@clazic/kordoc 2.6.0 → 2.6.1

package/dist/index.cjs

@@ -130,7 +130,7 @@ var VERSION, KordocError, SAFE_HREF_RE;
 var init_utils = __esm({
   "src/utils.ts"() {
     "use strict";
-    VERSION = true ? "2.5.2" : "0.0.0-dev";
+    VERSION = true ? "2.6.0" : "0.0.0-dev";
     KordocError = class extends Error {
       code;
       stage;
@@ -3196,7 +3196,7 @@ __export(index_exports, {
   runUnifiedOcrPipeline: () => runUnifiedOcrPipeline
 });
 module.exports = __toCommonJS(index_exports);
-var import_promises4 = require("fs/promises");
+var import_promises5 = require("fs/promises");
 
 // src/detect.ts
 var import_jszip = __toESM(require("jszip"), 1);
@@ -11265,7 +11265,7 @@ async function markdownToXlsx(markdown, options) {
 }
 
 // src/convert/index.ts
-var import_promises2 = require("fs/promises");
+var import_promises3 = require("fs/promises");
 init_utils();
 
 // src/convert/libreoffice.ts
@@ -11280,6 +11280,204 @@ var ConvertError = class extends Error {
   }
 };
 
+// src/convert/installer.ts
+var import_os3 = require("os");
+var import_path5 = require("path");
+var import_promises2 = require("fs/promises");
+var import_fs4 = require("fs");
+var import_child_process4 = require("child_process");
+var CACHE_DIR = (0, import_path5.join)((0, import_os3.homedir)(), ".cache", "kordoc", "libreoffice");
+var VERSION_FILE = (0, import_path5.join)(CACHE_DIR, "version");
+var PACKAGES = {
+  darwin: {
+    url: "https://download.documentfoundation.org/libreoffice/stable/24.8.4/mac/x86_64/LibreOffice_24.8.4_MacOS_x86-64.dmg",
+    binPath: "LibreOffice.app/Contents/MacOS/soffice",
+    sizeMb: 300
+  },
+  linux: {
+    url: "https://download.documentfoundation.org/libreoffice/stable/24.8.4/deb/x86_64/LibreOffice_24.8.4_Linux_x86-64_deb.tar.gz",
+    binPath: "opt/libreoffice24.8/program/soffice",
+    sizeMb: 200
+  },
+  win32: {
+    url: "https://download.documentfoundation.org/libreoffice/stable/24.8.4/win/x86_64/LibreOffice_24.8.4_Win_x86-64.msi",
+    binPath: "LibreOffice/program/soffice.exe",
+    sizeMb: 350
+  }
+};
+async function findInPath() {
+  try {
+    const { runCommand: runCommand2 } = await Promise.resolve().then(() => (init_utils(), utils_exports));
+    await runCommand2("soffice", ["--version"]);
+    return "soffice";
+  } catch {
+    return null;
+  }
+}
+async function findInCache() {
+  const cachedBin = (0, import_path5.join)(CACHE_DIR, "bin", "soffice");
+  try {
+    await (0, import_promises2.access)(cachedBin);
+    return cachedBin;
+  } catch {
+    return null;
+  }
+}
+async function downloadWithProgress(url, dest, totalBytes, onProgress) {
+  const response = await fetch(url);
+  if (!response.body) throw new Error("\uB2E4\uC6B4\uB85C\uB4DC \uC2E4\uD328: response body \uC5C6\uC74C");
+  const file = (0, import_fs4.createWriteStream)(dest);
+  const reader = response.body.getReader();
+  let downloaded = 0;
+  try {
+    while (true) {
+      const { done, value } = await reader.read();
+      if (done) break;
+      file.write(value);
+      downloaded += value.length;
+      onProgress?.(downloaded, totalBytes);
+    }
+  } finally {
+    file.end();
+    reader.releaseLock();
+  }
+}
+async function installForPlatform(pkg, onProgress) {
+  const platform = process.platform;
+  await (0, import_promises2.mkdir)(CACHE_DIR, { recursive: true });
+  const downloadPath = (0, import_path5.join)(CACHE_DIR, `download-${Date.now()}`);
+  await downloadWithProgress(pkg.url, downloadPath, pkg.sizeMb * 1024 * 1024, onProgress);
+  try {
+    if (platform === "darwin") {
+      return await installMacOS(pkg, downloadPath);
+    } else if (platform === "linux") {
+      return await installLinux(pkg, downloadPath);
+    } else if (platform === "win32") {
+      return await installWindows(pkg, downloadPath);
+    }
+  } catch (err) {
+    await (0, import_promises2.rm)(downloadPath, { force: true });
+    throw err;
+  }
