// MongoDB script to fix article categories
// Run with: docker exec -i munich-news-mongodb mongosh -u admin -p changeme --authenticationDatabase admin < fix-categories.js

use munich_news

print("=== RSS Feeds and their categories ===");
db.rss_feeds.find({}, {name: 1, category: 1, _id: 0}).forEach(feed => {
    print(`${feed.name}: ${feed.category || 'NO CATEGORY'}`);
});

print("\n=== Current article category distribution ===");
db.articles.aggregate([
    {$group: {_id: "$category", count: {$sum: 1}}},
    {$sort: {count: -1}}
]).forEach(result => {
    print(`${result._id || 'null'}: ${result.count} articles`);
});

print("\n=== Fixing null categories ===");

// Update articles based on their RSS feed source
var feedsUpdated = 0;
db.rss_feeds.find().forEach(function(feed) {
    if (feed.category) {
        var result = db.articles.updateMany(
            {source: feed.name, category: null},
            {$set: {category: feed.category}}
        );
        if (result.modifiedCount > 0) {
            print(`Updated ${result.modifiedCount} articles from ${feed.name} to category: ${feed.category}`);
            feedsUpdated += result.modifiedCount;
        }
    }
});

// Set remaining null categories to 'general'
var remainingNull = db.articles.updateMany(
    {category: null},
    {$set: {category: "general"}}
);

if (remainingNull.modifiedCount > 0) {
    print(`Set ${remainingNull.modifiedCount} remaining null articles to 'general'`);
}

print(`\nTotal articles updated: ${feedsUpdated + remainingNull.modifiedCount}`);

print("\n=== Updated article category distribution ===");
db.articles.aggregate([
    {$group: {_id: "$category", count: {$sum: 1}}},
    {$sort: {count: -1}}
]).forEach(result => {
    print(`${result._id || 'null'}: ${result.count} articles`);
});

print("\n✓ Done!");