@clazic/kordoc 2.1.6 → 2.2.0

package/dist/resolve-Y3KMGD3R.js

@@ -0,0 +1,187 @@
+#!/usr/bin/env node
+import "./chunk-ZWE3DS7E.js";
+
+// src/ocr/auto-detect.ts
+import { execSync } from "child_process";
+import { createRequire } from "module";
+var CLI_PRIORITY = ["gemini", "claude", "codex", "ollama"];
+function detectAvailableOcr() {
+  for (const cli of CLI_PRIORITY) {
+    if (isCliInstalled(cli)) return cli;
+  }
+  if (isTesseractAvailable()) return "tesseract";
+  return null;
+}
+function isCliInstalled(name) {
+  try {
+    const cmd = process.platform === "win32" ? "where" : "which";
+    execSync(`${cmd} ${name}`, { stdio: "ignore", timeout: 3e3 });
+    return true;
+  } catch {
+    return false;
+  }
+}
+function isTesseractAvailable() {
+  try {
+    const require2 = createRequire(import.meta.url);
+    require2.resolve("tesseract.js");
+    return true;
+  } catch {
+    return false;
+  }
+}
+function validateOcrMode(mode) {
+  if (mode === "auto" || mode === "off") return;
+  if (mode === "tesseract") {
+    if (!isTesseractAvailable()) {
+      throw new Error(
+        "tesseract.js\uAC00 \uC124\uCE58\uB418\uC9C0 \uC54A\uC558\uC2B5\uB2C8\uB2E4.\n\uC124\uCE58: npm install tesseract.js"
+      );
+    }
+    return;
+  }
+  if (!isCliInstalled(mode)) {
+    throw new Error(`'${mode}' CLI\uAC00 \uC124\uCE58\uB418\uC9C0 \uC54A\uC558\uC2B5\uB2C8\uB2E4.
+${getInstallGuide(mode)}`);
+  }
+}
+function getInstallGuide(mode) {
+  const guides = {
+    gemini: "\uC124\uCE58: https://ai.google.dev/gemini-api/docs/cli",
+    claude: "\uC124\uCE58: npm install -g @anthropic-ai/claude-code \uB610\uB294 https://claude.ai/code",
+    codex: "\uC124\uCE58: npm install -g @openai/codex \uB610\uB294 https://github.com/openai/codex",
+    ollama: "\uC124\uCE58: brew install ollama \uB610\uB294 https://ollama.com/download"
+  };
+  return guides[mode] || `'${mode}'\uC744(\uB97C) \uC124\uCE58\uD574\uC8FC\uC138\uC694.`;
+}
+
+// src/ocr/cli-provider.ts
+import { spawnSync } from "child_process";
+import { writeFileSync, unlinkSync, mkdtempSync } 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.\n\uADDC\uCE59:\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";
+var _tempDir = null;
+function getTempDir() {
+  if (!_tempDir) _tempDir = mkdtempSync(join(tmpdir(), "kordoc-ocr-"));
+  return _tempDir;
+}
+function createCliOcrProvider(mode) {
+  return async (pageImage, pageNumber) => {
+    const tempPath = join(getTempDir(), `page-${pageNumber}.png`);
+    try {
+      writeFileSync(tempPath, pageImage);
+      let output;
+      if (mode === "ollama") {
+        output = await callOllamaApi(tempPath);
+      } else {
+        output = callCli(mode, tempPath);
+      }
+      return { markdown: stripCodeFence(output.trim()) };
+    } finally {
+      try {
+        unlinkSync(tempPath);
+      } catch {
+      }
+    }
+  };
+}
+function callCli(mode, imagePath) {
+  const args = buildCliArgs(mode, imagePath);
+  const result = spawnSync(mode === "codex" ? "codex" : mode, args, {
+    encoding: "utf-8",
+    timeout: 6e4,
+    maxBuffer: 10 * 1024 * 1024
+  });
+  if (result.error) {
+    throw new Error(`${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: ${errMsg}`);
+  }
+  return result.stdout || "";
+}
+function buildCliArgs(mode, imagePath) {
+  const promptWithImage = `${OCR_PROMPT}
+
+\uC774\uBBF8\uC9C0: @${imagePath}`;
+  const promptOnly = OCR_PROMPT;
+  switch (mode) {
+    case "gemini":
+      return ["--prompt", promptWithImage, "--yolo"];
+    case "claude":
+      return ["--print", promptWithImage];
+    case "codex":
+      return ["exec", "--image", imagePath, promptOnly];
+    default:
+      throw new Error(`\uC9C0\uC6D0\uD558\uC9C0 \uC54A\uB294 CLI: ${mode}`);
+  }
+}
+async function callOllamaApi(imagePath) {
+  const { readFileSync } = await import("fs");
+  const imageBase64 = readFileSync(imagePath).toString("base64");
+  const model = process.env.KORDOC_OLLAMA_MODEL || "gemma4:27b";
+  const host = process.env.KORDOC_OLLAMA_HOST || "http://localhost:11434";
+  const response = await fetch(`${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(6e4)
+  });
+  if (!response.ok) {
+    throw new Error(`Ollama API \uC624\uB958: ${response.status} ${response.statusText}`);
+  }
+  const data = await response.json();
+  return data.message?.content || "";
+}
+function stripCodeFence(text) {
+  const match = text.match(/^```(?:markdown|md)?\s*\n([\s\S]*?)\n```\s*$/m);
+  return match ? match[1].trim() : text;
+}
+
+// src/ocr/resolve.ts
+async function resolveOcrProvider(mode, warnings) {
+  if (mode === "off") {
+    throw new Error("OCR\uC774 \uBE44\uD65C\uC131\uD654\uB418\uC5B4 \uC788\uC2B5\uB2C8\uB2E4 (--ocr off).");
+  }
+  if (mode !== "auto") {
+    validateOcrMode(mode);
+    if (mode === "tesseract") {
+      const { createTesseractProvider } = await import("./tesseract-provider-MZ37ZKQW.js");
+      return createTesseractProvider();
+    }
+    return createCliOcrProvider(mode);
+  }
+  const detected = detectAvailableOcr();
+  if (!detected) {
+    throw new Error("\uC0AC\uC6A9 \uAC00\uB2A5\uD55C OCR \uB3C4\uAD6C\uB97C \uCC3E\uC744 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4.");
