kordoc 4.18.0 → 4.18.2

package/dist/{chunk-ELTPM4HM.cjs → chunk-KDMAYOWH.cjs}

@@ -4,15 +4,15 @@
 
 
 
-var _chunkF4ESJBGUcjs = require('./chunk-F4ESJBGU.cjs');
+var _chunkQGFZU3JEcjs = require('./chunk-QGFZU3JE.cjs');
 
 
 
-var _chunkRLVALQZIcjs = require('./chunk-RLVALQZI.cjs');
+var _chunkYNLW6EGUcjs = require('./chunk-YNLW6EGU.cjs');
 
 
 
-var _chunkCKMGVW6Scjs = require('./chunk-CKMGVW6S.cjs');
+var _chunkUZGZRGYMcjs = require('./chunk-UZGZRGYM.cjs');
 
 // src/ocr/models.ts
 var _path = require('path');
@@ -40,10 +40,10 @@ var OCR_REC_DICT = {
 };
 var ALL_OCR_MODELS = [OCR_DET_MODEL, OCR_REC_MODEL, OCR_REC_DICT];
 function getOcrModelsDir() {
-  return _chunkRLVALQZIcjs.getModelsDir.call(void 0, "ppocr");
+  return _chunkYNLW6EGUcjs.getModelsDir.call(void 0, "ppocr");
 }
 async function ensureOcrModels(onProgress) {
-  return _chunkRLVALQZIcjs.ensureModelsIn.call(void 0, getOcrModelsDir(), ALL_OCR_MODELS, onProgress);
+  return _chunkYNLW6EGUcjs.ensureModelsIn.call(void 0, getOcrModelsDir(), ALL_OCR_MODELS, onProgress);
 }
 async function ocrModelsCached() {
   const dir = getOcrModelsDir();
@@ -2090,7 +2090,7 @@ function tagScripts(text, lines) {
     const [s, e, kind] = marks[m];
     out = out.slice(0, s) + `<${kind}>` + out.slice(s, e) + `</${kind}>` + out.slice(e);
   }
-  return _chunkF4ESJBGUcjs.tidyScriptTags.call(void 0, out);
+  return _chunkQGFZU3JEcjs.tidyScriptTags.call(void 0, out);
 }
 
 // src/pdf/line-wrap.ts
@@ -3477,7 +3477,9 @@ function findVerticalCutWithOutlierFilter(items, minGap) {
     const filtered = items.filter((i) => i.w >= narrowThreshold);
     if (filtered.length >= 2 && filtered.length < items.length && filtered.length >= items.length * 0.7) {
       const filteredCut = findVerticalCut(filtered);
-      if (filteredCut.gap > edgeCut.gap && filteredCut.gap >= minGap) {
+      const sameRow = (a, b) => Math.abs(a.y - b.y) <= Math.max(a.fontSize, b.fontSize) * 0.5;
+      const inLine = items.some((r) => r.w < narrowThreshold && r.x < filteredCut.position && r.x + r.w > filteredCut.position && filtered.some((o) => sameRow(o, r) && o.x + o.w <= r.x + 1) && filtered.some((o) => sameRow(o, r) && o.x >= r.x + r.w - 1));
+      if (filteredCut.gap > edgeCut.gap && filteredCut.gap >= minGap && !inLine) {
         return filteredCut;
       }
     }
@@ -3641,9 +3643,9 @@ function detectHeadings(blocks, medianFontSize) {
     if (/^\d+$/.test(text) && !isChapterNumber(block, blocks[bi + 1], medianFontSize)) continue;
     const ratio = block.style.fontSize / medianFontSize;
     let level = 0;
-    if (ratio >= _chunkF4ESJBGUcjs.HEADING_RATIO_H1) level = 1;
-    else if (ratio >= _chunkF4ESJBGUcjs.HEADING_RATIO_H2) level = 2;
-    else if (ratio >= _chunkF4ESJBGUcjs.HEADING_RATIO_H3) level = 3;
+    if (ratio >= _chunkQGFZU3JEcjs.HEADING_RATIO_H1) level = 1;
+    else if (ratio >= _chunkQGFZU3JEcjs.HEADING_RATIO_H2) level = 2;
+    else if (ratio >= _chunkQGFZU3JEcjs.HEADING_RATIO_H3) level = 3;
     if (level > 0) {
       block.type = "heading";
       block.level = level;
@@ -3656,7 +3658,7 @@ function isChapterNumber(block, next, medianFontSize) {
   const nextSize = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _94 => _94.style, 'optionalAccess', _95 => _95.fontSize]), () => ( 0));
   const text = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _96 => _96.text, 'optionalAccess', _97 => _97.trim, 'call', _98 => _98()]), () => ( ""));
   const a = block.bbox, b = _optionalChain([next, 'optionalAccess', _99 => _99.bbox]);
-  return /^\d{1,2}$/.test(_nullishCoalesce(_optionalChain([block, 'access', _100 => _100.text, 'optionalAccess', _101 => _101.trim, 'call', _102 => _102()]), () => ( ""))) && size >= medianFontSize * 2.5 && !!next && (next.type === "paragraph" || next.type === "heading") && next.pageNumber === block.pageNumber && nextSize >= medianFontSize * _chunkF4ESJBGUcjs.HEADING_RATIO_H1 && text.length > 0 && text.length <= 80 && !/^[\d\s.]+$/.test(text) && !!a && !!b && a.y > b.y && a.y - (b.y + b.height) <= size * 2;
