private async Task <string> GetGuestToken()
{
if (guestToken == null)
{
var requestRegistration = new CancellationTokenRegistration();
try
{
string url = await GetApiUrl(Blog.Url, 0, "", 0);
if (ShellService.Settings.LimitConnectionsApi)
{
CrawlerService.TimeconstraintApi.Acquire();
}
var headers = new Dictionary <string, string>();
headers.Add("Origin", "https://twitter.com");
headers.Add("Authorization", "Bearer " + BearerToken);
HttpWebRequest request = WebRequestFactory.CreatePostRequest(url, "https://twitter.com", headers);
CookieService.GetUriCookie(request.CookieContainer, new Uri("https://twitter.com"));
requestRegistration = Ct.Register(() => request.Abort());
var content = await WebRequestFactory.ReadRequestToEndAsync(request);
guestToken = ((JValue)((JObject)JsonConvert.DeserializeObject(content))["guest_token"]).Value <string>();
}
catch (Exception e)
{
Logger.Error("GetGuestToken: {0}", e);
}
finally
{
requestRegistration.Dispose();
}
}
return(guestToken);
}
protected virtual async Task <string> RequestPostAsync(int pageNumber)
{
var requestRegistration = new CancellationTokenRegistration();
try
{
string url = "https://www.tumblr.com/search/" + Blog.Name + "/post_page/" + pageNumber;
string referer = @"https://www.tumblr.com/search/" + Blog.Name;
var headers = new Dictionary <string, string> {
{ "X-tumblr-form-key", tumblrKey }, { "DNT", "1" }
};
HttpWebRequest request = WebRequestFactory.CreatePostXhrReqeust(url, referer, headers);
CookieService.GetUriCookie(request.CookieContainer, new Uri("https://www.tumblr.com/"));
//Example request body, searching for cars:
//q=cars&sort=top&post_view=masonry&blogs_before=8&num_blogs_shown=8&num_posts_shown=20&before=24&blog_page=2&safe_mode=true&post_page=2&filter_nsfw=true&filter_post_type=&next_ad_offset=0&ad_placement_id=0&more_posts=true
string requestBody = "q=" + Blog.Name + "&sort=top&post_view=masonry&num_posts_shown=" +
((pageNumber - 1) * Blog.PageSize) + "&before=" + ((pageNumber - 1) * Blog.PageSize) +
"&safe_mode=false&post_page=" + pageNumber +
"&filter_nsfw=false&filter_post_type=&next_ad_offset=0&ad_placement_id=0&more_posts=true";
await WebRequestFactory.PerformPostXHRReqeustAsync(request, requestBody);
requestRegistration = Ct.Register(() => request.Abort());
return(await WebRequestFactory.ReadReqestToEndAsync(request));
}
finally
{
requestRegistration.Dispose();
}
}
protected async Task <string> RequestApiDataAsync(string url, string bearerToken, Dictionary <string, string> headers = null,
IEnumerable <string> cookieHosts = null)
{
var requestRegistration = new CancellationTokenRegistration();
try
{
HttpWebRequest request = WebRequestFactory.CreateGetRequest(url, string.Empty, headers);
cookieHosts = cookieHosts ?? new List <string>();
foreach (string cookieHost in cookieHosts)
{
CookieService.GetUriCookie(request.CookieContainer, new Uri(cookieHost));
}
request.PreAuthenticate = true;
request.Headers.Add("Authorization", "Bearer " + bearerToken);
request.Accept = "application/json";
requestRegistration = Ct.Register(() => request.Abort());
return(await WebRequestFactory.ReadRequestToEndAsync(request));
}
finally
{
requestRegistration.Dispose();
}
}
protected async Task <string> RequestDataAsync(string url, Dictionary <string, string> headers = null,
IEnumerable <string> cookieHosts = null)
{
var requestRegistration = new CancellationTokenRegistration();
try
{
int redirects = 0;
ResponseDetails responseDetails;
do
{
HttpWebRequest request = WebRequestFactory.CreateGetRequest(url, string.Empty, headers, false);
cookieHosts = cookieHosts ?? new List <string>();
foreach (string cookieHost in cookieHosts)
{
CookieService.GetUriCookie(request.CookieContainer, new Uri(cookieHost));
}
requestRegistration = Ct.Register(() => request.Abort());
responseDetails = await WebRequestFactory.ReadRequestToEnd2Async(request);
if (responseDetails.HttpStatusCode == HttpStatusCode.Found)
{
url = request.RequestUri.GetLeftPart(UriPartial.Authority) + responseDetails.RedirectUrl;
}
} while (responseDetails.HttpStatusCode == HttpStatusCode.Found && redirects++ < 5);
if (responseDetails.HttpStatusCode == HttpStatusCode.Found)
{
throw new WebException("Too many automatic redirections were attempted.", WebExceptionStatus.ProtocolError);