+  throw new ConvertError("UNSUPPORTED_PLATFORM", `${platform}\uC740 \uC790\uB3D9 \uC124\uCE58\uB97C \uC9C0\uC6D0\uD558\uC9C0 \uC54A\uC2B5\uB2C8\uB2E4`);
+}
+async function installMacOS(pkg, downloadPath) {
+  const mountPoint = `/Volumes/LibreOffice_${Date.now()}`;
+  await new Promise((resolve4, reject) => {
+    const child = (0, import_child_process4.spawn)("hdiutil", ["attach", "-nobrowse", "-mountpoint", mountPoint, downloadPath]);
+    child.on("close", (code) => code === 0 ? resolve4() : reject(new Error("dmg \uB9C8\uC6B4\uD2B8 \uC2E4\uD328")));
+  });
+  try {
+    const appSource = (0, import_path5.join)(mountPoint, "LibreOffice.app");
+    const appDest = (0, import_path5.join)(CACHE_DIR, "LibreOffice.app");
+    await new Promise((resolve4, reject) => {
+      const child = (0, import_child_process4.spawn)("cp", ["-R", appSource, appDest]);
+      child.on("close", (code) => code === 0 ? resolve4() : reject(new Error(".app \uBCF5\uC0AC \uC2E4\uD328")));
+    });
+  } finally {
+    await new Promise((resolve4) => {
+      const child = (0, import_child_process4.spawn)("hdiutil", ["detach", mountPoint]);
+      child.on("close", () => resolve4());
+    });
+  }
+  await (0, import_promises2.rm)(downloadPath, { force: true });
+  return await createSymlink((0, import_path5.join)(CACHE_DIR, pkg.binPath));
+}
+async function installLinux(pkg, downloadPath) {
+  const extractDir = (0, import_path5.join)(CACHE_DIR, `extract-${Date.now()}`);
+  await (0, import_promises2.mkdir)(extractDir, { recursive: true });
+  await new Promise((resolve4, reject) => {
+    const child = (0, import_child_process4.spawn)("tar", ["xzf", downloadPath, "-C", extractDir]);
+    child.on("close", (code) => code === 0 ? resolve4() : reject(new Error("\uC555\uCD95 \uD574\uC81C \uC2E4\uD328")));
+  });
+  const debsDir = (0, import_path5.join)(extractDir, "DEBS");
+  try {
+    await (0, import_promises2.access)(debsDir);
+    const entries = await (await import("fs/promises")).readdir(debsDir);
+    for (const entry of entries) {
+      if (entry.endsWith(".deb")) {
+        await new Promise((resolve4, reject) => {
+          const child = (0, import_child_process4.spawn)("dpkg-deb", ["-x", (0, import_path5.join)(debsDir, entry), CACHE_DIR]);
+          child.on("close", (code) => code === 0 ? resolve4() : reject(new Error(`${entry} \uCD94\uCD9C \uC2E4\uD328`)));
+        });
+      }
+    }
+  } catch {
+  }
+  await (0, import_promises2.rm)(downloadPath, { force: true });
+  await (0, import_promises2.rm)(extractDir, { recursive: true, force: true });
+  return await createSymlink((0, import_path5.join)(CACHE_DIR, pkg.binPath));
+}
+async function installWindows(pkg, downloadPath) {
+  await new Promise((resolve4, reject) => {
+    const child = (0, import_child_process4.spawn)("msiexec", ["/a", downloadPath, "/qn", `TARGETDIR=${CACHE_DIR}`]);
+    child.on("close", (code) => code === 0 ? resolve4() : reject(new Error("MSI \uC124\uCE58 \uC2E4\uD328")));
+  });
+  await (0, import_promises2.rm)(downloadPath, { force: true });
+  return (0, import_path5.join)(CACHE_DIR, pkg.binPath);
+}
+async function createSymlink(actualBin) {
+  const binDir = (0, import_path5.join)(CACHE_DIR, "bin");
+  await (0, import_promises2.mkdir)(binDir, { recursive: true });
+  const linkBin = (0, import_path5.join)(binDir, "soffice");
+  try {
+    await (0, import_promises2.symlink)(actualBin, linkBin);
+  } catch {
+  }
+  process.env.PATH = `${binDir}${import_path5.delimiter}${process.env.PATH}`;
+  return linkBin;
+}
+async function installLibreOffice(onProgress) {
+  const platform = process.platform;
+  const pkg = PACKAGES[platform];
+  if (!pkg) {
+    throw new ConvertError(
+      "UNSUPPORTED_PLATFORM",
+      `${platform}\uC740 \uC790\uB3D9 \uC124\uCE58\uB97C \uC9C0\uC6D0\uD558\uC9C0 \uC54A\uC2B5\uB2C8\uB2E4. \uC218\uB3D9\uC73C\uB85C LibreOffice\uB97C \uC124\uCE58\uD574 \uC8FC\uC138\uC694.`
+    );
+  }
