Update app.py

app.py

@@ -257,7 +257,7 @@ def create_indexed_database():
     database_uploaded = progress.get("database_uploaded", False)
     indexing_complete = progress.get("indexing_complete", False)
     
-    # If fully complete, try to download
+    # If fully complete, download and return
     if indexing_complete:
         log_progress("Fully indexed database exists in HF!", "SUCCESS")
         log_progress(f"Downloading from {INDEXED_REPO_ID}...", "INFO")
@@ -271,8 +271,7 @@ def create_indexed_database():
             )
             log_progress(f"Downloaded: {indexed_path}", "SUCCESS")
             
-            # CRITICAL CHECK: Verify downloaded database actually has indices
-            log_progress("Verifying downloaded database has indices...", "INFO")
+            # Verify indices
             has_indices, idx_count = verify_database_has_indices(indexed_path)
             
             if has_indices:
@@ -280,32 +279,67 @@ def create_indexed_database():
                 return indexed_path
             else:
                 log_progress(f"  ❌ CORRUPTED: Only {idx_count} indices (expected 4+)", "ERROR")
-                log_progress("  The uploaded database is missing indices!", "ERROR")
-                log_progress("  This was caused by WAL mode not being checkpointed.", "ERROR")
                 log_progress("  Forcing re-indexing...", "WARN")
                 
-                # Reset progress to force rebuild
+                # Reset progress
                 indexing_complete = False
                 completed_indices = set()
                 analyzed_tables = set()
                 database_uploaded = False
-                
-                # Update remote to clear the bad state
                 update_remote_progress([], [], database_uploaded=False, indexing_complete=False)
             
         except Exception as e:
             log_progress(f"Download failed: {e}", "ERROR")
-            log_progress("Will create locally", "INFO")
     
-    # Check for partial progress
-    if completed_indices or analyzed_tables:
-        log_progress("Resuming from checkpoint:", "INFO")
+    # CRITICAL FIX: If ANY progress exists, download from OUR repo (not original!)
+    if (completed_indices or analyzed_tables or database_uploaded) and not os.path.exists(LOCAL_DB_PATH):
+        log_progress("Checkpoint detected - downloading PARTIALLY INDEXED database from our repo...", "INFO")
         log_progress(f"  Completed indices: {sorted(completed_indices)}", "INFO")
         log_progress(f"  Analyzed tables: {sorted(analyzed_tables)}", "INFO")
+        
+        try:
+            # Download from OUR repo (has indices!)
+            indexed_path = hf_hub_download(
+                repo_id=INDEXED_REPO_ID,
+                filename=INDEXED_DB_FILENAME,
+                repo_type="dataset",
+                token=HF_TOKEN
+            )
+            
+            log_progress(f"Downloaded partial DB: {indexed_path}", "SUCCESS")
+            
+            # Verify it has the expected indices
+            has_indices, idx_count = verify_database_has_indices(indexed_path)
+            log_progress(f"  Verified: {idx_count} indices present", "SUCCESS" if idx_count >= len(completed_indices) else "WARN")
+            
+            # Copy to working location
+            log_progress(f"Copying to {LOCAL_DB_PATH}...", "INFO")
+            start = time.time()
+            shutil.copy2(indexed_path, LOCAL_DB_PATH)
+            elapsed = time.time() - start
+            log_progress(f"Copied in {elapsed:.1f}s", "SUCCESS")
+            
+            # Successfully resumed from checkpoint
+            log_progress("Resuming from our partially indexed database ✅", "SUCCESS")
+            
+        except Exception as e:
+            log_progress(f"Could not download from our repo: {e}", "WARN")
+            log_progress("Will fall back to original database and rebuild indices", "WARN")
+            
+            # Reset since we can't continue
+            completed_indices = set()
+            analyzed_tables = set()
     
-    # Download original database if needed
+    # Only download ORIGINAL if we have NO local DB and NO checkpoint
     if not os.path.exists(LOCAL_DB_PATH):
-        log_progress("Downloading original database...", "INFO")
+        if completed_indices or analyzed_tables:
+            log_progress("WARNING: Should have downloaded from our repo but failed!", "ERROR")
+            log_progress("Clearing progress and starting from scratch...", "WARN")
+            completed_indices = set()
+            analyzed_tables = set()
+            update_remote_progress([], [], database_uploaded=False, indexing_complete=False)
+        
+        log_progress("Downloading ORIGINAL database (no checkpoint exists)...", "INFO")
         original_path = hf_hub_download(
             repo_id=ORIGINAL_REPO_ID,
             filename=ORIGINAL_DB_FILENAME,