updated app.py

app.py

@@ -6,19 +6,19 @@ import os
 from document_processor.file_handler import DocumentProcessor
 from retriever.builder import RetrieverBuilder
 from agents.workflow import AgentWorkflow
-from config import constants, settings
+from config import constants
 from utils.logging import logger
 
 # -------------------------
 # Utils
 # -------------------------
-def _get_file_hashes(uploaded_files: List) -> frozenset:
+def _get_file_hashes(uploaded_files: List) -> List[str]:
     """Generate SHA-256 hashes for uploaded files."""
-    hashes = set()
+    hashes = []
     for file in uploaded_files:
         with open(file.name, "rb") as f:
-            hashes.add(hashlib.sha256(f.read()).hexdigest())
-    return frozenset(hashes)
+            hashes.append(hashlib.sha256(f.read()).hexdigest())
+    return hashes
 
 # -------------------------
 # Main App
@@ -64,8 +64,8 @@ def main():
 
         # Session state
         session_state = gr.State({
-            "file_hashes": frozenset(),
-            "retriever": None
+            "file_hashes": [],
+            "retriever_ready": False  # track if retriever is ready
         })
 
         # -------------------------
@@ -93,19 +93,23 @@ def main():
 
                 current_hashes = _get_file_hashes(uploaded_files)
 
-                if state["retriever"] is None or current_hashes != state["file_hashes"]:
+                # Initialize retriever if not ready or files changed
+                if not state.get("retriever_ready", False) or current_hashes != state["file_hashes"]:
                     logger.info("Processing new/changed documents...")
                     chunks = processor.process(uploaded_files)
                     retriever = retriever_builder.build_hybrid_retriever(chunks)
-                    state.update({
-                        "file_hashes": current_hashes,
-                        "retriever": retriever
-                    })
+                    state["file_hashes"] = current_hashes
+                    state["retriever_ready"] = True
+                    state["_retriever_internal"] = retriever  # keep internal, not for output
+                else:
+                    retriever = state["_retriever_internal"]
 
+                # Run workflow
                 result = workflow.full_pipeline(
                     question=question_text,
-                    retriever=state["retriever"]
+                    retriever=retriever
                 )
+
                 return result["draft_answer"], result["verification_report"], state
 
             except Exception as e: