kordoc 4.16.0 → 4.16.2

package/dist/{parser-PXATL5DM.js → parser-ZQM3664Z.js}

@@ -4,7 +4,7 @@ import {
   blocksToPages,
   encodePng,
   escapeLiteralDollar
-} from "./chunk-NNYF2AGC.js";
+} from "./chunk-7KNGSQU7.js";
 import {
   hasRequestedPagesAfter,
   parsePageRange
@@ -41,15 +41,18 @@ import {
   ocrModelsCached,
   refineDocumentStyleHeadings,
   removeHeaderFooterBlocks,
+  startsNewItem,
   trimTrailingEmptyTableCols,
   wrapJoiner
-} from "./chunk-HIVK5V27.js";
-import "./chunk-QJERJZJ6.js";
-import "./chunk-PQRP4NSF.js";
+} from "./chunk-TXUKTOG2.js";
+import {
+  stripScriptTags
+} from "./chunk-3IB7LBHC.js";
+import "./chunk-F4NRLLY4.js";
 import {
   KordocError,
   sanitizeHref
-} from "./chunk-GOQBON77.js";
+} from "./chunk-MLIS253L.js";
 
 // src/pdf/ocr-region-merge.ts
 function mergeOcrImageRegions(blocks, page, regions, ocrBlocks) {
@@ -634,6 +637,40 @@ 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) => 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) {
@@ -670,7 +707,7 @@ 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;
@@ -726,13 +763,35 @@ function joinSplitParts(prev, pcx, curr, ccx, dx = 0, prevBottom) {
   }
   for (let r = 0; r < rows; r++) for (let c = 0; c < cols; c++) if (owner[r][c] === null) 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 = table.cells[u.r][u.c].text.trim().split(/\s+/).pop() ?? "", head = 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) {
@@ -740,18 +799,24 @@ 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 = CELL_LINES.get(a);
-    if (!U?.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 = CELL_LINES.get(a);
+      if (!U?.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;
@@ -787,11 +852,21 @@ 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) {
+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 = 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 = 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 (a.blocks?.length || b.blocks?.length) {
@@ -801,7 +876,22 @@ 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() || !!cell.blocks?.length || IMAGE_CELLS.has(cell);
-function mergeSplitRow(table, owner, first, colXs) {
+var CLAUSE_OPEN_ENDING = /(?:하는|되는|하고|하며|하여|되어|되고|되며|이며|으며|어야|아야|여야)$/;
+var batchim = (ch) => {
+  const k = ch.charCodeAt(0) - 44032;
+  return k >= 0 && k < 11172 && k % 28 !== 0;
+};
+function clauseContinues(u, d) {
+  const tail = u.text.trim().split(/\s+/).pop() ?? "", head = d.text.trim();
+  if (!/^[가-힣]{2,}$/.test(tail) || !/^[가-힣]/.test(head) || startsNewItem(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;
@@ -819,12 +909,12 @@ 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++) {
@@ -871,8 +961,8 @@ 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;
@@ -881,7 +971,7 @@ 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) ? 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) ? joinClipParts(prev, curr, pageHeights, lex) ?? false : null : null;
     if (joined) {
       placeJoined(blocks, i, j, joined);
       continue;
@@ -925,7 +1015,7 @@ function mergeCrossPageTables(blocks, pageHeights) {
     if (blocks[i].type === "table" && t && 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];
@@ -947,7 +1037,7 @@ 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);
@@ -956,12 +1046,12 @@ 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 (!CLIP_TABLES.has(pt) || !CLIP_TABLES.has(ct)) return null;
   const px = TABLE_COLXS.get(pt), cx = 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 (!pageHeights?.get(prev.pageNumber) || !pageHeights?.get(curr.pageNumber)) return null;
     dx = px[px.length - 1] - cx[cx.length - 1];
@@ -970,19 +1060,29 @@ 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(ct.cells[0] ?? []);
   }
-  const res = joinSplitParts(pt, px, ct, xs, dx, prev.bbox?.y);
-  if (!res || shifted && !res.split && !res.header) return null;
+  const res = joinSplitParts(pt, px, ct, xs, dx, prev.bbox?.y, lex);
+  if (!res || (shifted || foreign) && !res.split && !res.header) return null;
   TABLE_COLXS.set(res.table, res.colXs);
   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 = pageHeights?.get(prev.pageNumber), ch = pageHeights?.get(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 = curr.table.cells[0]?.find((c) => c.text.trim())?.text ?? "";
-  return !ANNEX_HEAD_RE.test(firstText);
+  const firstRow = curr.table.cells[0] ?? [];
+  const firstText = firstRow.find((c) => c.text.trim())?.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 = TABLE_COLXS.get(t);
@@ -995,6 +1095,7 @@ 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;
@@ -1841,6 +1942,52 @@ 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 === OPS5.setFont) font = String(args[0]);
+    else if (fn === OPS5.save || fn === OPS5.paintFormXObjectBegin) saved.push(font);
+    else if (fn === OPS5.restore || fn === OPS5.paintFormXObjectEnd) font = saved.pop() ?? font;
+    else if (fn === OPS5.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(normalizeUnicode3(g2.unicode));
+      }
+    }
+  }
+  const cursor = /* @__PURE__ */ new Map();
+  const failed = /* @__PURE__ */ new Set();
+  const marks = [];
+  for (const it of items) {
+    const f = it.fontName ?? "";
+    const glyphs = streams.get(f);
+    if (!glyphs || failed.has(f) || typeof it.str !== "string" || !it.str) continue;
+    let gi = 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(it.fontName ?? "")) it.synthetic = true;
+}
 
