kordoc 4.15.2 → 4.15.4

package/dist/{chunk-AS66S3HA.js → chunk-SIQAT7KT.js}

@@ -6,10 +6,10 @@ import {
   ensureOcrModels,
   getOcrModelsDir,
   parseCharacterDict
-} from "./chunk-VTUADXSS.js";
+} from "./chunk-IHVWE3C5.js";
 import {
   extractPageBlocksWithLines
-} from "./chunk-PTN7YG24.js";
+} from "./chunk-JO3MKOBE.js";
 
 // src/ocr/ruling-lines.ts
 var INK_LUMA_MAX = 205;
@@ -1307,4 +1307,4 @@ export {
   runPdfOcr,
   ocrItemsToBlocks
 };
-//# sourceMappingURL=chunk-AS66S3HA.js.map
+//# sourceMappingURL=chunk-SIQAT7KT.js.map

package/dist/{chunk-VSI5SRP2.cjs → chunk-SRITXAMQ.cjs}

@@ -6,7 +6,7 @@ var _chunkPY2M7YVNcjs = require('./chunk-PY2M7YVN.cjs');
 
 
 
-var _chunkVGU4FLLDcjs = require('./chunk-VGU4FLLD.cjs');
+var _chunkUUNEKQUUcjs = require('./chunk-UUNEKQUU.cjs');
 
 // src/pdf/image-regions.ts
 var _pdfmjs = require('pdfjs-dist/legacy/build/pdf.mjs');
@@ -824,9 +824,20 @@ function buildTableGrids(horizontals, verticals) {
     }
     const groupRadius = groupVertices.length > 0 ? groupVertices.reduce((max, v) => Math.max(max, v.radius), 1) : globalRadius;
     const coordMergeTol = Math.max(VERTEX_MERGE_FACTOR * groupRadius, MIN_COORD_MERGE_TOL);
+    const leftEdge = Math.min(...vLines.map((v) => v.x1));
+    const rightEdge = Math.max(...vLines.map((v) => v.x1));
+    const underlines = hLines.filter((h) => {
+      const crossings = /* @__PURE__ */ new Set();
+      for (const v of vLines) {
+        if (v.x1 >= h.x1 - CONNECT_TOL && v.x1 <= h.x2 + CONNECT_TOL && h.y1 >= v.y1 - CONNECT_TOL && h.y1 <= v.y2 + CONNECT_TOL) crossings.add(v.x1);
+      }
+      return crossings.size === 1 && Math.abs(h.x2 - rightEdge) <= CONNECT_TOL && h.x1 > (leftEdge + rightEdge) / 2 && h.x2 - h.x1 < (rightEdge - leftEdge) * 0.6;
+    });
+    const rowLines = underlines.length >= 3 && hLines.length - underlines.length <= 4 ? hLines.filter((h) => !underlines.includes(h)) : hLines;
+    const rowVertices = rowLines.length === hLines.length ? groupVertices : groupVertices.filter((v) => rowLines.some((h) => Math.abs(v.y - h.y1) <= CONNECT_TOL));
     const rawYs = [
-      ...hLines.map((l) => l.y1),
-      ...groupVertices.map((v) => v.y)
+      ...rowLines.map((l) => l.y1),
+      ...rowVertices.map((v) => v.y)
     ];
     const rowYs = clusterCoordinates(rawYs, coordMergeTol).sort((a, b) => b - a);
     const rawXs = [
@@ -861,7 +872,7 @@ function dropShadingClipGrids(clipGrids, lineGrids, fillRects, verticals = []) {
     if (!shaded) return true;
     const ruled = c.cells.every((cell) => cell.bbox.x2 >= c.bbox.x2 - SHADE_CLIP_TOL || verticals.some((v) => Math.abs(v.x1 - cell.bbox.x2) <= SHADE_CLIP_TOL * 2 && Math.min(v.y2, cell.bbox.y2) - Math.max(v.y1, cell.bbox.y1) >= (cell.bbox.y2 - cell.bbox.y1) * 0.75));
     if (!ruled) return true;
-    const host = lineGrids.find((l) => c.bbox.x1 >= l.bbox.x1 - SHADE_CLIP_TOL && c.bbox.x2 <= l.bbox.x2 + SHADE_CLIP_TOL && c.bbox.y1 >= l.bbox.y1 - SHADE_CLIP_TOL && c.bbox.y2 <= l.bbox.y2 + SHADE_CLIP_TOL && c.colXs.every((x) => l.colXs.some((lx) => near(lx, x))) && clipGrids.reduce((s, o) => s + inter(o.bbox, l.bbox), 0) <= area(l.bbox) * SHADE_CLIP_MAX_AREA);
+    const host = lineGrids.find((l) => c.bbox.x1 >= l.bbox.x1 - SHADE_CLIP_TOL && c.bbox.x2 <= l.bbox.x2 + SHADE_CLIP_TOL && c.bbox.y1 >= l.bbox.y1 - SHADE_CLIP_TOL && c.bbox.y2 <= l.bbox.y2 + SHADE_CLIP_TOL && c.colXs.every((x) => l.colXs.some((lx) => near(lx, x))) && (clipGrids.reduce((s, o) => s + inter(o.bbox, l.bbox), 0) <= area(l.bbox) * SHADE_CLIP_MAX_AREA || l.rowYs.length === c.rowYs.length + 1 && c.rowYs.every((y, i) => near(y, l.rowYs[i])) && l.colXs.length === c.colXs.length && c.colXs.every((x, i) => near(x, l.colXs[i])) && !clipGrids.some((o) => o !== c && inter(o.bbox, l.bbox) > 0)));
     return !host;
   });
 }
@@ -1148,10 +1159,28 @@ function buildClipCellGrids(rects, strokedH, strokedV, pageWidth, pageHeight, te
   const pageArea = pageWidth * pageHeight;
   const sameRect = (a, b) => Math.abs(a.x1 - b.x1) <= CLIP_EDGE_TOL && Math.abs(a.x2 - b.x2) <= CLIP_EDGE_TOL && Math.abs(a.y1 - b.y1) <= CLIP_EDGE_TOL && Math.abs(a.y2 - b.y2) <= CLIP_EDGE_TOL;
   const cells = [];
+  const buckets = /* @__PURE__ */ new Map();
   for (const r of rects) {
     if (r.x2 - r.x1 < CLIP_MIN_W2 || r.y2 - r.y1 < CLIP_MIN_H2) continue;
     if (pageArea > 0 && (r.x2 - r.x1) * (r.y2 - r.y1) >= pageArea * CLIP_MAX_PAGE_FRAC) continue;
-    if (!cells.some((c) => sameRect(c, r))) cells.push(r);
