@clazic/kordoc 2.4.5 → 2.4.7

package/dist/{batch-provider-PNDCSGQW.js → batch-provider-XI2EPODC.js}

@@ -2,7 +2,7 @@
 import "./chunk-ZWE3DS7E.js";
 
 // src/ocr/batch-provider.ts
-import { spawn } from "child_process";
+import { spawn, execSync } from "child_process";
 import { writeFileSync, readFileSync, unlinkSync, mkdirSync } from "fs";
 import { join } from "path";
 import { tmpdir } from "os";
@@ -15,8 +15,14 @@ var DEFAULT_BATCH_SIZES = {
 var _batchTempDir = null;
 function getBatchTempDir() {
   if (!_batchTempDir) {
-    _batchTempDir = join(process.cwd(), "_kordoc_ocr_tmp");
+    _batchTempDir = join(process.cwd(), ".kordoc_ocr_tmp");
     mkdirSync(_batchTempDir, { recursive: true });
+    if (process.platform === "win32") {
+      try {
+        execSync(`attrib +h "${_batchTempDir}"`, { stdio: "ignore" });
+      } catch {
+      }
+    }
   }
   return _batchTempDir;
 }
@@ -165,4 +171,4 @@ export {
   DEFAULT_BATCH_SIZES,
   createBatchCliProvider
 };
-//# sourceMappingURL=batch-provider-PNDCSGQW.js.map
+//# sourceMappingURL=batch-provider-XI2EPODC.js.map

package/dist/batch-provider-XI2EPODC.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/ocr/batch-provider.ts"],"sourcesContent":["/**\n * CLI 배치 OCR 프로바이더\n *\n * 여러 페이지 이미지를 단일 CLI 호출로 처리하여 API 호출 수를 대폭 감소.\n * gemini/claude: @file 멀티 참조, codex: --image 멀티 플래그\n *\n * 299페이지 기준:\n * - 기존: CLI 299회 호출 (~30분)\n * - 배치: CLI 3~6회 호출 (~3분)\n */\n\nimport { spawn, execSync } from \"child_process\"\nimport { writeFileSync, readFileSync, unlinkSync, mkdirSync } from \"fs\"\nimport { join } from \"path\"\nimport { tmpdir } from \"os\"\nimport type { StructuredOcrResult, BatchOcrProvider } from \"../types.js\"\n\n/** 배치 OCR 프롬프트 */\nconst BATCH_OCR_PROMPT =\n  \"다음 문서 페이지 이미지들을 OCR하여 순수 Markdown으로 변환하세요.\\n\\n\" +\n  \"규칙:\\n\" +\n  \"- 각 페이지 결과 사이에 반드시 이 구분자를 삽입: <!-- PAGE_BREAK -->\\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/** 모드별 기본 배치 크기 (CLI 내부 타임아웃 + 실측 기반)\n *\n * gemini CLI: 10장 이상에서 AbortError 발생 (내부 타임아웃).\n * 5장 배치가 안정적으로 동작 확인 (35초/배치).\n * 299페이지 = 60배치 = 기존 299회 대비 80% 감소.\n */\nexport const DEFAULT_BATCH_SIZES: Record<string, number> = {\n  gemini: 5,\n  claude: 5,\n  codex: 10,\n}\n\n/**\n * 임시 디렉토리 — gemini CLI는 cwd 하위 + gitignore 밖만 @참조 가능\n *\n * 숨김 처리:\n * - macOS/Linux: '.' 접두사로 기본 숨김 (ls -a 로만 표시)\n * - Windows: '.' 접두사 + attrib +h 로 숨김 속성 부여\n */\nlet _batchTempDir: string | null = null\nfunction getBatchTempDir(): string {\n  if (!