markpdfdown 0.1.7 → 0.1.8

package/dist/main/index.js

@@ -516,6 +516,69 @@ class ImagePathUtil {
     this.uploadsDir = null;
   }
 }
+class FileWaitUtil {
+  static MAX_ATTEMPTS = 5;
+  static DELAY_MS = 1e3;
+  /**
+   * Wait for a file to become available and non-empty.
+   *
+   * @param filePath - Full path to the file
+   * @param uploadsDir - Base uploads directory for diagnostic logging
+   * @param taskId - Task ID for diagnostic logging
+   * @param filename - Original filename for error messages
+   * @param label - Log label (e.g., 'PDFSplitter', 'ImageSplitter')
+   * @throws Error if the file is not found after all retries
+   */
+  static async waitForFile(filePath, uploadsDir, taskId, filename, label) {
+    for (let attempt = 1; attempt <= this.MAX_ATTEMPTS; attempt++) {
+      try {
+        await promises.access(filePath);
+        const stats = await promises.stat(filePath);
+        if (stats.size > 0) {
+          if (attempt > 1) {
+            console.log(
+              `[${label}] File became available on attempt ${attempt}: ${filePath}`
+            );
+          }
+          return;
+        }
+        console.warn(
+          `[${label}] File exists but is empty (attempt ${attempt}/${this.MAX_ATTEMPTS}): ${filePath}`
+        );
+      } catch {
+        console.warn(
+          `[${label}] File not accessible (attempt ${attempt}/${this.MAX_ATTEMPTS}): ${filePath}`
+        );
+      }
+      if (attempt < this.MAX_ATTEMPTS) {
+        await new Promise((resolve) => setTimeout(resolve, this.DELAY_MS));
+      }
+    }
+    await this.logDiagnostics(uploadsDir, taskId, filename, label);
+    throw new Error(
+      `${label === "PDFSplitter" ? "PDF" : "Image"} file not found: ${filename}. The file may have been moved or deleted.`
+    );
+  }
+  /**
+   * Log diagnostic information about the task directory contents.
+   */
+  static async logDiagnostics(uploadsDir, taskId, filename, label) {
+    const taskDir = path.join(uploadsDir, taskId);
+    try {
+      const dirExists = await promises.stat(taskDir).then(() => true).catch(() => false);
+      if (dirExists) {
+        const files = await promises.readdir(taskDir);
+        console.error(
+          `[${label}] Task directory exists but target file not found. Dir: ${taskDir}, Files in dir: [${files.join(", ")}], Expected: ${filename}`
+        );
+      } else {
+        console.error(`[${label}] Task directory does not exist: ${taskDir}`);
+      }
+    } catch (diagError) {
+      console.error(`[${label}] Diagnostic check failed:`, diagError);
+    }
+  }
+}
 class PageRangeParser {
   /**
    * Regular expression for validating page range format.
@@ -684,6 +747,7 @@ class PDFSplitter {
     const taskId = task.id;
     const filename = task.filename;
     const sourcePath = path.join(this.uploadsDir, taskId, filename);
+    await FileWaitUtil.waitForFile(sourcePath, this.uploadsDir, taskId, filename, "PDFSplitter");
     try {
       const totalPages = await this.getPDFPageCountWithRetry(sourcePath);
       const pageNumbers = PageRangeParser.parse(task.page_range, totalPages);
@@ -855,8 +919,8 @@ class ImageSplitter {
     const taskId = task.id;
     const filename = task.filename;
     const sourcePath = path.join(this.uploadsDir, taskId, filename);
+    await FileWaitUtil.waitForFile(sourcePath, this.uploadsDir, taskId, filename, "ImageSplitter");
     try {
-      await promises.access(sourcePath);
       const ext = path.extname(filename).toLowerCase();
       if (!ext) {
         throw new Error(`Image file has no extension: ${filename}`);
@@ -1119,6 +1183,22 @@ const findAll$2 = async () => {
     orderBy: [{ createdAt: "desc" }]
   });
 };
+const findAllIncludeDisabled = async () => {
+  return await prisma.provider.findMany({
+    select: {
+      id: true,
+      name: true,
+      type: true,
+      api_key: true,
+      base_url: true,
+      suffix: true,
+      status: true,
+      createdAt: true,
+      updatedAt: true
+    },
+    orderBy: [{ createdAt: "desc" }]
+  });
+};
 const findById$1 = async (id) => {
   return await prisma.provider.findUnique({
     where: { id },
@@ -1169,6 +1249,7 @@ const updateStatus = async (id, status) => {
 };
 const providerRepository = {
   findAll: findAll$2,
+  findAllIncludeDisabled,
   findById: findById$1,
   create: create$2,
   update: update$1,
@@ -1411,25 +1492,31 @@ class ConverterWorker extends WorkerBase {
   /**
    * Claim a PENDING page for processing using optimistic locking.