+  }
+  if (detected !== "gemini") {
+    warnings?.push({
+      message: `OCR: '${detected}' \uC0AC\uC6A9 \uC911 (gemini CLI\uAC00 \uC5C6\uC5B4 fallback). \uB354 \uB098\uC740 \uD488\uC9C8\uC744 \uC704\uD574 gemini CLI \uC124\uCE58\uB97C \uAD8C\uC7A5\uD569\uB2C8\uB2E4.`,
+      code: "OCR_CLI_FALLBACK"
+    });
+  }
+  if (detected === "tesseract") {
+    warnings?.push({
+      message: "tesseract.js\uB294 \uD14C\uC774\uBE14 \uAD6C\uC870\uB97C \uBCF5\uC6D0\uD558\uC9C0 \uBABB\uD569\uB2C8\uB2E4. Vision LLM CLI(gemini/claude/codex) \uC124\uCE58\uB97C \uAD8C\uC7A5\uD569\uB2C8\uB2E4.",
+      code: "OCR_CLI_FALLBACK"
+    });
+    const { createTesseractProvider } = await import("./tesseract-provider-MZ37ZKQW.js");
+    return createTesseractProvider();
+  }
+  return createCliOcrProvider(detected);
+}
+export {
+  resolveOcrProvider
+};
+//# sourceMappingURL=resolve-Y3KMGD3R.js.map

package/dist/resolve-Y3KMGD3R.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/ocr/auto-detect.ts","../src/ocr/cli-provider.ts","../src/ocr/resolve.ts"],"sourcesContent":["/**\n * OCR CLI 자동 탐색\n *\n * 탐색 순서: gemini → claude → codex → ollama → tesseract.js\n * CLI는 which(unix) / where(win) 명령어로 PATH 존재 확인.\n * tesseract.js는 createRequire로 resolve 시도.\n */\n\nimport { execSync } from \"child_process\"\nimport { createRequire } from \"module\"\nimport type { OcrMode } from \"../types.js\"\n\n/** CLI 탐색 우선순위 */\nconst CLI_PRIORITY = [\"gemini\", \"claude\", \"codex\", \"ollama\"] as const\n\n/**\n * 시스템에 설치된 OCR 도구를 우선순위대로 탐색.\n * @returns 사용 가능한 OcrMode 또는 null\n */\nexport function detectAvailableOcr(): OcrMode | null {\n  // 1. CLI 프로그램 탐색 (gemini → claude → codex → ollama)\n  for (const cli of CLI_PRIORITY) {\n    if (isCliInstalled(cli)) return cli\n  }\n\n  // 2. tesseract.js npm 패키지 탐색\n  if (isTesseractAvailable()) return \"tesseract\"\n\n  return null\n}\n\n/**\n * 특정 CLI가 시스템 PATH에 있는지 확인.\n * which(unix) 또는 where(win32) 사용.\n */\nfunction isCliInstalled(name: string): boolean {\n  try {\n    const cmd = process.platform === \"win32\" ? \"where\" : \"which\"\n    execSync(`${cmd} ${name}`, { stdio: \"ignore\", timeout: 3000 })\n    return true\n  } catch {\n    return false\n  }\n}\n\n/**\n * tesseract.js npm 패키지가 설치되어 있는지 확인.\n * ESM 환경에서 createRequire를 사용하여 resolve 시도.\n */\nfunction isTesseractAvailable(): boolean {\n  try {\n    const require = createRequire(import.meta.url)\n    require.resolve(\"tesseract.js\")\n    return true\n  } catch {\n    return false\n  }\n}\n\n/**\n * 수동 지정된 OcrMode 유효성 검증.\n * --ocr gemini 등 강제 지정 시 호출.\n * @throws 해당 CLI/패키지가 설치되지 않은 경우 Error\n */\nexport function validateOcrMode(mode: OcrMode): void {\n  if (mode === \"auto\" || mode === \"off\") return\n\n  if (mode === \"tesseract\") {\n    if (!isTesseractAvailable()) {\n      throw new Error(\n        \"tesseract.js가 설치되지 않았습니다.\\n\" +\n        \"설치: npm install tesseract.js\"\n      )\n    }\n    return\n  }\n\n  if (!isCliInstalled(mode)) {\n    throw new Error(`'${mode}' CLI가 설치되지 않았습니다.\\n${getInstallGuide(mode)}`)\n  }\n}\n\n/** CLI별 설치 안내 메시지 */\nfunction getInstallGuide(mode: string): string {\n  const guides: Record<string, string> = {\n    gemini: \"설치: https://ai.google.dev/gemini-api/docs/cli\",\n    claude: \"설치: npm install -g @anthropic-ai/claude-code 또는 https://claude.ai/code\",\n    codex:  \"설치: npm install -g @openai/codex 또는 https://github.com/openai/codex\",\n    ollama: \"설치: brew install ollama 또는 https://ollama.com/download\",\n  }\n  return guides[mode] || `'${mode}'을(를) 설치해주세요.`\n}\n\n/**\n * OCR 도구를 하나도 찾지 못했을 때 표시할 에러 메시지.\n */\nexport function getNoOcrErrorMessage(): string {\n  return [\n    \"이미지 기반 PDF입니다. OCR을 위해 다음 중 하나를 설치하세요:\",\n    \"\",\n    \"  [권장] Gemini CLI:  https://ai.google.dev/gemini-api/docs/cli\",\n    \"         Claude CLI:  npm install -g @anthropic-ai/claude-code\",\n    \"         Codex CLI:   npm install -g @openai/codex\",\n    \"         Ollama:      brew install ollama (+ ollama pull gemma4:27b)\",\n    \"         Tesseract:   npm install tesseract.js\",\n    \"\",\n    \"설치 후 다시 실행하면 자동으로 감지됩니다.\",\n    \"특정 도구 지정: kordoc parse <파일> --ocr gemini\",\n  ].join(\"\\n\")\n}\n","/**\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, unlinkSync, mkdtempSync } 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 =\n  \"이 PDF 페이지 이미지에서 텍스트와 테이블을 추출하여 순수 Markdown으로 변환하세요.\\n\" +\n  \"규칙:\\n\" +\n  \"- 테이블은 Markdown 테이블 문법 사용 (| 구분, |---|---| 헤더 구분선 포함)\\n\" +\n  \"- 병합된 셀은 해당 위치에 내용 기재\\n\" +\n  \"- 헤딩은 글자 크기에 따라 ## ~ ###### 사용\\n\" +\n  \"- 리스트는 - 또는 1. 사용\\n\" +\n  \"- 이미지, 도형 등 비텍스트 요소는 무시\\n\" +\n  \"- 원문의 읽기 순서와 구조를 유지\\n\" +\n  \"- ```로 감싸지 말고 순수 Markdown만 출력\"\n\n/** 임시 디렉토리 (프로세스당 1회 생성) */\nlet _tempDir: string | null = null\nfunction getTempDir(): string {\n  if (!