Fix tar extraction to handle nested directories

streamlit_app.py

@@ -65,12 +65,68 @@ def extract_packaged_data():
         # Extract the package
         try:
             with tarfile.open(PACKAGE_FILE, "r:gz") as tar:
+                print("Examining tar file contents before extraction:")
+                for member in tar.getmembers():
+                    print(f"  File in archive: {member.name}")
+                
+                # Extract files, handling potential nested directories
                 print("Extracting package...")
-                tar.extractall(path=PROCESSED_DATA_DIR)
+                for member in tar.getmembers():
+                    # Skip directories
+                    if member.isdir():
+                        continue
+                    
+                    # Get the basename and handle nested paths
+                    # If file is in processed_data/something, extract just "something"
+                    # If file is just something, extract as is
+                    basename = os.path.basename(member.name)
+                    
+                    # Determine target path
+                    if basename == "document_chunks.pkl":
+                        target_path = CHUNKS_FILE
+                    elif "qdrant_vectorstore" in member.name:
+                        # For Qdrant files, preserve the subdirectory structure
+                        if member.name.startswith("processed_data/"):
+                            # Remove 'processed_data/' prefix if it exists
+                            relative_path = member.name[len("processed_data/"):]
+                        else:
+                            relative_path = member.name
+                        
+                        target_path = PROCESSED_DATA_DIR / relative_path
+                    else:
+                        # Other files go directly in processed_data
+                        target_path = PROCESSED_DATA_DIR / basename
+                    
+                    # Create directories if needed
+                    os.makedirs(os.path.dirname(target_path), exist_ok=True)
+                    
+                    # Extract the file
+                    print(f"  Extracting {member.name} to {target_path}")
+                    f = tar.extractfile(member)
+                    if f is not None:
+                        with open(target_path, "wb") as out_file:
+                            out_file.write(f.read())
+                
                 print("Extraction complete")
+                
+                # Verify extraction worked
+                print("Checking extracted files:")
+                if os.path.exists(CHUNKS_FILE):
+                    print(f"  {CHUNKS_FILE} exists: ✓")
+                else:
+                    print(f"  {CHUNKS_FILE} exists: ✗")
+                
+                if os.path.exists(QDRANT_DIR):
+                    print(f"  {QDRANT_DIR} exists: ✓")
+                    print(f"  Contents: {os.listdir(QDRANT_DIR)}")
+                else:
+                    print(f"  {QDRANT_DIR} exists: ✗")
+                    
                 return True
         except Exception as e:
             print(f"Error extracting package: {str(e)}")
+            import traceback
+            traceback.print_exc()
             return False
     else:
         print(f"No packaged data found: {PACKAGE_FILE}")