_batchTempDir) {\n    _batchTempDir = join(process.cwd(), \".kordoc_ocr_tmp\")\n    mkdirSync(_batchTempDir, { recursive: true })\n    // Windows: dot-prefix만으로 숨김 처리 불충분 → attrib +h 추가\n    if (process.platform === \"win32\") {\n      try { execSync(`attrib +h \"${_batchTempDir}\"`, { stdio: \"ignore\" }) } catch { /* ignore */ }\n    }\n  }\n  return _batchTempDir\n}\n\n/**\n * 배치 CLI 프로바이더 생성\n */\nexport function createBatchCliProvider(\n  mode: \"gemini\" | \"claude\" | \"codex\",\n  batchSize: number\n): BatchOcrProvider {\n  return {\n    __batch: true as const,\n    batchSize,\n    async processBatch(pages) {\n      const results = new Map<number, StructuredOcrResult>()\n      const tempDir = getBatchTempDir()\n      const tempFiles: string[] = []\n\n      try {\n        // 1. Write all page images to temp files\n        for (const { image, pageNum } of pages) {\n          const path = join(tempDir, `batch-p${pageNum}.png`)\n          writeFileSync(path, image)\n          tempFiles.push(path)\n        }\n\n        // 2. Call CLI with all file references (비동기 — 병렬 배치 실행 가능)\n        let output: string\n        if (mode === \"codex\") {\n          output = await callBatchCodexCli(tempFiles)\n        } else {\n          output = await callBatchCli(mode, tempFiles)\n        }\n\n        // 3. Parse response by PAGE_BREAK separator\n        const cleaned = stripCodeFence(output.trim())\n        const parts = cleaned.split(/<!--\\s*PAGE_BREAK\\s*-->/)\n          .map(p => p.trim())\n          .filter(p => p.length > 0)\n\n        // 4. Map results to page numbers (best-effort if count mismatch)\n        for (let i = 0; i < pages.length; i++) {\n          const pageNum = pages[i].pageNum\n          if (i < parts.length) {\n            results.set(pageNum, { markdown: parts[i] })\n          }\n          // If fewer parts than pages, remaining pages get no result\n        }\n      } finally {\n        // 5. Clean up temp files\n        for (const f of tempFiles) {\n          try { unlinkSync(f) } catch { /* ignore */ }\n        }\n      }\n\n      return results\n    },\n  }\n}\n\n/**\n * 비동기 CLI 실행 헬퍼 — spawn + Promise 래핑.\n * spawnSync는 이벤트 루프를 차단하여 병렬 배치 실행 불가.\n */\nfunction spawnAsync(\n  cmd: string,\n  args: string[],\n  opts: { timeoutMs: number; cwd?: string; stdin?: string }\n): Promise<{ stdout: string; stderr: string; exitCode: number }> {\n  return new Promise((resolve, reject) => {\n    const child = spawn(cmd, args, {\n      cwd: opts.cwd,\n      env: process.env,\n      stdio: [\"pipe\", \"pipe\", \"pipe\"],\n    })\n\n    let stdout = \"\"\n    let stderr = \"\"\n    let killed = false\n\n    child.stdout.setEncoding(\"utf-8\")\n    child.stderr.setEncoding(\"utf-8\")\n    child.stdout.on(\"data\", (d: string) => { stdout += d })\n    child.stderr.on(\"data\", (d: string) => { stderr += d })\n\n    const timer = setTimeout(() => {\n      killed = true\n      child.kill(\"SIGTERM\")\n    }, opts.timeoutMs)\n\n    if (opts.stdin !