_tempDir) _tempDir = mkdtempSync(join(tmpdir(), \"kordoc-ocr-\"))\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-${pageNumber}.png`)\n\n    try {\n      writeFileSync(tempPath, pageImage)\n\n      let output: string\n      if (mode === \"ollama\") {\n        output = await callOllamaApi(tempPath)\n      } else {\n        output = callCli(mode, tempPath)\n      }\n\n      return { markdown: stripCodeFence(output.trim()) }\n    } finally {\n      try { unlinkSync(tempPath) } catch { /* 임시 파일 정리 실패 무시 */ }\n    }\n  }\n}\n\n/**\n * CLI 실행 — gemini / claude / codex\n *\n * @throws CLI 실행 실패 또는 타임아웃(60초) 시 Error\n */\nfunction callCli(mode: string, imagePath: string): string {\n  const args = buildCliArgs(mode, imagePath)\n\n  const result = spawnSync(mode === \"codex\" ? \"codex\" : mode, args, {\n    encoding: \"utf-8\",\n    timeout: 60_000,\n    maxBuffer: 10 * 1024 * 1024,\n  })\n\n  if (result.error) {\n    throw new Error(`${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: ${errMsg}`)\n  }\n\n  return result.stdout || \"\"\n}\n\n/**\n * CLI별 인자 배열 생성.\n *\n * gemini:  [\"--prompt\", \"프롬프트 @이미지경로\", \"--yolo\"]\n *   - -y/--yolo: 자동 승인 (OCR은 도구 사용 없으므로 실질적 영향 없음)\n *   - @ 파일 참조로 이미지를 컨텍스트에 포함\n *\n * claude:  [\"--print\", \"프롬프트 @이미지경로\"]\n *   - --print(-p): 비대화형 출력 모드\n *   - @ 파일 참조로 이미지를 컨텍스트에 포함\n *\n * codex:   [\"exec\", \"--image\", 이미지경로, \"프롬프트\"]\n *   - exec: 비대화형 실행 서브커맨드\n *   - -i/--image: 이미지 첨부 플래그 (codex exec --help 로 확인됨)\n *\n * ⚠️ CLI 버전에 따라 문법이 다를 수 있음. 업데이트 시 --help 재확인 필요.\n */\nfunction buildCliArgs(mode: string, imagePath: string): string[] {\n  const promptWithImage = `${OCR_PROMPT}\\n\\n이미지: @${imagePath}`\n  const promptOnly = OCR_PROMPT\n\n  switch (mode) {\n    case \"gemini\":\n      return [\"--prompt\", promptWithImage, \"--yolo\"]\n\n    case \"claude\":\n      return [\"--print\", promptWithImage]\n\n    case \"codex\":\n      return [\"exec\", \"--image\", imagePath, promptOnly]\n\n    default:\n      throw new Error(`지원하지 않는 CLI: ${mode}`)\n  }\n}\n\n/**\n * Ollama REST API 호출 — CLI는 이미지 입력을 지원하지 않으므로 API 직접 사용.\n *\n * 기본 모델: KORDOC_OLLAMA_MODEL 환경변수 또는 \"gemma4:27b\"\n * 기본 호스트: KORDOC_OLLAMA_HOST 환경변수 또는 \"http://localhost:11434\"\n *\n * @throws Ollama 서버 미실행 또는 응답 오류 시 Error\n */\nasync function callOllamaApi(imagePath: string): Promise<string> {\n  const { readFileSync } = await import(\"fs\")\n  const imageBase64 = readFileSync(imagePath).toString(\"base64\")\n\n  const model = process.env.KORDOC_OLLAMA_MODEL || \"gemma4:27b\"\n  const host = process.env.KORDOC_OLLAMA_HOST || \"http://localhost:11434\"\n\n  const response = await fetch(`${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(60_000),\n  })\n\n  if (!response.ok) {\n    throw new Error(`Ollama API 오류: ${response.status} ${response.statusText}`)\n  }\n\n  const data = await response.json() as { message?: { content?: string } }\n  return data.message?.content || \"\"\n}\n\n/**\n * LLM 출력에서 코드 펜스 제거.\n * LLM이 가끔 결과를 ```markdown ... ``` 으로 감싸는 경우 처리.\n */\nfunction stripCodeFence(text: string): string {\n  const match = text.match(/^```(?:markdown|md)?\\s*\\n([\\s\\S]*?)\\n```\\s*$/m)\n  return match ? match[1].trim() : text\n}\n","/**\n * OCR 프로바이더 팩토리\n *\n * ocrMode에 따라 적절한 OcrProvider를 생성하여 반환.\n * - \"auto\": 설치된 CLI 자동 탐색 (gemini → claude → codex → ollama → tesseract)\n * - 특정 CLI: 해당 CLI 사용 (미설치 시 에러)\n * - \"off\": 에러 throw\n */\n\nimport type { OcrMode, OcrProvider, ParseWarning } from \"../types.js\"\nimport { detectAvailableOcr, validateOcrMode } from \"./auto-detect.js\"\nimport { createCliOcrProvider } from \"./cli-provider.js\"\n\n/**\n * ocrMode에 따라 OcrProvider를 생성.\n *\n * @param mode - OCR 모드\n * @param warnings - 경고 수집 배열 (fallback 발생 시 경고 추가)\n * @returns OcrProvider 함수\n * @throws 사용 가능한 OCR 도구가 없거나 mode=\"off\"인 경우\n */\nexport async function resolveOcrProvider(\n  mode: OcrMode,\n  warnings?: ParseWarning[]\n): Promise<OcrProvider> {\n  if (mode === \"off\") {\n    throw new Error(\"OCR이 비활성화되어 있습니다 (--ocr off).\")\n  }\n\n  // ── 수동 지정 모드 ──────────────────────────────────\n  if (mode !== \"auto\") {\n    validateOcrMode(mode)\n\n    if (mode === \"tesseract\") {\n      const { createTesseractProvider } = await import(\"./tesseract-provider.js\")\n      return createTesseractProvider()\n    }\n\n    return createCliOcrProvider(mode)\n  }\n\n  // ── 자동 탐색 모드 ───────────────────────────────────\n  const detected = detectAvailableOcr()\n\n  if (!detected) {\n    throw new Error(\"사용 가능한 OCR 도구를 찾을 수 없습니다.\")\n  }\n\n  // gemini가 아닌 경우 fallback 경고\n  if (detected !== \"gemini\") {\n    warnings?.push({\n      message: `OCR: '${detected}' 사용 중 (gemini CLI가 없어 fallback). 더 나은 품질을 위해 gemini CLI 설치를 권장합니다.`,\n      code: \"OCR_CLI_FALLBACK\",\n    })\n  }\n\n  if (detected === \"tesseract\") {\n    warnings?.push({\n      message: \"tesseract.js는 테이블 구조를 복원하지 못합니다. Vision LLM CLI(gemini/claude/codex) 설치를 권장합니다.