+  return /^\d{1,2}$/.test(_nullishCoalesce(_optionalChain([block, 'access', _100 => _100.text, 'optionalAccess', _101 => _101.trim, 'call', _102 => _102()]), () => ( ""))) && size >= medianFontSize * 2.5 && !!next && (next.type === "paragraph" || next.type === "heading") && next.pageNumber === block.pageNumber && nextSize >= medianFontSize * _chunkQGFZU3JEcjs.HEADING_RATIO_H1 && text.length > 0 && text.length <= 80 && !/^[\d\s.]+$/.test(text) && !!a && !!b && a.y > b.y && a.y - (b.y + b.height) <= size * 2;
 }
 function mergeStackedHeadingLines(blocks, medianFontSize) {
   for (let i = 0; i < blocks.length - 1; ) {
@@ -4197,7 +4199,7 @@ function detectSpecialKoreanTables(blocks) {
 }
 function removeHeaderFooterBlocks(blocks, pageHeights, warnings, notes, tables = false) {
   const ZONE_RATIO = 0.12;
-  const MIN_REPEAT = 3;
+  const MIN_REPEAT2 = 3;
   const topEntries = [];
   const bottomEntries = [];
   for (let bi = 0; bi < blocks.length; bi++) {
@@ -4234,7 +4236,7 @@ function removeHeaderFooterBlocks(blocks, pageHeights, warnings, notes, tables =
       const pages = [..._nullishCoalesce(patternPages.get(p), () => ( []))];
       const span = pages.length ? Math.max(...pages) - Math.min(...pages) + 1 : 0;
       const pageNumbered = !tables && (_nullishCoalesce(_optionalChain([patternNumbers, 'access', _224 => _224.get, 'call', _225 => _225(p), 'optionalAccess', _226 => _226.size]), () => ( 0))) > 1;
-      if (count >= MIN_REPEAT && pages.length >= MIN_REPEAT && (pages.length >= span * 0.4 || pageNumbered)) {
+      if (count >= MIN_REPEAT2 && pages.length >= MIN_REPEAT2 && (pages.length >= span * 0.4 || pageNumbered)) {
         repeatedPatterns.add(p);
       }
     }
@@ -4382,6 +4384,78 @@ function mergeSliverColumns(grid, colXs, maxWidth) {
   return merged;
 }
 
+// src/pdf/side-tabs.ts
+var BAND = 0.12;
+var MAX_CHARS = 10;
+var MIN_REPEAT = 3;
+var RUN_GAP = 5;
+var TAB_CELL_CHARS = 24;
+var TAB_MIN_HEIGHT = 0.3;
+var SIDE_TAB_TABLES = /* @__PURE__ */ new WeakSet();
+function isSideTabTable(box, table, others, pageWidth, pageHeight) {
+  if (table.cols !== 1 || box.x2 - box.x1 > pageWidth * BAND || box.y2 - box.y1 < pageHeight * TAB_MIN_HEIGHT) return false;
+  const side = box.x1 >= pageWidth * (1 - BAND) ? "R" : box.x2 <= pageWidth * BAND ? "L" : null;
+  if (!side) return false;
+  const texts = table.cells.flat().map((c) => c.text.replace(/\s+/g, ""));
+  if (!texts.some(Boolean) || texts.some((t) => [...t].length > TAB_CELL_CHARS)) return false;
+  return others.every((it) => it.y < box.y1 || it.y > box.y2 || (side === "R" ? it.x + it.w <= box.x1 + 1 : it.x >= box.x2 - 1));
+}
+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', _250 => _250.text, 'optionalAccess', _251 => _251.trim, 'call', _252 => _252()]), () => ( ""))].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);
+    edges.set(b.pageNumber, e);
+  }
+  const q = (xs, p) => {
+    const v = [...xs].sort((a, b) => a - b);
+    return v[Math.min(v.length - 1, Math.floor(p * v.length))];
+  };
+  const body = /* @__PURE__ */ new Map();
+  for (const [page, e] of edges) body.set(page, { left: q(e.lefts, 0.2), right: q(e.rights, 0.8) });
+  const all = [...body.values()];
+  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', _253 => _253.text, 'optionalAccess', _254 => _254.trim, 'call', _255 => _255()]);
+    const area = _nullishCoalesce((b.pageNumber !== void 0 ? body.get(b.pageNumber) : void 0), () => ( fallback));
+    if (b.type === "table") {
+      if (!w || !b.bbox || !b.table || !SIDE_TAB_TABLES.has(b.table)) return null;
+      return `
${b.bbox.x >= w * (1 - BAND) ? "R" : "L"}${Math.round(b.bbox.x / 8)}`;
+    }
+    if (!w || !area || !b.bbox || !text || [...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;
+    return side && text.replace(/\s+/g, "");
+  };
+  const pages = /* @__PURE__ */ new Map();
+  for (const b of blocks) {
+    const k = key(b);
+    if (k) pages.set(k, (_nullishCoalesce(pages.get(k), () => ( /* @__PURE__ */ new Set()))).add(b.pageNumber));
+  }
+  const running = /* @__PURE__ */ new Set();
+  for (const [k, ps] of pages) {
+    const list = [...ps].sort((a, b) => a - b);
+    const steps = list.slice(1).map((p, i) => p - list[i]).sort((a, b) => a - b);
+    const gap = Math.max(RUN_GAP, (_nullishCoalesce(steps[steps.length >> 1], () => ( 1))) * 3);
+    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] > gap) flush();
+      run.push(p);
+    }
+    flush();
+  }
+  return running.size ? blocks.filter((b) => {
