kordoc 4.24.19 → 4.24.21

package/dist/{chunk-XGUX3JDV.js → chunk-BMXWY36E.js}

@@ -5,15 +5,15 @@ import {
   HEADING_RATIO_H3,
   tidyScriptTags,
   toTex
-} from "./chunk-JQJYIYMJ.js";
+} from "./chunk-PDOWSDNM.js";
 import {
   ensureModelsIn,
   getModelsDir
-} from "./chunk-DTNNWOBN.js";
+} from "./chunk-NE75RO62.js";
 import {
   safeMax,
   safeMin
-} from "./chunk-RQCGVE3R.js";
+} from "./chunk-UNYH6MTO.js";
 
 // src/ocr/models.ts
 import { join } from "path";
@@ -10510,4 +10510,4 @@ export {
   UNRULED_REGION_TABLES,
   mergeOcrImageRegions
 };
-//# sourceMappingURL=chunk-XGUX3JDV.js.map
+//# sourceMappingURL=chunk-BMXWY36E.js.map

package/dist/{chunk-DIDPIFGO.js → chunk-BNEKESI4.js}

@@ -2,7 +2,7 @@
 import {
   assignTableIds,
   blocksToMarkdown
-} from "./chunk-IXCXFGYB.js";
+} from "./chunk-B3POEH7X.js";
 
 // src/chunks.ts
 var LIST_MARKER_RE = /^(?:[□■◇◆○◎●◦ㅇ•▪▸▶※-]|\d{1,3}[.)]|[가나다라마바사아자차카타파하][.)]|\([가나다라마바사아자차카타파하0-9]{1,3}\)|[①-⑳㉮-㉻㈎-㈛])\s/;
@@ -82,4 +82,4 @@ function blocksToChunks(blocks, options) {
 export {
   blocksToChunks
 };
-//# sourceMappingURL=chunk-DIDPIFGO.js.map
+//# sourceMappingURL=chunk-BNEKESI4.js.map

package/dist/{chunk-UTDQAY35.js → chunk-BTBFKDPH.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import {
   blocksToMarkdown
-} from "./chunk-IXCXFGYB.js";
+} from "./chunk-B3POEH7X.js";
 
 // src/page-markdown.ts
 function blocksToPages(blocks, render = blocksToMarkdown) {
@@ -29,4 +29,4 @@ function blocksToPages(blocks, render = blocksToMarkdown) {
 export {
   blocksToPages
 };
-//# sourceMappingURL=chunk-UTDQAY35.js.map
+//# sourceMappingURL=chunk-BTBFKDPH.js.map

package/dist/{chunk-OJL4V7WL.cjs → chunk-BUFGW3UJ.cjs}

@@ -5,15 +5,15 @@
 
 
 
-var _chunkXQJNL7TDcjs = require('./chunk-XQJNL7TD.cjs');
+var _chunkB5M676H3cjs = require('./chunk-B5M676H3.cjs');
 
 
 
-var _chunk4OWHKA6Dcjs = require('./chunk-4OWHKA6D.cjs');
+var _chunkV7S36XCUcjs = require('./chunk-V7S36XCU.cjs');
 
 
 
-var _chunkU4QSRBEVcjs = require('./chunk-U4QSRBEV.cjs');
+var _chunkOMGCVZ4Wcjs = require('./chunk-OMGCVZ4W.cjs');
 
 // src/ocr/models.ts
 var _path = require('path');
@@ -65,7 +65,7 @@ var OCR_REC_DICT = {
   sizeMb: 1
 };
 function getOcrModelsDir() {
-  return _chunk4OWHKA6Dcjs.getModelsDir.call(void 0, "ppocr");
+  return _chunkV7S36XCUcjs.getModelsDir.call(void 0, "ppocr");
 }
 var DEFAULT_OCR_MODELS = Object.freeze({
   det: { spec: OCR_DET_MODEL },
@@ -107,7 +107,7 @@ async function ensureOcrModelsFor(sel, onProgress) {
     if (!stamp || verifiedInProcess.get(path) !== stamp) todo.push(s.spec);
   }
   if (!todo.length) return;
-  await _chunk4OWHKA6Dcjs.ensureModelsIn.call(void 0, dir, todo, onProgress);
+  await _chunkV7S36XCUcjs.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);
@@ -3154,16 +3154,16 @@ function recordOpenEndCellEdges(grid, cells, horizontals) {
 }
 function takeClipCellEdges(from, to) {
   const e = CLIP_CELL_EDGES.get(from);
-  if (e) _chunkXQJNL7TDcjs.CELL_EDGES.set(to, e);
+  if (e) _chunkB5M676H3cjs.CELL_EDGES.set(to, e);
 }
 function hasOpenCellSeam(a, b) {
-  const u = _chunkXQJNL7TDcjs.CELL_EDGES.get(a), d = _chunkXQJNL7TDcjs.CELL_EDGES.get(b);
+  const u = _chunkB5M676H3cjs.CELL_EDGES.get(a), d = _chunkB5M676H3cjs.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 = _chunkXQJNL7TDcjs.CELL_EDGES.get(a), d = _chunkXQJNL7TDcjs.CELL_EDGES.get(b);
+  const u = _chunkB5M676H3cjs.CELL_EDGES.get(a), d = _chunkB5M676H3cjs.CELL_EDGES.get(b);
   if (!u || !d) return;
-  _chunkXQJNL7TDcjs.CELL_EDGES.set(a, { t: u.t, b: d.b, l: u.l || d.l, r: u.r || d.r });
+  _chunkB5M676H3cjs.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
@@ -3339,7 +3339,7 @@ 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 _chunkXQJNL7TDcjs.tidyScriptTags.call(void 0, out);
+  return _chunkB5M676H3cjs.tidyScriptTags.call(void 0, out);
 }
 
