public LinkCrawler(IEnumerable<IOutput> outputs, IValidUrlParser validUrlParser, ISettings settings)
{
BaseUrl = settings.BaseUrl;
Outputs = outputs;
ValidUrlParser = validUrlParser;
CheckImages = settings.CheckImages;
VisitedUrlList = new List<string>();
RestRequest = new RestRequest(Method.GET).SetHeader("Accept", "*/*");
OnlyReportBrokenLinksToOutput = settings.OnlyReportBrokenLinksToOutput;
_settings = settings;
}
public LinkCrawler(ISlackClient slackClient, IValidUrlParser validUrlParser, ISettings settings)
{
SlackClient = slackClient;
BaseUrl = settings.BaseUrl;
RestClient = new RestClient();
ValidUrlParser = validUrlParser;
CheckImages = settings.CheckImages;
VisitedUrlList = new List <string>();
RestRequest = new RestRequest(Method.GET).SetHeader("Accept", "*/*");
OnlyReportBrokenLinksToOutput = settings.OnlyReportBrokenLinksToOutput;
}
public LinkCrawler(IEnumerable <IOutput> outputs, IValidUrlParser validUrlParser, ISettings settings)
{
BaseUrl = settings.BaseUrl;
Outputs = outputs;
ValidUrlParser = validUrlParser;
CheckImages = settings.CheckImages;
VisitedUrlList = new List <string>();
RestRequest = new RestRequest(Method.GET).SetHeader("Accept", "*/*");
OnlyReportBrokenLinksToOutput = settings.OnlyReportBrokenLinksToOutput;
_settings = settings;
}
/// <summary>
/// Get's a list of all urls in markup and tires to fix the urls that Restsharp will have a problem with
/// (i.e relative urls, urls with no sceme, mailto links..etc)
/// </summary>
/// <returns>List of urls that will work with restsharp for sending http get</returns>
public static List<string> GetValidUrlListFromMarkup(string markup, IValidUrlParser parser, bool checkImages)
{
var urlList = GetAllUrlsFromMarkup(markup, checkImages);
var validUrlList = new List<string>();
foreach (var url in urlList)
{
string validUrl;
if (parser.Parse(url, out validUrl))
{
validUrlList.Add(validUrl);
}
}
return validUrlList;
}
/// <summary>
/// Get's a list of all urls in markup and tires to fix the urls that Restsharp will have a problem with
/// (i.e relative urls, urls with no sceme, mailto links..etc)
/// </summary>
/// <returns>List of urls that will work with restsharp for sending http get</returns>
public static List <string> GetValidUrlListFromMarkup(string markup, IValidUrlParser parser, bool checkImages)
{
var urlList = GetAllUrlsFromMarkup(markup, checkImages);
var validUrlList = new List <string>();
foreach (var url in urlList)
{
string validUrl;
if (parser.Parse(url, out validUrl))
{
validUrlList.Add(validUrl);
}
}
return(validUrlList);
}
public LinkCrawler(IEnumerable <IOutput> outputs, IValidUrlParser validUrlParser, ISettings settings)
{
_httpClient = new HttpClient();
_httpClient.DefaultRequestHeaders.Add("User-Agent", "curl/7.54.0");
_httpClient.DefaultRequestHeaders.Add("Accept", "*/*");
msSleepBetweenRequests = settings.TimeMsBetweenRequests;
Outputs = outputs;
ValidUrlParser = validUrlParser;
UrlList = new HashSet <LinkModel>(new LinkListComparer()); //to have only unique addresses in the list
OnlyReportBrokenLinksToOutput = settings.OnlyReportBrokenLinksToOutput;
_settings = settings;
this.timer = new Stopwatch();
stopWordsList = ReadStopWords();
}