+    const k = key(b);
+    return !k || !running.has(`${b.pageNumber}\0${k}`);
+  }) : blocks;
+}
+
 // src/pdf/table-roles.ts
 var TOC_BLOCKS = /* @__PURE__ */ new WeakSet();
 var PAGE_LABEL = /^(?:\d{1,4}|[ivxlc]{1,7})$/i;
@@ -4432,7 +4506,7 @@ function isProseTable(table) {
   if (!(long > 0 && long >= total * 0.5 && longRows >= rows * 0.4)) return false;
   for (let c = 0; c < table.cols; c++) {
     const filled = table.cells.map((row) => row[c]).filter((cell) => cell && cell.colSpan < table.cols && cell.text.trim()).map((cell) => cell.text.trim());
-    if (filled.length >= 3 && filled.length >= rows * 0.6 && filled.every((text) => text.length <= 30 && (_nullishCoalesce(_optionalChain([text, 'access', _250 => _250.match, 'call', _251 => _251(/\p{L}/gu), 'optionalAccess', _252 => _252.length]), () => ( 0))) >= 2)) return false;
+    if (filled.length >= 3 && filled.length >= rows * 0.6 && filled.every((text) => text.length <= 30 && (_nullishCoalesce(_optionalChain([text, 'access', _256 => _256.match, 'call', _257 => _257(/\p{L}/gu), 'optionalAccess', _258 => _258.length]), () => ( 0))) >= 2)) return false;
   }
   return true;
 }
@@ -4459,7 +4533,7 @@ function isChartTable(table) {
   const numeric = texts.filter((text) => NUMERIC_CELL.test(text)).length;
   if (numeric < texts.length * 0.5) return false;
   for (let c = 0; c < table.cols; c++) {
-    const ticks = table.cells.flatMap((row) => (_nullishCoalesce(_optionalChain([row, 'access', _253 => _253[c], 'optionalAccess', _254 => _254.text]), () => ( ""))).split(/\s+/)).filter((token) => /^-?[\d,]+(?:\.\d+)?%?$/.test(token)).map((token) => Number(token.replace(/[,%]/g, "")));
+    const ticks = table.cells.flatMap((row) => (_nullishCoalesce(_optionalChain([row, 'access', _259 => _259[c], 'optionalAccess', _260 => _260.text]), () => ( ""))).split(/\s+/)).filter((token) => /^-?[\d,]+(?:\.\d+)?%?$/.test(token)).map((token) => Number(token.replace(/[,%]/g, "")));
     if (hasValueAxis(ticks)) return true;
   }
   const values = texts.filter((text) => text.split(/\s+/).every((token) => /^[-−–]?[\d,]*\.?\d+%?$/.test(token))).length;
@@ -4488,6 +4562,8 @@ var MIN_GAP_FACTOR = 2;
 var MIN_GAP_ABSOLUTE = 20;
 var MIN_COL_FILL_RATIO = 0.4;
 function detectClusterTables(items, pageNum, rejected) {
+  const tab = sideTabGlyphs(items);
+  if (tab.size) items = items.filter((i) => !tab.has(i));
   if (items.length < MIN_ROWS * MIN_COLS) return [];
   const { merged, originMap } = mergeEvenSpacedClusters(items);
   const rows = mergeOverlappingRows(groupByBaseline(merged));
@@ -4547,6 +4623,34 @@ function detectClusterTables(items, pageNum, rejected) {
     return !prose;
   });
 }
+var SIDE_TAB_MIN = 6;
+function sideTabGlyphs(items) {
+  const found = /* @__PURE__ */ new Set();
+  const singles = items.filter((i) => i.fontSize > 0 && [...i.text.trim()].length === 1);
+  if (singles.length < SIDE_TAB_MIN) return found;
+  const cols = [];
+  for (const it of [...singles].sort((a, b) => a.x - b.x)) {
+    const c = cols.find((col) => Math.abs(col[0].x - it.x) <= Math.max(1.5, it.fontSize * 0.25));
+    if (c) c.push(it);
+    else cols.push([it]);
+  }
+  for (const col of cols) {
+    if (col.length < SIDE_TAB_MIN) continue;
+    col.sort((a, b) => b.y - a.y);
+    const fs = [...col.map((i) => i.fontSize)].sort((a, b) => a - b)[col.length >> 1];
+    if (col.slice(1).some((it, k) => col[k].y - it.y > fs * 3)) continue;
+    const members = new Set(col);
+    const top = col[0].y + fs, bottom = col[col.length - 1].y - fs;
+    const others = items.filter((i) => !members.has(i) && i.y <= top && i.y >= bottom);
+    const left = Math.min(...col.map((i) => i.x)), right = Math.max(...col.map((i) => i.x + i.w));
+    const outside = others.every((i) => i.x >= right + fs) || others.every((i) => i.x + i.w <= left - fs);
+    if (!outside) continue;
+    const alone = col.filter((g) => !others.some((i) => Math.abs(i.y - g.y) <= Y_TOL)).length;
+    if (alone < col.length * 0.4) continue;
+    for (const g of col) found.add(g);
+  }
+  return found;
+}
 function detectCompactStatisticalTables(rows, pageNum) {
   const found = [];
   const parts = /* @__PURE__ */ new Map();
@@ -4613,7 +4717,7 @@ function detectAlignedTwoColumnTables(rows, pageNum) {
       continue;
     }
