kordoc 4.16.0 → 4.16.2

package/dist/{parser-YAQV7FBO.cjs → parser-2LU4EVU3.cjs}

@@ -4,7 +4,7 @@
 
 
 
-var _chunkDRY3L5KQcjs = require('./chunk-DRY3L5KQ.cjs');
+var _chunkXTFN5WKScjs = require('./chunk-XTFN5WKS.cjs');
 
 
 
@@ -43,13 +43,16 @@ var _chunkJ55UQC5Hcjs = require('./chunk-J55UQC5H.cjs');
 
 
 
-var _chunkMH6MAL64cjs = require('./chunk-MH6MAL64.cjs');
-require('./chunk-ZY5K3LPM.cjs');
-require('./chunk-H6KMTYYZ.cjs');
 
+var _chunkWDQTFAXYcjs = require('./chunk-WDQTFAXY.cjs');
 
 
-var _chunkNWODPYBDcjs = require('./chunk-NWODPYBD.cjs');
+var _chunk62W7QCEXcjs = require('./chunk-62W7QCEX.cjs');
+require('./chunk-CEIATRDW.cjs');
+
+
+
+var _chunkSTM3ORY7cjs = require('./chunk-STM3ORY7.cjs');
 
 
 var _chunkGS7T56RPcjs = require('./chunk-GS7T56RP.cjs');
@@ -243,7 +246,7 @@ async function extractPageImages(page, fnArray, argsArray, pageNumber, state, wa
     const filename = `image_${String(state.imageIndex).padStart(3, "0")}.png`;
     state.seen.set(hash, filename);
     if (withBytes) {
-      const png = _chunkDRY3L5KQcjs.encodePng.call(void 0, w, h, rgba);
+      const png = _chunkXTFN5WKScjs.encodePng.call(void 0, w, h, rgba);
       state.totalBytes += png.length;
       images.push({ filename, data: png, mimeType: "image/png" });
     }
@@ -637,6 +640,40 @@ function ocrVectorOps(opList, paths) {
   return { fnArray, argsArray };
 }
 
