Update app.py

app.py

@@ -75,14 +75,14 @@ data = loader.load()
 # (If you find that key fields are getting split, consider implementing a custom splitter.)
 from langchain_text_splitters import RecursiveCharacterTextSplitter
 text_splitter = RecursiveCharacterTextSplitter(
-    chunk_size=1250,  
-    chunk_overlap=150,  
+    chunk_size=1500,  
+    chunk_overlap=100,  
     add_start_index=True
 )
 def split_document_with_metadata(document):
-    # Split the document into chunks
+    # Split the document text into chunks.
     chunks = text_splitter.split_text(document.page_content)
-    # Attach the full metadata to every chunk
+    # Ensure every chunk has the complete original metadata.
     return [Document(page_content=chunk, metadata=document.metadata) for chunk in chunks]
 
 all_splits = []
@@ -94,6 +94,14 @@ for doc in data:
 
 # Create a Chroma vector store using the document splits.
 persist_directory = "./chroma_db"
+if os.path.exists(persist_directory):
+    print(f"Clearing existing vector store at {persist_directory}...")
+    shutil.rmtree(persist_directory)
+
+# -------------------------------
+# Set Up Retrievers
+# -------------------------------
+# Create a Chroma vector store using the document splits.
 vectorstore = Chroma.from_documents(
     documents=all_splits, 
     embedding=OpenAIEmbeddings(),