     const top = rows[i - 1];
-    const title = _optionalChain([top, 'optionalAccess', _255 => _255.items, 'access', _256 => _256.length]) === 1 && Math.abs(top.items[0].x - first[0].x) <= 5 && top.items[0].text.length <= 65 && top.y - first[0].y <= Math.max(24, first[0].fontSize * 2.8) ? top : void 0;
+    const title = _optionalChain([top, 'optionalAccess', _261 => _261.items, 'access', _262 => _262.length]) === 1 && Math.abs(top.items[0].x - first[0].x) <= 5 && top.items[0].text.length <= 65 && top.y - first[0].y <= Math.max(24, first[0].fontSize * 2.8) ? top : void 0;
     const tableRows = title ? [title, ...body] : body;
     const cells = tableRows.map((row, idx) => {
       if (title && idx === 0) return [{ text: row.items[0].text, colSpan: 2, rowSpan: 1 }, { text: "", colSpan: 1, rowSpan: 1 }];
@@ -4669,12 +4773,12 @@ 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', _257 => _257[c], 'optionalAccess', _258 => _258.text, 'access', _259 => _259.trim, 'call', _260 => _260()])).length;
+    const filled = t.cells.filter((row) => _optionalChain([row, 'access', _263 => _263[c], 'optionalAccess', _264 => _264.text, 'access', _265 => _265.trim, 'call', _266 => _266()])).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', _261 => _261[c], 'optionalAccess', _262 => _262.text, 'access', _263 => _263.replace, 'call', _264 => _264(/\s+/g, ""), 'access', _265 => _265.length]), () => ( 0))).filter((n) => n > 0);
+    const lens = t.cells.map((row) => _nullishCoalesce(_optionalChain([row, 'access', _267 => _267[c], 'optionalAccess', _268 => _268.text, 'access', _269 => _269.replace, 'call', _270 => _270(/\s+/g, ""), 'access', _271 => _271.length]), () => ( 0))).filter((n) => n > 0);
     const avg = lens.reduce((s, v) => s + v, 0) / (lens.length || 1);
     if (avg < 12) return false;
   }
@@ -5070,7 +5174,7 @@ function assignRowItems(items, columns, numCols, headerFace) {
   const runs = [];
   for (const it of sorted) {
     const last = runs[runs.length - 1];
-    const prev = _optionalChain([last, 'optionalAccess', _266 => _266[last.length - 1]]);
+    const prev = _optionalChain([last, 'optionalAccess', _272 => _272[last.length - 1]]);
     if (prev && it.x - (prev.x + prev.w) <= Math.max(0.5, it.fontSize * 0.05) && !it.hasSpaceBefore) last.push(it);
     else runs.push([it]);
   }
@@ -5144,7 +5248,7 @@ function buildClusterTable(rows, columns, pageNum) {
     () => Array.from({ length: numCols }, () => ({ text: "", colSpan: 1, rowSpan: 1 }))
   );
   const usedItems = /* @__PURE__ */ new Set();
-  const headFaces = new Set(_optionalChain([rows, 'access', _267 => _267[0], 'optionalAccess', _268 => _268.items, 'access', _269 => _269.map, 'call', _270 => _270((i) => i.fontName)]));
+  const headFaces = new Set(_optionalChain([rows, 'access', _273 => _273[0], 'optionalAccess', _274 => _274.items, 'access', _275 => _275.map, 'call', _276 => _276((i) => i.fontName)]));
   const headerFace = headFaces.size === 1 && rows.slice(1).every((row) => row.items.every((i) => !headFaces.has(i.fontName))) ? [...headFaces][0] : void 0;
   for (let r = 0; r < numRows; r++) {
     const row = rows[r];
@@ -5174,7 +5278,7 @@ 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', _271 => _271[r], 'access', _272 => _272.find, 'call', _273 => _273((c) => c.text.trim()), 'optionalAccess', _274 => _274.text, 'access', _275 => _275.trim, 'call', _276 => _276()]) || "";
+    const contentText = _optionalChain([cells, 'access', _277 => _277[r], 'access', _278 => _278.find, 'call', _279 => _279((c) => c.text.trim()), 'optionalAccess', _280 => _280.text, 'access', _281 => _281.trim, 'call', _282 => _282()]) || "";
     if (/^[○●▶\-·]/.test(contentText)) continue;
     for (let pr = r - 1; pr >= 0; pr--) {
       if (cells[pr].some((c) => c.text.trim())) {
@@ -5565,16 +5669,16 @@ 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 = _chunkCKMGVW6Scjs.safeMax.call(void 0, gaps);
+  const maxGap = _chunkUZGZRGYMcjs.safeMax.call(void 0, gaps);
   const avgLen = items.reduce((s, i) => s + i.text.length, 0) / items.length;
   if (items.length >= 4 && maxGap < 40 && avgLen < 5) return true;
-  const minGap = _chunkCKMGVW6Scjs.safeMin.call(void 0, gaps);
+  const minGap = _chunkUZGZRGYMcjs.safeMin.call(void 0, gaps);