 // src/pdf/line-wrap.ts
@@ -6598,9 +6598,9 @@ function detectHeadings(blocks, medianFontSize) {
     if (/^\d+$/.test(text) && !isChapterNumber(block, blocks[bi + 1], medianFontSize)) continue;
     const ratio = block.style.fontSize / medianFontSize;
     let level = 0;
-    if (ratio >= _chunkXQJNL7TDcjs.HEADING_RATIO_H1) level = 1;
-    else if (ratio >= _chunkXQJNL7TDcjs.HEADING_RATIO_H2) level = 2;
-    else if (ratio >= _chunkXQJNL7TDcjs.HEADING_RATIO_H3) level = 3;
+    if (ratio >= _chunkB5M676H3cjs.HEADING_RATIO_H1) level = 1;
+    else if (ratio >= _chunkB5M676H3cjs.HEADING_RATIO_H2) level = 2;
+    else if (ratio >= _chunkB5M676H3cjs.HEADING_RATIO_H3) level = 3;
     if (level > 0) {
       block.type = "heading";
       block.level = level;
@@ -6613,7 +6613,7 @@ function isChapterNumber(block, next, medianFontSize) {
   const nextSize = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _174 => _174.style, 'optionalAccess', _175 => _175.fontSize]), () => ( 0));
   const text = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _176 => _176.text, 'optionalAccess', _177 => _177.trim, 'call', _178 => _178()]), () => ( ""));
   const a = block.bbox, b = _optionalChain([next, 'optionalAccess', _179 => _179.bbox]);
-  return /^\d{1,2}$/.test(_nullishCoalesce(_optionalChain([block, 'access', _180 => _180.text, 'optionalAccess', _181 => _181.trim, 'call', _182 => _182()]), () => ( ""))) && size >= medianFontSize * 2.5 && !!next && (next.type === "paragraph" || next.type === "heading") && next.pageNumber === block.pageNumber && nextSize >= medianFontSize * _chunkXQJNL7TDcjs.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', _180 => _180.text, 'optionalAccess', _181 => _181.trim, 'call', _182 => _182()]), () => ( ""))) && size >= medianFontSize * 2.5 && !!next && (next.type === "paragraph" || next.type === "heading") && next.pageNumber === block.pageNumber && nextSize >= medianFontSize * _chunkB5M676H3cjs.HEADING_RATIO_H1 && text.length > 0 && text.length <= 80 && !/^[\d\s.]+$/.test(text) && !!a && !!b && a.y > b.y && a.y - (b.y + b.height) <= size * 2;
 }
 function mergeStackedHeadingLines(blocks, medianFontSize) {
   for (let i = 0; i < blocks.length - 1; ) {
@@ -8151,16 +8151,16 @@ function isProseSpread(items) {
   for (let i = 1; i < sorted.length; i++) {
     gaps.push(sorted[i].x - (sorted[i - 1].x + sorted[i - 1].w));
   }
-  const maxGap = _chunkU4QSRBEVcjs.safeMax.call(void 0, gaps);
+  const maxGap = _chunkOMGCVZ4Wcjs.safeMax.call(void 0, gaps);
   const avgLen = items.reduce((s, i) => s + i.text.length, 0) / items.length;
   if (items.length >= 4 && maxGap < 40 && avgLen < 5) return true;
-  const minGap = _chunkU4QSRBEVcjs.safeMin.call(void 0, gaps);
+  const minGap = _chunkOMGCVZ4Wcjs.safeMin.call(void 0, gaps);
   return minGap > 0 && maxGap - minGap <= Math.max(2, maxGap * 0.15) && (items.filter((i) => /^\P{L}*\p{Ll}/u.test(i.text)).length * 2 >= items.length || items.some((i) => /\p{L}[,.;:]["”’)]?$/u.test(i.text)));
 }
 function detectColumns(yLines) {
   const allItems = yLines.flat();
   if (allItems.length === 0) return null;
-  const pageWidth = _chunkU4QSRBEVcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkU4QSRBEVcjs.safeMin.call(void 0, allItems.map((i) => i.x));
+  const pageWidth = _chunkOMGCVZ4Wcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkOMGCVZ4Wcjs.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++) {
@@ -8558,13 +8558,13 @@ function partTex(group) {
   const base = Math.min(...g.filter((i) => i.fontSize >= size * SCRIPT_SIZE_RATIO).map((i) => i.y));
   let s = "", run = "";
