run.py

"""Orsync Scenarist — Local Run Script.

Prerequisites:
    Run ``python setup.py`` first to install dependencies and configure .env.

Usage:
    python run.py              # Start API server on port 7860
    python run.py --port 8000  # Custom port
    python run.py --reload     # Auto-reload on code changes (dev mode)
    python run.py --no-redis   # Skip embedded Redis startup
    python run.py --no-neo4j   # Skip embedded Neo4j startup
    python run.py --no-chroma  # Skip embedded ChromaDB startup
"""
from __future__ import annotations

import argparse
import errno
import os
import shutil
import socket
import subprocess
import sys
import time

ROOT_DIR = os.path.dirname(os.path.abspath(__file__))
PROJECT_PARENT = os.path.dirname(ROOT_DIR)
if PROJECT_PARENT not in sys.path:
    sys.path.insert(0, PROJECT_PARENT)

SETUP_MARKER = os.path.join(ROOT_DIR, ".setup_done")

_redis_proc: subprocess.Popen | None = None
_neo4j_proc: subprocess.Popen | None = None
_chroma_proc: subprocess.Popen | None = None


def _is_neo4j_ready(uri: str, username: str, password: str) -> bool:
    try:
        from neo4j import GraphDatabase

        driver = GraphDatabase.driver(uri, auth=(username, password))
        try:
            driver.verify_connectivity()
            return True
        finally:
            driver.close()
    except Exception:
        return False


def _is_chroma_ready(host: str, port: int) -> bool:
    try:
        import chromadb

        client = chromadb.HttpClient(host=host, port=port)
        client.heartbeat()
        return True
    except Exception:
        return False


def _is_port_available(host: str, port: int) -> bool:
    """Return True when the requested bind address can accept a TCP listener."""
    family = socket.AF_INET6 if ":" in host else socket.AF_INET
    probe = socket.socket(family, socket.SOCK_STREAM)
    try:
        probe.bind((host, port))
    except OSError as exc:
        if exc.errno in {errno.EADDRINUSE, errno.EACCES, 10013, 10048}:
            return False
        raise
    finally:
        probe.close()
    return True


def _display_host(host: str) -> str:
    if host in {"0.0.0.0", "::"}:
        return "127.0.0.1"
    return host


def _ensure_setup() -> None:
    """Run setup.py automatically if it hasn't been run yet."""
    if os.path.isfile(SETUP_MARKER):
        return

    print("[run] First run detected — launching setup ...\n")
    result = subprocess.run([sys.executable, os.path.join(ROOT_DIR, "setup.py")])
    if result.returncode != 0:
        print("\n[run] Setup failed. Fix the issues above and try again.")
        sys.exit(1)
    print()


def _start_redis() -> bool:
    """Start an embedded Redis server if redis-server is available locally."""
    global _redis_proc

    # First check if Redis is already running
    try:
        import redis as _redis
        r = _redis.from_url("redis://localhost:6379/0", socket_connect_timeout=1)
        r.ping()
        print("[run] Redis: already running on localhost:6379")
        return True
    except Exception:
        pass

    # Try to start redis-server
    redis_bin = shutil.which("redis-server")
    if not redis_bin:
        print("[run] Redis: redis-server not found — using in-memory fallback")
        print("[run]   Install Redis: https://redis.io/download")
        return False

    try:
        _redis_proc = subprocess.Popen(
            [redis_bin, "--daemonize", "no", "--save", "", "--appendonly", "no",
             "--maxmemory", "256mb", "--maxmemory-policy", "allkeys-lru"],
            stdout=subprocess.DEVNULL,
            stderr=subprocess.DEVNULL,
        )
        # Wait for it to be ready
        for _ in range(20):
            try:
                import redis as _redis
                r = _redis.from_url("redis://localhost:6379/0", socket_connect_timeout=1)
                r.ping()
                print("[run] Redis: started embedded server on localhost:6379")
                return True
            except Exception:
                time.sleep(0.25)
        print("[run] Redis: server started but not responding — using in-memory fallback")
        return False
    except Exception as exc:
        print(f"[run] Redis: failed to start — {exc}")
        return False


def _start_neo4j() -> bool:
    """Start an embedded Neo4j server if neo4j is available locally."""
    global _neo4j_proc

    from backend.app.core.config import settings
    neo4j_uri = settings.neo4j_uri
    neo4j_username = settings.neo4j_username
    neo4j_password = settings.neo4j_password

    # First check if Neo4j is already running
    if _is_neo4j_ready(neo4j_uri, neo4j_username, neo4j_password):
        print(f"[run] Neo4j: already running on {neo4j_uri}")
        return True