== undefined) {\n      child.stdin.end(opts.stdin)\n    } else {\n      child.stdin.end()\n    }\n\n    child.on(\"close\", (code) => {\n      clearTimeout(timer)\n      if (killed) {\n        reject(new Error(`타임아웃 (${Math.round(opts.timeoutMs / 1000)}초)`))\n      } else {\n        resolve({ stdout, stderr, exitCode: code ?? 1 })\n      }\n    })\n    child.on(\"error\", (err) => {\n      clearTimeout(timer)\n      reject(err)\n    })\n  })\n}\n\n/** gemini/claude 배치 호출 (비동기) */\nasync function callBatchCli(mode: \"gemini\" | \"claude\", imagePaths: string[]): Promise<string> {\n  const fileRefs = imagePaths.map(p => `@${p}`).join(\"\\n\")\n  const prompt = `${BATCH_OCR_PROMPT}\\n\\n${fileRefs}`\n\n  let args: string[]\n  if (mode === \"gemini\") {\n    const model = process.env.KORDOC_GEMINI_MODEL ?? \"gemini-2.5-flash\"\n    args = [\"--prompt\", prompt, \"--yolo\", \"--model\", model]\n  } else {\n    args = [\"--print\", prompt]\n    const model = process.env.KORDOC_CLAUDE_MODEL\n    if (model) args.push(\"--model\", model)\n  }\n\n  const timeoutMs = 60_000 + imagePaths.length * 20_000\n  const result = await spawnAsync(mode, args, {\n    timeoutMs,\n    ...(mode === \"claude\" ? { cwd: tmpdir() } : {}),\n  })\n\n  if (result.exitCode !== 0) {\n    const errMsg = result.stderr?.trim() || `exit code ${result.exitCode}`\n    throw new Error(`${mode} 배치 OCR 실패: ${errMsg}`)\n  }\n\n  return result.stdout || \"\"\n}\n\n/** codex 배치 호출 (비동기) — --image를 여러 번 지정 */\nasync function callBatchCodexCli(imagePaths: string[]): Promise<string> {\n  const outPath = join(tmpdir(), `kordoc-codex-batch-${Date.now()}-${Math.random().toString(36).slice(2)}.txt`)\n  try {\n    const args = [\"exec\", BATCH_OCR_PROMPT]\n    for (const p of imagePaths) {\n      args.push(\"--image\", p)\n    }\n    args.push(\"--output-last-message\", outPath)\n    const model = process.env.KORDOC_CODEX_MODEL\n    if (model) args.push(\"--model\", model)\n\n    const timeoutMs = 60_000 + imagePaths.length * 20_000\n    const result = await spawnAsync(\"codex\", args, {\n      timeoutMs,\n      stdin: \"\",\n    })\n\n    if (result.exitCode !== 0) {\n      const errMsg = result.stderr?.trim() || `exit code ${result.exitCode}`\n      throw new Error(`codex 배치 OCR 실패: ${errMsg}`)\n    }\n\n    try {\n      return readFileSync(outPath, \"utf-8\")\n    } catch {\n      return result.stdout || \"\"\n    }\n  } finally {\n    try { unlinkSync(outPath) } catch { /* ignore */ }\n  }\n}\n\n/** LLM 출력에서 코드 펜스 제거 (cli-provider.ts와 동일 로직) */\nfunction stripCodeFence(text: string): string {\n  const match = text.match(/^```(?:markdown|md)?\\s*\\n([\\s\\S]*?)