Newsbot.Collector/Newsbot.Collector.Services/Jobs/GithubWatcherJob.cs

using System.ServiceModel.Syndication;
using System.Xml;
using Newsbot.Collector.Database.Repositories;
using Newsbot.Collector.Domain.Interfaces;
using Newsbot.Collector.Domain.Models;
using Newsbot.Collector.Services.HtmlParser;

namespace Newsbot.Collector.Services.Jobs;

public class GithubWatcherJobOptions
{
    public string ConnectionString { get; set; } = "";
    public bool FeaturePullReleases { get; set; } = false;
    public bool FeaturePullCommits { get; set; } = false;
    public bool PullIssues { get; set; } = false;
}

public class GithubWatcherJob
{
    private IArticlesRepository _articles;
    private IDiscordQueueRepository _queue;
    private ISourcesRepository _source;

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

    public void Init(GithubWatcherJobOptions options)
    {
        _articles = new ArticlesTable(options.ConnectionString);
        _queue = new DiscordQueueTable(options.ConnectionString);
        _source = new SourcesTable(options.ConnectionString);
    }

    public void InitAndExecute(GithubWatcherJobOptions options)
    {
        Init(options);
        Execute();
    }

    private void Execute()
    {
        // query sources for things to pull
        var items = new List<ArticlesModel>();

        items.AddRange(Collect(new Uri("https://github.com/jtom38/dvb")));

        // query */commits/master.atom
        // query */commits/main.atom
    }

    public List<ArticlesModel> Collect(Uri url)
    {
        var items = new List<ArticlesModel>();

        Guid placeHolderId = Guid.NewGuid();
        // query */release.atom
        items.AddRange(CollectItems($"{url.AbsoluteUri}/releases.atom", placeHolderId));
        items.AddRange(CollectItems($"{url.AbsoluteUri}/master.atom", placeHolderId));

        return items;
    }

    private List<ArticlesModel> CollectItems(string baseUrl, Guid sourceId)
    {
        var items = new List<ArticlesModel>();

        using var reader = XmlReader.Create(baseUrl);
        var client = SyndicationFeed.Load(reader);

        foreach (var item in client.Items)
        {
            var itemUrl = item.Links[0].Uri.AbsoluteUri;
            var exits = _articles.GetByUrl(itemUrl);
            if (exits.ID != Guid.Empty)
            {
                continue;
            }

            var parser = new HtmlPageReader(itemUrl);
            parser.Parse();

            try
            {
                var a = new ArticlesModel
                {
                    SourceID = sourceId,
                    Tags = "github",
                    Title = item.Title.Text,
                    URL = itemUrl,
                    //PubDate = item.LastUpdatedTime.DateTime,
                    Thumbnail = parser.Data.Header.Image,
                    Description = $"'dvb' has released '{item.Title.Text}'!",
                    AuthorName = item.Authors[0].Name ?? "",
                    AuthorImage = item.Authors[0].Uri ?? ""
                };
                items.Add(a);
            }
            catch (Exception e)
            {
                Console.WriteLine(e);
            }
        }
        return items;
    }
}