+    const bx = Math.floor(r.x1 / CLIP_EDGE_TOL), by = Math.floor(r.y1 / CLIP_EDGE_TOL);
+    let duplicate = false;
+    for (let dx = -1; dx <= 1 && !duplicate; dx++) {
+      const ys2 = buckets.get(bx + dx);
+      for (let dy = -1; dy <= 1 && !duplicate; dy++) {
+        duplicate = _nullishCoalesce(_optionalChain([ys2, 'optionalAccess', _12 => _12.get, 'call', _13 => _13(by + dy), 'optionalAccess', _14 => _14.some, 'call', _15 => _15((c) => sameRect(c, r))]), () => ( false));
+      }
+    }
+    if (duplicate) continue;
+    cells.push(r);
+    let ys = buckets.get(bx);
+    if (!ys) {
+      ys = /* @__PURE__ */ new Map();
+      buckets.set(bx, ys);
+    }
+    const row = ys.get(by);
+    if (row) row.push(r);
+    else ys.set(by, [r]);
   }
   if (cells.length < 1) return { grids: [], containers: [], page: { lastCells: [], clips: [] } };
   const parent = new Array(cells.length).fill(-1);
@@ -1210,9 +1239,32 @@ function buildClipCellGrids(rects, strokedH, strokedV, pageWidth, pageHeight, te
     }
     return i;
   };
+  const edgeBin = (v) => Math.floor(v / CLIP_ADJ_GAP);
+  const startsX = /* @__PURE__ */ new Map(), endsX = /* @__PURE__ */ new Map();
+  const startsY = /* @__PURE__ */ new Map(), endsY = /* @__PURE__ */ new Map();
+  const addEdge = (map, v, i) => {
+    const k = edgeBin(v), row = map.get(k);
+    if (row) row.push(i);
+    else map.set(k, [i]);
+  };
+  for (let i = 0; i < cells.length; i++) {
+    addEdge(startsX, cells[i].x1, i);
+    addEdge(endsX, cells[i].x2, i);
+    addEdge(startsY, cells[i].y1, i);
+    addEdge(endsY, cells[i].y2, i);
+  }
   for (let i = 0; i < cells.length; i++) {
     if (tableClip[i]) continue;
-    for (let j = i + 1; j < cells.length; j++) {
+    const candidates = /* @__PURE__ */ new Set();
+    const near = (map, v) => {
+      const k = edgeBin(v);
+      for (let d = -1; d <= 1; d++) for (const j of _nullishCoalesce(map.get(k + d), () => ( []))) if (j > i) candidates.add(j);
+    };
+    near(startsX, cells[i].x2);
+    near(endsX, cells[i].x1);
+    near(startsY, cells[i].y2);
+    near(endsY, cells[i].y1);
+    for (const j of [...candidates].sort((a, b) => a - b)) {
       if (tableClip[j] || parent[i] !== parent[j]) continue;
       if (adjacent(cells[i], cells[j])) {
         const ra = find(i), rb = find(j);
@@ -1233,7 +1285,7 @@ function buildClipCellGrids(rects, strokedH, strokedV, pageWidth, pageHeight, te
     else groups.set(r, [i]);
   }
   const containers = cells.filter((_, i) => isContainer[i]);
-  const groupSize = (i) => _nullishCoalesce(_optionalChain([groups, 'access', _12 => _12.get, 'call', _13 => _13(find(i)), 'optionalAccess', _14 => _14.length]), () => ( 0));
+  const groupSize = (i) => _nullishCoalesce(_optionalChain([groups, 'access', _16 => _16.get, 'call', _17 => _17(find(i)), 'optionalAccess', _18 => _18.length]), () => ( 0));
   const isGridMember = (i) => groupSize(i) >= CLIP_MIN_GROUP;
   const framed = (r) => edgeStroked(strokedH, "h", r.y1, r.x1, r.x2) && edgeStroked(strokedH, "h", r.y2, r.x1, r.x2) && edgeStroked(strokedV, "v", r.x1, r.y1, r.y2) && edgeStroked(strokedV, "v", r.x2, r.y1, r.y2);
   const titledFrame = (r) => r.y1 >= pageHeight * TITLED_FRAME_MIN_TOP && r.x2 - r.x1 >= pageWidth * TITLED_FRAME_MIN_WIDTH && textPoints.some((p) => p.y < r.y1 && p.y > pageHeight * HEADER_BAND && p.x >= r.x1 - CLIP_EDGE_TOL && p.x <= r.x2 + CLIP_EDGE_TOL) && textPoints.some((p) => p.x > r.x1 && p.x < r.x2 && p.y > r.y1 && p.y < r.y2);
@@ -1664,7 +1716,7 @@ var WrapLexicon = (_class = class {constructor() { _class.prototype.__init.call(
     const n = left.length;
     if (n >= 2 && right.length >= 2) {
       const k1 = pairKey(left.charCodeAt(n - 2), left.charCodeAt(n - 1)), k2 = pairKey(right.charCodeAt(0), right.charCodeAt(1));
-      const v = decideCounts(_nullishCoalesce(_optionalChain([this, 'access', _15 => _15.joined2, 'access', _16 => _16.get, 'call', _17 => _17(k1), 'optionalAccess', _18 => _18.get, 'call', _19 => _19(k2)]), () => ( 0)), _nullishCoalesce(_optionalChain([this, 'access', _20 => _20.spaced2, 'access', _21 => _21.get, 'call', _22 => _22(k1), 'optionalAccess', _23 => _23.get, 'call', _24 => _24(k2)]), () => ( 0)));
+      const v = decideCounts(_nullishCoalesce(_optionalChain([this, 'access', _19 => _19.joined2, 'access', _20 => _20.get, 'call', _21 => _21(k1), 'optionalAccess', _22 => _22.get, 'call', _23 => _23(k2)]), () => ( 0)), _nullishCoalesce(_optionalChain([this, 'access', _24 => _24.spaced2, 'access', _25 => _25.get, 'call', _26 => _26(k1), 'optionalAccess', _27 => _27.get, 'call', _28 => _28(k2)]), () => ( 0)));
