@clazic/kordoc 2.2.9 → 2.3.1

package/dist/cli.js

@@ -4,11 +4,11 @@ import {
   markdownToHwpx,
   markdownToXlsx,
   parse
-} from "./chunk-OL2NDK3E.js";
+} from "./chunk-ZOEUKD77.js";
 import {
   VERSION,
   toArrayBuffer
-} from "./chunk-FF5M4SDK.js";
+} from "./chunk-W5KUC23B.js";
 import "./chunk-MOL7MDBG.js";
 import "./chunk-ZWE3DS7E.js";
 
@@ -66,6 +66,12 @@ async function runParse(files, opts) {
         if (n > 0) parseOptions.ocrConcurrency = n;
       } else if (parseOptions.ocrMode === "tesseract") {
         parseOptions.ocrConcurrency = cpus().length;
+      } else {
+        parseOptions.ocrConcurrency = 4;
+      }
+      if (opts.ocrBatchSize) {
+        const n = parseInt(opts.ocrBatchSize, 10);
+        if (n > 0) parseOptions.ocrBatchSize = n;
       }
       if (!opts.silent) {
         parseOptions.onProgress = (current, total) => {
@@ -128,7 +134,7 @@ async function runParse(files, opts) {
         saveImages(absPath);
       }
     } catch (err) {
-      const { sanitizeError } = await import("./utils-CU26KLDC.js");
+      const { sanitizeError } = await import("./utils-HSF5HI5T.js");
       process.stderr.write(`
 [kordoc] ERROR: ${fileName} \u2014 ${sanitizeError(err)}
 `);
@@ -137,7 +143,7 @@ async function runParse(files, opts) {
   }
 }
 function addParseOptions(cmd) {
-  return cmd.option("-o, --output <path>", "\uCD9C\uB825 \uD30C\uC77C \uACBD\uB85C (\uB2E8\uC77C \uD30C\uC77C \uC2DC)").option("-d, --out-dir <dir>", "\uCD9C\uB825 \uB514\uB809\uD1A0\uB9AC (\uB2E4\uC911 \uD30C\uC77C \uC2DC)").option("-p, --pages <range>", "\uD398\uC774\uC9C0/\uC139\uC158 \uBC94\uC704 (\uC608: 1-3, 1,3,5)").option("--format <type>", "\uCD9C\uB825 \uD615\uC2DD: markdown (\uAE30\uBCF8) \uB610\uB294 json", "markdown").option("--no-header-footer", "PDF \uBA38\uB9AC\uAE00/\uBC14\uB2E5\uAE00 \uC790\uB3D9 \uC81C\uAC70").option("--image-dir <dir>", "\uC774\uBBF8\uC9C0 \uC800\uC7A5 \uD3F4\uB354 (\uAE30\uBCF8: \uC785\uB825 \uD30C\uC77C\uBA85_images \uD3F4\uB354)").option("--silent", "\uC9C4\uD589 \uBA54\uC2DC\uC9C0 \uC228\uAE30\uAE30").option("--ocr <mode>", "OCR \uBAA8\uB4DC: auto(\uAE30\uBCF8), gemini, claude, codex, ollama, tesseract, off").option("--ocr-jobs <n>", "OCR \uBCD1\uB82C \uCC98\uB9AC \uC218 (\uAE30\uBCF8: CPU \uCF54\uC5B4 \uC218, tesseract \uC804\uC6A9)");
+  return cmd.option("-o, --output <path>", "\uCD9C\uB825 \uD30C\uC77C \uACBD\uB85C (\uB2E8\uC77C \uD30C\uC77C \uC2DC)").option("-d, --out-dir <dir>", "\uCD9C\uB825 \uB514\uB809\uD1A0\uB9AC (\uB2E4\uC911 \uD30C\uC77C \uC2DC)").option("-p, --pages <range>", "\uD398\uC774\uC9C0/\uC139\uC158 \uBC94\uC704 (\uC608: 1-3, 1,3,5)").option("--format <type>", "\uCD9C\uB825 \uD615\uC2DD: markdown (\uAE30\uBCF8) \uB610\uB294 json", "markdown").option("--no-header-footer", "PDF \uBA38\uB9AC\uAE00/\uBC14\uB2E5\uAE00 \uC790\uB3D9 \uC81C\uAC70").option("--image-dir <dir>", "\uC774\uBBF8\uC9C0 \uC800\uC7A5 \uD3F4\uB354 (\uAE30\uBCF8: \uC785\uB825 \uD30C\uC77C\uBA85_images \uD3F4\uB354)").option("--silent", "\uC9C4\uD589 \uBA54\uC2DC\uC9C0 \uC228\uAE30\uAE30").option("--ocr <mode>", "OCR \uBAA8\uB4DC: auto(\uAE30\uBCF8), gemini, claude, codex, ollama, tesseract, off").option("--ocr-jobs <n>", "OCR \uBCD1\uB82C \uCC98\uB9AC \uC218 (\uAE30\uBCF8: CPU \uCF54\uC5B4 \uC218, tesseract \uC804\uC6A9)").option("--ocr-batch-size <n>", "OCR \uBC30\uCE58 \uD06C\uAE30 \u2014 CLI\uB2F9 \uD398\uC774\uC9C0 \uC218 (\uAE30\uBCF8: gemini/claude 50, codex 100)");
 }
 program.enablePositionalOptions().name("kordoc").description("\uBAA8\uB450 \uD30C\uC2F1\uD574\uBC84\uB9AC\uACA0\uB2E4 \u2014 HWP, HWPX, PDF, XLSX, DOCX \u2192 Markdown").version(VERSION);
 addParseOptions(
@@ -212,7 +218,7 @@ program.command("convert <input>").description("\uB9C8\uD06C\uB2E4\uC6B4 \uD30C\
 `));
     }
   } catch (err) {
-    const { sanitizeError } = await import("./utils-CU26KLDC.js");
+    const { sanitizeError } = await import("./utils-HSF5HI5T.js");
     process.stderr.write(` FAIL
 `);
     process.stderr.write(`  \u2192 ${sanitizeError(err)}
@@ -221,7 +227,7 @@ 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-Z6SH4KRB.js");
+  const { watchDirectory } = await import("./watch-R2JHXDGF.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}\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): tesseract 기본값은 CPU 코어 수\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      }\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}\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;AAgB5B,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;AAGA,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;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;AACxE;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":[]}

package/dist/index.cjs

@@ -2011,7 +2011,7 @@ var init_auto_detect = __esm({
 // src/ocr/cli-provider.ts
 function getTempDir() {
   if (!_tempDir) {
-    _tempDir = (0, import_path.join)(process.cwd(), ".kordoc-tmp");
+    _tempDir = (0, import_path.join)(process.cwd(), "_kordoc_ocr_tmp");
     (0, import_fs.mkdirSync)(_tempDir, { recursive: true });
   }
   return _tempDir;
@@ -2111,8 +2111,8 @@ function buildCliArgs(mode, imagePath) {
   }
 }
 async function callOllamaApi(imagePath) {
-  const { readFileSync: readFileSync2 } = await import("fs");
-  const imageBase64 = readFileSync2(imagePath).toString("base64");
+  const { readFileSync: readFileSync3 } = await import("fs");
+  const imageBase64 = readFileSync3(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;
@@ -2213,12 +2213,185 @@ var init_tesseract_provider = __esm({
   }
 });
 
+// src/ocr/batch-provider.ts
+var batch_provider_exports = {};
+__export(batch_provider_exports, {
+  DEFAULT_BATCH_SIZES: () => DEFAULT_BATCH_SIZES,
+  createBatchCliProvider: () => createBatchCliProvider
+});
+function getBatchTempDir() {
+  if (!_batchTempDir) {
+    _batchTempDir = (0, import_path2.join)(process.cwd(), "_kordoc_ocr_tmp");
+    (0, import_fs2.mkdirSync)(_batchTempDir, { recursive: true });
+  }
+  return _batchTempDir;
+}
+function createBatchCliProvider(mode, batchSize) {
+  return {
+    __batch: true,
+    batchSize,
+    async processBatch(pages) {
