private Offer GetFullOffer(Offer teaser) { var client = new HtmlClient(); var html = client.GetHtml(teaser.Url); var dom = new CQ(html); var header = dom[".wspolny_naglowek_tytul"][0].InnerHTML; if (header.Contains("PRYWATNA")) { teaser.PrivateOffer = true; } var fullDescription = dom[".pokaz_ogloszenie_tresc"]; for (var i = 0; i < fullDescription.Length; i++) { teaser.Description = TextHelper.CleanText(fullDescription.RenderSelection()); } var kontakt = dom["ul.pokaz_ogloszenie"][0].OuterHTML; Regex rgx = new Regex("<script.+script>", RegexOptions.Singleline); Match match = rgx.Match(kontakt); if (match.Success) kontakt = rgx.Replace(kontakt, ""); teaser.Description += kontakt; var pictureEls = dom["img.pokaz_ogloszenie_obrazek"]; var pictures = new List<string>(); foreach (var picture in pictureEls) { var pictureUrl = "http://ogloszenia.przemysl.pl/" + picture.ParentNode.Attributes["href"]; pictures.Add(pictureUrl); } if (pictures.Count > 0) { teaser.Pictures = pictures; } teaser.Teaser = false; return teaser; }
private TeaserCrawrlResult GetTeasers(string url) { var result = new TeaserCrawrlResult(); var client = new HtmlClient(); var html = client.GetHtml(url); var dom = new CQ(html); result.TeaserUrls = GetUrls(dom).ToList(); result.Teasers = GetTeasers(dom); return result; }