\",\n      code: \"OCR_CLI_FALLBACK\",\n    })\n    const { createTesseractProvider } = await import(\"./tesseract-provider.js\")\n    return createTesseractProvider()\n  }\n\n  return createCliOcrProvider(detected)\n}\n"],"mappings":";;;;AAQA,SAAS,gBAAgB;AACzB,SAAS,qBAAqB;AAI9B,IAAM,eAAe,CAAC,UAAU,UAAU,SAAS,QAAQ;AAMpD,SAAS,qBAAqC;AAEnD,aAAW,OAAO,cAAc;AAC9B,QAAI,eAAe,GAAG,EAAG,QAAO;AAAA,EAClC;AAGA,MAAI,qBAAqB,EAAG,QAAO;AAEnC,SAAO;AACT;AAMA,SAAS,eAAe,MAAuB;AAC7C,MAAI;AACF,UAAM,MAAM,QAAQ,aAAa,UAAU,UAAU;AACrD,aAAS,GAAG,GAAG,IAAI,IAAI,IAAI,EAAE,OAAO,UAAU,SAAS,IAAK,CAAC;AAC7D,WAAO;AAAA,EACT,QAAQ;AACN,WAAO;AAAA,EACT;AACF;AAMA,SAAS,uBAAgC;AACvC,MAAI;AACF,UAAMA,WAAU,cAAc,YAAY,GAAG;AAC7C,IAAAA,SAAQ,QAAQ,cAAc;AAC9B,WAAO;AAAA,EACT,QAAQ;AACN,WAAO;AAAA,EACT;AACF;AAOO,SAAS,gBAAgB,MAAqB;AACnD,MAAI,SAAS,UAAU,SAAS,MAAO;AAEvC,MAAI,SAAS,aAAa;AACxB,QAAI,CAAC,qBAAqB,GAAG;AAC3B,YAAM,IAAI;AAAA,QACR;AAAA,MAEF;AAAA,IACF;AACA;AAAA,EACF;AAEA,MAAI,CAAC,eAAe,IAAI,GAAG;AACzB,UAAM,IAAI,MAAM,IAAI,IAAI;AAAA,EAAuB,gBAAgB,IAAI,CAAC,EAAE;AAAA,EACxE;AACF;AAGA,SAAS,gBAAgB,MAAsB;AAC7C,QAAM,SAAiC;AAAA,IACrC,QAAQ;AAAA,IACR,QAAQ;AAAA,IACR,OAAQ;AAAA,IACR,QAAQ;AAAA,EACV;AACA,SAAO,OAAO,IAAI,KAAK,IAAI,IAAI;AACjC;;;AC9EA,SAAS,iBAAiB;AAC1B,SAAS,eAAe,YAAY,mBAAmB;AACvD,SAAS,YAAY;AACrB,SAAS,cAAc;AAIvB,IAAM,aACJ;AAWF,IAAI,WAA0B;AAC9B,SAAS,aAAqB;AAC5B,MAAI,CAAC,SAAU,YAAW,YAAY,KAAK,OAAO,GAAG,aAAa,CAAC;AACnE,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;AACxD,QAAM,OAAO,aAAa,MAAM,SAAS;AAEzC,QAAM,SAAS,UAAU,SAAS,UAAU,UAAU,MAAM,MAAM;AAAA,IAChE,UAAU;AAAA,IACV,SAAS;AAAA,IACT,WAAW,KAAK,OAAO;AAAA,EACzB,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;AAmBA,SAAS,aAAa,MAAc,WAA6B;AAC/D,QAAM,kBAAkB,GAAG,UAAU;AAAA;AAAA,uBAAa,SAAS;AAC3D,QAAM,aAAa;AAEnB,UAAQ,MAAM;AAAA,IACZ,KAAK;AACH,aAAO,CAAC,YAAY,iBAAiB,QAAQ;AAAA,IAE/C,KAAK;AACH,aAAO,CAAC,WAAW,eAAe;AAAA,IAEpC,KAAK;AACH,aAAO,CAAC,QAAQ,WAAW,WAAW,UAAU;AAAA,IAElD;AACE,YAAM,IAAI,MAAM,8CAAgB,IAAI,EAAE;AAAA,EAC1C;AACF;AAUA,eAAe,cAAc,WAAoC;AAC/D,QAAM,EAAE,aAAa,IAAI,MAAM,OAAO,IAAI;AAC1C,QAAM,cAAc,aAAa,SAAS,EAAE,SAAS,QAAQ;AAE7D,QAAM,QAAQ,QAAQ,IAAI,uBAAuB;AACjD,QAAM,OAAO,QAAQ,IAAI,sBAAsB;AAE/C,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,GAAM;AAAA,EACpC,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,+CAA+C;AACxE,SAAO,QAAQ,MAAM,CAAC,EAAE,KAAK,IAAI;AACnC;;;ACxJA,eAAsB,mBACpB,MACA,UACsB;AACtB,MAAI,SAAS,OAAO;AAClB,UAAM,IAAI,MAAM,sFAA+B;AAAA,EACjD;AAGA,MAAI,SAAS,QAAQ;AACnB,oBAAgB,IAAI;AAEpB,QAAI,SAAS,aAAa;AACxB,YAAM,EAAE,wBAAwB,IAAI,MAAM,OAAO,kCAAyB;AAC1E,aAAO,wBAAwB;AAAA,IACjC;AAEA,WAAO,qBAAqB,IAAI;AAAA,EAClC;AAGA,QAAM,WAAW,mBAAmB;AAEpC,MAAI,CAAC,UAAU;AACb,UAAM,IAAI,MAAM,sGAA2B;AAAA,EAC7C;AAGA,MAAI,aAAa,UAAU;AACzB,cAAU,KAAK;AAAA,MACb,SAAS,SAAS,QAAQ;AAAA,MAC1B,MAAM;AAAA,IACR,CAAC;AAAA,EACH;AAEA,MAAI,aAAa,aAAa;AAC5B,cAAU,KAAK;AAAA,MACb,SAAS;AAAA,MACT,MAAM;AAAA,IACR,CAAC;AACD,UAAM,EAAE,wBAAwB,IAAI,MAAM,OAAO,kCAAyB;AAC1E,WAAO,wBAAwB;AAAA,EACjC;AAEA,SAAO,qBAAqB,QAAQ;AACtC;","names":["require"]}

package/dist/tesseract-provider-MZ37ZKQW.js

@@ -0,0 +1,31 @@
+#!/usr/bin/env node
+import "./chunk-ZWE3DS7E.js";
+
+// src/ocr/tesseract-provider.ts
+async function createTesseractProvider() {
+  let tesseract;
+  try {
+    tesseract = await import("tesseract.js");
+  } catch {
+    throw new Error(
+      "tesseract.js\uAC00 \uC124\uCE58\uB418\uC9C0 \uC54A\uC558\uC2B5\uB2C8\uB2E4.\n\uC124\uCE58: npm install tesseract.js"
+    );
+  }
+  const worker = await tesseract.createWorker("kor+eng");
+  let terminated = false;
+  const provider = async (pageImage, _pageNumber, _mimeType) => {
+    const { data } = await worker.recognize(pageImage);
+    return data.text;
+  };
+  provider.terminate = async () => {
+    if (!terminated) {
+      await worker.terminate();
+      terminated = true;
+    }
+  };
+  return provider;
+}
+export {
+  createTesseractProvider
+};
+//# sourceMappingURL=tesseract-provider-MZ37ZKQW.js.map

package/dist/tesseract-provider-MZ37ZKQW.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/ocr/tesseract-provider.ts"],"sourcesContent":["/**\n * Tesseract.js 기반 OCR 프로바이더\n *\n * tesseract.js를 선택적 의존성으로 동적 import하여 텍스트 추출.\n * Vision LLM CLI가 없을 때의 최후 fallback.\n *\n * 제한사항:\n * - 순수 텍스트만 반환 (테이블/헤딩 구조 복원 불가)\n * - 한글 인식률 약 85-90% (깨끗한 이미지 기준)\n * - 설치 필요: npm install tesseract.js\n */\n\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 필요.\n *\n * @returns OcrProvider 함수\n * @throws tesseract.js 미설치 시 Error\n */\nexport async function createTesseractProvider(): Promise<OcrProvider & { terminate: () => Promise<void> }> {\n  // 동적 import — 미설치 시 명확한 에러\n  let tesseract: typeof import(\"tesseract.js\")\n  try {\n    tesseract = await import(\"tesseract.js\")\n  } catch {\n    throw new Error(\n      \"tesseract.js가 설치되지 않았습니다.\\n\" +\n      \"설치: npm install tesseract.js\"\n    )\n  }\n\n  // kor+eng: 한글 + 영문 동시 인식 (한국 공문서 특성)\n  const worker = await tesseract.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  // 외부에서 호출하여 워커 정리\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"],"mappings":";;;;AAuBA,eAAsB,0BAAqF;AAEzG,MAAI;AACJ,MAAI;AACF,gBAAY,MAAM,OAAO,cAAc;AAAA,EACzC,QAAQ;AACN,UAAM,IAAI;AAAA,MACR;AAAA,IAEF;AAAA,EACF;AAGA,QAAM,SAAS,MAAM,UAAU,aAAa,SAAS;AACrD,MAAI,aAAa;AAEjB,QAAM,WAAW,OACf,WACA,aACA,cACoB;AACpB,UAAM,EAAE,KAAK,IAAI,MAAM,OAAO,UAAU,SAAS;AACjD,WAAO,KAAK;AAAA,EACd;AAGC,EAAC,SAA8D,YAAY,YAAY;AACtF,QAAI,CAAC,YAAY;AACf,YAAM,OAAO,UAAU;AACvB,mBAAa;AAAA,IACf;AAAA,EACF;AAEA,SAAO;AACT;","names":[]}

package/dist/{utils-7JE5SKSL.js → utils-4NP2VUFW.js}

@@ -8,7 +8,8 @@ import {
   sanitizeError,
   sanitizeHref,
   toArrayBuffer
-} from "./chunk-TFYOEQE2.js";
+} from "./chunk-XWET7ONC.js";
+import "./chunk-ZWE3DS7E.js";
 export {
   KordocError,
   VERSION,
@@ -19,4 +20,4 @@ export {
   sanitizeHref,
   toArrayBuffer
 };
-//# sourceMappingURL=utils-7JE5SKSL.js.map
+//# sourceMappingURL=utils-4NP2VUFW.js.map

