kordoc 4.14.2 → 4.14.3

package/dist/{model-bundle-FSPZZ2EM.js → model-bundle-57CLEBCO.js}

@@ -1,14 +1,14 @@
 #!/usr/bin/env node
 import {
   ALL_OCR_MODELS
-} from "./chunk-ZWD734EY.js";
+} from "./chunk-D6KQKMU3.js";
 import {
   ALL_FORMULA_MODELS,
   getModelStatusIn,
   getModelsDir
-} from "./chunk-NS55SM7A.js";
-import "./chunk-GDW4YP7K.js";
-import "./chunk-UET462HB.js";
+} from "./chunk-O2PGQNYE.js";
+import "./chunk-FAZR34CA.js";
+import "./chunk-I4NNYWIR.js";
 
 // src/shared/model-bundle.ts
 import { copyFile, mkdir } from "fs/promises";
@@ -101,4 +101,4 @@ export {
   importModels,
   modelCacheStatus
 };
-//# sourceMappingURL=model-bundle-FSPZZ2EM.js.map
+//# sourceMappingURL=model-bundle-57CLEBCO.js.map

package/dist/{models-CJPAQK5R.js → models-7ELIFTD7.js}

@@ -9,10 +9,10 @@ import {
   getOcrModelsDir,
   ocrModelPath,
   parseCharacterDict
-} from "./chunk-ZWD734EY.js";
-import "./chunk-NS55SM7A.js";
-import "./chunk-GDW4YP7K.js";
-import "./chunk-UET462HB.js";
+} from "./chunk-D6KQKMU3.js";
+import "./chunk-O2PGQNYE.js";
+import "./chunk-FAZR34CA.js";
+import "./chunk-I4NNYWIR.js";
 export {
   ALL_OCR_MODELS,
   OCR_DET_MODEL,
@@ -24,4 +24,4 @@ export {
   ocrModelPath,
   parseCharacterDict
 };
-//# sourceMappingURL=models-CJPAQK5R.js.map
+//# sourceMappingURL=models-7ELIFTD7.js.map

package/dist/{parser-SCUTZAIJ.js → parser-4NKAYWAX.js}

@@ -1,5 +1,11 @@
 #!/usr/bin/env node
 import {
+  CELL_LINES,
+  CLIP_TABLES,
+  EMPTY_PARTS,
+  FILLER_CELLS,
+  IMAGE_CELLS,
+  TABLE_COLXS,
   collapseEvenSpacing,
   computeMedianFontSizeFromFreq,
   detectHeadings,
@@ -9,24 +15,25 @@ import {
   extractImageRegions,
   extractPageBlocksWithLines,
   filterHiddenText,
-  mergeCrossPageTables,
   normalizeItems,
-  removeHeaderFooterBlocks
-} from "./chunk-3DUADAY2.js";
+  removeHeaderFooterBlocks,
+  trimTrailingEmptyTableCols
+} from "./chunk-A275ZQLX.js";
 import {
   encodePng
 } from "./chunk-7FCNOKOR.js";
-import "./chunk-7IAQ4HOK.js";
+import "./chunk-NQCUAGLR.js";
 import {
   parsePageRange
 } from "./chunk-MOL7MDBG.js";
 import {
+  WINGDINGS,
   blocksToMarkdown
-} from "./chunk-QC5T6D4J.js";
+} from "./chunk-QGW7NMTX.js";
 import {
   KordocError,
   sanitizeHref
-} from "./chunk-UET462HB.js";
+} from "./chunk-I4NNYWIR.js";
 
 // src/pdf/image-extract.ts
 import { OPS, ImageKind } from "pdfjs-dist/legacy/build/pdf.mjs";
@@ -110,7 +117,7 @@ function toRgba(img) {
   }
   return null;
 }
-async function extractPageImages(page, fnArray, argsArray, pageNumber, state, warnings) {
+async function extractPageImages(page, fnArray, argsArray, pageNumber, state, warnings, withBytes = true) {
   const blocks = [];
   const images = [];
   const pageSeenIds = /* @__PURE__ */ new Set();
@@ -149,12 +156,14 @@ async function extractPageImages(page, fnArray, argsArray, pageNumber, state, wa
     if (state.seen.has(hash)) continue;
     const rgba = toRgba(imgData);
     if (!rgba) continue;
-    const png = encodePng(w, h, rgba);
     state.imageIndex++;
-    state.totalBytes += png.length;
     const filename = `image_${String(state.imageIndex).padStart(3, "0")}.png`;
     state.seen.set(hash, filename);