       if (v !== null) return v;
     }
     if (n && right.length) {
@@ -1735,7 +1787,7 @@ function wrapJoiner(prevText, nextText, lex) {
   if (endsAsWord(left)) return " ";
   if (/[·ㆍ‧-]/.test(a) && !/[·ㆍ‧…]{2}$/.test(prev) && /[가-힣A-Za-z0-9]/.test(b)) return "";
   if (/[A-Za-z]/.test(a) && /[A-Za-z]/.test(b) || /\d/.test(a) && /\d/.test(b)) return " ";
-  const ev = _optionalChain([lex, 'optionalAccess', _25 => _25.evidence, 'call', _26 => _26(left, right)]);
+  const ev = _optionalChain([lex, 'optionalAccess', _29 => _29.evidence, 'call', _30 => _30(left, right)]);
   if (ev != null) return ev;
   if (lex && /[가-힣A-Za-z0-9]$/.test(left) && /^[가-힣A-Za-z0-9]/.test(right) && lex.joinedWord(left, right.replace(/[.,)」』’”;:]+$/, ""))) return "";
   if (/\d$/.test(left) && COUNTER.test(right)) return "";
@@ -1757,7 +1809,7 @@ function joinPageBreakWraps(blocks, lex) {
     if (a.type !== "paragraph" && a.type !== "list" || b.type !== "paragraph") continue;
     const last = a.bbox && PARA_LAST_LINE.get(a.bbox), fs = last ? last.fontSize : 0;
     if (!last || fs <= 0 || (_nullishCoalesce(pageRight.get(a.pageNumber), () => ( Infinity))) - last.right >= BODY_FULL_TOL * fs) continue;
-    if (last.width < BODY_MIN_WIDTH_EM * fs || Math.abs((_nullishCoalesce(_optionalChain([b, 'access', _27 => _27.style, 'optionalAccess', _28 => _28.fontSize]), () => ( 0))) - fs) > 0.15 * fs) continue;
+    if (last.width < BODY_MIN_WIDTH_EM * fs || Math.abs((_nullishCoalesce(_optionalChain([b, 'access', _31 => _31.style, 'optionalAccess', _32 => _32.fontSize]), () => ( 0))) - fs) > 0.15 * fs) continue;
     if (startsNewItem(a.text, b.text)) continue;
     a.text += wrapJoiner(a.text, b.text, lex) + b.text;
     const bl = b.bbox && PARA_LAST_LINE.get(b.bbox);
@@ -2051,7 +2103,7 @@ function mergeCellTextLines(textLines, wrap) {
 }
 
 // src/pdf/undersegmented.ts
-var MAX_UNDERSEGMENTED_ROWS = 2;
+var MAX_UNDERSEGMENTED_ROWS = 5;
 var MIN_UNDERSEGMENTED_COLUMNS = 3;
 var MIN_UNDERSEGMENTED_TEXT_LINES = 8;
 var MIN_ROW_BAND_MISMATCH = 2;
@@ -2099,11 +2151,30 @@ function groupItemsToVisualLines(items) {
   lines.push(cur);
   return lines;
 }
-function normalizeUndersegmentedTable(originalCells, colXs, items) {
+function normalizeUndersegmentedTable(originalCells, colXs, items, rowYs) {
   const numRows = originalCells.length;
   const numCols = colXs.length - 1;
   if (numRows > MAX_UNDERSEGMENTED_ROWS || numCols < MIN_UNDERSEGMENTED_COLUMNS) return null;
   if (items.length === 0) return null;
+  if (numRows >= 3) {
+    if (_optionalChain([rowYs, 'optionalAccess', _33 => _33.length]) !== numRows + 1) return null;
+    const rebuilt2 = [];
+    let expanded = 0;
+    for (let r = 0; r < numRows; r++) {
+      const row = originalCells[r].map((cell) => cell.text);
+      if (r === 0 || numRows === 3 && r === 1) {
+        rebuilt2.push(row);
+        continue;
+      }
+      const rowItems = items.filter((item) => itemCenterY(item) <= rowYs[r] + 1 && itemCenterY(item) >= rowYs[r + 1] - 1);
+      const body = normalizeUndersegmentedTable([originalCells[r]], colXs, rowItems);
+      if (body) {
+        rebuilt2.push(...body);
+        expanded++;
+      } else rebuilt2.push(row);
+    }
+    return expanded ? rebuilt2 : null;
+  }
   const itemsByCol = Array.from({ length: numCols }, () => []);
   for (const item of items) {
     if (!item.text.trim()) continue;
@@ -2235,7 +2306,7 @@ function dominantStyle(items) {
     }
   }
   if (dominantSize === 0) return void 0;
-  const fontName = _optionalChain([items, 'access', _29 => _29.find, 'call', _30 => _30((i) => i.fontSize === dominantSize), 'optionalAccess', _31 => _31.fontName]) || void 0;
+  const fontName = _optionalChain([items, 'access', _34 => _34.find, 'call', _35 => _35((i) => i.fontSize === dominantSize), 'optionalAccess', _36 => _36.fontName]) || void 0;
   return fontName ? { fontSize: dominantSize, fontName } : { fontSize: dominantSize };
 }
 function normalizeItems(rawItems) {
@@ -2344,7 +2415,7 @@ function splitAroundOverlays(c, overlays) {
   const ov = overlays.filter((o) => {
     for (let at = c.text.indexOf(o.text); at >= 0; at = c.text.indexOf(o.text, at + 1)) {
       const prefix = c.text.slice(0, at);
