update

src/legisqa_local/config/settings.py

@@ -39,6 +39,13 @@ def get_chroma_config():
 
 def setup_chromadb():
     """Setup ChromaDB - use persistent storage (/data) or download from HF Dataset if needed"""
+    # Use session state to prevent duplicate runs in Streamlit
+    import streamlit as st
+    if hasattr(st, 'session_state') and getattr(st.session_state, 'chromadb_setup_complete', False):
+        logger.info("ChromaDB setup already completed in this session")
+        chroma_config = get_chroma_config()
+        return chroma_config["persist_directory"]
+    
     logger.info("=== ChromaDB Setup Starting ===")
     
     chroma_config = get_chroma_config()
@@ -76,6 +83,9 @@ def setup_chromadb():
             logger.info(f"Updated CHROMA_PERSIST_DIRECTORY to: {persistent_chroma_path}")
             # Inspect the existing ChromaDB
             inspect_chromadb(persistent_chroma_path)
+            # Mark setup as complete
+            if hasattr(st, 'session_state'):
+                st.session_state.chromadb_setup_complete = True
             return persistent_chroma_path
         
         # ChromaDB not found in persistent storage, try to download from HF Dataset
@@ -96,6 +106,9 @@ def setup_chromadb():
                 logger.info(f"✅ ChromaDB download successful! Updated path to: {persistent_chroma_path}")
                 # Inspect the downloaded ChromaDB
                 inspect_chromadb(persistent_chroma_path)
+                # Mark setup as complete
+                if hasattr(st, 'session_state'):
+                    st.session_state.chromadb_setup_complete = True
                 return persistent_chroma_path
             else:
                 logger.error("❌ ChromaDB download from HF Dataset failed!")
@@ -115,6 +128,9 @@ def setup_chromadb():
             logger.info(f"✅ ChromaDB found at {chroma_path}")
             # Inspect the fallback ChromaDB
             inspect_chromadb(chroma_path)
+            # Mark setup as complete
+            if hasattr(st, 'session_state'):
+                st.session_state.chromadb_setup_complete = True
             return chroma_path
         else:
             logger.warning(f"ChromaDB directory exists but is empty: {chroma_path}")
@@ -123,6 +139,9 @@ def setup_chromadb():
     
     logger.warning(f"⚠️  Using default ChromaDB path: {chroma_path}")
     logger.info("=== ChromaDB Setup Complete ===")
+    # Mark setup as complete even if no ChromaDB found (prevents infinite retries)
+    if hasattr(st, 'session_state'):
+        st.session_state.chromadb_setup_complete = True
     return chroma_path
 
 def download_chromadb_from_hf_dataset(dataset_repo: str, local_path: str) -> bool:
@@ -141,8 +160,9 @@ def download_chromadb_from_hf_dataset(dataset_repo: str, local_path: str) -> boo
         logger.info(f"Downloading dataset: {dataset_repo}")
         logger.info("This may take several minutes for large datasets...")
         
-        # Download to a temporary location first
-        temp_download_path = f"{local_path}_temp"
+        # Download to a unique temporary location first to avoid conflicts
+        import uuid
+        temp_download_path = f"{local_path}_temp_{uuid.uuid4().hex[:8]}"
         
         downloaded_path = snapshot_download(
             repo_id=dataset_repo,
@@ -161,18 +181,42 @@ def download_chromadb_from_hf_dataset(dataset_repo: str, local_path: str) -> boo
         if os.path.exists(chromadb_subdir):
             logger.info(f"📁 Found ChromaDB subdirectory: {chromadb_subdir}")
             
-            # Move the ChromaDB files from the subdirectory to the target location
-            import shutil
-            if os.path.exists(local_path):
-                shutil.rmtree(local_path)
-            shutil.move(chromadb_subdir, local_path)
-            
-            # Clean up the temporary download directory
-            shutil.rmtree(temp_download_path)
-            
-            logger.info(f"✅ ChromaDB files moved to: {local_path}")
+            try:
+                # Move the ChromaDB files from the subdirectory to the target location
+                import shutil
+                if os.path.exists(local_path):
+                    logger.info(f"Removing existing target directory: {local_path}")
+                    shutil.rmtree(local_path)
+                
+                logger.info(f"Moving ChromaDB from {chromadb_subdir} to {local_path}")
+                shutil.move(chromadb_subdir, local_path)
+                
+                # Clean up the temporary download directory
+                logger.info(f"Cleaning up temporary directory: {temp_download_path}")
+                shutil.rmtree(temp_download_path)
+                
+                logger.info(f"✅ ChromaDB files moved to: {local_path}")
+            except Exception as e:
+                logger.error(f"❌ Error moving ChromaDB files: {e}")
+                # Clean up temporary directory on error
+                try:
+                    shutil.rmtree(temp_download_path)
+                except:
+                    pass
+                return False
         else:
             logger.error(f"❌ ChromaDB subdirectory not found in downloaded data: {chromadb_subdir}")
+            # List what we actually downloaded for debugging
+            if os.path.exists(temp_download_path):
+                logger.info(f"Contents of {temp_download_path}:")
+                for item in os.listdir(temp_download_path):
+                    logger.info(f"  {item}")
+            # Clean up temporary directory
+            try:
+                import shutil
+                shutil.rmtree(temp_download_path)
+            except:
+                pass
             return False
         
         # Verify the final ChromaDB structure