-    images.push({ filename, data: png, mimeType: "image/png" });
+    if (withBytes) {
+      const png = encodePng(w, h, rgba);
+      state.totalBytes += png.length;
+      images.push({ filename, data: png, mimeType: "image/png" });
+    }
     blocks.push({ type: "image", text: filename, pageNumber });
   }
   return { blocks, images };
@@ -325,232 +334,241 @@ function summarizeDocumentQuality(pages) {
   };
 }
 
+// src/pdf/table-parts.ts
+var PART_COL_TOL = 1;
+function anchorsOf(table) {
+  const out = [];
+  const covered = /* @__PURE__ */ new Set();
+  for (let r = 0; r < table.rows; r++) {
+    for (let c = 0; c < table.cols; c++) {
+      if (covered.has(r * 1e5 + c)) continue;
+      const cell = table.cells[r]?.[c];
+      if (!cell) continue;
+      out.push({ r, c, rs: cell.rowSpan, cs: cell.colSpan, cell });
+      for (let dr = 0; dr < cell.rowSpan; dr++) {
+        for (let dc = 0; dc < cell.colSpan; dc++) if (dr || dc) covered.add((r + dr) * 1e5 + (c + dc));
+      }
+    }
+  }
+  return out;
+}
+function unionCoords(a, b) {
+  const out = [...a];
+  for (const v of b) if (!a.some((x) => Math.abs(x - v) <= PART_COL_TOL)) out.push(v);
+  return out.sort((x, y) => x - y);
+}
+var indexOf = (coords, x) => {
+  let best = -1, bestD = Infinity;
+  for (let i = 0; i < coords.length; i++) {
+    const d = Math.abs(coords[i] - x);
+    if (d < bestD) {
+      bestD = d;
+      best = i;
+    }
+  }
+  return bestD <= PART_COL_TOL ? best : -1;
+};
+var rowText = (anchors, r) => anchors.filter((a) => a.r === r).map((a) => a.cell.text.replace(/\s+/g, "")).join("|");
+function joinSplitParts(prev, pcx, curr, ccx) {
+  if (pcx.length !== prev.cols + 1 || ccx.length !== curr.cols + 1) return null;
+  const U = unionCoords(pcx, ccx);
+  const cols = U.length - 1;
+  if (cols < 1) return null;
+  const pa = anchorsOf(prev), ca = anchorsOf(curr);
+  let skip = 0;
+  while (skip < Math.min(3, curr.rows - 1, prev.rows) && rowText(ca, skip) !== "" && rowText(ca, skip) === rowText(pa, skip)) skip++;
+  const rows = prev.rows + curr.rows - skip;
+  const grid = Array.from({ length: rows }, () => Array.from({ length: cols }, () => ({ text: "", colSpan: 1, rowSpan: 1 })));
+  const owner = Array.from({ length: rows }, () => new Array(cols).fill(null));
+  const place = (a, x, rowOff) => {
+    const c1 = indexOf(U, x[a.c]), c2 = indexOf(U, x[a.c + a.cs]);
+    if (c1 < 0 || c2 <= c1) return false;
+    const r = a.r + rowOff;
+    const placed = { r, c: c1, rs: a.rs, cs: c2 - c1, cell: a.cell };
+    for (let dr = 0; dr < a.rs; dr++) for (let dc = c1; dc < c2; dc++) {
+      if (r + dr < rows) owner[r + dr][dc] = placed;
+    }
+    grid[r][c1] = { ...a.cell, colSpan: c2 - c1, rowSpan: a.rs };
+    const lines = CELL_LINES.get(a.cell);
+    if (lines) CELL_LINES.set(grid[r][c1], lines);
+    if (IMAGE_CELLS.has(a.cell)) IMAGE_CELLS.add(grid[r][c1]);
+    return true;
+  };
+  for (const a of pa) if (!FILLER_CELLS.has(a.cell) && !place(a, pcx, 0)) return null;
+  for (const a of ca) {
+    if (a.r < skip || FILLER_CELLS.has(a.cell)) continue;
+    if (!place(a, ccx, prev.rows - skip)) return null;
+  }
+  const first = prev.rows;