-      const est = c.x + glyphUnits(prefix.replace(/\s+/g, "")) * a0 + (_nullishCoalesce(_optionalChain([prefix, 'access', _32 => _32.match, 'call', _33 => _33(/\s+/g), 'optionalAccess', _34 => _34.length]), () => ( 0))) * ws;
+      const est = c.x + glyphUnits(prefix.replace(/\s+/g, "")) * a0 + (_nullishCoalesce(_optionalChain([prefix, 'access', _37 => _37.match, 'call', _38 => _38(/\s+/g), 'optionalAccess', _39 => _39.length]), () => ( 0))) * ws;
       if (Math.abs(est - o.x) <= OVERLAY_DUP_EM * fs) return false;
     }
     return true;
@@ -2457,7 +2528,7 @@ function mergeSuperscriptLines(lines) {
   return result;
 }
 function mergeLineSimple(items) {
-  if (items.length <= 1) return _optionalChain([items, 'access', _35 => _35[0], 'optionalAccess', _36 => _36.text]) || "";
+  if (items.length <= 1) return _optionalChain([items, 'access', _40 => _40[0], 'optionalAccess', _41 => _41.text]) || "";
   const sorted = sortLineByX([...items]);
   const isEvenSpaced = detectEvenSpacedItems(sorted);
   let result = sorted[0].text;
@@ -2506,7 +2577,7 @@ function computeMedianFontSizeFromFreq(freq) {
 }
 function detectHeadings(blocks, medianFontSize) {
   for (const block of blocks) {
-    if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _37 => _37.style, 'optionalAccess', _38 => _38.fontSize])) continue;
+    if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _42 => _42.style, 'optionalAccess', _43 => _43.fontSize])) continue;
     const text = block.text.trim();
     if (text.length === 0 || text.length > 200) continue;
     if (/^\d+$/.test(text)) continue;
@@ -2522,6 +2593,43 @@ function detectHeadings(blocks, medianFontSize) {
     }
   }
 }
+function detectTypographyHeadings(blocks) {
+  const byPage = /* @__PURE__ */ new Map();
+  for (const block of blocks) {
+    if (!block.pageNumber) continue;
+    const page = _nullishCoalesce(byPage.get(block.pageNumber), () => ( []));
+    page.push(block);
+    byPage.set(block.pageNumber, page);
+  }
+  for (const page of byPage.values()) {
+    if (page.some((block) => block.type === "heading")) continue;
+    const faceWeight = /* @__PURE__ */ new Map();
+    for (const block of page) {
+      if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _44 => _44.style, 'optionalAccess', _45 => _45.fontName])) continue;
+      const face = block.style.fontName;
+      faceWeight.set(face, (_nullishCoalesce(faceWeight.get(face), () => ( 0))) + block.text.length);
+    }
+    const bodyFace = _optionalChain([[...faceWeight], 'access', _46 => _46.sort, 'call', _47 => _47((a, b) => b[1] - a[1]), 'access', _48 => _48[0], 'optionalAccess', _49 => _49[0]]);
+    if (!bodyFace || (_nullishCoalesce(faceWeight.get(bodyFace), () => ( 0))) < 250) continue;
+    for (let i = 0; i < page.length; i++) {
+      const block = page[i];
+      const { text, bbox, style } = block;
+      if (block.type !== "paragraph" || !text || !bbox || !_optionalChain([style, 'optionalAccess', _50 => _50.fontName]) || !style.fontSize) continue;
+      const title = text.trim();
+      const numbered = /^\d+(?:\.\d+)*\.?\s+[A-Z가-힣]/.test(title);
+      if (style.fontName === bodyFace || title.length < 3 || title.length > 120 || /^\d+$/.test(title) || /^(?:table|figure|fig\.?|표|그림)\s*\d/i.test(title) || /^(?:doi:|https?:|[•●○▪▫])/i.test(title) || /^(?:over|under)\s+\d+$/i.test(title) || /^[\d\s.,:%+\-–]+$/.test(title) || bbox.height > style.fontSize * (numbered ? 2.4 : 1.6) || title.includes("\n")) continue;
+      const centerX = bbox.x + bbox.width / 2;
+      const nearby = page.filter((other) => other !== block && other.bbox && other.type !== "image" && other.type !== "separator" && other.bbox.x < centerX && centerX < other.bbox.x + other.bbox.width);
+      const above = nearby.filter((other) => other.bbox.y >= bbox.y + bbox.height).sort((a, b) => a.bbox.y - b.bbox.y)[0];
+      const below = nearby.filter((other) => other.bbox.y + other.bbox.height <= bbox.y).sort((a, b) => b.bbox.y - a.bbox.y)[0];
+      const gapAbove = above ? above.bbox.y - bbox.y - bbox.height : Infinity;
+      const gapBelow = below ? bbox.y - below.bbox.y - below.bbox.height : Infinity;
+      if (Math.max(gapAbove, gapBelow) < style.fontSize * (numbered ? 0.7 : 1.2)) continue;
+      block.type = "heading";
+      block.level = 2;
+    }
+  }
+}
 function shouldDemoteTable(table) {
   const allCells = table.cells.flatMap((row) => row.map((c) => c.text.trim())).filter(Boolean);
   const allText = allCells.join(" ");
