@clazic/kordoc 2.3.1 → 2.3.2

package/dist/chunk-4PP34NVQ.js

@@ -0,0 +1,121 @@
+#!/usr/bin/env node
+
+// src/ocr/markdown-to-blocks.ts
+function markdownToBlocks(markdown, pageNumber) {
+  const blocks = [];
+  const lines = markdown.split("\n");
+  let i = 0;
+  while (i < lines.length) {
+    const line = lines[i];
+    if (line.trim() === "") {
+      i++;
+      continue;
+    }
+    const headingMatch = line.match(/^(#{1,6})\s+(.+)$/);
+    if (headingMatch) {
+      blocks.push({
+        type: "heading",
+        level: headingMatch[1].length,
+        text: headingMatch[2].trim(),
+        pageNumber
+      });
+      i++;
+      continue;
+    }
+    if (/^[-*_]{3,}\s*$/.test(line.trim())) {
+      blocks.push({ type: "separator", pageNumber });
+      i++;
+      continue;
+    }
+    if (line.trim().startsWith("|")) {
+      const tableLines = [];
+      while (i < lines.length && lines[i].trim().startsWith("|")) {
+        tableLines.push(lines[i]);
+        i++;
+      }
+      const table = parseMarkdownTable(tableLines);
+      if (table) {
+        blocks.push({ type: "table", table, pageNumber });
+      }
+      continue;
+    }
+    const ulMatch = line.match(/^(\s*)[-*+]\s+(.+)$/);
+    if (ulMatch) {
+      blocks.push({
+        type: "list",
+        listType: "unordered",
+        text: ulMatch[2].trim(),
+        pageNumber
+      });
+      i++;
+      continue;
+    }
+    const olMatch = line.match(/^(\s*)\d+\.\s+(.+)$/);
+    if (olMatch) {
+      blocks.push({
+        type: "list",
+        listType: "ordered",
+        text: olMatch[2].trim(),
+        pageNumber
+      });
+      i++;
+      continue;
+    }
+    const paraLines = [];
+    while (i < lines.length && lines[i].trim() !== "" && !isStructuralLine(lines[i])) {
+      paraLines.push(lines[i].trim());
+      i++;
+    }
+    if (paraLines.length > 0) {
+      blocks.push({
+        type: "paragraph",
+        text: paraLines.join("\n"),
+        pageNumber
+      });
+    }
+  }
+  return blocks;
+}
+function isStructuralLine(line) {
+  if (/^#{1,6}\s+/.test(line)) return true;
+  if (line.trim().startsWith("|")) return true;
+  if (/^[-*_]{3,}\s*$/.test(line.trim())) return true;
+  if (/^\s*[-*+]\s+/.test(line)) return true;
+  if (/^\s*\d+\.\s+/.test(line)) return true;
+  return false;
+}
+function parseMarkdownTable(lines) {
+  const hasSeparator = lines.some((line) => /^\|[\s:|-]+\|$/.test(line.trim()));
+  const rows = [];
+  let maxCols = 0;
+  for (const line of lines) {
+    if (/^\|\s*:?-+:?\s*(\|\s*:?-+:?\s*)+\|?\s*$/.test(line.trim())) continue;
+    const parts = line.split("|");
+    const cells = parts.slice(1, parts[parts.length - 1].trim() === "" ? -1 : void 0).map((cell) => ({
+      text: cell.trim(),
+      colSpan: 1,
+      rowSpan: 1
+    }));
+    if (cells.length > 0) {
+      rows.push(cells);
+      maxCols = Math.max(maxCols, cells.length);
+    }
+  }
+  if (rows.length === 0) return null;
+  for (const row of rows) {
+    while (row.length < maxCols) {
+      row.push({ text: "", colSpan: 1, rowSpan: 1 });
+    }
+  }
+  return {
+    rows: rows.length,
+    cols: maxCols,
+    cells: rows,
+    hasHeader: hasSeparator && rows.length > 1
+  };
+}
+
+export {
+  markdownToBlocks
+};
+//# sourceMappingURL=chunk-4PP34NVQ.js.map

package/dist/chunk-4PP34NVQ.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/ocr/markdown-to-blocks.ts"],"sourcesContent":["/**\n * Markdown → IRBlock[] 역파싱\n *\n * Vision LLM(gemini/claude/codex 등)이 반환한 Markdown 문자열을\n * kordoc의 IRBlock[] 중간 표현으로 변환.\n * 기존 blocksToMarkdown()의 역방향 처리.\n */\n\nimport type { IRBlock, IRTable, IRCell } from \"../types.js\"\n\n/**\n * Markdown 문자열을 IRBlock[] 배열로 변환.\n *\n * 지원 요소:\n * - 헤딩: # ~ ######\n * - 테이블: | col1 | col2 | (파이프 구분, |---|---| 구분선 포함)\n * - 순서/비순서 리스트: - / 1.\n * - 구분선: ---, ***, ___\n * - 일반 텍스트 (paragraph)\n */\nexport function markdownToBlocks(markdown: string, pageNumber: number): IRBlock[] {\n  const blocks: IRBlock[] = []\n  const lines = markdown.split(\"\\n\")\n  let i = 0\n\n  while (i < lines.length) {\n    const line = lines[i]\n\n    // 빈 줄 스킵\n    if (line.trim() === \"\") {\n      i++\n      continue\n    }\n\n    // 1. 헤딩: # ~ ######\n    const headingMatch = line.match(/^(#{1,6})\\s+(.+)$/)\n    if (headingMatch) {\n      blocks.push({\n        type: \"heading\",\n        level: headingMatch[1].length,\n        text: headingMatch[2].trim(),\n        pageNumber,\n      })\n      i++\n      continue\n    }\n\n    // 2. 구분선: ---, ***, ___\n    if (/^[-*_]{3,}\\s*$/.test(line.trim())) {\n      blocks.push({ type: \"separator\", pageNumber })\n      i++\n      continue\n    }\n\n    // 3. 테이블: | 로 시작하는 연속 행 수집\n    if (line.trim().startsWith(\"|\")) {\n      const tableLines: string[] = []\n      while (i < lines.length && lines[i].trim().startsWith(\"|\")) {\n        tableLines.push(lines[i])\n        i++\n      }\n      const table = parseMarkdownTable(tableLines)\n      if (table) {\n        blocks.push({ type: \"table\", table, pageNumber })\n      }\n      continue\n    }\n\n    // 4. 비순서 리스트: -, *, +\n    const ulMatch = line.match(/^(\\s*)[-*+]\\s+(.+)$/)\n    if (ulMatch) {\n      blocks.push({\n        type: \"list\",\n        listType: \"unordered\",\n        text: ulMatch[2].trim(),\n        pageNumber,\n      })\n      i++\n      continue\n    }\n\n    // 5. 순서 리스트: 1.\n    const olMatch = line.match(/^(\\s*)\\d+\\.\\s+(.+)$/)\n    if (olMatch) {\n      blocks.push({\n        type: \"list\",\n        listType: \"ordered\",\n        text: olMatch[2].trim(),\n        pageNumber,\n      })\n      i++\n      continue\n    }\n\n    // 6. 일반 텍스트 — 구조적 행이 나올 때까지 병합\n    const paraLines: string[] = []\n    while (i < lines.length && lines[i].trim() !== \"\" && !isStructuralLine(lines[i])) {\n      paraLines.push(lines[i].trim())\n      i++\n    }\n    if (paraLines.length > 0) {\n      blocks.push({\n        type: \"paragraph\",\n        text: paraLines.join(\"\\n\"),\n        pageNumber,\n      })\n    }\n  }\n\n  return blocks\n}\n\n/**\n * 구조적 행 판별 — paragraph 병합 중단 트리거.\n */\nfunction isStructuralLine(line: string): boolean {\n  if (/^#{1,6}\\s+/.test(line)) return true\n  if (line.trim().startsWith(\"|\")) return true\n  if (/^[-*_]{3,}\\s*$/.test(line.trim())) return true\n  if (/^\\s*[-*+]\\s+/.test(line)) return true\n  if (/^\\s*\\d+\\.\\s+/.test(line)) return true\n  return false\n}\n\n/**\n * Markdown 테이블 행 배열을 IRTable로 변환.\n *\n * 구분선 행(|---|---|)은 제거 후 데이터 행만 파싱.\n * hasHeader: 구분선이 있었으면 true.\n */\nfunction parseMarkdownTable(lines: string[]): IRTable | null {\n  const hasSeparator = lines.some(line => /^\\|[\\s:|-]+\\|$/.test(line.trim()))\n\n  const rows: IRCell[][] = []\n  let maxCols = 0\n\n  for (const line of lines) {\n    // 구분선 행 스킵: |---|---| 패턴\n    if (/^\\|\\s*:?-+:?\\s*(\\|\\s*:?-+:?\\s*)+\\|?\\s*$/.test(line.trim())) continue\n\n    const parts = line.split(\"|\")\n    // 앞뒤 빈 요소 제거 (| 로 시작/종료하는 행)\n    const cells: IRCell[] = parts\n      .slice(1, parts[parts.length - 1].trim() === \"\" ? -1 : undefined)\n      .map(cell => ({\n        text: cell.trim(),\n        colSpan: 1,\n        rowSpan: 1,\n      }))\n\n    if (cells.length > 0) {\n      rows.push(cells)\n      maxCols = Math.max(maxCols, cells.length)\n    }\n  }\n\n  if (rows.length === 0) return null\n\n  // 열 수 통일 (부족한 셀은 빈 셀로 채움)\n  for (const row of rows) {\n    while (row.length < maxCols) {\n      row.push({ text: \"\", colSpan: 1, rowSpan: 1 })\n    }\n  }\n\n  return {\n    rows: rows.length,\n    cols: maxCols,\n    cells: rows,\n    hasHeader: hasSeparator && rows.length > 1,\n  }\n}\n"],"mappings":";;;AAoBO,SAAS,iBAAiB,UAAkB,YAA+B;AAChF,QAAM,SAAoB,CAAC;AAC3B,QAAM,QAAQ,SAAS,MAAM,IAAI;AACjC,MAAI,IAAI;AAER,SAAO,IAAI,MAAM,QAAQ;AACvB,UAAM,OAAO,MAAM,CAAC;AAGpB,QAAI,KAAK,KAAK,MAAM,IAAI;AACtB;AACA;AAAA,IACF;AAGA,UAAM,eAAe,KAAK,MAAM,mBAAmB;AACnD,QAAI,cAAc;AAChB,aAAO,KAAK;AAAA,QACV,MAAM;AAAA,QACN,OAAO,aAAa,CAAC,EAAE;AAAA,QACvB,MAAM,aAAa,CAAC,EAAE,KAAK;AAAA,QAC3B;AAAA,MACF,CAAC;AACD;AACA;AAAA,IACF;AAGA,QAAI,iBAAiB,KAAK,KAAK,KAAK,CAAC,GAAG;AACtC,aAAO,KAAK,EAAE,MAAM,aAAa,WAAW,CAAC;AAC7C;AACA;AAAA,IACF;AAGA,QAAI,KAAK,KAAK,EAAE,WAAW,GAAG,GAAG;AAC/B,YAAM,aAAuB,CAAC;AAC9B,aAAO,IAAI,MAAM,UAAU,MAAM,CAAC,EAAE,KAAK,EAAE,WAAW,GAAG,GAAG;AAC1D,mBAAW,KAAK,MAAM,CAAC,CAAC;AACxB;AAAA,MACF;AACA,YAAM,QAAQ,mBAAmB,UAAU;AAC3C,UAAI,OAAO;AACT,eAAO,KAAK,EAAE,MAAM,SAAS,OAAO,WAAW,CAAC;AAAA,MAClD;AACA;AAAA,IACF;AAGA,UAAM,UAAU,KAAK,MAAM,qBAAqB;AAChD,QAAI,SAAS;AACX,aAAO,KAAK;AAAA,QACV,MAAM;AAAA,QACN,UAAU;AAAA,QACV,MAAM,QAAQ,CAAC,EAAE,KAAK;AAAA,QACtB;AAAA,MACF,CAAC;AACD;AACA;AAAA,IACF;AAGA,UAAM,UAAU,KAAK,MAAM,qBAAqB;AAChD,QAAI,SAAS;AACX,aAAO,KAAK;AAAA,QACV,MAAM;AAAA,QACN,UAAU;AAAA,QACV,MAAM,QAAQ,CAAC,EAAE,KAAK;AAAA,QACtB;AAAA,MACF,CAAC;AACD;AACA;AAAA,IACF;AAGA,UAAM,YAAsB,CAAC;AAC7B,WAAO,IAAI,MAAM,UAAU,MAAM,CAAC,EAAE,KAAK,MAAM,MAAM,CAAC,iBAAiB,MAAM,CAAC,CAAC,GAAG;AAChF,gBAAU,KAAK,MAAM,CAAC,EAAE,KAAK,CAAC;AAC9B;AAAA,IACF;AACA,QAAI,UAAU,SAAS,GAAG;AACxB,aAAO,KAAK;AAAA,QACV,MAAM;AAAA,QACN,MAAM,UAAU,KAAK,IAAI;AAAA,QACzB;AAAA,MACF,CAAC;AAAA,IACH;AAAA,EACF;AAEA,SAAO;AACT;AAKA,SAAS,iBAAiB,MAAuB;AAC/C,MAAI,aAAa,KAAK,IAAI,EAAG,QAAO;AACpC,MAAI,KAAK,KAAK,EAAE,WAAW,GAAG,EAAG,QAAO;AACxC,MAAI,iBAAiB,KAAK,KAAK,KAAK,CAAC,EAAG,QAAO;AAC/C,MAAI,eAAe,KAAK,IAAI,EAAG,QAAO;AACtC,MAAI,eAAe,KAAK,IAAI,EAAG,QAAO;AACtC,SAAO;AACT;AAQA,SAAS,mBAAmB,OAAiC;AAC3D,QAAM,eAAe,MAAM,KAAK,UAAQ,iBAAiB,KAAK,KAAK,KAAK,CAAC,CAAC;AAE1E,QAAM,OAAmB,CAAC;AAC1B,MAAI,UAAU;AAEd,aAAW,QAAQ,OAAO;AAExB,QAAI,0CAA0C,KAAK,KAAK,KAAK,CAAC,EAAG;AAEjE,UAAM,QAAQ,KAAK,MAAM,GAAG;AAE5B,UAAM,QAAkB,MACrB,MAAM,GAAG,MAAM,MAAM,SAAS,CAAC,EAAE,KAAK,MAAM,KAAK,KAAK,MAAS,EAC/D,IAAI,WAAS;AAAA,MACZ,MAAM,KAAK,KAAK;AAAA,MAChB,SAAS;AAAA,MACT,SAAS;AAAA,IACX,EAAE;AAEJ,QAAI,MAAM,SAAS,GAAG;AACpB,WAAK,KAAK,KAAK;AACf,gBAAU,KAAK,IAAI,SAAS,MAAM,MAAM;AAAA,IAC1C;AAAA,EACF;AAEA,MAAI,KAAK,WAAW,EAAG,QAAO;AAG9B,aAAW,OAAO,MAAM;AACtB,WAAO,IAAI,SAAS,SAAS;AAC3B,UAAI,KAAK,EAAE,MAAM,IAAI,SAAS,GAAG,SAAS,EAAE,CAAC;AAAA,IAC/C;AAAA,EACF;AAEA,SAAO;AAAA,IACL,MAAM,KAAK;AAAA,IACX,MAAM;AAAA,IACN,OAAO;AAAA,IACP,WAAW,gBAAgB,KAAK,SAAS;AAAA,EAC3C;AACF;","names":[]}

package/dist/{tesseract-provider-WCVJWBUT.js → chunk-7FMKAV4P.js}

@@ -1,5 +1,4 @@
 #!/usr/bin/env node
-import "./chunk-ZWE3DS7E.js";
 
