]$/.test(prev) && /^[가-힣(]/.test(curr) && !startsWithMarker(curr) && !isStandaloneHeader(prev) && !startsWithMarker(prev)) {
-      result[result.length - 1] = prev + _chunkW33QEDLYcjs.wrapJoiner.call(void 0, prev, curr) + curr;
+      result[result.length - 1] = prev + _chunkVSI5SRP2cjs.wrapJoiner.call(void 0, prev, curr) + curr;
     } else {
       result.push(curr);
     }
@@ -1155,7 +1178,7 @@ function applyLinkAnnotations(items, annots) {
   const wrapped = /* @__PURE__ */ new Set();
   for (const a of annots) {
     if (a.subtype !== "Link" || !a.url || !a.rect || a.rect.length < 4) continue;
-    const url = _chunkLBYKEJUJcjs.sanitizeHref.call(void 0, a.url);
+    const url = _chunkVGU4FLLDcjs.sanitizeHref.call(void 0, a.url);
     if (!url) continue;
     const x1 = Math.min(a.rect[0], a.rect[2]), x2 = Math.max(a.rect[0], a.rect[2]);
     const y1 = Math.min(a.rect[1], a.rect[3]), y2 = Math.max(a.rect[1], a.rect[3]);
@@ -1186,7 +1209,7 @@ function applyLinkAnnotations(items, annots) {
 
 // src/pdf/formula-ocr.ts
 async function applyFormulaOcr(buffer, blocks, pageFilter, effectivePageCount, warnings, _onProgress) {
-  const formulaMod = await Promise.resolve().then(() => _interopRequireWildcard(require("./formula-RGK6WGD2.cjs")));
+  const formulaMod = await Promise.resolve().then(() => _interopRequireWildcard(require("./formula-DYVJSHWN.cjs")));
   const { FormulaPipeline, ensureFormulaModels } = formulaMod;
   await ensureFormulaModels((p) => {
     if (p.phase === "download" && p.total) {
@@ -1341,7 +1364,8 @@ try {
 }
 async function loadPdfWithTimeout(buffer) {
   const loadingTask = _pdfmjs.getDocument.call(void 0, {
-    data: new Uint8Array(buffer),
+    // pdfjs transfers its input to the worker; retain the caller's buffer for reuse.
+    data: new Uint8Array(buffer.slice(0)),
     useSystemFonts: true,
     disableFontFace: true,
     isEvalSupported: false,
@@ -1352,11 +1376,14 @@ async function loadPdfWithTimeout(buffer) {
   let timer;
   try {
     return await Promise.race([
-      loadingTask.promise,
+      loadingTask.promise.catch((e) => {
+        if (e instanceof Error && e.name === "PasswordException") throw new (0, _chunkVGU4FLLDcjs.KordocError)("\uC554\uD638\uB85C \uBCF4\uD638\uB41C PDF \uD30C\uC77C\uC785\uB2C8\uB2E4 (PDF \uC5F4\uAE30 \uC554\uD638\uB294 \uC9C0\uC6D0\uD558\uC9C0 \uC54A\uC2B5\uB2C8\uB2E4)");
+        throw e;
+      }),
       new Promise((_, reject) => {
         timer = setTimeout(() => {
           loadingTask.destroy();
-          reject(new (0, _chunkLBYKEJUJcjs.KordocError)("PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (30\uCD08 \uCD08\uACFC)"));
+          reject(new (0, _chunkVGU4FLLDcjs.KordocError)("PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (30\uCD08 \uCD08\uACFC)"));
         }, PDF_LOAD_TIMEOUT_MS);
       })
     ]);
@@ -1365,12 +1392,12 @@ async function loadPdfWithTimeout(buffer) {
   }
 }
 async function parsePdfDocument(buffer, options) {
-  const formulaBuffer = _optionalChain([options, 'optionalAccess', _57 => _57.formulaOcr]) ? buffer.slice(0) : null;
-  const ocrBuffer = _optionalChain([options, 'optionalAccess', _58 => _58.ocr]) ? buffer.slice(0) : null;
