kordoc 4.17.6 → 4.18.1

package/README.md

@@ -342,11 +342,16 @@ for (const p of r.pageQuality ?? []) if (p.needsOcr) console.log(`p${p.page} 검
 
 ## 💻 CLI
 
+[병렬 변환 성능 측정과 제한 사항](docs/parallel-batch.md)
+
+`--jobs N`은 상주 프로세스 N개로 파일 단위 병렬 변환을 실행합니다(기본 1). `--jobs > 1`의 다중 입력은 `-d`가 필요하며, 확장자를 제외한 파일 이름이 서로 달라야 합니다. 일부 파일이 실패해도 나머지는 계속 처리하고 종료 코드는 실패로 설정합니다. 완료 로그와 실패 JSON은 완료 순서로 나옵니다. 프로세스 수만큼 메모리 사용이 늘고, 작은 파일은 오히려 느릴 수 있습니다. OCR은 자체 네이티브 스레드도 사용합니다. 단일 문서의 페이지 병렬화나 JavaScript/MCP API 변경은 포함하지 않습니다.
+
 ```bash
 # 변환
 npx kordoc 사업계획서.hwpx                           # 터미널 출력
 npx kordoc 보고서.hwp -o 보고서.md                   # 파일 저장 (그림은 images/보고서/)
 npx kordoc *.pdf -d ./변환결과/                      # 일괄 변환
+npx kordoc *.pdf --jobs 4 -d ./변환결과/             # 병렬 변환 (상주 프로세스 4개)
 npx kordoc 검토서.hwpx --format json                # JSON (blocks + pages + metadata)
 npx kordoc 검토서.pdf --format chunks               # RAG 구조 청크 JSON (breadcrumb + 표 독립 청크)
 npx kordoc 보고서.hwpx --pages 1-3                   # 쪽 범위

package/dist/{-Y6U3UTVZ.js → -X4KTLQC6.js}

@@ -50,17 +50,17 @@ import {
   unknownFontWarnings,
   usesGaejosikMunche,
   validateHwpx
-} from "./chunk-ZS5KKM4L.js";
+} from "./chunk-UTVMLJZE.js";
 import {
   blocksToPages
-} from "./chunk-7F7CUU22.js";
+} from "./chunk-PWRY4W6Q.js";
 import {
   applySplices,
   buildParagraphSplices,
   buildRangeSplices,
   placeSealHwpx,
   scanSectionXml
-} from "./chunk-PGHIQ6E5.js";
+} from "./chunk-Z4KCEERM.js";
 import {
   DEFAULT_REDACT_RULES,
   redactMarkdown,
@@ -68,7 +68,7 @@ import {
 } from "./chunk-3WSNLH4B.js";
 import {
   blocksToChunks
-} from "./chunk-B3RQHQCC.js";
+} from "./chunk-2VFZQ634.js";
 import {
   SPACE_EM_FIXED,
   SPACE_EM_FONT,
@@ -93,20 +93,20 @@ import {
   renderSceneToHtml,
   simulateWrap,
   simulateWrapKeepWord
-} from "./chunk-7R7NFGHY.js";
+} from "./chunk-CWJ5KY2F.js";
 import "./chunk-7FCNOKOR.js";
-import "./chunk-2WUE6JIT.js";
+import "./chunk-TMC4A3FM.js";
 import "./chunk-VQFA55IC.js";
-import "./chunk-6A4Z6VYA.js";
+import "./chunk-6N5BOPJI.js";
 import {
   blocksToMarkdown,
   flattenLayoutTables,
   hasStructuredCellContent
-} from "./chunk-5VHV6W7W.js";
-import "./chunk-MOM67IND.js";
+} from "./chunk-K35WIYBD.js";
+import "./chunk-AQUNBGXH.js";
 import {
   VERSION
-} from "./chunk-T5O6HEHG.js";
+} from "./chunk-I7YWXECO.js";
 export {
   BUILTIN_TEMPLATES,
   DEFAULT_REDACT_RULES,
@@ -196,4 +196,4 @@ export {
   usesGaejosikMunche,
   validateHwpx
 };
-//# sourceMappingURL=-Y6U3UTVZ.js.map
+//# sourceMappingURL=-X4KTLQC6.js.map

package/dist/{chunk-B3RQHQCC.js → chunk-2VFZQ634.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import {
   blocksToMarkdown
-} from "./chunk-5VHV6W7W.js";
+} from "./chunk-K35WIYBD.js";
 
 // src/chunks.ts
 var LIST_MARKER_RE = /^(?:[□■◇◆○◎●◦ㅇ•▪▸▶※-]|\d{1,3}[.)]|[가나다라마바사아자차카타파하][.)]|\([가나다라마바사아자차카타파하0-9]{1,3}\)|[①-⑳㉮-㉻㈎-㈛])\s/;
@@ -79,4 +79,4 @@ function blocksToChunks(blocks, options) {
 export {
   blocksToChunks
 };
-//# sourceMappingURL=chunk-B3RQHQCC.js.map
+//# sourceMappingURL=chunk-2VFZQ634.js.map

package/dist/{chunk-U6NLEEIN.js → chunk-5PNQIF3V.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import {
   extractPageBlocksWithLines
-} from "./chunk-QKEPDJ2D.js";
+} from "./chunk-WWOTHS2E.js";
 import {
   OCR_DET_MODEL,
   OCR_REC_DICT,
@@ -9,10 +9,10 @@ import {
   ensureOcrModels,
   getOcrModelsDir,
   parseCharacterDict
-} from "./chunk-MOVFO74W.js";