   return minGap > 0 && maxGap - minGap <= Math.max(2, maxGap * 0.15) && items.filter((i) => /^\P{L}*\p{Ll}/u.test(i.text)).length * 2 >= items.length;
 }
 function detectColumns(yLines) {
   const allItems = yLines.flat();
   if (allItems.length === 0) return null;
-  const pageWidth = _chunkCKMGVW6Scjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkCKMGVW6Scjs.safeMin.call(void 0, allItems.map((i) => i.x));
+  const pageWidth = _chunkUZGZRGYMcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkUZGZRGYMcjs.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++) {
@@ -5873,7 +5977,7 @@ function hasBoxTopPair(line, maxH, horizontals, cellItems) {
     if (ow <= 0) continue;
     const overlap2 = Math.min(line.x2, o.x2) - Math.max(line.x1, o.x1);
     if (overlap2 / Math.max(w, ow) < UNDER_BOX_PAIR_OVERLAP) continue;
-    const covered = _nullishCoalesce(_optionalChain([cellItems, 'optionalAccess', _277 => _277.reduce, 'call', _278 => _278((sum, item) => {
+    const covered = _nullishCoalesce(_optionalChain([cellItems, 'optionalAccess', _283 => _283.reduce, 'call', _284 => _284((sum, item) => {
       const below = item.y - o.y1;
       if (below < 0.5 || below > 3 || !item.text.trim()) return sum;
       return sum + Math.max(0, Math.min(o.x2, item.x + item.w) - Math.max(o.x1, item.x));
@@ -6003,8 +6107,8 @@ function edgesOf(box, h, v, hNear = EDGE_NEAR2) {
 }
 function recordClipCellEdges(grids, horizontals, verticals, nonRules) {
   if (!grids.some((g) => g.cells)) return;
-  const h = rules(_optionalChain([nonRules, 'optionalAccess', _279 => _279.size]) ? horizontals.filter((l) => !nonRules.has(l)) : horizontals, "h");
-  const v = rules(_optionalChain([nonRules, 'optionalAccess', _280 => _280.size]) ? verticals.filter((l) => !nonRules.has(l)) : verticals, "v");
+  const h = rules(_optionalChain([nonRules, 'optionalAccess', _285 => _285.size]) ? horizontals.filter((l) => !nonRules.has(l)) : horizontals, "h");
+  const v = rules(_optionalChain([nonRules, 'optionalAccess', _286 => _286.size]) ? verticals.filter((l) => !nonRules.has(l)) : verticals, "v");
   for (const g of grids) for (const c of _nullishCoalesce(g.cells, () => ( []))) {
     const near = c.filler ? Math.min(EDGE_NEAR2, (c.bbox.y2 - c.bbox.y1) / 3) : EDGE_NEAR2;
     CLIP_CELL_EDGES.set(c, edgesOf(c.bbox, h, v, near));
@@ -6012,12 +6116,12 @@ function recordClipCellEdges(grids, horizontals, verticals, nonRules) {
 }
 function takeClipCellEdges(from, to) {
   const e = CLIP_CELL_EDGES.get(from);
-  if (e) _chunkF4ESJBGUcjs.CELL_EDGES.set(to, e);
+  if (e) _chunkQGFZU3JEcjs.CELL_EDGES.set(to, e);
 }
 function joinCellEdges(a, b) {
-  const u = _chunkF4ESJBGUcjs.CELL_EDGES.get(a), d = _chunkF4ESJBGUcjs.CELL_EDGES.get(b);
+  const u = _chunkQGFZU3JEcjs.CELL_EDGES.get(a), d = _chunkQGFZU3JEcjs.CELL_EDGES.get(b);
   if (!u || !d) return;
-  _chunkF4ESJBGUcjs.CELL_EDGES.set(a, { t: u.t, b: d.b, l: u.l || d.l, r: u.r || d.r });
+  _chunkQGFZU3JEcjs.CELL_EDGES.set(a, { t: u.t, b: d.b, l: u.l || d.l, r: u.r || d.r });
 }
 
 // src/pdf/table-unit-row.ts
@@ -6033,12 +6137,12 @@ function rebuildUnitLine(items, grid) {
 function prependUnitRow(cells, cols, line, usedItems) {
   const text = cleanCellText(line.map((item) => item.text).join(""));
   cells.unshift(Array.from({ length: cols }, (_, c) => ({ text: c === 0 ? text : "", colSpan: c === 0 ? cols : 1, rowSpan: 1 })));
-  for (const cell of cells[0]) _chunkF4ESJBGUcjs.CELL_EDGES.set(cell, NO_EDGES);
+  for (const cell of cells[0]) _chunkQGFZU3JEcjs.CELL_EDGES.set(cell, NO_EDGES);
   for (const item of line) usedItems.add(item);
 }
 function attachUnitRow(items, grid, cells, cols, usedItems) {
   const all = cells.flat();
-  if (all.filter((c) => c.text.trim()).length * 2 < all.length || UNIT.test(_nullishCoalesce(_optionalChain([cells, 'access', _281 => _281[0], 'optionalAccess', _282 => _282[0], 'optionalAccess', _283 => _283.text]), () => ( "")))) return false;
+  if (all.filter((c) => c.text.trim()).length * 2 < all.length || UNIT.test(_nullishCoalesce(_optionalChain([cells, 'access', _287 => _287[0], 'optionalAccess', _288 => _288[0], 'optionalAccess', _289 => _289.text]), () => ( "")))) return false;
   const above = items.filter((it) => !usedItems.has(it) && it.y >= grid.bbox.y2 && it.y - grid.bbox.y2 <= 8 && it.x >= grid.bbox.x1 - 3 && it.x + it.w <= grid.bbox.x2 + 3);
