public static List<DownloadData> ParseSjOrgSite(ShowData showData, out string firstcover, UploadCache uploadCache = null)
{
if(String.IsNullOrWhiteSpace(showData?.Url) || String.IsNullOrWhiteSpace(showData.Name))
throw new ArgumentNullException();
string nextpage = showData.Url;
firstcover = "";
var episodes = new List<DownloadData>();
do
{
string cover;
var a = ParseSite(showData, nextpage, out nextpage, out cover,uploadCache);
if (firstcover == "")
{
firstcover = cover;
}
episodes.AddRange(a);
} while (nextpage != "");
return episodes;
}
private static List<DownloadData> ParseSite(ShowData showData, string url, out string nextpageurl, out string firstcover, UploadCache uploadCache)
{
nextpageurl = "";
firstcover = "";
WebResponse resp = null;
for (int i = 0; i <= 7; i++)
{
try
{
HttpWebRequest req = WebRequest.CreateHttp(url);
req.AutomaticDecompression = DecompressionMethods.GZip | DecompressionMethods.Deflate;
req.KeepAlive = false;
resp = req.GetResponse();
break;
}
catch (WebException ex)
{
if (ex.Status != WebExceptionStatus.Timeout)
throw;
}
}
if (resp == null)
{
throw new TimeoutException();
}
HtmlDocument doc = new HtmlDocument();
doc.OptionDefaultStreamEncoding=Encoding.UTF8;
doc.Load(resp.GetResponseStream());
resp.Dispose();
List<DownloadData> list = new List<DownloadData>();
HtmlNode content = doc.GetElementbyId("content");
HtmlNode nextLink = content.SelectSingleNode("//div[@class='navigation']//a[@href][@class='next']");
if (nextLink != null)
{
nextpageurl = nextLink.GetAttributeValue("href", null);
}
var posts = content.SelectNodes("div[@class='post']");
if (posts == null) return list;
foreach (var post in posts)
{
//--------------Season Title-----------------------------------
var title = post.SelectSingleNode("h2/a[@href]");
if (title == null)
{
Console.WriteLine("SjInfo Parser: No Title");
continue;
}
var seasonData = new SeasonData();
seasonData.Show = showData;
seasonData.Url = title.GetAttributeValue("href", null);
seasonData.Title = WebUtility.HtmlDecode(title.InnerText);
var postContent = post.SelectSingleNode("div[@class='post-content']");
if(postContent == null)
{
Console.WriteLine("SjInfo Parser: No Post content");
continue;
}
//----------------Season Cover------------------------------------------------
var cover = postContent.SelectSingleNode(".//p/img[@src]");
if (cover != null)
{
seasonData.CoverUrl = cover.GetAttributeValue("src", null);
if (String.IsNullOrEmpty(firstcover))
{
firstcover = seasonData.CoverUrl;
}
}
//----------------Season description-----------------------------------------
var desc = postContent.SelectSingleNode(".//p[count(node())=1][not(@class='post-info-co')]/text()");
if (desc != null)
{
seasonData.Description = WebUtility.HtmlDecode(desc.InnerText);
}
UploadData uploadData = null;
var ps = postContent.SelectNodes(".//node()[self::p|self::div][count(strong)>=2]");
if (ps == null)
{
Console.WriteLine("SjInfo Parser: no uploads/headers");
continue;
}
foreach (var p in ps)
{
//--------------- Upload Header ------------------------------
if (p.SelectSingleNode("self::node()[not(./a[@target])]") != null)
{
uploadData = new UploadData();
uploadData.Season = seasonData;
String c = WebUtility.HtmlDecode(p.InnerHtml);
MatchCollection mc = new Regex("<strong>\\s*(.+?)\\s*</strong>\\s*(.+?)\\s*(?:\\||$)").Matches(c);
foreach (Match match in mc)
{
String key = match.Groups[1].Value.ToLower();
String value = match.Groups[2].Value;
if (key.Contains("dauer") || key.Contains("runtime") || key.Contains("duration"))
{
uploadData.Runtime = value;
}
else if (key.Contains("grösse") || key.Contains("größe") || key.Contains("size"))
{
uploadData.Size = value;
}
else if (key.Contains("uploader"))
{
uploadData.Uploader = value;
}
else if (key.Contains("format"))
{
uploadData.Format = value;
}
else if (key.Contains("sprache") || key.Contains("language"))
{
value = value.ToLower();
if (value.Contains("deutsch") || value.Contains("german"))
{
uploadData.Language |= UploadLanguage.German;
}
if (value.Contains("englisch") || value.Contains("english"))
{
uploadData.Language |= UploadLanguage.English;
}
if (value.Contains("subbed"))
{
uploadData.Subbed = true;
}
}
}
} else if (uploadData != null) {
// ------------------ Links -------------------------
var ulTitle = p.SelectSingleNode("strong[position()=1][count(node())=1]/text()");
if (ulTitle == null)
{
Console.WriteLine("SjInfo Parser: No title for link? " + p.InnerHtml);
continue;
}
string titleStr = WebUtility.HtmlDecode(ulTitle.InnerText).Trim();
var links = p.SelectNodes("a[@href][following-sibling::text()]");
if (links == null) continue;
var downloads = new Dictionary<string, string>();
foreach (var link in links)
{
string ur = link.GetAttributeValue("href", null);
string keyOrg = WebUtility.HtmlDecode(link.NextSibling.InnerText.Trim());
if(keyOrg.StartsWith("|")) keyOrg = keyOrg.Substring(1).Trim();
String key = keyOrg;
int i = 1;
while (downloads.ContainsKey(key))
{
key = keyOrg + "(" + i++ + ")";
}
downloads.Add(key, ur);
}
if (titleStr.Contains("720p"))
{
uploadData.Format = "720p";
}
else if (titleStr.Contains("1080p"))
{
uploadData.Format = "1080p";
}
else if (titleStr.Contains("720i"))
{
uploadData.Format = "720i";
}
else if (titleStr.Contains("1080i"))
{
uploadData.Format = "1080i";
}
DownloadData dd = new DownloadData();
dd.Upload = uploadCache == null ? uploadData : uploadCache.GetUniqueUploadData(uploadData);
dd.Title = titleStr;
if (titleStr.ToLower().Contains("subbed"))
{
dd.Upload.Subbed = true;
}
foreach (var download in downloads)
{
dd.Links.Add(download.Key, download.Value);
}
list.Add(dd);
}
else
{
Console.WriteLine("SjInfo Parser: UploadData was null");
}
}
}
return list;
}