 // src/ocr/tesseract-provider.ts
 import { createWorker } from "tesseract.js";
@@ -49,8 +48,9 @@ async function createTesseractPoolProvider(concurrency) {
   };
   return provider;
 }
+
 export {
-  createTesseractPoolProvider,
-  createTesseractProvider
+  createTesseractProvider,
+  createTesseractPoolProvider
 };
-//# sourceMappingURL=tesseract-provider-WCVJWBUT.js.map
+//# sourceMappingURL=chunk-7FMKAV4P.js.map

package/dist/{tesseract-provider-WCVJWBUT.js.map → chunk-7FMKAV4P.js.map}

@@ -1 +1 @@
-{"version":3,"sources":["../src/ocr/tesseract-provider.ts"],"sourcesContent":["/**\n * Tesseract.js 기반 OCR 프로바이더\n *\n * tesseract.js는 bundled 의존성으로 별도 설치 불필요.\n * Vision LLM CLI가 없을 때의 최후 fallback으로 자동 사용.\n *\n * 특성:\n * - 순수 텍스트만 반환 (테이블/헤딩 구조 복원 불가)\n * - 한글 인식률 약 85-90% (깨끗한 이미지 기준)\n * - 완전 오프라인 동작 (API 키 불필요)\n */\n\nimport { createWorker } from \"tesseract.js\"\nimport type { OcrProvider } from \"../types.js\"\n\n/**\n * Tesseract.js OcrProvider 생성.\n *\n * 워커를 1회 생성하여 재사용 (매 페이지마다 초기화 방지).\n * 문서 처리 완료 후 terminate()로 워커 정리.\n *\n * @returns OcrProvider 함수 (+ terminate 메서드)\n */\nexport async function createTesseractProvider(): Promise<OcrProvider & { terminate: () => Promise<void> }> {\n  // kor+eng: 한글 + 영문 동시 인식 (한국 공문서 특성)\n  const worker = await createWorker(\"kor+eng\")\n  let terminated = false\n\n  const provider = async (\n    pageImage: Uint8Array,\n    _pageNumber: number,\n    _mimeType: \"image/png\"\n  ): Promise<string> => {\n    const { data } = await worker.recognize(pageImage)\n    return data.text\n  }\n\n  ;(provider as OcrProvider & { terminate: () => Promise<void> }).terminate = async () => {\n    if (!terminated) {\n      await worker.terminate()\n      terminated = true\n    }\n  }\n\n  return provider as OcrProvider & { terminate: () => Promise<void> }\n}\n\n/**\n * Tesseract.js 워커 풀 기반 병렬 OcrProvider 생성.\n *\n * N개 워커를 동시 생성하여 여러 페이지를 병렬로 OCR 처리.\n * acquire/release 패턴으로 동시 실행 안전성 보장.\n * 대량 이미지 기반 PDF(예: 300페이지) 처리 속도 향상에 효과적.\n *\n * @param concurrency - 동시 워커 수 (권장: CPU 코어 수)\n * @returns OcrProvider 함수 (+ terminate 메서드)\n */\nexport async function createTesseractPoolProvider(\n  concurrency: number\n): Promise<OcrProvider & { terminate: () => Promise<void> }> {\n  // N개 워커를 동시 초기화 (Promise.all로 병렬 생성)\n  const workers = await Promise.all(\n    Array.from({ length: concurrency }, () => createWorker(\"kor+eng\"))\n  )\n\n  // 유휴 워커 큐 (초기에는 모든 워커가 유휴 상태)\n  const idle: typeof workers = [...workers]\n  // 워커를 기다리는 콜백 큐\n  const waitQueue: ((w: (typeof workers)[0]) => void)[] = []\n\n  /** 유휴 워커를 획득 (없으면 대기) */\n  function acquire(): Promise<(typeof workers)[0]> {\n    if (idle.length > 0) return Promise.resolve(idle.pop()!)\n    return new Promise(resolve => waitQueue.push(resolve))\n  }\n\n  /** 사용 완료된 워커를 반환 (대기 중인 호출자에게 우선 전달) */\n  function release(w: (typeof workers)[0]): void {\n    if (waitQueue.length > 0) {\n      // 대기 중인 요청자에게 즉시 전달\n      waitQueue.shift()!(w)\n    } else {\n      idle.push(w)\n    }\n  }\n\n  // 병렬 호출 안전 프로바이더 — 각 호출이 acquire/release로 워커를 독점 사용\n  const provider = async (\n    pageImage: Uint8Array,\n    _pageNumber: number,\n    _mimeType: \"image/png\"\n  ): Promise<string> => {\n    const w = await acquire()\n    try {\n      const { data } = await w.recognize(pageImage)\n      return data.text\n    } finally {\n      // 성공/실패 무관하게 워커 반환\n      release(w)\n    }\n  }\n\n  // 모든 워커 정리 메서드\n  ;(provider as OcrProvider & { terminate: () => Promise<void> }).terminate = async () => {\n    await Promise.all(workers.map(w => w.terminate()))\n  }\n\n  return provider as OcrProvider & { terminate: () => Promise<void> }\n}\n"],"mappings":";;;;AAYA,SAAS,oBAAoB;AAW7B,eAAsB,0BAAqF;AAEzG,QAAM,SAAS,MAAM,aAAa,SAAS;AAC3C,MAAI,aAAa;AAEjB,QAAM,WAAW,OACf,WACA,aACA,cACoB;AACpB,UAAM,EAAE,KAAK,IAAI,MAAM,OAAO,UAAU,SAAS;AACjD,WAAO,KAAK;AAAA,EACd;AAEC,EAAC,SAA8D,YAAY,YAAY;AACtF,QAAI,CAAC,YAAY;AACf,YAAM,OAAO,UAAU;AACvB,mBAAa;AAAA,IACf;AAAA,EACF;AAEA,SAAO;AACT;AAYA,eAAsB,4BACpB,aAC2D;AAE3D,QAAM,UAAU,MAAM,QAAQ;AAAA,IAC5B,MAAM,KAAK,EAAE,QAAQ,YAAY,GAAG,MAAM,aAAa,SAAS,CAAC;AAAA,EACnE;AAGA,QAAM,OAAuB,CAAC,GAAG,OAAO;AAExC,QAAM,YAAkD,CAAC;AAGzD,WAAS,UAAwC;AAC/C,QAAI,KAAK,SAAS,EAAG,QAAO,QAAQ,QAAQ,KAAK,IAAI,CAAE;AACvD,WAAO,IAAI,QAAQ,aAAW,UAAU,KAAK,OAAO,CAAC;AAAA,EACvD;AAGA,WAAS,QAAQ,GAA8B;AAC7C,QAAI,UAAU,SAAS,GAAG;AAExB,gBAAU,MAAM,EAAG,CAAC;AAAA,IACtB,OAAO;AACL,WAAK,KAAK,CAAC;AAAA,IACb;AAAA,EACF;AAGA,QAAM,WAAW,OACf,WACA,aACA,cACoB;AACpB,UAAM,IAAI,MAAM,QAAQ;AACxB,QAAI;AACF,YAAM,EAAE,KAAK,IAAI,MAAM,EAAE,UAAU,SAAS;AAC5C,aAAO,KAAK;AAAA,IACd,UAAE;AAEA,cAAQ,CAAC;AAAA,IACX;AAAA,EACF;AAGC,EAAC,SAA8D,YAAY,YAAY;AACtF,UAAM,QAAQ,IAAI,QAAQ,IAAI,OAAK,EAAE,UAAU,CAAC,CAAC;AAAA,EACnD;AAEA,SAAO;AACT;","names":[]}
+{"version":3,"sources":["../src/ocr/tesseract-provider.ts"],"sourcesContent":["/**\n * Tesseract.js 기반 OCR 프로바이더\n *\n * tesseract.js는 bundled 의존성으로 별도 설치 불필요.\n * Vision LLM CLI가 없을 때의 최후 fallback으로 자동 사용.\n *\n * 특성:\n * - 순수 텍스트만 반환 (테이블/헤딩 구조 복원 불가)\n * - 한글 인식률 약 85-90% (깨끗한 이미지 기준)\n * - 완전 오프라인 동작 (API 키 불필요)\n */\n\nimport { createWorker } from \"tesseract.js\"\nimport type { OcrProvider } from \"../types.js\"\n\n/**\n * Tesseract.js OcrProvider 생성.\n *\n * 워커를 1회 생성하여 재사용 (매 페이지마다 초기화 방지).\n * 문서 처리 완료 후 terminate()로 워커 정리.\n *\n * @returns OcrProvider 함수 (+ terminate 메서드)\n */\nexport async function createTesseractProvider(): Promise<OcrProvider & { terminate: () => Promise<void> }> {\n  // kor+eng: 한글 + 영문 동시 인식 (한국 공문서 특성)\n  const worker = await createWorker(\"kor+eng\")\n  let terminated = false\n\n  const provider = async (\n    pageImage: Uint8Array,\n    _pageNumber: number,\n    _mimeType: \"image/png\"\n  ): Promise<string> => {\n    const { data } = await worker.recognize(pageImage)\n    return data.text\n  }\n\n  ;(provider as OcrProvider & { terminate: () => Promise<void> }).terminate = async () => {\n    if (!terminated) {\n      await worker.terminate()\n      terminated = true\n    }\n  }\n\n  return provider as OcrProvider & { terminate: () => Promise<void> }\n}\n\n/**\n * Tesseract.js 워커 풀 기반 병렬 OcrProvider 생성.\n *\n * N개 워커를 동시 생성하여 여러 페이지를 병렬로 OCR 처리.\n * acquire/release 패턴으로 동시 실행 안전성 보장.\n * 대량 이미지 기반 PDF(예: 300페이지) 처리 속도 향상에 효과적.\n *\n * @param concurrency - 동시 워커 수 (권장: CPU 코어 수)\n * @returns OcrProvider 함수 (+ terminate 메서드)\n */\nexport async function createTesseractPoolProvider(\n  concurrency: number\n): Promise<OcrProvider & { terminate: () => Promise<void> }> {\n  // N개 워커를 동시 초기화 (Promise.all로 병렬 생성)\n  const workers = await Promise.all(\n    Array.from({ length: concurrency }, () => createWorker(\"kor+eng\"))\n  )\n\n  // 유휴 워커 큐 (초기에는 모든 워커가 유휴 상태)\n  const idle: typeof workers = [...workers]\n  // 워커를 기다리는 콜백 큐\n  const waitQueue: ((w: (typeof workers)[0]) => void)[] = []\n\n  /** 유휴 워커를 획득 (없으면 대기) */\n  function acquire(): Promise<(typeof workers)[0]> {\n    if (idle.length > 0) return Promise.resolve(idle.pop()!)\n    return new Promise(resolve => waitQueue.push(resolve))\n  }\n\n  /** 사용 완료된 워커를 반환 (대기 중인 호출자에게 우선 전달) */\n  function release(w: (typeof workers)[0]): void {\n    if (waitQueue.length > 0) {\n      // 대기 중인 요청자에게 즉시 전달\n      waitQueue.shift()!(w)\n    } else {\n      idle.push(w)\n    }\n  }\n\n  // 병렬 호출 안전 프로바이더 — 각 호출이 acquire/release로 워커를 독점 사용\n  const provider = async (\n    pageImage: Uint8Array,\n    _pageNumber: number,\n    _mimeType: \"image/png\"\n  ): Promise<string> => {\n    const w = await acquire()\n    try {\n      const { data } = await w.recognize(pageImage)\n      return data.text\n    } finally {\n      // 성공/실패 무관하게 워커 반환\n      release(w)\n    }\n  }\n\n  // 모든 워커 정리 메서드\n  ;(provider as OcrProvider & { terminate: () => Promise<void> }).terminate = async () => {\n    await Promise.all(workers.map(w => w.terminate()))\n  }\n\n  return provider as OcrProvider & { terminate: () => Promise<void> }\n}\n"],"mappings":";;;AAYA,SAAS,oBAAoB;AAW7B,eAAsB,0BAAqF;AAEzG,QAAM,SAAS,MAAM,aAAa,SAAS;AAC3C,MAAI,aAAa;AAEjB,QAAM,WAAW,OACf,WACA,aACA,cACoB;AACpB,UAAM,EAAE,KAAK,IAAI,MAAM,OAAO,UAAU,SAAS;AACjD,WAAO,KAAK;AAAA,EACd;AAEC,EAAC,SAA8D,YAAY,YAAY;AACtF,QAAI,CAAC,YAAY;AACf,YAAM,OAAO,UAAU;AACvB,mBAAa;AAAA,IACf;AAAA,EACF;AAEA,SAAO;AACT;AAYA,eAAsB,4BACpB,aAC2D;AAE3D,QAAM,UAAU,MAAM,QAAQ;AAAA,IAC5B,MAAM,KAAK,EAAE,QAAQ,YAAY,GAAG,MAAM,aAAa,SAAS,CAAC;AAAA,EACnE;AAGA,QAAM,OAAuB,CAAC,GAAG,OAAO;AAExC,QAAM,YAAkD,CAAC;AAGzD,WAAS,UAAwC;AAC/C,QAAI,KAAK,SAAS,EAAG,QAAO,QAAQ,QAAQ,KAAK,IAAI,CAAE;AACvD,WAAO,IAAI,QAAQ,aAAW,UAAU,KAAK,OAAO,CAAC;AAAA,EACvD;AAGA,WAAS,QAAQ,GAA8B;AAC7C,QAAI,UAAU,SAAS,GAAG;AAExB,gBAAU,MAAM,EAAG,CAAC;AAAA,IACtB,OAAO;AACL,WAAK,KAAK,CAAC;AAAA,IACb;AAAA,EACF;AAGA,QAAM,WAAW,OACf,WACA,aACA,cACoB;AACpB,UAAM,IAAI,MAAM,QAAQ;AACxB,QAAI;AACF,YAAM,EAAE,KAAK,IAAI,MAAM,EAAE,UAAU,SAAS;AAC5C,aAAO,KAAK;AAAA,IACd,UAAE;AAEA,cAAQ,CAAC;AAAA,IACX;AAAA,EACF;AAGC,EAAC,SAA8D,YAAY,YAAY;AACtF,UAAM,QAAQ,IAAI,QAAQ,IAAI,OAAK,EAAE,UAAU,CAAC,CAAC;AAAA,EACnD;AAEA,SAAO;AACT;","names":[]}