+  return await installForPlatform(pkg, onProgress);
+}
+async function resolveSoffice(emitter, autoInstall = true) {
+  emitter.validate("soffice_check", "LibreOffice \uAC00\uC6A9\uC131 \uD655\uC778 \uC911...");
+  const inPath = await findInPath();
+  if (inPath) {
+    emitter.validate("soffice_found", "\uC2DC\uC2A4\uD15C PATH\uC5D0\uC11C LibreOffice \uBC1C\uACAC", { sofficePath: inPath });
+    return inPath;
+  }
+  const inCache = await findInCache();
+  if (inCache) {
+    emitter.validate("soffice_found", "\uCE90\uC2DC\uB41C LibreOffice \uBC1C\uACAC", { sofficePath: inCache });
+    return inCache;
+  }
+  if (!autoInstall) {
+    emitter.error(
+      "validate",
+      "SOFFICE_NOT_FOUND",
+      "LibreOffice\uB97C \uCC3E\uC744 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4",
+      "\uC218\uB3D9\uC73C\uB85C \uC124\uCE58\uD558\uAC70\uB098 autoInstallLibreOffice: true \uC635\uC158\uC744 \uC0AC\uC6A9\uD558\uC138\uC694."
+    );
+    throw new ConvertError("SOFFICE_NOT_FOUND", "LibreOffice\uAC00 \uC124\uCE58\uB418\uC9C0 \uC54A\uC558\uC2B5\uB2C8\uB2E4");
+  }
+  emitter.install("install_start", "LibreOffice \uC790\uB3D9 \uC124\uCE58\uB97C \uC2DC\uC791\uD569\uB2C8\uB2E4...");
+  try {
+    const installed = await installLibreOffice((downloaded, total) => {
+      const percent = Math.round(downloaded / total * 100);
+      emitter.install("download_progress", `\uB2E4\uC6B4\uB85C\uB4DC \uC911... ${percent}%`, {
+        percent,
+        downloadedBytes: downloaded,
+        totalBytes: total
+      });
+    });
+    emitter.install("install_complete", "\uC124\uCE58 \uC644\uB8CC", { installedPath: installed });
+    return installed;
+  } catch (err) {
+    const errorMsg = err instanceof Error ? err.message : String(err);
+    emitter.install("install_failed", "\uC124\uCE58 \uC2E4\uD328", { error: errorMsg });
+    throw err;
+  }
+}
+
 // src/convert/libreoffice.ts
 var libreConvert = import_libreoffice_convert.default.convert;
 async function assertSofficeAvailable() {
@@ -11316,6 +11514,54 @@ async function convertBuffer(buffer, targetExt, timeoutMs = 6e4) {
   });
 }
 
+// src/convert/events.ts
+var ConvertEventEmitter = class {
+  listener = null;
+  /** 이벤트 리스너 등록 */
+  setListener(listener) {
+    this.listener = listener;
+  }
+  /** 이벤트 발송 */
+  emit(event) {
+    try {
+      this.listener?.(event);
+    } catch {
+    }
+  }
+  /** 타입 안전한 헬퍼: detect 이벤트 */
+  detect(stage, message, meta) {
+    this.emit({ type: "detect", stage, message, ...meta });
+  }
+  /** 타입 안전한 헬퍼: validate 이벤트 */
+  validate(stage, message, meta) {
+    this.emit({ type: "validate", stage, message, ...meta });
+  }
+  /** 타입 안전한 헬퍼: install 이벤트 */
+  install(stage, message, meta) {
+    this.emit({ type: "install", stage, message, ...meta });
+  }
+  /** 타입 안전한 헬퍼: convert 진행 이벤트 */
+  progress(percent, message) {
+    this.emit({ type: "convert", stage: "convert_progress", message, percent });
+  }
+  /** 타입 안전한 헬퍼: convert 시작 */
+  convertStart(message) {
+    this.emit({ type: "convert", stage: "convert_start", message, percent: 0 });
+  }
+  /** 타입 안전한 헬퍼: convert 완료 */
+  convertDone(message) {
+    this.emit({ type: "convert", stage: "convert_done", message, percent: 100 });
+  }
+  /** 타입 안전한 헬퍼: 완료 이벤트 */
+  complete(result) {
+    this.emit({ type: "complete", stage: "success", message: "\uBCC0\uD658 \uC644\uB8CC", result });
+  }
+  /** 타입 안전한 헬퍼: 에러 이벤트 */
+  error(stage, code, message, suggestion) {
+    this.emit({ type: "error", stage, code, message, recoverable: true, suggestion });
+  }
+};
+
 // src/convert/index.ts
 var isConverting = false;
 var queue = [];
@@ -11340,81 +11586,129 @@ async function acquireConvertLock() {
   });
 }
 async function convertToPdf(input, options) {
-  let buffer;
-  try {
-    if (typeof input === "string") {
-      buffer = await (0, import_promises2.readFile)(input);
-    } else if (Buffer.isBuffer(input)) {
