#!/usr/bin/env python3
"""
Diagnose why data appears in API but not in costs.json.
"""

import json
from pathlib import Path


def main():
    target_email = 'ratoui@sonymusic-pde.com'
    target_date = '2025-10-18'

    print(f'=== Diagnosing missing data for {target_email} on {target_date} ===\n')

    # Step 1: Load users and find ratoui's user ID
    print('Step 1: Loading user metadata...')
    with open('output/usage/raw/users.json') as f:
        users_data = json.load(f)

    target_user = None
    for user in users_data['data']:
        if user['email'] == target_email:
            target_user = user
            break

    if not target_user:
        print(f'ERROR: User {target_email} not found!')
        return

    target_user_id = target_user['id']
    print(f'  User ID: {target_user_id}')
    print(f'  Name: {target_user["name"]}')

    # Step 2: Load API keys and find ratoui's keys
    print('\nStep 2: Loading API keys...')
    with open('output/usage/raw/api_keys.json') as f:
        api_keys_data = json.load(f)

    user_api_keys = []
    for key in api_keys_data['data']:
        if key.get('created_by', {}).get('id') == target_user_id:
            user_api_keys.append(key)

    user_key_ids = {key['id'] for key in user_api_keys}
    print(f'  Found {len(user_api_keys)} API keys')
    for key in user_api_keys:
        print(f'    - {key["name"]} ({key["id"]}) - {key["status"]}')

    # Step 3: Check raw usage file for target date
    print(f'\nStep 3: Checking raw usage file for {target_date}...')
    raw_file = Path(f'output/usage/raw/messages_{target_date}.json')

    if not raw_file.exists():
        print(f'  ERROR: Raw file {raw_file} does not exist!')
        print('  This means the date was never fetched. Run: make fetch-costs-update')
        return

    with open(raw_file) as f:
        raw_usage = json.load(f)

    # Parse the raw file structure
    usage_records = []
    data_array = raw_usage.get('data', [])
    if data_array and len(data_array) > 0:
        usage_records = data_array[0].get('results', [])

    print(f'  Raw file has {len(usage_records)} total usage records')

    # Find ratoui's records
    ratoui_records = [r for r in usage_records if r.get('api_key_id') in user_key_ids]
    print(f'  Found {len(ratoui_records)} records for ratoui')

    if ratoui_records:
        for record in ratoui_records:
            key_id = record['api_key_id']
            key_name = 'Unknown'
            for key in user_api_keys:
                if key['id'] == key_id:
                    key_name = key['name']
                    break

            cache_creation = record.get('cache_creation', {})
            cache_tokens = cache_creation.get('ephemeral_1h_input_tokens', 0) + cache_creation.get('ephemeral_5m_input_tokens', 0)

            print(f'\n    Record for key: {key_name}')
            print(f'      Model: {record["model"]}')
            print(f'      Uncached input: {record.get("uncached_input_tokens", 0):,}')
            print(f'      Cache creation: {cache_tokens:,}')
            print(f'      Cache read: {record.get("cache_read_input_tokens", 0):,}')
            print(f'      Output: {record.get("output_tokens", 0):,}')
    else:
        print('  ❌ No records found for ratoui in raw file!')
        return

    # Step 4: Check costs.json
    print('\nStep 4: Checking costs.json...')
    costs_file = Path('output/usage/costs.json')

    if not costs_file.exists():
        print(f'  ERROR: {costs_file} does not exist!')
        print('  Run: make regenerate-costs')
        return

    with open(costs_file) as f:
        costs_data = json.load(f)

    # Find ratoui's API keys in costs.json
    print(f'  Checking {len(costs_data["api_keys"])} API keys in costs.json...')

    found_in_costs = False
    for api_key_data in costs_data['api_keys']:
        if api_key_data['api_key_id'] in user_key_ids:
            found_in_costs = True
            print(f'\n  ✓ Found key in costs.json: {api_key_data["name"]}')
            print(f'    API Key ID: {api_key_data["api_key_id"]}')
            print(f'    Total cost: ${api_key_data.get("total_cost", 0):.2f}')
            print(f'    Created by: {api_key_data.get("created_by_name")} ({api_key_data.get("created_by_email")})')

            # Check time periods
            time_periods = api_key_data.get('time_periods', [])
            print(f'    Time periods: {len(time_periods)}')

            # Look for target date
            target_period = None
            for period in time_periods:
                if period['time_period'] == target_date:
                    target_period = period
                    break

            if target_period:
                print(f'\n    ✓ Found data for {target_date}:')
                print(f'      Cost: ${target_period["period_cost"]:.2f}')
                print(f'      Models: {len(target_period["models"])}')
                for model_data in target_period['models']:
                    print(f'        - {model_data["model"]}: ${model_data["cost"]["total"]:.2f}')
            else:
                print(f'\n    ❌ NO data for {target_date} in time_periods!')
                print('    Available periods:')
                for period in time_periods[:5]:  # Show first 5
                    print(f'      - {period["time_period"]}: ${period["period_cost"]:.2f}')

    if not found_in_costs:
        print('  ❌ No keys found for ratoui in costs.json!')
        print('  This means the rebuild process is filtering out the data.')

    # Step 5: Summary
    print('\n=== Summary ===')
    print(f'Raw file exists: {raw_file.exists()}')
    print(f'Records in raw file: {len(ratoui_records)}')
    print(f'Keys in costs.json: {found_in_costs}')

    if len(ratoui_records) > 0 and not found_in_costs:
        print('\n⚠️  DATA LOSS DETECTED!')
        print('Data exists in raw file but missing from costs.json.')
        print('Possible causes:')
        print('  1. The rebuild script is not processing the raw file correctly')
        print('  2. The API key ID is not matching between raw data and metadata')
        print('  3. The date/time grouping is incorrect')
        print('\nTo debug further:')
        print('  1. Run: make regenerate-costs')
        print('  2. Check the console output for warnings')
        print('  3. Verify API key IDs match between raw files')


if __name__ == '__main__':
    main()