+  for (let c = 0; c < cols; c++) {
+    const above = owner[first - 1]?.[c];
+    if (!above) continue;
+    let r = first;
+    while (r < rows && owner[r][c] === null) r++;
+    if (r === first) continue;
+    const span = r - first;
+    let ok = true;
+    for (let dc = above.c; dc < above.c + above.cs && ok; dc++) {
+      for (let rr = first; rr < first + span; rr++) if (owner[rr][dc] !== null) {
+        ok = false;
+        break;
+      }
+    }
+    if (!ok) continue;
+    const newRs = above.rs + span;
+    grid[above.r][above.c].rowSpan = newRs;
+    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) FILLER_CELLS.add(grid[r][c]);
+  const table = { rows, cols, cells: grid, hasHeader: prev.hasHeader, ...prev.caption ? { caption: prev.caption } : {} };
+  if (rows > prev.rows) mergeSplitRow(table, owner, prev.rows, U);
+  return { table, colXs: U };
+}
+var FULL_LINE_TOL = 1;
+var FULL_LINE_MIN_FRAC = 0.5;
+var FULL_LINE_EDGE_MIN = 12;
+var SINGLE_LINE_WORD_GAP = 3.5;
+var SHORT_LINE_GAP = 1.5;
+function continuesAcross(u, d, x1, x2) {
+  const U = CELL_LINES.get(u), D = CELL_LINES.get(d);
+  if (!U?.length || !D?.length) return false;
+  const last = U[U.length - 1];
+  const fs = last.h || 10;
+  let minL = Infinity, maxR = -Infinity;
+  for (const l of U) {
+    minL = Math.min(minL, l.l);
+    maxR = Math.max(maxR, l.r);
+  }
+  for (const l of D) {
+    minL = Math.min(minL, l.l);
+    maxR = Math.max(maxR, l.r);
+  }
+  if (maxR - last.r > FULL_LINE_TOL * fs) return false;
+  if (U.length === 1 && x2 - last.r > Math.max(FULL_LINE_EDGE_MIN, SINGLE_LINE_WORD_GAP * fs)) return false;
+  if (last.r - last.l < FULL_LINE_MIN_FRAC * (x2 - x1)) return false;
+  const leftAligned = (l) => l.l - minL <= fs && maxR - l.r > SHORT_LINE_GAP * fs;
+  return U.some(leftAligned) || D.some(leftAligned);
+}
+function mergeSplitRow(table, owner, first, colXs) {
+  const last = first - 1;
+  const pairs = [];
+  for (let c = 0; c < table.cols; ) {
+    const a = owner[last][c], b = owner[first][c];
+    if (!a || !b) return;
+    if (a !== b) {
+      if (a.c !== b.c || a.cs !== b.cs || a.r + a.rs - 1 !== last || b.r !== first || b.rs !== 1) return;
+      pairs.push([a, b]);
+    }
+    c = a.c + a.cs;
+  }
+  if (!pairs.some(([u, d]) => continuesAcross(u.cell, d.cell, colXs[u.c], colXs[u.c + u.cs]))) return;
+  for (const [u, d] of pairs) {
+    const a = table.cells[u.r][u.c], b = table.cells[d.r][d.c];
+    if (b.text.trim()) a.text = a.text.trim() ? a.text + "\n" + b.text : b.text;
+    if (b.blocks?.length) a.blocks = [...a.blocks ?? [], ...b.blocks];
+  }
+  for (let r = 0; r < first; r++) for (let c = 0; c < table.cols; c++) {
+    const o = owner[r][c];
+    if (o && o.r === r && o.c === c && r + table.cells[r][c].rowSpan > first) table.cells[r][c].rowSpan--;
+  }
+  table.cells.splice(first, 1);
+  table.rows--;
+}
+var NEIGHBOR_TABLE_EPSILON = 0.2;
+function besideBoth(b, p, c) {
+  if (b.type === "table" || !b.bbox) return false;
+  const bx1 = b.bbox.x, bx2 = b.bbox.x + b.bbox.width;
+  return [p.bbox, c.bbox].every((t) => bx2 <= t.x + 1 || bx1 >= t.x + t.width - 1);
+}
+function mergeCrossPageTables(blocks, pageHeights) {