+      const results = /* @__PURE__ */ new Map();
+      const tempDir = getBatchTempDir();
+      const tempFiles = [];
+      try {
+        for (const { image, pageNum } of pages) {
+          const path = (0, import_path2.join)(tempDir, `batch-p${pageNum}.png`);
+          (0, import_fs2.writeFileSync)(path, image);
+          tempFiles.push(path);
+        }
+        let output;
+        if (mode === "codex") {
+          output = await callBatchCodexCli(tempFiles);
+        } else {
+          output = await callBatchCli(mode, tempFiles);
+        }
+        const cleaned = stripCodeFence2(output.trim());
+        const parts = cleaned.split(/<!--\s*PAGE_BREAK\s*-->/).map((p) => p.trim()).filter((p) => p.length > 0);
+        for (let i = 0; i < pages.length; i++) {
+          const pageNum = pages[i].pageNum;
+          if (i < parts.length) {
+            results.set(pageNum, { markdown: parts[i] });
+          }
+        }
+      } finally {
+        for (const f of tempFiles) {
+          try {
+            (0, import_fs2.unlinkSync)(f);
+          } catch {
+          }
+        }
+      }
+      return results;
+    }
+  };
+}
+function spawnAsync(cmd, args, opts) {
+  return new Promise((resolve, reject) => {
+    const child = (0, import_child_process3.spawn)(cmd, args, {
+      cwd: opts.cwd,
+      env: process.env,
+      stdio: ["pipe", "pipe", "pipe"]
+    });
+    let stdout = "";
+    let stderr = "";
+    let killed = false;
+    child.stdout.setEncoding("utf-8");
+    child.stderr.setEncoding("utf-8");
+    child.stdout.on("data", (d) => {
+      stdout += d;
+    });
+    child.stderr.on("data", (d) => {
+      stderr += d;
+    });
+    const timer = setTimeout(() => {
+      killed = true;
+      child.kill("SIGTERM");
+    }, opts.timeoutMs);
+    if (opts.stdin !== void 0) {
+      child.stdin.end(opts.stdin);
+    } else {
+      child.stdin.end();
+    }
+    child.on("close", (code) => {
+      clearTimeout(timer);
+      if (killed) {
+        reject(new Error(`\uD0C0\uC784\uC544\uC6C3 (${Math.round(opts.timeoutMs / 1e3)}\uCD08)`));
+      } else {
+        resolve({ stdout, stderr, exitCode: code ?? 1 });
+      }
+    });
+    child.on("error", (err) => {
+      clearTimeout(timer);
+      reject(err);
+    });
+  });
+}
+async function callBatchCli(mode, imagePaths) {
+  const fileRefs = imagePaths.map((p) => `@${p}`).join("\n");
+  const prompt = `${BATCH_OCR_PROMPT}
+
+${fileRefs}`;
+  let args;
+  if (mode === "gemini") {
+    args = ["--prompt", prompt, "--yolo"];
+    const model = process.env.KORDOC_GEMINI_MODEL;
+    if (model) args.push("--model", model);
+  } else {
+    args = ["--print", prompt];
+    const model = process.env.KORDOC_CLAUDE_MODEL;
+    if (model) args.push("--model", model);
+  }
+  const timeoutMs = 6e4 + imagePaths.length * 2e4;
+  const result = await spawnAsync(mode, args, {
+    timeoutMs,
+    ...mode === "claude" ? { cwd: (0, import_os2.tmpdir)() } : {}
+  });
+  if (result.exitCode !== 0) {
+    const errMsg = result.stderr?.trim() || `exit code ${result.exitCode}`;