 // src/pdf/endnotes.ts
 var NOTE_MARK = /^\s*(문\d{1,5}[）)]|\d{1,5}\)|\(\d{1,5}\))/;
@@ -1978,6 +2125,7 @@ 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) {
@@ -2010,13 +2158,21 @@ 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;
 }
 
@@ -2184,7 +2340,9 @@ 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);
@@ -2261,7 +2419,7 @@ function cleanPdfText(text, opts) {
     }
     const mark = /^(?:#{1,6}|-) /.exec(line)?.[0] ?? "";
     return mark + collapseEvenSpacing(line.slice(mark.length), false);
-  }).replace(/\\~\\~/g, "~~").replace(/~~~~/g, "").replace(/<u>\s*<\/u>/g, "").replace(/\n{3,}/g, "\n\n").trim();
+  }).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();
@@ -2349,7 +2507,7 @@ function mergeLinkRuns(markdown) {
 
 // src/pdf/formula-ocr.ts
 async function applyFormulaOcr(buffer, blocks, pageFilter, effectivePageCount, warnings, _onProgress) {
-  const formulaMod = await import("./formula-FNQYHMIY.js");
+  const formulaMod = await import("./formula-EPKLPDU7.js");
   const { FormulaPipeline, ensureFormulaModels } = formulaMod;
   await ensureFormulaModels((p) => {
     if (p.phase === "download" && p.total) {
@@ -2489,42 +2647,15 @@ g.pdfjsWorker = pdfjsWorker;
 import { getDocument, GlobalWorkerOptions, OPS as OPS6 } from "pdfjs-dist/legacy/build/pdf.mjs";
 import { createRequire } from "module";
 import { dirname, join } from "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) => 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
 GlobalWorkerOptions.workerSrc = "";
 var MAX_PAGES = 5e3;
 var MAX_TOTAL_TEXT = 100 * 1024 * 1024;
@@ -2597,6 +2728,7 @@ 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();
@@ -2628,6 +2760,7 @@ async function parsePdfDocument(buffer, options) {
         const differencesOf = (loadedName) => fontObj(loadedName)?.differences;
         restoreNamedGlyphs(rawItems, rawOps.fnArray, rawOps.argsArray, differencesOf, (n) => fontObj(n)?.name);
         restoreTrackedSpacing(rawItems, rawOps.fnArray, rawOps.argsArray);
+        markSyntheticSpaces(rawItems, rawOps.fnArray, rawOps.argsArray);
         const occluded = occludedTextItems(rawItems, rawOps.fnArray, rawOps.argsArray);
         const items = normalizeItems(occluded.size ? rawItems.filter((it) => !occluded.has(it)) : rawItems);
         const filtered = filterHiddenText(items, pageW, pageH, viewX1, viewY1);
@@ -2691,6 +2824,7 @@ async function parsePdfDocument(buffer, options) {
                 const regions = uncoveredImageRegions.get(i) ?? [];
                 regions.push(r);
                 uncoveredImageRegions.set(i, regions);
+                if (large) largeImageRegions.add(r);
               }
             }
           }
@@ -2739,6 +2873,7 @@ async function parsePdfDocument(buffer, options) {
     const parsedPageCount = parsedPages || (pageFilter ? pageFilter.size : effectivePageCount);
     const isImageBased = pageFilter?.size !== 0 && totalChars / Math.max(parsedPageCount, 1) < 10;
     const ocrDone = /* @__PURE__ */ new Set();