+  for (let i = blocks.length - 2; i >= 0; i--) {
+    const prev = blocks[i];
+    if (prev.type !== "table" || !prev.table || !prev.bbox || !prev.pageNumber) continue;
+    let j = i + 1;
+    while (j < blocks.length && blocks[j].type !== "table" && (blocks[j].pageNumber ?? 0) <= prev.pageNumber + 1) j++;
+    const curr = blocks[j];
+    if (!curr || curr.type !== "table" || !curr.table || !curr.bbox || curr.pageNumber !== prev.pageNumber + 1) continue;
+    const joined = j === i + 1 || blocks.slice(i + 1, j).every((b) => besideBoth(b, prev, curr)) ? looksContinued(prev, curr, pageHeights) ? joinClipParts(prev, curr, pageHeights) ?? false : null : null;
+    if (joined) {
+      blocks[i] = { ...prev, table: joined };
+      blocks.splice(j, 1);
+      continue;
+    }
+    if (EMPTY_PARTS.has(curr.table)) {
+      blocks.splice(j, 1);
+      i++;
+      continue;
+    }
+    if (joined === null) continue;
+    if (prev.table.cols !== curr.table.cols || 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 = TABLE_COLXS.get(prev.table), cx = TABLE_COLXS.get(curr.table);
+    if (px && cx && (px.length !== cx.length || px.some((x, k) => Math.abs(x - cx[k]) > CONTINUATION_COL_TOL))) continue;
+    let currCells = curr.table.cells;
+    if (currCells.length > 1 && prev.table.cells.length > 0 && rowTextsEqual(prev.table.cells[0], currCells[0])) {
+      currCells = currCells.slice(1);
+    }
+    if (currCells.length === 0) {
+      blocks.splice(j, 1);
+      continue;
+    }
+    const merged = {
+      rows: prev.table.rows + currCells.length,
+      cols: prev.table.cols,
+      cells: [...prev.table.cells, ...currCells],
+      hasHeader: prev.table.hasHeader,
+      caption: prev.table.caption
+    };
+    if (px ?? cx) TABLE_COLXS.set(merged, px ?? cx);
+    if (CLIP_TABLES.has(prev.table) || CLIP_TABLES.has(curr.table)) CLIP_TABLES.add(merged);
+    blocks[i] = { ...prev, table: merged };
+    blocks.splice(j, 1);
+  }
+  for (let i = blocks.length - 1; i >= 0; i--) {
+    const t = blocks[i].table;
+    if (blocks[i].type === "table" && t && EMPTY_PARTS.has(t)) blocks.splice(i, 1);
+  }
+}
+function joinClipParts(prev, curr, pageHeights) {
+  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;
+  const sameCols = px.length === cx.length && px.every((x, k) => Math.abs(x - cx[k]) <= CONTINUATION_COL_TOL);
+  if (!sameCols) {
+    if (!pageHeights?.get(prev.pageNumber) || !pageHeights?.get(curr.pageNumber)) return null;
+    if (Math.abs(px[px.length - 1] - cx[cx.length - 1]) > CONTINUATION_COL_TOL) return null;
+    if (!px.some((x) => Math.abs(x - cx[0]) <= CONTINUATION_COL_TOL)) return null;
+  }
+  const res = joinSplitParts(pt, px, ct, cx);
+  if (!res) return null;
+  TABLE_COLXS.set(res.table, res.colXs);
+  CLIP_TABLES.add(res.table);
+  return res.table;
+}
+var ANNEX_HEAD_RE = /^\s*[<\[(【]?\s*(?:붙\s*임|참\s*고|별\s*첨|별\s*지|별\s*표|첨\s*부|부\s*록)(?:\s*\d|\s*$|\s*[>\])】])|^\s*■/;
+function looksContinued(prev, curr, pageHeights) {
+  const ph = 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);
+}
+function rowTextsEqual(a, b) {
+  if (a.length !== b.length) return false;