+// src/pdf/contact-table.ts
+var CONTACT_HEAD = /^(?:담당\s*부서|<[^<>]+>)$/;
+var CONTACT_ROLE = /^(?:책임자|담당자)$/;
+var PERSON = /^(.+?)\s+([가-힣]{2,4})\s+(\([^()]*\d[^()]*\)|\S+@\S+)$/;
+function splitContactTables(blocks) {
+  for (const b of blocks) {
+    const t = b.type === "table" ? b.table : void 0;
+    if (!t || t.cols !== 4 || t.rows < 2) continue;
+    const rows = t.cells;
+    if (!CONTACT_HEAD.test(rows[0][0].text.replace(/\s+/g, " ").trim())) continue;
+    const people = rows.map((r) => PERSON.exec(r[3].text.replace(/\s+/g, " ").trim()));
+    if (people.some((p) => !p) || rows.some((r) => !CONTACT_ROLE.test(r[2].text.replace(/\s+/g, "")) || r[2].rowSpan !== 1 || r[3].rowSpan !== 1)) continue;
+    const lead = [];
+    let ok = true;
+    for (const c of [0, 1]) {
+      const top = rows[0][c];
+      if (top.rowSpan === 1) {
+        lead.push(rows.map((r) => r[c].text));
+        continue;
+      }
+      const lines = top.text.split("\n").map((s) => s.trim()).filter(Boolean);
+      if (top.rowSpan !== t.rows || lines.length > t.rows) {
+        ok = false;
+        break;
+      }
+      lead.push(rows.map((_, r) => _nullishCoalesce(lines[r], () => ( ""))));
+    }
+    if (!ok) continue;
+    const one = (text) => ({ text, colSpan: 1, rowSpan: 1 });
+    t.cells = rows.map((r, i) => [one(lead[0][i]), one(lead[1][i]), one(r[2].text), one(people[i][1]), one(people[i][2]), one(people[i][3])]);
+    t.cols = 6;
+  }
+}
+
 // src/pdf/table-parts.ts
 var PART_COL_TOL = 1;
 function anchorsOf(table) {
@@ -673,7 +710,7 @@ var indexOf = (coords, x) => {
 };
 var rowText = (anchors, r) => anchors.filter((a) => a.r === r).map((a) => a.cell.text.replace(/\s+/g, "")).join("|");
 var hasText = (t) => t.replace(/\|/g, "") !== "";
-function joinSplitParts(prev, pcx, curr, ccx, dx = 0, prevBottom) {
+function joinSplitParts(prev, pcx, curr, ccx, dx = 0, prevBottom, lex) {
   if (pcx.length !== prev.cols + 1 || ccx.length !== curr.cols + 1) return null;
   const U = unionCoords(pcx, ccx);
   const cols = U.length - 1;
@@ -696,14 +733,14 @@ 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 = _chunkMH6MAL64cjs.CELL_LINES.get(a.cell);
-    if (lines) _chunkMH6MAL64cjs.CELL_LINES.set(grid[r][c1], shift ? lines.map((l) => ({ ...l, l: l.l + shift, r: l.r + shift })) : lines);
-    if (_chunkMH6MAL64cjs.IMAGE_CELLS.has(a.cell)) _chunkMH6MAL64cjs.IMAGE_CELLS.add(grid[r][c1]);
+    const lines = _chunkWDQTFAXYcjs.CELL_LINES.get(a.cell);
+    if (lines) _chunkWDQTFAXYcjs.CELL_LINES.set(grid[r][c1], shift ? lines.map((l) => ({ ...l, l: l.l + shift, r: l.r + shift })) : lines);
+    if (_chunkWDQTFAXYcjs.IMAGE_CELLS.has(a.cell)) _chunkWDQTFAXYcjs.IMAGE_CELLS.add(grid[r][c1]);
     return true;
   };
-  for (const a of pa) if (!_chunkMH6MAL64cjs.FILLER_CELLS.has(a.cell) && !place(a, pcx, 0)) return null;
+  for (const a of pa) if (!_chunkWDQTFAXYcjs.FILLER_CELLS.has(a.cell) && !place(a, pcx, 0)) return null;
   for (const a of ca) {
-    if (a.r < skip || _chunkMH6MAL64cjs.FILLER_CELLS.has(a.cell)) continue;
+    if (a.r < skip || _chunkWDQTFAXYcjs.FILLER_CELLS.has(a.cell)) continue;
     if (!place(a, ccx, prev.rows - skip, dx)) return null;
   }
   const first = prev.rows;
@@ -727,15 +764,37 @@ 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) _chunkMH6MAL64cjs.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) _chunkWDQTFAXYcjs.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);
+  let split = rows > prev.rows && mergeSplitRow(table, owner, prev.rows, U);
+  if (rows > prev.rows && !split && prevBottom !== void 0 && mergeStraddlingCells(table, owner, prev.rows, prevBottom, lex)) split = mergeSplitRow(table, owner, prev.rows, U, true);
   return { table, colXs: U, split, header: ca.filter((a) => a.r < skip && a.cell.text.trim()).length >= 2 };
 }
-function mergeStraddlingCells(table, owner, first, prevBottom) {
+function mergeStraddlingCells(table, owner, first, prevBottom, lex) {
   const last = first - 1;
-  if (!owner[first].some((o, c) => o !== null && o === owner[last][c])) return;
+  const straddles = owner[first].some((o, c) => o !== null && o === owner[last][c]);
+  let firstNew = table.cols;
+  for (let c = 0; c < table.cols; c++) {
+    const d = owner[first][c];
+    if (d && d.r === first && d !== owner[last][c] && hasContent(table.cells[d.r][d.c])) {
+      firstNew = c;
+      break;
+    }
+  }
+  let midWord = false;
+  if (lex) for (let c = 0; c < table.cols && !midWord; ) {
+    const u = owner[last][c], d = owner[first][c];
+    if (!u) {
+      c++;
+      continue;
+    }
+    c = u.c + u.cs;
+    if (!d || d === u || u.r + u.rs - 1 !== last || d.r !== first || d.c !== u.c || d.cs !== u.cs) continue;
+    if (u.r !== last || d.rs < 2) continue;
+    const tail = _nullishCoalesce(table.cells[u.r][u.c].text.trim().split(/\s+/).pop(), () => ( "")), head = _nullishCoalesce(table.cells[first][d.c].text.trim().split(/\s+/)[0], () => ( ""));
+    if (/^[가-힣]{2,}$/.test(tail) && /^[가-힣]{2,}/.test(head) && lex.evidence2(tail, head) === "") midWord = true;
+  }
+  let merged = false;
   for (let c = 0; c < table.cols; ) {
     const u = owner[last][c], d = owner[first][c];
     if (!u) {
@@ -743,18 +802,24 @@ function mergeStraddlingCells(table, owner, first, prevBottom) {
       continue;
     }
     c = u.c + u.cs;
-    if (!d || d === u || u.r >= last && d.rs < 2 || u.r + u.rs - 1 !== last || d.r !== first || d.c !== u.c || d.cs !== u.cs) continue;
+    if (!d || d === u || 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 = _chunkMH6MAL64cjs.CELL_LINES.get(a);
-    if (!_optionalChain([U, 'optionalAccess', _33 => _33.length]) || U.length > STRADDLE_MAX_LINES || !hasContent(b)) continue;
-    const lu = U[U.length - 1];
-    if (lu.y - prevBottom > (u.r >= last ? STRADDLE_BOTTOM_ONE_ROW : STRADDLE_BOTTOM) * (lu.h || 10)) continue;
+    if (!(u.rs >= 2 && u.c + u.cs <= firstNew && hasContent(a) && !hasContent(b))) {
+      if (!straddles && !midWord) continue;
+      if (u.r >= last && d.rs < 2) continue;
+      const U = _chunkWDQTFAXYcjs.CELL_LINES.get(a);
+      if (!_optionalChain([U, 'optionalAccess', _33 => _33.length]) || U.length > STRADDLE_MAX_LINES || !hasContent(b)) continue;
+      const lu = U[U.length - 1];
+      if (!midWord && lu.y - prevBottom > (u.r >= last ? STRADDLE_BOTTOM_ONE_ROW : STRADDLE_BOTTOM) * (lu.h || 10)) continue;
+    }
     appendCell(a, b);
     a.rowSpan = first + d.rs - u.r;
     table.cells[first][d.c] = { text: "", colSpan: 1, rowSpan: 1 };
     const grown = { ...u, rs: a.rowSpan };
     for (let rr = u.r; rr < u.r + grown.rs; rr++) for (let dc = u.c; dc < u.c + u.cs; dc++) owner[rr][dc] = grown;
+    merged = true;
   }
+  return merged;
 }
 var FULL_LINE_TOL = 1;
 var FULL_LINE_MIN_FRAC = 0.5;
@@ -769,7 +834,7 @@ var LABEL_ROOM = 1.5;
 var HANGING_MIN = 0.5;
 var HANGING_TOL = 1;
 function continuesAcross(u, d, x1, x2) {
-  const U = _chunkMH6MAL64cjs.CELL_LINES.get(u), D = _chunkMH6MAL64cjs.CELL_LINES.get(d);
+  const U = _chunkWDQTFAXYcjs.CELL_LINES.get(u), D = _chunkWDQTFAXYcjs.CELL_LINES.get(d);
   if (!_optionalChain([U, 'optionalAccess', _34 => _34.length]) || !_optionalChain([D, 'optionalAccess', _35 => _35.length])) return false;
   const last = U[U.length - 1];
   const fs = last.h || 10;
@@ -790,11 +855,21 @@ function continuesAcross(u, d, x1, x2) {
   const hanging = U.length >= 2 && U.every((l) => maxR - l.r <= FULL_LINE_TOL * fs) && last.l - minL >= HANGING_MIN * fs && Math.abs(D[0].l - last.l) <= HANGING_TOL;
   return U.some(leftAligned) || D.some(leftAligned) || hanging;
 }
-function lineEnded(c, x1, x2) {
-  const L = _chunkMH6MAL64cjs.CELL_LINES.get(c);
+var OUTLINE_HEAD = /^[□■]\s*\S/;
+var OUTLINE_ITEM = /^[ㅇ○◦\-‐–·․‧※*]\s*\S/;
+function outlineContinues(u, d) {
+  const U = u.text.split("\n").map((l) => l.trim()).filter(Boolean);
+  const D = d.text.split("\n").map((l) => l.trim()).filter(Boolean);
+  return U.length >= 2 && D.length >= 1 && U.some((l) => OUTLINE_HEAD.test(l)) && OUTLINE_ITEM.test(D[0]);
+}
+function lineEnded(c, x1, x2, next = "") {
+  const L = _chunkWDQTFAXYcjs.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);
+  const l = L[0], h = l.h || 10;
+  const word = _nullishCoalesce(next.trim().split(/\s+/)[0], () => ( ""));
+  let units = 0;
+  for (const ch of word) units += /[가-힣\u3400-\u9fff]/.test(ch) ? 1 : 0.55;
+  return x2 - l.r - (l.l - x1) >= Math.max(LABEL_ROOM, units) * h;
 }
 function appendCell(a, b) {
   if (_optionalChain([a, 'access', _36 => _36.blocks, 'optionalAccess', _37 => _37.length]) || _optionalChain([b, 'access', _38 => _38.blocks, 'optionalAccess', _39 => _39.length])) {
@@ -803,8 +878,23 @@ function appendCell(a, 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', _42 => _42.blocks, 'optionalAccess', _43 => _43.length]) || _chunkMH6MAL64cjs.IMAGE_CELLS.has(cell);
-function mergeSplitRow(table, owner, first, colXs) {
+var hasContent = (cell) => !!cell.text.trim() || !!_optionalChain([cell, 'access', _42 => _42.blocks, 'optionalAccess', _43 => _43.length]) || _chunkWDQTFAXYcjs.IMAGE_CELLS.has(cell);
+var CLAUSE_OPEN_ENDING = /(?:하는|되는|하고|하며|하여|되어|되고|되며|이며|으며|어야|아야|여야)$/;
+var batchim = (ch) => {
+  const k = ch.charCodeAt(0) - 44032;
+  return k >= 0 && k < 11172 && k % 28 !== 0;
+};
+function clauseContinues(u, d) {
+  const tail = _nullishCoalesce(u.text.trim().split(/\s+/).pop(), () => ( "")), head = d.text.trim();
+  if (!/^[가-힣]{2,}$/.test(tail) || !/^[가-힣]/.test(head) || _chunkWDQTFAXYcjs.startsNewItem.call(void 0, u.text, head)) return false;
+  const last = tail[tail.length - 1], before = tail[tail.length - 2];
+  return last === "\uC744" && batchim(before) || last === "\uB97C" && !batchim(before) || CLAUSE_OPEN_ENDING.test(tail);
+}
+function outlineOnly(pairs, cell) {
+  const filled = pairs.filter(([, d]) => hasContent(d.cell));
+  return filled.length === 1 && outlineContinues(cell(filled[0][0]), cell(filled[0][1]));
+}
+function mergeSplitRow(table, owner, first, colXs, textOnly = false) {
   const last = first - 1;
   const pairs = [];
   let carried = false;
@@ -822,12 +912,12 @@ function mergeSplitRow(table, owner, first, colXs) {
     } else if (a.r === last && a.rs === 2) carried = true;
     c = a.c + a.cs;
   }
-  const carriedSplit = carried && pairs.filter(([u, d]) => hasContent(u.cell) && hasContent(d.cell)).length <= 1 && !pairs.some(([u, d]) => !hasContent(u.cell) && hasContent(d.cell));
+  const carriedSplit = !textOnly && carried && pairs.filter(([u, d]) => hasContent(u.cell) && hasContent(d.cell)).length <= 1 && !pairs.some(([u, d]) => !hasContent(u.cell) && hasContent(d.cell));
   const cell = (a) => table.cells[a.r][a.c];
   if (!carriedSplit) {
     const norm = (c) => c.text.replace(/\s+/g, "");
-    if (pairs.some(([u, d]) => hasContent(d.cell) && norm(cell(d)) !== norm(cell(u)) && lineEnded(cell(u), colXs[u.c], colXs[u.c + u.cs]))) return false;
-    if (!pairs.some(([u, d]) => continuesAcross(cell(u), cell(d), colXs[u.c], colXs[u.c + u.cs]))) return false;
+    if (pairs.some(([u, d]) => hasContent(d.cell) && norm(cell(d)) !== norm(cell(u)) && lineEnded(cell(u), colXs[u.c], colXs[u.c + u.cs], cell(d).text))) return false;
+    if (!outlineOnly(pairs, cell) && !pairs.some(([u, d]) => continuesAcross(cell(u), cell(d), colXs[u.c], colXs[u.c + u.cs])) && !pairs.some(([u, d]) => clauseContinues(cell(u), cell(d)))) return false;
   }
   for (const [u, d] of pairs) appendCell(table.cells[u.r][u.c], table.cells[d.r][d.c]);
   for (let r = 0; r < first; r++) for (let c = 0; c < table.cols; c++) {
@@ -874,8 +964,8 @@ function sameHeadShape(a, b) {
   }
   return tall;
 }
-function mergeCrossPageTables(blocks, pageHeights) {
-  mergeColumnFlow(blocks, pageHeights);
+function mergeCrossPageTables(blocks, pageHeights, lex) {
+  mergeColumnFlow(blocks, pageHeights, lex);
   for (let i = blocks.length - 2; i >= 0; i--) {
     const prev = blocks[i];
     if (prev.type !== "table" || !prev.table || !prev.bbox || !prev.pageNumber) continue;
@@ -884,24 +974,24 @@ function mergeCrossPageTables(blocks, pageHeights) {
     const curr = blocks[j];
     if (!curr || curr.type !== "table" || !curr.table || !curr.bbox || curr.pageNumber !== prev.pageNumber + 1) continue;
     if (startsWithUnitRow(prev.table) && startsWithUnitRow(curr.table)) continue;
-    const joined = j === i + 1 || blocks.slice(i + 1, j).every((b) => besideOwn(b, prev, curr) || insideTable(b, prev)) ? looksContinued(prev, curr, pageHeights) && !restartsTable(blocks, i, curr.table, pageHeights) ? _nullishCoalesce(joinClipParts(prev, curr, pageHeights), () => ( false)) : null : null;
+    const joined = j === i + 1 || blocks.slice(i + 1, j).every((b) => besideOwn(b, prev, curr) || insideTable(b, prev)) ? looksContinued(prev, curr, pageHeights) && !restartsTable(blocks, i, curr.table, pageHeights) ? _nullishCoalesce(joinClipParts(prev, curr, pageHeights, lex), () => ( false)) : null : null;
     if (joined) {
       placeJoined(blocks, i, j, joined);
       continue;
     }
-    if (_chunkMH6MAL64cjs.EMPTY_PARTS.has(curr.table)) {
+    if (_chunkWDQTFAXYcjs.EMPTY_PARTS.has(curr.table)) {
       blocks.splice(j, 1);
       i++;
       continue;
     }
     if (joined === null) continue;
-    if (prev.table.cols !== curr.table.cols || prev.table.renderAsTable !== curr.table.renderAsTable || _chunkMH6MAL64cjs.EMPTY_PARTS.has(prev.table)) continue;
+    if (prev.table.cols !== curr.table.cols || prev.table.renderAsTable !== curr.table.renderAsTable || _chunkWDQTFAXYcjs.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 = _chunkMH6MAL64cjs.TABLE_COLXS.get(prev.table), cx = _chunkMH6MAL64cjs.TABLE_COLXS.get(curr.table);
-    if (px && cx && !shiftedSame(px, cx, !_chunkMH6MAL64cjs.CLIP_TABLES.has(prev.table) && !_chunkMH6MAL64cjs.CLIP_TABLES.has(curr.table))) continue;
+    const px = _chunkWDQTFAXYcjs.TABLE_COLXS.get(prev.table), cx = _chunkWDQTFAXYcjs.TABLE_COLXS.get(curr.table);
+    if (px && cx && !shiftedSame(px, cx, !_chunkWDQTFAXYcjs.CLIP_TABLES.has(prev.table) && !_chunkWDQTFAXYcjs.CLIP_TABLES.has(curr.table))) continue;
     if (!rowTextsEqual(prev.table.cells[0], curr.table.cells[0]) && sameHeadShape(prev.table.cells[0], curr.table.cells[0])) continue;
     let currCells = curr.table.cells;
     if (currCells.length > 1 && prev.table.cells.length > 0 && rowTextsEqual(prev.table.cells[0], currCells[0])) {
@@ -919,20 +1009,20 @@ function mergeCrossPageTables(blocks, pageHeights) {
       caption: prev.table.caption,
       ...prev.table.renderAsTable ? { renderAsTable: true } : {}
     };
-    if (_nullishCoalesce(px, () => ( cx))) _chunkMH6MAL64cjs.TABLE_COLXS.set(merged, _nullishCoalesce(px, () => ( cx)));
-    if (_chunkMH6MAL64cjs.CLIP_TABLES.has(prev.table) || _chunkMH6MAL64cjs.CLIP_TABLES.has(curr.table)) _chunkMH6MAL64cjs.CLIP_TABLES.add(merged);
+    if (_nullishCoalesce(px, () => ( cx))) _chunkWDQTFAXYcjs.TABLE_COLXS.set(merged, _nullishCoalesce(px, () => ( cx)));
+    if (_chunkWDQTFAXYcjs.CLIP_TABLES.has(prev.table) || _chunkWDQTFAXYcjs.CLIP_TABLES.has(curr.table)) _chunkWDQTFAXYcjs.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 && _chunkMH6MAL64cjs.EMPTY_PARTS.has(t)) blocks.splice(i, 1);
+    if (blocks[i].type === "table" && t && _chunkWDQTFAXYcjs.EMPTY_PARTS.has(t)) blocks.splice(i, 1);
   }
 }
-function mergeColumnFlow(blocks, pageHeights) {
+function mergeColumnFlow(blocks, pageHeights, lex) {
   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 || !_chunkMH6MAL64cjs.CLIP_TABLES.has(L.table) || _chunkMH6MAL64cjs.EMPTY_PARTS.has(L.table)) continue;
+    if (L.type !== "table" || !L.table || !L.bbox || !L.pageNumber || !_chunkWDQTFAXYcjs.CLIP_TABLES.has(L.table) || _chunkWDQTFAXYcjs.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;
@@ -950,7 +1040,7 @@ function mergeColumnFlow(blocks, pageHeights) {
     const R = blocks[k];
     if (Math.abs(R.bbox.width - L.bbox.width) > Math.max(R.bbox.width, L.bbox.width) * NEIGHBOR_TABLE_EPSILON) continue;
     if (!looksContinued(L, R) || restartsTable(blocks, i, R.table, pageHeights)) continue;
-    const joined = joinClipParts(L, R, pageHeights);
+    const joined = joinClipParts(L, R, pageHeights, lex);
     if (!joined) continue;
     const top = L.bbox.y + L.bbox.height;
     const at = Math.min(i, k);
@@ -959,12 +1049,12 @@ function mergeColumnFlow(blocks, pageHeights) {
     i = at - 1;
