#!/usr/bin/env python3
"""Sync dashboard data — generates dashboard-data.json from evolution.json (real data)."""
import json, os
from datetime import datetime, timezone

def build_dashboard_data():
    # Load real evolution data
    with open('agent-evolution/data/evolution.json') as f:
        evo = json.load(f)

    # Load model benchmarks for supplemental info
    try:
        with open('agent-evolution/data/model-benchmarks-verified.json') as f:
            bm = json.load(f)
    except:
        bm = {'models': []}

    # Build model lookup
    bench_map = {}
    for m in bm.get('models', []):
        mid = m.get('id', '')
        if mid:
            bench_map[f'ollama-cloud/{mid}'] = {
                'swe': m.get('swe_bench'),
                'name': m.get('name', mid),
                'params': m.get('parameters', ''),
                'ctx': m.get('context_window', '')
            }

    agents = []
    for name, a in evo.get('agents', {}).items():
        c = a.get('current', {})
        if not c.get('model'):
            continue
        model = c['model']
        b = c.get('benchmark', {})
        fit = b.get('fit_score', 0)
        if_ = b.get('instruction_following', 0)
        model_short = model.replace('ollama-cloud/', '')
        b_info = bench_map.get(model, {})

        # Get latest model change from history
        history = a.get('history', [])
        latest_change = None
        for h in reversed(history):
            if h.get('type') == 'model_change':
                latest_change = h
                break

        agents.append({
            'name': name,
            'model': model,
            'mode': c.get('mode', 'subagent'),
            'description': c.get('description', ''),
            'category': c.get('category', 'General'),
            'color': c.get('color', '#8B5CF6'),
            'provider': c.get('provider', 'Ollama'),
            'variant': c.get('variant', ''),
            'fit_score': fit,
            'instruction_following': if_,
            'swe_bench': b_info.get('swe'),
            'model_short': model_short,
            'model_name': b_info.get('name', model_short),
            'model_params': b_info.get('params', ''),
            'model_ctx': b_info.get('ctx', ''),
            'recommendations': len(c.get('recommendations', [])),
            'history_count': len(history),
            'latest_change': latest_change
        })

    # Build model summary from REAL agent assignments
    models = {}
    for a in agents:
        ms = a['model_short']
        if ms not in models:
            models[ms] = {
                'fit_agents': [],  # fit scores of agents on this model
                'name': a['model_name'],
                'params': a['model_params'],
                'ctx': a['model_ctx'],
                'agents': 0,
                'commands': 0,
                'avg_fit': 0
            }
        models[ms]['agents'] += 1
        models[ms]['fit_agents'].append(a['fit_score'])

    # Calculate averages
    for ms, m in models.items():
        fits = m['fit_agents']
        m['avg_fit'] = round(sum(fits) / len(fits), 1) if fits else 0
        m['min_fit'] = min(fits) if fits else 0
        m['max_fit'] = max(fits) if fits else 0
        del m['fit_agents']

    # Model distribution
    from collections import Counter
    model_dist = Counter(a['model'] for a in agents)

    return {
        'generated': datetime.now(timezone.utc).isoformat().replace('+00:00', 'Z'),
        'source': 'evolution.json',
        'total_agents': len(agents),
        'total_models': len(models),
        'agents': agents,
        'models': models,
        'model_distribution': dict(model_dist)
    }

if __name__ == '__main__':
    data = build_dashboard_data()
    out = 'agent-evolution/data/dashboard-data.json'
    os.makedirs(os.path.dirname(out), exist_ok=True)
    with open(out, 'w') as f:
        json.dump(data, f, ensure_ascii=False, indent=2)
    print(f'Written {out}: {len(data["agents"])} agents, {len(data["models"])} models')
    print('Sample agents:')
    for a in data['agents'][:5]:
        print(f'  {a["name"]:25} fit={a["fit_score"]:3d} if={a["instruction_following"]:3d} {a["model_short"]:25}')
    print('Models:')
    for ms, m in sorted(data['models'].items(), key=lambda x: -x[1]['agents']):
        print(f'  {ms:20} avg_fit={m["avg_fit"]:5.1f} agents={m["agents"]}')