public void CanParseHtmlPage()
{
var path = GetFullFilePath("Test_data\\page.html");
var html = File.ReadAllText(path);
var parser = new WykopTitlesExtractor();
var headers = parser.ExtractTitles(html);
Assert.Equal(35, headers.Count);
}
private static void TimerOnElapsed(object sender, ElapsedEventArgs e)
{
Task.Run(async() =>
{
var supplier = new ScratchSupplier("https://wykop.pl");
var htmlPage = await supplier.GetPage();
var extractor = new WykopTitlesExtractor();
var headers = extractor.ExtractNews(htmlPage);
bootstrapper.PublishNews(headers);
}).GetAwaiter().GetResult();
}
public void CanParseNewsLines()
{
var path = GetFullFilePath("Test_data\\page.html");
var html = File.ReadAllText(path);
var parser = new WykopTitlesExtractor();
var titles = parser.ExtractNews(html);
foreach (var newsLine in titles)
{
Assert.NotEmpty(newsLine.Title);
Assert.NotEmpty(newsLine.ImgUrl);
Assert.True(newsLine.Tags.Any());
Assert.NotEmpty(newsLine.Attributes.Single(e => e.Key == "id").Value);
var time = newsLine.Attributes.SingleOrDefault(a => a.Key == "time").Value;
Assert.True(DateTime.TryParse(time, out DateTime dt));
}
}