    # Find neo4j binary — check common locations
    neo4j_bin = shutil.which("neo4j")
    if not neo4j_bin:
        # Check common install paths
        for candidate in [
            "/opt/neo4j/bin/neo4j",
            os.path.expanduser("~/neo4j/bin/neo4j"),
            r"C:\Program Files\Neo4j\bin\neo4j.bat",
        ]:
            if os.path.isfile(candidate):
                neo4j_bin = candidate
                break

    if not neo4j_bin:
        print("[run] Neo4j: not found — graph queries will use no-op stub")
        print("[run]   Install: https://neo4j.com/download/")
        return False

    try:
        # Use 'neo4j console' which runs in foreground (suitable for subprocess)
        _neo4j_proc = subprocess.Popen(
            [neo4j_bin, "console"],
            stdout=subprocess.DEVNULL,
            stderr=subprocess.DEVNULL,
        )
        # Wait for bolt port to be ready
        print("[run] Neo4j: starting ...")
        for _ in range(60):
            if _is_neo4j_ready(neo4j_uri, neo4j_username, neo4j_password):
                print(f"[run] Neo4j: started on {neo4j_uri}")
                return True
            time.sleep(0.5)
        print("[run] Neo4j: started but not responding — using no-op stub")
        return False
    except Exception as exc:
        print(f"[run] Neo4j: failed to start — {exc}")
        return False


def _start_chroma() -> bool:
    """Start an embedded ChromaDB server if the chroma CLI is available."""
    global _chroma_proc

    from backend.app.core.config import settings
    chroma_bind_host = settings.chroma_host
    chroma_probe_host = _display_host(chroma_bind_host)
    chroma_port = settings.chroma_port

    # First check if ChromaDB is already running
    if _is_chroma_ready(chroma_probe_host, chroma_port):
        print(f"[run] ChromaDB: already running on {chroma_probe_host}:{chroma_port}")
        return True

    # Find chroma binary
    chroma_bin = shutil.which("chroma")
    if not chroma_bin:
        print("[run] ChromaDB: chroma CLI not found — using no-op stub")
        print("[run]   Install: pip install chromadb")
        return False

    try:
        data_dir = os.path.join(ROOT_DIR, ".chroma_data")
        os.makedirs(data_dir, exist_ok=True)
        _chroma_proc = subprocess.Popen(
            [chroma_bin, "run", "--path", data_dir, "--host", chroma_bind_host, "--port", str(chroma_port)],
            stdout=subprocess.DEVNULL,
            stderr=subprocess.DEVNULL,
        )
        print("[run] ChromaDB: starting ...")
        for _ in range(10):
            if _is_chroma_ready(chroma_probe_host, chroma_port):
                print(f"[run] ChromaDB: started on {chroma_probe_host}:{chroma_port}")
                return True
            if _chroma_proc.poll() is not None:
                print("[run] ChromaDB: process exited before becoming ready — using no-op stub")
                return False
            time.sleep(0.5)

        if _chroma_proc.poll() is None:
            print(
                "[run] ChromaDB: still warming up in background — "
                "vector search will connect automatically once ready"
            )
            return True

        print("[run] ChromaDB: process exited before becoming ready — using no-op stub")
        return False
    except Exception as exc:
        print(f"[run] ChromaDB: failed to start — {exc}")
        return False


def main() -> None:
    parser = argparse.ArgumentParser(description="Run Orsync Scenarist backend locally")
    default_host = "127.0.0.1" if os.name == "nt" else "0.0.0.0"
    parser.add_argument("--host", default=default_host, help=f"Bind address (default: {default_host})")
    parser.add_argument("--port", type=int, default=None, help="Port (default: from .env or 7860)")
    parser.add_argument("--reload", action="store_true", help="Enable auto-reload for development")
    parser.add_argument("--no-redis", action="store_true", help="Skip embedded Redis startup")
    parser.add_argument("--no-neo4j", action="store_true", help="Skip embedded Neo4j startup")
    parser.add_argument("--no-chroma", action="store_true", help="Skip embedded ChromaDB startup")
    args = parser.parse_args()

    os.chdir(ROOT_DIR)
    _ensure_setup()

    from backend.app.core.config import settings

    port = args.port or settings.port
    if not _is_port_available(args.host, port):
        display_host = _display_host(args.host)
        print(f"[run] Port {port} is already in use on {args.host}.")
        print(f"[run] Check the existing server: http://{display_host}:{port}/healthz")
        print(f"[run] Or start another instance on a different port: python run.py --port {port + 1}")
        sys.exit(1)

    # Start embedded services unless --no-* or running inside Docker
    is_docker = os.path.isfile("/.dockerenv")
    if not args.no_redis and not is_docker:
        _start_redis()
    if not args.no_neo4j and not is_docker:
        _start_neo4j()
    if not args.no_chroma and not is_docker:
        _start_chroma()

    print(f"[run] Starting Orsync Scenarist on http://{args.host}:{port}")
    print(f"[run] LLM model : {settings.ollama_model}")
    print(f"[run] Embeddings: {settings.embedding_model}")
    print(f"[run] Ollama host: {settings.ollama_host}")

    import uvicorn
    try:
        uvicorn.run(
            "backend.app.main:app",
            host=args.host,
            port=port,
            reload=args.reload,
            log_level="info",
            log_config=None,
            access_log=False,
            timeout_keep_alive=0,
        )
    finally:
        if _redis_proc is not None:
            _redis_proc.terminate()
            _redis_proc.wait(timeout=5)
        if _neo4j_proc is not None:
            _neo4j_proc.terminate()
            _neo4j_proc.wait(timeout=10)
        if _chroma_proc is not None:
            _chroma_proc.terminate()
            _chroma_proc.wait(timeout=5)


if __name__ == "__main__":
    main()