}
return(responseDetails.Response);
}
finally
{
requestRegistration.Dispose();
}
}
protected async Task <string> RequestDataAsync(string url, Dictionary <string, string> headers = null,
IEnumerable <string> cookieHosts = null)
{
var requestRegistration = new CancellationTokenRegistration();
try
{
HttpWebRequest request = WebRequestFactory.CreateGetReqeust(url, string.Empty, headers);
cookieHosts = cookieHosts ?? new List <string>();
foreach (string cookieHost in cookieHosts)
{
CookieService.GetUriCookie(request.CookieContainer, new Uri(cookieHost));
}
requestRegistration = Ct.Register(() => request.Abort());
return(await WebRequestFactory.ReadReqestToEndAsync(request));
}
finally
{
requestRegistration.Dispose();
}
}
protected string RetrieveOriginalImageUrl(string url, int width, int height)
{
if (width > height)
{
(width, height) = (height, width);
}
if (ShellService.Settings.ImageSize != "best" ||
!url.Contains("/s1280x1920/") ||
(width <= 1280 && height <= 1920))
{
return(url);
}
url = url.Replace("/s1280x1920/", (width <= 2048 && height <= 3072) ? "/s2048x3072/" : "/s99999x99999/");
string pageContent = "";
try
{
HttpWebRequest request = WebRequestFactory.CreateGetRequest(url, "",
new Dictionary <string, string>()
{
{ "Accept-Language", "en-US" }, { "Accept-Encoding", "gzip, deflate" }
}, false);
request.Accept = "text/html, application/xhtml+xml, */*";
request.UserAgent = ShellService.Settings.UserAgent;
request.AutomaticDecompression = DecompressionMethods.Deflate | DecompressionMethods.GZip;
pageContent = WebRequestFactory.ReadRequestToEndAsync(request).GetAwaiter().GetResult();
}
catch (Exception e)
{
Logger.Error("TumblrBlogCrawler:RetrieveRawImageUrl:Exception {0}", e);
return(url);
}
pageContent = extractJsonFromPage.Match(pageContent).Groups[1].Value;
ImageResponse imgRsp = ConvertJsonToClass <ImageResponse>(pageContent);
Image img = imgRsp.Images.FirstOrDefault(x => x.HasOriginalDimensions = true);
return(string.IsNullOrEmpty(img.MediaKey) ? url : img.Url);
}
protected virtual async Task <string> RequestDataAsync(string limit, string offset)
{
var requestRegistration = new CancellationTokenRegistration();
try
{
string url = @"https://www.tumblr.com/svc/indash_blog?tumblelog_name_or_id=" + Blog.Name +
@"&post_id=&limit=" + limit + "&offset=" + offset + "&should_bypass_safemode=true";
string referer = @"https://www.tumblr.com/dashboard/blog/" + Blog.Name;
var headers = new Dictionary <string, string> {
{ "X-tumblr-form-key", tumblrKey }
};
HttpWebRequest request = WebRequestFactory.CreateGetXhrRequest(url, referer, headers);
CookieService.GetUriCookie(request.CookieContainer, new Uri("https://www.tumblr.com/"));
CookieService.GetUriCookie(request.CookieContainer, new Uri("https://" + Blog.Name.Replace("+", "-") + ".tumblr.com"));
requestRegistration = Ct.Register(() => request.Abort());
return(await WebRequestFactory.ReadRequestToEndAsync(request));
}
finally
{
requestRegistration.Dispose();
}
}
protected async Task <string> RequestDataAsync(string url, Dictionary <string, string> headers = null,
IEnumerable <string> cookieHosts = null)
{
var requestRegistration = new CancellationTokenRegistration();
try
{
int redirects = 0;
ResponseDetails responseDetails;
do
{
HttpWebRequest request = WebRequestFactory.CreateGetRequest(url, string.Empty, headers, false);
cookieHosts = cookieHosts ?? new List <string>();
foreach (string cookieHost in cookieHosts)
{
CookieService.GetUriCookie(request.CookieContainer, new Uri(cookieHost));
}
requestRegistration = Ct.Register(() => request.Abort());
responseDetails = await WebRequestFactory.ReadRequestToEnd2Async(request);
url = responseDetails.RedirectUrl ?? url;
if (responseDetails.HttpStatusCode == HttpStatusCode.Found)
{
if (url.Contains("privacy/consent"))
{
var ex = new Exception("Acceptance of privacy consent needed!");
ShellService.ShowError(new TumblrPrivacyConsentException(ex), Resources.ConfirmationTumblrPrivacyConsentNeeded);
throw ex;
}
if (!url.StartsWith("http", StringComparison.InvariantCultureIgnoreCase))
{
url = request.RequestUri.GetLeftPart(UriPartial.Authority) + url;
}
}
if (responseDetails.HttpStatusCode == HttpStatusCode.Moved)
{
Uri uri = new Uri(url);
if (!uri.Authority.Contains(".tumblr."))