package/dist/chunk-JOGAFNIL.js

@@ -0,0 +1,153 @@
+#!/usr/bin/env node
+
+// src/ocr/cli-provider.ts
+import { spawnSync } from "child_process";
+import { writeFileSync, readFileSync, unlinkSync, mkdirSync } from "fs";
+import { join } from "path";
+import { tmpdir } from "os";
+var OCR_PROMPT = `\uC774 PDF \uD398\uC774\uC9C0 \uC774\uBBF8\uC9C0\uC5D0\uC11C \uD14D\uC2A4\uD2B8\uC640 \uD14C\uC774\uBE14\uC744 \uCD94\uCD9C\uD558\uC5EC \uC21C\uC218 Markdown\uC73C\uB85C \uBCC0\uD658\uD558\uC138\uC694.
+\uADDC\uCE59:
+- \uD14C\uC774\uBE14\uC740 Markdown \uD14C\uC774\uBE14 \uBB38\uBC95 \uC0AC\uC6A9 (| \uAD6C\uBD84, |---|---| \uD5E4\uB354 \uAD6C\uBD84\uC120 \uD3EC\uD568)
+- \uBCD1\uD569\uB41C \uC140\uC740 \uD574\uB2F9 \uC704\uCE58\uC5D0 \uB0B4\uC6A9 \uAE30\uC7AC
+- \uD5E4\uB529\uC740 \uAE00\uC790 \uD06C\uAE30\uC5D0 \uB530\uB77C ## ~ ###### \uC0AC\uC6A9
+- \uB9AC\uC2A4\uD2B8\uB294 - \uB610\uB294 1. \uC0AC\uC6A9
+- \uC774\uBBF8\uC9C0, \uB3C4\uD615 \uB4F1 \uBE44\uD14D\uC2A4\uD2B8 \uC694\uC18C\uB294 \uBB34\uC2DC
+- \uC6D0\uBB38\uC758 \uC77D\uAE30 \uC21C\uC11C\uC640 \uAD6C\uC870\uB97C \uC720\uC9C0
+- \`\`\`\uB85C \uAC10\uC2F8\uC9C0 \uB9D0\uACE0 \uC21C\uC218 Markdown\uB9CC \uCD9C\uB825`;
+var _tempDir = null;
+function getTempDir() {
+  if (!_tempDir) {
+    _tempDir = join(process.cwd(), "_kordoc_ocr_tmp");
+    mkdirSync(_tempDir, { recursive: true });
+  }
+  return _tempDir;
+}
+function createCliOcrProvider(mode) {
+  return async (pageImage, pageNumber) => {
+    const tempPath = join(getTempDir(), `page-${pageNumber}.png`);
+    try {
+      writeFileSync(tempPath, pageImage);
+      let output;
+      if (mode === "ollama") {
+        output = await callOllamaApi(tempPath);
+      } else {
+        output = callCli(mode, tempPath);
+      }
+      return { markdown: stripCodeFence(output.trim()) };
+    } finally {
+      try {
+        unlinkSync(tempPath);
+      } catch {
+      }
+    }
+  };
+}
+function callCli(mode, imagePath) {
+  if (mode === "codex") {
+    return callCodexCli(imagePath);
+  }
+  const args = buildCliArgs(mode, imagePath);
+  const result = spawnSync(mode, args, {
+    encoding: "utf-8",
+    timeout: 6e5,
+    maxBuffer: 10 * 1024 * 1024,
+    // claude: /tmp에서 실행하여 프로젝트 CLAUDE.md의 규칙 간섭 방지
+    ...mode === "claude" ? { cwd: tmpdir() } : {}
+  });
+  if (result.error) {
+    throw new Error(`${mode} CLI \uC2E4\uD589 \uC2E4\uD328: ${result.error.message}`);
+  }
+  if (result.status !== 0) {
+    const errMsg = result.stderr?.trim() || `exit code ${result.status}`;
+    throw new Error(`${mode} OCR \uC2E4\uD328: ${errMsg}`);
+  }
+  return result.stdout || "";
+}
+function callCodexCli(imagePath) {
+  const outPath = join(tmpdir(), `kordoc-codex-out-${Date.now()}.txt`);
+  try {
+    const args = ["exec", OCR_PROMPT, "--image", imagePath, "--output-last-message", outPath];
+    const model = process.env.KORDOC_CODEX_MODEL;
+    if (model) args.push("--model", model);
+    const result = spawnSync("codex", args, {
+      encoding: "utf-8",
+      timeout: 18e4,
+      maxBuffer: 10 * 1024 * 1024,
+      input: ""
+      // stdin EOF 즉시 전달 (대화형 입력 차단)
+    });
+    if (result.error) {
+      throw new Error(`codex CLI \uC2E4\uD589 \uC2E4\uD328: ${result.error.message}`);
+    }
+    if (result.status !== 0) {
+      const errMsg = result.stderr?.trim() || `exit code ${result.status}`;
+      throw new Error(`codex OCR \uC2E4\uD328: ${errMsg}`);
+    }
+    try {
+      return readFileSync(outPath, "utf-8");
+    } catch {
+      return result.stdout || "";
+    }
+  } finally {
+    try {
+      unlinkSync(outPath);
+    } catch {
+    }
+  }
+}
+function buildCliArgs(mode, imagePath) {
+  const promptWithImage = `${OCR_PROMPT}
+
+\uC774\uBBF8\uC9C0: @${imagePath}`;
+  switch (mode) {
+    case "gemini": {
+      const args = ["--prompt", promptWithImage, "--yolo"];
+      const model = process.env.KORDOC_GEMINI_MODEL;
+      if (model) args.push("--model", model);
+      return args;
+    }
+    case "claude": {
+      const args = ["--print", promptWithImage];
+      const model = process.env.KORDOC_CLAUDE_MODEL;
+      if (model) args.push("--model", model);
+      return args;
+    }
+    default:
+      throw new Error(`\uC9C0\uC6D0\uD558\uC9C0 \uC54A\uB294 CLI: ${mode}`);
+  }
+}
+async function callOllamaApi(imagePath) {
+  const { readFileSync: readFileSync2 } = await import("fs");
+  const imageBase64 = readFileSync2(imagePath).toString("base64");
+  const model = process.env.KORDOC_OLLAMA_MODEL || "qwen3-vl:8b";
+  const host = process.env.KORDOC_OLLAMA_HOST || "http://localhost:11434";
+  const timeoutMs = Number(process.env.KORDOC_OLLAMA_TIMEOUT) || 12e4;
+  const response = await fetch(`${host}/api/chat`, {
+    method: "POST",
+    headers: { "Content-Type": "application/json" },
+    body: JSON.stringify({
+      model,
+      messages: [{
+        role: "user",
+        content: OCR_PROMPT,
+        images: [imageBase64]
+      }],
+      stream: false
+    }),
+    signal: AbortSignal.timeout(timeoutMs)
+  });
+  if (!response.ok) {
+    throw new Error(`Ollama API \uC624\uB958: ${response.status} ${response.statusText}`);
+  }
+  const data = await response.json();
+  return data.message?.content || "";
+}
+function stripCodeFence(text) {
+  const match = text.match(/^```(?:markdown|md)?\s*([\s\S]*?)```\s*$/m);
+  return match ? match[1].trim() : text;
+}
+
+export {
+  createCliOcrProvider
+};
+//# sourceMappingURL=chunk-JOGAFNIL.js.map

