public async Task<WebClientStringResult> GetString(WebRequest request)
{
logger.Debug(string.Format("UnixLibCurlWebClient:GetString(Url:{0})", request.Url));
var result = await GetBytes(request);
var sresult = new WebClientStringResult()
{
Content = Encoding.UTF8.GetString(result.Content),
Cookies = result.Cookies,
Status = result.Status
};
return sresult;
}
protected async Task FollowIfRedirect(WebClientStringResult response, string referrer = null, string overrideRedirectUrl = null, string overrideCookies = null)
{
var byteResult = new WebClientByteResult();
// Map to byte
Mapper.Map(response, byteResult);
await FollowIfRedirect(byteResult, referrer, overrideRedirectUrl, overrideCookies);
// Map to string
Mapper.Map(byteResult, response);
}
public override async Task<ConfigurationData> GetConfigurationForSetup()
{
var Login = Definition.Login;
if (Login == null || Login.Method != "form")
return configData;
var LoginUrl = SiteLink + Login.Path;
configData.CookieHeader.Value = null;
landingResult = await RequestStringWithCookies(LoginUrl, null, SiteLink);
var htmlParser = new HtmlParser();
landingResultDocument = htmlParser.Parse(landingResult.Content);
var grecaptcha = landingResultDocument.QuerySelector(".g-recaptcha");
if (grecaptcha != null)
{
var CaptchaItem = new RecaptchaItem();
CaptchaItem.Name = "Captcha";
CaptchaItem.Version = "2";
CaptchaItem.SiteKey = grecaptcha.GetAttribute("data-sitekey");
if (CaptchaItem.SiteKey == null) // some sites don't store the sitekey in the .g-recaptcha div (e.g. cloudflare captcha challenge page)
CaptchaItem.SiteKey = landingResultDocument.QuerySelector("[data-sitekey]").GetAttribute("data-sitekey");
configData.AddDynamic("Captcha", CaptchaItem);
}
return configData;
}
protected bool CheckIfLoginIsNeeded(WebClientStringResult Result, IHtmlDocument document)
{
if (Result.IsRedirect)
{
return true;
}
if (Definition.Login == null || Definition.Login.Test == null)
return false;
if (Definition.Login.Test.Selector != null)
{
var selection = document.QuerySelectorAll(Definition.Login.Test.Selector);
if (selection.Length == 0)
{
return true;
}
}
return false;
}
protected async Task<bool> DoLogin()
{
var Login = Definition.Login;
if (Login == null)
return false;
if (Login.Method == "post")
{
var pairs = new Dictionary<string, string>();
foreach (var Input in Definition.Login.Inputs)
{
var value = applyGoTemplateText(Input.Value);
pairs.Add(Input.Key, value);
}
var LoginUrl = SiteLink + Login.Path;
configData.CookieHeader.Value = null;
var loginResult = await RequestLoginAndFollowRedirect(LoginUrl, pairs, null, true, null, SiteLink, true);
configData.CookieHeader.Value = loginResult.Cookies;
checkForLoginError(loginResult);
}
else if (Login.Method == "form")
{
var LoginUrl = SiteLink + Login.Path;
var pairs = new Dictionary<string, string>();
var CaptchaConfigItem = (RecaptchaItem)configData.GetDynamic("Captcha");
if (CaptchaConfigItem != null)
{
if (!string.IsNullOrWhiteSpace(CaptchaConfigItem.Cookie))
{
// for remote users just set the cookie and return
CookieHeader = CaptchaConfigItem.Cookie;
return true;
}
var CloudFlareCaptchaChallenge = landingResultDocument.QuerySelector("script[src=\"/cdn-cgi/scripts/cf.challenge.js\"]");
if (CloudFlareCaptchaChallenge != null)
{
var CloudFlareQueryCollection = new NameValueCollection();
CloudFlareQueryCollection["id"] = CloudFlareCaptchaChallenge.GetAttribute("data-ray");
CloudFlareQueryCollection["g-recaptcha-response"] = CaptchaConfigItem.Value;
var ClearanceUrl = resolvePath("/cdn-cgi/l/chk_captcha?" + CloudFlareQueryCollection.GetQueryString());
var ClearanceResult = await RequestStringWithCookies(ClearanceUrl.ToString(), null, SiteLink);
if (ClearanceResult.IsRedirect) // clearance successfull
{
// request real login page again
landingResult = await RequestStringWithCookies(LoginUrl, null, SiteLink);
var htmlParser = new HtmlParser();
landingResultDocument = htmlParser.Parse(landingResult.Content);