+  const formulaBuffer = _optionalChain([options, 'optionalAccess', _66 => _66.formulaOcr]) ? buffer : null;
+  const ocrBuffer = _optionalChain([options, 'optionalAccess', _67 => _67.ocr]) ? buffer : null;
   const doc = await loadPdfWithTimeout(buffer);
   try {
     const pageCount = doc.numPages;
-    if (pageCount === 0) throw new (0, _chunkLBYKEJUJcjs.KordocError)("PDF\uC5D0 \uD398\uC774\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
+    if (pageCount === 0) throw new (0, _chunkVGU4FLLDcjs.KordocError)("PDF\uC5D0 \uD398\uC774\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
     const metadata = { pageCount, pageMode: "layout" };
     await extractPdfMetadata(doc, metadata);
     const blocks = [];
@@ -1379,7 +1406,10 @@ async function parsePdfDocument(buffer, options) {
     let totalChars = 0;
     let totalTextBytes = 0;
     const effectivePageCount = Math.min(pageCount, MAX_PAGES);
-    const pageFilter = _optionalChain([options, 'optionalAccess', _59 => _59.pages]) ? _chunkDCZVOIEOcjs.parsePageRange.call(void 0, options.pages, effectivePageCount) : null;
+    const pageFilter = _optionalChain([options, 'optionalAccess', _68 => _68.pages]) ? _chunkJ55UQC5Hcjs.parsePageRange.call(void 0, options.pages, effectivePageCount) : null;
+    if (pageCount > MAX_PAGES && (!_optionalChain([options, 'optionalAccess', _69 => _69.pages]) || _chunkJ55UQC5Hcjs.hasRequestedPagesAfter.call(void 0, options.pages, MAX_PAGES, pageCount))) {
+      warnings.push({ message: `${pageCount}\uCABD \uC911 \uC55E ${MAX_PAGES}\uCABD\uB9CC \uD30C\uC2F1\uD588\uC2B5\uB2C8\uB2E4 (\uCABD \uC218 \uC0C1\uD55C)`, code: "PARTIAL_PARSE" });
+    }
     const totalTarget = pageFilter ? pageFilter.size : effectivePageCount;
     const fontSizeFreq = /* @__PURE__ */ new Map();
     const pageHeights = /* @__PURE__ */ new Map();
@@ -1390,18 +1420,21 @@ async function parsePdfDocument(buffer, options) {
     const pageImageBlocks = /* @__PURE__ */ new Map();
     const carry = {};
     const vectorPageOps = /* @__PURE__ */ new Map();
-    const wrapLexicon = new (0, _chunkW33QEDLYcjs.WrapLexicon)();
+    const wrapLexicon = new (0, _chunkVSI5SRP2cjs.WrapLexicon)();
     let parsedPages = 0;
     for (let i = 1; i <= effectivePageCount; i++) {
       if (pageFilter && !pageFilter.has(i)) continue;
+      let loadedPage;
       try {
         const page = await doc.getPage(i);
+        loadedPage = page;
         const tc = await page.getTextContent();
-        const viewport = page.getViewport({ scale: 1 });
-        pageHeights.set(i, viewport.height);
+        const [viewX1, viewY1, viewX2, viewY2] = page.view;
+        const pageW = viewX2 - viewX1, pageH = viewY2 - viewY1;
+        pageHeights.set(i, pageH);
         const rawItems = tc.items;
-        const items = _chunkW33QEDLYcjs.normalizeItems.call(void 0, rawItems);
-        const { visible, hiddenCount } = _chunkW33QEDLYcjs.filterHiddenText.call(void 0, items, viewport.width, viewport.height);
+        const items = _chunkVSI5SRP2cjs.normalizeItems.call(void 0, rawItems);
+        const { visible, hiddenCount } = _chunkVSI5SRP2cjs.filterHiddenText.call(void 0, items, pageW, pageH, viewX1, viewY1);