@@ -2567,7 +2675,7 @@ function detectMarkerHeadings(blocks) {
       block.level = 4;
       continue;
     }
-    if (/^[가-힣]{2,6}$/.test(text) && _optionalChain([block, 'access', _39 => _39.style, 'optionalAccess', _40 => _40.fontSize])) {
+    if (/^[가-힣]{2,6}$/.test(text) && _optionalChain([block, 'access', _51 => _51.style, 'optionalAccess', _52 => _52.fontSize])) {
       const prev = blocks[i - 1];
       const next = blocks[i + 1];
       const prevIsStructural = !prev || prev.type === "table" || prev.type === "heading" || prev.type === "separator";
@@ -2822,7 +2930,7 @@ function removeHeaderFooterBlocks(blocks, pageHeights, warnings) {
   const bottomEntries = [];
   for (let bi = 0; bi < blocks.length; bi++) {
     const b = blocks[bi];
-    if (!b.bbox || !b.pageNumber || !_optionalChain([b, 'access', _41 => _41.text, 'optionalAccess', _42 => _42.trim, 'call', _43 => _43()])) continue;
+    if (!b.bbox || !b.pageNumber || !_optionalChain([b, 'access', _53 => _53.text, 'optionalAccess', _54 => _54.trim, 'call', _55 => _55()])) continue;
     const ph = pageHeights.get(b.bbox.page) || pageHeights.get(b.pageNumber);
     if (!ph) continue;
     const blockTop = ph - (b.bbox.y + b.bbox.height);
@@ -2845,7 +2953,7 @@ function removeHeaderFooterBlocks(blocks, pageHeights, warnings) {
     }
     const repeatedPatterns = /* @__PURE__ */ new Set();
     for (const [p, count] of patternCount) {
-      if (count >= MIN_REPEAT && (_nullishCoalesce(_optionalChain([patternPages, 'access', _44 => _44.get, 'call', _45 => _45(p), 'optionalAccess', _46 => _46.size]), () => ( 0))) >= MIN_REPEAT) {
+      if (count >= MIN_REPEAT && (_nullishCoalesce(_optionalChain([patternPages, 'access', _56 => _56.get, 'call', _57 => _57(p), 'optionalAccess', _58 => _58.size]), () => ( 0))) >= MIN_REPEAT) {
         repeatedPatterns.add(p);
       }
     }
@@ -2865,7 +2973,7 @@ function removeHeaderFooterBlocks(blocks, pageHeights, warnings) {
         if (bi === e.blockIdx) continue;
         const o = blocks[bi];
         if (!o.bbox || o.bbox.page !== cb.page) continue;
-        const oNorm = _optionalChain([o, 'access', _47 => _47.text, 'optionalAccess', _48 => _48.trim, 'call', _49 => _49(), 'access', _50 => _50.replace, 'call', _51 => _51(/\d+/g, "#")]);
+        const oNorm = _optionalChain([o, 'access', _59 => _59.text, 'optionalAccess', _60 => _60.trim, 'call', _61 => _61(), 'access', _62 => _62.replace, 'call', _63 => _63(/\d+/g, "#")]);
         if (oNorm && repeatedPatterns.has(oNorm)) continue;
         const overlap2 = Math.min(cb.y + cb.height, o.bbox.y + o.bbox.height) - Math.max(cb.y, o.bbox.y);
         if (overlap2 >= cb.height * 0.5) {
@@ -2908,7 +3016,7 @@ function recordCellLines(cell, items) {
 function markImageCell(cell) {
   IMAGE_CELLS.add(cell);
 }
-var emptyCell = (cell) => !cell || !_optionalChain([cell, 'access', _52 => _52.text, 'optionalAccess', _53 => _53.trim, 'call', _54 => _54()]) && !_optionalChain([cell, 'access', _55 => _55.blocks, 'optionalAccess', _56 => _56.length]) && !IMAGE_CELLS.has(cell);
+var emptyCell = (cell) => !cell || !_optionalChain([cell, 'access', _64 => _64.text, 'optionalAccess', _65 => _65.trim, 'call', _66 => _66()]) && !_optionalChain([cell, 'access', _67 => _67.blocks, 'optionalAccess', _68 => _68.length]) && !IMAGE_CELLS.has(cell);
 function trimTable(table) {
   let cols = table.cols;
   while (cols > 0 && table.cells.every((row) => emptyCell(row[cols - 1]))) cols--;
@@ -2924,7 +3032,7 @@ function trimTable(table) {
 function trimTrailingEmptyTableCols(blocks) {
   for (const b of blocks) {
     if (b.type !== "table" || !b.table) continue;
-    for (const row of b.table.cells) for (const cell of row) if (_optionalChain([cell, 'access', _57 => _57.blocks, 'optionalAccess', _58 => _58.length])) trimTrailingEmptyTableCols(cell.blocks);
+    for (const row of b.table.cells) for (const cell of row) if (_optionalChain([cell, 'access', _69 => _69.blocks, 'optionalAccess', _70 => _70.length])) trimTrailingEmptyTableCols(cell.blocks);
     trimTable(b.table);
   }
 }
@@ -2983,12 +3091,12 @@ function isTwoColumnProse(r) {
   if (t.rows < 8) return false;
   const dense = [];
   for (let c = 0; c < t.cols; c++) {
-    const filled = t.cells.filter((row) => _optionalChain([row, 'access', _59 => _59[c], 'optionalAccess', _60 => _60.text, 'access', _61 => _61.trim, 'call', _62 => _62()])).length;