}
else
{
throw new ExceptionWithConfigData(string.Format("Login failed: Cloudflare clearance failed using cookies {0}: {1}", CookieHeader, ClearanceResult.Content), configData);
}
}
else
{
pairs.Add("g-recaptcha-response", CaptchaConfigItem.Value);
}
}
var FormSelector = Login.Form;
if (FormSelector == null)
FormSelector = "form";
// landingResultDocument might not be initiated if the login is caused by a relogin during a query
if (landingResultDocument == null)
{
await GetConfigurationForSetup();
}
var form = landingResultDocument.QuerySelector(FormSelector);
if (form == null)
{
throw new ExceptionWithConfigData(string.Format("Login failed: No form found on {0} using form selector {1}", LoginUrl, FormSelector), configData);
}
var inputs = form.QuerySelectorAll("input");
if (inputs == null)
{
throw new ExceptionWithConfigData(string.Format("Login failed: No inputs found on {0} using form selector {1}", LoginUrl, FormSelector), configData);
}
var submitUrl = resolvePath(form.GetAttribute("action"));
foreach (var input in inputs)
{
var name = input.GetAttribute("name");
if (name == null)
continue;
var value = input.GetAttribute("value");
if (value == null)
value = "";
pairs[name] = value;
}
foreach (var Input in Definition.Login.Inputs)
{
var value = applyGoTemplateText(Input.Value);
pairs[Input.Key] = value;
}
// automatically solve simpleCaptchas, if used
var simpleCaptchaPresent = landingResultDocument.QuerySelector("script[src*=\"simpleCaptcha\"]");
if(simpleCaptchaPresent != null)
{
var captchaUrl = resolvePath("simpleCaptcha.php?numImages=1");
var simpleCaptchaResult = await RequestStringWithCookies(captchaUrl.ToString(), null, LoginUrl);
var simpleCaptchaJSON = JObject.Parse(simpleCaptchaResult.Content);
var captchaSelection = simpleCaptchaJSON["images"][0]["hash"].ToString();
pairs["captchaSelection"] = captchaSelection;
pairs["submitme"] = "X";
}
// clear landingResults/Document, otherwise we might use an old version for a new relogin (if GetConfigurationForSetup() wasn't called before)
landingResult = null;
landingResultDocument = null;
WebClientStringResult loginResult = null;
var enctype = form.GetAttribute("enctype");
if (enctype == "multipart/form-data")
{
var headers = new Dictionary<string, string>();
var boundary = "---------------------------" + (DateTime.UtcNow.Subtract(new DateTime(1970, 1, 1))).TotalSeconds.ToString().Replace(".", "");
var bodyParts = new List<string>();
foreach (var pair in pairs)
{
var part = "--" + boundary + "\r\n" +
"Content-Disposition: form-data; name=\"" + pair.Key + "\"\r\n" +
"\r\n" +
pair.Value;
bodyParts.Add(part);
}
bodyParts.Add("--" + boundary + "--");
headers.Add("Content-Type", "multipart/form-data; boundary=" + boundary);
var body = string.Join("\r\n", bodyParts);
loginResult = await PostDataWithCookies(submitUrl.ToString(), pairs, configData.CookieHeader.Value, SiteLink, headers, body);
} else {
loginResult = await RequestLoginAndFollowRedirect(submitUrl.ToString(), pairs, configData.CookieHeader.Value, true, null, SiteLink, true);
}
configData.CookieHeader.Value = loginResult.Cookies;
checkForLoginError(loginResult);
}
else if (Login.Method == "cookie")
{
configData.CookieHeader.Value = ((StringItem)configData.GetDynamic("cookie")).Value;
}
else
{
throw new NotImplementedException("Login method " + Definition.Login.Method + " not implemented");
}
return true;
}
protected bool checkForLoginError(WebClientStringResult loginResult)
{
var ErrorBlocks = Definition.Login.Error;
if (ErrorBlocks == null)
return true; // no error
var loginResultParser = new HtmlParser();
var loginResultDocument = loginResultParser.Parse(loginResult.Content);
foreach (errorBlock error in ErrorBlocks)
{
var selection = loginResultDocument.QuerySelector(error.Selector);
if (selection != null)
{
string errorMessage = selection.TextContent;
if (error.Message != null)
{
var errorSubMessage = loginResultDocument.QuerySelector(error.Message.Selector);