+    throw new Error(`${mode} \uBC30\uCE58 OCR \uC2E4\uD328: ${errMsg}`);
+  }
+  return result.stdout || "";
+}
+async function callBatchCodexCli(imagePaths) {
+  const outPath = (0, import_path2.join)((0, import_os2.tmpdir)(), `kordoc-codex-batch-${Date.now()}-${Math.random().toString(36).slice(2)}.txt`);
+  try {
+    const args = ["exec", BATCH_OCR_PROMPT];
+    for (const p of imagePaths) {
+      args.push("--image", p);
+    }
+    args.push("--output-last-message", outPath);
+    const model = process.env.KORDOC_CODEX_MODEL;
+    if (model) args.push("--model", model);
+    const timeoutMs = 6e4 + imagePaths.length * 2e4;
+    const result = await spawnAsync("codex", args, {
+      timeoutMs,
+      stdin: ""
+    });
+    if (result.exitCode !== 0) {
+      const errMsg = result.stderr?.trim() || `exit code ${result.exitCode}`;
+      throw new Error(`codex \uBC30\uCE58 OCR \uC2E4\uD328: ${errMsg}`);
+    }
+    try {
+      return (0, import_fs2.readFileSync)(outPath, "utf-8");
+    } catch {
+      return result.stdout || "";
+    }
+  } finally {
+    try {
+      (0, import_fs2.unlinkSync)(outPath);
+    } catch {
+    }
+  }
+}
+function stripCodeFence2(text) {
+  const match = text.match(/^```(?:markdown|md)?\s*\n([\s\S]*?)\n```\s*$/m);
+  return match ? match[1].trim() : text;
+}
+var import_child_process3, import_fs2, import_path2, import_os2, BATCH_OCR_PROMPT, DEFAULT_BATCH_SIZES, _batchTempDir;
+var init_batch_provider = __esm({
+  "src/ocr/batch-provider.ts"() {
+    "use strict";
+    import_child_process3 = require("child_process");
+    import_fs2 = require("fs");
+    import_path2 = require("path");
+    import_os2 = require("os");
+    BATCH_OCR_PROMPT = "\uB2E4\uC74C \uBB38\uC11C \uD398\uC774\uC9C0 \uC774\uBBF8\uC9C0\uB4E4\uC744 OCR\uD558\uC5EC \uC21C\uC218 Markdown\uC73C\uB85C \uBCC0\uD658\uD558\uC138\uC694.\n\n\uADDC\uCE59:\n- \uAC01 \uD398\uC774\uC9C0 \uACB0\uACFC \uC0AC\uC774\uC5D0 \uBC18\uB4DC\uC2DC \uC774 \uAD6C\uBD84\uC790\uB97C \uC0BD\uC785: <!-- PAGE_BREAK -->\n- \uD14C\uC774\uBE14\uC740 Markdown \uD14C\uC774\uBE14 \uBB38\uBC95 \uC0AC\uC6A9 (| \uAD6C\uBD84, |---|---| \uD5E4\uB354 \uAD6C\uBD84\uC120 \uD3EC\uD568)\n- \uBCD1\uD569\uB41C \uC140\uC740 \uD574\uB2F9 \uC704\uCE58\uC5D0 \uB0B4\uC6A9 \uAE30\uC7AC\n- \uD5E4\uB529\uC740 \uAE00\uC790 \uD06C\uAE30\uC5D0 \uB530\uB77C ## ~ ###### \uC0AC\uC6A9\n- \uB9AC\uC2A4\uD2B8\uB294 - \uB610\uB294 1. \uC0AC\uC6A9\n- \uC774\uBBF8\uC9C0, \uB3C4\uD615 \uB4F1 \uBE44\uD14D\uC2A4\uD2B8 \uC694\uC18C\uB294 \uBB34\uC2DC\n- \uC6D0\uBB38\uC758 \uC77D\uAE30 \uC21C\uC11C\uC640 \uAD6C\uC870\uB97C \uC720\uC9C0\n- ```\uB85C \uAC10\uC2F8\uC9C0 \uB9D0\uACE0 \uC21C\uC218 Markdown\uB9CC \uCD9C\uB825";
+    DEFAULT_BATCH_SIZES = {
+      gemini: 5,
+      claude: 5,
+      codex: 10
+    };
+    _batchTempDir = null;
+  }
+});
+
 // src/ocr/resolve.ts
 var resolve_exports = {};
 __export(resolve_exports, {
   resolveOcrProvider: () => resolveOcrProvider
 });
-async function resolveOcrProvider(mode, warnings, concurrency) {
