@arela/uploader 0.2.10 → 0.2.11

package/.envbackup

@@ -0,0 +1,37 @@
+# SUPABASE_URL=https://supabase.prod.app.trader-docs.com
+# SUPABASE_KEY=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJyb2xlIjoic2VydmljZV9yb2xlIiwiaXNzIjoic3VwYWJhc2UiLCJpYXQiOjE3NDU3MzM2MDAsImV4cCI6MTkwMzUwMDAwMH0.eAoXvaBZzZa31txItwEkoLnplWG10ee-U5GMc265xwk
+# SUPABASE_BUCKET=sample
+
+# PATH_VARS=o:/ExpedienteElectronicko/expediente/archivos/:rfcPatente/:patente-:aduana/:year/:pedimento/ALSJDKHF ASD,KFHJA SDFK,ASLDKJFH
+
+SUPABASE_URL=http://127.0.0.1:54321
+SUPABASE_KEY=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJzdXBhYmFzZS1kZW1vIiwicm9sZSI6InNlcnZpY2Vfcm9sZSIsImV4cCI6MTk4MzgxMjk5Nn0.EGIM96RAZx35lJzdJsyH-qQwv8Hdp7fsn3W0YpN81IU
+SUPABASE_BUCKET=arela
+
+# Arela API Configuration
+ARELA_API_URL=http://localhost:3010
+ARELA_API_TOKEN=f0608f83e5faa4c6be32c19975635c4e9012b31f7249a9f0a9270f54d74ec599
+
+# Upload Configuration
+UPLOAD_BASE_PATH=./sample
+UPLOAD_SOURCES=2023|2024
+
+# RFC Upload Configuration
+# Pipe-separated list of RFCs to upload files for
+# Example: MMJ0810145N1|ABC1234567XY|DEF9876543ZZ
+UPLOAD_RFCS=AKS151005E46|IMS030409FZ0|RDG1107154L7|SHP031226BV2|CSM9301219B4|LIN960124HT8|LME971009SW4|AKM9707151B6|FEL000822AG2|FDM060802J54|MTM9807279B4|AUM9207011CA|MMJ0810145N1|ACC010328EQ6|PED781129JT6|CAD890407NK7|SME140411IK7|JME1903121C2|EIJ110429NF9|PTJ080414TM6|TME050503BM4
+
+# AKS151005E46|IMS030409FZ0|RDG1107154L7|SHP031226BV2|CSM9301219B4|LIN960124HT8|LME971009SW4|AKM9707151B6|FEL000822AG2|PTJ080414TM6|TME050503BM4
+# FDM060802J54|MTM9807279B4|AUM9207011CA|MMJ0810145N1|ACC010328EQ6|PED781129JT6|CAD890407NK7|SME140411IK7|JME1903121C2|EIJ110429NF9
+
+# Logging and Verbosity
+VERBOSE_LOGGING=true          # Disable verbose path logging for better performance
+BATCH_DELAY=50                 # Reduce delay between batches (default: 100ms)
+PROGRESS_UPDATE_INTERVAL=20    # Update progress every 20 items (default: 10)
+
+# Log Buffering
+LOG_BUFFER_SIZE=200           # Increase buffer size for fewer I/O ops (default: 100)
+LOG_FLUSH_INTERVAL=3000       # Flush logs every 3 seconds (default: 5000ms)
+
+# Example for maximum performance
+# VERBOSE_LOGGING=false BATCH_DELAY=25 LOG_BUFFER_SIZE=500 arela --stats-only /path/to/files

