kordoc 4.24.0 → 4.24.1

package/dist/{chunk-OHDGEGDJ.cjs → chunk-4VNIEHWM.cjs}

@@ -4,15 +4,15 @@
 
 
 
-var _chunkIHNIPVPJcjs = require('./chunk-IHNIPVPJ.cjs');
+var _chunk5QBTJ4LAcjs = require('./chunk-5QBTJ4LA.cjs');
 
 
 
-var _chunkDKWEZWCFcjs = require('./chunk-DKWEZWCF.cjs');
+var _chunkRIQVVWVFcjs = require('./chunk-RIQVVWVF.cjs');
 
 
 
-var _chunkHOXHBG4Ccjs = require('./chunk-HOXHBG4C.cjs');
+var _chunkTYXXTWPHcjs = require('./chunk-TYXXTWPH.cjs');
 
 // src/ocr/models.ts
 var _path = require('path');
@@ -56,7 +56,7 @@ var OCR_REC_DICT = {
   sizeMb: 1
 };
 function getOcrModelsDir() {
-  return _chunkDKWEZWCFcjs.getModelsDir.call(void 0, "ppocr");
+  return _chunkRIQVVWVFcjs.getModelsDir.call(void 0, "ppocr");
 }
 var DEFAULT_OCR_MODELS = Object.freeze({
   det: { spec: OCR_DET_MODEL },
@@ -98,7 +98,7 @@ async function ensureOcrModelsFor(sel, onProgress) {
     if (!stamp || verifiedInProcess.get(path) !== stamp) todo.push(s.spec);
   }
   if (!todo.length) return;
-  await _chunkDKWEZWCFcjs.ensureModelsIn.call(void 0, dir, todo, onProgress);
+  await _chunkRIQVVWVFcjs.ensureModelsIn.call(void 0, dir, todo, onProgress);
   for (const spec of todo) {
     const path = _path.join.call(void 0, dir, spec.filename), stamp = await fileStamp(path);
     if (stamp) verifiedInProcess.set(path, stamp);
@@ -2147,16 +2147,16 @@ function recordOpenEndCellEdges(grid, cells, horizontals) {
 }
 function takeClipCellEdges(from, to) {
   const e = CLIP_CELL_EDGES.get(from);
-  if (e) _chunkIHNIPVPJcjs.CELL_EDGES.set(to, e);
+  if (e) _chunk5QBTJ4LAcjs.CELL_EDGES.set(to, e);
 }
 function hasOpenCellSeam(a, b) {
-  const u = _chunkIHNIPVPJcjs.CELL_EDGES.get(a), d = _chunkIHNIPVPJcjs.CELL_EDGES.get(b);
+  const u = _chunk5QBTJ4LAcjs.CELL_EDGES.get(a), d = _chunk5QBTJ4LAcjs.CELL_EDGES.get(b);
   return !!u && !!d && u.t && !u.b && !d.t && d.b && u.l && u.r && d.l && d.r;
 }
 function joinCellEdges(a, b) {
-  const u = _chunkIHNIPVPJcjs.CELL_EDGES.get(a), d = _chunkIHNIPVPJcjs.CELL_EDGES.get(b);
+  const u = _chunk5QBTJ4LAcjs.CELL_EDGES.get(a), d = _chunk5QBTJ4LAcjs.CELL_EDGES.get(b);
   if (!u || !d) return;
-  _chunkIHNIPVPJcjs.CELL_EDGES.set(a, { t: u.t, b: d.b, l: u.l || d.l, r: u.r || d.r });
+  _chunk5QBTJ4LAcjs.CELL_EDGES.set(a, { t: u.t, b: d.b, l: u.l || d.l, r: u.r || d.r });
 }
 
 // src/pdf/cell-extract.ts
@@ -2332,7 +2332,7 @@ 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 _chunkIHNIPVPJcjs.tidyScriptTags.call(void 0, out);
+  return _chunk5QBTJ4LAcjs.tidyScriptTags.call(void 0, out);
 }
 
