@clazic/kordoc 2.4.3 → 2.4.4

package/dist/mcp.js

@@ -10,17 +10,17 @@ import {
   markdownToHwpx,
   markdownToXlsx,
   parse
-} from "./chunk-IAU7NTTA.js";
+} from "./chunk-5AXJRBBK.js";
+import "./chunk-4PP34NVQ.js";
 import {
   KordocError,
   VERSION,
   sanitizeError,
   toArrayBuffer
-} from "./chunk-HOUVJPR7.js";
+} from "./chunk-KEDUF24M.js";
 import "./chunk-MOL7MDBG.js";
 import "./chunk-7FMKAV4P.js";
 import "./chunk-JOGAFNIL.js";
-import "./chunk-4PP34NVQ.js";
 import "./chunk-ZWE3DS7E.js";
 
 // src/mcp.ts

package/dist/{resolve-UOAOPQ4H.js → resolve-TZVGVOVD.js}

@@ -2,54 +2,13 @@
 import {
   createCliOcrProvider
 } from "./chunk-JOGAFNIL.js";
+import {
+  detectAvailableOcr,
+  getTesseractFallbackMessage,
+  validateOcrMode
+} from "./chunk-7NOZFYH6.js";
 import "./chunk-ZWE3DS7E.js";
 
-// src/ocr/auto-detect.ts
-import { execSync } from "child_process";
-var CLI_PRIORITY = ["codex", "gemini", "claude", "ollama"];
-function detectAvailableOcr() {
-  for (const cli of CLI_PRIORITY) {
-    if (isCliInstalled(cli)) return cli;
-  }
-  return "tesseract";
-}
-function isCliInstalled(name) {
-  try {
-    const cmd = process.platform === "win32" ? "where" : "which";
-    execSync(`${cmd} ${name}`, { stdio: "ignore", timeout: 3e3 });
-    return true;
-  } catch {
-    return false;
-  }
-}
-function validateOcrMode(mode) {
-  if (mode === "auto" || mode === "off" || mode === "tesseract") return;
-  if (!isCliInstalled(mode)) {
-    throw new Error(`'${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.`;
-}
-function getTesseractFallbackMessage() {
-  return [
-    "\uC124\uCE58\uB41C AI CLI\uAC00 \uC5C6\uC5B4 \uB0B4\uC7A5 tesseract.js\uB85C OCR\uC744 \uC218\uD589\uD569\uB2C8\uB2E4.",
-    "\uB354 \uB098\uC740 \uD488\uC9C8(\uD14C\uC774\uBE14/\uD5E4\uB529 \uAD6C\uC870 \uBCF4\uC874)\uC744 \uC704\uD574 AI CLI \uC124\uCE58\uB97C \uAD8C\uC7A5\uD569\uB2C8\uB2E4:",
-    "",
-    "  [\uAD8C\uC7A5] Codex CLI:   npm install -g @openai/codex",
-    "         Gemini CLI:  https://ai.google.dev/gemini-api/docs/cli",
-    "         Claude CLI:  npm install -g @anthropic-ai/claude-code",
-    "         Ollama:      brew install ollama (+ ollama pull gemma4:27b)"
-  ].join("\n");
-}
-
 // src/ocr/resolve.ts
 async function resolveOcrProvider(mode, warnings, concurrency, batchSize) {
   if (mode === "off") {
@@ -108,4 +67,4 @@ async function resolveOcrProvider(mode, warnings, concurrency, batchSize) {
 export {
   resolveOcrProvider
 };
-//# sourceMappingURL=resolve-UOAOPQ4H.js.map
+//# sourceMappingURL=resolve-TZVGVOVD.js.map