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@arela/uploader",
-  "version": "0.2.10",
+  "version": "0.2.11",
   "description": "CLI to upload files/directories to Arela",
   "bin": {
     "arela": "./src/index.js"

package/src/config/config.js

@@ -28,10 +28,10 @@ class Config {
       const __dirname = path.dirname(__filename);
       const packageJsonPath = path.resolve(__dirname, '../../package.json');
       const packageJson = JSON.parse(fs.readFileSync(packageJsonPath, 'utf-8'));
-      return packageJson.version || '0.2.10';
+      return packageJson.version || '0.2.11';
     } catch (error) {
       console.warn('⚠️ Could not read package.json version, using fallback');
-      return '0.2.10';
+      return '0.2.11';
     }
   }
 

package/src/services/DatabaseService.js

@@ -316,7 +316,6 @@ export class DatabaseService {
                 filename: record.filename,
                 file_extension: record.file_extension,
                 is_like_simplificado: record.is_like_simplificado,
-                year: record.year,
               })
               .eq('original_path', record.original_path);
 
@@ -669,6 +668,7 @@ export class DatabaseService {
               totalErrors++;
               break;
             }
+            // console.log(`query by basePath: ${basePath} count: [${relatedFilesPage.length}]`);
 
             if (!relatedFilesPage || relatedFilesPage.length === 0) {
               hasMoreRelatedFiles = false;
@@ -682,6 +682,8 @@ export class DatabaseService {
               `Processing related files page ${relatedFilesPageNumber}: ${relatedFilesPage.length} files for ${pedimento.filename}`,
             );
 
+            // Track if any updates occurred in this page
+            let updatesOccurred = false;
             // Process this page of related files in batches
             const pageFileIds = relatedFilesPage.map((f) => f.id);
 
@@ -713,6 +715,7 @@ export class DatabaseService {
                 } else {
                   totalUpdated += batchIds.length;
                   totalRelatedFilesProcessed += batchIds.length;
+                  updatesOccurred = true; // Mark that updates occurred
                   logger.info(
                     `Successfully updated batch ${batchNumber}: ${batchIds.length} files with arela_path and year`,
                   );
@@ -727,16 +730,26 @@ export class DatabaseService {
 
             // Check if we need to fetch the next page of related files
             if (relatedFilesPage.length < relatedFilesPageSize) {
+              console.log('no more related files for ', basePath);
               hasMoreRelatedFiles = false;
               logger.info(
                 `Completed processing related files for ${pedimento.filename}. Total processed: ${totalRelatedFilesProcessed}`,
               );
             } else {
-              relatedFilesFrom += relatedFilesPageSize;
+              // If updates occurred, reset pagination to start from beginning
+              // since records that matched the query may no longer match after update
+              if (updatesOccurred) {
+                relatedFilesFrom = 0;
+                logger.info(
+                  `Page ${relatedFilesPageNumber} complete with updates: ${relatedFilesPage.length} files processed, restarting pagination from beginning due to query condition changes...`,
+                );
+              } else {
+                relatedFilesFrom += relatedFilesPageSize - 1;
+                logger.info(
+                  `Page ${relatedFilesPageNumber} complete: ${relatedFilesPage.length} files processed, continuing to next page...`,
+                );
+              }
               relatedFilesPageNumber++;
-              logger.info(
-                `Page ${relatedFilesPageNumber - 1} complete: ${relatedFilesPage.length} files processed, continuing to next page...`,
-              );
             }
           }
         } catch (error) {
@@ -830,33 +843,75 @@ export class DatabaseService {
       logger.info(`Total files for specified RFCs: ${totalRfcFiles || 0}`);
     }
 
-    // Step 1: Get all pedimento_simplificado records that match the specified RFCs and have arela_path
+    // Step 1: Get all pedimento_simplificado records that match the specified RFCs and have arela_path using pagination
     console.log(
       '🎯 Finding pedimento_simplificado records for specified RFCs...',
     );
 
-    let pedimentoQuery = supabase
-      .from('uploader')
-      .select('arela_path')
-      .eq('document_type', 'pedimento_simplificado')
-      .in('rfc', appConfig.upload.rfcs)
-      .not('arela_path', 'is', null);
+    let allPedimentoRecords = [];
+    let offset = 0;
+    const pageSize = 500; // Process pedimento records in pages
+    let hasMorePedimentoData = true;
+    let pageNumber = 1;
 
