Add analysis trigger

util/dailyS3Scheduler.ts

@@ -21,11 +21,11 @@ export class DailyS3Scheduler {
     }
 
     // Test S3 connection before starting scheduler
-    const connectionTest = await this.s3Sync.testConnection();
-    if (!connectionTest) {
-      logger.error("S3 connection test failed. S3 sync scheduler will not be started.");
-      return;
-    }
+    // const connectionTest = await this.s3Sync.testConnection();
+    // if (!connectionTest) {
+    //   logger.error("S3 connection test failed. S3 sync scheduler will not be started.");
+    //   return;
+    // }
 
     // Cron expression for midnight PST
     // Note: This uses PST timezone. During PDT (daylight time), it will still run at midnight local time
@@ -35,10 +35,11 @@ export class DailyS3Scheduler {
     this.cronJob = cron.schedule(
       cronExpression,
       async () => {
-        await this.performDailySync();
+        //await this.performDailySync();
+        await this.triggerJobAnalysis();
       },
       {
-        timezone: timezone,
+        timezone: timezone
       }
     );
 
@@ -92,6 +93,21 @@ export class DailyS3Scheduler {
     await this.performDailySync();
   }
 
+  async triggerJobAnalysis(): Promise<void> {
+    if (!this.s3Sync) {
+      logger.error("S3 sync not configured");
+      return;
+    }
+
+    logger.info("Triggering jobs directory analysis...");
+    try {
+      const analysis = await this.s3Sync.analyzeJobsDirectory();
+      logger.info("Jobs directory analysis completed:", analysis);
+    } catch (error) {
+      logger.error("Jobs directory analysis failed:", error);
+    }
+  }
+
   /**
    * Get the next scheduled run time
    */
@@ -103,7 +119,7 @@ export class DailyS3Scheduler {
     // Create a date object for midnight PST today
     const now = new Date();
     const pstNow = new Date(now.toLocaleString("en-US", { timeZone: "America/Los_Angeles" }));
-    
+
     // If it's past midnight today, next run is tomorrow at midnight
     const nextRun = new Date(pstNow);
     if (pstNow.getHours() > 0 || pstNow.getMinutes() > 0 || pstNow.getSeconds() > 0) {
@@ -111,7 +127,7 @@ export class DailyS3Scheduler {
     }
     nextRun.setHours(0, 0, 0, 0);
 
-    return nextRun.toLocaleString("en-US", { 
+    return nextRun.toLocaleString("en-US", {
       timeZone: "America/Los_Angeles",
       weekday: "long",
       year: "numeric",
@@ -145,10 +161,10 @@ export class DailyS3Scheduler {
       isConfigured: this.s3Sync !== null,
       isRunning: this.cronJob !== null,
       nextRun: this.getNextRunTime(),
-      syncStats,
+      syncStats
     };
   }
 }
 
 // Export a singleton instance
-export const dailyS3Scheduler = new DailyS3Scheduler();
+export const dailyS3Scheduler = new DailyS3Scheduler();

util/s3Sync.ts

@@ -187,7 +187,7 @@ export class S3Sync {
           jobStats.push(folderStats);
           totalDocuments += folderStats.documentCount;
           totalSizeBytes += folderStats.totalSizeBytes;
-          
+
           // Aggregate file type stats
           for (const [ext, count] of Object.entries(folderStats.fileTypeStats)) {
             aggregatedFileTypeStats[ext] = (aggregatedFileTypeStats[ext] || 0) + count;
@@ -236,7 +236,9 @@ export class S3Sync {
   /**
    * Recursively get document count and total size for a directory
    */
-  private async getDirectoryStats(dirPath: string): Promise<{ documentCount: number; totalSizeBytes: number; fileTypeStats: { [extension: string]: number } }> {
+  private async getDirectoryStats(
+    dirPath: string
+  ): Promise<{ documentCount: number; totalSizeBytes: number; fileTypeStats: { [extension: string]: number } }> {
     let documentCount = 0;
     let totalSizeBytes = 0;
     const fileTypeStats: { [extension: string]: number } = {};
@@ -253,7 +255,7 @@ export class S3Sync {
           const subStats = await this.getDirectoryStats(itemPath);
           documentCount += subStats.documentCount;
           totalSizeBytes += subStats.totalSizeBytes;
-          
+
           // Merge file type stats
           for (const [ext, count] of Object.entries(subStats.fileTypeStats)) {
             fileTypeStats[ext] = (fileTypeStats[ext] || 0) + count;
@@ -262,9 +264,9 @@ export class S3Sync {
           // Count files as documents
           documentCount++;
           totalSizeBytes += stat.size;
-          
+
           // Track file extension
-          const ext = path.extname(item).toLowerCase() || 'no-extension';
+          const ext = path.extname(item).toLowerCase() || "no-extension";
           fileTypeStats[ext] = (fileTypeStats[ext] || 0) + 1;
         }
       }
@@ -340,7 +342,7 @@ export async function analyzeJobsDirectory(): Promise<JobsDirectoryAnalysis> {
         jobStats.push(folderStats);
         totalDocuments += folderStats.documentCount;
         totalSizeBytes += folderStats.totalSizeBytes;
-        
+
         // Aggregate file type stats
         for (const [ext, count] of Object.entries(folderStats.fileTypeStats)) {
           aggregatedFileTypeStats[ext] = (aggregatedFileTypeStats[ext] || 0) + count;
@@ -360,6 +362,9 @@ export async function analyzeJobsDirectory(): Promise<JobsDirectoryAnalysis> {
     logger.info(
       `Jobs directory analysis complete: ${analysis.totalJobs} jobs, ${analysis.totalDocuments} documents, ${analysis.totalSizeMB} MB`
     );
+
+    //Add an upload to the IO database to categorize all of this.
+
     return analysis;
   } catch (error) {
     logger.error("Failed to analyze Jobs directory:", error);
@@ -389,7 +394,9 @@ async function analyzeJobFolder(jobsPath: string, jobId: string): Promise<JobFol
 /**
  * Recursively get document count and total size for a directory (standalone helper function)
  */
-async function getDirectoryStats(dirPath: string): Promise<{ documentCount: number; totalSizeBytes: number; fileTypeStats: { [extension: string]: number } }> {
+async function getDirectoryStats(
+  dirPath: string
+): Promise<{ documentCount: number; totalSizeBytes: number; fileTypeStats: { [extension: string]: number } }> {
   let documentCount = 0;
   let totalSizeBytes = 0;
   const fileTypeStats: { [extension: string]: number } = {};
@@ -406,7 +413,7 @@ async function getDirectoryStats(dirPath: string): Promise<{ documentCount: numb
         const subStats = await getDirectoryStats(itemPath);
         documentCount += subStats.documentCount;
         totalSizeBytes += subStats.totalSizeBytes;
-        
+
         // Merge file type stats
         for (const [ext, count] of Object.entries(subStats.fileTypeStats)) {
           fileTypeStats[ext] = (fileTypeStats[ext] || 0) + count;
@@ -415,9 +422,9 @@ async function getDirectoryStats(dirPath: string): Promise<{ documentCount: numb
         // Count files as documents
         documentCount++;
         totalSizeBytes += stat.size;
-        
+
         // Track file extension
-        const ext = path.extname(item).toLowerCase() || 'no-extension';
+        const ext = path.extname(item).toLowerCase() || "no-extension";
         fileTypeStats[ext] = (fileTypeStats[ext] || 0) + 1;
       }
     }