    *
-   * Query conditions:
-   * - task.status = PROCESSING
-   * - task.status != CANCELLED
-   * - page.status = PENDING
-   * - page.worker_id = null
+   * Strategy: First find tasks in valid states (PROCESSING/COMPLETED),
+   * then find PENDING pages only within those tasks. This avoids the problem
+   * where orphaned PENDING pages from terminal tasks exhaust claim attempts.
    *
    * Order: retry_count ASC, page ASC (prioritize fresh pages)
    */
   async claimPage() {
     const maxAttempts = 5;
-    const checkedTaskIds = [];
+    const activeTasks = await prisma.task.findMany({
+      where: {
+        status: { in: [TaskStatus.PROCESSING, TaskStatus.COMPLETED] }
+      },
+      select: { id: true }
+    });
+    if (activeTasks.length === 0) {
+      return null;
+    }
+    const activeTaskIds = activeTasks.map((t) => t.id);
     for (let attempt = 0; attempt < maxAttempts; attempt++) {
       try {
         const candidate = await prisma.taskDetail.findFirst({
           where: {
             status: PageStatus.PENDING,
             worker_id: null,
-            // Exclude pages from tasks we've already checked and found not in PROCESSING state
-            ...checkedTaskIds.length > 0 && { task: { notIn: checkedTaskIds } }
+            task: { in: activeTaskIds }
           },
           orderBy: [
             { retry_count: "asc" },
@@ -1439,14 +1526,6 @@ class ConverterWorker extends WorkerBase {
         if (!candidate) {
           return null;
         }
-        const task = await prisma.task.findUnique({
-          where: { id: candidate.task },
-          select: { status: true }
-        });
-        if (!task || task.status !== TaskStatus.PROCESSING && task.status !== TaskStatus.COMPLETED) {
-          checkedTaskIds.push(candidate.task);
-          continue;
-        }
         const result = await prisma.taskDetail.updateMany({
           where: {
             id: candidate.id,
@@ -2303,11 +2382,43 @@ class WorkerOrchestrator {
       if (orphanedMergingTasks.count > 0) {
         console.log(`[WorkerOrchestrator] Reset ${orphanedMergingTasks.count} orphaned MERGING tasks to READY_TO_MERGE`);
       }
+      const terminalTaskStatuses = [
+        TaskStatus.CREATED,
+        TaskStatus.FAILED,
+        TaskStatus.CANCELLED,
+        TaskStatus.COMPLETED,
+        TaskStatus.PARTIAL_FAILED
+      ];
+      const terminalTasks = await prisma.task.findMany({
+        where: {
+          status: { in: terminalTaskStatuses }
+        },
+        select: { id: true }
+      });
+      let orphanedPendingPages = 0;
+      if (terminalTasks.length > 0) {
+        const terminalTaskIds = terminalTasks.map((t) => t.id);
+        const result2 = await prisma.taskDetail.updateMany({
+          where: {
+            task: { in: terminalTaskIds },
+            status: PageStatus.PENDING
+          },
+          data: {
+            status: PageStatus.FAILED,
+            error: "Orphaned: parent task no longer active"
+          }
+        });
+        orphanedPendingPages = result2.count;
+        if (orphanedPendingPages > 0) {
+          console.log(`[WorkerOrchestrator] Marked ${orphanedPendingPages} orphaned PENDING pages as FAILED (parent task in terminal state)`);
+        }
+      }
       const result = {
         orphanedPages: orphanedPages.count,
         orphanedSplittingTasks: orphanedSplittingTasks.count,
         orphanedMergingTasks: orphanedMergingTasks.count,
-        total: orphanedPages.count + orphanedSplittingTasks.count + orphanedMergingTasks.count
+        orphanedPendingPages,
+        total: orphanedPages.count + orphanedSplittingTasks.count + orphanedMergingTasks.count + orphanedPendingPages
       };
       if (result.total === 0) {
         console.log("[WorkerOrchestrator] No orphaned work found");
@@ -2321,6 +2432,7 @@ class WorkerOrchestrator {
         orphanedPages: 0,
         orphanedSplittingTasks: 0,
         orphanedMergingTasks: 0,
+        orphanedPendingPages: 0,
         total: 0
       };
     }
@@ -2367,7 +2479,6 @@ const IPC_CHANNELS = {
     DOWNLOAD_MARKDOWN: "file:downloadMarkdown",
     SELECT_DIALOG: "file:selectDialog",
     UPLOAD: "file:upload",
-    UPLOAD_MULTIPLE: "file:uploadMultiple",
     UPLOAD_FILE_CONTENT: "file:uploadFileContent"