-      buffer = input;
-    } else {
-      buffer = Buffer.from(input);
-    }
-  } catch (err) {
-    return {
-      success: false,
-      code: "PARSE_ERROR",
-      error: `\uC785\uB825 \uC77D\uAE30 \uC2E4\uD328: ${err instanceof Error ? err.message : String(err)}`,
-      stage: "detect"
-    };
-  }
-  const MAX_FILE_SIZE = 500 * 1024 * 1024;
-  if (buffer.length > MAX_FILE_SIZE) {
-    return {
-      success: false,
-      code: "FILE_TOO_LARGE",
-      error: `\uD30C\uC77C \uD06C\uAE30 \uCD08\uACFC: ${(buffer.length / 1024 / 1024).toFixed(1)}MB (\uCD5C\uB300 500MB)`,
-      stage: "detect"
-    };
+  const emitter = new ConvertEventEmitter();
+  if (options?.onEvent) {
+    emitter.setListener(options.onEvent);
   }
-  const format = detectFormat(toArrayBuffer(buffer));
-  if (format !== "hwp" && format !== "hwpx") {
-    return {
-      success: false,
-      code: "UNSUPPORTED_FORMAT",
-      error: `\uC9C0\uC6D0\uD558\uC9C0 \uC54A\uB294 \uD3EC\uB9F7\uC785\uB2C8\uB2E4: ${format}`,
-      stage: "detect"
-    };
+  if (options?.onProgress) {
+    const legacyProgress = options.onProgress;
+    emitter.setListener((event) => {
+      if (event.type === "convert" && event.stage === "convert_progress") {
+        legacyProgress(event.percent, event.message);
+      }
+    });
   }
   try {
-    await assertSofficeAvailable();
-  } catch (err) {
-    if (err instanceof ConvertError) {
+    emitter.detect("reading", "\uC785\uB825 \uD30C\uC77C \uC77D\uB294 \uC911...");
+    let buffer;
+    try {
+      if (typeof input === "string") {
+        buffer = await (0, import_promises3.readFile)(input);
+      } else if (Buffer.isBuffer(input)) {
+        buffer = input;
+      } else {
+        buffer = Buffer.from(input);
+      }
+    } catch (err) {
+      emitter.error(
+        "detect",
+        "PARSE_ERROR",
+        `\uC785\uB825 \uC77D\uAE30 \uC2E4\uD328: ${err instanceof Error ? err.message : String(err)}`
+      );
       return {
         success: false,
-        code: err.code,
-        error: err.message,
-        stage: "validate"
+        code: "PARSE_ERROR",
+        error: `\uC785\uB825 \uC77D\uAE30 \uC2E4\uD328: ${err instanceof Error ? err.message : String(err)}`,
+        stage: "detect"
       };
     }
-    throw err;
-  }
-  const releaseLock = await acquireConvertLock();
-  try {
-    options?.onProgress?.(10, "convert");
-    const pdf = await convertBuffer(buffer, ".pdf", options?.timeoutMs);
-    options?.onProgress?.(100, "done");
-    return {
-      success: true,
-      pdf: new Uint8Array(pdf),
-      sourceFormat: format
-    };
-  } catch (err) {
-    if (err instanceof ConvertError) {
+    const MAX_FILE_SIZE = 500 * 1024 * 1024;
+    if (buffer.length > MAX_FILE_SIZE) {
+      emitter.error(
+        "detect",
+        "FILE_TOO_LARGE",
+        `\uD30C\uC77C \uD06C\uAE30 \uCD08\uACFC: ${(buffer.length / 1024 / 1024).toFixed(1)}MB (\uCD5C\uB300 500MB)`
+      );
+      return {
+        success: false,
+        code: "FILE_TOO_LARGE",
+        error: `\uD30C\uC77C \uD06C\uAE30 \uCD08\uACFC: ${(buffer.length / 1024 / 1024).toFixed(1)}MB (\uCD5C\uB300 500MB)`,
+        stage: "detect"
+      };
+    }
+    const format = detectFormat(toArrayBuffer(buffer));
+    emitter.detect("format_detected", `\uD3EC\uB9F7 \uAC10\uC9C0 \uC644\uB8CC: ${format}`, { format });
+    if (format !== "hwp" && format !== "hwpx") {
+      emitter.error("detect", "UNSUPPORTED_FORMAT", `\uC9C0\uC6D0\uD558\uC9C0 \uC54A\uB294 \uD3EC\uB9F7\uC785\uB2C8\uB2E4: ${format}`);
       return {
         success: false,
-        code: err.code,
-        error: err.message,
+        code: "UNSUPPORTED_FORMAT",
+        error: `\uC9C0\uC6D0\uD558\uC9C0 \uC54A\uB294 \uD3EC\uB9F7\uC785\uB2C8\uB2E4: ${format}`,
+        stage: "detect"
+      };
+    }
+    emitter.validate("soffice_check", "LibreOffice \uAC00\uC6A9\uC131 \uD655\uC778 \uC911...");
+    let sofficePath;