+} from "./chunk-R2IFWAHI.js";
 import {
   OPTIONAL_DEP_INSTALL_HINT
-} from "./chunk-T5O6HEHG.js";
+} from "./chunk-I7YWXECO.js";
 
 // src/ocr/ruling-lines.ts
 var INK_LUMA_MAX = 205;
@@ -2061,4 +2061,4 @@ export {
   runPdfOcr,
   ocrItemsToBlocks
 };
-//# sourceMappingURL=chunk-U6NLEEIN.js.map
+//# sourceMappingURL=chunk-5PNQIF3V.js.map

package/dist/{chunk-6A4Z6VYA.js → chunk-6N5BOPJI.js}

@@ -2,7 +2,7 @@
 import {
   KordocError,
   unzipLimitBytes
-} from "./chunk-T5O6HEHG.js";
+} from "./chunk-I7YWXECO.js";
 
 // src/hwpx/parser-shared.ts
 import { DOMParser } from "@xmldom/xmldom";
@@ -336,4 +336,4 @@ export {
   hwpFaceToCssStack,
   parseRenderStyles
 };
-//# sourceMappingURL=chunk-6A4Z6VYA.js.map
+//# sourceMappingURL=chunk-6N5BOPJI.js.map

package/dist/{chunk-YCMPQRHV.js → chunk-6V7QRQY3.js}

@@ -4,15 +4,15 @@ import {
   HEADING_RATIO_H2,
   HEADING_RATIO_H3,
   tidyScriptTags
-} from "./chunk-ZQA7Q3GV.js";
+} from "./chunk-HUGBJVAA.js";
 import {
   ensureModelsIn,
   getModelsDir
-} from "./chunk-AYMTRCLC.js";
+} from "./chunk-KXR4CTFT.js";
 import {
   safeMax,
   safeMin
-} from "./chunk-BPLEMV4B.js";
+} from "./chunk-VDIXNEWV.js";
 
 // src/ocr/models.ts
 import { join } from "path";
@@ -7611,4 +7611,4 @@ export {
   joinCellEdges,
   extractPageBlocksWithLines
 };
-//# sourceMappingURL=chunk-YCMPQRHV.js.map
+//# sourceMappingURL=chunk-6V7QRQY3.js.map

package/dist/{chunk-CZOPNDTO.cjs → chunk-7DC2KL3V.cjs}

@@ -4,15 +4,15 @@
 
 
 
-var _chunkL5JX2GAMcjs = require('./chunk-L5JX2GAM.cjs');
+var _chunkQ5N4YS3Tcjs = require('./chunk-Q5N4YS3T.cjs');
 
 
 
-var _chunk4LU6NGTLcjs = require('./chunk-4LU6NGTL.cjs');
+var _chunkSIZ63E5Gcjs = require('./chunk-SIZ63E5G.cjs');
 
 
 
-var _chunkFVESLY6Vcjs = require('./chunk-FVESLY6V.cjs');
+var _chunkPCSOJ523cjs = require('./chunk-PCSOJ523.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 _chunk4LU6NGTLcjs.getModelsDir.call(void 0, "ppocr");
+  return _chunkSIZ63E5Gcjs.getModelsDir.call(void 0, "ppocr");
 }
 async function ensureOcrModels(onProgress) {
-  return _chunk4LU6NGTLcjs.ensureModelsIn.call(void 0, getOcrModelsDir(), ALL_OCR_MODELS, onProgress);
+  return _chunkSIZ63E5Gcjs.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 _chunkL5JX2GAMcjs.tidyScriptTags.call(void 0, out);
+  return _chunkQ5N4YS3Tcjs.tidyScriptTags.call(void 0, out);
 }
 
 // src/pdf/line-wrap.ts
@@ -3641,9 +3641,9 @@ function detectHeadings(blocks, medianFontSize) {
     if (/^\d+$/.test(text) && !isChapterNumber(block, blocks[bi + 1], medianFontSize)) continue;
     const ratio = block.style.fontSize / medianFontSize;
     let level = 0;
-    if (ratio >= _chunkL5JX2GAMcjs.HEADING_RATIO_H1) level = 1;
-    else if (ratio >= _chunkL5JX2GAMcjs.HEADING_RATIO_H2) level = 2;
-    else if (ratio >= _chunkL5JX2GAMcjs.HEADING_RATIO_H3) level = 3;
+    if (ratio >= _chunkQ5N4YS3Tcjs.HEADING_RATIO_H1) level = 1;
+    else if (ratio >= _chunkQ5N4YS3Tcjs.HEADING_RATIO_H2) level = 2;
+    else if (ratio >= _chunkQ5N4YS3Tcjs.HEADING_RATIO_H3) level = 3;
     if (level > 0) {
       block.type = "heading";
       block.level = level;
@@ -3656,7 +3656,7 @@ function isChapterNumber(block, next, medianFontSize) {
   const nextSize = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _94 => _94.style, 'optionalAccess', _95 => _95.fontSize]), () => ( 0));
   const text = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _96 => _96.text, 'optionalAccess', _97 => _97.trim, 'call', _98 => _98()]), () => ( ""));