+    const filled = t.cells.filter((row) => _optionalChain([row, 'access', _71 => _71[c], 'optionalAccess', _72 => _72.text, 'access', _73 => _73.trim, 'call', _74 => _74()])).length;
     if (filled / t.rows >= 0.3) dense.push(c);
   }
   if (dense.length !== 2) return false;
   for (const c of dense) {
-    const lens = t.cells.map((row) => _nullishCoalesce(_optionalChain([row, 'access', _63 => _63[c], 'optionalAccess', _64 => _64.text, 'access', _65 => _65.replace, 'call', _66 => _66(/\s+/g, ""), 'access', _67 => _67.length]), () => ( 0))).filter((n) => n > 0);
+    const lens = t.cells.map((row) => _nullishCoalesce(_optionalChain([row, 'access', _75 => _75[c], 'optionalAccess', _76 => _76.text, 'access', _77 => _77.replace, 'call', _78 => _78(/\s+/g, ""), 'access', _79 => _79.length]), () => ( 0))).filter((n) => n > 0);
     const avg = lens.reduce((s, v) => s + v, 0) / (lens.length || 1);
     if (avg < 12) return false;
   }
@@ -3463,7 +3571,7 @@ function buildClusterTable(rows, columns, pageNum) {
     const nonEmptyCols = cells[r].filter((c) => c.text.trim()).length;
     if (nonEmptyCols !== 1) continue;
     if (cells[r][0].text.trim() !== "") continue;
-    const contentText = _optionalChain([cells, 'access', _68 => _68[r], 'access', _69 => _69.find, 'call', _70 => _70((c) => c.text.trim()), 'optionalAccess', _71 => _71.text, 'access', _72 => _72.trim, 'call', _73 => _73()]) || "";
+    const contentText = _optionalChain([cells, 'access', _80 => _80[r], 'access', _81 => _81.find, 'call', _82 => _82((c) => c.text.trim()), 'optionalAccess', _83 => _83.text, 'access', _84 => _84.trim, 'call', _85 => _85()]) || "";
     if (/^[○●▶\-·]/.test(contentText)) continue;
     for (let pr = r - 1; pr >= 0; pr--) {
       if (cells[pr].some((c) => c.text.trim())) {
@@ -3852,14 +3960,14 @@ function isProseSpread(items) {
   for (let i = 1; i < sorted.length; i++) {
     gaps.push(sorted[i].x - (sorted[i - 1].x + sorted[i - 1].w));
   }
-  const maxGap = _chunkVGU4FLLDcjs.safeMax.call(void 0, gaps);
+  const maxGap = _chunkUUNEKQUUcjs.safeMax.call(void 0, gaps);
   const avgLen = items.reduce((s, i) => s + i.text.length, 0) / items.length;
   return maxGap < 40 && avgLen < 5;
 }
 function detectColumns(yLines) {
   const allItems = yLines.flat();
   if (allItems.length === 0) return null;
-  const pageWidth = _chunkVGU4FLLDcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkVGU4FLLDcjs.safeMin.call(void 0, allItems.map((i) => i.x));
+  const pageWidth = _chunkUUNEKQUUcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkUUNEKQUUcjs.safeMin.call(void 0, allItems.map((i) => i.x));
   if (pageWidth < 100) return null;
   let bigoLineIdx = -1;
   for (let i = 0; i < yLines.length; i++) {
@@ -4187,7 +4295,7 @@ function extractPageBlocksWithLines(items, pageNum, opList, pageWidth, pageHeigh
   for (const line of streamLines(items)) lex.addLine(mergeLineSimple(line));
   const extracted = extractLines(opList.fnArray, opList.argsArray);
   let { horizontals, verticals } = extracted;
-  const prevPage = _optionalChain([carry, 'optionalAccess', _74 => _74.page]) === pageNum - 1 ? carry.clip : void 0;
+  const prevPage = _optionalChain([carry, 'optionalAccess', _86 => _86.page]) === pageNum - 1 ? carry.clip : void 0;
   const clipResult = detectTables ? buildClipCellGrids(extracted.clipRects, horizontals, verticals, pageWidth, pageHeight, items.map((it) => ({ x: it.x + it.w / 2, y: it.y + it.h / 2 })), extracted.fillRects, prevPage) : { grids: [], containers: [], page: void 0 };
   if (carry) {
     carry.page = pageNum;
@@ -4422,8 +4530,21 @@ function extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals, i
     let finalGrid = irGrid;
     let finalRows = numRows;
     let rebuiltUsed = false;
-    if (!grid.cells && numRows <= 2 && numCols >= 3 && !nestedAttached) {
-      const rebuilt = normalizeUndersegmentedTable(irGrid, grid.colXs, textItems);
+    let unitLine = [];
+    let unitText = "";
+    if (!grid.cells && numRows >= 3 && numRows <= 5 && numCols >= 3) {
+      const nearby = items.filter((item) => item.y >= grid.bbox.y2 && item.y - grid.bbox.y2 <= 18 && item.x >= grid.bbox.x1 - 3 && item.x + item.w <= grid.bbox.x2 + 3);
+      for (const y of [...new Set(nearby.map((item) => Math.round(item.y)))].sort((a, b) => a - b)) {
+        const line = nearby.filter((item) => Math.abs(item.y - y) <= 1).sort((a, b) => a.x - b.x);
+        const text = line.map((item) => item.text).join("");
+        if (!/^\s*\(\s*단위\s*[:：]/.test(text)) continue;
+        unitLine = line;