+    try {
+      sofficePath = await resolveSoffice(emitter, options?.autoInstallLibreOffice ?? true);
+    } catch (err) {
+      if (err instanceof ConvertError) {
+        return {
+          success: false,
+          code: err.code,
+          error: err.message,
+          stage: "validate"
+        };
+      }
+      throw err;
+    }
+    const releaseLock = await acquireConvertLock();
+    try {
+      emitter.convertStart("\uBCC0\uD658 \uC2DC\uC791...");
+      emitter.progress(10, "\uBCC0\uD658 \uC911...");
+      const pdf = await convertBuffer(buffer, ".pdf", options?.timeoutMs);
+      emitter.progress(100, "\uBCC0\uD658 \uC644\uB8CC");
+      emitter.convertDone("\uBCC0\uD658 \uC644\uB8CC");
+      const result = {
+        success: true,
+        pdf: new Uint8Array(pdf),
+        sourceFormat: format
+      };
+      emitter.complete({
+        sourceFormat: format,
+        pdfSize: pdf.length
+      });
+      return result;
+    } catch (err) {
+      if (err instanceof ConvertError) {
+        emitter.error("convert", err.code, err.message);
+        return {
+          success: false,
+          code: err.code,
+          error: err.message,
+          stage: "convert"
+        };
+      }
+      const errorMsg = err instanceof Error ? err.message : "\uBCC0\uD658 \uC2E4\uD328";
+      emitter.error("convert", classifyError(err), errorMsg);
+      return {
+        success: false,
+        code: classifyError(err),
+        error: errorMsg,
         stage: "convert"
       };
+    } finally {
+      releaseLock();
     }
+  } catch (unexpectedErr) {
+    const errorMsg = unexpectedErr instanceof Error ? unexpectedErr.message : "\uC608\uC0C1\uCE58 \uBABB\uD55C \uC624\uB958";
+    emitter.error("convert", "PARSE_ERROR", errorMsg);
     return {
       success: false,
-      code: classifyError(err),
-      error: err instanceof Error ? err.message : "\uBCC0\uD658 \uC2E4\uD328",
+      code: "PARSE_ERROR",
+      error: errorMsg,
       stage: "convert"
     };
-  } finally {
-    releaseLock();
   }
 }
 async function convertHwpToPdf(input, options) {
@@ -11539,9 +11833,9 @@ var ApiKeyRotationPool = class _ApiKeyRotationPool {
 };
 
 // src/pipeline/unified-ocr.ts
-var import_promises3 = require("fs/promises");
-var import_path5 = require("path");
-var import_child_process4 = require("child_process");
+var import_promises4 = require("fs/promises");
+var import_path6 = require("path");
+var import_child_process5 = require("child_process");
 var import_node_perf_hooks = require("perf_hooks");
 init_logger();
 
@@ -11675,15 +11969,15 @@ function elapsedMs(startAt) {
   return Math.round(import_node_perf_hooks.performance.now() - startAt);
 }
 async function runUnifiedOcrPipeline(inputPath, options = {}) {
-  const absInput = (0, import_path5.resolve)(inputPath);
-  const stem = (0, import_path5.basename)(absInput, (0, import_path5.extname)(absInput));
-  const workspaceDir = (0, import_path5.resolve)(options.workspaceDir ?? (0, import_path5.join)((0, import_path5.dirname)(absInput), `${stem}_ocr_workspace`));
-  const imagesDir = (0, import_path5.join)(workspaceDir, "images");
-  const rawDir = (0, import_path5.join)(workspaceDir, "ocr", "raw");
-  const diffDir = (0, import_path5.join)(workspaceDir, "ocr", "diff");
-  const outputPath = (0, import_path5.resolve)(options.outputPath ?? (0, import_path5.join)((0, import_path5.dirname)(absInput), `${stem}.md`));
-  const reportPath = (0, import_path5.join)(workspaceDir, "run-report.json");
-  const modelCachePath = (0, import_path5.join)((0, import_path5.dirname)(absInput), ".kordoc-model-cache.json");
+  const absInput = (0, import_path6.resolve)(inputPath);
+  const stem = (0, import_path6.basename)(absInput, (0, import_path6.extname)(absInput));
+  const workspaceDir = (0, import_path6.resolve)(options.workspaceDir ?? (0, import_path6.join)((0, import_path6.dirname)(absInput), `${stem}_ocr_workspace`));