+    const ocrRegions = options?.ocr === true || autoOcr ? ocrImageRegions(uncoveredImageRegions, largeImageRegions, options?.ocr === true) : /* @__PURE__ */ new Map();
     if (ocrBuffer) {
       const inScope = (p) => !pageFilter || pageFilter.has(p);
       const targets = /* @__PURE__ */ new Set();
@@ -2751,14 +2886,15 @@ async function parsePdfDocument(buffer, options) {
           if (autoOcr && pq.ocrReason !== "low_text" && pq.ocrReason !== "vector_text") continue;
           targets.add(pq.page);
         }
-        if (options?.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 import("./pdf-ocr-2TTO5PX2.js");
+          const { runPdfOcr } = await import("./pdf-ocr-6DEE23IP.js");
           const mode = typeof options?.ocr === "function" ? options.ocr : "builtin";
-          const regionPages = new Map([...uncoveredImageRegions].filter(([p]) => options?.ocr !== "force" && !isImageBased && !pageQuality.find((q) => q.page === p)?.needsOcr));
+          const regionPages = new Map([...ocrRegions].filter(([p]) => options?.ocr !== "force" && !isImageBased && !pageQuality.find((q) => q.page === p)?.needsOcr));
           const ocrPageBlocks = await runPdfOcr(ocrBuffer, targets, mode, warnings, options?.onProgress, options?.tables !== false, vectorPageOps, regionPages);
+          for (const obs of ocrPageBlocks.values()) stripScriptTags({ blocks: obs });
           if (ocrPageBlocks.size > 0) {
             const replacePages = /* @__PURE__ */ new Set();
             for (const [p, obs] of ocrPageBlocks) {
@@ -2768,7 +2904,7 @@ async function parsePdfDocument(buffer, options) {
                 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 (pageImageBlocks.get(p)?.length === 1) pageImageBlocks.delete(p);
@@ -2822,7 +2958,7 @@ async function parsePdfDocument(buffer, options) {
       }
     }
     if (options?.removeHeaderFooter !== false && parsedPageCount >= 3) {
-      const removed = removeHeaderFooterBlocks(blocks, pageHeights, warnings);
+      const removed = removeHeaderFooterBlocks(blocks, pageHeights, warnings, noteMarks);
       for (let ri = removed.length - 1; ri >= 0; ri--) {
         blocks.splice(removed[ri], 1);
       }
@@ -2833,7 +2969,11 @@ async function parsePdfDocument(buffer, options) {
       }
     }
     mergeContinuedCells(blocks, pageHeights);
-    mergeCrossPageTables(blocks, pageHeights);
+    mergeCrossPageTables(blocks, pageHeights, wrapLexicon);
+    if (options?.removeHeaderFooter !== false && parsedPageCount >= 3) {
+      const boxes = removeHeaderFooterBlocks(blocks, pageHeights, warnings, noteMarks, true);
+      for (let ri = boxes.length - 1; ri >= 0; ri--) blocks.splice(boxes[ri], 1);
+    }
     splitContactTables(blocks);
     if (!options?.keepTrailingEmptyCols) trimTrailingEmptyTableCols(blocks);
     joinPageBreakWraps(blocks, wrapLexicon);
@@ -2946,7 +3086,8 @@ export {
   detectTableCaptions,
   extractPdfMetadataOnly,
   mergeCrossPageTables,
+  ocrImageRegions,
   parsePdfDocument,
   removeHeaderFooterBlocks
 };
-//# sourceMappingURL=parser-PXATL5DM.js.map
+//# sourceMappingURL=parser-ZQM3664Z.js.map