         if (hiddenCount > 0) {
           warnings.push({ page: i, message: `${hiddenCount}\uAC1C \uC228\uACA8\uC9C4 \uD14D\uC2A4\uD2B8 \uC694\uC18C \uD544\uD130\uB9C1\uB428`, code: "HIDDEN_TEXT_FILTERED" });
         }
@@ -1413,18 +1446,27 @@ async function parsePdfDocument(buffer, options) {
           applyLinkAnnotations(visible, annots);
         } catch (e4) {
         }
-        const opList = await page.getOperatorList();
+        const rawOps = await page.getOperatorList();
+        const shifted = viewX1 !== 0 || viewY1 !== 0;
+        if (shifted) for (const item of visible) {
+          item.x -= viewX1;
+          item.y -= viewY1;
+        }
+        const opList = shifted ? {
+          fnArray: [_pdfmjs.OPS.transform, ...rawOps.fnArray],
+          argsArray: [[1, 0, 0, 1, -viewX1, -viewY1], ...rawOps.argsArray]
+        } : rawOps;
         remapSymbolFontItems(visible, (loadedName) => {
           try {
-            return page.commonObjs.has(loadedName) ? _optionalChain([page, 'access', _60 => _60.commonObjs, 'access', _61 => _61.get, 'call', _62 => _62(loadedName), 'optionalAccess', _63 => _63.name]) : void 0;
+            return page.commonObjs.has(loadedName) ? _optionalChain([page, 'access', _70 => _70.commonObjs, 'access', _71 => _71.get, 'call', _72 => _72(loadedName), 'optionalAccess', _73 => _73.name]) : void 0;
           } catch (e5) {
             return void 0;
           }
         });
-        for (const it of visible) if (it.text.includes("$")) it.text = _chunk3CWXYYCLcjs.escapeLiteralDollar.call(void 0, it.text);
-        const pageArea = viewport.width * viewport.height;
+        for (const it of visible) if (it.text.includes("$")) it.text = _chunk3Y26ICMCcjs.escapeLiteralDollar.call(void 0, it.text);
+        const pageArea = pageW * pageH;
         if (pageArea > 0) {
-          const imageRegions = _chunkW33QEDLYcjs.extractImageRegions.call(void 0, opList.fnArray, opList.argsArray);
+          const imageRegions = _chunkVSI5SRP2cjs.extractImageRegions.call(void 0, opList.fnArray, opList.argsArray);
           let uncovered = 0;
           for (const r of imageRegions) {
             const area = (r.x2 - r.x1) * (r.y2 - r.y1);
@@ -1439,10 +1481,10 @@ async function parsePdfDocument(buffer, options) {
           }
           if (uncovered > 0) skippedImagePages.set(i, uncovered);
         }
-        const pageBlocks = _chunkW33QEDLYcjs.extractPageBlocksWithLines.call(void 0, visible, i, opList, viewport.width, viewport.height, void 0, _optionalChain([options, 'optionalAccess', _64 => _64.tables]) !== false, carry, wrapLexicon);
+        const pageBlocks = _chunkVSI5SRP2cjs.extractPageBlocksWithLines.call(void 0, visible, i, opList, pageW, pageH, void 0, _optionalChain([options, 'optionalAccess', _74 => _74.tables]) !== false, carry, wrapLexicon);
         for (const b of pageBlocks) blocks.push(b);
         try {
-          const { blocks: imgBlocks, images: pageImages } = await extractPageImages(page, opList.fnArray, opList.argsArray, i, imageState, warnings, _optionalChain([options, 'optionalAccess', _65 => _65.images]) !== false);