+        unitText = text;
+        break;
+      }
+    }
+    if (!grid.cells && numRows <= 5 && numCols >= 3 && !nestedAttached && (numRows <= 2 || unitLine.length > 0)) {
+      const rebuilt = normalizeUndersegmentedTable(irGrid, grid.colXs, textItems, grid.rowYs);
       if (rebuilt) {
         rebuiltUsed = true;
         finalGrid = rebuilt.map((row) => row.map((rawText) => ({ text: cleanCellText(rawText), colSpan: 1, rowSpan: 1 })));
@@ -4435,6 +4556,11 @@ function extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals, i
         if (!assignedItems.has(textItems[ti])) usedItems.delete(tableItems[ti]);
       }
     }
+    if (unitLine.length > 0 && rebuiltUsed && !/^\s*\(\s*단위\s*[:：]/.test(_nullishCoalesce(_optionalChain([finalGrid, 'access', _87 => _87[0], 'optionalAccess', _88 => _88[0], 'optionalAccess', _89 => _89.text]), () => ( "")))) {
+      finalGrid.unshift(Array.from({ length: numCols }, (_, c) => ({ text: c === 0 ? cleanCellText(unitText) : "", colSpan: c === 0 ? numCols : 1, rowSpan: 1 })));
+      finalRows++;
+      for (const item of unitLine) usedItems.add(item);
+    }
     const irTable = {
       rows: finalRows,
       cols: numCols,
@@ -4517,7 +4643,7 @@ function extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals, i
     if (gutterX !== null) {
       const gx = gutterX;
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkVGU4FLLDcjs.safeMax.call(void 0, allY) - _chunkVGU4FLLDcjs.safeMin.call(void 0, allY);
+      const pageH = _chunkUUNEKQUUcjs.safeMax.call(void 0, allY) - _chunkUUNEKQUUcjs.safeMin.call(void 0, allY);
       const gapThreshold = Math.max(15, pageH * 0.03);
       const sides = [
         remaining.filter((i) => i.x + i.w <= gx),
@@ -4537,7 +4663,7 @@ function extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals, i
       finalTextBlocks = detectListBlocks(textBlocks);
     } else {
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkVGU4FLLDcjs.safeMax.call(void 0, allY) - _chunkVGU4FLLDcjs.safeMin.call(void 0, allY);
+      const pageH = _chunkUUNEKQUUcjs.safeMax.call(void 0, allY) - _chunkUUNEKQUUcjs.safeMin.call(void 0, allY);
       const groups = xyCutOrder(remaining, Math.max(15, pageH * 0.03));
       const textBlocks = [];
       for (const group of groups) {
@@ -4671,7 +4797,7 @@ function pushLineParagraphs(out, yLines, pageNum, lex) {
   const lines = yLines.map((items) => ({ items, text: mergeLineSimple(items) })).filter((l) => l.text.trim());
   const geo = lines.map((l) => {
     const b = computeBBox(l.items, pageNum);
-    return { text: l.text, left: b.x, right: b.x + b.width, y: l.items.reduce((s, i) => s + i.y, 0) / l.items.length, fontSize: _nullishCoalesce(_optionalChain([dominantStyle, 'call', _75 => _75(l.items), 'optionalAccess', _76 => _76.fontSize]), () => ( 0)) };
+    return { text: l.text, left: b.x, right: b.x + b.width, y: l.items.reduce((s, i) => s + i.y, 0) / l.items.length, fontSize: _nullishCoalesce(_optionalChain([dominantStyle, 'call', _90 => _90(l.items), 'optionalAccess', _91 => _91.fontSize]), () => ( 0)) };
   });
   const joins = bodyLineJoins(geo, lex);
   for (let i = 0; i < lines.length; ) {
@@ -4687,6 +4813,43 @@ function pushLineParagraphs(out, yLines, pageNum, lex) {
     i++;
   }
 }
+function columnTextToBlocks(text, pageNum, bbox, style) {
+  const lines = text.split("\n");
+  const blocks = [];
+  const prose = [];
+  const flushProse = () => {
+    const content = prose.join("\n").trim();
+    if (content) blocks.push({ type: "paragraph", text: content, pageNumber: pageNum, bbox, style });
+    prose.length = 0;
+  };
+  const cells = (line) => line.trim().slice(1, -1).split(/(?<!\\)\|/).map((c) => c.trim().replace(/\\\|/g, "|"));
+  const tableLine = (line) => /^\|.*\|\s*$/.test(line.trim());
+  for (let i = 0; i < lines.length; ) {
+    if (tableLine(lines[i]) && i + 2 < lines.length && tableLine(lines[i + 1]) && cells(lines[i + 1]).every((c) => /^:?-{3,}:?$/.test(c))) {
+      let end = i + 2;
+      while (end < lines.length && tableLine(lines[end])) end++;
+      const rows = [cells(lines[i]), ...lines.slice(i + 2, end).map(cells)];
+      const cols = rows[0].length;
+      const values = rows.slice(1).flat().filter(Boolean);
+      const numeric = values.filter((value) => value.length <= 24 && /\d/.test(value)).length;
+      if (cols >= 3 && rows.length >= 3 && rows.every((row) => row.length === cols) && rows[0].every((value) => value.length <= 80) && numeric >= 3 && numeric / values.length >= 0.25) {
