rag.py

from langchain.embeddings.openai import OpenAIEmbeddings
from langchain.vectorstores import Chroma
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.document_loaders import TextLoader

# We'll assume you have a documentation text file. If not, we can use some sample Python docs.
# Let's create a sample if the file doesn't exist, or load it.

def load_documents():
    # Load the documents from a file (or multiple files)
    # For demonstration, we'll create a sample document if it doesn't exist.
    doc_path = "python_docs.txt"
    if not os.path.exists(doc_path):
        # Create a sample documentation about Python functions
        with open(doc_path, 'w') as f:
            f.write("""
            Functions in Python are defined using the def keyword.
            For example: def hello_world(): print("Hello, world!")
            Functions can take parameters and return values.
            """)
    loader = TextLoader(doc_path)
    documents = loader.load()
    return documents

def create_vector_store(documents):
    text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
    texts = text_splitter.split_documents(documents)
    embeddings = OpenAIEmbeddings()
    vectorstore = Chroma.from_documents(documents=texts, embedding=embeddings)
    return vectorstore

def retrieve_relevant_docs(vectorstore, query, k=3):
    """
    Retrieve relevant documents for the query.
    """
    docs = vectorstore.similarity_search(query, k=k)
    return "\n".join([doc.page_content for doc in docs])

# Initialize the vector store once (for performance)
documents = load_documents()
vectorstore = create_vector_store(documents)

def get_rag_context(query):
    return retrieve_relevant_docs(vectorstore, query)