package/dist/chunk-JOGAFNIL.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/ocr/cli-provider.ts"],"sourcesContent":["/**\n * CLI 기반 OCR 프로바이더\n *\n * gemini / claude / codex / ollama CLI를 subprocess로 호출하여\n * PDF 페이지 이미지를 Markdown으로 변환.\n *\n * 이미지 전달 방식:\n * - gemini:  -p \"프롬프트 @이미지경로\"  (@ 파일 참조)\n * - claude:  -p \"프롬프트 @이미지경로\"  (@ 파일 참조, --print 모드)\n * - codex:   exec -i 이미지경로 \"프롬프트\"  (-i/--image 플래그)\n * - ollama:  REST API (localhost:11434) — CLI는 이미지 입력 미지원\n */\n\nimport { spawnSync } from \"child_process\"\nimport { writeFileSync, readFileSync, unlinkSync, mkdirSync } from \"fs\"\nimport { join } from \"path\"\nimport { tmpdir } from \"os\"\nimport type { OcrMode, StructuredOcrResult } from \"../types.js\"\n\n/** OCR 프롬프트 — 모든 CLI 공통 */\nconst OCR_PROMPT = `이 PDF 페이지 이미지에서 텍스트와 테이블을 추출하여 순수 Markdown으로 변환하세요.\n규칙:\n- 테이블은 Markdown 테이블 문법 사용 (| 구분, |---|---| 헤더 구분선 포함)\n- 병합된 셀은 해당 위치에 내용 기재\n- 헤딩은 글자 크기에 따라 ## ~ ###### 사용\n- 리스트는 - 또는 1. 사용\n- 이미지, 도형 등 비텍스트 요소는 무시\n- 원문의 읽기 순서와 구조를 유지\n- \\`\\`\\`로 감싸지 말고 순수 Markdown만 출력`\n\n/** 임시 디렉토리 (프로세스당 1회 생성)\n *\n * gemini CLI는 /tmp/ 등 시스템 임시 디렉토리를 워크스페이스 외부로 간주하여\n * @파일참조 시 접근을 거부할 수 있음. cwd 하위 폴더를 사용하면 모든 CLI에서 접근 가능.\n *\n * ⚠️ .gitignore에 포함된 경로(예: .kordoc-tmp/)는 gemini CLI가 무시하므로\n * 반드시 gitignore되지 않는 이름 사용. 파일은 try/finally로 즉시 정리.\n */\nlet _tempDir: string | null = null\nfunction getTempDir(): string {\n  if (!_tempDir) {\n    _tempDir = join(process.cwd(), \"_kordoc_ocr_tmp\")\n    mkdirSync(_tempDir, { recursive: true })\n  }\n  return _tempDir\n}\n\n/**\n * CLI OcrProvider 생성.\n *\n * @param mode - 사용할 CLI (gemini, claude, codex, ollama)\n * @returns OcrProvider 함수 (StructuredOcrResult 반환)\n */\nexport function createCliOcrProvider(\n  mode: Exclude<OcrMode, \"auto\" | \"off\" | \"tesseract\">\n): (pageImage: Uint8Array, pageNumber: number, mimeType: \"image/png\") => Promise<StructuredOcrResult> {\n  return async (pageImage: Uint8Array, pageNumber: number): Promise<StructuredOcrResult> => {\n    const tempPath = join(getTempDir(), `page-${pageNumber}.png`)\n\n    try {\n      writeFileSync(tempPath, pageImage)\n\n      let output: string\n      if (mode === \"ollama\") {\n        output = await callOllamaApi(tempPath)\n      } else {\n        output = callCli(mode, tempPath)\n      }\n\n      return { markdown: stripCodeFence(output.trim()) }\n    } finally {\n      try { unlinkSync(tempPath) } catch { /* 임시 파일 정리 실패 무시 */ }\n    }\n  }\n}\n\n/**\n * CLI 실행 — gemini / claude / codex\n *\n * @throws CLI 실행 실패 또는 타임아웃(180초) 시 Error\n */\nfunction callCli(mode: string, imagePath: string): string {\n  // codex는 --output-last-message로 대화 헤더 없는 깔끔한 출력 사용\n  if (mode === \"codex\") {\n    return callCodexCli(imagePath)\n  }\n\n  const args = buildCliArgs(mode, imagePath)\n\n  const result = spawnSync(mode, args, {\n    encoding: \"utf-8\",\n    timeout: 600_000,\n    maxBuffer: 10 * 1024 * 1024,\n    // claude: /tmp에서 실행하여 프로젝트 CLAUDE.md의 규칙 간섭 방지\n    ...(mode === \"claude\" ? { cwd: tmpdir() } : {}),\n  })\n\n  if (result.error) {\n    throw new Error(`${mode} CLI 실행 실패: ${result.error.message}`)\n  }\n  if (result.status !== 0) {\n    const errMsg = result.stderr?.trim() || `exit code ${result.status}`\n    throw new Error(`${mode} OCR 실패: ${errMsg}`)\n  }\n\n  return result.stdout || \"\"\n}\n\n/**\n * codex exec 실행 — --output-last-message로 대화 헤더 없는 깔끔한 출력.\n * 인자 순서: `codex exec <prompt> --image <file> --output-last-message <outfile>`\n */\nfunction callCodexCli(imagePath: string): string {\n  // 출력 파일은 /tmp/ 사용 — codex sandbox는 cwd 내 쓰기를 막을 수 있음\n  const outPath = join(tmpdir(), `kordoc-codex-out-${Date.now()}.txt`)\n  try {\n    const args = [\"exec\", OCR_PROMPT, \"--image\", imagePath, \"--output-last-message\", outPath]\n    const model = process.env.KORDOC_CODEX_MODEL\n    if (model) args.push(\"--model\", model)\n\n    const result = spawnSync(\"codex\", args, {\n      encoding: \"utf-8\",\n      timeout: 180_000,\n      maxBuffer: 10 * 1024 * 1024,\n      input: \"\",  // stdin EOF 즉시 전달 (대화형 입력 차단)\n    })\n\n    if (result.error) {\n      throw new Error(`codex CLI 실행 실패: ${result.error.message}`)\n    }\n    if (result.status !== 0) {\n      const errMsg = result.stderr?.trim() || `exit code ${result.status}`\n      throw new Error(`codex OCR 실패: ${errMsg}`)\n    }\n\n    // --output-last-message 파일에서 읽기 (없으면 stdout 폴백)\n    try {\n      return readFileSync(outPath, \"utf-8\")\n    } catch {\n      return result.stdout || \"\"\n    }\n  } finally {\n    try { unlinkSync(outPath) } catch { /* 무시 */ }\n  }\n}\n\n/**\n * CLI별 인자 배열 생성.\n *\n * gemini:  [\"--prompt\", \"프롬프트 @이미지경로\", \"--yolo\"]\n *   - -y/--yolo: 자동 승인 (OCR은 도구 사용 없으므로 실질적 영향 없음)\n *   - @ 파일 참조로 이미지를 컨텍스트에 포함\n *\n * claude:  [\"--print\", \"프롬프트 @이미지경로\"]\n *   - --print(-p): 비대화형 출력 모드\n *   - @ 파일 참조로 이미지를 컨텍스트에 포함\n *\n * codex: callCodexCli()에서 별도 처리\n *   - `codex exec <prompt> --image <file> --output-last-message <outfile>`\n *   - 프롬프트가 --image보다 앞에 위치해야 함 (인자 순서 중요)\n *\n * ⚠️ CLI 버전에 따라 문법이 다를 수 있음. 업데이트 시 --help 재확인 필요.\n */\nfunction buildCliArgs(mode: string, imagePath: string): string[] {\n  const promptWithImage = `${OCR_PROMPT}\n\n이미지: @${imagePath}`\n\n  switch (mode) {\n    case \"gemini\": {\n      const args = [\"--prompt\", promptWithImage, \"--yolo\"]\n      const model = process.env.KORDOC_GEMINI_MODEL\n      if (model) args.push(\"--model\", model)\n      return args\n    }\n\n    case \"claude\": {\n      const args = [\"--print\", promptWithImage]\n      const model = process.env.KORDOC_CLAUDE_MODEL\n      if (model) args.push(\"--model\", model)\n      return args\n    }\n\n    default:\n      throw new Error(`지원하지 않는 CLI: ${mode}`)\n  }\n}\n\n/**\n * Ollama REST API 호출 — CLI는 이미지 입력을 지원하지 않으므로 API 직접 사용.\n *\n * 기본 모델: KORDOC_OLLAMA_MODEL 환경변수 또는 \"gemma4:27b\"\n * 기본 호스트: KORDOC_OLLAMA_HOST 환경변수 또는 \"http://localhost:11434\"\n *\n * @throws Ollama 서버 미실행 또는 응답 오류 시 Error\n */\nasync function callOllamaApi(imagePath: string): Promise<string> {\n  const { readFileSync } = await import(\"fs\")\n  const imageBase64 = readFileSync(imagePath).toString(\"base64\")\n\n  const model = process.env.KORDOC_OLLAMA_MODEL || \"qwen3-vl:8b\"\n  const host = process.env.KORDOC_OLLAMA_HOST || \"http://localhost:11434\"\n  const timeoutMs = Number(process.env.KORDOC_OLLAMA_TIMEOUT) || 120_000\n\n  const response = await fetch(`${host}/api/chat`, {\n    method: \"POST\",\n    headers: { \"Content-Type\": \"application/json\" },\n    body: JSON.stringify({\n      model,\n      messages: [{\n        role: \"user\",\n        content: OCR_PROMPT,\n        images: [imageBase64],\n      }],\n      stream: false,\n    }),\n    signal: AbortSignal.timeout(timeoutMs),\n  })\n\n  if (!response.ok) {\n    throw new Error(`Ollama API 오류: ${response.status} ${response.statusText}`)\n  }\n\n  const data = await response.json() as { message?: { content?: string } }\n  return data.message?.content || \"\"\n}\n\n/**\n * LLM 출력에서 코드 펜스 제거.\n * LLM이 가끔 결과를 ```markdown ... ``` 으로 감싸는 경우 처리.\n */\nfunction stripCodeFence(text: string): string {\n  const match = text.match(/^```(?:markdown|md)?\\s*([\\s\\S]*?)```\\s*$/m)\n  return match ? match[1].trim() : text\n}\n"],"mappings":";;;AAaA,SAAS,iBAAiB;AAC1B,SAAS,eAAe,cAAc,YAAY,iBAAiB;AACnE,SAAS,YAAY;AACrB,SAAS,cAAc;AAIvB,IAAM,aAAa;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAkBnB,IAAI,WAA0B;AAC9B,SAAS,aAAqB;AAC5B,MAAI,CAAC,UAAU;AACb,eAAW,KAAK,QAAQ,IAAI,GAAG,iBAAiB;AAChD,cAAU,UAAU,EAAE,WAAW,KAAK,CAAC;AAAA,EACzC;AACA,SAAO;AACT;AAQO,SAAS,qBACd,MACoG;AACpG,SAAO,OAAO,WAAuB,eAAqD;AACxF,UAAM,WAAW,KAAK,WAAW,GAAG,QAAQ,UAAU,MAAM;AAE5D,QAAI;AACF,oBAAc,UAAU,SAAS;AAEjC,UAAI;AACJ,UAAI,SAAS,UAAU;AACrB,iBAAS,MAAM,cAAc,QAAQ;AAAA,MACvC,OAAO;AACL,iBAAS,QAAQ,MAAM,QAAQ;AAAA,MACjC;AAEA,aAAO,EAAE,UAAU,eAAe,OAAO,KAAK,CAAC,EAAE;AAAA,IACnD,UAAE;AACA,UAAI;AAAE,mBAAW,QAAQ;AAAA,MAAE,QAAQ;AAAA,MAAuB;AAAA,IAC5D;AAAA,EACF;AACF;AAOA,SAAS,QAAQ,MAAc,WAA2B;AAExD,MAAI,SAAS,SAAS;AACpB,WAAO,aAAa,SAAS;AAAA,EAC/B;AAEA,QAAM,OAAO,aAAa,MAAM,SAAS;AAEzC,QAAM,SAAS,UAAU,MAAM,MAAM;AAAA,IACnC,UAAU;AAAA,IACV,SAAS;AAAA,IACT,WAAW,KAAK,OAAO;AAAA;AAAA,IAEvB,GAAI,SAAS,WAAW,EAAE,KAAK,OAAO,EAAE,IAAI,CAAC;AAAA,EAC/C,CAAC;AAED,MAAI,OAAO,OAAO;AAChB,UAAM,IAAI,MAAM,GAAG,IAAI,mCAAe,OAAO,MAAM,OAAO,EAAE;AAAA,EAC9D;AACA,MAAI,OAAO,WAAW,GAAG;AACvB,UAAM,SAAS,OAAO,QAAQ,KAAK,KAAK,aAAa,OAAO,MAAM;AAClE,UAAM,IAAI,MAAM,GAAG,IAAI,sBAAY,MAAM,EAAE;AAAA,EAC7C;AAEA,SAAO,OAAO,UAAU;AAC1B;AAMA,SAAS,aAAa,WAA2B;AAE/C,QAAM,UAAU,KAAK,OAAO,GAAG,oBAAoB,KAAK,IAAI,CAAC,MAAM;AACnE,MAAI;AACF,UAAM,OAAO,CAAC,QAAQ,YAAY,WAAW,WAAW,yBAAyB,OAAO;AACxF,UAAM,QAAQ,QAAQ,IAAI;AAC1B,QAAI,MAAO,MAAK,KAAK,WAAW,KAAK;AAErC,UAAM,SAAS,UAAU,SAAS,MAAM;AAAA,MACtC,UAAU;AAAA,MACV,SAAS;AAAA,MACT,WAAW,KAAK,OAAO;AAAA,MACvB,OAAO;AAAA;AAAA,IACT,CAAC;AAED,QAAI,OAAO,OAAO;AAChB,YAAM,IAAI,MAAM,wCAAoB,OAAO,MAAM,OAAO,EAAE;AAAA,IAC5D;AACA,QAAI,OAAO,WAAW,GAAG;AACvB,YAAM,SAAS,OAAO,QAAQ,KAAK,KAAK,aAAa,OAAO,MAAM;AAClE,YAAM,IAAI,MAAM,2BAAiB,MAAM,EAAE;AAAA,IAC3C;AAGA,QAAI;AACF,aAAO,aAAa,SAAS,OAAO;AAAA,IACtC,QAAQ;AACN,aAAO,OAAO,UAAU;AAAA,IAC1B;AAAA,EACF,UAAE;AACA,QAAI;AAAE,iBAAW,OAAO;AAAA,IAAE,QAAQ;AAAA,IAAW;AAAA,EAC/C;AACF;AAmBA,SAAS,aAAa,MAAc,WAA6B;AAC/D,QAAM,kBAAkB,GAAG,UAAU;AAAA;AAAA,uBAE/B,SAAS;AAEf,UAAQ,MAAM;AAAA,IACZ,KAAK,UAAU;AACb,YAAM,OAAO,CAAC,YAAY,iBAAiB,QAAQ;AACnD,YAAM,QAAQ,QAAQ,IAAI;AAC1B,UAAI,MAAO,MAAK,KAAK,WAAW,KAAK;AACrC,aAAO;AAAA,IACT;AAAA,IAEA,KAAK,UAAU;AACb,YAAM,OAAO,CAAC,WAAW,eAAe;AACxC,YAAM,QAAQ,QAAQ,IAAI;AAC1B,UAAI,MAAO,MAAK,KAAK,WAAW,KAAK;AACrC,aAAO;AAAA,IACT;AAAA,IAEA;AACE,YAAM,IAAI,MAAM,8CAAgB,IAAI,EAAE;AAAA,EAC1C;AACF;AAUA,eAAe,cAAc,WAAoC;AAC/D,QAAM,EAAE,cAAAA,cAAa,IAAI,MAAM,OAAO,IAAI;AAC1C,QAAM,cAAcA,cAAa,SAAS,EAAE,SAAS,QAAQ;AAE7D,QAAM,QAAQ,QAAQ,IAAI,uBAAuB;AACjD,QAAM,OAAO,QAAQ,IAAI,sBAAsB;AAC/C,QAAM,YAAY,OAAO,QAAQ,IAAI,qBAAqB,KAAK;AAE/D,QAAM,WAAW,MAAM,MAAM,GAAG,IAAI,aAAa;AAAA,IAC/C,QAAQ;AAAA,IACR,SAAS,EAAE,gBAAgB,mBAAmB;AAAA,IAC9C,MAAM,KAAK,UAAU;AAAA,MACnB;AAAA,MACA,UAAU,CAAC;AAAA,QACT,MAAM;AAAA,QACN,SAAS;AAAA,QACT,QAAQ,CAAC,WAAW;AAAA,MACtB,CAAC;AAAA,MACD,QAAQ;AAAA,IACV,CAAC;AAAA,IACD,QAAQ,YAAY,QAAQ,SAAS;AAAA,EACvC,CAAC;AAED,MAAI,CAAC,SAAS,IAAI;AAChB,UAAM,IAAI,MAAM,4BAAkB,SAAS,MAAM,IAAI,SAAS,UAAU,EAAE;AAAA,EAC5E;AAEA,QAAM,OAAO,MAAM,SAAS,KAAK;AACjC,SAAO,KAAK,SAAS,WAAW;AAClC;AAMA,SAAS,eAAe,MAAsB;AAC5C,QAAM,QAAQ,KAAK,MAAM,2CAA2C;AACpE,SAAO,QAAQ,MAAM,CAAC,EAAE,KAAK,IAAI;AACnC;","names":["readFileSync"]}

package/dist/{chunk-W5KUC23B.js → chunk-STIKJGEA.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 
 // src/utils.ts
-var VERSION = true ? "2.3.1" : "0.0.0-dev";
+var VERSION = true ? "2.3.2" : "0.0.0-dev";
 function toArrayBuffer(buf) {
   if (buf.byteOffset === 0 && buf.byteLength === buf.buffer.byteLength) {
     return buf.buffer;
@@ -90,4 +90,4 @@ export {
   sanitizeHref,
   classifyError
 };
-//# sourceMappingURL=chunk-W5KUC23B.js.map
+//# sourceMappingURL=chunk-STIKJGEA.js.map

package/dist/cli.js

@@ -4,12 +4,15 @@ import {
   markdownToHwpx,
   markdownToXlsx,
   parse
-} from "./chunk-ZOEUKD77.js";
+} from "./chunk-2GFJFTKS.js";
 import {
   VERSION,
   toArrayBuffer
-} from "./chunk-W5KUC23B.js";
+} from "./chunk-STIKJGEA.js";
 import "./chunk-MOL7MDBG.js";
+import "./chunk-7FMKAV4P.js";
+import "./chunk-JOGAFNIL.js";
+import "./chunk-4PP34NVQ.js";
 import "./chunk-ZWE3DS7E.js";
 
 // src/cli.ts
@@ -134,7 +137,7 @@ async function runParse(files, opts) {
         saveImages(absPath);
       }
     } catch (err) {
-      const { sanitizeError } = await import("./utils-HSF5HI5T.js");
+      const { sanitizeError } = await import("./utils-FFUQJTTI.js");
       process.stderr.write(`
 [kordoc] ERROR: ${fileName} \u2014 ${sanitizeError(err)}
 `);
@@ -218,7 +221,7 @@ program.command("convert <input>").description("\uB9C8\uD06C\uB2E4\uC6B4 \uD30C\
 `));
     }
   } catch (err) {
-    const { sanitizeError } = await import("./utils-HSF5HI5T.js");
+    const { sanitizeError } = await import("./utils-FFUQJTTI.js");
     process.stderr.write(` FAIL
 `);
     process.stderr.write(`  \u2192 ${sanitizeError(err)}
@@ -227,7 +230,7 @@ program.command("convert <input>").description("\uB9C8\uD06C\uB2E4\uC6B4 \uD30C\
   }
 });
 program.command("watch <dir>").description("\uB514\uB809\uD1A0\uB9AC \uAC10\uC2DC \u2014 \uC0C8 \uBB38\uC11C \uC790\uB3D9 \uBCC0\uD658").option("--webhook <url>", "\uACB0\uACFC \uC804\uC1A1 \uC6F9\uD6C5 URL").option("-d, --out-dir <dir>", "\uBCC0\uD658 \uACB0\uACFC \uCD9C\uB825 \uB514\uB809\uD1A0\uB9AC").option("-p, --pages <range>", "\uD398\uC774\uC9C0/\uC139\uC158 \uBC94\uC704").option("--format <type>", "\uCD9C\uB825 \uD615\uC2DD: markdown \uB610\uB294 json", "markdown").option("--silent", "\uC9C4\uD589 \uBA54\uC2DC\uC9C0 \uC228\uAE30\uAE30").action(async (dir, opts) => {
-  const { watchDirectory } = await import("./watch-R2JHXDGF.js");
+  const { watchDirectory } = await import("./watch-2O32L6IF.js");
   await watchDirectory({
     dir,
     outDir: opts.outDir,

