"""Tools used by the RAG agent.

Two tools are provided:

* ``search_local_kb`` – semantic search in the local ChromaDB store.
* ``web_search`` – real‑time web search via Tavily.
"""

from typing import List

from langchain_ollama import ChatOllama
from langchain_tavily import TavilySearchResults
from langchain_chroma import Chroma
from langchain_text_splitters import RecursiveCharacterTextSplitter
from langchain.docstore.document import Document

from langchain.tools import tool

# The LLM used for generating answers.  Using the same model as the embeddings
# keeps the pipeline consistent.
_llm = ChatOllama(model="llama3")

# Tavily client – the API key is read from the environment variable
# ``TAVILY_API_KEY`` by the TavilySearchResults class.
_tavily = TavilySearchResults()


@tool("search_local_kb")
def search_local_kb(query: str, top_k: int = 3, vectorstore: Chroma = None) -> str:
    """Semantic search in the local ChromaDB vector store.

    Args:
        query: User question.
        top_k: Number of results to return.
        vectorstore: The Chroma instance to query.

    Returns:
        A string containing the concatenated top results.
    """
    if vectorstore is None:
        raise ValueError("vectorstore must be provided")
    retriever = vectorstore.as_retriever(search_kwargs={"k": top_k})
    docs: List[Document] = retriever.invoke(query)
    return "\n\n".join(doc.page_content for doc in docs)


@tool("web_search")
def web_search(query: str, top_k: int = 3) -> str:
    """Search the web using Tavily.

    Args:
        query: User question.
        top_k: Number of results to return.

    Returns:
        Concatenated snippets from the search results.
    """
    results = _tavily.run(query, max_results=top_k)
    # TavilySearchResults returns a list of dicts with keys like 'title',
    # 'content', 'url'.  We return the content for simplicity.
    return "\n\n".join(r.get("content", "") for r in results)

*** End of File ***