-    // Add year filter if UPLOAD_YEARS is configured
-    if (appConfig.upload.years && appConfig.upload.years.length > 0) {
-      pedimentoQuery = pedimentoQuery.in('year', appConfig.upload.years);
-    }
+    // Process pedimento records page by page for memory efficiency
+    while (hasMorePedimentoData) {
+      logger.info(
+        `Fetching pedimento records page ${pageNumber} (records ${offset + 1} to ${offset + pageSize})...`,
+      );
 
-    const { data: pedimentoRfcRecords, error: pedimentoRfcError } =
-      await pedimentoQuery;
+      let pedimentoQuery = supabase
+        .from('uploader')
+        .select('arela_path')
+        .eq('document_type', 'pedimento_simplificado')
+        .in('rfc', appConfig.upload.rfcs)
+        .not('arela_path', 'is', null);
 
-    if (pedimentoRfcError) {
-      const errorMsg = `Error fetching pedimento RFC records: ${pedimentoRfcError.message}`;
-      logger.error(errorMsg);
-      throw new Error(errorMsg);
+      // Add year filter if UPLOAD_YEARS is configured
+      if (appConfig.upload.years && appConfig.upload.years.length > 0) {
+        pedimentoQuery = pedimentoQuery.in('year', appConfig.upload.years);
+      }
+
+      const { data: pedimentoPage, error: pedimentoError } =
+        await pedimentoQuery
+          .range(offset, offset + pageSize - 1)
+          .order('created_at');
+
+      if (pedimentoError) {
+        const errorMsg = `Error fetching pedimento RFC records page ${pageNumber}: ${pedimentoError.message}`;
+        logger.error(errorMsg);
+        throw new Error(errorMsg);
+      }
+
+      if (!pedimentoPage || pedimentoPage.length === 0) {
+        hasMorePedimentoData = false;
+        logger.info('No more pedimento records found');
+        break;
+      }
+
+      logger.info(
+        `Processing pedimento page ${pageNumber}: ${pedimentoPage.length} pedimento records`,
+      );
+
+      // Add this page to our collection
+      allPedimentoRecords = allPedimentoRecords.concat(pedimentoPage);
+
+      // Check if we need to fetch the next page
+      if (pedimentoPage.length < pageSize) {
+        hasMorePedimentoData = false;
+        logger.info(
+          `Completed fetching pedimento records. Last page ${pageNumber} had ${pedimentoPage.length} records`,
+        );
+      } else {
+        offset += pageSize;
+        pageNumber++;
+        logger.info(
+          `Page ${pageNumber - 1} complete: ${pedimentoPage.length} records fetched, moving to next page...`,
+        );
+      }
     }
 
-    if (!pedimentoRfcRecords || pedimentoRfcRecords.length === 0) {
+    if (!allPedimentoRecords || allPedimentoRecords.length === 0) {
       console.log(
         'ℹ️  No pedimento_simplificado records found for the specified RFCs with arela_path',
       );
@@ -864,15 +919,15 @@ export class DatabaseService {
       return { processedCount: 0, uploadedCount: 0, errorCount: 0 };
     }
 
-    // Get unique arela_paths from pedimento records
+    // Get unique arela_paths from all pedimento records
     const uniqueArelaPaths = [
-      ...new Set(pedimentoRfcRecords.map((r) => r.arela_path)),
+      ...new Set(allPedimentoRecords.map((r) => r.arela_path)),
     ];
     console.log(
-      `📋 Found ${pedimentoRfcRecords.length} pedimento records with ${uniqueArelaPaths.length} unique arela_paths for specified RFCs`,
+      `📋 Found ${allPedimentoRecords.length} pedimento records with ${uniqueArelaPaths.length} unique arela_paths for specified RFCs across ${pageNumber - 1} pages`,
     );
     logger.info(
-      `Found ${pedimentoRfcRecords.length} pedimento records with ${uniqueArelaPaths.length} unique arela_paths`,
+      `Found ${allPedimentoRecords.length} pedimento records with ${uniqueArelaPaths.length} unique arela_paths across ${pageNumber - 1} pages`,
     );
 