+  const imagesDir = (0, import_path6.join)(workspaceDir, "images");
+  const rawDir = (0, import_path6.join)(workspaceDir, "ocr", "raw");
+  const diffDir = (0, import_path6.join)(workspaceDir, "ocr", "diff");
+  const outputPath = (0, import_path6.resolve)(options.outputPath ?? (0, import_path6.join)((0, import_path6.dirname)(absInput), `${stem}.md`));
+  const reportPath = (0, import_path6.join)(workspaceDir, "run-report.json");
+  const modelCachePath = (0, import_path6.join)((0, import_path6.dirname)(absInput), ".kordoc-model-cache.json");
   const baseUrl = options.baseUrl ?? "https://integrate.api.nvidia.com/v1/chat/completions";
   const timeoutMs = options.timeoutMs ?? 6e4;
   const maxRetriesPerPage = options.maxRetriesPerPage ?? 5;
@@ -11697,9 +11991,9 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
   const keyPool = ApiKeyRotationPool.fromEnv();
   const runId = options.runId ?? generateRunId("ocr");
   const logger = (options.logger ?? createLoggerFromEnv()).withRun(runId).child({ component: "pipeline/unified-ocr.ts" });
-  await (0, import_promises3.mkdir)(imagesDir, { recursive: true });
-  await (0, import_promises3.mkdir)(rawDir, { recursive: true });
-  await (0, import_promises3.mkdir)(diffDir, { recursive: true });
+  await (0, import_promises4.mkdir)(imagesDir, { recursive: true });
+  await (0, import_promises4.mkdir)(rawDir, { recursive: true });
+  await (0, import_promises4.mkdir)(diffDir, { recursive: true });
   const timingsMs = {};
   const markStageStart = (stage, message) => emitProgress(options.onEvent, stage, 0, stageWeights, { message, type: "stage_start" });
   const markStageProgress = (stage, stagePercent, current, total, message, model) => emitProgress(options.onEvent, stage, stagePercent, stageWeights, { type: "stage_progress", current, total, message, model });
@@ -11715,12 +12009,12 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
     currentStage = "convert";
     markStageStart("convert", "\uBB38\uC11C\uB97C PDF\uB85C \uBCC0\uD658 \uC911");
     logStage("info", "convert", "start", "\uBB38\uC11C\uB97C PDF\uB85C \uBCC0\uD658 \uC2DC\uC791", { input: absInput });
-    if ((0, import_path5.extname)(absInput).toLowerCase() !== ".pdf") {
+    if ((0, import_path6.extname)(absInput).toLowerCase() !== ".pdf") {
       await assertSofficeAvailable();
-      workingPdfPath = (0, import_path5.join)(workspaceDir, `${stem}.pdf`);
-      const inputBuffer = await (0, import_promises3.readFile)(absInput);
+      workingPdfPath = (0, import_path6.join)(workspaceDir, `${stem}.pdf`);
+      const inputBuffer = await (0, import_promises4.readFile)(absInput);
       const out = await convertBuffer(inputBuffer, ".pdf");
-      await (0, import_promises3.writeFile)(workingPdfPath, out);
+      await (0, import_promises4.writeFile)(workingPdfPath, out);
     }
     timingsMs.convert = elapsedMs(convertStart);
     markStageDone("convert", "PDF \uBCC0\uD658 \uC644\uB8CC");
@@ -11731,10 +12025,10 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
     if (totalPages === 0) throw new UnifiedOcrError("RENDER_FAILED", "render", "\uD398\uC774\uC9C0 \uC218\uB97C \uD655\uC778\uD560 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4.");
     markStageStart("render", "PDF \uD398\uC774\uC9C0 \uC774\uBBF8\uC9C0 \uC0DD\uC131 \uC911");
     logStage("info", "render", "start", "PDF \uD398\uC774\uC9C0 \uB80C\uB354\uB9C1 \uC2DC\uC791", { pdf: workingPdfPath, dpi, totalPages });
-    await runCommand("pdftoppm", ["-png", "-r", String(dpi), "-f", "1", "-l", "1", workingPdfPath, (0, import_path5.join)(imagesDir, "page")]);
-    const firstFiles = (await (0, import_promises3.readdir)(imagesDir)).filter((f) => f.endsWith(".png")).sort((a, b) => naturalPageSort(a, b));
+    await runCommand("pdftoppm", ["-png", "-r", String(dpi), "-f", "1", "-l", "1", workingPdfPath, (0, import_path6.join)(imagesDir, "page")]);
