from langchain.tools import tool
from qdrant_client import QdrantClient
from langchain.embeddings.ollama import OllamaEmbeddings
from langchain.text_splitter import RecursiveCharacterTextSplitter
import os

# Initialize Qdrant client and collection
qdrant_url = os.getenv("QDRANT_URL", "http://localhost:6333")
client = QdrantClient(url=qdrant_url)
collection_name = "knowledge_base"
if not client.has_collection(collection_name):
    client.create_collection(name=collection_name, vectors_config={"size": 384, "distance": "Cosine"})

embeddings = OllamaEmbeddings(model="nomic-embed-text")
text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)

@tool("search_knowledge_base", description="Semantic search in knowledge base")
def search_knowledge_base(query: str, max_results: int = 5):
    vector = embeddings.embed_query(query)
    results = client.search(collection_name=collection_name, query_vector=vector, limit=max_results)
    return [hit.payload["text"] for hit in results]

@tool("add_to_knowledge_base", description="Add document to knowledge base")
def add_to_knowledge_base(content: str, title: str):
    docs = text_splitter.split_text(content)
    vectors = embeddings.embed_documents(docs)
    ids = [title + f"_{i}" for i in range(len(docs))]
    client.upsert(collection_name=collection_name, points=[{"id": id_, "vector": vec, "payload": {"text": doc}} for id_, vec, doc in zip(ids, vectors, docs)])
    return f"Added {len(docs)} chunks to knowledge base"