     // Step 2: Get all files with these arela_paths that haven't been uploaded yet
@@ -937,16 +992,24 @@ export class DatabaseService {
       `Found ${rfcRecords.length} files ready for upload, ${skipped} skipped`,
     );
 
-    // Step 3: Get ALL files that have these arela_paths (including supporting documents)
-    let allRelatedFiles = [];
+    // Step 3: Process files with streaming pagination to avoid memory overload
+    let totalProcessed = 0;
+    let totalUploaded = 0;
+    let totalErrors = 0;
+    let globalFileCount = 0;
     const arelaPathChunkSize = 50;
-    const queryBatchSize = 1000;
+    const queryBatchSize = 500; // Reduced batch size for better memory management
+    const batchSize = parseInt(options.batchSize) || 10;
+
+    // Import performance configuration
+    const { performance: perfConfig } = appConfig;
+    const maxConcurrency = perfConfig?.maxApiConnections || 3;
 
     console.log(
-      '📥 Fetching all related files (processing arela_paths in chunks to avoid URI limits)...',
+      '📥 Processing files with streaming pagination (processing arela_paths in chunks to avoid URI limits and memory overload)...',
     );
 
-    // Process arela_paths in chunks
+    // Process arela_paths in chunks and upload files as we fetch them
     for (let i = 0; i < uploadableArelaPaths.length; i += arelaPathChunkSize) {
       const arelaPathChunk = uploadableArelaPaths.slice(
         i,
@@ -961,9 +1024,10 @@ export class DatabaseService {
         `   Processing arela_path chunk ${chunkNumber}/${totalChunks} (${arelaPathChunk.length} paths)`,
       );
 
-      // For each chunk of arela_paths, use pagination to get all related files
+      // For each chunk of arela_paths, use pagination to get related files and process them immediately
       let hasMore = true;
       let offset = 0;
+      let chunkFileCount = 0;
 
       while (hasMore) {
         const { data: batch, error: queryError } = await supabase
@@ -972,7 +1036,8 @@ export class DatabaseService {
           .in('arela_path', arelaPathChunk)
           .not('original_path', 'is', null)
           .neq('status', 'file-uploaded')
-          .range(offset, offset + queryBatchSize - 1);
+          .range(offset, offset + queryBatchSize - 1)
+          .order('created_at');
 
         if (queryError) {
           const errorMsg = `Error fetching related files for chunk ${chunkNumber}: ${queryError.message}`;
@@ -980,74 +1045,94 @@ export class DatabaseService {
           throw new Error(errorMsg);
         }
 
-        if (batch && batch.length > 0) {
-          allRelatedFiles = allRelatedFiles.concat(batch);
+        if (!batch || batch.length === 0) {
+          hasMore = false;
+          break;
         }
 
-        hasMore = batch && batch.length === queryBatchSize;
-        offset += queryBatchSize;
-      }
-    }
+        chunkFileCount += batch.length;
+        globalFileCount += batch.length;
 
-    if (!allRelatedFiles || allRelatedFiles.length === 0) {
-      console.log('ℹ️  No related files found to upload');
-      logger.info('No related files found to upload');
-      return { processedCount: 0, uploadedCount: 0, errorCount: 0 };
-    }
+        console.log(
+          `   📦 Processing batch within chunk ${chunkNumber}: ${batch.length} files (total processed so far: ${globalFileCount})`,
+        );
 
-    console.log(`📋 Total files to upload: ${allRelatedFiles.length}`);
-    logger.info(`Total files to upload: ${allRelatedFiles.length}`);
+        // Track if any uploads occurred in this batch
+        let uploadsOccurred = false;
 
-    // Step 4: Upload all related files using concurrent batch processing
-    let totalProcessed = 0;
-    let totalUploaded = 0;
-    let totalErrors = 0;
-    const batchSize = parseInt(options.batchSize) || 10;
+        // Process this batch of files immediately using concurrent processing
+        // Split batch into upload batches
+        for (let j = 0; j < batch.length; j += batchSize) {
+          const uploadBatch = batch.slice(j, j + batchSize);
+          const batchNum = Math.floor(globalFileCount / batchSize) + 1;
 