   let nearest = Infinity;
   for (const it of above) nearest = Math.min(nearest, it.y);
@@ -6578,7 +6682,7 @@ function stackedTableBands(items) {
   const caption = (start, end) => lines.slice(start, end).some((line) => /^Table\s+\d+\s*[:.]/i.test(mergeLineSimple(line).trim()));
   while (runs.length > 0 && !caption(runs[runs.length - 1].end, lines.length)) runs.pop();
   if (runs.length < 2) return null;
-  if (runs.some((run, i) => !caption(run.end, _nullishCoalesce(_optionalChain([runs, 'access', _284 => _284[i + 1], 'optionalAccess', _285 => _285.start]), () => ( lines.length))))) return null;
+  if (runs.some((run, i) => !caption(run.end, _nullishCoalesce(_optionalChain([runs, 'access', _290 => _290[i + 1], 'optionalAccess', _291 => _291.start]), () => ( lines.length))))) return null;
   const tables = runs.map((run) => lines.slice(run.start, run.end).flat());
   const between = runs.slice(0, -1).map((run, i) => lines.slice(run.end, runs[i + 1].start).flat());
   const tail = lines.slice(runs[runs.length - 1].end);
@@ -6734,6 +6838,11 @@ function extractPageBlocksWithLines(items, pageNum, opList, pageWidth, pageHeigh
     if (carry) carry.clip = void 0;
     return [];
   }
+  const tab = sideTabGlyphs(items);
+  if (tab.size) {
+    const glyphs = [...tab].sort((a, b) => b.y - a.y).map((g) => ({ type: "paragraph", text: g.text.trim(), pageNumber: pageNum, bbox: computeBBox([g], pageNum), style: dominantStyle([g]) }));
+    return [...glyphs, ...extractPageBlocksWithLines(items.filter((i) => !tab.has(i)), pageNum, opList, pageWidth, pageHeight, extraLines, detectTables, carry, lexicon)];
+  }
   const lex = _nullishCoalesce(lexicon, () => ( new WrapLexicon()));
   for (const line of streamLines(items)) lex.addLine(mergeLineSimple(line));
   const extracted = extractLines(opList.fnArray, opList.argsArray);
@@ -6742,7 +6851,7 @@ function extractPageBlocksWithLines(items, pageNum, opList, pageWidth, pageHeigh
   const filled = fillBlanks(items, horizontals, verticals);
   items = filled.items;
   horizontals = filled.horizontals;
-  const prevPage = _optionalChain([carry, 'optionalAccess', _286 => _286.page]) === pageNum - 1 ? carry.clip : void 0;
+  const prevPage = _optionalChain([carry, 'optionalAccess', _292 => _292.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;
@@ -7030,14 +7139,14 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
         if (!assignedItems.has(textItems[ti])) usedItems.delete(tableItems[ti]);
       }
     }
-    if (unitLine.length > 0 && rebuiltUsed && !/^\s*\(\s*단위\s*[:：]/.test(_nullishCoalesce(_optionalChain([finalGrid, 'access', _287 => _287[0], 'optionalAccess', _288 => _288[0], 'optionalAccess', _289 => _289.text]), () => ( "")))) {
+    if (unitLine.length > 0 && rebuiltUsed && !/^\s*\(\s*단위\s*[:：]/.test(_nullishCoalesce(_optionalChain([finalGrid, 'access', _293 => _293[0], 'optionalAccess', _294 => _294[0], 'optionalAccess', _295 => _295.text]), () => ( "")))) {
       prependUnitRow(finalGrid, numCols, unitLine, usedItems);
       finalRows++;
     }
     if (!rebuiltUsed && attachUnitRow(items, grid, finalGrid, numCols, usedItems)) finalRows++;
     let semanticOneColumn = false;
     if (!grid.cells && numCols === 1 && numGridRows >= 2) {
-      const populatedRows = finalGrid.filter((row) => _optionalChain([row, 'access', _290 => _290[0], 'optionalAccess', _291 => _291.text, 'access', _292 => _292.trim, 'call', _293 => _293()]));
+      const populatedRows = finalGrid.filter((row) => _optionalChain([row, 'access', _296 => _296[0], 'optionalAccess', _297 => _297.text, 'access', _298 => _298.trim, 'call', _299 => _299()]));
       if (populatedRows.length >= 4) {
         const fontSizes = tableItems.map((item) => item.fontSize).filter((size) => size > 0).sort((a, b) => a - b);
         const medianFont = _nullishCoalesce(fontSizes[Math.floor(fontSizes.length / 2)], () => ( 0));
@@ -7095,6 +7204,11 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
     const emptyPart = !hasContent && !!grid.cells && !grid.clipParent && !nestedAttached;
     if (!hasContent && !emptyPart) continue;
     if (emptyPart) EMPTY_PARTS.add(irTable);
+    if (numCols === 1 && isSideTabTable(grid.bbox, irTable, items.filter((it) => !tableItems.includes(it)), pageWidth, pageHeight)) {