   }
 }
-function joinClipParts(prev, curr, pageHeights) {
+function joinClipParts(prev, curr, pageHeights, lex) {
   const pt = prev.table, ct = curr.table;
-  if (!_chunkMH6MAL64cjs.CLIP_TABLES.has(pt) || !_chunkMH6MAL64cjs.CLIP_TABLES.has(ct)) return null;
-  const px = _chunkMH6MAL64cjs.TABLE_COLXS.get(pt), cx = _chunkMH6MAL64cjs.TABLE_COLXS.get(ct);
+  if (!_chunkWDQTFAXYcjs.CLIP_TABLES.has(pt) || !_chunkWDQTFAXYcjs.CLIP_TABLES.has(ct)) return null;
+  const px = _chunkWDQTFAXYcjs.TABLE_COLXS.get(pt), cx = _chunkWDQTFAXYcjs.TABLE_COLXS.get(ct);
   if (!px || !cx) return null;
-  let xs = cx, shifted = false, dx = 0;
+  let xs = cx, shifted = false, dx = 0, foreign = false;
   if (!shiftedSame(px, cx, false)) {
     if (!_optionalChain([pageHeights, 'optionalAccess', _54 => _54.get, 'call', _55 => _55(prev.pageNumber)]) || !_optionalChain([pageHeights, 'optionalAccess', _56 => _56.get, 'call', _57 => _57(curr.pageNumber)])) return null;
     dx = px[px.length - 1] - cx[cx.length - 1];
@@ -973,22 +1063,32 @@ function joinClipParts(prev, curr, pageHeights) {
       shifted = true;
     } else dx = 0;
     if (!px.some((x) => Math.abs(x - xs[0]) <= CONTINUATION_COL_TOL)) return null;
+    const inner = xs.slice(1, -1);
+    foreign = inner.length > 0 && !inner.some((x) => px.some((p) => Math.abs(p - x) <= CONTINUATION_COL_TOL)) && headingRow(_nullishCoalesce(ct.cells[0], () => ( [])));
   }
-  const res = joinSplitParts(pt, px, ct, xs, dx, _optionalChain([prev, 'access', _58 => _58.bbox, 'optionalAccess', _59 => _59.y]));
-  if (!res || shifted && !res.split && !res.header) return null;
-  _chunkMH6MAL64cjs.TABLE_COLXS.set(res.table, res.colXs);
-  _chunkMH6MAL64cjs.CLIP_TABLES.add(res.table);
+  const res = joinSplitParts(pt, px, ct, xs, dx, _optionalChain([prev, 'access', _58 => _58.bbox, 'optionalAccess', _59 => _59.y]), lex);
+  if (!res || (shifted || foreign) && !res.split && !res.header) return null;
+  _chunkWDQTFAXYcjs.TABLE_COLXS.set(res.table, res.colXs);
+  _chunkWDQTFAXYcjs.CLIP_TABLES.add(res.table);
   return res.table;
 }
+function headingRow(row) {
+  const heads = row.map((c) => c.text.replace(/\s+/g, "")).filter(Boolean);
+  if (heads.length === 1) return heads[0].length <= 8 || /^[①-⑳❶-❿]/.test(heads[0]);
+  return heads.length === 2 && /^(?:\d{1,2}|[ⅠⅡⅢⅣⅤⅥⅦⅧⅨⅩ]+|[①-⑳❶-❿])$/.test(heads[0]);
+}
 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', _60 => _60.get, 'call', _61 => _61(prev.pageNumber)]), ch = _optionalChain([pageHeights, 'optionalAccess', _62 => _62.get, 'call', _63 => _63(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', _64 => _64.table, 'access', _65 => _65.cells, 'access', _66 => _66[0], 'optionalAccess', _67 => _67.find, 'call', _68 => _68((c) => c.text.trim()), 'optionalAccess', _69 => _69.text]), () => ( ""));
-  return !ANNEX_HEAD_RE.test(firstText);
+  const firstRow = _nullishCoalesce(curr.table.cells[0], () => ( []));
+  const firstText = _nullishCoalesce(_optionalChain([firstRow, 'access', _64 => _64.find, 'call', _65 => _65((c) => c.text.trim()), 'optionalAccess', _66 => _66.text]), () => ( ""));
+  if (!ANNEX_HEAD_RE.test(firstText)) return true;
+  const filled = firstRow.map((c) => c.text.replace(/\s+/g, "")).filter(Boolean);
+  return filled.length >= 2 && filled.every((t) => t === filled[0]);
 }
 function firstRowSig(t) {
-  const x = _chunkMH6MAL64cjs.TABLE_COLXS.get(t);
+  const x = _chunkWDQTFAXYcjs.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, "") }));
 }
@@ -998,6 +1098,7 @@ function restartsTable(blocks, i, curr, pageHeights) {
   const head = blocks[chainHead(blocks, i, pageHeights)].table;
   if (cs.length === 1 && restartsTitledForm(head, curr)) return true;
   if (cs.length < RESTART_MIN_ANCHORS) return false;
+  if (CONTACT_HEAD.test(cs[0].t) && cs.some((c) => CONTACT_ROLE.test(c.t))) return false;
   const hs = firstRowSig(head);
   if (!hs || hs.length !== cs.length) return false;
   const dx = hs[hs.length - 1].x2 - cs[cs.length - 1].x2;
@@ -1012,9 +1113,9 @@ function restartsTable(blocks, i, curr, pageHeights) {
 }
 function restartsTitledForm(head, curr) {
   if (head.rows < 2 || curr.rows < 2 || head.cols !== curr.cols) return false;
-  const full = (t) => _optionalChain([t, 'access', _70 => _70.cells, 'access', _71 => _71[0], 'access', _72 => _72[0], 'optionalAccess', _73 => _73.colSpan]) === t.cols;
+  const full = (t) => _optionalChain([t, 'access', _67 => _67.cells, 'access', _68 => _68[0], 'access', _69 => _69[0], 'optionalAccess', _70 => _70.colSpan]) === t.cols;
   if (!full(head) || !full(curr)) return false;
-  const norm = (c) => (_nullishCoalesce(_optionalChain([c, 'optionalAccess', _74 => _74.text]), () => ( ""))).replace(/\s+/g, "");
+  const norm = (c) => (_nullishCoalesce(_optionalChain([c, 'optionalAccess', _71 => _71.text]), () => ( ""))).replace(/\s+/g, "");
   const ht = norm(head.cells[0][0]), ct = norm(curr.cells[0][0]);
   if (!ht || !ct || ht === ct) return false;
   const titleLike = (t) => t.length <= TITLE_ROW_MAX_CHARS && !/^[ㅇ○◦•·\-–※*□■▪☞]/.test(t);
@@ -1040,7 +1141,7 @@ 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 = _chunkMH6MAL64cjs.TABLE_COLXS.get(pb.table), cx = _chunkMH6MAL64cjs.TABLE_COLXS.get(cur.table);
+    const px = _chunkWDQTFAXYcjs.TABLE_COLXS.get(pb.table), cx = _chunkWDQTFAXYcjs.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;
@@ -1071,12 +1172,12 @@ var RESTART_LOOKBACK_PAGES = 5;
 var COL_MATCH_TOL = 0.5;
 var CONTINUED_TABLE_PAGE_BAND = 0.16;
 function lastRowCell(t, x1, x2) {
-  const xs = _chunkMH6MAL64cjs.TABLE_COLXS.get(t);
+  const xs = _chunkWDQTFAXYcjs.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', _75 => _75.cells, 'access', _76 => _76[r], 'optionalAccess', _77 => _77[c]]);
+      const cell = _optionalChain([t, 'access', _72 => _72.cells, 'access', _73 => _73[r], 'optionalAccess', _74 => _74[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;
@@ -1089,19 +1190,19 @@ 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 ? _chunkMH6MAL64cjs.CONT_PARTS.get(part.table) : void 0;
+    const from = part.table ? _chunkWDQTFAXYcjs.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', _78 => _78.table, 'access', _79 => _79.cells, 'access', _80 => _80[0], 'optionalAccess', _81 => _81[0]]);
+    const add = _optionalChain([part, 'access', _75 => _75.table, 'access', _76 => _76.cells, 'access', _77 => _77[0], 'optionalAccess', _78 => _78[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', _82 => _82.blocks, 'optionalAccess', _83 => _83.length])) _chunkMH6MAL64cjs.EMPTY_PARTS.delete(prev.table);
+    if (cell.text.trim() || _optionalChain([cell, 'access', _79 => _79.blocks, 'optionalAccess', _80 => _80.length])) _chunkWDQTFAXYcjs.EMPTY_PARTS.delete(prev.table);
     blocks.splice(j, 1);
     if (cell.blocks && at > 0) {
       const nextAt = cell.blocks[at];
@@ -1113,24 +1214,24 @@ function mergeContinuedCells(blocks, pageHeights) {
 }
 function unwrapContinuedTable(blocks, at, pageHeights) {
   const prev = blocks[at - 1], frame = blocks[at];
-  if (!_optionalChain([prev, 'optionalAccess', _84 => _84.table]) || !_optionalChain([frame, 'optionalAccess', _85 => _85.table]) || frame.table.rows !== 1 || frame.table.cols !== 1 || !_chunkMH6MAL64cjs.CLIP_TABLES.has(frame.table)) return;
-  const inner = _optionalChain([frame, 'access', _86 => _86.table, 'access', _87 => _87.cells, 'access', _88 => _88[0], 'optionalAccess', _89 => _89[0], 'optionalAccess', _90 => _90.blocks]);
-  const next = _optionalChain([inner, 'optionalAccess', _91 => _91[0]]);
-  if (!inner || !_optionalChain([next, 'optionalAccess', _92 => _92.table]) || !prev.bbox || !next.bbox || !prev.pageNumber || next.pageNumber !== prev.pageNumber + 1 || !_chunkMH6MAL64cjs.CLIP_TABLES.has(prev.table) || !_chunkMH6MAL64cjs.CLIP_TABLES.has(next.table) || prev.table.cols !== next.table.cols) return;
-  const px = _chunkMH6MAL64cjs.TABLE_COLXS.get(prev.table), nx = _chunkMH6MAL64cjs.TABLE_COLXS.get(next.table);
+  if (!_optionalChain([prev, 'optionalAccess', _81 => _81.table]) || !_optionalChain([frame, 'optionalAccess', _82 => _82.table]) || frame.table.rows !== 1 || frame.table.cols !== 1 || !_chunkWDQTFAXYcjs.CLIP_TABLES.has(frame.table)) return;
+  const inner = _optionalChain([frame, 'access', _83 => _83.table, 'access', _84 => _84.cells, 'access', _85 => _85[0], 'optionalAccess', _86 => _86[0], 'optionalAccess', _87 => _87.blocks]);
+  const next = _optionalChain([inner, 'optionalAccess', _88 => _88[0]]);
+  if (!inner || !_optionalChain([next, 'optionalAccess', _89 => _89.table]) || !prev.bbox || !next.bbox || !prev.pageNumber || next.pageNumber !== prev.pageNumber + 1 || !_chunkWDQTFAXYcjs.CLIP_TABLES.has(prev.table) || !_chunkWDQTFAXYcjs.CLIP_TABLES.has(next.table) || prev.table.cols !== next.table.cols) return;
+  const px = _chunkWDQTFAXYcjs.TABLE_COLXS.get(prev.table), nx = _chunkWDQTFAXYcjs.TABLE_COLXS.get(next.table);
   if (!px || !nx || px.length !== nx.length || px.some((x, i) => Math.abs(x - nx[i]) > COL_MATCH_TOL)) return;
-  const ph = _optionalChain([pageHeights, 'optionalAccess', _93 => _93.get, 'call', _94 => _94(prev.pageNumber)]), nh = _optionalChain([pageHeights, 'optionalAccess', _95 => _95.get, 'call', _96 => _96(next.pageNumber)]);
+  const ph = _optionalChain([pageHeights, 'optionalAccess', _90 => _90.get, 'call', _91 => _91(prev.pageNumber)]), nh = _optionalChain([pageHeights, 'optionalAccess', _92 => _92.get, 'call', _93 => _93(next.pageNumber)]);
   if (!ph || !nh || prev.bbox.y > ph * CONTINUED_TABLE_PAGE_BAND || next.bbox.y + next.bbox.height < nh * (1 - CONTINUED_TABLE_PAGE_BAND)) return;
   blocks.splice(at, 1, ...inner);
 }
 function fillNestedContinuation(blocks, at) {
   const prev = blocks[at - 1], next = blocks[at];
-  const t = _optionalChain([prev, 'optionalAccess', _97 => _97.table]), u = _optionalChain([next, 'optionalAccess', _98 => _98.table]);
-  if (!t || !u || u.rows !== 1 || u.cols !== 1 || !_chunkMH6MAL64cjs.CLIP_TABLES.has(t) || !_chunkMH6MAL64cjs.CLIP_TABLES.has(u) || next.pageNumber !== (_nullishCoalesce(prev.pageNumber, () => ( 0))) + 1) return;
-  const xs = _chunkMH6MAL64cjs.TABLE_COLXS.get(u);
+  const t = _optionalChain([prev, 'optionalAccess', _94 => _94.table]), u = _optionalChain([next, 'optionalAccess', _95 => _95.table]);
+  if (!t || !u || u.rows !== 1 || u.cols !== 1 || !_chunkWDQTFAXYcjs.CLIP_TABLES.has(t) || !_chunkWDQTFAXYcjs.CLIP_TABLES.has(u) || next.pageNumber !== (_nullishCoalesce(prev.pageNumber, () => ( 0))) + 1) return;
+  const xs = _chunkWDQTFAXYcjs.TABLE_COLXS.get(u);
   const cell = xs ? lastRowCell(t, xs[0], xs[xs.length - 1]) : void 0;
-  const add = _optionalChain([u, 'access', _99 => _99.cells, 'access', _100 => _100[0], 'optionalAccess', _101 => _101[0]]);
-  if (!cell || !add || cell.text.trim() || _optionalChain([cell, 'access', _102 => _102.blocks, 'optionalAccess', _103 => _103.length])) return;
+  const add = _optionalChain([u, 'access', _96 => _96.cells, 'access', _97 => _97[0], 'optionalAccess', _98 => _98[0]]);
+  if (!cell || !add || cell.text.trim() || _optionalChain([cell, 'access', _99 => _99.blocks, 'optionalAccess', _100 => _100.length])) return;
   cell.text = add.text;
   if (add.blocks) cell.blocks = add.blocks;
   blocks.splice(at, 1);
@@ -1168,7 +1269,7 @@ var WINANSI_REVERSE = {
 };
 function symbolFontTable(fontName) {
   if (!fontName) return void 0;
-  if (/wingdings(?![\s-]*[23])/i.test(fontName)) return _chunkDRY3L5KQcjs.WINGDINGS;
+  if (/wingdings(?![\s-]*[23])/i.test(fontName)) return _chunkXTFN5WKScjs.WINGDINGS;
   return void 0;
 }
 function remapSymbolText(text, table) {
@@ -1296,7 +1397,7 @@ function layout(toks) {
     const bar = rest.find((t) => t.bar && t.x >= rt.x && t.x <= rt.x + rt.w + 2 && Math.abs(t.y - rt.y) <= size * 1.5);
     if (!bar) continue;
     const inner = rest.filter((t) => !t.bar && t !== rt && t.x + t.w / 2 > bar.x && t.x + t.w / 2 < bar.x + bar.w && bar.y - t.y > 0 && bar.y - t.y <= size * 1.4);
-    const tok = { text: `\\sqrt{${linear(inner)}}`, x: rt.x, w: bar.x + bar.w - rt.x, y: _nullishCoalesce(_optionalChain([inner, 'access', _104 => _104[0], 'optionalAccess', _105 => _105.y]), () => ( rt.y)), size: Math.max(rt.size, ...inner.map((t) => t.size)), bar: false };
+    const tok = { text: `\\sqrt{${linear(inner)}}`, x: rt.x, w: bar.x + bar.w - rt.x, y: _nullishCoalesce(_optionalChain([inner, 'access', _101 => _101[0], 'optionalAccess', _102 => _102.y]), () => ( rt.y)), size: Math.max(rt.size, ...inner.map((t) => t.size)), bar: false };
     rest = [...rest.filter((t) => t !== rt && t !== bar && !inner.includes(t)), tok];
   }
   const bars = rest.filter((t) => t.bar).sort((a, b) => a.w - b.w);
@@ -1398,7 +1499,7 @@ function wrapEquationRuns(items, faceName) {
     for (const i of mains) ys.set(i.y, (_nullishCoalesce(ys.get(i.y), () => ( 0))) + 1);
     const top = [...ys].sort((p, q) => q[1] - p[1]);
     const bar = cur.find(isBarItem);
-    const lineY = bar && top.length > 1 && top[0][1] === top[1][1] ? Math.round(bar.y + size * 0.3) : _nullishCoalesce(_optionalChain([top, 'access', _106 => _106[0], 'optionalAccess', _107 => _107[0]]), () => ( cur[0].y));
+    const lineY = bar && top.length > 1 && top[0][1] === top[1][1] ? Math.round(bar.y + size * 0.3) : _nullishCoalesce(_optionalChain([top, 'access', _103 => _103[0], 'optionalAccess', _104 => _104[0]]), () => ( cur[0].y));
     const base = { ..._nullishCoalesce(mains[0], () => ( cur[0])), y: lineY, fontSize: size || cur[0].fontSize };
     const x1 = Math.min(...cur.map((i) => i.x)), x2 = Math.max(...cur.map((i) => i.x + i.w));
     added.push({ ...base, text: `$${tex}$`, x: x1, w: x2 - x1, hasSpaceBefore: true });
@@ -1744,7 +1845,7 @@ function joinVerticalColumns(items) {
   const groups = [];
   for (const col of columns) {
     const g2 = groups[groups.length - 1];
-    const last = _optionalChain([g2, 'optionalAccess', _108 => _108[g2.length - 1]]);
+    const last = _optionalChain([g2, 'optionalAccess', _105 => _105[g2.length - 1]]);
     const fs = col.items[0].fontSize;
     if (last && last.x - col.x >= fs * 1.2 && last.x - col.x <= fs * 4 && Math.abs(last.items[0].y - col.items[0].y) <= fs * 1.2) g2.push(col);
     else groups.push([col]);
@@ -1844,6 +1945,52 @@ function restoreTrackedSpacing(items, fnArray, argsArray) {
   }
   return changed;
 }
+function markSyntheticSpaces(items, fnArray, argsArray) {
+  if (!items.some((it) => typeof it.str === "string" && it.str.length > 0 && !it.str.trim())) return;
+  const streams = /* @__PURE__ */ new Map();
+  const saved = [];
+  let font = "";
+  for (let i = 0; i < fnArray.length; i++) {
+    const fn = fnArray[i];
+    const args = argsArray[i];
+    if (fn === _pdfmjs.OPS.setFont) font = String(args[0]);
+    else if (fn === _pdfmjs.OPS.save || fn === _pdfmjs.OPS.paintFormXObjectBegin) saved.push(font);
+    else if (fn === _pdfmjs.OPS.restore || fn === _pdfmjs.OPS.paintFormXObjectEnd) font = _nullishCoalesce(saved.pop(), () => ( font));
+    else if (fn === _pdfmjs.OPS.showText) {
+      let list = streams.get(font);
+      if (!list) streams.set(font, list = []);
+      for (const g2 of args[0]) {
+        if (g2 && typeof g2 === "object" && typeof g2.unicode === "string") list.push(_pdfmjs.normalizeUnicode.call(void 0, g2.unicode));
+      }
+    }
+  }
+  const cursor = /* @__PURE__ */ new Map();
+  const failed = /* @__PURE__ */ new Set();
+  const marks = [];
+  for (const it of items) {
+    const f = _nullishCoalesce(it.fontName, () => ( ""));
+    const glyphs = streams.get(f);
+    if (!glyphs || failed.has(f) || typeof it.str !== "string" || !it.str) continue;
+    let gi = _nullishCoalesce(cursor.get(f), () => ( 0));
+    if (!it.str.trim()) {
+      if (gi < glyphs.length && /^\s+$/.test(glyphs[gi])) {
+        while (gi < glyphs.length && /^\s+$/.test(glyphs[gi])) gi++;
+        cursor.set(f, gi);
+      } else marks.push(it);
+      continue;
+    }
+    const want = it.str.replace(/\s+/g, "");
+    while (gi < glyphs.length && /^\s*$/.test(glyphs[gi])) gi++;
+    let got = "";
+    while (gi < glyphs.length && got.length < want.length) got += glyphs[gi++].replace(/\s+/g, "");
+    if (got !== want) {
+      failed.add(f);
+      continue;
+    }
+    cursor.set(f, gi);
+  }
+  for (const it of marks) if (!failed.has(_nullishCoalesce(it.fontName, () => ( "")))) it.synthetic = true;
+}
 