 // src/pdf/line-wrap.ts
@@ -5391,9 +5391,9 @@ function detectHeadings(blocks, medianFontSize) {
     if (/^\d+$/.test(text) && !isChapterNumber(block, blocks[bi + 1], medianFontSize)) continue;
     const ratio = block.style.fontSize / medianFontSize;
     let level = 0;
-    if (ratio >= _chunkIHNIPVPJcjs.HEADING_RATIO_H1) level = 1;
-    else if (ratio >= _chunkIHNIPVPJcjs.HEADING_RATIO_H2) level = 2;
-    else if (ratio >= _chunkIHNIPVPJcjs.HEADING_RATIO_H3) level = 3;
+    if (ratio >= _chunk5QBTJ4LAcjs.HEADING_RATIO_H1) level = 1;
+    else if (ratio >= _chunk5QBTJ4LAcjs.HEADING_RATIO_H2) level = 2;
+    else if (ratio >= _chunk5QBTJ4LAcjs.HEADING_RATIO_H3) level = 3;
     if (level > 0) {
       block.type = "heading";
       block.level = level;
@@ -5406,7 +5406,7 @@ function isChapterNumber(block, next, medianFontSize) {
   const nextSize = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _159 => _159.style, 'optionalAccess', _160 => _160.fontSize]), () => ( 0));
   const text = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _161 => _161.text, 'optionalAccess', _162 => _162.trim, 'call', _163 => _163()]), () => ( ""));
   const a = block.bbox, b = _optionalChain([next, 'optionalAccess', _164 => _164.bbox]);
-  return /^\d{1,2}$/.test(_nullishCoalesce(_optionalChain([block, 'access', _165 => _165.text, 'optionalAccess', _166 => _166.trim, 'call', _167 => _167()]), () => ( ""))) && size >= medianFontSize * 2.5 && !!next && (next.type === "paragraph" || next.type === "heading") && next.pageNumber === block.pageNumber && nextSize >= medianFontSize * _chunkIHNIPVPJcjs.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', _165 => _165.text, 'optionalAccess', _166 => _166.trim, 'call', _167 => _167()]), () => ( ""))) && size >= medianFontSize * 2.5 && !!next && (next.type === "paragraph" || next.type === "heading") && next.pageNumber === block.pageNumber && nextSize >= medianFontSize * _chunk5QBTJ4LAcjs.HEADING_RATIO_H1 && text.length > 0 && text.length <= 80 && !/^[\d\s.]+$/.test(text) && !!a && !!b && a.y > b.y && a.y - (b.y + b.height) <= size * 2;
 }
 function mergeStackedHeadingLines(blocks, medianFontSize) {
   for (let i = 0; i < blocks.length - 1; ) {
@@ -6944,16 +6944,16 @@ function isProseSpread(items) {
   for (let i = 1; i < sorted.length; i++) {
     gaps.push(sorted[i].x - (sorted[i - 1].x + sorted[i - 1].w));
   }
-  const maxGap = _chunkHOXHBG4Ccjs.safeMax.call(void 0, gaps);
+  const maxGap = _chunkTYXXTWPHcjs.safeMax.call(void 0, gaps);
   const avgLen = items.reduce((s, i) => s + i.text.length, 0) / items.length;
   if (items.length >= 4 && maxGap < 40 && avgLen < 5) return true;
-  const minGap = _chunkHOXHBG4Ccjs.safeMin.call(void 0, gaps);
+  const minGap = _chunkTYXXTWPHcjs.safeMin.call(void 0, gaps);
   return minGap > 0 && maxGap - minGap <= Math.max(2, maxGap * 0.15) && (items.filter((i) => /^\P{L}*\p{Ll}/u.test(i.text)).length * 2 >= items.length || items.some((i) => /\p{L}[,.;:]["”’)]?$/u.test(i.text)));
 }
 function detectColumns(yLines) {
   const allItems = yLines.flat();
   if (allItems.length === 0) return null;
-  const pageWidth = _chunkHOXHBG4Ccjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkHOXHBG4Ccjs.safeMin.call(void 0, allItems.map((i) => i.x));
+  const pageWidth = _chunkTYXXTWPHcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkTYXXTWPHcjs.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++) {
@@ -7445,7 +7445,7 @@ 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]) _chunkIHNIPVPJcjs.CELL_EDGES.set(cell, NO_EDGES);
+  for (const cell of cells[0]) _chunk5QBTJ4LAcjs.CELL_EDGES.set(cell, NO_EDGES);
   for (const item of line) usedItems.add(item);
 }
 function attachUnitRow(items, grid, cells, cols, usedItems) {
@@ -7977,7 +7977,7 @@ function recordFrameTitle(table, source, pageNum) {
   const [label, spacer, title] = cells;
   if (!/^(?:붙임|별첨|첨부|부록)\s*\d{1,3}$/.test(label.text.trim()) || spacer.text.trim() || !title.text.trim()) return;
   if (cells.some((c) => {
-    const e = _chunkIHNIPVPJcjs.CELL_EDGES.get(c);
+    const e = _chunk5QBTJ4LAcjs.CELL_EDGES.get(c);
     return !e || e.t || e.b || e.l || e.r;
   })) return;
   const xs = TABLE_COLXS.get(table), style = dominantStyle(source), fs = _nullishCoalesce(_optionalChain([style, 'optionalAccess', _358 => _358.fontSize]), () => ( 0));
@@ -8558,7 +8558,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
         groupSizes.push(1);
       }
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkHOXHBG4Ccjs.safeMax.call(void 0, allY) - _chunkHOXHBG4Ccjs.safeMin.call(void 0, allY);
+      const pageH = _chunkTYXXTWPHcjs.safeMax.call(void 0, allY) - _chunkTYXXTWPHcjs.safeMin.call(void 0, allY);
       const gapThreshold = Math.max(15, pageH * 0.03);
       const sides = frameColumnTextBands(remaining, spanning, gx);
       const textBlocks = [];
@@ -8574,7 +8574,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       finalTextBlocks.push(...detectListBlocks(textBlocks));
     } else {
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkHOXHBG4Ccjs.safeMax.call(void 0, allY) - _chunkHOXHBG4Ccjs.safeMin.call(void 0, allY);
+      const pageH = _chunkTYXXTWPHcjs.safeMax.call(void 0, allY) - _chunkTYXXTWPHcjs.safeMin.call(void 0, allY);
       const gapThreshold = Math.max(15, pageH * 0.03);
       const groups = xyCutOrder(remaining, gapThreshold);
       const textBlocks = [];
@@ -8902,7 +8902,7 @@ 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 = _chunkHOXHBG4Ccjs.safeMax.call(void 0, allY) - _chunkHOXHBG4Ccjs.safeMin.call(void 0, allY);
+      const pageHeight = _chunkTYXXTWPHcjs.safeMax.call(void 0, allY) - _chunkTYXXTWPHcjs.safeMin.call(void 0, allY);
       const gapThreshold = Math.max(15, pageHeight * 0.03);
       const orderedGroups = proseCutX !== null ? splitTwoColumnProse(items, proseCutX) : xyCutOrder(items, gapThreshold, 0, void 0, minGap);
       for (const group of orderedGroups) {
@@ -9118,4 +9118,4 @@ function rowParagraphs(block) {
 