package/dist/utils-4NP2VUFW.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":[],"sourcesContent":[],"mappings":"","names":[]}

package/dist/{watch-XALC6VOR.js → watch-4VVWG2WC.js}

@@ -2,11 +2,12 @@
 import {
   detectFormat,
   parse
-} from "./chunk-H7HMKSLX.js";
+} from "./chunk-JJMA5HGQ.js";
 import {
   toArrayBuffer
-} from "./chunk-TFYOEQE2.js";
+} from "./chunk-XWET7ONC.js";
 import "./chunk-MOL7MDBG.js";
+import "./chunk-ZWE3DS7E.js";
 
 // src/watch.ts
 import { watch, readFileSync, writeFileSync, mkdirSync, statSync, existsSync } from "fs";
@@ -125,4 +126,4 @@ async function sendWebhook(url, payload) {
 export {
   watchDirectory
 };
-//# sourceMappingURL=watch-XALC6VOR.js.map
+//# sourceMappingURL=watch-4VVWG2WC.js.map

package/dist/{watch-XALC6VOR.js.map → watch-4VVWG2WC.js.map}

@@ -1 +1 @@
-{"version":3,"sources":["../src/watch.ts"],"sourcesContent":["/** 디렉토리 감시 모드 — 새 문서 자동 변환 + Webhook 알림 */\n\nimport { watch, readFileSync, writeFileSync, mkdirSync, statSync, existsSync } from \"fs\"\nimport { basename, resolve, extname } from \"path\"\nimport { parse, detectFormat } from \"./index.js\"\nimport { toArrayBuffer } from \"./utils.js\"\nimport type { WatchOptions } from \"./types.js\"\n\nconst SUPPORTED_EXTENSIONS = new Set([\".hwp\", \".hwpx\", \".pdf\", \".xlsx\", \".docx\"])\nconst DEBOUNCE_MS = 1000\n/** 파일 쓰기 완료 판정: 연속 2회 동일 크기 확인 간격 */\nconst STABLE_CHECK_MS = 300\nconst MAX_FILE_SIZE = 500 * 1024 * 1024\n\n/**\n * 디렉토리를 감시하여 새 문서 파일을 자동 변환.\n *\n * @example\n * ```bash\n * kordoc watch ./incoming -d ./output --webhook https://api.example.com/docs\n * ```\n */\nexport async function watchDirectory(options: WatchOptions): Promise<void> {\n  const { dir, outDir, webhook, format = \"markdown\", pages, silent } = options\n\n  if (!existsSync(dir)) throw new Error(`디렉토리를 찾을 수 없습니다: ${dir}`)\n  if (webhook) validateWebhookUrl(webhook)\n  if (outDir) mkdirSync(outDir, { recursive: true })\n\n  const log = silent ? () => {} : (msg: string) => process.stderr.write(msg + \"\\n\")\n  log(`[kordoc watch] 감시 시작: ${resolve(dir)}`)\n  if (outDir) log(`[kordoc watch] 출력: ${resolve(outDir)}`)\n  if (webhook) log(`[kordoc watch] 웹훅: ${webhook}`)\n\n  // 디바운스 맵\n  const pending = new Map<string, ReturnType<typeof setTimeout>>()\n\n  /** 파일 크기가 안정화될 때까지 대기 (쓰기 완료 감지) */\n  const waitForStableSize = async (absPath: string): Promise<number> => {\n    let prevSize = statSync(absPath).size\n    await new Promise(r => setTimeout(r, STABLE_CHECK_MS))\n    if (!existsSync(absPath)) return 0\n    const currSize = statSync(absPath).size\n    if (currSize !== prevSize) {\n      // 크기가 변했으면 한 번 더 대기\n      await new Promise(r => setTimeout(r, STABLE_CHECK_MS))\n      if (!existsSync(absPath)) return 0\n      return statSync(absPath).size\n    }\n    return currSize\n  }\n\n  const processFile = async (filePath: string) => {\n    const ext = extname(filePath).toLowerCase()\n    if (!SUPPORTED_EXTENSIONS.has(ext)) return\n\n    const fileName = basename(filePath)\n    try {\n      const absPath = resolve(dir, filePath)\n      // 경로 순회 방지 — 감시 디렉토리 외부 파일 차단\n      const realDir = resolve(dir)\n      if (!absPath.startsWith(realDir)) return\n      if (!existsSync(absPath)) return\n\n      const fileSize = await waitForStableSize(absPath)\n      if (fileSize > MAX_FILE_SIZE || fileSize === 0) return\n\n      log(`[kordoc watch] 변환 중: ${fileName}`)\n\n      const buffer = readFileSync(absPath)\n      const arrayBuffer = toArrayBuffer(buffer)\n      const parseOptions = pages ? { pages } : undefined\n      const result = await parse(arrayBuffer, parseOptions)\n\n      if (!result.success) {\n        log(`[kordoc watch] 실패: ${fileName} — ${result.error}`)\n        await sendWebhook(webhook, { file: fileName, format: detectFormat(arrayBuffer), success: false, error: result.error })\n        return\n      }\n\n      const output = format === \"json\" ? JSON.stringify(result, null, 2) : result.markdown\n\n      if (outDir) {\n        const outExt = format === \"json\" ? \".json\" : \".md\"\n        const outPath = resolve(outDir, fileName.replace(/\\.[^.]+$/, outExt))\n        writeFileSync(outPath, output, \"utf-8\")\n        log(`[kordoc watch] 완료: ${fileName} → ${basename(outPath)}`)\n      } else {\n        process.stdout.write(output + \"\\n\")\n      }\n\n      await sendWebhook(webhook, {\n        file: fileName,\n        format: result.fileType,\n        success: true,\n        markdown: format === \"markdown\" ? output.substring(0, 1000) : undefined,\n      })\n    } catch (err) {\n      log(`[kordoc watch] 에러: ${fileName} — ${err instanceof Error ? err.message : err}`)\n    }\n  }\n\n  // fs.watch recursive (Node 18+ Windows/macOS, Node 19+ Linux)\n  watch(dir, { recursive: true }, (event, filename) => {\n    if (!filename) return\n    const filePath = filename.toString()\n\n    // 디바운스\n    const existing = pending.get(filePath)\n    if (existing) clearTimeout(existing)\n    pending.set(filePath, setTimeout(() => {\n      pending.delete(filePath)\n      processFile(filePath).catch(() => {})\n    }, DEBOUNCE_MS))\n  })\n\n  // 프로세스 종료 방지 (Ctrl+C로 종료)\n  return new Promise(() => {})\n}\n\n/** Webhook URL 검증 — SSRF 방지: http/https만 허용, localhost/private IP 차단 */\nfunction validateWebhookUrl(url: string): void {\n  let parsed: URL\n  try {\n    parsed = new URL(url)\n  } catch {\n    throw new Error(`유효하지 않은 webhook URL: ${url}`)\n  }\n  if (parsed.protocol !