code/src/api/main.py

# src/api/main.py
"""
FastAPI application with Agentic Adaptive RAG System.

Backends: Qdrant (hybrid search), Neo4j (knowledge graph), Mem0 (memory)
Agent:   LangGraph state machine with self-reflective retrieval loop
"""

from fastapi import FastAPI
from fastapi.middleware.cors import CORSMiddleware
import logging
import sys
import os

sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname(__file__))))

from src.api.routes import router, app_state

logging.basicConfig(
    level=logging.INFO,
    format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
)
logger = logging.getLogger(__name__)

app = FastAPI(
    title="CVE-KGRAG Agentic RAG API",
    description="Hybrid search + knowledge graph + agentic RAG with long-term memory",
    version="3.0.0-agentic",
)

app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

app.include_router(router, prefix="/api/v1")


@app.on_event("startup")
async def startup():
    logger.info("CVE-KGRAG Agentic RAG API starting up ...")

    # ── 1. Qdrant RAG system ──────────────────────────────────────────────────
    try:
        from src.generators.rag_system import CVERAGSystem
        rag_system = CVERAGSystem()
        app_state["rag_system"] = rag_system
        stats = rag_system.get_collection_stats()
        total = stats.get("total_documents", 0)
        if total == 0:
            logger.warning("Qdrant collections empty. Run: python -m src.generators.rag_system --build")
        else:
            logger.info("Qdrant ready: %s total chunks  %s", f"{total:,}", stats.get("collections", {}))
    except Exception as e:
        logger.error("Qdrant init failed: %s", e)
        app_state["rag_system"] = None

    # ── 2. Neo4j graph service ────────────────────────────────────────────────
    try:
        from src.constructors.neo4j_graph_service import Neo4jGraphService
        graph_service = Neo4jGraphService()
        graph_service.driver.verify_connectivity()
        app_state["graph_service"] = graph_service
        neo4j_stats = graph_service.get_stats()
        logger.info("Neo4j ready: CVE nodes=%s", f"{neo4j_stats.get('CVE', 0):,}")
    except Exception as e:
        logger.warning("Neo4j unavailable: %s. Graph endpoints → 503.", e)
        app_state["graph_service"] = None

    # ── 3. LLM client ─────────────────────────────────────────────────────────
    try:
        from llms.factory import LLMFactory
        from src.generators.rag_config import get_llm_config
        llm_config = get_llm_config()
        if llm_config.get("enabled"):
            app_state["llm_client"] = LLMFactory.create_from_config(llm_config)
            if app_state["llm_client"]:
                info = app_state["llm_client"].get_model_info()
                logger.info("LLM ready: provider=%s model=%s", info.get("provider"), info.get("model_name"))
                from src.agents.llm_runnable import BaseLLMRunnable
                app_state["llm_runnable"] = BaseLLMRunnable(app_state["llm_client"])
            else:
                logger.warning("LLM config enabled but init failed.")
                app_state["llm_runnable"] = None
        else:
            logger.info("LLM not enabled. Set LLM_ENABLED=true.")
            app_state["llm_client"] = None
            app_state["llm_runnable"] = None
    except Exception as e:
        logger.warning("LLM client unavailable: %s", e)
        app_state["llm_client"] = None
        app_state["llm_runnable"] = None

    # ── 4. Reranker (grading backend) ────────────────────────────────────────
    try:
        from src.agents.reranker import load_reranker
        app_state["reranker"] = load_reranker()
        if app_state["reranker"] is not None:
            logger.info("Reranker ready: backend=%s threshold=%.2f",
                        app_state["reranker"].backend, app_state["reranker"].threshold)
    except Exception as e:
        logger.warning("Reranker init failed: %s", e)
        app_state["reranker"] = None

    # ── 5. Mem0 long-term memory ──────────────────────────────────────────────
    try:
        from src.agents.memory import CVEKGMemory
        memory = CVEKGMemory()
        app_state["memory"] = memory
        logger.info("Mem0 memory: %s", "enabled" if memory.enabled else "disabled")
    except Exception as e:
        logger.warning("Mem0 init failed: %s — memory disabled", e)
        app_state["memory"] = None

    # ── 5.5. Cheat-sheet taxonomy (CWE↔CAPEC↔MITRE) ─────────────────────────
    try:
        from pathlib import Path
        from src.agents.cheat_sheet import init_cheat_sheet
        project_root = Path(__file__).resolve().parents[2]
        app_state["cheat_sheet"] = init_cheat_sheet(project_root)
        logger.info("Cheat sheet loaded ✓")
    except Exception as e:
        logger.warning("Cheat sheet init failed: %s — running without taxonomy hints", e)
        app_state["cheat_sheet"] = None

    # ── 6. Agent tools ────────────────────────────────────────────────────────
    try:
        from src.agents.tools import create_agent_tools
        if app_state["rag_system"]:
            app_state["agent_tools"] = create_agent_tools(
                app_state["rag_system"],
                app_state.get("graph_service"),
                llm_client=app_state.get("llm_client"),
                cheat_sheet=app_state.get("cheat_sheet"),
            )
            logger.info("Agent tools: %d ready", len(app_state["agent_tools"]))
        else:
            app_state["agent_tools"] = []
    except Exception as e:
        logger.warning("Agent tools init failed: %s", e)
        app_state["agent_tools"] = []

    # ── 6.5. Chat-model adapter for ReAct retrieve ───────────────────────────
    try:
        from src.agents.llm_runnable import get_chat_model
        app_state["chat_model"] = get_chat_model(app_state.get("llm_client"))
        if app_state["chat_model"] is not None:
            logger.info("Chat model adapter ready (ReAct retrieve enabled)")
        else:
            logger.info("No chat model adapter — retrieve will use legacy pipeline")
    except Exception as e:
        logger.warning("Chat model adapter init failed: %s", e)
        app_state["chat_model"] = None

    # ── 7. Compile agent graph ────────────────────────────────────────────────
    try:
        from src.agents.adaptive_rag import compile_agent_graph
        if app_state["rag_system"]:
            app_state["agent_graph"] = compile_agent_graph(
                llm_client=app_state.get("llm_client"),
                rag_system=app_state["rag_system"],
                graph_service=app_state.get("graph_service"),
                memory=app_state.get("memory"),
                llm_runnable=app_state.get("llm_runnable"),
                reranker=app_state.get("reranker"),
                cheat_sheet=app_state.get("cheat_sheet"),
                agent_tools=app_state.get("agent_tools"),
                chat_model=app_state.get("chat_model"),
            )
            logger.info("Agent graph compiled ✓  (10 nodes, self-reflective, adaptive retrieval)")
        else:
            app_state["agent_graph"] = None
    except Exception as e:
        logger.warning("Agent graph compilation failed: %s", e)
        app_state["agent_graph"] = None

    logger.info("API ready — endpoints: /api/v1/{health,search,query,query/agent,query/stream,summary,stats,graph/*}")


@app.on_event("shutdown")
async def shutdown():
    logger.info("CVE-KGRAG Agentic RAG API shutting down ...")


if __name__ == "__main__":
    import uvicorn
    uvicorn.run(
        "src.api.main:app",
        host="0.0.0.0",
        port=8000,
        reload=True,
        log_level="info",
    )