public void Download(DownloadServiceOption options)
{
var worker = new WebRequestWorker();
foreach (var urlAndPattern in options.UrlsAndPatterns)
{
var uri = urlAndPattern.Key;
var configFile = urlAndPattern.Value;
Console.WriteLine("Start processing " + uri);
var categoryResult = worker.DownloadResponse(new CrawlingOption(uri));
if (categoryResult.StatusCode == HttpStatusCode.OK)
{
var configs = SerializationHelper.DeserializeFrom<List<DataItem>>(configFile);
var pageLinkPattern = configs.Get(Constants.PagePattern).Value;
var fileLinkPattern = configs.Get(Constants.FilePattern).Value;
var categoryPages = GetArticleLinks(categoryResult.ReadAsText(), pageLinkPattern);
ProcessAllPages(categoryPages, fileLinkPattern, options.TargetFolder);
}
else
{
Console.WriteLine(categoryResult.StatusDescription);
}
}
}
public static IHtmlString GetRemoteContent(this HtmlHelper html, string fileUri)
{
if (string.IsNullOrWhiteSpace(fileUri))
{
return MvcHtmlString.Empty;
}
var fileContent = new WebRequestWorker().DownloadResponse(new CrawlingOption(fileUri)).ReadAsText();
return html.Raw(fileContent);
}