 // src/pdf/endnotes.ts
 var NOTE_MARK = /^\s*(문\d{1,5}[）)]|\d{1,5}\)|\(\d{1,5}\))/;
@@ -1884,7 +2031,7 @@ function relocateEndnotes(blocks) {
   }
   const shapes = /* @__PURE__ */ new Map();
   for (const h of heads) shapes.set(h.shape, (_nullishCoalesce(shapes.get(h.shape), () => ( 0))) + 1);
-  const main = _optionalChain([[...shapes], 'access', _109 => _109.sort, 'call', _110 => _110((a, b) => b[1] - a[1]), 'access', _111 => _111[0], 'optionalAccess', _112 => _112[0]]);
+  const main = _optionalChain([[...shapes], 'access', _106 => _106.sort, 'call', _107 => _107((a, b) => b[1] - a[1]), 'access', _108 => _108[0], 'optionalAccess', _109 => _109[0]]);
   if (!main || !/#/.test(main)) return blocks;
   const lastRun = (hs) => {
     let run2 = [];
@@ -1981,10 +2128,11 @@ function orderTwoUpPage(blocks, pageWidth, pageHeight) {
 var BAND = 0.12;
 var MAX_CHARS = 10;
 var MIN_REPEAT = 3;
+var RUN_GAP = 5;
 function removeSideTabs(blocks, pageWidths) {
   const edges = /* @__PURE__ */ new Map();
   for (const b of blocks) {
-    if (!b.bbox || b.pageNumber === void 0 || [..._nullishCoalesce(_optionalChain([b, 'access', _113 => _113.text, 'optionalAccess', _114 => _114.trim, 'call', _115 => _115()]), () => ( ""))].length <= MAX_CHARS) continue;
+    if (!b.bbox || b.pageNumber === void 0 || [..._nullishCoalesce(_optionalChain([b, 'access', _110 => _110.text, 'optionalAccess', _111 => _111.trim, 'call', _112 => _112()]), () => ( ""))].length <= MAX_CHARS) continue;
     const e = _nullishCoalesce(edges.get(b.pageNumber), () => ( { lefts: [], rights: [] }));
     e.lefts.push(b.bbox.x);
     e.rights.push(b.bbox.x + b.bbox.width);
@@ -2000,7 +2148,7 @@ function removeSideTabs(blocks, pageWidths) {
   const fallback = all.length ? { left: q(all.map((e) => e.left), 0.5), right: q(all.map((e) => e.right), 0.5) } : void 0;
   const key = (b) => {
     const w = b.pageNumber !== void 0 ? pageWidths.get(b.pageNumber) : void 0;
-    const text = _optionalChain([b, 'access', _116 => _116.text, 'optionalAccess', _117 => _117.trim, 'call', _118 => _118()]);
+    const text = _optionalChain([b, 'access', _113 => _113.text, 'optionalAccess', _114 => _114.trim, 'call', _115 => _115()]);
     const area = _nullishCoalesce((b.pageNumber !== void 0 ? body.get(b.pageNumber) : void 0), () => ( fallback));
     if (!w || !area || !b.bbox || !text || b.type === "table" || [...text].length > MAX_CHARS) return null;
     const side = b.bbox.x + b.bbox.width <= Math.min(w * BAND, area.left - 1) ? "L" : b.bbox.x >= Math.max(w * (1 - BAND), area.right + 1) ? "R" : null;
@@ -2013,13 +2161,21 @@ function removeSideTabs(blocks, pageWidths) {
   }
   const running = /* @__PURE__ */ new Set();
   for (const [k, ps] of pages) {
-    const list = [...ps];
-    const span = Math.max(...list) - Math.min(...list) + 1;
-    if (list.length >= MIN_REPEAT && list.length >= span * 0.4) running.add(k);
+    const list = [...ps].sort((a, b) => a - b);
+    let run = [];
+    const flush = () => {
+      if (run.length >= MIN_REPEAT && run.length >= (run[run.length - 1] - run[0] + 1) * 0.4) for (const p of run) running.add(`${p}\0${k}`);
+      run = [];
+    };
+    for (const p of list) {
+      if (run.length && p - run[run.length - 1] > RUN_GAP) flush();
+      run.push(p);
+    }
+    flush();
   }
   return running.size ? blocks.filter((b) => {
     const k = key(b);
-    return !k || !running.has(k);
+    return !k || !running.has(`${b.pageNumber}\0${k}`);
   }) : blocks;
 }
 