-    // Import performance configuration
-    const { performance: perfConfig } = appConfig;
-    const maxConcurrency = perfConfig?.maxApiConnections || 3;
+          console.log(
+            `📦 Processing upload batch ${batchNum} (${uploadBatch.length} files)`,
+          );
 
-    console.log(
-      `🚀 Starting batch upload: ${allRelatedFiles.length} files in batches of ${batchSize}`,
-    );
-    console.log(
-      `⚡ Concurrent processing: up to ${maxConcurrency} parallel operations`,
-    );
+          // Process batch using concurrent processing similar to UploadCommand
+          const batchResults = await this.#processRfcBatch(
+            uploadBatch,
+            uploadService,
+            supabase,
+            options,
+            maxConcurrency,
+          );
 
-    // Process files in batches with concurrent processing
-    for (let i = 0; i < allRelatedFiles.length; i += batchSize) {
-      const batch = allRelatedFiles.slice(i, i + batchSize);
-      const batchNum = Math.floor(i / batchSize) + 1;
-      const totalBatches = Math.ceil(allRelatedFiles.length / batchSize);
+          totalProcessed += batchResults.processed;
+          totalUploaded += batchResults.uploaded;
+          totalErrors += batchResults.errors;
 
-      console.log(
-        `📦 Processing batch ${batchNum}/${totalBatches} (${batch.length} files)`,
-      );
+          // Track if uploads occurred (status changes from non-uploaded to uploaded)
+          if (batchResults.uploaded > 0) {
+            uploadsOccurred = true;
+          }
 
-      // Process batch using concurrent processing similar to UploadCommand
-      const batchResults = await this.#processRfcBatch(
-        batch,
-        uploadService,
-        supabase,
-        options,
-        maxConcurrency,
-      );
+          console.log(
+            `📊 Upload batch complete - Total progress: ${totalUploaded} uploaded, ${totalErrors} errors`,
+          );
+        }
+
+        // Check if we need more data from this chunk
+        if (batch.length < queryBatchSize) {
+          hasMore = false;
+        } else {
+          // If uploads occurred, reset pagination to start from beginning
+          // since records that matched the query may no longer match after upload
+          if (uploadsOccurred) {
+            offset = 0;
+            console.log(
+              `   📄 Batch complete with uploads: ${batch.length} files processed, restarting pagination from beginning due to query condition changes...`,
+            );
+          } else {
+            offset += queryBatchSize;
+            console.log(
+              `   📄 Batch complete: ${batch.length} files processed, continuing to next page (offset: ${offset})...`,
+            );
+          }
+        }
 
-      totalProcessed += batchResults.processed;
-      totalUploaded += batchResults.uploaded;
-      totalErrors += batchResults.errors;
+        if (hasMore) {
+          console.log(
+            `   📄 Fetching more files for chunk ${chunkNumber}... (offset: ${offset})`,
+          );
+        }
+      }
 
-      // Progress update
-      const progress = (
-        ((i + batch.length) / allRelatedFiles.length) *
-        100
-      ).toFixed(1);
       console.log(
-        `📊 Batch ${batchNum} complete - Progress: ${progress}% (${totalUploaded}/${allRelatedFiles.length} uploaded)`,
+        `   ✅ Chunk ${chunkNumber}/${totalChunks} complete: ${chunkFileCount} files processed`,
       );
     }
 
+    if (globalFileCount === 0) {
+      console.log('ℹ️  No related files found to upload');
+      logger.info('No related files found to upload');
+      return { processedCount: 0, uploadedCount: 0, errorCount: 0 };
+    }
+
+    console.log(`📋 Total files processed: ${globalFileCount}`);
+    logger.info(`Total files processed: ${globalFileCount}`);
+
     const result = {
       processedCount: totalProcessed,
       uploadedCount: totalUploaded,