errorMessage = errorSubMessage.TextContent;
}
throw new ExceptionWithConfigData(string.Format("Login failed: {0}", errorMessage.Trim()), configData);
}
}
return true; // no error
}
virtual public async Task <WebClientStringResult> GetString(WebRequest request)
{
logger.Debug(string.Format("IWebClient.GetString(Url:{0})", request.Url));
PrepareRequest(request);
DelayRequest(request);
var result = await Run(request);
result.Request = request;
WebClientStringResult stringResult = Mapper.Map <WebClientStringResult>(result);
Encoding encoding = null;
if (request.Encoding != null)
{
encoding = request.Encoding;
}
else if (result.Headers.ContainsKey("content-type"))
{
Regex CharsetRegex = new Regex(@"charset=([\w-]+)", RegexOptions.Compiled);
var CharsetRegexMatch = CharsetRegex.Match(result.Headers["content-type"][0]);
if (CharsetRegexMatch.Success)
{
var charset = CharsetRegexMatch.Groups[1].Value;
try
{
encoding = Encoding.GetEncoding(charset);
}
catch (Exception ex)
{
logger.Error(string.Format("IWebClient.GetString(Url:{0}): Error loading encoding {0} based on header {1}: {2}", request.Url, charset, result.Headers["content-type"][0], ex));
}
}
else
{
logger.Error(string.Format("IWebClient.GetString(Url:{0}): Got header without charset: {0}", request.Url, result.Headers["content-type"][0]));
}
}
if (encoding == null)
{
logger.Error(string.Format("IWebClient.GetString(Url:{0}): No encoding detected, defaulting to UTF-8", request.Url));
encoding = Encoding.UTF8;
}
string decodedContent = null;
if (result.Content != null)
{
decodedContent = encoding.GetString(result.Content);
}
stringResult.Content = decodedContent;
logger.Debug(string.Format("IWebClient: Returning {0} => {1}", result.Status, (result.IsRedirect ? result.RedirectingTo + " " : "") + (decodedContent == null ? "<NULL>" : decodedContent)));
string[] server;
if (stringResult.Headers.TryGetValue("server", out server))
{
if (server[0] == "cloudflare-nginx")
{
stringResult.Content = BrowserUtil.DecodeCloudFlareProtectedEmailFromHTML(stringResult.Content);
}
}
return(stringResult);
}
public async Task<WebClientStringResult> GetString(WebRequest request)
{
logger.Debug(string.Format("WindowsWebClient:GetString(Url:{0})", request.Url));
var cookies = new CookieContainer();
if (!string.IsNullOrEmpty(request.Cookies))
{
var uri = new Uri(request.Url);
foreach (var c in request.Cookies.Split(';'))
{
try
{
cookies.SetCookies(uri, c);
}
catch (CookieException ex)
{
logger.Info("(Non-critical) Problem loading cookie {0}, {1}, {2}", uri, c, ex.Message);
}
}
}
var client = new HttpClient(new HttpClientHandler
{
CookieContainer = cookies,
AllowAutoRedirect = false, // Do not use this - Bugs ahoy! Lost cookies and more.
UseCookies = true,
});
client.DefaultRequestHeaders.Add("User-Agent", BrowserUtil.ChromeUserAgent);
HttpResponseMessage response = null;
if (request.Type == RequestType.POST)
{
var content = new FormUrlEncodedContent(request.PostData);
response = await client.PostAsync(request.Url, content);
} else
{
response = await client.GetAsync(request.Url);
}
var result = new WebClientStringResult();
result.Content = await response.Content.ReadAsStringAsync();
// Compatiblity issue between the cookie format and httpclient
// Pull it out manually ignoring the expiry date then set it manually
// http://stackoverflow.com/questions/14681144/httpclient-not-storing-cookies-in-cookiecontainer
IEnumerable<string> cookieHeaders;
if (response.Headers.TryGetValues("set-cookie", out cookieHeaders))
{
var cookieBuilder = new StringBuilder();
foreach (var c in cookieHeaders)
{
if (cookieBuilder.Length > 0)
{
cookieBuilder.Append("; ");
}
cookieBuilder.Append( c.Substring(0, c.IndexOf(';')));
}
result.Cookies = cookieBuilder.ToString();
}
result.Status = response.StatusCode;
if (null != response.Headers.Location)
{
result.RedirectingTo = response.Headers.Location.ToString();
}
return result;
}
