faq_retriever.py

from langchain.docstore.document import Document
from langchain_community.retrievers import BM25Retriever
from smolagents import Tool
from faq_data import faq_entries

# Convert the FAQ data into LangChain Documents
docs = [
    Document(
        page_content="\n".join([
            f"Category: {faq['category']}",
            f"Question: {faq['question']}",
            f"Answer: {faq['answer']}"
        ]),
        metadata={"category": faq["category"]}
    )
    for faq in faq_entries
]

# Create the custom FAQ retriever tool
class FAQRetrieverTool(Tool):
    name = "faq_retriever"
    description = "Answers company FAQ questions using internal documentation."
    inputs = {
        "query": {
            "type": "string",
            "description": "The user question to answer from internal FAQ documents."
        }
    }
    output_type = "string"

    def __init__(self, docs):
        self.is_initialized = True  # <-- questo mancava!
        self.retriever = BM25Retriever.from_documents(docs)

    def forward(self, query: str):
        results = self.retriever.get_relevant_documents(query)
        if results:
            return "\n\n".join([doc.page_content for doc in results[:3]])
        else:
            return "Sorry, I couldn't find a matching FAQ."


# Initialize the retriever tool
faq_tool = FAQRetrieverTool(docs)