Update app.py

app.py

@@ -5,26 +5,35 @@ from typing import List
 from unified_document_processor import UnifiedDocumentProcessor, CustomEmbeddingFunction
 import chromadb
 from chromadb.config import Settings
+from groq import Groq
+
+def initialize_session_state():
+    """Initialize all session state variables"""
+    if 'CHROMADB_DIR' not in st.session_state:
+        st.session_state.CHROMADB_DIR = os.path.join(os.getcwd(), 'chromadb_data')
+        os.makedirs(st.session_state.CHROMADB_DIR, exist_ok=True)
+    
+    if 'processed_files' not in st.session_state:
+        st.session_state.processed_files = set()
+    
+    if 'processor' not in st.session_state:
+        try:
+            st.session_state.processor = None  # Will be initialized in StreamlitDocProcessor
+        except Exception as e:
+            st.error(f"Error initializing processor: {str(e)}")
 
 class StreamlitDocProcessor:
     def __init__(self):
-        # Create a persistent directory for ChromaDB
-        if 'CHROMADB_DIR' not in st.session_state:
-            st.session_state.CHROMADB_DIR = os.path.join(os.getcwd(), 'chromadb_data')
-            os.makedirs(st.session_state.CHROMADB_DIR, exist_ok=True)
-
-        # Initialize session state
-        if 'processor' not in st.session_state:
+        if st.session_state.processor is None:
             try:
                 groq_api_key = st.secrets["GROQ_API_KEY"]
                 # Initialize processor with persistent ChromaDB
                 st.session_state.processor = self.initialize_processor(groq_api_key)
+                # Update processed files after initializing processor
+                st.session_state.processed_files = self.get_processed_files()
             except Exception as e:
                 st.error(f"Error initializing processor: {str(e)}")
                 return
-        
-        if 'processed_files' not in st.session_state:
-            st.session_state.processed_files = self.get_processed_files()
 
     def initialize_processor(self, groq_api_key):
         """Initialize the processor with persistent ChromaDB"""
@@ -65,8 +74,10 @@ class StreamlitDocProcessor:
     def get_processed_files(self) -> set:
         """Get list of processed files from ChromaDB"""
         try:
-            available_files = st.session_state.processor.get_available_files()
-            return set(available_files['pdf'] + available_files['xml'])
+            if st.session_state.processor:
+                available_files = st.session_state.processor.get_available_files()
+                return set(available_files['pdf'] + available_files['xml'])
+            return set()
         except Exception as e:
             st.error(f"Error getting processed files: {str(e)}")
             return set()
@@ -181,6 +192,10 @@ class StreamlitDocProcessor:
             st.error(f"Error in Q&A interface: {str(e)}")
 
 def main():
+    # Initialize session state
+    initialize_session_state()
+    
+    # Create and run app
     app = StreamlitDocProcessor()
     app.run()