== \"http:\" && parsed.protocol !== \"https:\") {\n    throw new Error(`허용되지 않는 webhook 프로토콜: ${parsed.protocol}`)\n  }\n  const hostname = parsed.hostname.toLowerCase()\n  if (\n    hostname === \"localhost\" ||\n    hostname === \"[::1]\" ||\n    hostname.startsWith(\"127.\") ||\n    hostname.startsWith(\"10.\") ||\n    hostname.startsWith(\"192.168.\") ||\n    /^172\\.(1[6-9]|2\\d|3[01])\\./.test(hostname) ||\n    hostname === \"0.0.0.0\" ||\n    hostname.startsWith(\"169.254.\") ||\n    hostname.endsWith(\".local\") ||\n    // IPv6 사설 대역\n    hostname.startsWith(\"[fc\") ||\n    hostname.startsWith(\"[fd\") ||\n    hostname.startsWith(\"[fe80:\") ||\n    hostname === \"[::0]\" ||\n    hostname === \"[::]\" ||\n    // 클라우드 메타데이터 엔드포인트\n    hostname === \"metadata.google.internal\" ||\n    hostname === \"metadata.google\" ||\n    // 16진수/8진수 IP 인코딩 우회 방지\n    /^0x[0-9a-f]+$/i.test(hostname) ||\n    /^0[0-7]+$/.test(hostname)\n  ) {\n    throw new Error(`내부 네트워크 대상 webhook은 허용되지 않습니다: ${hostname}`)\n  }\n}\n\nasync function sendWebhook(url: string | undefined, payload: Record<string, unknown>): Promise<void> {\n  if (!url) return\n  try {\n    validateWebhookUrl(url)\n    await fetch(url, {\n      method: \"POST\",\n      headers: { \"Content-Type\": \"application/json\" },\n      body: JSON.stringify({ ...payload, timestamp: new Date().toISOString() }),\n    })\n  } catch {\n    // webhook 실패는 조용히 무시\n  }\n}\n"],"mappings":";;;;;;;;;;;AAEA,SAAS,OAAO,cAAc,eAAe,WAAW,UAAU,kBAAkB;AACpF,SAAS,UAAU,SAAS,eAAe;AAK3C,IAAM,uBAAuB,oBAAI,IAAI,CAAC,QAAQ,SAAS,QAAQ,SAAS,OAAO,CAAC;AAChF,IAAM,cAAc;AAEpB,IAAM,kBAAkB;AACxB,IAAM,gBAAgB,MAAM,OAAO;AAUnC,eAAsB,eAAe,SAAsC;AACzE,QAAM,EAAE,KAAK,QAAQ,SAAS,SAAS,YAAY,OAAO,OAAO,IAAI;AAErE,MAAI,CAAC,WAAW,GAAG,EAAG,OAAM,IAAI,MAAM,gFAAoB,GAAG,EAAE;AAC/D,MAAI,QAAS,oBAAmB,OAAO;AACvC,MAAI,OAAQ,WAAU,QAAQ,EAAE,WAAW,KAAK,CAAC;AAEjD,QAAM,MAAM,SAAS,MAAM;AAAA,EAAC,IAAI,CAAC,QAAgB,QAAQ,OAAO,MAAM,MAAM,IAAI;AAChF,MAAI,6CAAyB,QAAQ,GAAG,CAAC,EAAE;AAC3C,MAAI,OAAQ,KAAI,gCAAsB,QAAQ,MAAM,CAAC,EAAE;AACvD,MAAI,QAAS,KAAI,gCAAsB,OAAO,EAAE;AAGhD,QAAM,UAAU,oBAAI,IAA2C;AAG/D,QAAM,oBAAoB,OAAO,YAAqC;AACpE,QAAI,WAAW,SAAS,OAAO,EAAE;AACjC,UAAM,IAAI,QAAQ,OAAK,WAAW,GAAG,eAAe,CAAC;AACrD,QAAI,CAAC,WAAW,OAAO,EAAG,QAAO;AACjC,UAAM,WAAW,SAAS,OAAO,EAAE;AACnC,QAAI,aAAa,UAAU;AAEzB,YAAM,IAAI,QAAQ,OAAK,WAAW,GAAG,eAAe,CAAC;AACrD,UAAI,CAAC,WAAW,OAAO,EAAG,QAAO;AACjC,aAAO,SAAS,OAAO,EAAE;AAAA,IAC3B;AACA,WAAO;AAAA,EACT;AAEA,QAAM,cAAc,OAAO,aAAqB;AAC9C,UAAM,MAAM,QAAQ,QAAQ,EAAE,YAAY;AAC1C,QAAI,CAAC,qBAAqB,IAAI,GAAG,EAAG;AAEpC,UAAM,WAAW,SAAS,QAAQ;AAClC,QAAI;AACF,YAAM,UAAU,QAAQ,KAAK,QAAQ;AAErC,YAAM,UAAU,QAAQ,GAAG;AAC3B,UAAI,CAAC,QAAQ,WAAW,OAAO,EAAG;AAClC,UAAI,CAAC,WAAW,OAAO,EAAG;AAE1B,YAAM,WAAW,MAAM,kBAAkB,OAAO;AAChD,UAAI,WAAW,iBAAiB,aAAa,EAAG;AAEhD,UAAI,uCAAwB,QAAQ,EAAE;AAEtC,YAAM,SAAS,aAAa,OAAO;AACnC,YAAM,cAAc,cAAc,MAAM;AACxC,YAAM,eAAe,QAAQ,EAAE,MAAM,IAAI;AACzC,YAAM,SAAS,MAAM,MAAM,aAAa,YAAY;AAEpD,UAAI,CAAC,OAAO,SAAS;AACnB,YAAI,gCAAsB,QAAQ,WAAM,OAAO,KAAK,EAAE;AACtD,cAAM,YAAY,SAAS,EAAE,MAAM,UAAU,QAAQ,aAAa,WAAW,GAAG,SAAS,OAAO,OAAO,OAAO,MAAM,CAAC;AACrH;AAAA,MACF;AAEA,YAAM,SAAS,WAAW,SAAS,KAAK,UAAU,QAAQ,MAAM,CAAC,IAAI,OAAO;AAE5E,UAAI,QAAQ;AACV,cAAM,SAAS,WAAW,SAAS,UAAU;AAC7C,cAAM,UAAU,QAAQ,QAAQ,SAAS,QAAQ,YAAY,MAAM,CAAC;AACpE,sBAAc,SAAS,QAAQ,OAAO;AACtC,YAAI,gCAAsB,QAAQ,WAAM,SAAS,OAAO,CAAC,EAAE;AAAA,MAC7D,OAAO;AACL,gBAAQ,OAAO,MAAM,SAAS,IAAI;AAAA,MACpC;AAEA,YAAM,YAAY,SAAS;AAAA,QACzB,MAAM;AAAA,QACN,QAAQ,OAAO;AAAA,QACf,SAAS;AAAA,QACT,UAAU,WAAW,aAAa,OAAO,UAAU,GAAG,GAAI,IAAI;AAAA,MAChE,CAAC;AAAA,IACH,SAAS,KAAK;AACZ,UAAI,gCAAsB,QAAQ,WAAM,eAAe,QAAQ,IAAI,UAAU,GAAG,EAAE;AAAA,IACpF;AAAA,EACF;AAGA,QAAM,KAAK,EAAE,WAAW,KAAK,GAAG,CAAC,OAAO,aAAa;AACnD,QAAI,CAAC,SAAU;AACf,UAAM,WAAW,SAAS,SAAS;AAGnC,UAAM,WAAW,QAAQ,IAAI,QAAQ;AACrC,QAAI,SAAU,cAAa,QAAQ;AACnC,YAAQ,IAAI,UAAU,WAAW,MAAM;AACrC,cAAQ,OAAO,QAAQ;AACvB,kBAAY,QAAQ,EAAE,MAAM,MAAM;AAAA,MAAC,CAAC;AAAA,IACtC,GAAG,WAAW,CAAC;AAAA,EACjB,CAAC;AAGD,SAAO,IAAI,QAAQ,MAAM;AAAA,EAAC,CAAC;AAC7B;AAGA,SAAS,mBAAmB,KAAmB;AAC7C,MAAI;AACJ,MAAI;AACF,aAAS,IAAI,IAAI,GAAG;AAAA,EACtB,QAAQ;AACN,UAAM,IAAI,MAAM,sDAAwB,GAAG,EAAE;AAAA,EAC/C;AACA,MAAI,OAAO,aAAa,WAAW,OAAO,aAAa,UAAU;AAC/D,UAAM,IAAI,MAAM,2EAAyB,OAAO,QAAQ,EAAE;AAAA,EAC5D;AACA,QAAM,WAAW,OAAO,SAAS,YAAY;AAC7C,MACE,aAAa,eACb,aAAa,WACb,SAAS,WAAW,MAAM,KAC1B,SAAS,WAAW,KAAK,KACzB,SAAS,WAAW,UAAU,KAC9B,6BAA6B,KAAK,QAAQ,KAC1C,aAAa,aACb,SAAS,WAAW,UAAU,KAC9B,SAAS,SAAS,QAAQ;AAAA,EAE1B,SAAS,WAAW,KAAK,KACzB,SAAS,WAAW,KAAK,KACzB,SAAS,WAAW,QAAQ,KAC5B,aAAa,WACb,aAAa;AAAA,EAEb,aAAa,8BACb,aAAa;AAAA,EAEb,iBAAiB,KAAK,QAAQ,KAC9B,YAAY,KAAK,QAAQ,GACzB;AACA,UAAM,IAAI,MAAM,uHAAkC,QAAQ,EAAE;AAAA,EAC9D;AACF;AAEA,eAAe,YAAY,KAAyB,SAAiD;AACnG,MAAI,CAAC,IAAK;AACV,MAAI;AACF,uBAAmB,GAAG;AACtB,UAAM,MAAM,KAAK;AAAA,MACf,QAAQ;AAAA,MACR,SAAS,EAAE,gBAAgB,mBAAmB;AAAA,MAC9C,MAAM,KAAK,UAAU,EAAE,GAAG,SAAS,YAAW,oBAAI,KAAK,GAAE,YAAY,EAAE,CAAC;AAAA,IAC1E,CAAC;AAAA,EACH,QAAQ;AAAA,EAER;AACF;","names":[]}