+async function resolveOcrProvider(mode, warnings, concurrency, batchSize) {
   if (mode === "off") {
     throw new Error("OCR\uC774 \uBE44\uD65C\uC131\uD654\uB418\uC5B4 \uC788\uC2B5\uB2C8\uB2E4 (--ocr off).");
   }
@@ -2231,6 +2404,14 @@ async function resolveOcrProvider(mode, warnings, concurrency) {
       }
       return createTesseractProvider2();
     }
+    if (mode === "gemini" || mode === "claude" || mode === "codex") {
+      const { createBatchCliProvider: createBatchCliProvider2, DEFAULT_BATCH_SIZES: DEFAULT_BATCH_SIZES2 } = await Promise.resolve().then(() => (init_batch_provider(), batch_provider_exports));
+      const effectiveBatch = batchSize ?? DEFAULT_BATCH_SIZES2[mode];
+      if (effectiveBatch > 1) {
+        return createBatchCliProvider2(mode, effectiveBatch);
+      }
+      return createCliOcrProvider(mode);
+    }
     return createCliOcrProvider(mode);
   }
   const detected = detectAvailableOcr();
@@ -2254,6 +2435,14 @@ async function resolveOcrProvider(mode, warnings, concurrency) {
     }
     return createTesseractProvider2();
   }
+  if (detected === "gemini" || detected === "codex" || detected === "claude") {
+    const { createBatchCliProvider: createBatchCliProvider2, DEFAULT_BATCH_SIZES: DEFAULT_BATCH_SIZES2 } = await Promise.resolve().then(() => (init_batch_provider(), batch_provider_exports));
+    const effectiveBatch = batchSize ?? DEFAULT_BATCH_SIZES2[detected];
+    if (effectiveBatch > 1) {
+      return createBatchCliProvider2(detected, effectiveBatch);
+    }
+    return createCliOcrProvider(detected);
+  }
   return createCliOcrProvider(detected);
 }
 var init_resolve = __esm({
@@ -2416,8 +2605,14 @@ function ocrResultToBlocks(result, pageNum) {
   }
   return pageBlocks;
 }
-async function ocrPages(doc, provider, pageFilter, effectivePageCount, warnings, concurrency = 1) {
+function isBatchProvider(p) {
+  return !!p && typeof p === "object" && "__batch" in p && p.__batch === true;
+}
+async function ocrPages(doc, provider, pageFilter, effectivePageCount, warnings, concurrency = 1, onProgress) {
   const blocks = [];
+  if (isBatchProvider(provider)) {
+    return ocrPagesBatch(doc, provider, pageFilter, effectivePageCount, warnings, concurrency, onProgress);
+  }
   if (concurrency <= 1) {
     for (let i = 1; i <= effectivePageCount; i++) {
       if (pageFilter && !pageFilter.has(i)) continue;
@@ -2463,6 +2658,58 @@ async function ocrPages(doc, provider, pageFilter, effectivePageCount, warnings,
   }
   return blocks;
 }
+async function ocrPagesBatch(doc, provider, pageFilter, effectivePageCount, warnings, concurrency = 1, onProgress) {
+  const pageNumbers = [];
+  for (let i = 1; i <= effectivePageCount; i++) {
+    if (pageFilter && !pageFilter.has(i)) continue;
+    pageNumbers.push(i);
+  }
+  const pageImages = [];
+  for (const pageNum of pageNumbers) {
+    const page = await doc.getPage(pageNum);
+    const image = await renderPageToPng(page);
+    pageImages.push({ image, pageNum });
+  }
+  const batches = [];
+  for (let i = 0; i < pageImages.length; i += provider.batchSize) {