+          const { blocks: imgBlocks, images: pageImages } = await extractPageImages(page, opList.fnArray, opList.argsArray, i, imageState, warnings, _optionalChain([options, 'optionalAccess', _75 => _75.images]) !== false);
           if (imgBlocks.length > 0) pageImageBlocks.set(i, imgBlocks);
           extractedImages.push(...pageImages);
         } catch (e6) {
@@ -1461,21 +1503,23 @@ async function parsePdfDocument(buffer, options) {
         const vector = scanVectorGlyphs(opList.fnArray, opList.argsArray, visible);
         const quality = computePageQuality(i, pageText, vector.glyphs);
         pageQuality.push(quality);
-        if (_optionalChain([options, 'optionalAccess', _66 => _66.ocr]) && quality.ocrReason === "vector_text" && page.rotate % 360 === 0 && page.view[0] === 0 && page.view[1] === 0) {
+        if (_optionalChain([options, 'optionalAccess', _76 => _76.ocr]) && quality.ocrReason === "vector_text" && page.rotate % 360 === 0 && page.view[0] === 0 && page.view[1] === 0) {
           vectorPageOps.set(i, ocrVectorOps(opList, vector.paths));
         }
-        if (totalTextBytes > MAX_TOTAL_TEXT) throw new (0, _chunkLBYKEJUJcjs.KordocError)("\uD14D\uC2A4\uD2B8 \uCD94\uCD9C \uD06C\uAE30 \uCD08\uACFC");
+        if (totalTextBytes > MAX_TOTAL_TEXT) throw new (0, _chunkVGU4FLLDcjs.KordocError)("\uD14D\uC2A4\uD2B8 \uCD94\uCD9C \uD06C\uAE30 \uCD08\uACFC");
         parsedPages++;
-        _optionalChain([options, 'optionalAccess', _67 => _67.onProgress, 'optionalCall', _68 => _68(parsedPages, totalTarget)]);
+        _optionalChain([options, 'optionalAccess', _77 => _77.onProgress, 'optionalCall', _78 => _78(parsedPages, totalTarget)]);
       } catch (pageErr) {
-        if (pageErr instanceof _chunkLBYKEJUJcjs.KordocError) throw pageErr;
+        if (pageErr instanceof _chunkVGU4FLLDcjs.KordocError) throw pageErr;
         warnings.push({ page: i, message: `\uD398\uC774\uC9C0 ${i} \uD30C\uC2F1 \uC2E4\uD328: ${pageErr instanceof Error ? pageErr.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958"}`, code: "PARTIAL_PARSE" });
+      } finally {
+        _optionalChain([loadedPage, 'optionalAccess', _79 => _79.cleanup, 'call', _80 => _80()]);
       }
     }
     const parsedPageCount = parsedPages || (pageFilter ? pageFilter.size : effectivePageCount);
-    const isImageBased = totalChars / Math.max(parsedPageCount, 1) < 10;
+    const isImageBased = _optionalChain([pageFilter, 'optionalAccess', _81 => _81.size]) !== 0 && totalChars / Math.max(parsedPageCount, 1) < 10;
     const ocrDone = /* @__PURE__ */ new Set();
-    if (_optionalChain([options, 'optionalAccess', _69 => _69.ocr]) && ocrBuffer) {
+    if (_optionalChain([options, 'optionalAccess', _82 => _82.ocr]) && ocrBuffer) {
       const inScope = (p) => !pageFilter || pageFilter.has(p);
       const targets = /* @__PURE__ */ new Set();
       if (options.ocr === "force" || isImageBased) {
@@ -1489,16 +1533,16 @@ async function parsePdfDocument(buffer, options) {
       }
       if (targets.size > 0) {
         try {
-          const { runPdfOcr } = await Promise.resolve().then(() => _interopRequireWildcard(require("./pdf-ocr-E3IZ3EI5.cjs")));
