Exemple #1
0
        /// <summary>Gets resource urls.</summary>
        /// <param name="endsWith">Resource pattern if any.</param>
        private List <Uri> GetResources(string[] endsWith = null)
        {
            var res = new List <Uri>();

            foreach (var uri in this.urls)
            {
                bool isResource = UrlHelper.IsResource(uri);
                if (!isResource)
                {
                    continue;
                }

                if (Sitemap.EndsWith(uri.LocalPath, endsWith))
                {
                    res.Add(uri);
                }
            }

            return(res);
        }
Exemple #2
0
        public Scraper(Sitemap sitemap, string htmlDownloadPath, Action <double> reportProgress, CancellationToken token)
        {
            if (sitemap == null)
            {
                throw new NullReferenceException("sitemap");
            }

            if (htmlDownloadPath == null)
            {
                throw new NullReferenceException("htmlDownloadPath");
            }

            if (Directory.Exists(htmlDownloadPath))
            {
                Directory.Delete(htmlDownloadPath, true);
            }
            Directory.CreateDirectory(htmlDownloadPath);

            this.sitemap           = sitemap;
            this.htmlDownloadPath  = htmlDownloadPath;
            this.reportProgress    = reportProgress;
            this.cancellationToken = token;
        }