@@ -2068,7 +2224,7 @@ function inlineFootnotes(blocks, pages) {
         open = true;
         continue;
       }
-      const prev = pb[j - 1], line = (_nullishCoalesce(_nullishCoalesce(_optionalChain([b, 'access', _119 => _119.style, 'optionalAccess', _120 => _120.fontSize]), () => ( _optionalChain([b, 'access', _121 => _121.bbox, 'optionalAccess', _122 => _122.height]))), () => ( 10))) * 1.5;
+      const prev = pb[j - 1], line = (_nullishCoalesce(_nullishCoalesce(_optionalChain([b, 'access', _116 => _116.style, 'optionalAccess', _117 => _117.fontSize]), () => ( _optionalChain([b, 'access', _118 => _118.bbox, 'optionalAccess', _119 => _119.height]))), () => ( 10))) * 1.5;
       if (open && b.bbox && prev.bbox && prev.bbox.y - top(b) <= line && !/^[-–—\s\d]+$/.test(b.text)) {
         notes[notes.length - 1].text += " " + b.text.trim();
         notes[notes.length - 1].blocks.push(b);
@@ -2091,7 +2247,7 @@ var CAPTION = /^(?:Table|Figure|Fig\.?)\s*\d+(?:\.\d+)*\s*[.:]/i;
 var EQUATION_NUMBER = /\t\(\d{1,3}[a-z]?\)\s*$/;
 var DISPLAY_MATH = /=|[√∑∏∫∂∇≤≥≈≠∈∀∃]/;
 function isRunningHead(block, page, pageHeight) {
-  const box = block.bbox, text = _optionalChain([block, 'access', _123 => _123.text, 'optionalAccess', _124 => _124.trim, 'call', _125 => _125()]);
+  const box = block.bbox, text = _optionalChain([block, 'access', _120 => _120.text, 'optionalAccess', _121 => _121.trim, 'call', _122 => _122()]);
   if (!box || !text || !pageHeight) return false;
   const top = box.y + box.height >= pageHeight * 0.9;
   const bottom = box.y <= pageHeight * 0.1;
@@ -2099,7 +2255,7 @@ function isRunningHead(block, page, pageHeight) {
   const others = page.filter((o) => o !== block && o.bbox && o.type !== "image" && o.type !== "separator");
   if (top && others.some((o) => o.bbox.y + o.bbox.height > box.y + box.height)) return false;
   if (bottom && others.some((o) => o.bbox.y < box.y)) return false;
-  if (bottom && box.height <= (_nullishCoalesce(_optionalChain([block, 'access', _126 => _126.style, 'optionalAccess', _127 => _127.fontSize]), () => ( 0))) * 1.6) return true;
+  if (bottom && box.height <= (_nullishCoalesce(_optionalChain([block, 'access', _123 => _123.style, 'optionalAccess', _124 => _124.fontSize]), () => ( 0))) * 1.6) return true;
   if (!text.includes("	")) return false;
   const parts = text.split(/\t+/).map((part) => part.trim()).filter(Boolean);
   if (parts.length >= 2 && (PAGE_NUMBER.test(parts[0]) || PAGE_NUMBER.test(parts[parts.length - 1]))) return true;
@@ -2108,7 +2264,7 @@ function isRunningHead(block, page, pageHeight) {
   return right > left && box.width >= (right - left) * 0.6;
 }
 function unbalancedClose(text) {
-  return (_nullishCoalesce(_optionalChain([text, 'access', _128 => _128.match, 'call', _129 => _129(/\)/g), 'optionalAccess', _130 => _130.length]), () => ( 0))) > (_nullishCoalesce(_optionalChain([text, 'access', _131 => _131.match, 'call', _132 => _132(/\(/g), 'optionalAccess', _133 => _133.length]), () => ( 0)));
+  return (_nullishCoalesce(_optionalChain([text, 'access', _125 => _125.match, 'call', _126 => _126(/\)/g), 'optionalAccess', _127 => _127.length]), () => ( 0))) > (_nullishCoalesce(_optionalChain([text, 'access', _128 => _128.match, 'call', _129 => _129(/\(/g), 'optionalAccess', _130 => _130.length]), () => ( 0)));
 }
 function demoteNonHeadingRoles(blocks, pageHeights, faceNames) {
   const byPage = /* @__PURE__ */ new Map();
@@ -2121,7 +2277,7 @@ function demoteNonHeadingRoles(blocks, pageHeights, faceNames) {
   for (const [pageNumber, page] of byPage) {
     const chars = /* @__PURE__ */ new Map();
     for (const b of page) {
-      if (b.type !== "paragraph" && b.type !== "heading" || !b.text || !_optionalChain([b, 'access', _134 => _134.style, 'optionalAccess', _135 => _135.fontName]) || !b.style.fontSize) continue;
+      if (b.type !== "paragraph" && b.type !== "heading" || !b.text || !_optionalChain([b, 'access', _131 => _131.style, 'optionalAccess', _132 => _132.fontName]) || !b.style.fontSize) continue;
       const key2 = `${b.style.fontName}:${b.style.fontSize}`;
       chars.set(key2, (_nullishCoalesce(chars.get(key2), () => ( 0))) + b.text.length);
     }
@@ -2132,48 +2288,48 @@ function demoteNonHeadingRoles(blocks, pageHeights, faceNames) {
     const block = blocks[i];
     if (block.type !== "heading" || !block.text) continue;
     const text = block.text.replace(/<[^>]+>/g, "").trim();
-    const fsz = _nullishCoalesce(_optionalChain([block, 'access', _136 => _136.style, 'optionalAccess', _137 => _137.fontSize]), () => ( 0));
-    const tocEntry = i > 0 && _chunkMH6MAL64cjs.TOC_BLOCKS.has(blocks[i - 1]) && _chunkMH6MAL64cjs.TOC_BLOCKS.has(blocks[i + 1]);
+    const fsz = _nullishCoalesce(_optionalChain([block, 'access', _133 => _133.style, 'optionalAccess', _134 => _134.fontSize]), () => ( 0));
+    const tocEntry = i > 0 && _chunkWDQTFAXYcjs.TOC_BLOCKS.has(blocks[i - 1]) && _chunkWDQTFAXYcjs.TOC_BLOCKS.has(blocks[i + 1]);
     const page = _nullishCoalesce(byPage.get(_nullishCoalesce(block.pageNumber, () => ( 0))), () => ( []));
-    const proseStyle = _optionalChain([block, 'access', _138 => _138.style, 'optionalAccess', _139 => _139.fontName]) && bodyStyle.get(_nullishCoalesce(block.pageNumber, () => ( 0))) === `${block.style.fontName}:${block.style.fontSize}` && text.length > 60;
+    const proseStyle = _optionalChain([block, 'access', _135 => _135.style, 'optionalAccess', _136 => _136.fontName]) && bodyStyle.get(_nullishCoalesce(block.pageNumber, () => ( 0))) === `${block.style.fontName}:${block.style.fontSize}` && text.length > 60;
     const next = blocks[i + 1];
-    const nb = _optionalChain([next, 'optionalAccess', _140 => _140.bbox]), bb = block.bbox;
-    const chapterNumber = !!nb && !!bb && fsz >= (_nullishCoalesce(_optionalChain([next, 'optionalAccess', _141 => _141.style, 'optionalAccess', _142 => _142.fontSize]), () => ( 0))) * 1.2 && Math.abs(bb.x - nb.x) > fsz && Math.abs(bb.x + bb.width / 2 - (nb.x + nb.width / 2)) < nb.width * 0.1;
-    if (/^\d+(?:\.\d+)*\.?$/.test(text) && _optionalChain([next, 'optionalAccess', _143 => _143.type]) === "heading" && next.text && next.pageNumber === block.pageNumber && !chapterNumber) {
+    const nb = _optionalChain([next, 'optionalAccess', _137 => _137.bbox]), bb = block.bbox;
+    const chapterNumber = !!nb && !!bb && fsz >= (_nullishCoalesce(_optionalChain([next, 'optionalAccess', _138 => _138.style, 'optionalAccess', _139 => _139.fontSize]), () => ( 0))) * 1.2 && Math.abs(bb.x - nb.x) > fsz && Math.abs(bb.x + bb.width / 2 - (nb.x + nb.width / 2)) < nb.width * 0.1;
+    if (/^\d+(?:\.\d+)*\.?$/.test(text) && _optionalChain([next, 'optionalAccess', _140 => _140.type]) === "heading" && next.text && next.pageNumber === block.pageNumber && !chapterNumber) {
       next.text = `${block.text.trim()} ${next.text.trim()}`;
       blocks.splice(i--, 1);
       continue;
     }
-    if ((_optionalChain([next, 'optionalAccess', _144 => _144.type]) === "paragraph" || _optionalChain([next, 'optionalAccess', _145 => _145.type]) === "heading" && !CAPTION.test(text)) && next.pageNumber === block.pageNumber && next.text && next.bbox && block.bbox && _optionalChain([next, 'access', _146 => _146.style, 'optionalAccess', _147 => _147.fontName]) === _optionalChain([block, 'access', _148 => _148.style, 'optionalAccess', _149 => _149.fontName]) && _optionalChain([next, 'access', _150 => _150.style, 'optionalAccess', _151 => _151.fontSize]) === _optionalChain([block, 'access', _152 => _152.style, 'optionalAccess', _153 => _153.fontSize]) && fsz > 0 && /^[a-z]/.test(next.text.trim()) && next.text.trim().length <= 40 && !/[.:;!?]$/.test(next.text.trim()) && block.bbox.y - (next.bbox.y + next.bbox.height) < fsz * 1.2 && Math.abs(next.bbox.x - block.bbox.x) < fsz) {
+    if ((_optionalChain([next, 'optionalAccess', _141 => _141.type]) === "paragraph" || _optionalChain([next, 'optionalAccess', _142 => _142.type]) === "heading" && !CAPTION.test(text)) && next.pageNumber === block.pageNumber && next.text && next.bbox && block.bbox && _optionalChain([next, 'access', _143 => _143.style, 'optionalAccess', _144 => _144.fontName]) === _optionalChain([block, 'access', _145 => _145.style, 'optionalAccess', _146 => _146.fontName]) && _optionalChain([next, 'access', _147 => _147.style, 'optionalAccess', _148 => _148.fontSize]) === _optionalChain([block, 'access', _149 => _149.style, 'optionalAccess', _150 => _150.fontSize]) && fsz > 0 && /^[a-z]/.test(next.text.trim()) && next.text.trim().length <= 40 && !/[.:;!?]$/.test(next.text.trim()) && block.bbox.y - (next.bbox.y + next.bbox.height) < fsz * 1.2 && Math.abs(next.bbox.x - block.bbox.x) < fsz) {
       block.text = `${block.text.trim()} ${next.text.trim()}`;
       block.bbox = { ...block.bbox, y: next.bbox.y, height: block.bbox.y + block.bbox.height - next.bbox.y };
       blocks.splice(i + 1, 1);
     }
     const after = blocks[i + 2];
-    if ((_optionalChain([next, 'optionalAccess', _154 => _154.type]) === "paragraph" || _optionalChain([next, 'optionalAccess', _155 => _155.type]) === "heading") && /^(?:&|\+|and|or)$/i.test(_nullishCoalesce(_optionalChain([next, 'access', _156 => _156.text, 'optionalAccess', _157 => _157.trim, 'call', _158 => _158()]), () => ( ""))) && _optionalChain([after, 'optionalAccess', _159 => _159.type]) === "heading" && after.text && after.pageNumber === block.pageNumber && next.pageNumber === block.pageNumber && _optionalChain([after, 'access', _160 => _160.style, 'optionalAccess', _161 => _161.fontSize]) === _optionalChain([block, 'access', _162 => _162.style, 'optionalAccess', _163 => _163.fontSize])) {
+    if ((_optionalChain([next, 'optionalAccess', _151 => _151.type]) === "paragraph" || _optionalChain([next, 'optionalAccess', _152 => _152.type]) === "heading") && /^(?:&|\+|and|or)$/i.test(_nullishCoalesce(_optionalChain([next, 'access', _153 => _153.text, 'optionalAccess', _154 => _154.trim, 'call', _155 => _155()]), () => ( ""))) && _optionalChain([after, 'optionalAccess', _156 => _156.type]) === "heading" && after.text && after.pageNumber === block.pageNumber && next.pageNumber === block.pageNumber && _optionalChain([after, 'access', _157 => _157.style, 'optionalAccess', _158 => _158.fontSize]) === _optionalChain([block, 'access', _159 => _159.style, 'optionalAccess', _160 => _160.fontSize])) {
       block.text = `${block.text.trim()} ${next.text.trim()} ${after.text.trim()}`;
       blocks.splice(i + 1, 2);
     }
     const tiny = fsz > 0 && fsz < 7.5;
     const prev = blocks[i - 1];
-    if (_optionalChain([prev, 'optionalAccess', _164 => _164.type]) === "paragraph" && prev.pageNumber === block.pageNumber && /^[a-z]{1,3}[.)]$/.test(_nullishCoalesce(_optionalChain([prev, 'access', _165 => _165.text, 'optionalAccess', _166 => _166.trim, 'call', _167 => _167()]), () => ( ""))) && prev.bbox && block.bbox && Math.abs(prev.bbox.y - block.bbox.y) < 2 && prev.bbox.x + prev.bbox.width <= block.bbox.x) {
+    if (_optionalChain([prev, 'optionalAccess', _161 => _161.type]) === "paragraph" && prev.pageNumber === block.pageNumber && /^[a-z]{1,3}[.)]$/.test(_nullishCoalesce(_optionalChain([prev, 'access', _162 => _162.text, 'optionalAccess', _163 => _163.trim, 'call', _164 => _164()]), () => ( ""))) && prev.bbox && block.bbox && Math.abs(prev.bbox.y - block.bbox.y) < 2 && prev.bbox.x + prev.bbox.width <= block.bbox.x) {
       prev.text = `${prev.text.trim()} ${block.text.trim()}`;
       prev.bbox = { ...prev.bbox, width: block.bbox.x + block.bbox.width - prev.bbox.x };
       blocks.splice(i--, 1);
       continue;
     }
-    if (_optionalChain([prev, 'optionalAccess', _168 => _168.type]) === "paragraph" && prev.pageNumber === block.pageNumber && /^\d{1,2}(?:\.\d{1,2})*\.?$/.test(_nullishCoalesce(_optionalChain([prev, 'access', _169 => _169.text, 'optionalAccess', _170 => _170.trim, 'call', _171 => _171()]), () => ( ""))) && prev.bbox && block.bbox && fsz > 0 && (Math.abs(prev.bbox.y - block.bbox.y) < 2 && prev.bbox.x + prev.bbox.width <= block.bbox.x && block.bbox.x - (prev.bbox.x + prev.bbox.width) < fsz * 3 || Math.abs(prev.bbox.x - block.bbox.x) < 2 && (_nullishCoalesce(_optionalChain([prev, 'access', _172 => _172.style, 'optionalAccess', _173 => _173.fontSize]), () => ( 0))) >= fsz && prev.bbox.y - (block.bbox.y + block.bbox.height) < fsz * 1.5)) {