   const flush = () => {
-    if (run) s += _chunkXQJNL7TDcjs.toTex.call(void 0, run);
+    if (run) s += _chunkB5M676H3cjs.toTex.call(void 0, run);
     run = "";
   };
   g.forEach((it, k) => {
     if (it.fontSize < size * SCRIPT_SIZE_RATIO && it.y - base >= size * SCRIPT_RISE_EM) {
       flush();
-      s += `^{${_chunkXQJNL7TDcjs.toTex.call(void 0, it.text)}}`;
+      s += `^{${_chunkB5M676H3cjs.toTex.call(void 0, it.text)}}`;
       return;
     }
     if (k && it.x - (g[k - 1].x + g[k - 1].w) > spaceGapThreshold(it.fontSize)) run += " ";
@@ -8736,7 +8736,7 @@ 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]) _chunkXQJNL7TDcjs.CELL_EDGES.set(cell, NO_EDGES);
+  for (const cell of cells[0]) _chunkB5M676H3cjs.CELL_EDGES.set(cell, NO_EDGES);
   for (const item of line) usedItems.add(item);
 }
 function attachUnitRow(items, grid, cells, cols, usedItems) {
@@ -9318,7 +9318,7 @@ 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 = _chunkXQJNL7TDcjs.CELL_EDGES.get(c);
+    const e = _chunkB5M676H3cjs.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', _373 => _373.fontSize]), () => ( 0));
@@ -9920,7 +9920,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
         groupSizes.push(1);
       }
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkU4QSRBEVcjs.safeMax.call(void 0, allY) - _chunkU4QSRBEVcjs.safeMin.call(void 0, allY);
+      const pageH = _chunkOMGCVZ4Wcjs.safeMax.call(void 0, allY) - _chunkOMGCVZ4Wcjs.safeMin.call(void 0, allY);
       const gapThreshold = Math.max(15, pageH * 0.03);
       const sides = frameColumnTextBands(remaining, spanning, gx);
       const textBlocks = [];
@@ -9936,7 +9936,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       finalTextBlocks.push(...detectListBlocks(textBlocks));
     } else {
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkU4QSRBEVcjs.safeMax.call(void 0, allY) - _chunkU4QSRBEVcjs.safeMin.call(void 0, allY);
+      const pageH = _chunkOMGCVZ4Wcjs.safeMax.call(void 0, allY) - _chunkOMGCVZ4Wcjs.safeMin.call(void 0, allY);
       const gapThreshold = Math.max(15, pageH * 0.03);
       const groups = xyCutOrder(remaining, gapThreshold);
       const textBlocks = [];
@@ -10268,7 +10268,7 @@ 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 = _chunkU4QSRBEVcjs.safeMax.call(void 0, allY) - _chunkU4QSRBEVcjs.safeMin.call(void 0, allY);
+      const pageHeight = _chunkOMGCVZ4Wcjs.safeMax.call(void 0, allY) - _chunkOMGCVZ4Wcjs.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) {
@@ -10510,4 +10510,4 @@ 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.grayCrop = grayCrop; exports.inkStats = inkStats; exports.splitRowBands = splitRowBands; exports.leadingTriangle = leadingTriangle; exports.arrowGlyph = arrowGlyph; exports.leaderRuns = leaderRuns; exports.inkBounds = inkBounds; exports.bracketFeatures = bracketFeatures; exports.slantedPair = slantedPair; exports.romanStems = romanStems; exports.circledAt = circledAt; exports.quoteHead = quoteHead; exports.edgeTrim = edgeTrim; exports.tallInkCount = tallInkCount; exports.gapGlyphs = gapGlyphs; exports.ringBullet = ringBullet; exports.dashRatio = dashRatio; exports.leadingRing = leadingRing; exports.discAt = discAt; exports.serifOne = serifOne; exports.starRun = starRun; exports.bracketShape = bracketShape; exports.leadingBullet = leadingBullet; exports.rememberOcrTextGaps = rememberOcrTextGaps; exports.copyOcrTextGaps = copyOcrTextGaps; exports.findOcrTextGaps = findOcrTextGaps; 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-OJL4V7WL.cjs.map
+//# sourceMappingURL=chunk-BUFGW3UJ.cjs.map