Newsbot.Collector/Newsbot.Collector.Services/Jobs/RssWatcherJob.cs

using System.Runtime.InteropServices;
using System.ServiceModel.Syndication;
using System.Xml;
using Microsoft.Extensions.Configuration;
using Newsbot.Collector.Database.Repositories;
using Newsbot.Collector.Domain.Consts;
using Newsbot.Collector.Domain.Interfaces;
using Newsbot.Collector.Domain.Models;

namespace Newsbot.Collector.Services.Jobs;

public class RssWatcherJobOptions
{
    public string ConnectionString { get; set; } = "";
}

// This class was made to work with Hangfire and it does not support constructors.
public class RssWatcherJob : IHangfireJob
{

    private IArticlesRepository _articles;
    private IDiscordQueueRepository _queue;
    private ISourcesRepository _source;

    public RssWatcherJob()
    {
        _articles = new ArticlesTable("");
        _queue = new DiscordQueueTable("");
        _source = new SourcesTable("");
    }

    public void InitAndExecute(RssWatcherJobOptions options)
    {
        Console.WriteLine("Job was triggered");
        Console.WriteLine("Setting up the job");
        Init(options.ConnectionString);

        var articles = new List<ArticlesModel>();

        Console.WriteLine("Requesting sources");
        var sources = _source.ListByType(SourceTypes.Rss);
        Console.WriteLine($"Got {sources.Count()} back");
        foreach (var source in sources)
        {
            Console.WriteLine("Starting to request feed to be processed");
            var results = Collect(source.Url);

            articles.AddRange(results);
        }

        UpdateDatabase(articles);
    }

    public void InitAndExecute(IConfiguration config)
    {
        // reach out to the db and find all the rss feeds
        var connectionString = config.GetConnectionString("database");
        if (connectionString is null)
        {
            connectionString = "";
        }
        Init(connectionString);

        var articles = new List<ArticlesModel>();

        var sources = _source.ListByType(SourceTypes.Rss);
        foreach (var source in sources)
        {
            var results = Collect(source.Url);

            articles.AddRange(results);
        }

        UpdateDatabase(articles);
    }

    public void Init(string connectionString)
    {
        _articles = new ArticlesTable(connectionString);
        _queue = new DiscordQueueTable(connectionString);
        _source = new SourcesTable(connectionString);
    }

    public List<ArticlesModel> Collect(string url, int sleep = 3000)
    {
        var CollectedPosts = new List<ArticlesModel>();

        using var reader = XmlReader.Create(url);
        var feed = SyndicationFeed.Load(reader);

        foreach (var post in feed.Items.ToList())
        {
            var articleUrl = post.Links[0].Uri.AbsoluteUri;

            // Check if we have seen the url before
            // If we have, skip and save the site bandwidth
            if (IsThisUrlKnown(articleUrl) == true)
            {
                continue;
            }

            var meta = new HtmlPageReader(articleUrl);

            var article = new ArticlesModel
            {
                Title = post.Title.Text,
                Tags = FetchTags(post),
                URL = articleUrl,
                PubDate = post.PublishDate.DateTime,
                Thumbnail = meta.Data.Header.Meta.Image,
                Description = meta.Data.Header.Meta.Description,
            };

            CollectedPosts.Add(article);

            // try to not be too greedy
            Thread.Sleep(sleep);
        }
        return CollectedPosts;
    }

    public void UpdateDatabase(List<ArticlesModel> items)
    {
        foreach (var item in items)
        {
            if (IsThisUrlKnown(item.URL) == false)
            {
                continue;
            }

            var p = _articles.New(item);
            _queue.New(new DiscordQueueModel
            {
                ArticleID = p.ID
            });
        }
    }

    private bool IsThisUrlKnown(string url)
    {
        var isKnown = _articles.GetByUrl(url);
        if (isKnown.URL == url)
        {
            return true;
        }
        return false;
    }

    private string FetchTags(SyndicationItem post)
    {
        string result = "";
        foreach (var tag in post.Categories)
        {
            result += $"{tag.Name},";
        }
        return result;
    }
}