\\n```\\s*$/m)\n  return match ? match[1].trim() : text\n}\n"],"mappings":";;;;AAWA,SAAS,OAAO,gBAAgB;AAChC,SAAS,eAAe,cAAc,YAAY,iBAAiB;AACnE,SAAS,YAAY;AACrB,SAAS,cAAc;AAIvB,IAAM,mBACJ;AAiBK,IAAM,sBAA8C;AAAA,EACzD,QAAQ;AAAA,EACR,QAAQ;AAAA,EACR,OAAO;AACT;AASA,IAAI,gBAA+B;AACnC,SAAS,kBAA0B;AACjC,MAAI,CAAC,eAAe;AAClB,oBAAgB,KAAK,QAAQ,IAAI,GAAG,iBAAiB;AACrD,cAAU,eAAe,EAAE,WAAW,KAAK,CAAC;AAE5C,QAAI,QAAQ,aAAa,SAAS;AAChC,UAAI;AAAE,iBAAS,cAAc,aAAa,KAAK,EAAE,OAAO,SAAS,CAAC;AAAA,MAAE,QAAQ;AAAA,MAAe;AAAA,IAC7F;AAAA,EACF;AACA,SAAO;AACT;AAKO,SAAS,uBACd,MACA,WACkB;AAClB,SAAO;AAAA,IACL,SAAS;AAAA,IACT;AAAA,IACA,MAAM,aAAa,OAAO;AACxB,YAAM,UAAU,oBAAI,IAAiC;AACrD,YAAM,UAAU,gBAAgB;AAChC,YAAM,YAAsB,CAAC;AAE7B,UAAI;AAEF,mBAAW,EAAE,OAAO,QAAQ,KAAK,OAAO;AACtC,gBAAM,OAAO,KAAK,SAAS,UAAU,OAAO,MAAM;AAClD,wBAAc,MAAM,KAAK;AACzB,oBAAU,KAAK,IAAI;AAAA,QACrB;AAGA,YAAI;AACJ,YAAI,SAAS,SAAS;AACpB,mBAAS,MAAM,kBAAkB,SAAS;AAAA,QAC5C,OAAO;AACL,mBAAS,MAAM,aAAa,MAAM,SAAS;AAAA,QAC7C;AAGA,cAAM,UAAU,eAAe,OAAO,KAAK,CAAC;AAC5C,cAAM,QAAQ,QAAQ,MAAM,yBAAyB,EAClD,IAAI,OAAK,EAAE,KAAK,CAAC,EACjB,OAAO,OAAK,EAAE,SAAS,CAAC;AAG3B,iBAAS,IAAI,GAAG,IAAI,MAAM,QAAQ,KAAK;AACrC,gBAAM,UAAU,MAAM,CAAC,EAAE;AACzB,cAAI,IAAI,MAAM,QAAQ;AACpB,oBAAQ,IAAI,SAAS,EAAE,UAAU,MAAM,CAAC,EAAE,CAAC;AAAA,UAC7C;AAAA,QAEF;AAAA,MACF,UAAE;AAEA,mBAAW,KAAK,WAAW;AACzB,cAAI;AAAE,uBAAW,CAAC;AAAA,UAAE,QAAQ;AAAA,UAAe;AAAA,QAC7C;AAAA,MACF;AAEA,aAAO;AAAA,IACT;AAAA,EACF;AACF;AAMA,SAAS,WACP,KACA,MACA,MAC+D;AAC/D,SAAO,IAAI,QAAQ,CAAC,SAAS,WAAW;AACtC,UAAM,QAAQ,MAAM,KAAK,MAAM;AAAA,MAC7B,KAAK,KAAK;AAAA,MACV,KAAK,QAAQ;AAAA,MACb,OAAO,CAAC,QAAQ,QAAQ,MAAM;AAAA,IAChC,CAAC;AAED,QAAI,SAAS;AACb,QAAI,SAAS;AACb,QAAI,SAAS;AAEb,UAAM,OAAO,YAAY,OAAO;AAChC,UAAM,OAAO,YAAY,OAAO;AAChC,UAAM,OAAO,GAAG,QAAQ,CAAC,MAAc;AAAE,gBAAU;AAAA,IAAE,CAAC;AACtD,UAAM,OAAO,GAAG,QAAQ,CAAC,MAAc;AAAE,gBAAU;AAAA,IAAE,CAAC;AAEtD,UAAM,QAAQ,WAAW,MAAM;AAC7B,eAAS;AACT,YAAM,KAAK,SAAS;AAAA,IACtB,GAAG,KAAK,SAAS;AAEjB,QAAI,KAAK,UAAU,QAAW;AAC5B,YAAM,MAAM,IAAI,KAAK,KAAK;AAAA,IAC5B,OAAO;AACL,YAAM,MAAM,IAAI;AAAA,IAClB;AAEA,UAAM,GAAG,SAAS,CAAC,SAAS;AAC1B,mBAAa,KAAK;AAClB,UAAI,QAAQ;AACV,eAAO,IAAI,MAAM,6BAAS,KAAK,MAAM,KAAK,YAAY,GAAI,CAAC,SAAI,CAAC;AAAA,MAClE,OAAO;AACL,gBAAQ,EAAE,QAAQ,QAAQ,UAAU,QAAQ,EAAE,CAAC;AAAA,MACjD;AAAA,IACF,CAAC;AACD,UAAM,GAAG,SAAS,CAAC,QAAQ;AACzB,mBAAa,KAAK;AAClB,aAAO,GAAG;AAAA,IACZ,CAAC;AAAA,EACH,CAAC;AACH;AAGA,eAAe,aAAa,MAA2B,YAAuC;AAC5F,QAAM,WAAW,WAAW,IAAI,OAAK,IAAI,CAAC,EAAE,EAAE,KAAK,IAAI;AACvD,QAAM,SAAS,GAAG,gBAAgB;AAAA;AAAA,EAAO,QAAQ;AAEjD,MAAI;AACJ,MAAI,SAAS,UAAU;AACrB,UAAM,QAAQ,QAAQ,IAAI,uBAAuB;AACjD,WAAO,CAAC,YAAY,QAAQ,UAAU,WAAW,KAAK;AAAA,EACxD,OAAO;AACL,WAAO,CAAC,WAAW,MAAM;AACzB,UAAM,QAAQ,QAAQ,IAAI;AAC1B,QAAI,MAAO,MAAK,KAAK,WAAW,KAAK;AAAA,EACvC;AAEA,QAAM,YAAY,MAAS,WAAW,SAAS;AAC/C,QAAM,SAAS,MAAM,WAAW,MAAM,MAAM;AAAA,IAC1C;AAAA,IACA,GAAI,SAAS,WAAW,EAAE,KAAK,OAAO,