+  const norm = (t) => t.replace(/\s+/g, "");
+  for (let i = 0; i < a.length; i++) {
+    if (norm(a[i].text) !== norm(b[i].text)) return false;
+  }
+  return a.some((c) => c.text.trim() !== "");
+}
+var CONTINUATION_COL_TOL = 2;
+var PAGE_EDGE_BAND = 0.16;
+
 // src/pdf/symbol-fonts.ts
-var WINGDINGS = [
-  "\u{1F589}",
-  "\u2702",
-  "\u2701",
-  "\u{1F453}",
-  "\u{1F56D}",
-  "\u{1F56E}",
-  "\u{1F56F}",
-  "\u{1F57F}",
-  "\u2706",
-  "\u{1F582}",
-  "\u{1F583}",
-  "\u{1F4EA}",
-  "\u{1F4EB}",
-  "\u{1F4EC}",
-  "\u{1F4ED}",
-  "\u{1F4C1}",
-  "\u{1F4C2}",
-  "\u{1F4C4}",
-  "\u{1F5CF}",
-  "\u{1F5D0}",
-  "\u{1F5C4}",
-  "\u231B",
-  "\u{1F5AE}",
-  "\u{1F5B0}",
-  "\u{1F5B2}",
-  "\u{1F5B3}",
-  "\u{1F5B4}",
-  "\u{1F5AB}",
-  "\u{1F5AC}",
-  "\u2707",
-  "\u270D",
-  "\u{1F58E}",
-  "\u270C",
-  "\u{1F44C}",
-  "\u{1F44D}",
-  "\u{1F44E}",
-  "\u261C",
-  "\u261E",
-  "\u261D",
-  "\u261F",
-  "\u{1F590}",
-  "\u263A",
-  "\u{1F610}",
-  "\u2639",
-  "\u{1F4A3}",
-  "\u2620",
-  "\u{1F3F3}",
-  "\u{1F3F1}",
-  "\u2708",
-  "\u263C",
-  "\u{1F4A7}",
-  "\u2744",
-  "\u{1F546}",
-  "\u271E",
-  "\u{1F548}",
-  "\u2720",
-  "\u2721",
-  "\u262A",
-  "\u262F",
-  "\u0950",
-  "\u2638",
-  "\u2648",
-  "\u2649",
-  "\u264A",
-  "\u264B",
-  "\u264C",
-  "\u264D",
-  "\u264E",
-  "\u264F",
-  "\u2650",
-  "\u2651",
-  "\u2652",
-  "\u2653",
-  "\u{1F670}",
-  "\u{1F675}",
-  "\u25CF",
-  "\u25CB",
-  "\u25A0",
-  "\u25A1",
-  "\u25A1",
-  "\u2751",
-  "\u2752",
-  "\u2B27",
-  "\u29EB",
-  "\u25C6",
-  "\u2756",
-  "\u2B25",
-  "\u2327",
-  "\u2BB9",
-  "\u2318",
-  "\u{1F3F5}",
-  "\u{1F3F6}",
-  "\u{1F676}",
-  "\u{1F677}",
-  "",
-  "\u24EA",
-  "\u2460",
-  "\u2461",
-  "\u2462",
-  "\u2463",
-  "\u2464",
-  "\u2465",
-  "\u2466",
-  "\u2467",
-  "\u2468",
-  "\u2469",
-  "\u24FF",
-  "\u2776",
-  "\u2777",
-  "\u2778",
-  "\u2779",
-  "\u277A",
-  "\u277B",
-  "\u277C",
-  "\u277D",
-  "\u277E",
-  "\u277F",
-  "\u{1F662}",
-  "\u{1F660}",
-  "\u{1F661}",
-  "\u{1F663}",
-  "\u{1F65E}",
-  "\u{1F65C}",
-  "\u{1F65D}",
-  "\u{1F65F}",
-  "\xB7",
-  "\u2022",
-  "\u25AA",
-  "\u26AA",
-  "\u25CB",
-  "\u25EF",
-  "\u25C9",
-  "\u25CE",
-  "\u{1F53F}",
-  "\u25AA",
-  "\u25FB",
-  "\u{1F7C2}",
-  "\u2726",
-  "\u2605",
-  "\u2736",
-  "\u2734",
-  "\u2739",
-  "\u2735",
-  "\u2BD0",
-  "\u2316",
-  "\u27E1",
-  "\u2311",
-  "\u2BD1",
-  "\u272A",
-  "\u2730",
-  "\u{1F550}",
-  "\u{1F551}",
-  "\u{1F552}",
-  "\u{1F553}",
-  "\u{1F554}",
-  "\u{1F555}",
-  "\u{1F556}",
-  "\u{1F557}",
-  "\u{1F558}",
-  "\u{1F559}",
-  "\u{1F55A}",
-  "\u{1F55B}",
-  "\u2BB0",
-  "\u2BB1",
-  "\u2BB2",
-  "\u2BB3",
-  "\u2BB4",
-  "\u2BB5",
-  "\u2BB6",
-  "\u2BB7",
-  "\u{1F66A}",
-  "\u{1F66B}",
-  "\u{1F655}",
-  "\u{1F654}",
-  "\u{1F657}",
-  "\u{1F656}",
-  "\u{1F650}",
-  "\u{1F651}",
-  "\u{1F652}",
-  "\u{1F653}",
-  "\u232B",
-  "\u2326",
-  "\u2B98",
-  "\u2B9A",
-  "\u2B99",
-  "\u2B9B",
-  "\u2B88",
-  "\u2B8A",
-  "\u2B89",
-  "\u2B8B",
-  "\u2190",
-  "\u2192",
-  "\u2191",
-  "\u2193",
-  "\u2196",
-  "\u2197",
-  "\u2199",
-  "\u2198",
-  "\u2B05",
-  "\u2794",
-  "\u2B06",
-  "\u2B07",
-  "\u2B09",
-  "\u2B08",
-  "\u2B0B",
