"""
CLI commands for variance collection.

Commands:
- nyx-probe variance collect <TERM> --runs 1000
- nyx-probe variance batch <FILE> --runs 1000
- nyx-probe variance stats <SESSION_ID>
- nyx-probe variance analyze <TERM>
"""

import click
import uuid
import json
from pathlib import Path
from rich.console import Console
from rich.table import Table

from ..core.model import NyxModel
from ..runners.variance_runner import VarianceRunner
from nyx_substrate.database import PhoebeConnection, VarianceProbeDAO

console = Console()

# Global model instance (lazy loaded)
_model = None


def get_model() -> NyxModel:
    """Get or initialize model (lazy loading)."""
    global _model
    if _model is None:
        console.print("[yellow]Loading model...[/yellow]")
        _model = NyxModel()
        _model.load()
        console.print("[green]✓ Model loaded[/green]")
    return _model


@click.group()
def variance():
    """Variance collection commands."""
    pass


@variance.command()
@click.argument("term")
@click.option("--runs", "-n", default=1000, help="Number of runs (default: 1000)")
@click.option("--max-rounds", "-r", default=3, help="Max echo rounds (default: 3)")
@click.option("--temperature", "-t", default=0.8, help="Sampling temperature (default: 0.8)")
@click.option("--max-tokens", default=50, help="Max new tokens (default: 50)")
def collect(term: str, runs: int, max_rounds: int, temperature: float, max_tokens: int):
    """
    Collect variance data for a single term.

    Runs Echo Probe N times and stores results in phoebe.

    Example:
        nyx-probe variance collect "Geworfenheit" --runs 1000
    """
    # Initialize
    model = get_model()
    conn = PhoebeConnection()
    dao = VarianceProbeDAO(conn)

    # Create runner
    runner = VarianceRunner(
        model=model,
        dao=dao,
        max_rounds=max_rounds,
        max_new_tokens=max_tokens,
        temperature=temperature,
    )

    # Run session
    session_id = runner.run_session(term, runs=runs)

    # Display stats
    runner.display_session_stats(session_id)

    # Close connection
    conn.close()


@variance.command()
@click.argument("glossary_file", type=click.Path(exists=True))
@click.option("--runs", "-n", default=1000, help="Number of runs per term (default: 1000)")
@click.option("--max-rounds", "-r", default=3, help="Max echo rounds (default: 3)")
@click.option("--temperature", "-t", default=0.8, help="Sampling temperature (default: 0.8)")
def batch(glossary_file: str, runs: int, max_rounds: int, temperature: float):
    """
    Collect variance data from a glossary file.

    Glossary file should be JSON with a "terms" key containing list of strings.

    Example:
        nyx-probe variance batch data/glossary/philosophical.json --runs 1000
    """
    # Load glossary
    with open(glossary_file, 'r') as f:
        data = json.load(f)

    terms = data.get("terms", [])
    if not terms:
        console.print("[red]Error: No terms found in glossary file[/red]")
        return

    console.print(f"[cyan]Loaded {len(terms)} terms from {glossary_file}[/cyan]\n")

    # Initialize
    model = get_model()
    conn = PhoebeConnection()
    dao = VarianceProbeDAO(conn)

    # Create runner
    runner = VarianceRunner(
        model=model,
        dao=dao,
        max_rounds=max_rounds,
        max_new_tokens=50,
        temperature=temperature,
    )

    # Run batch
    sessions = runner.run_batch(terms, runs_per_term=runs)

    # Summary table
    table = Table(title="Batch Summary")
    table.add_column("Term", style="cyan")
    table.add_column("Session ID", style="yellow")

    for term, session_id in sessions.items():
        table.add_row(term, str(session_id))

    console.print(table)

    # Close connection
    conn.close()


@variance.command()
@click.argument("session_id")
def stats(session_id: str):
    """
    Display statistics for a variance collection session.

    Example:
        nyx-probe variance stats 550e8400-e29b-41d4-a716-446655440000
    """
    # Parse UUID
    try:
        session_uuid = uuid.UUID(session_id)
    except ValueError:
        console.print(f"[red]Error: Invalid UUID: {session_id}[/red]")
        return

    # Initialize
    conn = PhoebeConnection()
    dao = VarianceProbeDAO(conn)

    # Create dummy runner just for display method
    # (we don't need the model for stats)
    class StatsRunner:
        def __init__(self, dao):
            self.dao = dao
        def get_session_summary(self, sid):
            return dao.get_session_stats(sid)
        def display_session_stats(self, sid):
            stats = self.get_session_summary(sid)
            if stats['total_runs'] == 0:
                console.print(f"[yellow]No runs found for session {sid}[/yellow]")
                return

            console.print(f"\n[bold cyan]📊 Session Statistics[/bold cyan]")
            console.print(f"Session ID: {sid}")
            console.print(f"Term: [bold]{stats['term']}[/bold]")
            console.print(f"Total runs: {stats['total_runs']}")
            console.print(f"Average depth: {stats['avg_depth']:.2f}")
            console.print(f"Average rounds: {stats['avg_rounds']:.2f}")

            console.print("\n[bold]Depth Distribution:[/bold]")
            dist = stats['depth_distribution']
            for depth_val in ['0', '1', '2', '3']:
                count = dist.get(depth_val, 0)
                pct = (count / stats['total_runs'] * 100) if stats['total_runs'] > 0 else 0
                console.print(f"  Depth {depth_val}: {count:4d} ({pct:5.1f}%)")

            console.print("\n[bold]Most Common Echo Types:[/bold]")
            for echo_info in stats['most_common_echo_types'][:5]:
                console.print(f"  {echo_info['type']:12s}: {echo_info['count']:4d}")

            console.print()

    runner = StatsRunner(dao)
    runner.display_session_stats(session_uuid)

    # Close connection
    conn.close()


@variance.command()
@click.argument("term")
@click.option("--limit", "-l", default=1000, help="Max runs to analyze (default: 1000)")
def analyze(term: str, limit: int):
    """
    Analyze variance across ALL sessions for a term.

    Shows aggregate statistics across all variance collections.

    Example:
        nyx-probe variance analyze "Geworfenheit" --limit 1000
    """
    # Initialize
    conn = PhoebeConnection()
    dao = VarianceProbeDAO(conn)

    # Get distribution
    dist = dao.get_term_distribution(term, limit=limit)

    if dist['total_runs'] == 0:
        console.print(f"[yellow]No variance data found for term: {term}[/yellow]")
        conn.close()
        return

    # Display
    console.print(f"\n[bold cyan]📊 Variance Analysis: {term}[/bold cyan]")
    console.print(f"Total runs: {dist['total_runs']}")
    console.print(f"Total sessions: {dist['total_sessions']}")
    console.print(f"Average depth: {dist['avg_depth']:.2f}")
    console.print(f"Depth std dev: {dist['depth_std_dev']:.2f}")
    console.print(f"Average rounds: {dist['avg_rounds']:.2f}")
    console.print(f"Rounds std dev: {dist['rounds_std_dev']:.2f}")

    console.print("\n[bold]Depth Distribution:[/bold]")
    depth_dist = dist['depth_distribution']
    for depth_val in ['0', '1', '2', '3']:
        count = depth_dist.get(depth_val, 0)
        pct = (count / dist['total_runs'] * 100) if dist['total_runs'] > 0 else 0
        console.print(f"  Depth {depth_val}: {count:4d} ({pct:5.1f}%)")

    console.print()

    # Close connection
    conn.close()


# Export for registration
__all__ = ["variance"]