feat: implement project scanning timeout and enhance metadata extraction

- Added scanProjectWithTimeout method to ProjectScanner to prevent long scans from blocking the entire batch.
- Introduced cleanup logic in metadata extraction functions to ensure proper resource management during timeouts.
- Updated logging to provide insights on scan durations and session statistics for better performance monitoring.
- Enhanced repository group fetching logic in DashboardView and DateGroupedSessions components to handle loading and error states more effectively.

src/main/services/discovery/ProjectScanner.ts

@@ -168,7 +168,7 @@ export class ProjectScanner {
       const projectArrays = await this.collectFulfilledInBatches(
         projectDirs,
         this.fsProvider.type === 'ssh' ? 8 : ProjectScanner.LOCAL_PROJECT_BATCH,
-        async (dir) => this.scanProject(dir.name)
+        async (dir) => this.scanProjectWithTimeout(dir.name)
       );
 
       // Flatten and sort by most recent
@@ -312,6 +312,31 @@ export class ProjectScanner {
   // Project Scanning (continued)
   // ===========================================================================
 
+  // Per-project scan timeout: prevents a single slow directory from blocking
+  // the entire scan batch (e.g. a project with 1000+ session files on slow I/O).
+  private static readonly SCAN_PROJECT_TIMEOUT_MS = 15_000;
+
+  /**
+   * Scans a single project directory with a timeout guard.
+   * Returns empty array if the scan exceeds the timeout.
+   */
+  private async scanProjectWithTimeout(encodedName: string): Promise<Project[]> {
+    let timer: ReturnType<typeof setTimeout> | null = null;
+    const timeout = new Promise<Project[]>((resolve) => {
+      timer = setTimeout(() => {
+        logger.warn(
+          `[scanProject] timeout after ${ProjectScanner.SCAN_PROJECT_TIMEOUT_MS}ms project=${encodedName}`
+        );
+        resolve([]);
+      }, ProjectScanner.SCAN_PROJECT_TIMEOUT_MS);
+    });
+    try {
+      return await Promise.race([this.scanProject(encodedName), timeout]);
+    } finally {
+      if (timer) clearTimeout(timer);
+    }
+  }
+
   /**
    * Scans a single project directory and returns project metadata.
    * If sessions have different cwd values, splits into multiple projects.
@@ -319,7 +344,9 @@ export class ProjectScanner {
   private async scanProject(encodedName: string): Promise<Project[]> {
     try {
       const projectPath = path.join(this.projectsDir, encodedName);
+      const readdirStart = Date.now();
       const entries = await this.fsProvider.readdir(projectPath);
+      const readdirMs = Date.now() - readdirStart;
 
       // Get session files (.jsonl at root level)
       const sessionFiles = entries.filter(
@@ -330,6 +357,12 @@ export class ProjectScanner {
         return [];
       }
 
+      if (sessionFiles.length > 200 || readdirMs > 500) {
+        logger.debug(
+          `[scanProject] ${encodedName} readdir=${readdirMs}ms entries=${entries.length} jsonl=${sessionFiles.length}`
+        );
+      }
+
       // Collect file stats and cwd for each session
       interface SessionInfo {
         sessionId: string;
@@ -346,6 +379,8 @@ export class ProjectScanner {
       const MAX_CWD_SPLIT_FILES = 80;
       const shouldSplitByCwd =
         this.fsProvider.type !== 'ssh' && sessionFiles.length <= MAX_CWD_SPLIT_FILES;
+
+      const sessionStatStart = Date.now();
       const sessionInfos = await this.collectFulfilledInBatches(
         sessionFiles,
         this.fsProvider.type === 'ssh' ? 32 : ProjectScanner.LOCAL_SESSION_BATCH,
@@ -377,6 +412,13 @@ export class ProjectScanner {
         return [];
       }
 
+      const sessionStatMs = Date.now() - sessionStatStart;
+      if (sessionFiles.length > 200 || sessionStatMs > 1000) {
+        logger.debug(
+          `[scanProject] ${encodedName} sessionStat=${sessionStatMs}ms files=${sessionFiles.length} infos=${sessionInfos.length}`
+        );
+      }
+
       // Group sessions by cwd
       const cwdGroups = new Map<string, SessionInfo[]>();
       const firstCwd = sessionInfos.find((s) => s.cwd)?.cwd ?? undefined;

src/main/utils/metadataExtraction.ts

@@ -51,23 +51,37 @@ export async function extractCwd(
   }
 
   const fileStream = fsProvider.createReadStream(filePath, { encoding: 'utf8' });
-  let bytes = 0;
-  let timedOut = false;
-  const timer = setTimeout(() => {
-    timedOut = true;
-    fileStream.destroy();
-  }, JSONL_HEAD_TIMEOUT_MS);
-  fileStream.on('data', (chunk: string) => {
-    bytes += byteLen(chunk);
-    if (bytes > JSONL_HEAD_MAX_BYTES) {
-      fileStream.destroy();
-    }
-  });
   const rl = readline.createInterface({
     input: fileStream,
     crlfDelay: Infinity,
   });
 