+      SIDE_TAB_TABLES.add(irTable);
+      blocks.push({ type: "table", table: irTable, pageNumber: pageNum, bbox: { page: pageNum, x: grid.bbox.x1, y: grid.bbox.y1, width: gridW, height: grid.bbox.y2 - grid.bbox.y1 } });
+      continue;
+    }
     if (grid.clipParent) {
       const nb = { page: pageNum, x: grid.bbox.x1, y: grid.bbox.y1, width: grid.bbox.x2 - grid.bbox.x1, height: grid.bbox.y2 - grid.bbox.y1 };
       pendingNested.push({ parent: grid.clipParent, block: { type: "table", table: irTable, pageNumber: pageNum, bbox: nb } });
@@ -7116,7 +7230,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       height: grid.bbox.y2 - grid.bbox.y1
     };
     if (numRows === 1 && numCols === 1 && gridW < pageWidth * 0.35) {
-      const prose = _nullishCoalesce(_optionalChain([finalGrid, 'access', _294 => _294[0], 'optionalAccess', _295 => _295[0], 'optionalAccess', _296 => _296.text]), () => ( ""));
+      const prose = _nullishCoalesce(_optionalChain([finalGrid, 'access', _300 => _300[0], 'optionalAccess', _301 => _301[0], 'optionalAccess', _302 => _302.text]), () => ( ""));
       const besideLines = (side) => new Set(side.map((it) => Math.round(it.y / 3))).size;
       const rightLines = items.filter((it) => it.x >= grid.bbox.x2 + 3 && it.y >= grid.bbox.y1 && it.y <= grid.bbox.y2);
       const leftLines = items.filter((it) => it.x + it.w <= grid.bbox.x1 - 3 && it.y >= grid.bbox.y1 && it.y <= grid.bbox.y2);
@@ -7170,7 +7284,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
     const proseColumns = findTwoColumnProseCutX(clusterItems) !== null || detectPersistentColumnGutter(remaining.map((i) => ({ x: i.x, y: i.y, w: i.w, h: i.h > 0 ? i.h : i.fontSize }))) !== null;
     const clusterResults = (proseColumns ? [] : detectClusterTables(clusterItems, pageNum)).filter((cr) => {
       const b = cr.bbox;
-      return !blocks.some((x) => x.type === "paragraph" && x.bbox && (_nullishCoalesce(_optionalChain([x, 'access', _297 => _297.text, 'optionalAccess', _298 => _298.match, 'call', _299 => _299(/\p{L}/gu), 'optionalAccess', _300 => _300.length]), () => ( 0))) >= 10 && Math.min(x.bbox.x + x.bbox.width, b.x + b.width) - Math.max(x.bbox.x, b.x) >= 0.8 * b.width && x.bbox.y >= b.y - 2 && x.bbox.y + x.bbox.height <= b.y + b.height + 2);
+      return !blocks.some((x) => x.type === "paragraph" && x.bbox && (_nullishCoalesce(_optionalChain([x, 'access', _303 => _303.text, 'optionalAccess', _304 => _304.match, 'call', _305 => _305(/\p{L}/gu), 'optionalAccess', _306 => _306.length]), () => ( 0))) >= 10 && Math.min(x.bbox.x + x.bbox.width, b.x + b.width) - Math.max(x.bbox.x, b.x) >= 0.8 * b.width && x.bbox.y >= b.y - 2 && x.bbox.y + x.bbox.height <= b.y + b.height + 2);
     });
     if (clusterResults.length > 0) {
       const ciToIdx = /* @__PURE__ */ new Map();
@@ -7204,7 +7318,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
     if (gutterX !== null) {
       const gx = gutterX;
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkCKMGVW6Scjs.safeMax.call(void 0, allY) - _chunkCKMGVW6Scjs.safeMin.call(void 0, allY);
+      const pageH = _chunkUZGZRGYMcjs.safeMax.call(void 0, allY) - _chunkUZGZRGYMcjs.safeMin.call(void 0, allY);
       const gapThreshold = Math.max(15, pageH * 0.03);
       const sides = [
         remaining.filter((i) => i.x + i.w <= gx),
@@ -7224,7 +7338,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       finalTextBlocks = detectListBlocks(textBlocks);
     } else {
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkCKMGVW6Scjs.safeMax.call(void 0, allY) - _chunkCKMGVW6Scjs.safeMin.call(void 0, allY);
+      const pageH = _chunkUZGZRGYMcjs.safeMax.call(void 0, allY) - _chunkUZGZRGYMcjs.safeMin.call(void 0, allY);
       const groups = xyCutOrder(remaining, Math.max(15, pageH * 0.03));
       const textBlocks = [];
       for (const group of groups) {
@@ -7543,7 +7657,7 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false, detectTable
       blocks.push(...columnTextToBlocks(tableText, pageNum, bbox, dominantStyle(items)));
     } else {
       const allY = items.map((i) => i.y);
-      const pageHeight = _chunkCKMGVW6Scjs.safeMax.call(void 0, allY) - _chunkCKMGVW6Scjs.safeMin.call(void 0, allY);