+    if (_optionalChain([prev, 'optionalAccess', _165 => _165.type]) === "paragraph" && prev.pageNumber === block.pageNumber && /^\d{1,2}(?:\.\d{1,2})*\.?$/.test(_nullishCoalesce(_optionalChain([prev, 'access', _166 => _166.text, 'optionalAccess', _167 => _167.trim, 'call', _168 => _168()]), () => ( ""))) && prev.bbox && block.bbox && fsz > 0 && (Math.abs(prev.bbox.y - block.bbox.y) < 2 && prev.bbox.x + prev.bbox.width <= block.bbox.x && block.bbox.x - (prev.bbox.x + prev.bbox.width) < fsz * 3 || Math.abs(prev.bbox.x - block.bbox.x) < 2 && (_nullishCoalesce(_optionalChain([prev, 'access', _169 => _169.style, 'optionalAccess', _170 => _170.fontSize]), () => ( 0))) >= fsz && prev.bbox.y - (block.bbox.y + block.bbox.height) < fsz * 1.5)) {
       block.text = `${prev.text.trim()} ${block.text.trim()}`;
       blocks.splice(i - 1, 1);
       i -= 2;
       continue;
     }
-    const box = block.bbox, size = _nullishCoalesce(_optionalChain([block, 'access', _174 => _174.style, 'optionalAccess', _175 => _175.fontSize]), () => ( 0));
-    const kicker = !!box && size > 0 && _optionalChain([next, 'optionalAccess', _176 => _176.type]) === "heading" && next.pageNumber === block.pageNumber && !!next.bbox && (_nullishCoalesce(_optionalChain([next, 'access', _177 => _177.style, 'optionalAccess', _178 => _178.fontSize]), () => ( 0))) >= size * 1.3 && box.y - (next.bbox.y + next.bbox.height) <= size * 3 && Math.min(box.x + box.width, next.bbox.x + next.bbox.width) - Math.max(box.x, next.bbox.x) >= Math.min(box.width, next.bbox.width) * 0.5 && !page.some((o) => o !== block && o.bbox && o.bbox.y > box.y + box.height && o.type !== "image");
-    const face = _nullishCoalesce(_optionalChain([faceNames, 'optionalAccess', _179 => _179.get, 'call', _180 => _180(_nullishCoalesce(_optionalChain([block, 'access', _181 => _181.style, 'optionalAccess', _182 => _182.fontName]), () => ( "")))]), () => ( ""));
+    const box = block.bbox, size = _nullishCoalesce(_optionalChain([block, 'access', _171 => _171.style, 'optionalAccess', _172 => _172.fontSize]), () => ( 0));
+    const kicker = !!box && size > 0 && _optionalChain([next, 'optionalAccess', _173 => _173.type]) === "heading" && next.pageNumber === block.pageNumber && !!next.bbox && (_nullishCoalesce(_optionalChain([next, 'access', _174 => _174.style, 'optionalAccess', _175 => _175.fontSize]), () => ( 0))) >= size * 1.3 && box.y - (next.bbox.y + next.bbox.height) <= size * 3 && Math.min(box.x + box.width, next.bbox.x + next.bbox.width) - Math.max(box.x, next.bbox.x) >= Math.min(box.width, next.bbox.width) * 0.5 && !page.some((o) => o !== block && o.bbox && o.bbox.y > box.y + box.height && o.type !== "image");
+    const face = _nullishCoalesce(_optionalChain([faceNames, 'optionalAccess', _176 => _176.get, 'call', _177 => _177(_nullishCoalesce(_optionalChain([block, 'access', _178 => _178.style, 'optionalAccess', _179 => _179.fontName]), () => ( "")))]), () => ( ""));
     const [bodyFace, bodySize] = (_nullishCoalesce(bodyStyle.get(_nullishCoalesce(block.pageNumber, () => ( 0))), () => ( ""))).split(":");
-    const byline = /Italic|Oblique/i.test(face) && !/Bold|Black|Heavy|Semibold/i.test(face) && !!bodyFace && size <= Number(bodySize) + 0.5 && _optionalChain([blocks, 'access', _183 => _183[i - 1], 'optionalAccess', _184 => _184.type]) === "heading" && blocks[i - 1].pageNumber === block.pageNumber;
-    const ocrFragment = _optionalChain([block, 'access', _185 => _185.style, 'optionalAccess', _186 => _186.fontName]) === "ocr" && (/^[.,;:…·•-]/.test(text) || text.includes("\u2026") || (_nullishCoalesce(_optionalChain([text, 'access', _187 => _187.match, 'call', _188 => _188(/\p{L}/gu), 'optionalAccess', _189 => _189.length]), () => ( 0))) < 2 || text.split(/\s+/).length > 7);
+    const byline = /Italic|Oblique/i.test(face) && !/Bold|Black|Heavy|Semibold/i.test(face) && !!bodyFace && size <= Number(bodySize) + 0.5 && _optionalChain([blocks, 'access', _180 => _180[i - 1], 'optionalAccess', _181 => _181.type]) === "heading" && blocks[i - 1].pageNumber === block.pageNumber;
+    const ocrFragment = _optionalChain([block, 'access', _182 => _182.style, 'optionalAccess', _183 => _183.fontName]) === "ocr" && (/^[.,;:…·•-]/.test(text) || text.includes("\u2026") || (_nullishCoalesce(_optionalChain([text, 'access', _184 => _184.match, 'call', _185 => _185(/\p{L}/gu), 'optionalAccess', _186 => _186.length]), () => ( 0))) < 2 || text.split(/\s+/).length > 7);
     if (ocrFragment || tocEntry || proseStyle || kicker || byline || tiny || !/\p{L}/u.test(text) && !chapterNumber || /^[a-z]/.test(text) || CAPTION.test(text) || EQUATION_NUMBER.test(block.text) || DISPLAY_MATH.test(text) || // 닫는 괄호가 여는 괄호보다 많으면 앞 줄에서 이어진 문장 조각이다 ("Fact-checking) and is used …") — "1)"·"가)" 앞머리 번호는 빼고 센다
     unbalancedClose(text.replace(/^\s*[\dA-Za-z가-힣ⅰ-ⅹ]{1,3}\)\s*/, "")) || isRunningHead(block, page, pageHeights.get(_nullishCoalesce(block.pageNumber, () => ( 0))))) {
       block.type = "paragraph";
@@ -2187,7 +2343,9 @@ function normalizeAraea(text) {
   return text.replace(/(?<![\u1100-\u115F])\u119E/g, "\u318D");
 }
 var stripNoUnicodeGlyph = (text) => text.includes("\uF000") ? text.replace(/\uF000/g, "") : text;
-var cleanChars = (text) => normalizeAraea(stripNoUnicodeGlyph(stripControlChars(text)));
+var CHOSEONG_COMPAT = "\u3131\u3132\u3134\u3137\u3138\u3139\u3141\u3142\u3143\u3145\u3146\u3147\u3148\u3149\u314A\u314B\u314C\u314D\u314E";
+var normalizeLoneChoseong = (text) => /[\u1100-\u1112]/.test(text) ? text.replace(/[\u1100-\u1112](?![\u1160-\u11A7\uD7B0-\uD7C6])/g, (c) => CHOSEONG_COMPAT[c.charCodeAt(0) - 4352]) : text;
+var cleanChars = (text) => normalizeLoneChoseong(normalizeAraea(stripNoUnicodeGlyph(stripControlChars(text))));
 function sanitizeBlockControlChars(blocks) {
   for (const b of blocks) {
     if (b.text) b.text = cleanChars(b.text);
@@ -2206,28 +2364,28 @@ var CAPTION_LABEL = /^(?:figure|fig\.|diagram|table|chart|graph|exhibit|box)\s*[
 function captionTableText(t) {
   if (!t || t.rows !== 1 || t.cols !== 2) return null;
   const [label, body] = t.cells[0];
-  if (!label || !body || _optionalChain([label, 'access', _190 => _190.blocks, 'optionalAccess', _191 => _191.length]) || _optionalChain([body, 'access', _192 => _192.blocks, 'optionalAccess', _193 => _193.length]) || !CAPTION_LABEL.test(label.text.trim()) || !body.text.trim()) return null;
+  if (!label || !body || _optionalChain([label, 'access', _187 => _187.blocks, 'optionalAccess', _188 => _188.length]) || _optionalChain([body, 'access', _189 => _189.blocks, 'optionalAccess', _190 => _190.length]) || !CAPTION_LABEL.test(label.text.trim()) || !body.text.trim()) return null;
   return `${label.text.trim()} ${body.text.replace(/\s*\n\s*/g, " ").trim()}`;
 }
 function splitSingleCellTables(blocks) {
   const out = [];
   for (const b of blocks) {
     const t = b.type === "table" ? b.table : void 0;
-    const only = t && t.rows === 1 && t.cols === 1 && _optionalChain([t, 'access', _194 => _194.cells, 'access', _195 => _195[0], 'optionalAccess', _196 => _196[0], 'optionalAccess', _197 => _197.blocks, 'optionalAccess', _198 => _198.length]) === 1 ? t.cells[0][0].blocks[0] : void 0;
+    const only = t && t.rows === 1 && t.cols === 1 && _optionalChain([t, 'access', _191 => _191.cells, 'access', _192 => _192[0], 'optionalAccess', _193 => _193[0], 'optionalAccess', _194 => _194.blocks, 'optionalAccess', _195 => _195.length]) === 1 ? t.cells[0][0].blocks[0] : void 0;
     const squash = (s) => s.replace(/\s/g, "");
     const inner = only && (!squash(t.cells[0][0].text) || only.type === "table" && squash(t.cells[0][0].text) === squash(only.table.cells.flat().map((c) => c.text).join(""))) ? only : void 0;
-    const caption = _nullishCoalesce(captionTableText(t), () => ( (_optionalChain([inner, 'optionalAccess', _199 => _199.type]) === "table" ? captionTableText(inner.table) : null)));
+    const caption = _nullishCoalesce(captionTableText(t), () => ( (_optionalChain([inner, 'optionalAccess', _196 => _196.type]) === "table" ? captionTableText(inner.table) : null)));
     if (caption) {
       out.push({ type: "paragraph", text: caption, pageNumber: b.pageNumber, bbox: b.bbox });
       continue;
     }
-    const cell = t && t.rows === 1 && t.cols === 1 ? _optionalChain([t, 'access', _200 => _200.cells, 'access', _201 => _201[0], 'optionalAccess', _202 => _202[0]]) : void 0;
-    if (!cell || _optionalChain([cell, 'access', _203 => _203.blocks, 'optionalAccess', _204 => _204.some, 'call', _205 => _205((x) => x.type === "table")])) {
+    const cell = t && t.rows === 1 && t.cols === 1 ? _optionalChain([t, 'access', _197 => _197.cells, 'access', _198 => _198[0], 'optionalAccess', _199 => _199[0]]) : void 0;
+    if (!cell || _optionalChain([cell, 'access', _200 => _200.blocks, 'optionalAccess', _201 => _201.some, 'call', _202 => _202((x) => x.type === "table")])) {
       out.push(b);
       continue;
     }
     const lines = (_nullishCoalesce(cell.text, () => ( ""))).split(/\n/).map((l) => l.trim()).filter(Boolean);
-    if (_optionalChain([t, 'optionalAccess', _206 => _206.caption, 'optionalAccess', _207 => _207.trim, 'call', _208 => _208()])) out.push({ type: "paragraph", text: t.caption.trim(), pageNumber: b.pageNumber, bbox: b.bbox });
+    if (_optionalChain([t, 'optionalAccess', _203 => _203.caption, 'optionalAccess', _204 => _204.trim, 'call', _205 => _205()])) out.push({ type: "paragraph", text: t.caption.trim(), pageNumber: b.pageNumber, bbox: b.bbox });
     for (const text of lines) out.push({ type: "paragraph", text, pageNumber: b.pageNumber, bbox: b.bbox });
   }
   return out;
@@ -2240,7 +2398,7 @@ function joinLatinCellWraps(blocks) {
         const lines = cell.text.split("\n");
         const out = [lines[0]];
         for (const line of lines.slice(1)) {
-          if (_chunkMH6MAL64cjs.latinSoftWrap.call(void 0, out[out.length - 1], line)) out[out.length - 1] = out[out.length - 1].trimEnd() + " " + line.trim();
+          if (_chunkWDQTFAXYcjs.latinSoftWrap.call(void 0, out[out.length - 1], line)) out[out.length - 1] = out[out.length - 1].trimEnd() + " " + line.trim();
           else out.push(line);
         }
         cell.text = out.join("\n");
@@ -2251,20 +2409,20 @@ function joinLatinCellWraps(blocks) {
 }
 function cleanPdfText(text, opts) {
   let clean = normalizeAraea(stripControlChars(text));
-  if (!_optionalChain([opts, 'optionalAccess', _209 => _209.keepLoneNumbers])) {
+  if (!_optionalChain([opts, 'optionalAccess', _206 => _206.keepLoneNumbers])) {
     clean = clean.replace(/^\d{1,4}\n/, "").replace(/\n\d{1,4}\n/g, "\n").replace(/\n\d{1,4}$/, "");
   }
-  if (!_optionalChain([opts, 'optionalAccess', _210 => _210.keepLoneNumbers])) clean = clean.replace(/^#{1,6}\s*\d{1,4}\s*$/gm, "");
+  if (!_optionalChain([opts, 'optionalAccess', _207 => _207.keepLoneNumbers])) clean = clean.replace(/^#{1,6}\s*\d{1,4}\s*$/gm, "");
   return mergeKoreanLines(
     clean.replace(/^[\s]*[-–—]\s*[-–—]?\d+[-–—]?[\s]*[-–—]?[\s]*$/gm, "").replace(/^\s*\d+\s*\/\s*\d+\s*$/gm, "").replace(/\.(?: \.){3,}/g, (m) => ".".repeat(m.length + 1 >> 1))
   ).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) => _chunkMH6MAL64cjs.collapseEvenSpacing.call(void 0, t, false))).join("<br>"));
+      return line.replace(/(?:\\.|[^|\\])+/g, (cell) => cell.split("<br>").map((seg) => seg.replace(/\S(?:.*\S)?/, (t) => _chunkWDQTFAXYcjs.collapseEvenSpacing.call(void 0, t, false))).join("<br>"));
     }
-    const mark = _nullishCoalesce(_optionalChain([/^(?:#{1,6}|-) /, 'access', _211 => _211.exec, 'call', _212 => _212(line), 'optionalAccess', _213 => _213[0]]), () => ( ""));
-    return mark + _chunkMH6MAL64cjs.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();
+    const mark = _nullishCoalesce(_optionalChain([/^(?:#{1,6}|-) /, 'access', _208 => _208.exec, 'call', _209 => _209(line), 'optionalAccess', _210 => _210[0]]), () => ( ""));
+    return mark + _chunkWDQTFAXYcjs.collapseEvenSpacing.call(void 0, line.slice(mark.length), false);
+  }).replace(/\\~\\~/g, "~~").replace(/~~~~/g, "").replace(/<u>\s*<\/u>/g, "").replace(/<(sup|sub)>\s*<\/\1>/g, "").replace(/\n{3,}/g, "\n\n").trim();
 }
 function startsWithMarker(line) {
   const t = line.trimStart();
@@ -2295,7 +2453,7 @@ function mergeKoreanLines(text) {
       continue;
     }
     if (/[가-힣·,\-]$/.test(prev) && /^[가-힣(]/.test(curr) && !startsWithMarker(curr) && !isStandaloneHeader(prev) && !startsWithMarker(prev)) {
-      result[result.length - 1] = prev + _chunkMH6MAL64cjs.wrapJoiner.call(void 0, prev, curr) + curr;
+      result[result.length - 1] = prev + _chunkWDQTFAXYcjs.wrapJoiner.call(void 0, prev, curr) + curr;
     } else {
       result.push(curr);
     }
@@ -2311,7 +2469,7 @@ 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 = _chunkNWODPYBDcjs.sanitizeHref.call(void 0, a.url);
+    const url = _chunkSTM3ORY7cjs.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]);
@@ -2352,7 +2510,7 @@ function mergeLinkRuns(markdown) {
 
