//parse page(s) containing relations
/// <summary>
/// Parses page containing the relation data
/// </summary>
/// <param name="url">URL of the page to parse</param>
/// <param name="Showname">Showname</param>
public static void GetRelations(string url, string Showname, TvShowInfoProvider provider, BackgroundWorker worker, DoWorkEventArgs dwea)
{
if (provider == null) {
log.Error("GetRelations: No relation provider found/selected");
return;
}
if (String.IsNullOrEmpty(url)) {
log.Error("GetRelations: URL is null or empty");
return;
}
if (String.IsNullOrEmpty(Showname)) {
log.Error("GetRelations: Showname is null or empty");
return;
}
log.Debug("Trying to get relations from " + url);
//if episode infos are stored on a new page for each season, this should be marked with %S in url, so we can iterate through all those pages
int season = 1;
string url2 = url;
//Create new RelationCollection
TvShow rc = new TvShow(Showname);
while (true) {
if (url2.Contains("%S")) {
url = url2.Replace("%S", season.ToString());
}
if (url == null || url == "")
return;
// request
url = System.Web.HttpUtility.UrlPathEncode(url);
log.Debug("Trying to get relations for season " + season + " from " + url);
HttpWebRequest requestHtml = null;
try {
requestHtml = (HttpWebRequest)(HttpWebRequest.Create(url));
} catch (Exception ex) {
log.Error(ex.Message);
requestHtml.Abort();
return;
}
requestHtml.Timeout = Helper.ReadInt(AppProperties.CONNECTION_TIMEOUT_KEY);
// get response
HttpWebResponse responseHtml = null;
try {
responseHtml = (HttpWebResponse)(requestHtml.GetResponse());
} catch (WebException ex) {
//Serienjunkies returns "(300) Mehrdeutige Umleitung" when an inexistant season is requested
if (ex.Message.Contains("(300)"))
break;
log.Error(ex.Message);
if (responseHtml != null) {
responseHtml.Close();
}
return;
}
log.Debug("Response URL: " + responseHtml.ResponseUri.AbsoluteUri);
//if we get redirected, lets assume this page does not exist
if (responseHtml.ResponseUri.AbsoluteUri != url) {
log.Debug("Response URL doesn't match request URL, page doesn't seem to exist");
responseHtml.Close();
requestHtml.Abort();
break;
}
// and download
//Logger.Instance.LogMessage("charset=" + responseHtml.CharacterSet, LogLevel.INFO);
Encoding enc;
if (provider.Encoding != null && provider.Encoding != "") {
try {
enc = Encoding.GetEncoding(provider.Encoding);
} catch (Exception ex) {
log.Error("Invalid encoding in config file: " + ex.Message);
enc = Encoding.GetEncoding(responseHtml.CharacterSet);
}
} else {
enc = Encoding.GetEncoding(responseHtml.CharacterSet);
}
StreamReader r = new StreamReader(responseHtml.GetResponseStream(), enc);
string source = r.ReadToEnd();
r.Close();
responseHtml.Close();
//Source cropping
source = source.Substring(Math.Max(source.IndexOf(provider.RelationsStart), 0));
source = source.Substring(0, Math.Max(source.LastIndexOf(provider.RelationsEnd), 0));
string pattern = provider.RelationsRegExp;
//for some reason, responseHtml.ResponseUri is null for some providers when running on mono
if (!Settings.Instance.isMonoCompatibilityMode) {
log.Debug("Trying to match source from " + responseHtml.ResponseUri.AbsoluteUri + " with " + pattern);
}
RegexOptions ro = RegexOptions.IgnoreCase | RegexOptions.Singleline;
if (provider.RelationsRightToLeft)
ro |= RegexOptions.RightToLeft;
MatchCollection mc = Regex.Matches(source, pattern, ro);
string CleanupRegex = provider.RelationsRemove;
for (int i = 0; i < mc.Count; i++) {
Match m = mc[i];
string result = Regex.Replace(System.Web.HttpUtility.HtmlDecode(m.Groups["Title"].Value), CleanupRegex, "");
//RELATIONS_REMOVE_KEY may be used to filter out results completely, for example if they are a html tag
if (result != "") {
//parse season and episode numbers
int s, e;
Int32.TryParse(m.Groups["Season"].Value, out s);
Int32.TryParse(m.Groups["Episode"].Value, out e);
//if we are iterating through season pages, take season from page url directly
if (url != url2) {
rc.AddEpisode(new ShowEpisode(season, e, result));
log.Debug("Found Relation: " + "S" + s.ToString() + "E" + e.ToString() + " - " + result);
} else {
rc.AddEpisode(new ShowEpisode(s, e, result));
log.Debug("Found Relation: " + "S" + s.ToString() + "E" + e.ToString() + " - " + result);
}
}
}
TvShowManager.Instance.addTvShow(rc);
// THOU SHALL NOT FORGET THE BREAK
if (!url2.Contains("%S"))
break;
season++;
}
log.Debug("" + (season - 1) + " Seasons, " + rc.CountEpisodes + " relations found");
}