+{"version":3,"sources":["../src/watch.ts"],"sourcesContent":["/** 디렉토리 감시 모드 — 새 문서 자동 변환 + Webhook 알림 */\n\nimport { watch, readFileSync, writeFileSync, mkdirSync, statSync, existsSync } from \"fs\"\nimport { basename, resolve, extname } from \"path\"\nimport { parse, detectFormat } from \"./index.js\"\nimport { toArrayBuffer } from \"./utils.js\"\nimport type { WatchOptions } from \"./types.js\"\n\nconst SUPPORTED_EXTENSIONS = new Set([\".hwp\", \".hwpx\", \".pdf\", \".xlsx\", \".docx\"])\nconst DEBOUNCE_MS = 1000\n/** 파일 쓰기 완료 판정: 연속 2회 동일 크기 확인 간격 */\nconst STABLE_CHECK_MS = 300\nconst MAX_FILE_SIZE = 500 * 1024 * 1024\n\n/**\n * 디렉토리를 감시하여 새 문서 파일을 자동 변환.\n *\n * @example\n * ```bash\n * kordoc watch ./incoming -d ./output --webhook https://api.example.com/docs\n * ```\n */\nexport async function watchDirectory(options: WatchOptions): Promise<void> {\n  const { dir, outDir, webhook, format = \"markdown\", pages, silent } = options\n\n  if (!existsSync(dir)) throw new Error(`디렉토리를 찾을 수 없습니다: ${dir}`)\n  if (webhook) validateWebhookUrl(webhook)\n  if (outDir) mkdirSync(outDir, { recursive: true })\n\n  const log = silent ? () => {} : (msg: string) => process.stderr.write(msg + \"\\n\")\n  log(`[kordoc watch] 감시 시작: ${resolve(dir)}`)\n  if (outDir) log(`[kordoc watch] 출력: ${resolve(outDir)}`)\n  if (webhook) log(`[kordoc watch] 웹훅: ${webhook}`)\n\n  // 디바운스 맵\n  const pending = new Map<string, ReturnType<typeof setTimeout>>()\n\n  /** 파일 크기가 안정화될 때까지 대기 (쓰기 완료 감지) */\n  const waitForStableSize = async (absPath: string): Promise<number> => {\n    let prevSize = statSync(absPath).size\n    await new Promise(r => setTimeout(r, STABLE_CHECK_MS))\n    if (!existsSync(absPath)) return 0\n    const currSize = statSync(absPath).size\n    if (currSize !== prevSize) {\n      // 크기가 변했으면 한 번 더 대기\n      await new Promise(r => setTimeout(r, STABLE_CHECK_MS))\n      if (!existsSync(absPath)) return 0\n      return statSync(absPath).size\n    }\n    return currSize\n  }\n\n  const processFile = async (filePath: string) => {\n    const ext = extname(filePath).toLowerCase()\n    if (!SUPPORTED_EXTENSIONS.has(ext)) return\n\n    const fileName = basename(filePath)\n    try {\n      const absPath = resolve(dir, filePath)\n      // 경로 순회 방지 — 감시 디렉토리 외부 파일 차단\n      const realDir = resolve(dir)\n      if (!absPath.startsWith(realDir)) return\n      if (!existsSync(absPath)) return\n\n      const fileSize = await waitForStableSize(absPath)\n      if (fileSize > MAX_FILE_SIZE || fileSize === 0) return\n\n      log(`[kordoc watch] 변환 중: ${fileName}`)\n\n      const buffer = readFileSync(absPath)\n      const arrayBuffer = toArrayBuffer(buffer)\n      const parseOptions = pages ? { pages } : undefined\n      const result = await parse(arrayBuffer, parseOptions)\n\n      if (!result.success) {\n        log(`[kordoc watch] 실패: ${fileName} — ${result.error}`)\n        await sendWebhook(webhook, { file: fileName, format: detectFormat(arrayBuffer), success: false, error: result.error })\n        return\n      }\n\n      const output = format === \"json\" ? JSON.stringify(result, null, 2) : result.markdown\n\n      if (outDir) {\n        const outExt = format === \"json\" ? \".json\" : \".md\"\n        const outPath = resolve(outDir, fileName.replace(/\\.[^.]+$/, outExt))\n        writeFileSync(outPath, output, \"utf-8\")\n        log(`[kordoc watch] 완료: ${fileName} → ${basename(outPath)}`)\n      } else {\n        process.stdout.write(output + \"\\n\")\n      }\n\n      await sendWebhook(webhook, {\n        file: fileName,\n        format: result.fileType,\n        success: true,\n        markdown: format === \"markdown\" ? output.substring(0, 1000) : undefined,\n      })\n    } catch (err) {\n      log(`[kordoc watch] 에러: ${fileName} — ${err instanceof Error ? err.message : err}`)\n    }\n  }\n\n  // fs.watch recursive (Node 18+ Windows/macOS, Node 19+ Linux)\n  watch(dir, { recursive: true }, (event, filename) => {\n    if (!filename) return\n    const filePath = filename.toString()\n\n    // 디바운스\n    const existing = pending.get(filePath)\n    if (existing) clearTimeout(existing)\n    pending.set(filePath, setTimeout(() => {\n      pending.delete(filePath)\n      processFile(filePath).catch(() => {})\n    }, DEBOUNCE_MS))\n  })\n\n  // 프로세스 종료 방지 (Ctrl+C로 종료)\n  return new Promise(() => {})\n}\n\n/** Webhook URL 검증 — SSRF 방지: http/https만 허용, localhost/private IP 차단 */\nfunction validateWebhookUrl(url: string): void {\n  let parsed: URL\n  try {\n    parsed = new URL(url)\n  } catch {\n    throw new Error(`유효하지 않은 webhook URL: ${url}`)\n  }\n  if (parsed.protocol !== \"http:\" && parsed.protocol !== \"https:\") {\n    throw new Error(`허용되지 않는 webhook 프로토콜: ${parsed.protocol}`)\n  }\n  const hostname = parsed.hostname.toLowerCase()\n  if (\n    hostname === \"localhost\" ||\n    hostname === \"[::1]\" ||\n    hostname.startsWith(\"127.\") ||\n    hostname.startsWith(\"10.\") ||\n    hostname.startsWith(\"192.168.\") ||\n    /^172\\.(1[6-9]|2\\d|3[01])\\./.test(hostname) ||\n    hostname === \"0.0.0.0\" ||\n    hostname.startsWith(\"169.254.