| | | 1 | | using GistBackend.Exceptions; |
| | | 2 | | using GistBackend.Types; |
| | | 3 | | using HtmlAgilityPack; |
| | | 4 | | using static System.Net.WebUtility; |
| | | 5 | | using static GistBackend.Types.FeedType; |
| | | 6 | | using static GistBackend.Types.Language; |
| | | 7 | | using static GistBackend.Utils.RssFeedUtils; |
| | | 8 | | |
| | | 9 | | namespace GistBackend.Handlers.RssFeedHandler.Feeds; |
| | | 10 | | |
| | | 11 | | public record DarkReading : RssFeed |
| | | 12 | | { |
| | 0 | 13 | | public override Uri RssUrl => new("https://www.darkreading.com/rss.xml"); |
| | 0 | 14 | | public override Language Language => En; |
| | 0 | 15 | | public override FeedType Type => News; |
| | | 16 | | |
| | | 17 | | public override string ExtractText(string content) |
| | | 18 | | { |
| | 0 | 19 | | var doc = new HtmlDocument(); |
| | 0 | 20 | | doc.LoadHtml(content); |
| | | 21 | | |
| | 0 | 22 | | var entryContentSingleNode = |
| | 0 | 23 | | doc.DocumentNode.SelectSingleNode($"//div[{ContainsClassSpecifier("ArticleBase-BodyContent")}]"); |
| | | 24 | | string textContent; |
| | | 25 | | |
| | | 26 | | // ReSharper disable once ConditionIsAlwaysTrueOrFalseAccordingToNullableAPIContract |
| | 0 | 27 | | if (entryContentSingleNode == null) |
| | | 28 | | { |
| | 0 | 29 | | var entryContentMultiNode = |
| | 0 | 30 | | doc.DocumentNode.SelectNodes( |
| | 0 | 31 | | $"//div[{ContainsClassSpecifier("ArticleMultiSectionBody-SectionContainer")}]"); |
| | 0 | 32 | | if (entryContentMultiNode == null || entryContentMultiNode.Count == 0) |
| | | 33 | | { |
| | 0 | 34 | | throw new ExtractingEntryTextException("Missing single and multi node container elements"); |
| | | 35 | | } |
| | | 36 | | |
| | 0 | 37 | | textContent = string.Join("", entryContentMultiNode.Select(node => node.InnerText)); |
| | 0 | 38 | | if (string.IsNullOrWhiteSpace(textContent)) |
| | | 39 | | { |
| | 0 | 40 | | throw new ExtractingEntryTextException("No text found in multi node container"); |
| | | 41 | | } |
| | | 42 | | } |
| | | 43 | | else |
| | | 44 | | { |
| | 0 | 45 | | textContent = entryContentSingleNode.InnerText; |
| | | 46 | | } |
| | | 47 | | |
| | 0 | 48 | | if (string.IsNullOrWhiteSpace(textContent)) |
| | | 49 | | { |
| | 0 | 50 | | throw new ExtractingEntryTextException("No text found in container"); |
| | | 51 | | } |
| | | 52 | | |
| | 0 | 53 | | var decodedText = HtmlDecode(textContent); |
| | 0 | 54 | | return decodedText.Trim(); |
| | | 55 | | } |
| | | 56 | | } |
| | | 57 | | |