private void CrawlDetails(CrawledMovie movie) { string html = CrawlerUtility.Crawl("http://www.paradisbio.dk/" + movie.CrawledMovieUrl); const string detailsSelector = @"table#top_tabel td.forside_bg table table"; var htmlDocument = new HtmlDocument(); htmlDocument.LoadHtml(html); HtmlNode detailsNode = htmlDocument.DocumentNode.QuerySelectorAll(detailsSelector).Last(); foreach (HtmlNode rowNode in detailsNode.QuerySelectorAll("tr")) { IEnumerable<HtmlNode> columnNodes = rowNode.QuerySelectorAll("td"); string key = columnNodes.First().InnerText; if (key.StartsWith("Originaltitel")) movie.OriginalTitle = columnNodes.Last().InnerText; if (key.StartsWith("Læs mere")) { foreach (HtmlNode link in columnNodes.Last().QuerySelectorAll("a")) { if (link.InnerText.Contains("IMDb")) { string imdbUrl = link.Attributes["href"].Value; movie.ImdbId = GoogleImdbCrawler.ExtractImdbId(imdbUrl); } } } } }
private void CrawlImdbId(CrawledMovie movie) { movie.ImdbId = _googleImdbCrawler.QueryImdbId(movie.OriginalTitle ?? movie.Title); }