feat: move prompts to text files and reference them in config.yaml

src/mvt/config.yaml

@@ -12,10 +12,13 @@ html_files: "html_files"
 persist_directory: "faiss_index"
 host: "127.0.0.1"
 port: 8080
-system_prompt: "You are an assistant for question-answering tasks that are related to Linux Foundation Decentralized Trust former Hyperledger blockchain. Use the following pieces of retrieved context to answer the question. If you don't know the answer, say that you don't know."
 use_query_rewriting: false # Set to true to enable query rewriting
-query_rewriting_prompt: "You are an assistant helping to rewrite user queries to make them more specific and effective for searching documents. The context is the Linux Foundation Decentralized Trust former Hyperledger. Please rewrite the human query to be more specific, detailed, and optimized for document retrieval, considering the context mentioned."
 logo_pth: "https://upload.wikimedia.org/wikipedia/en/thumb/e/e2/The_Founder_Institute_Logo.png/250px-The_Founder_Institute_Logo.png"
 nr_retrieved_documents: 5
 max_download_retries: 3        # Number of retry attempts for failed downloads
-retry_delay_seconds: 2         # Base delay between retries (increases with each attempt)
+retry_delay_seconds: 2         # Base delay between retries (increases with each attempt)
+
+# Write prompts to files
+prompts:
+  system_prompt: src/mvt/prompts/system_prompt.txt
+  query_rewriting_prompt: src/mvt/prompts/query_rewriting_prompt.txt

src/mvt/main.py

@@ -7,10 +7,18 @@
 from langchain_core.prompts import ChatPromptTemplate
 from langchain.chains import create_retrieval_chain
 from langchain_mistralai.embeddings import MistralAIEmbeddings
-#from langchain.retrievers import BM25Retriever, EnsembleRetriever
 from langchain_core.documents import Document
 from typing import List
 
+def load_prompt_from_file(file_path):
+    """Load a prompt template from a file."""
+    try:
+        with open(file_path, 'r') as file:
+            prompt_text = file.read().strip()
+        return prompt_text
+    except Exception as e:
+        print(f"Error loading prompt from {file_path}: {e}")
+        return None
 
 def get_ragchain(filter):
     # Read config data with database prompt overrides
@@ -42,15 +50,19 @@ def get_ragchain(filter):
             model=config_data["model_name"],
             temperature=0.7
         )
-    
+
     # Load local vector db
     docsearch = FAISS.load_local(config_data["persist_directory"], embeddings, allow_dangerous_deserialization=True)
 
     # Define a retriever interface
     retriever = docsearch.as_retriever(search_kwargs={"k": config_data["nr_retrieved_documents"], "filter": filter})
 
-    # read prompt string from config file
-    prompt_str = config_data["system_prompt"]
+    # Load system prompt from file or use the one in config
+    prompt_path = config_data.get("prompts", {}).get("system_prompt")
+    if prompt_path and os.path.exists(prompt_path):
+        prompt_str = load_prompt_from_file(prompt_path)
+    else:
+        prompt_str = config_data["system_prompt"]
 
     # Answer question
     qa_system_prompt = (
@@ -65,6 +77,7 @@ def get_ragchain(filter):
             ("user", "{input}"),
         ]
     )
+
     question_answer_chain = create_stuff_documents_chain(model, qa_prompt)
 
     rag_chain = create_retrieval_chain(retriever, question_answer_chain)  

src/mvt/prompts/query_rewriting_prompt.txt

@@ -0,0 +1 @@
+You are an assistant helping to rewrite user queries to make them more specific and effective for searching documents. The context is the Linux Foundation Decentralized Trust former Hyperledger. Please rewrite the human query to be more specific, detailed, and optimized for document retrieval, considering the context mentioned.

src/mvt/prompts/system_prompt.txt

@@ -0,0 +1 @@
+You are an assistant for question-answering tasks that are related to Linux Foundation Decentralized Trust former Hyperledger blockchain. Use the following pieces of retrieved context to answer the question. If you don't know the answer, say that you don't know.

src/mvt/query_rewriting.py

@@ -2,6 +2,7 @@
 from utils import load_yaml_file_with_db_prompts
 from dotenv import load_dotenv, find_dotenv
 from langchain_mistralai.chat_models import ChatMistralAI
+from main import load_prompt_from_file
 
 def query_rewriting_llm(user_query, context="Founder Institute Keystone Chapter"):
     """
@@ -38,7 +39,12 @@ def query_rewriting_llm(user_query, context="Founder Institute Keystone Chapter"
             temperature=0.7
         )
 
-    query_rewriting_prompt = config_data["query_rewriting_prompt"]
+    # Load query rewriting prompt from file or use the one in config
+    prompt_path = config_data.get("prompts", {}).get("query_rewriting_prompt")
+    if prompt_path and os.path.exists(prompt_path):
+        query_rewriting_prompt = load_prompt_from_file(prompt_path)
+    else:
+        query_rewriting_prompt = config_data["query_rewriting_prompt"]
 
     messages = [
         ("system", query_rewriting_prompt),