package/dist/resolve-TZVGVOVD.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/ocr/resolve.ts"],"sourcesContent":["/**\n * OCR 프로바이더 팩토리\n *\n * ocrMode에 따라 적절한 OcrProvider를 생성하여 반환.\n * - \"auto\": 설치된 CLI 자동 탐색 (gemini → claude → codex → ollama → tesseract)\n *           tesseract.js는 bundled 의존성이므로 항상 사용 가능 (null 반환 없음)\n * - 특정 CLI: 해당 CLI 사용 (미설치 시 에러)\n * - \"tesseract\": 내장 tesseract.js 직접 사용\n * - \"off\": 에러 throw\n */\n\nimport type { OcrMode, OcrProvider, ParseWarning, BatchOcrProvider } from \"../types.js\"\nimport { detectAvailableOcr, validateOcrMode, getTesseractFallbackMessage } 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 * @param concurrency - 병렬 처리 수 (tesseract 전용, 기본: 1=순차)\n * @returns OcrProvider 함수\n * @throws mode=\"off\"이거나 지정 CLI 미설치 시 Error\n */\nexport async function resolveOcrProvider(\n  mode: OcrMode,\n  warnings?: ParseWarning[],\n  concurrency?: number,\n  batchSize?: number\n): Promise<OcrProvider | BatchOcrProvider> {\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)  // tesseract는 항상 통과\n\n    if (mode === \"tesseract\") {\n      const { createTesseractProvider, createTesseractPoolProvider } = await import(\"./tesseract-provider.js\")\n      // concurrency > 1이면 워커 풀 사용, 그 외 단일 워커 사용\n      if (concurrency && concurrency > 1) {\n        return createTesseractPoolProvider(concurrency)\n      }\n      return createTesseractProvider()\n    }\n\n    // gemini/claude/codex: 배치 크기 > 1이면 배치 프로바이더 사용\n    if (mode === \"gemini\" || mode === \"claude\" || mode === \"codex\") {\n      const { createBatchCliProvider, DEFAULT_BATCH_SIZES } = await import(\"./batch-provider.js\")\n      const effectiveBatch = batchSize ?? DEFAULT_BATCH_SIZES[mode]\n      if (effectiveBatch > 1) {\n        return createBatchCliProvider(mode, effectiveBatch)\n      }\n      return createCliOcrProvider(mode)\n    }\n\n    // CLI 프로바이더는 rate limit 보호를 위해 concurrency 무시 (항상 순차)\n    return createCliOcrProvider(mode)\n  }\n\n  // ── 자동 탐색 모드 ───────────────────────────────────\n  // detectAvailableOcr()는 항상 값을 반환 (tesseract fallback으로 null 없음)\n  const detected = detectAvailableOcr()\n\n  // codex가 아닌 경우 fallback 경고\n  if (detected !== \"codex\") {\n    if (detected === \"tesseract\") {\n      // 내장 tesseract로 fallback — 구조 복원 제한 안내\n      warnings?.push({\n        message: getTesseractFallbackMessage(),\n        code: \"OCR_CLI_FALLBACK\",\n      })\n    } else {\n      warnings?.push({\n        message: `OCR: '${detected}' 사용 중 (codex CLI가 없어 fallback). 더 나은 품질을 위해 codex CLI 설치를 권장합니다.`,\n        code: \"OCR_CLI_FALLBACK\",\n      })\n    }\n  }\n\n  if (detected === \"tesseract\") {\n    const { createTesseractProvider, createTesseractPoolProvider } = await import(\"./tesseract-provider.js\")\n    // concurrency > 1이면 워커 풀 사용, 그 외 단일 워커 사용\n    if (concurrency && concurrency > 1) {\n      return createTesseractPoolProvider(concurrency)\n    }\n    return createTesseractProvider()\n  }\n\n  // gemini/claude/codex: 배치 크기 > 1이면 배치 프로바이더 사용\n  if (detected === \"gemini\" || detected === \"codex\" || detected === \"claude\") {\n    const { createBatchCliProvider, DEFAULT_BATCH_SIZES } = await import(\"./batch-provider.js\")\n    const effectiveBatch = batchSize ?? DEFAULT_BATCH_SIZES[detected]\n    if (effectiveBatch > 1) {\n      return createBatchCliProvider(detected, effectiveBatch)\n    }\n    return createCliOcrProvider(detected)\n  }\n\n  // CLI 프로바이더는 rate limit 보호를 위해 concurrency 무시 (항상 순차)\n  return createCliOcrProvider(detected)\n}\n"],"mappings":";;;;;;;;;;;;AAwBA,eAAsB,mBACpB,MACA,UACA,aACA,WACyC;AACzC,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,yBAAyB,4BAA4B,IAAI,MAAM,OAAO,kCAAyB;AAEvG,UAAI,eAAe,cAAc,GAAG;AAClC,eAAO,4BAA4B,WAAW;AAAA,MAChD;AACA,aAAO,wBAAwB;AAAA,IACjC;AAGA,QAAI,SAAS,YAAY,SAAS,YAAY,SAAS,SAAS;AAC9D,YAAM,EAAE,wBAAwB,oBAAoB,IAAI,MAAM,OAAO,8BAAqB;AAC1F,YAAM,iBAAiB,aAAa,oBAAoB,IAAI;AAC5D,UAAI,iBAAiB,GAAG;AACtB,eAAO,uBAAuB,MAAM,cAAc;AAAA,MACpD;AACA,aAAO,qBAAqB,IAAI;AAAA,IAClC;AAGA,WAAO,qBAAqB,IAAI;AAAA,EAClC;AAIA,QAAM,WAAW,mBAAmB;AAGpC,MAAI,aAAa,SAAS;AACxB,QAAI,aAAa,aAAa;AAE5B,gBAAU,KAAK;AAAA,QACb,SAAS,4BAA4B;AAAA,QACrC,MAAM;AAAA,MACR,CAAC;AAAA,IACH,OAAO;AACL,gBAAU,KAAK;AAAA,QACb,SAAS,SAAS,QAAQ;AAAA,QAC1B,MAAM;AAAA,MACR,CAAC;AAAA,IACH;AAAA,EACF;AAEA,MAAI,aAAa,aAAa;AAC5B,UAAM,EAAE,yBAAyB,4BAA4B,IAAI,MAAM,OAAO,kCAAyB;AAEvG,QAAI,eAAe,cAAc,GAAG;AAClC,aAAO,4BAA4B,WAAW;AAAA,IAChD;AACA,WAAO,wBAAwB;AAAA,EACjC;AAGA,MAAI,aAAa,YAAY,aAAa,WAAW,aAAa,UAAU;AAC1E,UAAM,EAAE,wBAAwB,oBAAoB,IAAI,MAAM,OAAO,8BAAqB;AAC1F,UAAM,iBAAiB,aAAa,oBAAoB,QAAQ;AAChE,QAAI,iBAAiB,GAAG;AACtB,aAAO,uBAAuB,UAAU,cAAc;AAAA,IACxD;AACA,WAAO,qBAAqB,QAAQ;AAAA,EACtC;AAGA,SAAO,qBAAqB,QAAQ;AACtC;","names":[]}