package/dist/cli.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/cli.ts"],"sourcesContent":["/** kordoc CLI — 모두 파싱해버리겠다 */\n\nimport { readFileSync, writeFileSync, mkdirSync, statSync, existsSync, readdirSync } from \"fs\"\nimport { basename, resolve, extname } from \"path\"\nimport { cpus } from \"os\"\nimport { Command } from \"commander\"\nimport { parse, detectFormat, markdownToHwpx, markdownToXlsx } from \"./index.js\"\nimport type { ParseOptions, OcrMode } from \"./types.js\"\nimport { VERSION, toArrayBuffer } from \"./utils.js\"\n\nconst program = new Command()\n\n/** 공통 parse 옵션 타입 */\ninterface ParseOpts {\n  output?: string\n  outDir?: string\n  pages?: string\n  format: string\n  headerFooter: boolean\n  imageDir?: string\n  silent?: boolean\n  ocr?: string\n  ocrJobs?: string\n  ocrBatchSize?: string\n}\n\n/** parse 액션 공통 구현 — 루트 커맨드와 `parse` 서브커맨드가 공유 */\nasync function runParse(files: string[], opts: ParseOpts) {\n  const validFormats = [\"markdown\", \"json\"]\n  if (!validFormats.includes(opts.format)) {\n    process.stderr.write(`[kordoc] 지원하지 않는 형식: ${opts.format} (markdown 또는 json)\\n`)\n    process.exit(1)\n  }\n  for (let fi = 0; fi < files.length; fi++) {\n    const filePath = files[fi]\n    const absPath = resolve(filePath)\n    const fileName = basename(absPath)\n    const filePrefix = files.length > 1 ? `[${fi + 1}/${files.length}] ` : \"\"\n\n    try {\n      const fileSize = statSync(absPath).size\n      if (fileSize > 500 * 1024 * 1024) {\n        process.stderr.write(`\\n[kordoc] SKIP: ${fileName} — 파일이 너무 큽니다 (${(fileSize / 1024 / 1024).toFixed(1)}MB)\\n`)\n        process.exitCode = 1\n        continue\n      }\n      const buffer = readFileSync(absPath)\n      const arrayBuffer = toArrayBuffer(buffer)\n      const format = detectFormat(arrayBuffer)\n\n      if (!opts.silent) {\n        process.stderr.write(`[kordoc] ${filePrefix}${fileName} (${format}) ...`)\n      }\n\n      const parseOptions: ParseOptions = {}\n      if (opts.pages) parseOptions.pages = opts.pages as string\n      if (opts.headerFooter === false) parseOptions.removeHeaderFooter = false\n\n      // OCR 모드: CLI 기본값 \"auto\" (라이브러리 API는 undefined 유지)\n      const validOcrModes = [\"auto\", \"gemini\", \"claude\", \"codex\", \"ollama\", \"tesseract\", \"off\"]\n      if (opts.ocr) {\n        if (!validOcrModes.includes(opts.ocr)) {\n          process.stderr.write(`[kordoc] 지원하지 않는 OCR 모드: ${opts.ocr}\\n`)\n          process.stderr.write(`  사용 가능: ${validOcrModes.join(\", \")}\\n`)\n          process.exit(1)\n        }\n        parseOptions.ocrMode = opts.ocr as OcrMode\n      } else {\n        parseOptions.ocrMode = \"auto\"\n      }\n\n      // OCR 병렬 처리 수 (--ocr-jobs)\n      // - tesseract: CPU 코어 수 (워커 스레드 병렬)\n      // - gemini/claude/codex/auto: 4 (배치 병렬 실행)\n      if (opts.ocrJobs) {\n        const n = parseInt(opts.ocrJobs, 10)\n        if (n > 0) parseOptions.ocrConcurrency = n\n      } else if (parseOptions.ocrMode === \"tesseract\") {\n        parseOptions.ocrConcurrency = cpus().length\n      } else {\n        parseOptions.ocrConcurrency = 4\n      }\n\n      // OCR 배치 크기 (--ocr-batch-size)\n      if (opts.ocrBatchSize) {\n        const n = parseInt(opts.ocrBatchSize, 10)\n        if (n > 0) parseOptions.ocrBatchSize = n\n      }\n\n      if (!opts.silent) {\n        parseOptions.onProgress = (current: number, total: number) => {\n          process.stderr.write(`\\r[kordoc] ${filePrefix}${fileName} (${format}) [${current}/${total}]`)\n        }\n      }\n      const result = await parse(arrayBuffer, parseOptions)\n\n      if (!result.success) {\n        process.stderr.write(` FAIL\\n`)\n        process.stderr.write(`  → ${result.error}\\n`)\n        process.exitCode = 1\n        continue\n      }\n\n      if (!opts.silent) process.stderr.write(` OK\\n`)\n\n      // 이미지 기반 PDF OCR 결과 표시\n      if (!opts.silent && result.success && result.isImageBased) {\n        process.stderr.write(`  → 이미지 기반 PDF — OCR 처리됨\\n`)\n      }\n\n      // 경고 표시\n      if (!opts.silent && result.success && result.warnings?.length) {\n        for (const w of result.warnings) {\n          process.stderr.write(`  ⚠ ${w.message}\\n`)\n        }\n      }\n\n      let markdown = result.markdown\n      // --out-dir 시 이미지 참조 경로에 images/ 접두사 추가\n      if (opts.outDir && result.images?.length) {\n        markdown = markdown.replace(/!\\[image\\]\\(image_/g, \"![image](images/image_\")\n      }\n      const output = opts.format === \"json\"\n        ? JSON.stringify(result, null, 2)\n        : markdown\n\n      // 이미지 저장: 출력 MD 파일 기준 폴더 사용 (convert와 일치)\n      const saveImages = (outFilePath: string) => {\n        if (!result.images?.length) return\n        const stem = basename(outFilePath).replace(/\\.[^.]+$/, \"\")\n        const defaultDir = resolve(outFilePath, \"..\", stem + \"_images\")\n        const imgDir = opts.imageDir ? resolve(opts.imageDir) : defaultDir\n        mkdirSync(imgDir, { recursive: true })\n        for (const img of result.images) {\n          writeFileSync(resolve(imgDir, img.filename), img.data)\n        }\n        if (!opts.silent) process.stderr.write(`  → ${result.images.length}개 이미지 → ${imgDir}\\n`)\n      }\n\n      if (opts.output && files.length === 1) {\n        writeFileSync(opts.output, output, \"utf-8\")\n        if (!opts.silent) process.stderr.write(`  → ${opts.output}\\n`)\n        saveImages(resolve(opts.output))\n      } else if (opts.outDir) {\n        mkdirSync(opts.outDir, { recursive: true })\n        const outExt = opts.format === \"json\" ? \".json\" : \".md\"\n        const outPath = resolve(opts.outDir, fileName.replace(/\\.[^.]+$/, outExt))\n        writeFileSync(outPath, output, \"utf-8\")\n        if (!opts.silent) process.stderr.write(`  → ${outPath}\\n`)\n        saveImages(outPath)\n      } else {\n        process.stdout.write(output + \"\\n\")\n        saveImages(absPath)  // stdout 출력 시 입력 파일 기준\n      }\n    } catch (err) {\n      const { sanitizeError } = await import(\"./utils.js\")\n      process.stderr.write(`\\n[kordoc] ERROR: ${fileName} — ${sanitizeError(err)}\\n`)\n      process.exitCode = 1\n    }\n  }\n}\n\n/** 공통 parse 옵션 등록 헬퍼 */\nfunction addParseOptions(cmd: Command): Command {\n  return cmd\n    .option(\"-o, --output <path>\", \"출력 파일 경로 (단일 파일 시)\")\n    .option(\"-d, --out-dir <dir>\", \"출력 디렉토리 (다중 파일 시)\")\n    .option(\"-p, --pages <range>\", \"페이지/섹션 범위 (예: 1-3, 1,3,5)\")\n    .option(\"--format <type>\", \"출력 형식: markdown (기본) 또는 json\", \"markdown\")\n    .option(\"--no-header-footer\", \"PDF 머리글/바닥글 자동 제거\")\n    .option(\"--image-dir <dir>\", \"이미지 저장 폴더 (기본: 입력 파일명_images 폴더)\")\n    .option(\"--silent\", \"진행 메시지 숨기기\")\n    .option(\"--ocr <mode>\", \"OCR 모드: auto(기본), gemini, claude, codex, ollama, tesseract, off\")\n    .option(\"--ocr-jobs <n>\", \"OCR 병렬 처리 수 (기본: CPU 코어 수, tesseract 전용)\")\n    .option(\"--ocr-batch-size <n>\", \"OCR 배치 크기 — CLI당 페이지 수 (기본: gemini/claude 50, codex 100)\")\n}\n\nprogram\n  .enablePositionalOptions()\n  .name(\"kordoc\")\n  .description(\"모두 파싱해버리겠다 — HWP, HWPX, PDF, XLSX, DOCX → Markdown\")\n  .version(VERSION)\n\n// `kordoc parse <files>` 서브커맨드 (권장)\naddParseOptions(\n  program\n    .command(\"parse\")\n    .description(\"파일을 마크다운으로 파싱 (HWP, HWPX, PDF, XLSX, DOCX)\")\n    .argument(\"<files...>\", \"변환할 파일 경로\")\n).action(runParse)\n\n// `kordoc <files>` 루트 커맨드 (하위 호환)\naddParseOptions(\n  program\n    .argument(\"<files...>\", \"변환할 파일 경로 (HWP, HWPX, PDF, XLSX, DOCX)\")\n).action(runParse)\n\nprogram\n  .command(\"convert <input>\")\n  .description(\"마크다운 파일을 HWPX 또는 XLSX로 변환\")\n  .option(\"-f, --format <type>\", \"출력 포맷: hwpx | xlsx\", \"hwpx\")\n  .option(\"-o, --output <path>\", \"출력 파일 경로 (기본: 입력명 + 포맷 확장자)\")\n  .option(\"--image-dir <dir>\", \"이미지 폴더 경로 (기본: 입력 MD 파일명_images 폴더)\")\n  .option(\"--images\", \"이미지 포함 (기본: 생략, 레이아웃 문제 방지)\")\n  .option(\"--template <path>\", \"HWPX 템플릿 파일 경로 (hwpx 전용)\")\n  .option(\"--silent\", \"진행 메시지 숨기기\")\n  .action(async (input: string, opts) => {\n    const validFormats = [\"hwpx\", \"xlsx\"]\n    if (!validFormats.includes(opts.format)) {\n      process.stderr.write(`[kordoc] 지원하지 않는 포맷: ${opts.format} (hwpx 또는 xlsx)\\n`)\n      process.exit(1)\n    }\n\n    const absInput = resolve(input)\n    if (!existsSync(absInput)) {\n      process.stderr.write(`[kordoc] 파일을 찾을 수 없습니다: ${input}\\n`)\n      process.exit(1)\n    }\n\n    const stem = basename(absInput).replace(/\\.[^.]+$/, \"\")\n    const outPath = opts.output\n      ? resolve(opts.output)\n      : resolve(absInput, \"..\", `${stem}.${opts.format}`)\n\n    if (!opts.silent) process.stderr.write(`[kordoc] ${basename(absInput)} → ${basename(outPath)} ...`)\n\n    try {\n      const markdown = readFileSync(absInput, \"utf-8\")\n\n      // 이미지 폴더에서 이미지 로드 (--images 플래그 필요)\n      const imgDir = opts.imageDir ? resolve(opts.imageDir) : resolve(absInput, \"..