#!/usr/bin/env python3
import json
from collections import Counter

# Read the articles data
with open('articles_temp.json', 'r') as f:
    articles = json.load(f)

# Count thumbnail usage
thumbnail_counts = Counter()
thumbnail_to_articles = {}

for article in articles:
    thumbnail = article.get('thumbnail', '')
    if thumbnail:
        thumbnail_counts[thumbnail] += 1
        if thumbnail not in thumbnail_to_articles:
            thumbnail_to_articles[thumbnail] = []
        thumbnail_to_articles[thumbnail].append({
            'id': article['id'], 
            'title': article['title'][:60] + '...' if len(article['title']) > 60 else article['title'],
            'outletId': article['outletId']
        })

# Find duplicates
print("=== DUPLICATE THUMBNAILS ===")
duplicates_found = False
for thumbnail, count in thumbnail_counts.most_common():
    if count > 1:
        duplicates_found = True
        print(f"\n{thumbnail} (used {count} times):")
        for article in thumbnail_to_articles[thumbnail]:
            print(f"  - {article['outletId']}: {article['title']} (ID: {article['id']})")

if not duplicates_found:
    print("No duplicate thumbnails found!")

# Show some stats
print(f"\n=== STATISTICS ===")
print(f"Total articles: {len(articles)}")
print(f"Unique thumbnails: {len(thumbnail_counts)}")
print(f"Articles with duplicates: {sum(count for count in thumbnail_counts.values() if count > 1)}")