+    const firstFiles = (await (0, import_promises4.readdir)(imagesDir)).filter((f) => f.endsWith(".png")).sort((a, b) => naturalPageSort(a, b));
     if (firstFiles.length === 0) throw new UnifiedOcrError("RENDER_FAILED", "render", "\uCCAB \uD398\uC774\uC9C0 \uC774\uBBF8\uC9C0 \uC0DD\uC131 \uC2E4\uD328");
-    const probeImage = (0, import_path5.join)(imagesDir, firstFiles[0]);
+    const probeImage = (0, import_path6.join)(imagesDir, firstFiles[0]);
     markStageProgress("render", Math.round(1 / totalPages * 100), 1, totalPages, `\uD398\uC774\uC9C0 1/${totalPages} \uB80C\uB354\uB9C1`);
     const probeStart = import_node_perf_hooks.performance.now();
     currentStage = "probe";
@@ -11780,7 +12074,7 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
       try {
         await queue2.enqueue({ pageNumber: 1, imagePath: probeImage });
         if (totalPages > 1) {
-          for await (const item of renderPdfToPngStream(workingPdfPath, (0, import_path5.join)(imagesDir, "page"), dpi, totalPages, 2)) {
+          for await (const item of renderPdfToPngStream(workingPdfPath, (0, import_path6.join)(imagesDir, "page"), dpi, totalPages, 2)) {
             await queue2.enqueue(item);
             renderDone++;
             markStageProgress("render", Math.round(renderDone / totalPages * 100), renderDone, totalPages, `\uD398\uC774\uC9C0 ${renderDone}/${totalPages} \uB80C\uB354\uB9C1`);
@@ -11830,8 +12124,8 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
     const sortedEntries = Array.from(pageResultsMap.entries()).sort((a, b) => a[0] - b[0]);
     const rawPagePaths = [];
     for (const [pageNum, markdown] of sortedEntries) {
-      const pagePath = (0, import_path5.join)(rawDir, `page_${String(pageNum).padStart(4, "0")}.md`);
-      await (0, import_promises3.writeFile)(pagePath, markdown, "utf-8");
+      const pagePath = (0, import_path6.join)(rawDir, `page_${String(pageNum).padStart(4, "0")}.md`);
+      await (0, import_promises4.writeFile)(pagePath, markdown, "utf-8");
       rawPagePaths.push(pagePath);
     }
     const mergeStart = import_node_perf_hooks.performance.now();
@@ -11839,7 +12133,7 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
     markStageStart("merge", "\uCD5C\uC885 Markdown \uBCD1\uD569 \uC911");
     logStage("info", "merge", "start", "\uCD5C\uC885 \uBCD1\uD569 \uC2DC\uC791", { pages: rawPagePaths.length });
     const merged = await mergeMarkdownPages(rawPagePaths);
-    await (0, import_promises3.writeFile)(outputPath, merged, "utf-8");
+    await (0, import_promises4.writeFile)(outputPath, merged, "utf-8");
     timingsMs.merge = elapsedMs(mergeStart);
     markStageDone("merge", "\uBCD1\uD569 \uC644\uB8CC");
     logStage("info", "merge", "done", "\uCD5C\uC885 \uBCD1\uD569 \uC644\uB8CC", { outputPath, elapsedMs: timingsMs.merge });
@@ -11855,7 +12149,7 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
       timingsMs,
       modelCachePath
     };
-    await (0, import_promises3.writeFile)(reportPath, JSON.stringify(report, null, 2), "utf-8");
+    await (0, import_promises4.writeFile)(reportPath, JSON.stringify(report, null, 2), "utf-8");
     logStage("info", "finalize", "done", "run-report \uC800\uC7A5 \uC644\uB8CC", { reportPath });
     return { outputPath, reportPath, selectedModel };
   } catch (err) {
@@ -11930,7 +12224,7 @@ async function getPdfPageCount(pdfPath) {
   return n;
 }
 async function* renderPdfToPngStream(pdfPath, prefixPath, dpi, totalPages, startPage = 1) {
-  const imagesDir = (0, import_path5.dirname)(prefixPath);
+  const imagesDir = (0, import_path6.dirname)(prefixPath);
   for (let page = startPage; page <= totalPages; page++) {
     try {
       await runCommand("pdftoppm", [
@@ -11944,9 +12238,9 @@ async function* renderPdfToPngStream(pdfPath, prefixPath, dpi, totalPages, start
         pdfPath,
         prefixPath
       ]);
-      const files = await (0, import_promises3.readdir)(imagesDir);
