from pathlib import Path

# ---------- LLM & Embeddings ----------
from langchain_ollama import Ollama, OllamaEmbeddings

llm = Ollama(
    model="openai/gpt-oss-20b",  # e.g. "llama3"
    base_url="http://localhost:11434",
    temperature=0.7,
)

embeddings = OllamaEmbeddings(model="nomic-embed-text")

# ---------- Qdrant Vector Store ----------
from qdrant_client import QdrantClient
from qdrant_client.http.models import Distance, VectorParams
from langchain_qdrant import QdrantVectorStore

client = QdrantClient(":memory:")  # in‑memory for demo; replace with path or URL as needed
# Determine vector size from the embedding model
sample_vector = embeddings.embed_query("test")[0]
vector_size = len(sample_vector)

client.create_collection(
    collection_name="knowledge_base",
    vectors_config=VectorParams(size=vector_size, distance=Distance.COSINE),
)

vector_store = QdrantVectorStore(
    client=client,
    collection_name="knowledge_base",
    embedding=embeddings,
)

# ---------- Text Splitter ----------
from langchain_text_splitters import RecursiveCharacterTextSplitter

splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)

# ---------- Tools ----------
from langchain.tools import tool
from langchain_core.documents import Document

@tool("search_knowledge_base")
def search_knowledge_base(query: str, max_results: int = 5) -> str:
    """Search the knowledge base for relevant documents."""
    docs = vector_store.similarity_search_with_score(query, k=max_results)
    if not docs:
        return "No results found."
    response_lines = []
    for i, (doc, score) in enumerate(docs, start=1):
        title = doc.metadata.get("title", "Untitled")
        snippet = doc.page_content[:200] + ("..." if len(doc.page_content) > 200 else "")
        response_lines.append(f"{i}. [{score:.2f}] {title}\n{snippet}")
    return "\n\n".join(response_lines)

@tool("add_to_knowledge_base")
def add_to_knowledge_base(content: str, title: str = "Untitled") -> str:
    """Add a new document to the knowledge base."""
    chunks = splitter.split_text(content)
    docs = [Document(page_content=c, metadata={"title": title}) for c in chunks]
    vector_store.add_documents(docs)
    return f"Added {len(chunks)} chunks under title '{title}'."

# ---------- Agent ----------
from langchain.agents import create_agent
from langchain_core.messages import HumanMessage

agent = create_agent(
    model=llm,
    tools=[search_knowledge_base, add_to_knowledge_base],
    system_message="You are a helpful assistant that can search and update the knowledge base.",
)

# ---------- Document Loader ----------
def load_documents_from_dir(directory: str):
    """Load all .txt files from directory into vector store."""
    for file_path in Path(directory).glob("*.txt"):
        text = file_path.read_text(encoding="utf-8")
        add_to_knowledge_base(text, title=file_path.stem)

# ---------- CLI ----------
def main():
    print(
        "Welcome to the RAG Agent. Commands:\n"
        "/add <title> <file>\n"
        "/search <query>\n"
        "/quit\n"
    )
    while True:
        user_input = input("> ").strip()
        if not user_input:
            continue
        if user_input.lower() in ("/quit", "exit"):
            break

        if user_input.startswith("/add"):
            try:
                _, title, file_path = user_input.split(maxsplit=2)
                content = Path(file_path).read_text(encoding="utf-8")
                print(add_to_knowledge_base(content, title))
            except Exception as e:
                print(f"Error adding document: {e}")

        elif user_input.startswith("/search"):
            query = user_input[len("/search") :].strip()
            if not query:
                print("Please provide a search query.")
                continue
            result = agent.invoke({"messages": [HumanMessage(content=query)]})
            for msg in result["messages"]:
                if hasattr(msg, "content"):
                    print(msg.content)

        else:
            # Treat as normal chat message
            result = agent.invoke({"messages": [HumanMessage(content=user_input)]})
            for msg in result["messages"]:
                if hasattr(msg, "content"):
                    print(msg.content)

if __name__ == "__main__":
    # Optional: load initial docs from a folder named 'docs'
    if Path("docs").exists():
        load_documents_from_dir("docs")
    main()