{
Blog.Url = uri.GetLeftPart(UriPartial.Authority);
}
}
} while ((responseDetails.HttpStatusCode == HttpStatusCode.Found || responseDetails.HttpStatusCode == HttpStatusCode.Moved) && redirects++ < 5);
if (responseDetails.HttpStatusCode == HttpStatusCode.Found)
{
throw new WebException("Too many automatic redirections were attempted.", WebExceptionStatus.ProtocolError);
}
return(responseDetails.Response);
}
catch (Exception e)
{
Logger.Error("AbstractCrawler.RequestDataAsync: {0}", e);
throw;
}
finally
{
requestRegistration.Dispose();
}
}
protected string RetrieveOriginalImageUrl(string url, int width, int height)
{
if (width > height)
{
(width, height) = (height, width);
}
if (ShellService.Settings.ImageSize != "best" ||
!url.Contains("/s1280x1920/") ||
(width <= 1280 && height <= 1920))
{
return(url);
}
url = url.Replace("/s1280x1920/", (width <= 2048 && height <= 3072) ? "/s2048x3072/" : "/s99999x99999/");
string pageContent = "";
int errCnt = 0;
Exception lastError = null;
do
{
try
{
HttpWebRequest request = WebRequestFactory.CreateGetRequest(url, "",
new Dictionary <string, string>()
{
{ "Accept-Language", "en-US" }, { "Accept-Encoding", "gzip, deflate" }
}, false);
request.Accept = "text/html, application/xhtml+xml, */*";
request.UserAgent = ShellService.Settings.UserAgent;
request.AutomaticDecompression = DecompressionMethods.Deflate | DecompressionMethods.GZip;
pageContent = WebRequestFactory.ReadRequestToEndAsync(request).GetAwaiter().GetResult();
errCnt = 9;
}
catch (WebException we)
{
if (we.Response != null && ((HttpWebResponse)we.Response).StatusCode == HttpStatusCode.NotFound)
{
return(url);
}
}
catch (Exception e)
{
errCnt++;
Logger.Error("AbstractTumblrCrawler:RetrieveOriginalImageUrl: {0}", e);
lastError = e;
if (errCnt < 3)
{
Thread.Sleep(errCnt * 10000);
}
}
} while (errCnt < 3);
if (errCnt == 3)
{
ShellService.ShowError(lastError, Resources.PostNotParsable, Blog.Name);
throw new NullReferenceException("RetrieveOriginalImageUrl download", lastError);
}
try
{
var extracted = extractJsonFromPage.Match(pageContent).Groups[1].Value;
extracted = new Regex("/.*/").Replace(extracted, "\"\"");
ImageResponse imgRsp = DeserializeImageResponse(extracted);
int maxWidth = imgRsp.Images.Max(x => x.Width);
Image img = imgRsp.Images.FirstOrDefault(x => x.Width == maxWidth);
return(string.IsNullOrEmpty(img?.MediaKey) ? url : img.Url);
}
catch (Exception ex)
{
Logger.Error("AbstractTumblrCrawler:RetrieveOriginalImageUrl: {0}", ex);
ShellService.ShowError(ex, Resources.PostNotParsable, Blog.Name);
throw new NullReferenceException("RetrieveOriginalImageUrl parsing", ex);
}
}