 
 exports.DEFAULT_OCR_MODELS = DEFAULT_OCR_MODELS; exports.resolveOcrModels = resolveOcrModels; exports.ocrSelectionKey = ocrSelectionKey; exports.ocrSourcePath = ocrSourcePath; exports.ensureOcrModelsFor = ensureOcrModelsFor; exports.ocrModelsCached = ocrModelsCached; exports.parseCharacterDict = parseCharacterDict; exports.extractLines = extractLines; exports.extractImageRegions = extractImageRegions; exports.NO_EDGES = NO_EDGES; exports.hasOpenCellSeam = hasOpenCellSeam; exports.joinCellEdges = joinCellEdges; exports.filterHiddenText = filterHiddenText; exports.collapseEvenSpacing = collapseEvenSpacing; exports.computeBBox = computeBBox; exports.dominantStyle = dominantStyle; exports.normalizeItems = normalizeItems; exports.groupByY = groupByY; exports.mergeLineSimple = mergeLineSimple; exports.WrapLexicon = WrapLexicon; exports.startsNewItem = startsNewItem; exports.wrapJoiner = wrapJoiner; exports.joinPageBreakWraps = joinPageBreakWraps; exports.splitPageBreakWraps = splitPageBreakWraps; 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.runningLinePatterns = runningLinePatterns; exports.stripAbsorbedRunningLines = stripAbsorbedRunningLines; exports.CLIP_TABLES = CLIP_TABLES; exports.FRAME_TITLE_BLOCKS = FRAME_TITLE_BLOCKS; 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.restoreImageBullets = restoreImageBullets; exports.detectRightArrowRegions = detectRightArrowRegions; exports.removeSideTabs = removeSideTabs; exports.extractPageBlocksWithLines = extractPageBlocksWithLines; exports.UNRULED_REGION_TABLES = UNRULED_REGION_TABLES; exports.mergeOcrImageRegions = mergeOcrImageRegions;
-//# sourceMappingURL=chunk-OHDGEGDJ.cjs.map
+//# sourceMappingURL=chunk-4VNIEHWM.cjs.map