@clazic/kordoc 2.4.9 → 2.4.11

package/dist/mcp.js

@@ -10,17 +10,17 @@ import {
   markdownToHwpx,
   markdownToXlsx,
   parse
-} from "./chunk-MPMKWVV2.js";
+} from "./chunk-JGMLDBW5.js";
 import "./chunk-YW5G6BCJ.js";
 import {
   KordocError,
   VERSION,
   sanitizeError,
   toArrayBuffer
-} from "./chunk-SHM3PYVA.js";
+} from "./chunk-PJSXZBZB.js";
 import "./chunk-MOL7MDBG.js";
 import "./chunk-7FMKAV4P.js";
-import "./chunk-YC2MEB7R.js";
+import "./chunk-34WIGIQC.js";
 import "./chunk-ZWE3DS7E.js";
 
 // src/mcp.ts

package/dist/{resolve-BGOGWG6E.js → resolve-4I65IGMM.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import {
   createCliOcrProvider
-} from "./chunk-YC2MEB7R.js";
+} from "./chunk-34WIGIQC.js";
 import {
   detectAvailableOcr,
   getTesseractFallbackMessage,
@@ -24,7 +24,7 @@ async function resolveOcrProvider(mode, warnings, concurrency, batchSize) {
       return createTesseractProvider();
     }
     if (mode === "gemini" || mode === "claude" || mode === "codex") {
-      const { createBatchCliProvider, DEFAULT_BATCH_SIZES } = await import("./batch-provider-VX7CY6UH.js");
+      const { createBatchCliProvider, DEFAULT_BATCH_SIZES } = await import("./batch-provider-5BFJRKAZ.js");
       const effectiveBatch = batchSize ?? DEFAULT_BATCH_SIZES[mode];
       if (effectiveBatch > 1) {
         return createBatchCliProvider(mode, effectiveBatch);
@@ -55,7 +55,7 @@ async function resolveOcrProvider(mode, warnings, concurrency, batchSize) {
     return createTesseractProvider();
   }
   if (detected === "gemini" || detected === "codex" || detected === "claude") {
-    const { createBatchCliProvider, DEFAULT_BATCH_SIZES } = await import("./batch-provider-VX7CY6UH.js");
+    const { createBatchCliProvider, DEFAULT_BATCH_SIZES } = await import("./batch-provider-5BFJRKAZ.js");
     const effectiveBatch = batchSize ?? DEFAULT_BATCH_SIZES[detected];
     if (effectiveBatch > 1) {
       return createBatchCliProvider(detected, effectiveBatch);
@@ -67,4 +67,4 @@ async function resolveOcrProvider(mode, warnings, concurrency, batchSize) {
 export {
   resolveOcrProvider
 };
-//# sourceMappingURL=resolve-BGOGWG6E.js.map
+//# sourceMappingURL=resolve-4I65IGMM.js.map

package/dist/{utils-IUWGWQWV.js → utils-HKVOS2O3.js}

@@ -8,7 +8,7 @@ import {
   sanitizeError,
   sanitizeHref,
   toArrayBuffer
-} from "./chunk-SHM3PYVA.js";
+} from "./chunk-PJSXZBZB.js";
 import "./chunk-ZWE3DS7E.js";
 export {
   KordocError,
@@ -20,4 +20,4 @@ export {
   sanitizeHref,
   toArrayBuffer
 };
-//# sourceMappingURL=utils-IUWGWQWV.js.map
+//# sourceMappingURL=utils-HKVOS2O3.js.map

package/dist/{watch-CRENPZU5.js → watch-EYOGF3HY.js}

@@ -2,14 +2,14 @@
 import {
   detectFormat,
   parse
-} from "./chunk-MPMKWVV2.js";
+} from "./chunk-JGMLDBW5.js";
 import "./chunk-YW5G6BCJ.js";
 import {
   toArrayBuffer
-} from "./chunk-SHM3PYVA.js";
+} from "./chunk-PJSXZBZB.js";
 import "./chunk-MOL7MDBG.js";
 import "./chunk-7FMKAV4P.js";
-import "./chunk-YC2MEB7R.js";
+import "./chunk-34WIGIQC.js";
 import "./chunk-ZWE3DS7E.js";
 