+        flushProse();
+        blocks.push({ type: "table", pageNumber: pageNum, bbox, table: {
+          rows: rows.length,
+          cols,
+          hasHeader: true,
+          cells: rows.map((row) => row.map((value) => ({ text: value, rowSpan: 1, colSpan: 1 })))
+        } });
+        i = end;
+        continue;
+      }
+    }
+    prose.push(lines[i]);
+    i++;
+  }
+  flushProse();
+  return blocks;
+}
 function extractPageBlocksFallback(items, pageNum, fullPage = false, detectTables = true, lex) {
   if (items.length === 0) return [];
   const blocks = [];
@@ -4729,10 +4892,10 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false, detectTable
     if (columns && columns.length >= 3) {
       const tableText = extractWithColumns(allYLines, columns);
       const bbox = computeBBox(items, pageNum);
-      blocks.push({ type: "paragraph", text: tableText, pageNumber: pageNum, bbox, style: dominantStyle(items) });
+      blocks.push(...columnTextToBlocks(tableText, pageNum, bbox, dominantStyle(items)));
     } else {
       const allY = items.map((i) => i.y);
-      const pageHeight = _chunkVGU4FLLDcjs.safeMax.call(void 0, allY) - _chunkVGU4FLLDcjs.safeMin.call(void 0, allY);
+      const pageHeight = _chunkUUNEKQUUcjs.safeMax.call(void 0, allY) - _chunkUUNEKQUUcjs.safeMin.call(void 0, allY);
       const gapThreshold = Math.max(15, pageHeight * 0.03);
       const orderedGroups = proseCutX !== null ? splitTwoColumnProse(items, proseCutX) : xyCutOrder(items, gapThreshold);
       for (const group of orderedGroups) {
@@ -4742,7 +4905,7 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false, detectTable
         if (groupColumns && groupColumns.length >= 3) {
           const tableText = extractWithColumns(yLines, groupColumns);
           const bbox = computeBBox(group, pageNum);
-          blocks.push({ type: "paragraph", text: tableText, pageNumber: pageNum, bbox, style: dominantStyle(group) });
+          blocks.push(...columnTextToBlocks(tableText, pageNum, bbox, dominantStyle(group)));
         } else {
           pushLineParagraphs(blocks, yLines, pageNum, lex);
         }
@@ -4775,5 +4938,6 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false, detectTable
 
 
 
-exports.extractImageRegions = extractImageRegions; exports.filterHiddenText = filterHiddenText; exports.collapseEvenSpacing = collapseEvenSpacing; exports.normalizeItems = normalizeItems; exports.WrapLexicon = WrapLexicon; exports.wrapJoiner = wrapJoiner; exports.joinPageBreakWraps = joinPageBreakWraps; exports.computeMedianFontSizeFromFreq = computeMedianFontSizeFromFreq; exports.detectHeadings = detectHeadings; exports.detectMarkerHeadings = detectMarkerHeadings; exports.detectTableCaptions = detectTableCaptions; exports.detectKoreanListBlocks = detectKoreanListBlocks; exports.removeHeaderFooterBlocks = removeHeaderFooterBlocks; exports.CLIP_TABLES = CLIP_TABLES; exports.TABLE_COLXS = TABLE_COLXS; exports.EMPTY_PARTS = EMPTY_PARTS; exports.IMAGE_CELLS = IMAGE_CELLS; exports.FILLER_CELLS = FILLER_CELLS; exports.CONT_PARTS = CONT_PARTS; exports.CELL_LINES = CELL_LINES; exports.trimTrailingEmptyTableCols = trimTrailingEmptyTableCols; exports.extractPageBlocksWithLines = extractPageBlocksWithLines;
-//# sourceMappingURL=chunk-VSI5SRP2.cjs.map
+
+exports.extractImageRegions = extractImageRegions; exports.filterHiddenText = filterHiddenText; exports.collapseEvenSpacing = collapseEvenSpacing; exports.normalizeItems = normalizeItems; exports.WrapLexicon = WrapLexicon; exports.wrapJoiner = wrapJoiner; exports.joinPageBreakWraps = joinPageBreakWraps; exports.computeMedianFontSizeFromFreq = computeMedianFontSizeFromFreq; exports.detectHeadings = detectHeadings; exports.detectTypographyHeadings = detectTypographyHeadings; exports.detectMarkerHeadings = detectMarkerHeadings; exports.detectTableCaptions = detectTableCaptions; exports.detectKoreanListBlocks = detectKoreanListBlocks; exports.removeHeaderFooterBlocks = removeHeaderFooterBlocks; exports.CLIP_TABLES = CLIP_TABLES; exports.TABLE_COLXS = TABLE_COLXS; exports.EMPTY_PARTS = EMPTY_PARTS; exports.IMAGE_CELLS = IMAGE_CELLS; exports.FILLER_CELLS = FILLER_CELLS; exports.CONT_PARTS = CONT_PARTS; exports.CELL_LINES = CELL_LINES; exports.trimTrailingEmptyTableCols = trimTrailingEmptyTableCols; exports.extractPageBlocksWithLines = extractPageBlocksWithLines;
+//# sourceMappingURL=chunk-SRITXAMQ.cjs.map