+      const pageHeight = _chunkUZGZRGYMcjs.safeMax.call(void 0, allY) - _chunkUZGZRGYMcjs.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) {
@@ -7610,5 +7724,6 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false, detectTable
 
 
 
-exports.OCR_DET_MODEL = OCR_DET_MODEL; exports.OCR_REC_MODEL = OCR_REC_MODEL; exports.OCR_REC_DICT = OCR_REC_DICT; exports.getOcrModelsDir = getOcrModelsDir; exports.ensureOcrModels = ensureOcrModels; exports.ocrModelsCached = ocrModelsCached; exports.parseCharacterDict = parseCharacterDict; exports.extractLines = extractLines; exports.extractImageRegions = extractImageRegions; exports.filterHiddenText = filterHiddenText; exports.collapseEvenSpacing = collapseEvenSpacing; exports.normalizeItems = normalizeItems; exports.WrapLexicon = WrapLexicon; exports.startsNewItem = startsNewItem; exports.wrapJoiner = wrapJoiner; exports.joinPageBreakWraps = joinPageBreakWraps; exports.latinSoftWrap = latinSoftWrap; exports.TOC_BLOCKS = TOC_BLOCKS; exports.computeMedianFontSizeFromFreq = computeMedianFontSizeFromFreq; exports.detectHeadings = detectHeadings; exports.mergeStackedHeadingLines = mergeStackedHeadingLines; exports.detectTypographyHeadings = detectTypographyHeadings; exports.detectDocumentStyleHeadings = detectDocumentStyleHeadings; exports.detectSiblingStyleHeadings = detectSiblingStyleHeadings; exports.detectRepeatedPageLabels = detectRepeatedPageLabels; exports.detectPageLeadHeadings = detectPageLeadHeadings; exports.refineDocumentStyleHeadings = refineDocumentStyleHeadings; exports.detectMarkerHeadings = detectMarkerHeadings; exports.detectTableCaptions = detectTableCaptions; exports.detectKoreanListBlocks = detectKoreanListBlocks; exports.removeHeaderFooterBlocks = removeHeaderFooterBlocks; exports.CLIP_TABLES = CLIP_TABLES; exports.TABLE_COLXS = TABLE_COLXS; exports.TABLE_ROWYS = TABLE_ROWYS; exports.TABLE_TAIL = TABLE_TAIL; exports.PART_COLXS = PART_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.ROW_RULES = ROW_RULES; exports.trimTrailingEmptyTableCols = trimTrailingEmptyTableCols; exports.NO_EDGES = NO_EDGES; exports.joinCellEdges = joinCellEdges; exports.extractPageBlocksWithLines = extractPageBlocksWithLines;
-//# sourceMappingURL=chunk-ELTPM4HM.cjs.map
+
+exports.OCR_DET_MODEL = OCR_DET_MODEL; exports.OCR_REC_MODEL = OCR_REC_MODEL; exports.OCR_REC_DICT = OCR_REC_DICT; exports.getOcrModelsDir = getOcrModelsDir; exports.ensureOcrModels = ensureOcrModels; exports.ocrModelsCached = ocrModelsCached; exports.parseCharacterDict = parseCharacterDict; exports.extractLines = extractLines; exports.extractImageRegions = extractImageRegions; exports.filterHiddenText = filterHiddenText; exports.collapseEvenSpacing = collapseEvenSpacing; exports.normalizeItems = normalizeItems; exports.WrapLexicon = WrapLexicon; exports.startsNewItem = startsNewItem; exports.wrapJoiner = wrapJoiner; exports.joinPageBreakWraps = joinPageBreakWraps; exports.latinSoftWrap = latinSoftWrap; exports.TOC_BLOCKS = TOC_BLOCKS; exports.computeMedianFontSizeFromFreq = computeMedianFontSizeFromFreq; exports.detectHeadings = detectHeadings; exports.mergeStackedHeadingLines = mergeStackedHeadingLines; exports.detectTypographyHeadings = detectTypographyHeadings; exports.detectDocumentStyleHeadings = detectDocumentStyleHeadings; exports.detectSiblingStyleHeadings = detectSiblingStyleHeadings; exports.detectRepeatedPageLabels = detectRepeatedPageLabels; exports.detectPageLeadHeadings = detectPageLeadHeadings; exports.refineDocumentStyleHeadings = refineDocumentStyleHeadings; exports.detectMarkerHeadings = detectMarkerHeadings; exports.detectTableCaptions = detectTableCaptions; exports.detectKoreanListBlocks = detectKoreanListBlocks; exports.removeHeaderFooterBlocks = removeHeaderFooterBlocks; exports.CLIP_TABLES = CLIP_TABLES; exports.TABLE_COLXS = TABLE_COLXS; exports.TABLE_ROWYS = TABLE_ROWYS; exports.TABLE_TAIL = TABLE_TAIL; exports.PART_COLXS = PART_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.ROW_RULES = ROW_RULES; exports.trimTrailingEmptyTableCols = trimTrailingEmptyTableCols; exports.NO_EDGES = NO_EDGES; exports.joinCellEdges = joinCellEdges; exports.removeSideTabs = removeSideTabs; exports.extractPageBlocksWithLines = extractPageBlocksWithLines;
+//# sourceMappingURL=chunk-KDMAYOWH.cjs.map