   const a = block.bbox, b = _optionalChain([next, 'optionalAccess', _99 => _99.bbox]);
-  return /^\d{1,2}$/.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 * _chunkL5JX2GAMcjs.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 * _chunkQ5N4YS3Tcjs.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; ) {
@@ -5565,16 +5565,16 @@ 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 = _chunkFVESLY6Vcjs.safeMax.call(void 0, gaps);
+  const maxGap = _chunkPCSOJ523cjs.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 = _chunkFVESLY6Vcjs.safeMin.call(void 0, gaps);
+  const minGap = _chunkPCSOJ523cjs.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 = _chunkFVESLY6Vcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkFVESLY6Vcjs.safeMin.call(void 0, allItems.map((i) => i.x));
+  const pageWidth = _chunkPCSOJ523cjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkPCSOJ523cjs.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++) {
@@ -6012,12 +6012,12 @@ function recordClipCellEdges(grids, horizontals, verticals, nonRules) {
 }
 function takeClipCellEdges(from, to) {
   const e = CLIP_CELL_EDGES.get(from);
-  if (e) _chunkL5JX2GAMcjs.CELL_EDGES.set(to, e);
+  if (e) _chunkQ5N4YS3Tcjs.CELL_EDGES.set(to, e);
 }
 function joinCellEdges(a, b) {
-  const u = _chunkL5JX2GAMcjs.CELL_EDGES.get(a), d = _chunkL5JX2GAMcjs.CELL_EDGES.get(b);
+  const u = _chunkQ5N4YS3Tcjs.CELL_EDGES.get(a), d = _chunkQ5N4YS3Tcjs.CELL_EDGES.get(b);
   if (!u || !d) return;
-  _chunkL5JX2GAMcjs.CELL_EDGES.set(a, { t: u.t, b: d.b, l: u.l || d.l, r: u.r || d.r });
+  _chunkQ5N4YS3Tcjs.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
@@ -6033,7 +6033,7 @@ 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]) _chunkL5JX2GAMcjs.CELL_EDGES.set(cell, NO_EDGES);
+  for (const cell of cells[0]) _chunkQ5N4YS3Tcjs.CELL_EDGES.set(cell, NO_EDGES);
   for (const item of line) usedItems.add(item);
 }
 function attachUnitRow(items, grid, cells, cols, usedItems) {
@@ -7204,7 +7204,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
     if (gutterX !== null) {
       const gx = gutterX;
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkFVESLY6Vcjs.safeMax.call(void 0, allY) - _chunkFVESLY6Vcjs.safeMin.call(void 0, allY);
+      const pageH = _chunkPCSOJ523cjs.safeMax.call(void 0, allY) - _chunkPCSOJ523cjs.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 +7224,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       finalTextBlocks = detectListBlocks(textBlocks);
     } else {
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkFVESLY6Vcjs.safeMax.call(void 0, allY) - _chunkFVESLY6Vcjs.safeMin.call(void 0, allY);
+      const pageH = _chunkPCSOJ523cjs.safeMax.call(void 0, allY) - _chunkPCSOJ523cjs.safeMin.call(void 0, allY);
       const groups = xyCutOrder(remaining, Math.max(15, pageH * 0.03));
       const textBlocks = [];
       for (const group of groups) {
@@ -7543,7 +7543,7 @@ 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 = _chunkFVESLY6Vcjs.safeMax.call(void 0, allY) - _chunkFVESLY6Vcjs.safeMin.call(void 0, allY);
+      const pageHeight = _chunkPCSOJ523cjs.safeMax.call(void 0, allY) - _chunkPCSOJ523cjs.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) {
@@ -7611,4 +7611,4 @@ 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-CZOPNDTO.cjs.map
+//# sourceMappingURL=chunk-7DC2KL3V.cjs.map