EAAE,IAAI,CAAC;AAAA,EAC/C,CAAC;AAED,MAAI,OAAO,aAAa,GAAG;AACzB,UAAM,SAAS,OAAO,QAAQ,KAAK,KAAK,aAAa,OAAO,QAAQ;AACpE,UAAM,IAAI,MAAM,GAAG,IAAI,mCAAe,MAAM,EAAE;AAAA,EAChD;AAEA,SAAO,OAAO,UAAU;AAC1B;AAGA,eAAe,kBAAkB,YAAuC;AACtE,QAAM,UAAU,KAAK,OAAO,GAAG,sBAAsB,KAAK,IAAI,CAAC,IAAI,KAAK,OAAO,EAAE,SAAS,EAAE,EAAE,MAAM,CAAC,CAAC,MAAM;AAC5G,MAAI;AACF,UAAM,OAAO,CAAC,QAAQ,gBAAgB;AACtC,eAAW,KAAK,YAAY;AAC1B,WAAK,KAAK,WAAW,CAAC;AAAA,IACxB;AACA,SAAK,KAAK,yBAAyB,OAAO;AAC1C,UAAM,QAAQ,QAAQ,IAAI;AAC1B,QAAI,MAAO,MAAK,KAAK,WAAW,KAAK;AAErC,UAAM,YAAY,MAAS,WAAW,SAAS;AAC/C,UAAM,SAAS,MAAM,WAAW,SAAS,MAAM;AAAA,MAC7C;AAAA,MACA,OAAO;AAAA,IACT,CAAC;AAED,QAAI,OAAO,aAAa,GAAG;AACzB,YAAM,SAAS,OAAO,QAAQ,KAAK,KAAK,aAAa,OAAO,QAAQ;AACpE,YAAM,IAAI,MAAM,wCAAoB,MAAM,EAAE;AAAA,IAC9C;AAEA,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,IAAe;AAAA,EACnD;AACF;AAGA,SAAS,eAAe,MAAsB;AAC5C,QAAM,QAAQ,KAAK,MAAM,+CAA+C;AACxE,SAAO,QAAQ,MAAM,CAAC,EAAE,KAAK,IAAI;AACnC;","names":[]}

package/dist/{chunk-CG3DV7QG.js → chunk-5DD75UUX.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 
 // src/utils.ts
-var VERSION = true ? "2.4.5" : "0.0.0-dev";
+var VERSION = true ? "2.4.7" : "0.0.0-dev";
 function toArrayBuffer(buf) {
   if (buf.byteOffset === 0 && buf.byteLength === buf.buffer.byteLength) {
     return buf.buffer;
@@ -90,4 +90,4 @@ export {
   sanitizeHref,
   classifyError
 };
-//# sourceMappingURL=chunk-CG3DV7QG.js.map
+//# sourceMappingURL=chunk-5DD75UUX.js.map

package/dist/{chunk-ATB6T3SG.js → chunk-GITPGCCA.js}

@@ -9,7 +9,7 @@ import {
   precheckZipSize,
   sanitizeHref,
   toArrayBuffer
-} from "./chunk-CG3DV7QG.js";
+} from "./chunk-5DD75UUX.js";
 import {
   parsePageRange
 } from "./chunk-MOL7MDBG.js";
@@ -5462,7 +5462,7 @@ async function parsePdfDocument(buffer, options) {
       if (ocrMode === "off") {
         throw Object.assign(new KordocError(`\uC774\uBBF8\uC9C0 \uAE30\uBC18 PDF (${pageCount}\uD398\uC774\uC9C0, ${totalChars}\uC790)`), { isImageBased: true });
       }
-      const { resolveOcrProvider } = await import("./resolve-TZVGVOVD.js");
+      const { resolveOcrProvider } = await import("./resolve-SUU5Q6DJ.js");
       const { ocrPages } = await import("./provider-7F7NEDTN.js");
       const tryProvider = async (provider, filter) => {
         try {
@@ -5510,6 +5510,7 @@ async function parsePdfDocument(buffer, options) {
           } catch {
           }
         }
+        allOcrBlocks.sort((a, b) => (a.pageNumber ?? 0) - (b.pageNumber ?? 0));
         ocrBlocks = allOcrBlocks;
       } else {
         try {
@@ -9812,4 +9813,4 @@ export {
 cfb/cfb.js:
   (*! crc32.js (C) 2014-present SheetJS -- http://sheetjs.com *)
 */
-//# sourceMappingURL=chunk-ATB6T3SG.js.map
+//# sourceMappingURL=chunk-GITPGCCA.js.map