Update app.py

app.py

@@ -181,18 +181,19 @@ def self_reasoning(query, context):
     reasoning_prompt = f"""
     You are an AI assistant that analyzes the context provided to answer the user's query comprehensively and clearly. 
     Answer in a concise, factual way using the terminology from the context. Avoid extra explanation unless explicitly asked.
-    YOU MUST mention the page number. 
+    YOU MUST mention the document file name (e.g., tools.html, refguide.html) in your answer.
+
     ### Example 1:
-    **Question:** What is the purpose of the MODTRAN GUI?
+    **Question:** What is the purpose of the Monte GUI?
     **Context:**
-    [Page 10 of the docuemnt] The MODTRAN GUI helps users set parameters and visualize the model's output.
-    **Answer:** The MODTRAN GUI assists users in parameter setup and output visualization. You can find the answer at Page 10 of the document provided.
+    [From `tools.html`] The Monte GUI provides interfaces for setting up trajectory parameters and viewing output results.
+    **Answer:** The Monte GUI helps users configure trajectory parameters and visualize results. (From `tools.html`)
 
     ### Example 2:
-    **Question:** How do you run MODTRAN on Linux? Answer with page number. 
+    **Question:** How do you perform covariance analysis in Monte?
     **Context:**
-    [Page 15 of the docuemnt] On Linux systems, MODTRAN can be run using the `mod6c` binary via terminal.
-    **Answer:** Use the `mod6c` binary via terminal. (Page 15 of the document)
+    [From `designEdition.html`] The Monte Design Edition includes support for statistical maneuver and covariance analysis during the design phase.
+    **Answer:** Monte supports covariance analysis through the Design Edition. (From `designEdition.html`)
 
     ### Now answer:
     **Question:** {query}
@@ -200,6 +201,7 @@ def self_reasoning(query, context):
     {context}
 
     **Answer:**
+
     """
     try:
         result = llm._call(reasoning_prompt)
@@ -218,7 +220,7 @@ def faiss_search_with_keywords(query):
     refined_query = " ".join([keyword[0] for keyword in keywords])
     retriever = vectorstore_global.as_retriever(search_kwargs={"k": 13})
     docs = retriever.get_relevant_documents(refined_query)
-    context= '\n\n'.join([f"[Page {doc.metadata.get('page', 'Unknown')}] {doc.page_content}" for doc in docs])
+    context = '\n\n'.join([f"[From `{doc.metadata.get('document', 'unknown.html')}`] {doc.page_content}" for doc in docs])
     return self_reasoning(query, context)
 
 def get_reranker():
@@ -238,7 +240,7 @@ def faiss_search_with_reasoning(query):
     scores = reranker.predict(pairs)
     reranked_docs = sorted(zip(scores, docs), key=lambda x: x[0], reverse=True)
     top_docs = [doc for _, doc in reranked_docs[:5]]
-    context = '\n\n'.join([f"[Page {doc.metadata.get('page', 'Unknown')}] {doc.page_content.strip()}" for doc in top_docs])
+    context = '\n\n'.join([f"[From `{doc.metadata.get('document', 'unknown.html')}`] {doc.page_content.strip()}" for doc in top_docs])
     return self_reasoning(query, context)
 
 faiss_keyword_tool = Tool(