\", stem + \"_images\")\n      const images: import(\"./types.js\").ExtractedImage[] = []\n      if (opts.images && existsSync(imgDir)) {\n        const mimeMap: Record<string, string> = {\n          png: \"image/png\", jpg: \"image/jpeg\", jpeg: \"image/jpeg\",\n          gif: \"image/gif\", bmp: \"image/bmp\",\n        }\n        for (const entry of readdirSync(imgDir, { withFileTypes: true })) {\n          if (!entry.isFile()) continue\n          const fname = entry.name\n          const ext = extname(fname).slice(1).toLowerCase()\n          if (!mimeMap[ext]) continue\n          const data = readFileSync(resolve(imgDir, fname))\n          images.push({ filename: fname, data: new Uint8Array(data), mimeType: mimeMap[ext] })\n        }\n        if (!opts.silent) process.stderr.write(`  → 이미지 ${images.length}개 로드\\n`)\n      }\n\n      const warnings: string[] = []\n\n      let buf: ArrayBuffer\n      if (opts.format === \"xlsx\") {\n        if (opts.template && !opts.silent) {\n          process.stderr.write(`\\n[kordoc] 경고: --template은 hwpx 전용입니다. 무시됩니다.\\n`)\n        }\n        buf = await markdownToXlsx(markdown, { warnings, images: images.length ? images : undefined })\n      } else {\n        let templateArrayBuffer: ArrayBuffer | undefined\n        if (opts.template) {\n          const tmplBuf = readFileSync(resolve(opts.template))\n          templateArrayBuffer = tmplBuf.buffer.slice(tmplBuf.byteOffset, tmplBuf.byteOffset + tmplBuf.byteLength)\n        }\n        buf = await markdownToHwpx(markdown, { warnings, images: images.length ? images : undefined, templateArrayBuffer })\n      }\n\n      writeFileSync(outPath, Buffer.from(buf))\n\n      if (!opts.silent) {\n        process.stderr.write(` OK\\n`)\n        process.stderr.write(`  → ${outPath}\\n`)\n        if (warnings.length) warnings.forEach(w => process.stderr.write(`  ${w}\\n`))\n      }\n    } catch (err) {\n      const { sanitizeError } = await import(\"./utils.js\")\n      process.stderr.write(` FAIL\\n`)\n      process.stderr.write(`  → ${sanitizeError(err)}\\n`)\n      process.exit(1)\n    }\n  })\n\nprogram\n  .command(\"watch <dir>\")\n  .description(\"디렉토리 감시 — 새 문서 자동 변환\")\n  .option(\"--webhook <url>\", \"결과 전송 웹훅 URL\")\n  .option(\"-d, --out-dir <dir>\", \"변환 결과 출력 디렉토리\")\n  .option(\"-p, --pages <range>\", \"페이지/섹션 범위\")\n  .option(\"--format <type>\", \"출력 형식: markdown 또는 json\", \"markdown\")\n  .option(\"--silent\", \"진행 메시지 숨기기\")\n  .action(async (dir: string, opts) => {\n    const { watchDirectory } = await import(\"./watch.js\")\n    await watchDirectory({\n      dir,\n      outDir: opts.outDir,\n      webhook: opts.webhook,\n      format: opts.format,\n      pages: opts.pages,\n      silent: opts.silent,\n    })\n  })\n\nprogram.parse()\n"],"mappings":";;;;;;;;;;;;;;;AAEA,SAAS,cAAc,eAAe,WAAW,UAAU,YAAY,mBAAmB;AAC1F,SAAS,UAAU,SAAS,eAAe;AAC3C,SAAS,YAAY;AACrB,SAAS,eAAe;AAKxB,IAAM,UAAU,IAAI,QAAQ;AAiB5B,eAAe,SAAS,OAAiB,MAAiB;AACxD,QAAM,eAAe,CAAC,YAAY,MAAM;AACxC,MAAI,CAAC,aAAa,SAAS,KAAK,MAAM,GAAG;AACvC,YAAQ,OAAO,MAAM,gEAAwB,KAAK,MAAM;AAAA,CAAuB;AAC/E,YAAQ,KAAK,CAAC;AAAA,EAChB;AACA,WAAS,KAAK,GAAG,KAAK,MAAM,QAAQ,MAAM;AACxC,UAAM,WAAW,MAAM,EAAE;AACzB,UAAM,UAAU,QAAQ,QAAQ;AAChC,UAAM,WAAW,SAAS,OAAO;AACjC,UAAM,aAAa,MAAM,SAAS,IAAI,IAAI,KAAK,CAAC,IAAI,MAAM,MAAM,OAAO;AAEvE,QAAI;AACF,YAAM,WAAW,SAAS,OAAO,EAAE;AACnC,UAAI,WAAW,MAAM,OAAO,MAAM;AAChC,gBAAQ,OAAO,MAAM;AAAA,iBAAoB,QAAQ,gEAAmB,WAAW,OAAO,MAAM,QAAQ,CAAC,CAAC;AAAA,CAAO;AAC7G,gBAAQ,WAAW;AACnB;AAAA,MACF;AACA,YAAM,SAAS,aAAa,OAAO;AACnC,YAAM,cAAc,cAAc,MAAM;AACxC,YAAM,SAAS,aAAa,WAAW;AAEvC,UAAI,CAAC,KAAK,QAAQ;AAChB,gBAAQ,OAAO,MAAM,YAAY,UAAU,GAAG,QAAQ,KAAK,MAAM,OAAO;AAAA,MAC1E;AAEA,YAAM,eAA6B,CAAC;AACpC,UAAI,KAAK,MAAO,cAAa,QAAQ,KAAK;AAC1C,UAAI,KAAK,iBAAiB,MAAO,cAAa,qBAAqB;AAGnE,YAAM,gBAAgB,CAAC,QAAQ,UAAU,UAAU,SAAS,UAAU,aAAa,KAAK;AACxF,UAAI,KAAK,KAAK;AACZ,YAAI,CAAC,cAAc,SAAS,KAAK,GAAG,GAAG;AACrC,kBAAQ,OAAO,MAAM,oEAA4B,KAAK,GAAG;AAAA,CAAI;AAC7D,kBAAQ,OAAO,MAAM,gCAAY,cAAc,KAAK,IAAI,CAAC;AAAA,CAAI;AAC7D,kBAAQ,KAAK,CAAC;AAAA,QAChB;AACA,qBAAa,UAAU,KAAK;AAAA,MAC9B,OAAO;AACL,qBAAa,UAAU;AAAA,MACzB;AAKA,UAAI,KAAK,SAAS;AAChB,cAAM,IAAI,SAAS,KAAK,SAAS,EAAE;AACnC,YAAI,IAAI,EAAG,cAAa,iBAAiB;AAAA,MAC3C,WAAW,aAAa,YAAY,aAAa;AAC/C,qBAAa,iBAAiB,KAAK,EAAE;AAAA,MACvC,OAAO;AACL,qBAAa,iBAAiB;AAAA,MAChC;AAGA,UAAI,KAAK,cAAc;AACrB,cAAM,IAAI,SAAS,KAAK,cAAc,EAAE;AACxC,YAAI,IAAI,EAAG,cAAa,eAAe;AAAA,MACzC;AAEA,UAAI,CAAC,KAAK,QAAQ;AAChB,qBAAa,aAAa,CAAC,SAAiB,UAAkB;AAC5D,kBAAQ,OAAO,MAAM,cAAc,UAAU,GAAG,QAAQ,KAAK,MAAM,MAAM,OAAO,IAAI,KAAK,GAAG;AAAA,QAC9F;AAAA,MACF;AACA,YAAM,SAAS,MAAM,MAAM,aAAa,YAAY;AAEpD,UAAI,CAAC,OAAO,SAAS;AACnB,gBAAQ,OAAO,MAAM;AAAA,CAAS;AAC9B,gBAAQ,OAAO,MAAM,YAAO,OAAO,KAAK;AAAA,CAAI;AAC5C,gBAAQ,WAAW;AACnB;AAAA,MACF;AAEA,UAAI,CAAC,KAAK,OAAQ,SAAQ,OAAO,MAAM;AAAA,CAAO;AAG9C,UAAI,CAAC,KAAK,UAAU,OAAO,WAAW,OAAO,cAAc;AACzD,gBAAQ,OAAO,MAAM;AAAA,CAA4B;AAAA,MACnD;AAGA,UAAI,CAAC,KAAK,UAAU,OAAO,WAAW,OAAO,UAAU,QAAQ;AAC7D,mBAAW,KAAK,OAAO,UAAU;AAC/B,kBAAQ,OAAO,MAAM,YAAO,EAAE,OAAO;AAAA,CAAI;AAAA,QAC3C;AAAA,MACF;AAEA,UAAI,WAAW,OAAO;AAEtB,UAAI,KAAK,UAAU,OAAO,QAAQ,QAAQ;AACxC,mBAAW,SAAS,QAAQ,uBAAuB,wBAAwB;AAAA,MAC7E;AACA,YAAM,SAAS,KAAK,WAAW,SAC3B,KAAK,UAAU,QAAQ,MAAM,CAAC,IAC9B;AAGJ,YAAM,aAAa,CAAC,gBAAwB;AAC1C,YAAI,CAAC,OAAO,QAAQ,OAAQ;AAC5B,cAAM,OAAO,SAAS,WAAW,EAAE,QAAQ,YAAY,EAAE;AACzD,cAAM,aAAa,QAAQ,aAAa,MAAM,OAAO,SAAS;AAC9D,cAAM,SAAS,KAAK,WAAW,QAAQ,KAAK,QAAQ,IAAI;AACxD,kBAAU,QAAQ,EAAE,WAAW,KAAK,CAAC;AACrC,mBAAW,OAAO,OAAO,QAAQ;AAC/B,wBAAc,QAAQ,QAAQ,IAAI,QAAQ,GAAG,IAAI,IAAI;AAAA,QACvD;AACA,YAAI,CAAC,KAAK,OAAQ,SAAQ,OAAO,MAAM,YAAO,OAAO,OAAO,MAAM,oCAAW,MAAM;AAAA,CAAI;AAAA,MACzF;AAEA,UAAI,KAAK,UAAU,MAAM,WAAW,GAAG;AACrC,sBAAc,KAAK,QAAQ,QAAQ,OAAO;AAC1C,YAAI,CAAC,KAAK,OAAQ,SAAQ,OAAO,MAAM,YAAO,KAAK,MAAM;AAAA,CAAI;AAC7D,mBAAW,QAAQ,KAAK,MAAM,CAAC;AAAA,MACjC,WAAW,KAAK,QAAQ;AACtB,kBAAU,KAAK,QAAQ,EAAE,WAAW,KAAK,CAAC;AAC1C,cAAM,SAAS,KAAK,WAAW,SAAS,UAAU;AAClD,cAAM,UAAU,QAAQ,KAAK,QAAQ,SAAS,QAAQ,YAAY,MAAM,CAAC;AACzE,sBAAc,SAAS,QAAQ,OAAO;AACtC,YAAI,CAAC,KAAK,OAAQ,SAAQ,OAAO,MAAM,YAAO,OAAO;AAAA,CAAI;AACzD,mBAAW,OAAO;AAAA,MACpB,OAAO;AACL,gBAAQ,OAAO,MAAM,SAAS,IAAI;AAClC,mBAAW,OAAO;AAAA,MACpB;AAAA,IACF,SAAS,KAAK;AACZ,YAAM,EAAE,cAAc,IAAI,MAAM,OAAO,qBAAY;AACnD,cAAQ,OAAO,MAAM;AAAA,kBAAqB,QAAQ,WAAM,cAAc,GAAG,CAAC;AAAA,CAAI;AAC9E,cAAQ,WAAW;AAAA,IACrB;AAAA,EACF;AACF;AAGA,SAAS,gBAAgB,KAAuB;AAC9C,SAAO,IACJ,OAAO,uBAAuB,2EAAoB,EAClD,OAAO,uBAAuB,0EAAmB,EACjD,OAAO,uBAAuB,mEAA2B,EACzD,OAAO,mBAAmB,wEAAgC,UAAU,EACpE,OAAO,sBAAsB,qEAAmB,EAChD,OAAO,qBAAqB,kHAAkC,EAC9D,OAAO,YAAY,oDAAY,EAC/B,OAAO,gBAAgB,qFAAiE,EACxF,OAAO,kBAAkB,sGAA0C,EACnE,OAAO,wBAAwB,sHAA0D;AAC9F;AAEA,QACG,wBAAwB,EACxB,KAAK,QAAQ,EACb,YAAY,2GAAoD,EAChE,QAAQ,OAAO;AAGlB;AAAA,EACE,QACG,QAAQ,OAAO,EACf,YAAY,mGAA4C,EACxD,SAAS,cAAc,8CAAW;AACvC,EAAE,OAAO,QAAQ;AAGjB;AAAA,EACE,QACG,SAAS,cAAc,2EAAwC;AACpE,EAAE,OAAO,QAAQ;AAEjB,QACG,QAAQ,iBAAiB,EACzB,YAAY,uFAA2B,EACvC,OAAO,uBAAuB,0CAAsB,MAAM,EAC1D,OAAO,uBAAuB,6GAA6B,EAC3D,OAAO,qBAAqB,qHAAqC,EACjE,OAAO,YAAY,kHAA6B,EAChD,OAAO,qBAAqB,uEAA0B,EACtD,OAAO,YAAY,oDAAY,EAC/B,OAAO,OAAO,OAAe,SAAS;AACrC,QAAM,eAAe,CAAC,QAAQ,MAAM;AACpC,MAAI,CAAC,aAAa,SAAS,KAAK,MAAM,GAAG;AACvC,YAAQ,OAAO,MAAM,gEAAwB,KAAK,MAAM;AAAA,CAAmB;AAC3E,YAAQ,KAAK,CAAC;AAAA,EAChB;AAEA,QAAM,WAAW,QAAQ,KAAK;AAC9B,MAAI,CAAC,WAAW,QAAQ,GAAG;AACzB,YAAQ,OAAO,MAAM,6EAA2B,KAAK;AAAA,CAAI;AACzD,YAAQ,KAAK,CAAC;AAAA,EAChB;AAEA,QAAM,OAAO,SAAS,QAAQ,EAAE,QAAQ,YAAY,EAAE;AACtD,QAAM,UAAU,KAAK,SACjB,QAAQ,KAAK,MAAM,IACnB,QAAQ,UAAU,MAAM,GAAG,IAAI,IAAI,KAAK,MAAM,EAAE;AAEpD,MAAI,CAAC,KAAK,OAAQ,SAAQ,OAAO,MAAM,YAAY,SAAS,QAAQ,CAAC,WAAM,SAAS,OAAO,CAAC,MAAM;AAElG,MAAI;AACF,UAAM,WAAW,aAAa,UAAU,OAAO;AAG/C,UAAM,SAAS,KAAK,WAAW,QAAQ,KAAK,QAAQ,IAAI,QAAQ,UAAU,MAAM,OAAO,SAAS;AAChG,UAAM,SAAgD,CAAC;AACvD,QAAI,KAAK,UAAU,WAAW,MAAM,GAAG;AACrC,YAAM,UAAkC;AAAA,QACtC,KAAK;AAAA,QAAa,KAAK;AAAA,QAAc,MAAM;AAAA,QAC3C,KAAK;AAAA,QAAa,KAAK;AAAA,MACzB;AACA,iBAAW,SAAS,YAAY,QAAQ,EAAE,eAAe,KAAK,CAAC,GAAG;AAChE,YAAI,CAAC,MAAM,OAAO,EAAG;AACrB,cAAM,QAAQ,MAAM;AACpB,cAAM,MAAM,QAAQ,KAAK,EAAE,MAAM,CAAC,EAAE,YAAY;AAChD,YAAI,CAAC,QAAQ,GAAG,EAAG;AACnB,cAAM,OAAO,aAAa,QAAQ,QAAQ,KAAK,CAAC;AAChD,eAAO,KAAK,EAAE,UAAU,OAAO,MAAM,IAAI,WAAW,IAAI,GAAG,UAAU,QAAQ,GAAG,EAAE,CAAC;AAAA,MACrF;AACA,UAAI,CAAC,KAAK,OAAQ,SAAQ,OAAO,MAAM,+BAAW,OAAO,MAAM;AAAA,CAAQ;AAAA,IACzE;AAEA,UAAM,WAAqB,CAAC;AAE5B,QAAI;AACJ,QAAI,KAAK,WAAW,QAAQ;AAC1B,UAAI,KAAK,YAAY,CAAC,KAAK,QAAQ;AACjC,gBAAQ,OAAO,MAAM;AAAA;AAAA,CAAiD;AAAA,MACxE;AACA,YAAM,MAAM,eAAe,UAAU,EAAE,UAAU,QAAQ,OAAO,SAAS,SAAS,OAAU,CAAC;AAAA,IAC/F,OAAO;AACL,UAAI;AACJ,UAAI,KAAK,UAAU;AACjB,cAAM,UAAU,aAAa,QAAQ,KAAK,QAAQ,CAAC;AACnD,8BAAsB,QAAQ,OAAO,MAAM,QAAQ,YAAY,QAAQ,aAAa,QAAQ,UAAU;AAAA,MACxG;AACA,YAAM,MAAM,eAAe,UAAU,EAAE,UAAU,QAAQ,OAAO,SAAS,SAAS,QAAW,oBAAoB,CAAC;AAAA,IACpH;AAEA,kBAAc,SAAS,OAAO,KAAK,GAAG,CAAC;AAEvC,QAAI,CAAC,KAAK,QAAQ;AAChB,cAAQ,OAAO,MAAM;AAAA,CAAO;AAC5B,cAAQ,OAAO,MAAM,YAAO,OAAO;AAAA,CAAI;AACvC,UAAI,SAAS,OAAQ,UAAS,QAAQ,OAAK,QAAQ,OAAO,MAAM,KAAK,CAAC;AAAA,CAAI,CAAC;AAAA,IAC7E;AAAA,EACF,SAAS,KAAK;AACZ,UAAM,EAAE,cAAc,IAAI,MAAM,OAAO,qBAAY;AACnD,YAAQ,OAAO,MAAM;AAAA,CAAS;AAC9B,YAAQ,OAAO,MAAM,YAAO,cAAc,GAAG,CAAC;AAAA,CAAI;AAClD,YAAQ,KAAK,CAAC;AAAA,EAChB;AACF,CAAC;AAEH,QACG,QAAQ,aAAa,EACrB,YAAY,4FAAsB,EAClC,OAAO,mBAAmB,4CAAc,EACxC,OAAO,uBAAuB,iEAAe,EAC7C,OAAO,uBAAuB,8CAAW,EACzC,OAAO,mBAAmB,yDAA2B,UAAU,EAC/D,OAAO,YAAY,oDAAY,EAC/B,OAAO,OAAO,KAAa,SAAS;AACnC,QAAM,EAAE,eAAe,IAAI,MAAM,OAAO,qBAAY;AACpD,QAAM,eAAe;AAAA,IACnB;AAAA,IACA,QAAQ,KAAK;AAAA,IACb,SAAS,KAAK;AAAA,IACd,QAAQ,KAAK;AAAA,IACb,OAAO,KAAK;AAAA,IACZ,QAAQ,KAAK;AAAA,EACf,CAAC;AACH,CAAC;AAEH,QAAQ,MAAM;","names":[]}