//parse page(s) containing relations
/// <summary>
/// Parses page containing the relation data
/// </summary>
/// <param name="url">URL of the page to parse</param>
/// <param name="Showname">Showname</param>
public static void GetRelations(string url, string Showname, TvShowInfoProvider provider, BackgroundWorker worker, DoWorkEventArgs dwea)
{
if (provider == null)
{
log.Error("GetRelations: No relation provider found/selected");
return;
}
if (String.IsNullOrEmpty(url))
{
log.Error("GetRelations: URL is null or empty");
return;
}
if (String.IsNullOrEmpty(Showname))
{
log.Error("GetRelations: Showname is null or empty");
return;
}
log.Debug("Trying to get relations from " + url);
//if episode infos are stored on a new page for each season, this should be marked with %S in url, so we can iterate through all those pages
int season = 1;
string url2 = url;
//Create new RelationCollection
TvShow rc = new TvShow(Showname);
while (true)
{
if (url2.Contains("%S"))
{
url = url2.Replace("%S", season.ToString());
}
if (url == null || url == "")
{
return;
}
// request
url = System.Web.HttpUtility.UrlPathEncode(url);
log.Debug("Trying to get relations for season " + season + " from " + url);
HttpWebRequest requestHtml = null;
try {
requestHtml = (HttpWebRequest)(HttpWebRequest.Create(url));
} catch (Exception ex) {
log.Error(ex.Message);
requestHtml.Abort();
return;
}
requestHtml.Timeout = Helper.ReadInt(AppProperties.CONNECTION_TIMEOUT_KEY);
// get response
HttpWebResponse responseHtml = null;
try {
responseHtml = (HttpWebResponse)(requestHtml.GetResponse());
} catch (WebException ex) {
//Serienjunkies returns "(300) Mehrdeutige Umleitung" when an inexistant season is requested
if (ex.Message.Contains("(300)"))
{
break;
}
log.Error(ex.Message);
if (responseHtml != null)
{
responseHtml.Close();
}
return;
}
log.Debug("Response URL: " + responseHtml.ResponseUri.AbsoluteUri);
//if we get redirected, lets assume this page does not exist
if (responseHtml.ResponseUri.AbsoluteUri != url)
{
log.Debug("Response URL doesn't match request URL, page doesn't seem to exist");
responseHtml.Close();
requestHtml.Abort();
break;
}
// and download
//Logger.Instance.LogMessage("charset=" + responseHtml.CharacterSet, LogLevel.INFO);
Encoding enc;
if (provider.Encoding != null && provider.Encoding != "")
{
try {
enc = Encoding.GetEncoding(provider.Encoding);
} catch (Exception ex) {
log.Error("Invalid encoding in config file: " + ex.Message);
enc = Encoding.GetEncoding(responseHtml.CharacterSet);
}
}
else
{
enc = Encoding.GetEncoding(responseHtml.CharacterSet);
}
StreamReader r = new StreamReader(responseHtml.GetResponseStream(), enc);
string source = r.ReadToEnd();
r.Close();
responseHtml.Close();
//Source cropping
source = source.Substring(Math.Max(source.IndexOf(provider.RelationsStart), 0));
source = source.Substring(0, Math.Max(source.LastIndexOf(provider.RelationsEnd), 0));
string pattern = provider.RelationsRegExp;
//for some reason, responseHtml.ResponseUri is null for some providers when running on mono
if (!Settings.Instance.isMonoCompatibilityMode)
{
log.Debug("Trying to match source from " + responseHtml.ResponseUri.AbsoluteUri + " with " + pattern);
}
RegexOptions ro = RegexOptions.IgnoreCase | RegexOptions.Singleline;
if (provider.RelationsRightToLeft)
{
ro |= RegexOptions.RightToLeft;
}
MatchCollection mc = Regex.Matches(source, pattern, ro);
string CleanupRegex = provider.RelationsRemove;
for (int i = 0; i < mc.Count; i++)
{
Match m = mc[i];
string result = Regex.Replace(System.Web.HttpUtility.HtmlDecode(m.Groups["Title"].Value), CleanupRegex, "");
//RELATIONS_REMOVE_KEY may be used to filter out results completely, for example if they are a html tag
if (result != "")
{
//parse season and episode numbers
int s, e;
Int32.TryParse(m.Groups["Season"].Value, out s);
Int32.TryParse(m.Groups["Episode"].Value, out e);
//if we are iterating through season pages, take season from page url directly
if (url != url2)
{
rc.AddEpisode(new ShowEpisode(season, e, result));
log.Debug("Found Relation: " + "S" + s.ToString() + "E" + e.ToString() + " - " + result);
}
else
{
rc.AddEpisode(new ShowEpisode(s, e, result));
log.Debug("Found Relation: " + "S" + s.ToString() + "E" + e.ToString() + " - " + result);
}
}
}
TvShowManager.Instance.addTvShow(rc);
// THOU SHALL NOT FORGET THE BREAK
if (!url2.Contains("%S"))
{
break;
}
season++;
}
log.Debug("" + (season - 1) + " Seasons, " + rc.CountEpisodes + " relations found");
}