package/dist/{utils-PYEEPTPM.js → utils-BB2CDSTB.js}

@@ -8,7 +8,7 @@ import {
   sanitizeError,
   sanitizeHref,
   toArrayBuffer
-} from "./chunk-HOUVJPR7.js";
+} from "./chunk-KEDUF24M.js";
 import "./chunk-ZWE3DS7E.js";
 export {
   KordocError,
@@ -20,4 +20,4 @@ export {
   sanitizeHref,
   toArrayBuffer
 };
-//# sourceMappingURL=utils-PYEEPTPM.js.map
+//# sourceMappingURL=utils-BB2CDSTB.js.map

package/dist/utils-BB2CDSTB.js.map

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

package/dist/{watch-IQLSW2OB.js → watch-6QVK32X7.js}

@@ -2,14 +2,14 @@
 import {
   detectFormat,
   parse
-} from "./chunk-IAU7NTTA.js";
+} from "./chunk-5AXJRBBK.js";
+import "./chunk-4PP34NVQ.js";
 import {
   toArrayBuffer
-} from "./chunk-HOUVJPR7.js";
+} from "./chunk-KEDUF24M.js";
 import "./chunk-MOL7MDBG.js";
 import "./chunk-7FMKAV4P.js";
 import "./chunk-JOGAFNIL.js";
-import "./chunk-4PP34NVQ.js";
 import "./chunk-ZWE3DS7E.js";
 
 // src/watch.ts
@@ -129,4 +129,4 @@ async function sendWebhook(url, payload) {
 export {
   watchDirectory
 };
-//# sourceMappingURL=watch-IQLSW2OB.js.map
+//# sourceMappingURL=watch-6QVK32X7.js.map

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@clazic/kordoc",
-  "version": "2.4.3",
+  "version": "2.4.4",
   "description": "Parse Korean documents (HWP, HWPX, PDF, XLSX, DOCX) to Markdown",
   "type": "module",
   "exports": {