public WebsiteModel()
{
LastOpenConfig = 1;
// vytvořit instance subobjektů- pro případ nové website
CrawlerConfig = new CrawlerModel();
AnalyzerConfig = new AnalyzerModel();
OrganizerConfig = new OrganizerModel();
ExportConfig = new ExportModel();
}
/// <summary>
/// Creates object of type DefaultCrawler needed for Crawler process
/// </summary>
/// <param name="crawlerConfig">Crawler configuration</param>
/// <returns>
/// Crawler
/// </returns>
public DefaultCrawler ConfigureCrawler(CrawlerModel crawlerConfig)
{
var crawlerCfg = new CrawlerConfiguration
{
Uri = new Uri(crawlerConfig.SiteUrl),
CountLimit = crawlerConfig.CountLimit,
DepthLimit = crawlerConfig.DepthLimit,
IgnoredPrefixes = crawlerConfig.IgnoredPrefixes,
IgnoredPaths = crawlerConfig.IgnoredPaths
};
var directory = crawlerConfig.Directory;
var imageHandler = new ImageHandler(directory, null);
var scriptHandler = new ScriptHandler(directory, imageHandler);
var loader = new HtmlAgilityPackLoader(scriptHandler);
return new DefaultCrawler(crawlerCfg, loader);
}