+      const files = await (0, import_promises4.readdir)(imagesDir);
       const pageFiles = files.filter((f) => f.endsWith(".png")).sort((a, b) => naturalPageSort(a, b));
-      const imagePath = (0, import_path5.join)(imagesDir, pageFiles[pageFiles.length - 1]);
+      const imagePath = (0, import_path6.join)(imagesDir, pageFiles[pageFiles.length - 1]);
       yield { pageNumber: page, imagePath };
     } catch (err) {
       yield {
@@ -11959,7 +12253,7 @@ async function* renderPdfToPngStream(pdfPath, prefixPath, dpi, totalPages, start
 }
 async function runCommand(cmd, args) {
   await new Promise((resolvePromise, reject) => {
-    const child = (0, import_child_process4.spawn)(cmd, args, { stdio: "pipe" });
+    const child = (0, import_child_process5.spawn)(cmd, args, { stdio: "pipe" });
     let stderr = "";
     child.stderr.on("data", (d) => {
       stderr += String(d);
@@ -11973,7 +12267,7 @@ async function runCommand(cmd, args) {
 }
 async function runCommandWithStdout(cmd, args) {
   return await new Promise((resolvePromise, reject) => {
-    const child = (0, import_child_process4.spawn)(cmd, args, { stdio: "pipe" });
+    const child = (0, import_child_process5.spawn)(cmd, args, { stdio: "pipe" });
     let stdout = "";
     let stderr = "";
     child.stdout.on("data", (d) => {
@@ -12062,7 +12356,7 @@ function startParallelProbeRuns(input) {
 }
 async function loadModelCache(path) {
   try {
-    const raw = await (0, import_promises3.readFile)(path, "utf-8");
+    const raw = await (0, import_promises4.readFile)(path, "utf-8");
     return JSON.parse(raw);
   } catch {
     return null;
@@ -12093,7 +12387,7 @@ async function updateModelCache(path, probes) {
     }
   }
   current.updatedAt = (/* @__PURE__ */ new Date()).toISOString();
-  await (0, import_promises3.writeFile)(path, JSON.stringify(current, null, 2), "utf-8");
+  await (0, import_promises4.writeFile)(path, JSON.stringify(current, null, 2), "utf-8");
 }
 async function ocrWorkerPool(input) {
   const { queue: queue2, workerCount, ocrInput, onPageDone } = input;
@@ -12153,7 +12447,7 @@ async function ocrImageWithFallback(input) {
 async function mergeMarkdownPages(paths) {
   const out = [];
   for (let i = 0; i < paths.length; i++) {
-    const txt = (await (0, import_promises3.readFile)(paths[i], "utf-8")).trim();
+    const txt = (await (0, import_promises4.readFile)(paths[i], "utf-8")).trim();
     if (!txt) continue;
     out.push(txt);
   }
@@ -12269,7 +12563,7 @@ async function ocrImageViaNim(input) {
   throw new UnifiedOcrError("OCR_FAILED", "ocr", `OCR \uC7AC\uC2DC\uB3C4 \uCD08\uACFC: ${lastErr}`);
 }
 async function encodeBase64(path) {
-  const b = await (0, import_promises3.readFile)(path);
+  const b = await (0, import_promises4.readFile)(path);
   return b.toString("base64");
 }
 function stripCodeFence3(text) {
@@ -12281,7 +12575,7 @@ async function delay(ms) {
   await new Promise((resolvePromise) => setTimeout(resolvePromise, ms));
 }
 function ensureSupportedInput(path) {
-  const ext = (0, import_path5.extname)(path).toLowerCase();
+  const ext = (0, import_path6.extname)(path).toLowerCase();
   const allowed = /* @__PURE__ */ new Set([".pdf", ".hwp", ".hwpx", ".docx", ".xlsx"]);
   if (!allowed.has(ext)) {
     throw new UnifiedOcrError("UNSUPPORTED_INPUT", "convert", `\uC9C0\uC6D0\uD558\uC9C0 \uC54A\uB294 \uC785\uB825 \uD3EC\uB9F7: ${ext}`);
@@ -12308,7 +12602,7 @@ async function parse2(input, options) {
   let buffer;
   if (typeof input === "string") {
     try {
-      const buf = await (0, import_promises4.readFile)(input);
+      const buf = await (0, import_promises5.readFile)(input);
       buffer = toArrayBuffer(buf);
     } catch (err) {
       const msg = err instanceof Error && "code" in err && err.code === "ENOENT" ? `\uD30C\uC77C\uC744 \uCC3E\uC744 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4: ${input}` : `\uD30C\uC77C \uC77D\uAE30 \uC2E4\uD328: ${input}`;