   },
   // Completion (LLM) channels
@@ -2379,7 +2490,7 @@ const IPC_CHANNELS = {
 function registerProviderHandlers() {
   ipcMain.handle(IPC_CHANNELS.PROVIDER.GET_ALL, async () => {
     try {
-      const providers = await providerRepository.findAll();
+      const providers = await providerRepository.findAllIncludeDisabled();
       return { success: true, data: providers };
     } catch (error) {
       console.error("[IPC] provider:getAll error:", error);
@@ -2597,7 +2708,7 @@ const findById = async (id) => {
 const create = async (task) => {
   return await prisma.task.create({
     data: {
-      id: v4(),
+      id: task?.id || v4(),
       filename: task?.filename || "",
       type: task?.type || "",
       page_range: task?.page_range || "",
@@ -2605,8 +2716,8 @@ const create = async (task) => {
       provider: task?.provider || 0,
       model: task?.model || "",
       model_name: task?.model_name || "",
-      progress: 0,
-      status: 0
+      progress: task?.progress ?? 0,
+      status: task?.status ?? 0
     }
   });
 };
@@ -3132,6 +3243,7 @@ function registerFileHandlers() {
           return { success: false, error: "Task ID and file path are required" };
         }
         if (!fs.existsSync(filePath)) {
+          console.error(`[IPC] file:upload - Source file does not exist: ${filePath}`);
           return { success: false, error: "File does not exist" };
         }
         const baseUploadDir = fileLogic.getUploadDir();
@@ -3139,10 +3251,26 @@ function registerFileHandlers() {
         if (!fs.existsSync(uploadDir)) {
           fs.mkdirSync(uploadDir, { recursive: true });
         }
+        try {
+          fs.accessSync(uploadDir, fs.constants.W_OK);
+        } catch {
+          console.error(`[IPC] file:upload - Upload directory is not writable: ${uploadDir}`);
+          return { success: false, error: `Upload directory is not writable: ${uploadDir}` };
+        }
         const fileName = path.basename(filePath);
         const destPath = path.join(uploadDir, fileName);
+        console.log(`[IPC] file:upload - Copying file: ${filePath} -> ${destPath}`);
         fs.copyFileSync(filePath, destPath);
+        if (!fs.existsSync(destPath)) {
+          console.error(`[IPC] file:upload - File copy verification failed, destination not found: ${destPath}`);
+          return { success: false, error: "File copy failed: destination file not found after copy" };
+        }
         const stats = fs.statSync(destPath);
+        if (stats.size === 0) {
+          console.error(`[IPC] file:upload - Copied file is empty: ${destPath}`);
+          return { success: false, error: "File copy failed: destination file is empty" };
+        }
+        console.log(`[IPC] file:upload - File copied successfully: ${destPath} (${stats.size} bytes)`);
         const fileInfo = {
           originalName: fileName,
           savedName: fileName,
@@ -3157,45 +3285,6 @@ function registerFileHandlers() {
       }
     }
   );
-  ipcMain.handle(
-    IPC_CHANNELS.FILE.UPLOAD_MULTIPLE,
-    async (_, taskId, filePaths) => {
-      try {
-        if (!taskId || !Array.isArray(filePaths) || filePaths.length === 0) {
-          return { success: false, error: "Task ID and file path list are required" };
-        }
-        const uploadResults = [];
-        for (const filePath of filePaths) {
-          if (!fs.existsSync(filePath)) {
-            continue;
-          }
-          const baseUploadDir = fileLogic.getUploadDir();
-          const uploadDir = path.join(baseUploadDir, taskId);
-          if (!fs.existsSync(uploadDir)) {
-            fs.mkdirSync(uploadDir, { recursive: true });
-          }
-          const fileName = path.basename(filePath);
-          const destPath = path.join(uploadDir, fileName);
-          fs.copyFileSync(filePath, destPath);
-          const stats = fs.statSync(destPath);
-          uploadResults.push({
-            originalName: fileName,
-            savedName: fileName,
-            path: destPath,
-            size: stats.size,
-            taskId
-          });
-        }
-        return {
-          success: true,
-          data: { message: "Files uploaded successfully", files: uploadResults }
-        };
-      } catch (error) {