-  "\u2B0A",
-  "\u21E6",
-  "\u21E8",
-  "\u21E7",
-  "\u21E9",
-  "\u2B04",
-  "\u21F3",
-  "\u2B00",
-  "\u2B01",
-  "\u2B03",
-  "\u2B02",
-  "\u{1F8AC}",
-  "\u{1F8AD}",
-  "\u2717",
-  "\u2714",
-  "\u2612",
-  "\u2611",
-  ""
-];
 var WINANSI_REVERSE = {
   8364: 128,
   8218: 130,
@@ -623,7 +641,8 @@ function remapSymbolFontItems(items, resolveFontName) {
 function normalizeAraea(text) {
   return text.replace(/(?<![\u1100-\u115F])\u119E/g, "\u318D");
 }
-var cleanChars = (text) => normalizeAraea(stripControlChars(text));
+var stripNoUnicodeGlyph = (text) => text.includes("\uF000") ? text.replace(/\uF000/g, "") : text;
+var cleanChars = (text) => normalizeAraea(stripNoUnicodeGlyph(stripControlChars(text)));
 function sanitizeBlockControlChars(blocks) {
   for (const b of blocks) {
     if (b.text) b.text = cleanChars(b.text);
@@ -737,7 +756,7 @@ function applyLinkAnnotations(items, annots) {
 
 // src/pdf/formula-ocr.ts
 async function applyFormulaOcr(buffer, blocks, pageFilter, effectivePageCount, warnings, _onProgress) {
-  const formulaMod = await import("./formula-5S7WWXGH.js");
+  const formulaMod = await import("./formula-NOIDZEBA.js");
   const { FormulaPipeline, ensureFormulaModels } = formulaMod;
   await ensureFormulaModels((p) => {
     if (p.phase === "download" && p.total) {
@@ -987,7 +1006,7 @@ async function parsePdfDocument(buffer, options) {
         const pageBlocks = extractPageBlocksWithLines(visible, i, opList, viewport.width, viewport.height, void 0, options?.tables !== false);
         for (const b of pageBlocks) blocks.push(b);
         try {
-          const { blocks: imgBlocks, images: pageImages } = await extractPageImages(page, opList.fnArray, opList.argsArray, i, imageState, warnings);
+          const { blocks: imgBlocks, images: pageImages } = await extractPageImages(page, opList.fnArray, opList.argsArray, i, imageState, warnings, options?.images !== false);
           if (imgBlocks.length > 0) pageImageBlocks.set(i, imgBlocks);
           extractedImages.push(...pageImages);
         } catch {
@@ -1029,7 +1048,7 @@ async function parsePdfDocument(buffer, options) {
       }
       if (targets.size > 0) {
         try {
-          const { runPdfOcr } = await import("./pdf-ocr-CPU55CVF.js");
+          const { runPdfOcr } = await import("./pdf-ocr-QX2GB33Q.js");
           const mode = typeof options.ocr === "function" ? options.ocr : "builtin";
           const ocrPageBlocks = await runPdfOcr(ocrBuffer, targets, mode, warnings, options.onProgress, options.tables !== false);
           if (ocrPageBlocks.size > 0) {
@@ -1084,7 +1103,8 @@ async function parsePdfDocument(buffer, options) {
         blocks.splice(removed[ri], 1);
       }
     }
-    mergeCrossPageTables(blocks);
+    mergeCrossPageTables(blocks, pageHeights);
+    if (!options?.keepTrailingEmptyCols) trimTrailingEmptyTableCols(blocks);
     injectPageImageBlocks(blocks, pageImageBlocks);
     if (options?.formulaOcr && formulaBuffer) {
       try {
@@ -1166,4 +1186,4 @@ export {
   parsePdfDocument,
   removeHeaderFooterBlocks
 };
-//# sourceMappingURL=parser-SCUTZAIJ.js.map
+//# sourceMappingURL=parser-4NKAYWAX.js.map