\") ||\n    hostname.endsWith(\".local\") ||\n    // IPv6 사설 대역\n    hostname.startsWith(\"[fc\") ||\n    hostname.startsWith(\"[fd\") ||\n    hostname.startsWith(\"[fe80:\") ||\n    hostname === \"[::0]\" ||\n    hostname === \"[::]\" ||\n    // 클라우드 메타데이터 엔드포인트\n    hostname === \"metadata.google.internal\" ||\n    hostname === \"metadata.google\" ||\n    // 16진수/8진수 IP 인코딩 우회 방지\n    /^0x[0-9a-f]+$/i.test(hostname) ||\n    /^0[0-7]+$/.test(hostname)\n  ) {\n    throw new Error(`내부 네트워크 대상 webhook은 허용되지 않습니다: ${hostname}`)\n  }\n}\n\nasync function sendWebhook(url: string | undefined, payload: Record<string, unknown>): Promise<void> {\n  if (!url) return\n  try {\n    validateWebhookUrl(url)\n    await fetch(url, {\n      method: \"POST\",\n      headers: { \"Content-Type\": \"application/json\" },\n      body: JSON.stringify({ ...payload, timestamp: new Date().toISOString() }),\n    })\n  } catch {\n    // webhook 실패는 조용히 무시\n  }\n}\n"],"mappings":";;;;;;;;;;;;AAEA,SAAS,OAAO,cAAc,eAAe,WAAW,UAAU,kBAAkB;AACpF,SAAS,UAAU,SAAS,eAAe;AAK3C,IAAM,uBAAuB,oBAAI,IAAI,CAAC,QAAQ,SAAS,QAAQ,SAAS,OAAO,CAAC;AAChF,IAAM,cAAc;AAEpB,IAAM,kBAAkB;AACxB,IAAM,gBAAgB,MAAM,OAAO;AAUnC,eAAsB,eAAe,SAAsC;AACzE,QAAM,EAAE,KAAK,QAAQ,SAAS,SAAS,YAAY,OAAO,OAAO,IAAI;AAErE,MAAI,CAAC,WAAW,GAAG,EAAG,OAAM,IAAI,MAAM,gFAAoB,GAAG,EAAE;AAC/D,MAAI,QAAS,oBAAmB,OAAO;AACvC,MAAI,OAAQ,WAAU,QAAQ,EAAE,WAAW,KAAK,CAAC;AAEjD,QAAM,MAAM,SAAS,MAAM;AAAA,EAAC,IAAI,CAAC,QAAgB,QAAQ,OAAO,MAAM,MAAM,IAAI;AAChF,MAAI,6CAAyB,QAAQ,GAAG,CAAC,EAAE;AAC3C,MAAI,OAAQ,KAAI,gCAAsB,QAAQ,MAAM,CAAC,EAAE;AACvD,MAAI,QAAS,KAAI,gCAAsB,OAAO,EAAE;AAGhD,QAAM,UAAU,oBAAI,IAA2C;AAG/D,QAAM,oBAAoB,OAAO,YAAqC;AACpE,QAAI,WAAW,SAAS,OAAO,EAAE;AACjC,UAAM,IAAI,QAAQ,OAAK,WAAW,GAAG,eAAe,CAAC;AACrD,QAAI,CAAC,WAAW,OAAO,EAAG,QAAO;AACjC,UAAM,WAAW,SAAS,OAAO,EAAE;AACnC,QAAI,aAAa,UAAU;AAEzB,YAAM,IAAI,QAAQ,OAAK,WAAW,GAAG,eAAe,CAAC;AACrD,UAAI,CAAC,WAAW,OAAO,EAAG,QAAO;AACjC,aAAO,SAAS,OAAO,EAAE;AAAA,IAC3B;AACA,WAAO;AAAA,EACT;AAEA,QAAM,cAAc,OAAO,aAAqB;AAC9C,UAAM,MAAM,QAAQ,QAAQ,EAAE,YAAY;AAC1C,QAAI,CAAC,qBAAqB,IAAI,GAAG,EAAG;AAEpC,UAAM,WAAW,SAAS,QAAQ;AAClC,QAAI;AACF,YAAM,UAAU,QAAQ,KAAK,QAAQ;AAErC,YAAM,UAAU,QAAQ,GAAG;AAC3B,UAAI,CAAC,QAAQ,WAAW,OAAO,EAAG;AAClC,UAAI,CAAC,WAAW,OAAO,EAAG;AAE1B,YAAM,WAAW,MAAM,kBAAkB,OAAO;AAChD,UAAI,WAAW,iBAAiB,aAAa,EAAG;AAEhD,UAAI,uCAAwB,QAAQ,EAAE;AAEtC,YAAM,SAAS,aAAa,OAAO;AACnC,YAAM,cAAc,cAAc,MAAM;AACxC,YAAM,eAAe,QAAQ,EAAE,MAAM,IAAI;AACzC,YAAM,SAAS,MAAM,MAAM,aAAa,YAAY;AAEpD,UAAI,CAAC,OAAO,SAAS;AACnB,YAAI,gCAAsB,QAAQ,WAAM,OAAO,KAAK,EAAE;AACtD,cAAM,YAAY,SAAS,EAAE,MAAM,UAAU,QAAQ,aAAa,WAAW,GAAG,SAAS,OAAO,OAAO,OAAO,MAAM,CAAC;AACrH;AAAA,MACF;AAEA,YAAM,SAAS,WAAW,SAAS,KAAK,UAAU,QAAQ,MAAM,CAAC,IAAI,OAAO;AAE5E,UAAI,QAAQ;AACV,cAAM,SAAS,WAAW,SAAS,UAAU;AAC7C,cAAM,UAAU,QAAQ,QAAQ,SAAS,QAAQ,YAAY,MAAM,CAAC;AACpE,sBAAc,SAAS,QAAQ,OAAO;AACtC,YAAI,gCAAsB,QAAQ,WAAM,SAAS,OAAO,CAAC,EAAE;AAAA,MAC7D,OAAO;AACL,gBAAQ,OAAO,MAAM,SAAS,IAAI;AAAA,MACpC;AAEA,YAAM,YAAY,SAAS;AAAA,QACzB,MAAM;AAAA,QACN,QAAQ,OAAO;AAAA,QACf,SAAS;AAAA,QACT,UAAU,WAAW,aAAa,OAAO,UAAU,GAAG,GAAI,IAAI;AAAA,MAChE,CAAC;AAAA,IACH,SAAS,KAAK;AACZ,UAAI,gCAAsB,QAAQ,WAAM,eAAe,QAAQ,IAAI,UAAU,GAAG,EAAE;AAAA,IACpF;AAAA,EACF;AAGA,QAAM,KAAK,EAAE,WAAW,KAAK,GAAG,CAAC,OAAO,aAAa;AACnD,QAAI,CAAC,SAAU;AACf,UAAM,WAAW,SAAS,SAAS;AAGnC,UAAM,WAAW,QAAQ,IAAI,QAAQ;AACrC,QAAI,SAAU,cAAa,QAAQ;AACnC,YAAQ,IAAI,UAAU,WAAW,MAAM;AACrC,cAAQ,OAAO,QAAQ;AACvB,kBAAY,QAAQ,EAAE,MAAM,MAAM;AAAA,MAAC,CAAC;AAAA,IACtC,GAAG,WAAW,CAAC;AAAA,EACjB,CAAC;AAGD,SAAO,IAAI,QAAQ,MAAM;AAAA,EAAC,CAAC;AAC7B;AAGA,SAAS,mBAAmB,KAAmB;AAC7C,MAAI;AACJ,MAAI;AACF,aAAS,IAAI,IAAI,GAAG;AAAA,EACtB,QAAQ;AACN,UAAM,IAAI,MAAM,sDAAwB,GAAG,EAAE;AAAA,EAC/C;AACA,MAAI,OAAO,aAAa,WAAW,OAAO,aAAa,UAAU;AAC/D,UAAM,IAAI,MAAM,2EAAyB,OAAO,QAAQ,EAAE;AAAA,EAC5D;AACA,QAAM,WAAW,OAAO,SAAS,YAAY;AAC7C,MACE,aAAa,eACb,aAAa,WACb,SAAS,WAAW,MAAM,KAC1B,SAAS,WAAW,KAAK,KACzB,SAAS,WAAW,UAAU,KAC9B,6BAA6B,KAAK,QAAQ,KAC1C,aAAa,aACb,SAAS,WAAW,UAAU,KAC9B,SAAS,SAAS,QAAQ;AAAA,EAE1B,SAAS,WAAW,KAAK,KACzB,SAAS,WAAW,KAAK,KACzB,SAAS,WAAW,QAAQ,KAC5B,aAAa,WACb,aAAa;AAAA,EAEb,aAAa,8BACb,aAAa;AAAA,EAEb,iBAAiB,KAAK,QAAQ,KAC9B,YAAY,KAAK,QAAQ,GACzB;AACA,UAAM,IAAI,MAAM,uHAAkC,QAAQ,EAAE;AAAA,EAC9D;AACF;AAEA,eAAe,YAAY,KAAyB,SAAiD;AACnG,MAAI,CAAC,IAAK;AACV,MAAI;AACF,uBAAmB,GAAG;AACtB,UAAM,MAAM,KAAK;AAAA,MACf,QAAQ;AAAA,MACR,SAAS,EAAE,gBAAgB,mBAAmB;AAAA,MAC9C,MAAM,KAAK,UAAU,EAAE,GAAG,SAAS,YAAW,oBAAI,KAAK,GAAE,YAAY,EAAE,CAAC;AAAA,IAC1E,CAAC;AAAA,EACH,QAAQ;AAAA,EAER;AACF;","names":[]}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@clazic/kordoc",
-  "version": "2.1.6",
+  "version": "2.2.0",
   "description": "Parse Korean documents (HWP, HWPX, PDF, XLSX, DOCX) to Markdown",
   "type": "module",
   "exports": {
@@ -26,7 +26,8 @@
     "test": "node --import tsx --test tests/*.test.ts",
     "prepublishOnly": "npm run test && npm run build",
     "build:worker": "tsup --config tsup.worker.config.ts",
-    "deploy:worker": "npm run build:worker && npx wrangler deploy"
+    "deploy:worker": "npm run build:worker && npx wrangler deploy",
+    "build:bin": "npm run build && pkg dist-binary/cli-bin.cjs --targets node18-macos-arm64,node18-macos-x64 --output kordoc"
   },
   "keywords": [
     "hwp",
@@ -58,6 +59,7 @@
     "@xmldom/xmldom": "^0.9.8",
     "cfb": "1.2.2",
     "commander": "^13.0.0",
+    "exceljs": "^4.4.0",
     "jszip": "^3.10.1",
     "pdfjs-dist": "^4.10.38",
     "zod": "^3.23.0"