-        console.error("[IPC] file:uploadMultiple error:", error);
-        return { success: false, error: error.message };
-      }
-    }
-  );
   ipcMain.handle(
     IPC_CHANNELS.FILE.UPLOAD_FILE_CONTENT,
     async (_, taskId, fileName, fileBuffer) => {
@@ -3208,13 +3297,30 @@ function registerFileHandlers() {
         if (!fs.existsSync(uploadDir)) {
           fs.mkdirSync(uploadDir, { recursive: true });
         }
-        const destPath = path.join(uploadDir, fileName);
+        try {
+          fs.accessSync(uploadDir, fs.constants.W_OK);
+        } catch {
+          console.error(`[IPC] file:uploadFileContent - Upload directory is not writable: ${uploadDir}`);
+          return { success: false, error: `Upload directory is not writable: ${uploadDir}` };
+        }
+        const safeName = path.basename(fileName);
+        const destPath = path.join(uploadDir, safeName);
         const buffer = Buffer.from(fileBuffer);
+        console.log(`[IPC] file:uploadFileContent - Writing file: ${destPath} (${buffer.length} bytes)`);
         fs.writeFileSync(destPath, buffer);
+        if (!fs.existsSync(destPath)) {
+          console.error(`[IPC] file:uploadFileContent - File write verification failed, not found: ${destPath}`);
+          return { success: false, error: "File write failed: file not found after write" };
+        }
         const stats = fs.statSync(destPath);
+        if (stats.size === 0) {
+          console.error(`[IPC] file:uploadFileContent - Written file is empty: ${destPath}`);
+          return { success: false, error: "File write failed: file is empty" };
+        }
+        console.log(`[IPC] file:uploadFileContent - File written successfully: ${destPath} (${stats.size} bytes)`);
         const fileInfo = {
-          originalName: fileName,
-          savedName: fileName,
+          originalName: safeName,
+          savedName: safeName,
           path: destPath,
           size: stats.size,
           taskId
@@ -3234,7 +3340,10 @@ function registerCompletionHandlers() {
     async (_, providerId, modelId, url) => {
       try {
         if (!providerId || !modelId || !url) {
-          return { success: false, error: "providerId, modelId, and url are required" };
+          return {
+            success: false,
+            error: "providerId, modelId, and url are required"
+          };
         }
         const result = await modelLogic.completion(providerId, {
           model: modelId,
@@ -3266,11 +3375,15 @@ function registerCompletionHandlers() {
     async (_, providerId, modelId) => {
       try {
         if (!providerId || !modelId) {
-          return { success: false, error: "providerId and modelId are required" };
+          return {
+            success: false,
+            error: "providerId and modelId are required"
+          };
         }
-        const testImageBase64 = "iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAYAAAAfFcSJAAAADUlEQVR42mNk+M9QDwADhgGAWjR9awAAAABJRU5ErkJggg==";
+        const testImageBase64 = "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";
         const result = await modelLogic.completion(providerId, {
           model: modelId,
+          maxTokens: 8,
           messages: [
             {
               role: "user",
@@ -3283,7 +3396,7 @@ function registerCompletionHandlers() {
                 },
                 {
                   type: "text",
-                  text: "Test connection."
+                  text: "Please identify the largest letter in the image."
                 }
               ]
             }

package/dist/preload/index.js

@@ -37,7 +37,6 @@ electron.contextBridge.exposeInMainWorld("api", {
   file: {
     selectDialog: () => electron.ipcRenderer.invoke("file:selectDialog"),
     upload: (taskId, filePath) => electron.ipcRenderer.invoke("file:upload", taskId, filePath),
-    uploadMultiple: (taskId, filePaths) => electron.ipcRenderer.invoke("file:uploadMultiple", taskId, filePaths),
     uploadFileContent: (taskId, fileName, fileBuffer) => electron.ipcRenderer.invoke("file:uploadFileContent", taskId, fileName, fileBuffer),
     getImagePath: (taskId, page) => electron.ipcRenderer.invoke("file:getImagePath", taskId, page),
     downloadMarkdown: (taskId) => electron.ipcRenderer.invoke("file:downloadMarkdown", taskId)