 // src/pdf/formula-ocr.ts
 async function applyFormulaOcr(buffer, blocks, pageFilter, effectivePageCount, warnings, _onProgress) {
-  const formulaMod = await Promise.resolve().then(() => _interopRequireWildcard(require("./formula-VEFKYXPB.cjs")));
+  const formulaMod = await Promise.resolve().then(() => _interopRequireWildcard(require("./formula-WIMV5TUX.cjs")));
   const { FormulaPipeline, ensureFormulaModels } = formulaMod;
   await ensureFormulaModels((p) => {
     if (p.phase === "download" && p.total) {
@@ -2492,42 +2650,15 @@ g.pdfjsWorker = pdfjsWorker;
 
 var _module = require('module');
 var _path = require('path');
-
-// src/pdf/contact-table.ts
-var HEAD = /^(?:담당\s*부서|<[^<>]+>)$/;
-var ROLE = /^(?:책임자|담당자)$/;
-var PERSON = /^(.+?)\s+([가-힣]{2,4})\s+(\([^()]*\d[^()]*\)|\S+@\S+)$/;
-function splitContactTables(blocks) {
-  for (const b of blocks) {
-    const t = b.type === "table" ? b.table : void 0;
-    if (!t || t.cols !== 4 || t.rows < 2) continue;
-    const rows = t.cells;
-    if (!HEAD.test(rows[0][0].text.replace(/\s+/g, " ").trim())) continue;
-    const people = rows.map((r) => PERSON.exec(r[3].text.replace(/\s+/g, " ").trim()));
-    if (people.some((p) => !p) || rows.some((r) => !ROLE.test(r[2].text.replace(/\s+/g, "")) || r[2].rowSpan !== 1 || r[3].rowSpan !== 1)) continue;
-    const lead = [];
-    let ok = true;
-    for (const c of [0, 1]) {
-      const top = rows[0][c];
-      if (top.rowSpan === 1) {
-        lead.push(rows.map((r) => r[c].text));
-        continue;
-      }
-      const lines = top.text.split("\n").map((s) => s.trim()).filter(Boolean);
-      if (top.rowSpan !== t.rows || lines.length > t.rows) {
-        ok = false;
-        break;
-      }
-      lead.push(rows.map((_, r) => _nullishCoalesce(lines[r], () => ( ""))));
-    }
-    if (!ok) continue;
-    const one = (text) => ({ text, colSpan: 1, rowSpan: 1 });
-    t.cells = rows.map((r, i) => [one(lead[0][i]), one(lead[1][i]), one(r[2].text), one(people[i][1]), one(people[i][2]), one(people[i][3])]);
-    t.cols = 6;
+function ocrImageRegions(regions, large, all) {
+  if (all) return regions;
+  const out = /* @__PURE__ */ new Map();
+  for (const [p, rs] of regions) {
+    const big = rs.filter((r) => large.has(r));
+    if (big.length) out.set(p, big);
   }
+  return out;
 }
-
-// src/pdf/parser.ts
 _pdfmjs.GlobalWorkerOptions.workerSrc = "";
 var MAX_PAGES = 5e3;
 var MAX_TOTAL_TEXT = 100 * 1024 * 1024;
@@ -2558,13 +2689,13 @@ async function loadPdfWithTimeout(buffer) {
   try {
     return await Promise.race([
       loadingTask.promise.catch((e) => {
-        if (e instanceof Error && e.name === "PasswordException") throw new (0, _chunkNWODPYBDcjs.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)");
+        if (e instanceof Error && e.name === "PasswordException") throw new (0, _chunkSTM3ORY7cjs.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, _chunkNWODPYBDcjs.KordocError)("PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (30\uCD08 \uCD08\uACFC)"));
+          reject(new (0, _chunkSTM3ORY7cjs.KordocError)("PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (30\uCD08 \uCD08\uACFC)"));
         }, PDF_LOAD_TIMEOUT_MS);
       })
     ]);
@@ -2573,13 +2704,13 @@ async function loadPdfWithTimeout(buffer) {
   }
 }
 async function parsePdfDocument(buffer, options) {
-  const formulaBuffer = _optionalChain([options, 'optionalAccess', _214 => _214.formulaOcr]) ? buffer : null;
-  const autoOcr = _optionalChain([options, 'optionalAccess', _215 => _215.ocr]) === void 0 && await _chunkMH6MAL64cjs.ocrModelsCached.call(void 0, );
-  const ocrBuffer = _optionalChain([options, 'optionalAccess', _216 => _216.ocr]) || autoOcr ? buffer : null;
+  const formulaBuffer = _optionalChain([options, 'optionalAccess', _211 => _211.formulaOcr]) ? buffer : null;
+  const autoOcr = _optionalChain([options, 'optionalAccess', _212 => _212.ocr]) === void 0 && await _chunkWDQTFAXYcjs.ocrModelsCached.call(void 0, );
+  const ocrBuffer = _optionalChain([options, 'optionalAccess', _213 => _213.ocr]) || autoOcr ? buffer : null;
   const doc = await loadPdfWithTimeout(buffer);
   try {
     const pageCount = doc.numPages;
-    if (pageCount === 0) throw new (0, _chunkNWODPYBDcjs.KordocError)("PDF\uC5D0 \uD398\uC774\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
+    if (pageCount === 0) throw new (0, _chunkSTM3ORY7cjs.KordocError)("PDF\uC5D0 \uD398\uC774\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
     const metadata = { pageCount, pageMode: "layout" };
     await extractPdfMetadata(doc, metadata);
     const blocks = [];
@@ -2588,8 +2719,8 @@ async function parsePdfDocument(buffer, options) {
     let totalChars = 0;
     let totalTextBytes = 0;
     const effectivePageCount = Math.min(pageCount, MAX_PAGES);
-    const pageFilter = _optionalChain([options, 'optionalAccess', _217 => _217.pages]) ? _chunkJ55UQC5Hcjs.parsePageRange.call(void 0, options.pages, effectivePageCount) : null;
-    if (pageCount > MAX_PAGES && (!_optionalChain([options, 'optionalAccess', _218 => _218.pages]) || _chunkJ55UQC5Hcjs.hasRequestedPagesAfter.call(void 0, options.pages, MAX_PAGES, pageCount))) {
+    const pageFilter = _optionalChain([options, 'optionalAccess', _214 => _214.pages]) ? _chunkJ55UQC5Hcjs.parsePageRange.call(void 0, options.pages, effectivePageCount) : null;
+    if (pageCount > MAX_PAGES && (!_optionalChain([options, 'optionalAccess', _215 => _215.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;
@@ -2600,12 +2731,13 @@ async function parsePdfDocument(buffer, options) {
     const pagesWithLargeImage = /* @__PURE__ */ new Set();
     const skippedImagePages = /* @__PURE__ */ new Map();
     const uncoveredImageRegions = /* @__PURE__ */ new Map();
+    const largeImageRegions = /* @__PURE__ */ new Set();
     const imageState = createPdfImageState();
     const extractedImages = [];
     const pageImageBlocks = /* @__PURE__ */ new Map();
     const carry = {};
     const vectorPageOps = /* @__PURE__ */ new Map();
-    const wrapLexicon = new (0, _chunkMH6MAL64cjs.WrapLexicon)();
+    const wrapLexicon = new (0, _chunkWDQTFAXYcjs.WrapLexicon)();
     const noteMarks = /* @__PURE__ */ new Map();
     let parsedPages = 0;
     for (let i = 1; i <= effectivePageCount; i++) {
@@ -2628,12 +2760,13 @@ async function parsePdfDocument(buffer, options) {
             return void 0;
           }
         };
-        const differencesOf = (loadedName) => _optionalChain([fontObj, 'call', _219 => _219(loadedName), 'optionalAccess', _220 => _220.differences]);
-        restoreNamedGlyphs(rawItems, rawOps.fnArray, rawOps.argsArray, differencesOf, (n) => _optionalChain([fontObj, 'call', _221 => _221(n), 'optionalAccess', _222 => _222.name]));
+        const differencesOf = (loadedName) => _optionalChain([fontObj, 'call', _216 => _216(loadedName), 'optionalAccess', _217 => _217.differences]);
+        restoreNamedGlyphs(rawItems, rawOps.fnArray, rawOps.argsArray, differencesOf, (n) => _optionalChain([fontObj, 'call', _218 => _218(n), 'optionalAccess', _219 => _219.name]));
         restoreTrackedSpacing(rawItems, rawOps.fnArray, rawOps.argsArray);
+        markSyntheticSpaces(rawItems, rawOps.fnArray, rawOps.argsArray);
         const occluded = occludedTextItems(rawItems, rawOps.fnArray, rawOps.argsArray);
-        const items = _chunkMH6MAL64cjs.normalizeItems.call(void 0, occluded.size ? rawItems.filter((it) => !occluded.has(it)) : rawItems);
-        const filtered = _chunkMH6MAL64cjs.filterHiddenText.call(void 0, items, pageW, pageH, viewX1, viewY1);
+        const items = _chunkWDQTFAXYcjs.normalizeItems.call(void 0, occluded.size ? rawItems.filter((it) => !occluded.has(it)) : rawItems);
+        const filtered = _chunkWDQTFAXYcjs.filterHiddenText.call(void 0, items, pageW, pageH, viewX1, viewY1);
         const joined = joinVerticalColumns(filtered.visible);
         const visible = dropTabLeaderDots(joined);
         const hiddenCount = filtered.hiddenCount + occluded.size;
@@ -2657,7 +2790,7 @@ async function parsePdfDocument(buffer, options) {
           fnArray: [_pdfmjs.OPS.transform, ...rawOps.fnArray],
           argsArray: [[1, 0, 0, 1, -viewX1, -viewY1], ...rawOps.argsArray]
         } : rawOps;
-        remapSymbolFontItems(visible, (loadedName) => _optionalChain([fontObj, 'call', _223 => _223(loadedName), 'optionalAccess', _224 => _224.name]));
+        remapSymbolFontItems(visible, (loadedName) => _optionalChain([fontObj, 'call', _220 => _220(loadedName), 'optionalAccess', _221 => _221.name]));
         remapControlGlyphs(visible, differencesOf);
         const faces = /* @__PURE__ */ new Map();
         for (const it of visible) {
@@ -2665,17 +2798,17 @@ async function parsePdfDocument(buffer, options) {
           let face = faces.get(it.fontName);
           if (face === void 0) {
             const obj = fontObj(it.fontName);
-            face = _optionalChain([obj, 'optionalAccess', _225 => _225.isType3Font]) && /^[A-Z]{6}\+./.test(_nullishCoalesce(obj.name, () => ( ""))) && obj.name.slice(7) || it.fontName;
+            face = _optionalChain([obj, 'optionalAccess', _222 => _222.isType3Font]) && /^[A-Z]{6}\+./.test(_nullishCoalesce(obj.name, () => ( ""))) && obj.name.slice(7) || it.fontName;
             faces.set(it.fontName, face);
           }
           it.fontName = face;
-          if (!faceNames.has(face)) faceNames.set(face, (_nullishCoalesce(_optionalChain([fontObj, 'call', _226 => _226(face), 'optionalAccess', _227 => _227.name]), () => ( ""))).replace(/^[A-Z]{6}\+/, ""));
+          if (!faceNames.has(face)) faceNames.set(face, (_nullishCoalesce(_optionalChain([fontObj, 'call', _223 => _223(face), 'optionalAccess', _224 => _224.name]), () => ( ""))).replace(/^[A-Z]{6}\+/, ""));
         }
-        for (const it of visible) if (it.text.includes("$")) it.text = _chunkDRY3L5KQcjs.escapeLiteralDollar.call(void 0, it.text);
+        for (const it of visible) if (it.text.includes("$")) it.text = _chunkXTFN5WKScjs.escapeLiteralDollar.call(void 0, it.text);
         wrapEquationRuns(visible, (face) => faceNames.get(face));
         const pageArea = pageW * pageH;
         if (pageArea > 0) {
-          const imageRegions = _chunkMH6MAL64cjs.extractImageRegions.call(void 0, opList.fnArray, opList.argsArray);
+          const imageRegions = _chunkWDQTFAXYcjs.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);
@@ -2694,20 +2827,21 @@ async function parsePdfDocument(buffer, options) {
                 const regions = _nullishCoalesce(uncoveredImageRegions.get(i), () => ( []));
                 regions.push(r);
                 uncoveredImageRegions.set(i, regions);
+                if (large) largeImageRegions.add(r);
               }
             }
           }
           if (uncovered > 0) skippedImagePages.set(i, uncovered);
         }
         const marks = superscriptNoteMarks(visible);
-        if (marks.length) noteMarks.set(i, { marks, seps: footnoteSeparators(_chunkMH6MAL64cjs.extractLines.call(void 0, opList.fnArray, opList.argsArray).horizontals, pageW, pageH) });
+        if (marks.length) noteMarks.set(i, { marks, seps: footnoteSeparators(_chunkWDQTFAXYcjs.extractLines.call(void 0, opList.fnArray, opList.argsArray).horizontals, pageW, pageH) });
         const stamp = marginStamp(visible);
         const flow = stamp.length ? visible.filter((it) => !stamp.includes(it)) : visible;
-        const pageBlocks = orderTwoUpPage(_chunkMH6MAL64cjs.extractPageBlocksWithLines.call(void 0, flow, i, opList, pageW, pageH, void 0, _optionalChain([options, 'optionalAccess', _228 => _228.tables]) !== false, carry, wrapLexicon), pageW, pageH);
+        const pageBlocks = orderTwoUpPage(_chunkWDQTFAXYcjs.extractPageBlocksWithLines.call(void 0, flow, i, opList, pageW, pageH, void 0, _optionalChain([options, 'optionalAccess', _225 => _225.tables]) !== false, carry, wrapLexicon), pageW, pageH);
         if (stamp.length) pageBlocks.unshift({ type: "paragraph", text: [...stamp].sort((a, b) => a.y - b.y).map((it) => it.text).join(" "), pageNumber: i });
         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', _229 => _229.images]) !== false);
+          const { blocks: imgBlocks, images: pageImages } = await extractPageImages(page, opList.fnArray, opList.argsArray, i, imageState, warnings, _optionalChain([options, 'optionalAccess', _226 => _226.images]) !== false);
           if (imgBlocks.length > 0) pageImageBlocks.set(i, imgBlocks);
           extractedImages.push(...pageImages);
         } catch (e6) {
@@ -2726,26 +2860,27 @@ 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', _230 => _230.ocr]) && quality.ocrReason === "vector_text" && page.rotate % 360 === 0 && page.view[0] === 0 && page.view[1] === 0) {
+        if (_optionalChain([options, 'optionalAccess', _227 => _227.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, _chunkNWODPYBDcjs.KordocError)("\uD14D\uC2A4\uD2B8 \uCD94\uCD9C \uD06C\uAE30 \uCD08\uACFC");
+        if (totalTextBytes > MAX_TOTAL_TEXT) throw new (0, _chunkSTM3ORY7cjs.KordocError)("\uD14D\uC2A4\uD2B8 \uCD94\uCD9C \uD06C\uAE30 \uCD08\uACFC");
         parsedPages++;
-        _optionalChain([options, 'optionalAccess', _231 => _231.onProgress, 'optionalCall', _232 => _232(parsedPages, totalTarget)]);
+        _optionalChain([options, 'optionalAccess', _228 => _228.onProgress, 'optionalCall', _229 => _229(parsedPages, totalTarget)]);
       } catch (pageErr) {
-        if (pageErr instanceof _chunkNWODPYBDcjs.KordocError) throw pageErr;
+        if (pageErr instanceof _chunkSTM3ORY7cjs.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', _233 => _233.cleanup, 'call', _234 => _234()]);
+        _optionalChain([loadedPage, 'optionalAccess', _230 => _230.cleanup, 'call', _231 => _231()]);
       }
     }
     const parsedPageCount = parsedPages || (pageFilter ? pageFilter.size : effectivePageCount);
-    const isImageBased = _optionalChain([pageFilter, 'optionalAccess', _235 => _235.size]) !== 0 && totalChars / Math.max(parsedPageCount, 1) < 10;
+    const isImageBased = _optionalChain([pageFilter, 'optionalAccess', _232 => _232.size]) !== 0 && totalChars / Math.max(parsedPageCount, 1) < 10;
     const ocrDone = /* @__PURE__ */ new Set();
+    const ocrRegions = _optionalChain([options, 'optionalAccess', _233 => _233.ocr]) === true || autoOcr ? ocrImageRegions(uncoveredImageRegions, largeImageRegions, _optionalChain([options, 'optionalAccess', _234 => _234.ocr]) === true) : /* @__PURE__ */ new Map();
     if (ocrBuffer) {
       const inScope = (p) => !pageFilter || pageFilter.has(p);
       const targets = /* @__PURE__ */ new Set();
-      if (_optionalChain([options, 'optionalAccess', _236 => _236.ocr]) === "force" || isImageBased) {
+      if (_optionalChain([options, 'optionalAccess', _235 => _235.ocr]) === "force" || isImageBased) {
         for (let i = 1; i <= effectivePageCount; i++) if (inScope(i)) targets.add(i);
       } else {
         for (const pq of pageQuality) {
@@ -2754,27 +2889,28 @@ async function parsePdfDocument(buffer, options) {
           if (autoOcr && pq.ocrReason !== "low_text" && pq.ocrReason !== "vector_text") continue;
           targets.add(pq.page);
         }
-        if (_optionalChain([options, 'optionalAccess', _237 => _237.ocr]) === true) for (const p of uncoveredImageRegions.keys()) targets.add(p);
+        for (const p of ocrRegions.keys()) targets.add(p);
       }
       if (targets.size > 0) {
         try {
-          const { runPdfOcr } = await Promise.resolve().then(() => _interopRequireWildcard(require("./pdf-ocr-T2JXQATA.cjs")));
-          const mode = typeof _optionalChain([options, 'optionalAccess', _238 => _238.ocr]) === "function" ? options.ocr : "builtin";
-          const regionPages = new Map([...uncoveredImageRegions].filter(([p]) => _optionalChain([options, 'optionalAccess', _239 => _239.ocr]) !== "force" && !isImageBased && !_optionalChain([pageQuality, 'access', _240 => _240.find, 'call', _241 => _241((q) => q.page === p), 'optionalAccess', _242 => _242.needsOcr])));
-          const ocrPageBlocks = await runPdfOcr(ocrBuffer, targets, mode, warnings, _optionalChain([options, 'optionalAccess', _243 => _243.onProgress]), _optionalChain([options, 'optionalAccess', _244 => _244.tables]) !== false, vectorPageOps, regionPages);
+          const { runPdfOcr } = await Promise.resolve().then(() => _interopRequireWildcard(require("./pdf-ocr-GHEJVPX2.cjs")));
+          const mode = typeof _optionalChain([options, 'optionalAccess', _236 => _236.ocr]) === "function" ? options.ocr : "builtin";
+          const regionPages = new Map([...ocrRegions].filter(([p]) => _optionalChain([options, 'optionalAccess', _237 => _237.ocr]) !== "force" && !isImageBased && !_optionalChain([pageQuality, 'access', _238 => _238.find, 'call', _239 => _239((q) => q.page === p), 'optionalAccess', _240 => _240.needsOcr])));
+          const ocrPageBlocks = await runPdfOcr(ocrBuffer, targets, mode, warnings, _optionalChain([options, 'optionalAccess', _241 => _241.onProgress]), _optionalChain([options, 'optionalAccess', _242 => _242.tables]) !== false, vectorPageOps, regionPages);
+          for (const obs of ocrPageBlocks.values()) _chunk62W7QCEXcjs.stripScriptTags.call(void 0, { blocks: obs });
           if (ocrPageBlocks.size > 0) {
             const replacePages = /* @__PURE__ */ new Set();
             for (const [p, obs] of ocrPageBlocks) {
-              const needsOcr = _optionalChain([pageQuality, 'access', _245 => _245.find, 'call', _246 => _246((q) => q.page === p), 'optionalAccess', _247 => _247.needsOcr]);
-              if (_optionalChain([options, 'optionalAccess', _248 => _248.ocr]) === "force" || isImageBased || needsOcr) {
+              const needsOcr = _optionalChain([pageQuality, 'access', _243 => _243.find, 'call', _244 => _244((q) => q.page === p), 'optionalAccess', _245 => _245.needsOcr]);
+              if (_optionalChain([options, 'optionalAccess', _246 => _246.ocr]) === "force" || isImageBased || needsOcr) {
                 replacePages.add(p);
                 ocrDone.add(p);
                 continue;
               }
-              const regions = uncoveredImageRegions.get(p);
+              const regions = ocrRegions.get(p);
               if (!regions || mergeOcrImageRegions(blocks, p, regions, obs) === 0) continue;
               ocrDone.add(p);
-              if (_optionalChain([pageImageBlocks, 'access', _249 => _249.get, 'call', _250 => _250(p), 'optionalAccess', _251 => _251.length]) === 1) pageImageBlocks.delete(p);
+              if (_optionalChain([pageImageBlocks, 'access', _247 => _247.get, 'call', _248 => _248(p), 'optionalAccess', _249 => _249.length]) === 1) pageImageBlocks.delete(p);
             }
             if (replacePages.size) {
               const merged = blocks.filter((b) => !(b.pageNumber && replacePages.has(b.pageNumber)));
@@ -2824,8 +2960,8 @@ 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 \u2014 \uADF8\uB9BC \uC18D \uAE00\uC740 ocr: true / CLI --ocr \uB85C \uC77D\uC2B5\uB2C8\uB2E4)`, code: "SKIPPED_IMAGE" });
       }
     }
-    if (_optionalChain([options, 'optionalAccess', _252 => _252.removeHeaderFooter]) !== false && parsedPageCount >= 3) {
-      const removed = _chunkMH6MAL64cjs.removeHeaderFooterBlocks.call(void 0, blocks, pageHeights, warnings);
+    if (_optionalChain([options, 'optionalAccess', _250 => _250.removeHeaderFooter]) !== false && parsedPageCount >= 3) {
+      const removed = _chunkWDQTFAXYcjs.removeHeaderFooterBlocks.call(void 0, blocks, pageHeights, warnings, noteMarks);
       for (let ri = removed.length - 1; ri >= 0; ri--) {
         blocks.splice(removed[ri], 1);
       }
@@ -2836,12 +2972,16 @@ async function parsePdfDocument(buffer, options) {
       }
     }
     mergeContinuedCells(blocks, pageHeights);
-    mergeCrossPageTables(blocks, pageHeights);
+    mergeCrossPageTables(blocks, pageHeights, wrapLexicon);
+    if (_optionalChain([options, 'optionalAccess', _251 => _251.removeHeaderFooter]) !== false && parsedPageCount >= 3) {
+      const boxes = _chunkWDQTFAXYcjs.removeHeaderFooterBlocks.call(void 0, blocks, pageHeights, warnings, noteMarks, true);
+      for (let ri = boxes.length - 1; ri >= 0; ri--) blocks.splice(boxes[ri], 1);
+    }
     splitContactTables(blocks);
-    if (!_optionalChain([options, 'optionalAccess', _253 => _253.keepTrailingEmptyCols])) _chunkMH6MAL64cjs.trimTrailingEmptyTableCols.call(void 0, blocks);
-    _chunkMH6MAL64cjs.joinPageBreakWraps.call(void 0, blocks, wrapLexicon);
+    if (!_optionalChain([options, 'optionalAccess', _252 => _252.keepTrailingEmptyCols])) _chunkWDQTFAXYcjs.trimTrailingEmptyTableCols.call(void 0, blocks);
+    _chunkWDQTFAXYcjs.joinPageBreakWraps.call(void 0, blocks, wrapLexicon);
     injectPageImageBlocks(blocks, pageImageBlocks);
-    if (_optionalChain([options, 'optionalAccess', _254 => _254.formulaOcr]) && formulaBuffer) {
+    if (_optionalChain([options, 'optionalAccess', _253 => _253.formulaOcr]) && formulaBuffer) {
       try {
         await applyFormulaOcr(formulaBuffer, blocks, pageFilter, effectivePageCount, warnings, options.onProgress);
       } catch (e) {
@@ -2851,38 +2991,38 @@ async function parsePdfDocument(buffer, options) {
         });
       }
     }
-    const medianFontSize = _chunkMH6MAL64cjs.computeMedianFontSizeFromFreq.call(void 0, fontSizeFreq);
+    const medianFontSize = _chunkWDQTFAXYcjs.computeMedianFontSizeFromFreq.call(void 0, fontSizeFreq);
     if (medianFontSize > 0) {
-      _chunkMH6MAL64cjs.detectHeadings.call(void 0, blocks, medianFontSize);
-    }
-    _chunkMH6MAL64cjs.detectDocumentStyleHeadings.call(void 0, blocks);
-    _chunkMH6MAL64cjs.detectTypographyHeadings.call(void 0, blocks);
-    _chunkMH6MAL64cjs.detectSiblingStyleHeadings.call(void 0, blocks);
-    _chunkMH6MAL64cjs.detectRepeatedPageLabels.call(void 0, blocks);
-    _chunkMH6MAL64cjs.detectPageLeadHeadings.call(void 0, blocks);
-    _chunkMH6MAL64cjs.mergeStackedHeadingLines.call(void 0, blocks, medianFontSize);
-    _chunkMH6MAL64cjs.refineDocumentStyleHeadings.call(void 0, blocks);
-    _chunkMH6MAL64cjs.detectMarkerHeadings.call(void 0, blocks);
+      _chunkWDQTFAXYcjs.detectHeadings.call(void 0, blocks, medianFontSize);
+    }
+    _chunkWDQTFAXYcjs.detectDocumentStyleHeadings.call(void 0, blocks);
+    _chunkWDQTFAXYcjs.detectTypographyHeadings.call(void 0, blocks);
+    _chunkWDQTFAXYcjs.detectSiblingStyleHeadings.call(void 0, blocks);
+    _chunkWDQTFAXYcjs.detectRepeatedPageLabels.call(void 0, blocks);
+    _chunkWDQTFAXYcjs.detectPageLeadHeadings.call(void 0, blocks);
+    _chunkWDQTFAXYcjs.mergeStackedHeadingLines.call(void 0, blocks, medianFontSize);
+    _chunkWDQTFAXYcjs.refineDocumentStyleHeadings.call(void 0, blocks);
+    _chunkWDQTFAXYcjs.detectMarkerHeadings.call(void 0, blocks);
     demoteNonHeadingRoles(blocks, pageHeights, faceNames);
-    _chunkMH6MAL64cjs.detectTableCaptions.call(void 0, blocks);
-    _chunkMH6MAL64cjs.detectKoreanListBlocks.call(void 0, blocks);
+    _chunkWDQTFAXYcjs.detectTableCaptions.call(void 0, blocks);
+    _chunkWDQTFAXYcjs.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);
     let outBlocks = splitSingleCellTables(blocks);
     outBlocks = inlineFootnotes(outBlocks, noteMarks);
     outBlocks = relocateEndnotes(outBlocks);
-    if (_optionalChain([options, 'optionalAccess', _255 => _255.removeHeaderFooter]) !== false) outBlocks = outBlocks.filter((b) => {
+    if (_optionalChain([options, 'optionalAccess', _254 => _254.removeHeaderFooter]) !== false) outBlocks = outBlocks.filter((b) => {
       if (b.type !== "paragraph" || !/^\s*\d{1,4}\s*$/.test(_nullishCoalesce(b.text, () => ( "")))) return true;
       const h = pageHeights.get(_nullishCoalesce(b.pageNumber, () => ( 0)));
       if (!b.bbox || !h) return false;
       return b.bbox.y > h * 0.1 && b.bbox.y + b.bbox.height < h * 0.9;
     });
     joinLatinCellWraps(outBlocks);
-    const finishMarkdown = (bs) => mergeLinkRuns(cleanPdfText(_chunkDRY3L5KQcjs.blocksToMarkdown.call(void 0, bs), { keepLoneNumbers: true }));
+    const finishMarkdown = (bs) => mergeLinkRuns(cleanPdfText(_chunkXTFN5WKScjs.blocksToMarkdown.call(void 0, bs), { keepLoneNumbers: true }));
     let markdown = finishMarkdown(outBlocks);
     return {
       markdown,
-      pages: _chunkDRY3L5KQcjs.blocksToPages.call(void 0, outBlocks, finishMarkdown),
+      pages: _chunkXTFN5WKScjs.blocksToPages.call(void 0, outBlocks, finishMarkdown),
       blocks,
       metadata,
       outline: outline.length > 0 ? outline : void 0,
@@ -2900,7 +3040,7 @@ async function parsePdfDocument(buffer, options) {
 async function extractPdfMetadata(doc, metadata) {
   try {
     const result = await doc.getMetadata();
-    if (!_optionalChain([result, 'optionalAccess', _256 => _256.info])) return;
+    if (!_optionalChain([result, 'optionalAccess', _255 => _255.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();
@@ -2951,5 +3091,6 @@ function marginStamp(items) {
 