+    batches.push(pageImages.slice(i, i + provider.batchSize));
+  }
+  let processed = 0;
+  const batchTasks = batches.map((batch, batchIdx) => async () => {
+    const pageBlocks = [];
+    try {
+      const results = await provider.processBatch(batch);
+      for (const { pageNum } of batch) {
+        const result = results.get(pageNum);
+        pageBlocks.push({
+          pageNum,
+          blocks: result ? ocrResultToBlocks(result, pageNum) : []
+        });
+      }
+    } catch (err) {
+      const range = `${batch[0].pageNum}-${batch[batch.length - 1].pageNum}`;
+      warnings?.push({
+        message: `\uBC30\uCE58 OCR \uC2E4\uD328 (\uD398\uC774\uC9C0 ${range}): ${err instanceof Error ? err.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958"}`,
+        code: "OCR_PAGE_FAILED"
+      });
+      for (const { pageNum } of batch) {
+        pageBlocks.push({ pageNum, blocks: [] });
+      }
+    }
+    processed += batch.length;
+    onProgress?.(processed, pageNumbers.length);
+    return { batchIdx, pageBlocks };
+  });
+  const effectiveConcurrency = Math.max(1, concurrency);
+  const batchResults = await runWithConcurrency(batchTasks, effectiveConcurrency);
+  const blocks = [];
+  for (const result of batchResults) {
+    for (const { blocks: pageBlks } of result.pageBlocks) {
+      for (const b of pageBlks) blocks.push(b);
+    }
+  }
+  return blocks;
+}
 async function renderPageToPng(page) {
   const { createCanvas } = await import("@napi-rs/canvas");
   const scale = 2;
@@ -2551,7 +2798,7 @@ var import_jszip2 = __toESM(require("jszip"), 1);
 var import_xmldom = require("@xmldom/xmldom");
 
 // src/utils.ts
-var VERSION = true ? "2.2.9" : "0.0.0-dev";
+var VERSION = true ? "2.3.1" : "0.0.0-dev";
 function toArrayBuffer(buf) {
   if (buf.byteOffset === 0 && buf.byteLength === buf.buffer.byteLength) {
     return buf.buffer;
@@ -6069,7 +6316,8 @@ async function parsePdfDocument(buffer, options) {
         try {
           const { resolveOcrProvider: resolveOcrProvider2 } = await Promise.resolve().then(() => (init_resolve(), resolve_exports));
           const concurrency = options?.ocrConcurrency ?? 1;
-          ocrProvider = await resolveOcrProvider2(ocrMode, warnings, concurrency);
+          const batchSize = options?.ocrBatchSize;
+          ocrProvider = await resolveOcrProvider2(ocrMode, warnings, concurrency, batchSize);
         } catch (resolveErr) {
           if (ocrMode !== "auto") {
             throw Object.assign(
@@ -6084,7 +6332,7 @@ async function parsePdfDocument(buffer, options) {
         try {
           const { ocrPages: ocrPages2 } = await Promise.resolve().then(() => (init_provider(), provider_exports));
           const concurrency = options?.ocrConcurrency ?? 1;
-          ocrBlocks = await ocrPages2(doc, ocrProvider, pageFilter, effectivePageCount, warnings, concurrency);
+          ocrBlocks = await ocrPages2(doc, ocrProvider, pageFilter, effectivePageCount, warnings, concurrency, options?.onProgress);
         } catch {
         } finally {
           const terminable = ocrProvider;