+{"version":3,"sources":["../src/cli.ts"],"sourcesContent":["/** kordoc CLI — 모두 파싱해버리겠다 */\n\nimport { readFileSync, writeFileSync, mkdirSync, statSync, existsSync, readdirSync } from \"fs\"\nimport { basename, resolve, extname } from \"path\"\nimport { cpus } from \"os\"\nimport { Command } from \"commander\"\nimport { parse, detectFormat, markdownToHwpx, markdownToXlsx } from \"./index.js\"\nimport type { ParseOptions, OcrMode } from \"./types.js\"\nimport { VERSION, toArrayBuffer } from \"./utils.js\"\n\nconst program = new Command()\n\n/** 공통 parse 옵션 타입 */\ninterface ParseOpts {\n  output?: string\n  outDir?: string\n  pages?: string\n  format: string\n  headerFooter: boolean\n  imageDir?: string\n  silent?: boolean\n  ocr?: string\n  ocrJobs?: string\n  ocrBatchSize?: string\n}\n\n/** parse 액션 공통 구현 — 루트 커맨드와 `parse` 서브커맨드가 공유 */\nasync function runParse(files: string[], opts: ParseOpts) {\n  const validFormats = [\"markdown\", \"json\"]\n  if (!validFormats.includes(opts.format)) {\n    process.stderr.write(`[kordoc] 지원하지 않는 형식: ${opts.format} (markdown 또는 json)\\n`)\n    process.exit(1)\n  }\n  for (let fi = 0; fi < files.length; fi++) {\n    const filePath = files[fi]\n    const absPath = resolve(filePath)\n    const fileName = basename(absPath)\n    const filePrefix = files.length > 1 ? `[${fi + 1}/${files.length}] ` : \"\"\n\n    try {\n      const fileSize = statSync(absPath).size\n      if (fileSize > 500 * 1024 * 1024) {\n        process.stderr.write(`\\n[kordoc] SKIP: ${fileName} — 파일이 너무 큽니다 (${(fileSize / 1024 / 1024).toFixed(1)}MB)\\n`)\n        process.exitCode = 1\n        continue\n      }\n      const buffer = readFileSync(absPath)\n      const arrayBuffer = toArrayBuffer(buffer)\n      const format = detectFormat(arrayBuffer)\n\n      if (!opts.silent) {\n        process.stderr.write(`[kordoc] ${filePrefix}${fileName} (${format}) ...`)\n      }\n\n      const parseOptions: ParseOptions = {}\n      if (opts.pages) parseOptions.pages = opts.pages as string\n      if (opts.headerFooter === false) parseOptions.removeHeaderFooter = false\n\n      // OCR 모드: CLI 기본값 \"auto\" (라이브러리 API는 undefined 유지)\n      const validOcrModes = [\"auto\", \"gemini\", \"claude\", \"codex\", \"ollama\", \"tesseract\", \"off\"]\n      if (opts.ocr) {\n        if (!validOcrModes.includes(opts.ocr)) {\n          process.stderr.write(`[kordoc] 지원하지 않는 OCR 모드: ${opts.ocr}\\n`)\n          process.stderr.write(`  사용 가능: ${validOcrModes.join(\", \")}\\n`)\n          process.exit(1)\n        }\n        parseOptions.ocrMode = opts.ocr as OcrMode\n      } else {\n        parseOptions.ocrMode = \"auto\"\n      }\n\n      // OCR 병렬 처리 수 (--ocr-jobs)\n      // - tesseract: CPU 코어 수 (워커 스레드 병렬)\n      // - gemini/claude/codex/auto: 4 (배치 병렬 실행)\n      if (opts.ocrJobs) {\n        const n = parseInt(opts.ocrJobs, 10)\n        if (n > 0) parseOptions.ocrConcurrency = n\n      } else if (parseOptions.ocrMode === \"tesseract\") {\n        parseOptions.ocrConcurrency = cpus().length\n      } else {\n        parseOptions.ocrConcurrency = 4\n      }\n\n      // OCR 배치 크기 (--ocr-batch-size)\n      if (opts.ocrBatchSize) {\n        const n = parseInt(opts.ocrBatchSize, 10)\n        if (n > 0) parseOptions.ocrBatchSize = n\n      }\n\n      if (!opts.silent) {\n        parseOptions.onProgress = (current: number, total: number) => {\n          process.stderr.write(`\\r[kordoc] ${filePrefix}${fileName} (${format}) [${current}/${total}]`)\n        }\n      }\n      const result = await parse(arrayBuffer, parseOptions)\n\n      if (!result.success) {\n        process.stderr.write(` FAIL\\n`)\n        process.stderr.write(`  → ${result.error}\\n`)\n        process.exitCode = 1\n        continue\n      }\n\n      if (!opts.silent) process.stderr.write(` OK\\n`)\n\n      // 이미지 기반 PDF OCR 결과 표시\n      if (!opts.silent && result.success && result.isImageBased) {\n        process.stderr.write(`  → 이미지 기반 PDF — OCR 처리됨\\n`)\n      }\n\n      // 경고 표시\n      if (!opts.silent && result.success && result.warnings?.length) {\n        for (const w of result.warnings) {\n          process.stderr.write(`  ⚠ ${w.message}\\n`)\n        }\n      }\n\n      let markdown = result.markdown\n      // --out-dir 시 이미지 참조 경로에 images/ 접두사 추가\n      if (opts.outDir && result.images?.length) {\n        markdown = markdown.replace(/!\\[image\\]\\(image_/g, \"![image](images/image_\")\n      }\n      const output = opts.format === \"json\"\n        ? JSON.stringify(result, null, 2)\n        : markdown\n\n      // 이미지 저장: 출력 MD 파일 기준 폴더 사용 (convert와 일치)\n      const saveImages = (outFilePath: string) => {\n        if (!result.images?.length) return\n        const stem = basename(outFilePath).replace(/\\.[^.]+$/, \"\")\n        const defaultDir = resolve(outFilePath, \"..\", stem + \"_images\")\n        const imgDir = opts.imageDir ? resolve(opts.imageDir) : defaultDir\n        mkdirSync(imgDir, { recursive: true })\n        for (const img of result.images) {\n          writeFileSync(resolve(imgDir, img.filename), img.data)\n        }\n        if (!opts.silent) process.stderr.write(`  → ${result.images.length}개 이미지 → ${imgDir}\\n`)\n      }\n\n      if (opts.output && files.length === 1) {\n        writeFileSync(opts.output, output, \"utf-8\")\n        if (!opts.silent) process.stderr.write(`  → ${opts.output}\\n`)\n        saveImages(resolve(opts.output))\n      } else if (opts.outDir) {\n        mkdirSync(opts.outDir, { recursive: true })\n        const outExt = opts.format === \"json\" ? \".json\" : \".md\"\n        const outPath = resolve(opts.outDir, fileName.replace(/\\.[^.]+$/, outExt))\n        writeFileSync(outPath, output, \"utf-8\")\n        if (!opts.silent) process.stderr.write(`  → ${outPath}\\n`)\n        saveImages(outPath)\n      } else {\n        process.stdout.write(output + \"\\n\")\n        saveImages(absPath)  // stdout 출력 시 입력 파일 기준\n      }\n    } catch (err) {\n      const { sanitizeError } = await import(\"./utils.js\")\n      process.stderr.write(`\\n[kordoc] ERROR: ${fileName} — ${sanitizeError(err)}\\n`)\n      process.exitCode = 1\n    }\n  }\n}\n\n/** 공통 parse 옵션 등록 헬퍼 */\nfunction addParseOptions(cmd: Command): Command {\n  return cmd\n    .option(\"-o, --output <path>\", \"출력 파일 경로 (단일 파일 시)\")\n    .option(\"-d, --out-dir <dir>\", \"출력 디렉토리 (다중 파일 시)\")\n    .option(\"-p, --pages <range>\", \"페이지/섹션 범위 (예: 1-3, 1,3,5)\")\n    .option(\"--format <type>\", \"출력 형식: markdown (기본) 또는 json\", \"markdown\")\n    .option(\"--no-header-footer\", \"PDF 머리글/바닥글 자동 제거\")\n    .option(\"--image-dir <dir>\", \"이미지 저장 폴더 (기본: 입력 파일명_images 폴더)\")\n    .option(\"--silent\", \"진행 메시지 숨기기\")\n    .option(\"--ocr <mode>\", \"OCR 모드: auto(기본), gemini, claude, codex, ollama, tesseract, off\")\n    .option(\"--ocr-jobs <n>\", \"OCR 병렬 처리 수 (기본: CPU 코어 수, tesseract 전용)\")\n    .option(\"--ocr-batch-size <n>\", \"OCR 배치 크기 — CLI당 페이지 수 (기본: gemini/claude 50, codex 100)\")\n}\n\nprogram\n  .enablePositionalOptions()\n  .name(\"kordoc\")\n  .description(\"모두 파싱해버리겠다 — HWP, HWPX, PDF, XLSX, DOCX → Markdown\")\n  .version(VERSION)\n\n// `kordoc parse <files>` 서브커맨드 (권장)\naddParseOptions(\n  program\n    .command(\"parse\")\n    .description(\"파일을 마크다운으로 파싱 (HWP, HWPX, PDF, XLSX, DOCX)\")\n    .argument(\"<files...>\", \"변환할 파일 경로\")\n).action(runParse)\n\n// `kordoc <files>` 루트 커맨드 (하위 호환)\naddParseOptions(\n  program\n    .argument(\"<files...>\", \"변환할 파일 경로 (HWP, HWPX, PDF, XLSX, DOCX)\")\n).action(runParse)\n\nprogram\n  .command(\"convert <input>\")\n  .description(\"마크다운 파일을 HWPX 또는 XLSX로 변환\")\n  .option(\"-f, --format <type>\", \"출력 포맷: hwpx | xlsx\", \"hwpx\")\n  .option(\"-o, --output <path>\", \"출력 파일 경로 (기본: 입력명 + 포맷 확장자)\")\n  .option(\"--image-dir <dir>\", \"이미지 폴더 경로 (기본: 입력 MD 파일명_images 폴더)\")\n  .option(\"--images\", \"이미지 포함 (기본: 생략, 레이아웃 문제 방지)\")\n  .option(\"--template <path>\", \"HWPX 템플릿 파일 경로 (hwpx 전용)\")\n  .option(\"--silent\", \"진행 메시지 숨기기\")\n  .action(async (input: string, opts) => {\n    const validFormats = [\"hwpx\", \"xlsx\"]\n    if (!validFormats.includes(opts.format)) {\n      process.stderr.write(`[kordoc] 지원하지 않는 포맷: ${opts.format} (hwpx 또는 xlsx)\\n`)\n      process.exit(1)\n    }\n\n    const absInput = resolve(input)\n    if (!existsSync(absInput)) {\n      process.stderr.write(`[kordoc] 파일을 찾을 수 없습니다: ${input}\\n`)\n      process.exit(1)\n    }\n\n    const stem = basename(absInput).replace(/\\.[^.]+$/, \"\")\n    const outPath = opts.output\n      ? resolve(opts.output)\n      : resolve(absInput, \"..\", `${stem}.${opts.format}`)\n\n    if (!opts.silent) process.stderr.write(`[kordoc] ${basename(absInput)} → ${basename(outPath)} ...`)\n\n    try {\n      const markdown = readFileSync(absInput, \"utf-8\")\n\n      // 이미지 폴더에서 이미지 로드 (--images 플래그 필요)\n      const imgDir = opts.imageDir ? resolve(opts.imageDir) : resolve(absInput, \"..