 
 
-exports.cleanPdfText = cleanPdfText; exports.detectKoreanListBlocks = _chunkMH6MAL64cjs.detectKoreanListBlocks; exports.detectTableCaptions = _chunkMH6MAL64cjs.detectTableCaptions; exports.extractPdfMetadataOnly = extractPdfMetadataOnly; exports.mergeCrossPageTables = mergeCrossPageTables; exports.parsePdfDocument = parsePdfDocument; exports.removeHeaderFooterBlocks = _chunkMH6MAL64cjs.removeHeaderFooterBlocks;
-//# sourceMappingURL=parser-YAQV7FBO.cjs.map
+
+exports.cleanPdfText = cleanPdfText; exports.detectKoreanListBlocks = _chunkWDQTFAXYcjs.detectKoreanListBlocks; exports.detectTableCaptions = _chunkWDQTFAXYcjs.detectTableCaptions; exports.extractPdfMetadataOnly = extractPdfMetadataOnly; exports.mergeCrossPageTables = mergeCrossPageTables; exports.ocrImageRegions = ocrImageRegions; exports.parsePdfDocument = parsePdfDocument; exports.removeHeaderFooterBlocks = _chunkWDQTFAXYcjs.removeHeaderFooterBlocks;
+//# sourceMappingURL=parser-2LU4EVU3.cjs.map