inspect_dataset.py

"""
Inspect and verify the games dataset loader and normalizer.

Run this script to:
1. Load the dataset
2. Normalize records
3. Print one normalized record per game type
4. Verify schema consistency
"""

import json
import sys
from pathlib import Path
from app.services.retrieval import load_games_dataset, normalize_game_record

def main():
    # Load dataset
    dataset_path = Path("app/data/games_dataset.json")
    print(f"Loading dataset from: {dataset_path}")
    
    try:
        raw_records = load_games_dataset(str(dataset_path))
    except FileNotFoundError:
        print(f"ERROR: Dataset not found at {dataset_path}")
        sys.exit(1)
    
    print(f"\n✓ Loaded {len(raw_records)} game records\n")
    
    # Normalize and group by game type
    normalized_records = []
    game_types_seen = set()
    
    for record in raw_records:
        try:
            normalized = normalize_game_record(record)
            normalized_records.append(normalized)
        except Exception as e:
            print(f"ERROR normalizing record {record.get('id')}: {e}")
            continue
    
    print(f"✓ Normalized {len(normalized_records)} records\n")
    
    # Print summary
    print("=" * 80)
    print("DATASET SUMMARY")
    print("=" * 80)
    
    game_types = {}
    difficulties = set()
    age_groups = set()
    locations = set()
    
    for norm in normalized_records:
        gt = norm['game_type']
        game_types[gt] = game_types.get(gt, 0) + 1
        difficulties.add(norm['difficulty'])
        age_groups.add(norm['age_group'])
        locations.add((norm['city'], norm['area']))
    
    print(f"\nGame Types: {dict(game_types)}")
    print(f"Difficulties: {sorted(difficulties)}")
    print(f"Age Groups: {sorted(age_groups)}")
    print(f"Unique Locations: {len(locations)}")
    for city, area in sorted(locations):
        print(f"  - {city}: {area}")
    
    # Print one example per game type
    print("\n" + "=" * 80)
    print("SAMPLE NORMALIZED RECORDS (one per game type)")
    print("=" * 80)
    
    printed = set()
    for norm in normalized_records:
        gt = norm['game_type']
        if gt not in printed:
            print(f"\n--- {gt.upper()} (ID: {norm['id']}) ---")
            print(f"City: {norm['city']}")
            print(f"Area: {norm['area']}")
            print(f"Duration: {norm['duration_minutes']} min | Players: {norm['num_players']}")
            print(f"Difficulty: {norm['difficulty']} | Age Group: {norm['age_group']}")
            print(f"Tasks: {norm['num_tasks']} | Rules: {norm['num_rules']} | Hints: {norm['num_hints']}")
            print(f"Location Type: {norm['location_type']}")
            print(f"Safety: {norm['is_safe']} | Quality Score: {norm['quality_score']}")
            print(f"Notes: {norm['notes']}")
            
            # Print first task as example
            if norm['tasks']:
                task = norm['tasks'][0]
                print(f"\nFirst Task Example:")
                print(f"  Task ID: {task.get('task_id')}")
                print(f"  Description: {task.get('description')[:80]}...")
                print(f"  Points: {task.get('points')} | Time: {task.get('time_limit_minutes')} min")
            
            printed.add(gt)
    
    # Schema validation
    print("\n" + "=" * 80)
    print("SCHEMA VALIDATION")
    print("=" * 80)
    
    required_fields = [
        'id', 'game_type', 'city', 'area', 'location_type', 
        'duration_minutes', 'num_players', 'difficulty', 'age_group',
        'num_tasks', 'task_ids', 'num_rules', 'num_hints', 
        'is_safe', 'quality_score'
    ]
    
    all_valid = True
    for norm in normalized_records:
        for field in required_fields:
            if field not in norm:
                print(f"✗ Record {norm.get('id')} missing field: {field}")
                all_valid = False
    
    if all_valid:
        print("✓ All records have required fields")
    
    print("\n" + "=" * 80)
    print("INSPECTION COMPLETE")
    print("=" * 80)

if __name__ == "__main__":
    main()