+  let bytes = 0;
+  let timedOut = false;
+  let cleaned = false;
+
+  // Close readline FIRST so `for await` exits, then destroy the stream.
+  // Calling only stream.destroy() can leave readline hanging when it has
+  // a partial line buffered (e.g. a 400KB+ JSONL line read in 64KB chunks).
+  const cleanup = (): void => {
+    if (cleaned) return;
+    cleaned = true;
+    rl.close();
+    fileStream.destroy();
+  };
+
+  const timer = setTimeout(() => {
+    timedOut = true;
+    cleanup();
+  }, JSONL_HEAD_TIMEOUT_MS);
+
+  fileStream.on('data', (chunk: string) => {
+    bytes += byteLen(chunk);
+    if (bytes > JSONL_HEAD_MAX_BYTES) {
+      cleanup();
+    }
+  });
+
   try {
     let lines = 0;
     for await (const line of rl) {
@@ -84,8 +98,6 @@ export async function extractCwd(
       }
       // Only conversational entries have cwd
       if ('cwd' in entry && entry.cwd) {
-        rl.close();
-        fileStream.destroy();
         return entry.cwd;
       }
     }
@@ -95,8 +107,7 @@ export async function extractCwd(
     }
   } finally {
     clearTimeout(timer);
-    rl.close();
+    cleanup();
-    fileStream.destroy();
   }
 
   return null;
@@ -122,23 +133,34 @@ export async function extractFirstUserMessagePreview(
   }
 
   const fileStream = fsProvider.createReadStream(filePath, { encoding: 'utf8' });
-  let bytes = 0;
-  let timedOut = false;
-  const timer = setTimeout(() => {
-    timedOut = true;
-    fileStream.destroy();
-  }, JSONL_HEAD_TIMEOUT_MS);
-  fileStream.on('data', (chunk: string) => {
-    bytes += byteLen(chunk);
-    if (bytes > JSONL_HEAD_MAX_BYTES) {
-      fileStream.destroy();
-    }
-  });
   const rl = readline.createInterface({
     input: fileStream,
     crlfDelay: Infinity,
   });
 
+  let bytes = 0;
+  let timedOut = false;
+  let cleaned = false;
+
+  const cleanup = (): void => {
+    if (cleaned) return;
+    cleaned = true;
+    rl.close();
+    fileStream.destroy();
+  };
+
+  const timer = setTimeout(() => {
+    timedOut = true;
+    cleanup();
+  }, JSONL_HEAD_TIMEOUT_MS);
+
+  fileStream.on('data', (chunk: string) => {
+    bytes += byteLen(chunk);
+    if (bytes > JSONL_HEAD_MAX_BYTES) {
+      cleanup();
+    }
+  });
+
   let commandFallback: { text: string; timestamp: string } | null = null;
   let linesRead = 0;
 
@@ -182,8 +204,7 @@ export async function extractFirstUserMessagePreview(
     return commandFallback;
   } finally {
     clearTimeout(timer);
-    rl.close();
+    cleanup();
-    fileStream.destroy();
   }
 
   return commandFallback;

src/renderer/components/dashboard/DashboardView.tsx

@@ -518,10 +518,15 @@ const ProjectsGrid = ({
   const [visibleProjects, setVisibleProjects] = useState(maxProjects);
 
   useEffect(() => {
-    if (repositoryGroups.length === 0 && !repositoryGroupsLoading) {
+    if (repositoryGroups.length === 0 && !repositoryGroupsLoading && !repositoryGroupsError) {
       void fetchRepositoryGroups();
     }
-  }, [repositoryGroups.length, repositoryGroupsLoading, fetchRepositoryGroups]);
+  }, [
+    repositoryGroups.length,
+    repositoryGroupsLoading,
+    repositoryGroupsError,
+    fetchRepositoryGroups,
+  ]);
 
   useEffect(() => {
     if (repositoryGroups.length > 0 && !hasFetchedTasksRef.current && !repositoryGroupsLoading) {

src/renderer/components/sidebar/DateGroupedSessions.tsx

@@ -241,11 +241,18 @@ export const DateGroupedSessions = (): React.JSX.Element => {
   // Loading guards in the store actions prevent duplicate IPC calls
   // when the centralized init chain has already started a fetch.
   const repositoryGroupsLoading = useStore((s) => s.repositoryGroupsLoading);
+  const repositoryGroupsError = useStore((s) => s.repositoryGroupsError);
   const projectsLoading = useStore((s) => s.projectsLoading);
+  const projectsError = useStore((s) => s.projectsError);
   useEffect(() => {
-    if (viewMode === 'grouped' && repositoryGroups.length === 0 && !repositoryGroupsLoading) {
+    if (
+      viewMode === 'grouped' &&
+      repositoryGroups.length === 0 &&
+      !repositoryGroupsLoading &&
+      !repositoryGroupsError
+    ) {
       void fetchRepositoryGroups();
-    } else if (viewMode === 'flat' && projects.length === 0 && !projectsLoading) {
+    } else if (viewMode === 'flat' && projects.length === 0 && !projectsLoading && !projectsError) {
       void fetchProjects();
     }
   }, [
@@ -253,7 +260,9 @@ export const DateGroupedSessions = (): React.JSX.Element => {
     repositoryGroups.length,
     projects.length,
     repositoryGroupsLoading,
+    repositoryGroupsError,
     projectsLoading,
+    projectsError,
     fetchRepositoryGroups,
     fetchProjects,
   ]);