 // src/watch.ts
@@ -136,4 +136,4 @@ async function sendWebhook(url, payload) {
 export {
   watchDirectory
 };
-//# sourceMappingURL=watch-CRENPZU5.js.map
+//# sourceMappingURL=watch-EYOGF3HY.js.map

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@clazic/kordoc",
-  "version": "2.4.9",
+  "version": "2.4.11",
   "description": "Parse Korean documents (HWP, HWPX, PDF, XLSX, DOCX) to Markdown",
   "type": "module",
   "exports": {

package/dist/chunk-YC2MEB7R.js.map

@@ -1 +0,0 @@
-{"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-${pageNumber}.png`)\n\n    try {\n      writeFileSync(tempPath, pageImage)\n\n      let output: string\n      if (mode === \"ollama\") {\n        output = await callOllamaApi(tempPath)\n      } else {\n        output = callCli(mode, tempPath)\n      }\n\n      return { markdown: stripCodeFence(output.trim()) }\n    } finally {\n      try { unlinkSync(tempPath) } catch { /* 임시 파일 정리 실패 무시 */ }\n    }\n  }\n}\n\n/**\n * CLI 실행 — gemini / claude / codex\n *\n * @throws CLI 실행 실패 또는 타임아웃(180초) 시 Error\n */\nfunction callCli(mode: string, imagePath: string): string {\n  // codex는 --output-last-message로 대화 헤더 없는 깔끔한 출력 사용\n  if (mode === \"codex\") {\n    return callCodexCli(imagePath)\n  }\n\n  const args = buildCliArgs(mode, imagePath)\n\n  const result = spawnSync(mode, args, {\n    encoding: \"utf-8\",\n    timeout: 600_000,\n    maxBuffer: 10 * 1024 * 1024,\n    shell: process.platform === \"win32\",\n    // claude: /tmp에서 실행하여 프로젝트 CLAUDE.md의 규칙 간섭 방지\n    ...(mode === \"claude\" ? { cwd: tmpdir() } : {}),\n  })\n\n  if (result.error) {\n    throw new Error(`${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      shell: process.platform === \"win32\",\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: ${errMsg}`)\n    }\n\n    // --output-last-message 파일에서 읽기 (없으면 stdout 폴백)\n    try {\n      return readFileSync(outPath, \"utf-8\")\n    } catch {\n      return result.stdout || \"\"\n    }\n  } finally {\n    try { unlinkSync(outPath) } catch { /* 무시 */ }\n  }\n}\n\n/**\n * CLI별 인자 배열 생성.\n *\n * gemini:  [\"--prompt\", \"프롬프트 @이미지경로\", \"--yolo\"]\n *   - -y/--yolo: 자동 승인 (OCR은 도구 사용 없으므로 실질적 영향 없음)\n *   - @ 파일 참조로 이미지를 컨텍스트에 포함\n *\n * claude:  [\"--print\", \"프롬프트 @이미지경로\"]\n *   - --print(-p): 비대화형 출력 모드\n *   - @ 파일 참조로 이미지를 컨텍스트에 포함\n *\n * codex: callCodexCli()에서 별도 처리\n *   - `codex exec <prompt> --image <file> --output-last-message <outfile>`\n *   - 프롬프트가 --image보다 앞에 위치해야 함 (인자 순서 중요)\n *\n * ⚠️ CLI 버전에 따라 문법이 다를 수 있음. 업데이트 시 --help 재확인 필요.\n */\nfunction buildCliArgs(mode: string, imagePath: string): string[] {\n  const normalizedPath = imagePath.replace(/\\\\/g, \"/\")\n  const promptWithImage = `${OCR_PROMPT}\n\n이미지: @${normalizedPath}`\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,IACvB,OAAO,QAAQ,aAAa;AAAA;AAAA,IAE5B,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,MACP,OAAO,QAAQ,aAAa;AAAA,IAC9B,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,iBAAiB,UAAU,QAAQ,OAAO,GAAG;AACnD,QAAM,kBAAkB,GAAG,UAAU;AAAA;AAAA,uBAE/B,cAAc;AAEpB,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"]}