+          const { runPdfOcr } = await Promise.resolve().then(() => _interopRequireWildcard(require("./pdf-ocr-K7MJFRBS.cjs")));
           const mode = typeof options.ocr === "function" ? options.ocr : "builtin";
           const ocrPageBlocks = await runPdfOcr(ocrBuffer, targets, mode, warnings, options.onProgress, options.tables !== false, vectorPageOps);
           if (ocrPageBlocks.size > 0) {
             for (const p of ocrPageBlocks.keys()) ocrDone.add(p);
             const merged = blocks.filter((b) => !(b.pageNumber && ocrDone.has(b.pageNumber)));
-            for (const obs of ocrPageBlocks.values()) merged.push(...obs);
+            for (const obs of ocrPageBlocks.values()) for (const b of obs) merged.push(b);
             merged.sort((a, b) => (_nullishCoalesce(a.pageNumber, () => ( 0))) - (_nullishCoalesce(b.pageNumber, () => ( 0))));
             blocks.length = 0;
-            blocks.push(...merged);
+            for (const b of merged) blocks.push(b);
             for (const pq of pageQuality) if (ocrDone.has(pq.page)) pq.ocrApplied = true;
             warnings.push({
               message: `${ocrDone.size}\uAC1C \uD398\uC774\uC9C0\uC5D0 OCR \uC801\uC6A9 (${mode === "builtin" ? "\uB0B4\uC7A5 PP-OCRv5" : "\uC0AC\uC6A9\uC790 \uD504\uB85C\uBC14\uC774\uB354"})`,
@@ -1539,18 +1583,18 @@ async function parsePdfDocument(buffer, options) {
         warnings.push({ page, message: `${count}\uAC1C \uC774\uBBF8\uC9C0 \uC601\uC5ED\uC5D0 \uCD94\uCD9C \uAC00\uB2A5\uD55C \uD14D\uC2A4\uD2B8 \uC5C6\uC74C (\uADF8\uB9BC/\uCC28\uD2B8/\uB3C4\uC7A5 \uB0B4\uC6A9 \uB204\uB77D \uAC00\uB2A5)`, code: "SKIPPED_IMAGE" });
       }
     }
-    if (_optionalChain([options, 'optionalAccess', _70 => _70.removeHeaderFooter]) !== false && parsedPageCount >= 3) {
-      const removed = _chunkW33QEDLYcjs.removeHeaderFooterBlocks.call(void 0, blocks, pageHeights, warnings);
+    if (_optionalChain([options, 'optionalAccess', _83 => _83.removeHeaderFooter]) !== false && parsedPageCount >= 3) {
+      const removed = _chunkVSI5SRP2cjs.removeHeaderFooterBlocks.call(void 0, blocks, pageHeights, warnings);
       for (let ri = removed.length - 1; ri >= 0; ri--) {
         blocks.splice(removed[ri], 1);
       }
     }
     mergeContinuedCells(blocks, pageHeights);
     mergeCrossPageTables(blocks, pageHeights);
-    if (!_optionalChain([options, 'optionalAccess', _71 => _71.keepTrailingEmptyCols])) _chunkW33QEDLYcjs.trimTrailingEmptyTableCols.call(void 0, blocks);
-    _chunkW33QEDLYcjs.joinPageBreakWraps.call(void 0, blocks, wrapLexicon);
+    if (!_optionalChain([options, 'optionalAccess', _84 => _84.keepTrailingEmptyCols])) _chunkVSI5SRP2cjs.trimTrailingEmptyTableCols.call(void 0, blocks);
+    _chunkVSI5SRP2cjs.joinPageBreakWraps.call(void 0, blocks, wrapLexicon);
     injectPageImageBlocks(blocks, pageImageBlocks);
-    if (_optionalChain([options, 'optionalAccess', _72 => _72.formulaOcr]) && formulaBuffer) {
+    if (_optionalChain([options, 'optionalAccess', _85 => _85.formulaOcr]) && formulaBuffer) {
       try {
         await applyFormulaOcr(formulaBuffer, blocks, pageFilter, effectivePageCount, warnings, options.onProgress);
       } catch (e) {