\", stem + \"_images\")\n      const images: import(\"./types.js\").ExtractedImage[] = []\n      if (opts.images && existsSync(imgDir)) {\n        const mimeMap: Record<string, string> = {\n          png: \"image/png\", jpg: \"image/jpeg\", jpeg: \"image/jpeg\",\n          gif: \"image/gif\", bmp: \"image/bmp\",\n        }\n        for (const entry of readdirSync(imgDir, { withFileTypes: true })) {\n          if (!entry.isFile()) continue\n          const fname = entry.name\n          const ext = extname(fname).slice(1).toLowerCase()\n          if (!mimeMap[ext]) continue\n          const data = readFileSync(resolve(imgDir, fname))\n          images.push({ filename: fname, data: new Uint8Array(data), mimeType: mimeMap[ext] })\n        }\n        if (!opts.silent) process.stderr.write(`  → 이미지 ${images.length}개 로드\\n`)\n      }\n\n      const warnings: string[] = []\n\n      let buf: ArrayBuffer\n      if (opts.format === \"xlsx\") {\n        if (opts.template && !opts.silent) {\n          process.stderr.write(`\\n[kordoc] 경고: --template은 hwpx 전용입니다. 무시됩니다.\\n`)\n        }\n        buf = await markdownToXlsx(markdown, { warnings, images: images.length ? images : undefined })\n      } else {\n        let templateArrayBuffer: ArrayBuffer | undefined\n        if (opts.template) {\n          const tmplBuf = readFileSync(resolve(opts.template))\n          templateArrayBuffer = tmplBuf.buffer.slice(tmplBuf.byteOffset, tmplBuf.byteOffset + tmplBuf.byteLength)\n        }\n        buf = await markdownToHwpx(markdown, { warnings, images: images.length ? images : undefined, templateArrayBuffer })\n      }\n\n      writeFileSync(outPath, Buffer.from(buf))\n\n      if (!opts.silent) {\n        process.stderr.write(` OK\\n`)\n        process.stderr.write(`  → ${outPath}\\n`)\n        if (warnings.length) warnings.forEach(w => process.stderr.write(`  ${w}\\n`))\n      }\n    } catch (err) {\n      const { sanitizeError } = await import(\"./utils.js\")\n      process.stderr.write(` FAIL\\n`)\n      process.stderr.write(`  → ${sanitizeError(err)}\\n`)\n      process.exit(1)\n    }\n  })\n\nprogram\n  .command(\"watch <dir>\")\n  .description(\"디렉토리 감시 — 새 문서 자동 변환\")\n  .option(\"--webhook <url>\", \"결과 전송 웹훅 URL\")\n  .option(\"-d, --out-dir <dir>\", \"변환 결과 출력 디렉토리\")\n  .option(\"-p, --pages <range>\", \"페이지/섹션 범위\")\n  .option(\"--format <type>\", \"출력 형식: markdown 또는 json\", \"markdown\")\n  .option(\"--silent\", \"진행 메시지 숨기기\")\n  .action(async (dir: string, opts) => {\n    const { watchDirectory } = await import(\"./watch.js\")\n    await watchDirectory({\n      dir,\n      outDir: opts.outDir,\n      webhook: opts.webhook,\n      format: opts.format,\n      pages: opts.pages,\n      silent: opts.silent,\n    })\n  })\n\nprogram.parse()\n"],"mappings":";;;;;;;;;;;;;;;;;;AAEA,SAAS,cAAc,eAAe,WAAW,UAAU,YAAY,mBAAmB;AAC1F,SAAS,UAAU,SAAS,eAAe;AAC3C,SAAS,YAAY;AACrB,SAAS,eAAe;AAKxB,IAAM,UAAU,IAAI,QAAQ;AAiB5B,eAAe,SAAS,OAAiB,MAAiB;AACxD,QAAM,eAAe,CAAC,YAAY,MAAM;AACxC,MAAI,CAAC,aAAa,SAAS,KAAK,MAAM,GAAG;AACvC,YAAQ,OAAO,MAAM,gEAAwB,KAAK,MAAM;AAAA,CAAuB;AAC/E,YAAQ,KAAK,CAAC;AAAA,EAChB;AACA,WAAS,KAAK,GAAG,KAAK,MAAM,QAAQ,MAAM;AACxC,UAAM,WAAW,MAAM,EAAE;AACzB,UAAM,UAAU,QAAQ,QAAQ;AAChC,UAAM,WAAW,SAAS,OAAO;AACjC,UAAM,aAAa,MAAM,SAAS,IAAI,IAAI,KAAK,CAAC,IAAI,MAAM,MAAM,OAAO;AAEvE,QAAI;AACF,YAAM,WAAW,SAAS,OAAO,EAAE;AACnC,UAAI,WAAW,MAAM,OAAO,MAAM;AAChC,gBAAQ,OAAO,MAAM;AAAA,iBAAoB,QAAQ,gEAAmB,WAAW,OAAO,MAAM,QAAQ,CAAC,CAAC;AAAA,CAAO;AAC7G,gBAAQ,WAAW;AACnB;AAAA,MACF;AACA,YAAM,SAAS,aAAa,OAAO;AACnC,YAAM,cAAc,cAAc,MAAM;AACxC,YAAM,SAAS,aAAa,WAAW;AAEvC,UAAI,CAAC,KAAK,QAAQ;AAChB,gBAAQ,OAAO,MAAM,YAAY,UAAU,GAAG,QAAQ,KAAK,MAAM,OAAO;AAAA,MAC1E;AAEA,YAAM,eAA6B,CAAC;AACpC,UAAI,KAAK,MAAO,cAAa,QAAQ,KAAK;AAC1C,UAAI,KAAK,iBAAiB,MAAO,cAAa,qBAAqB;AAGnE,YAAM,gBAAgB,CAAC,QAAQ,UAAU,UAAU,SAAS,UAAU,aAAa,KAAK;AACxF,UAAI,KAAK,KAAK;AACZ,YAAI,CAAC,cAAc,SAAS,KAAK,GAAG,GAAG;AACrC,kBAAQ,OAAO,MAAM,oEAA4B,KAAK,GAAG;AAAA,CAAI;AAC7D,kBAAQ,OAAO,MAAM,gCAAY,cAAc,KAAK,IAAI,CAAC;AAAA,CAAI;AAC7D,kBAAQ,KAAK,CAAC;AAAA,QAChB;AACA,qBAAa,UAAU,KAAK;AAAA,MAC9B,OAAO;AACL,qBAAa,UAAU;AAAA,MACzB;AAKA,UAAI,KAAK,SAAS;AAChB,cAAM,IAAI,SAAS,KAAK,SAAS,EAAE;AACnC,YAAI,IAAI,EAAG,cAAa,iBAAiB;AAAA,MAC3C,WAAW,aAAa,YAAY,aAAa;AAC/C,qBAAa,iBAAiB,KAAK,EAAE;AAAA,MACvC,OAAO;AACL,qBAAa,iBAAiB;AAAA,MAChC;AAGA,UAAI,KAAK,cAAc;AACrB,cAAM,IAAI,SAAS,KAAK,cAAc,EAAE;AACxC,YAAI,IAAI,EAAG,cAAa,eAAe;AAAA,MACzC;AAEA,UAAI,CAAC,KAAK,QAAQ;AAChB,qBAAa,aAAa,CAAC,SAAiB,UAAkB;AAC5D,kBAAQ,OAAO,MAAM,cAAc,UAAU,GAAG,QAAQ,KAAK,MAAM,MAAM,OAAO,IAAI,KAAK,GAAG;AAAA,QAC9F;AAAA,MACF;AACA,YAAM,SAAS,MAAM,MAAM,aAAa,YAAY;AAEpD,UAAI,CAAC,OAAO,SAAS;AACnB,gBAAQ,OAAO,MAAM;AAAA,CAAS;AAC9B,gBAAQ,OAAO,MAAM,YAAO,OAAO,KAAK;AAAA,CAAI;AAC5C,gBAAQ,WAAW;AACnB;AAAA,MACF;AAEA,UAAI,CAAC,KAAK,OAAQ,SAAQ,OAAO,MAAM;AAAA,CAAO;AAG9C,UAAI,CAAC,KAAK,UAAU,OAAO,WAAW,OAAO,cAAc;AACzD,gBAAQ,OAAO,MAAM;AAAA,CAA4B;AAAA,MACnD;AAGA,UAAI,CAAC,KAAK,UAAU,OAAO,WAAW,OAAO,UAAU,QAAQ;AAC7D,mBAAW,KAAK,OAAO,UAAU;AAC/B,kBAAQ,OAAO,MAAM,YAAO,EAAE,OAAO;AAAA,CAAI;AAAA,QAC3C;AAAA,MACF;AAEA,UAAI,WAAW,OAAO;AAEtB,UAAI,KAAK,UAAU,OAAO,QAAQ,QAAQ;AACxC,mBAAW,SAAS,QAAQ,uBAAuB,wBAAwB;AAAA,MAC7E;AACA,YAAM,SAAS,KAAK,WAAW,SAC3B,KAAK,UAAU,QAAQ,MAAM,CAAC,IAC9B;AAGJ,YAAM,aAAa,CAAC,gBAAwB;AAC1C,YAAI,CAAC,OAAO,QAAQ,OAAQ;AAC5B,cAAM,OAAO,SAAS,WAAW,EAAE,QAAQ,YAAY,EAAE;AACzD,cAAM,aAAa,QAAQ,aAAa,MAAM,OAAO,SAAS;AAC9D,cAAM,SAAS,KAAK,WAAW,QAAQ,KAAK,QAAQ,IAAI;AACxD,kBAAU,QAAQ,EAAE,WAAW,KAAK,CAAC;AACrC,mBAAW,OAAO,OAAO,QAAQ;AAC/B,wBAAc,QAAQ,QAAQ,IAAI,QAAQ,GAAG,IAAI,IAAI;AAAA,QACvD;AACA,YAAI,CAAC,KAAK,OAAQ,SAAQ,OAAO,MAAM,YAAO,OAAO,OAAO,MAAM,oCAAW,MAAM;AAAA,CAAI;AAAA,MACzF;AAEA,UAAI,KAAK,UAAU,MAAM,WAAW,GAAG;AACrC,sBAAc,KAAK,QAAQ,QAAQ,OAAO;AAC1C,YAAI,CAAC,KAAK,OAAQ,SAAQ,OAAO,MAAM,YAAO,KAAK,MAAM;AAAA,CAAI;AAC7D,mBAAW,QAAQ,KAAK,MAAM,CAAC;AAAA,MACjC,WAAW,KAAK,QAAQ;AACtB,kBAAU,KAAK,QAAQ,EAAE,WAAW,KAAK,CAAC;AAC1C,cAAM,SAAS,KAAK,WAAW,SAAS,UAAU;AAClD,cAAM,UAAU,QAAQ,KAAK,QAAQ,SAAS,QAAQ,YAAY,MAAM,CAAC;AACzE,sBAAc,SAAS,QAAQ,OAAO;AACtC,YAAI,CAAC,KAAK,OAAQ,SAAQ,OAAO,MAAM,YAAO,OAAO;AAAA,CAAI;AACzD,mBAAW,OAAO;AAAA,MACpB,OAAO;AACL,gBAAQ,OAAO,MAAM,SAAS,IAAI;AAClC,mBAAW,OAAO;AAAA,MACpB;AAAA,IACF,SAAS,KAAK;AACZ,YAAM,EAAE,cAAc,IAAI,MAAM,OAAO,qBAAY;AACnD,cAAQ,OAAO,MAAM;AAAA,kBAAqB,QAAQ,WAAM,cAAc,GAAG,CAAC;AAAA,CAAI;AAC9E,cAAQ,WAAW;AAAA,IACrB;AAAA,EACF;AACF;AAGA,SAAS,gBAAgB,KAAuB;AAC9C,SAAO,IACJ,OAAO,uBAAuB,2EAAoB,EAClD,OAAO,uBAAuB,0EAAmB,EACjD,OAAO,uBAAuB,mEAA2B,EACzD,OAAO,mBAAmB,wEAAgC,UAAU,EACpE,OAAO,sBAAsB,qEAAmB,EAChD,OAAO,qBAAqB,kHAAkC,EAC9D,OAAO,YAAY,oDAAY,EAC/B,OAAO,gBAAgB,qFAAiE,EACxF,OAAO,kBAAkB,sGAA0C,EACnE,OAAO,wBAAwB,sHAA0D;AAC9F;AAEA,QACG,wBAAwB,EACxB,KAAK,QAAQ,EACb,YAAY,2GAAoD,EAChE,QAAQ,OAAO;AAGlB;AAAA,EACE,QACG,QAAQ,OAAO,EACf,YAAY,mGAA4C,EACxD,SAAS,cAAc,8CAAW;AACvC,EAAE,OAAO,QAAQ;AAGjB;AAAA,EACE,QACG,SAAS,cAAc,2EAAwC;AACpE,EAAE,OAAO,QAAQ;AAEjB,QACG,QAAQ,iBAAiB,EACzB,YAAY,uFAA2B,EACvC,OAAO,uBAAuB,0CAAsB,MAAM,EAC1D,OAAO,uBAAuB,6GAA6B,EAC3D,OAAO,qBAAqB,qHAAqC,EACjE,OAAO,YAAY,kHAA6B,EAChD,OAAO,qBAAqB,uEAA0B,EACtD,OAAO,YAAY,oDAAY,EAC/B,OAAO,OAAO,OAAe,SAAS;AACrC,QAAM,eAAe,CAAC,QAAQ,MAAM;AACpC,MAAI,CAAC,aAAa,SAAS,KAAK,MAAM,GAAG;AACvC,YAAQ,OAAO,MAAM,gEAAwB,KAAK,MAAM;AAAA,CAAmB;AAC3E,YAAQ,KAAK,CAAC;AAAA,EAChB;AAEA,QAAM,WAAW,QAAQ,KAAK;AAC9B,MAAI,CAAC,WAAW,QAAQ,GAAG;AACzB,YAAQ,OAAO,MAAM,6EAA2B,KAAK;AAAA,CAAI;AACzD,YAAQ,KAAK,CAAC;AAAA,EAChB;AAEA,QAAM,OAAO,SAAS,QAAQ,EAAE,QAAQ,YAAY,EAAE;AACtD,QAAM,UAAU,KAAK,SACjB,QAAQ,KAAK,MAAM,IACnB,QAAQ,UAAU,MAAM,GAAG,IAAI,IAAI,KAAK,MAAM,EAAE;AAEpD,MAAI,CAAC,KAAK,OAAQ,SAAQ,OAAO,MAAM,YAAY,SAAS,QAAQ,CAAC,WAAM,SAAS,OAAO,CAAC,MAAM;AAElG,MAAI;AACF,UAAM,WAAW,aAAa,UAAU,OAAO;AAG/C,UAAM,SAAS,KAAK,WAAW,QAAQ,KAAK,QAAQ,IAAI,QAAQ,UAAU,MAAM,OAAO,SAAS;AAChG,UAAM,SAAgD,CAAC;AACvD,QAAI,KAAK,UAAU,WAAW,MAAM,GAAG;AACrC,YAAM,UAAkC;AAAA,QACtC,KAAK;AAAA,QAAa,KAAK;AAAA,QAAc,MAAM;AAAA,QAC3C,KAAK;AAAA,QAAa,KAAK;AAAA,MACzB;AACA,iBAAW,SAAS,YAAY,QAAQ,EAAE,eAAe,KAAK,CAAC,GAAG;AAChE,YAAI,CAAC,MAAM,OAAO,EAAG;AACrB,cAAM,QAAQ,MAAM;AACpB,cAAM,MAAM,QAAQ,KAAK,EAAE,MAAM,CAAC,EAAE,YAAY;AAChD,YAAI,CAAC,QAAQ,GAAG,EAAG;AACnB,cAAM,OAAO,aAAa,QAAQ,QAAQ,KAAK,CAAC;AAChD,eAAO,KAAK,EAAE,UAAU,OAAO,MAAM,IAAI,WAAW,IAAI,GAAG,UAAU,QAAQ,GAAG,EAAE,CAAC;AAAA,MACrF;AACA,UAAI,CAAC,KAAK,OAAQ,SAAQ,OAAO,MAAM,+BAAW,OAAO,MAAM;AAAA,CAAQ;AAAA,IACzE;AAEA,UAAM,WAAqB,CAAC;AAE5B,QAAI;AACJ,QAAI,KAAK,WAAW,QAAQ;AAC1B,UAAI,KAAK,YAAY,CAAC,KAAK,QAAQ;AACjC,gBAAQ,OAAO,MAAM;AAAA;AAAA,CAAiD;AAAA,MACxE;AACA,YAAM,MAAM,eAAe,UAAU,EAAE,UAAU,QAAQ,OAAO,SAAS,SAAS,OAAU,CAAC;AAAA,IAC/F,OAAO;AACL,UAAI;AACJ,UAAI,KAAK,UAAU;AACjB,cAAM,UAAU,aAAa,QAAQ,KAAK,QAAQ,CAAC;AACnD,8BAAsB,QAAQ,OAAO,MAAM,QAAQ,YAAY,QAAQ,aAAa,QAAQ,UAAU;AAAA,MACxG;AACA,YAAM,MAAM,eAAe,UAAU,EAAE,UAAU,QAAQ,OAAO,SAAS,SAAS,QAAW,oBAAoB,CAAC;AAAA,IACpH;AAEA,kBAAc,SAAS,OAAO,KAAK,GAAG,CAAC;AAEvC,QAAI,CAAC,KAAK,QAAQ;AAChB,cAAQ,OAAO,MAAM;AAAA,CAAO;AAC5B,cAAQ,OAAO,MAAM,YAAO,OAAO;AAAA,CAAI;AACvC,UAAI,SAAS,OAAQ,UAAS,QAAQ,OAAK,QAAQ,OAAO,MAAM,KAAK,CAAC;AAAA,CAAI,CAAC;AAAA,IAC7E;AAAA,EACF,SAAS,KAAK;AACZ,UAAM,EAAE,cAAc,IAAI,MAAM,OAAO,qBAAY;AACnD,YAAQ,OAAO,MAAM;AAAA,CAAS;AAC9B,YAAQ,OAAO,MAAM,YAAO,cAAc,GAAG,CAAC;AAAA,CAAI;AAClD,YAAQ,KAAK,CAAC;AAAA,EAChB;AACF,CAAC;AAEH,QACG,QAAQ,aAAa,EACrB,YAAY,4FAAsB,EAClC,OAAO,mBAAmB,4CAAc,EACxC,OAAO,uBAAuB,iEAAe,EAC7C,OAAO,uBAAuB,8CAAW,EACzC,OAAO,mBAAmB,yDAA2B,UAAU,EAC/D,OAAO,YAAY,oDAAY,EAC/B,OAAO,OAAO,KAAa,SAAS;AACnC,QAAM,EAAE,eAAe,IAAI,MAAM,OAAO,qBAAY;AACpD,QAAM,eAAe;AAAA,IACnB;AAAA,IACA,QAAQ,KAAK;AAAA,IACb,SAAS,KAAK;AAAA,IACd,QAAQ,KAAK;AAAA,IACb,OAAO,KAAK;AAAA,IACZ,QAAQ,KAAK;AAAA,EACf,CAAC;AACH,CAAC;AAEH,QAAQ,MAAM;","names":[]}