@@ -1560,19 +1604,21 @@ async function parsePdfDocument(buffer, options) {
         });
       }
     }
-    const medianFontSize = _chunkW33QEDLYcjs.computeMedianFontSizeFromFreq.call(void 0, fontSizeFreq);
+    const medianFontSize = _chunkVSI5SRP2cjs.computeMedianFontSizeFromFreq.call(void 0, fontSizeFreq);
     if (medianFontSize > 0) {
-      _chunkW33QEDLYcjs.detectHeadings.call(void 0, blocks, medianFontSize);
+      _chunkVSI5SRP2cjs.detectHeadings.call(void 0, blocks, medianFontSize);
     }
-    _chunkW33QEDLYcjs.detectMarkerHeadings.call(void 0, blocks);
-    _chunkW33QEDLYcjs.detectTableCaptions.call(void 0, blocks);
-    _chunkW33QEDLYcjs.detectKoreanListBlocks.call(void 0, blocks);
+    _chunkVSI5SRP2cjs.detectMarkerHeadings.call(void 0, blocks);
+    _chunkVSI5SRP2cjs.detectTableCaptions.call(void 0, blocks);
+    _chunkVSI5SRP2cjs.detectKoreanListBlocks.call(void 0, blocks);
     const outline = blocks.filter((b) => b.type === "heading" && b.level && b.text).map((b) => ({ level: b.level, text: b.text, pageNumber: b.pageNumber }));
     sanitizeBlockControlChars(blocks);
     const outBlocks = splitSingleCellTables(blocks);
-    let markdown = cleanPdfText(_chunk3CWXYYCLcjs.blocksToMarkdown.call(void 0, outBlocks));
+    const finishMarkdown = (bs) => cleanPdfText(_chunk3Y26ICMCcjs.blocksToMarkdown.call(void 0, bs));
+    let markdown = finishMarkdown(outBlocks);
     return {
       markdown,
+      pages: _chunk3Y26ICMCcjs.blocksToPages.call(void 0, outBlocks, finishMarkdown),
       blocks,
       metadata,
       outline: outline.length > 0 ? outline : void 0,
@@ -1590,7 +1636,7 @@ async function parsePdfDocument(buffer, options) {
 async function extractPdfMetadata(doc, metadata) {
   try {
     const result = await doc.getMetadata();
-    if (!_optionalChain([result, 'optionalAccess', _73 => _73.info])) return;
+    if (!_optionalChain([result, 'optionalAccess', _86 => _86.info])) return;
     const info = result.info;
     if (typeof info.Title === "string" && info.Title.trim()) metadata.title = info.Title.trim();
     if (typeof info.Author === "string" && info.Author.trim()) metadata.author = info.Author.trim();
@@ -1629,5 +1675,5 @@ async function extractPdfMetadataOnly(buffer) {
 
 
 
-exports.cleanPdfText = cleanPdfText; exports.detectKoreanListBlocks = _chunkW33QEDLYcjs.detectKoreanListBlocks; exports.detectTableCaptions = _chunkW33QEDLYcjs.detectTableCaptions; exports.extractPdfMetadataOnly = extractPdfMetadataOnly; exports.mergeCrossPageTables = mergeCrossPageTables; exports.parsePdfDocument = parsePdfDocument; exports.removeHeaderFooterBlocks = _chunkW33QEDLYcjs.removeHeaderFooterBlocks;
-//# sourceMappingURL=parser-Y5NX55GC.cjs.map
+exports.cleanPdfText = cleanPdfText; exports.detectKoreanListBlocks = _chunkVSI5SRP2cjs.detectKoreanListBlocks; exports.detectTableCaptions = _chunkVSI5SRP2cjs.detectTableCaptions; exports.extractPdfMetadataOnly = extractPdfMetadataOnly; exports.mergeCrossPageTables = mergeCrossPageTables; exports.parsePdfDocument = parsePdfDocument; exports.removeHeaderFooterBlocks = _chunkVSI5SRP2cjs.removeHeaderFooterBlocks;
+//# sourceMappingURL=parser-XCXDSKQ3.cjs.map