public void GetPageCrawlerDetailTest()
{
string page = string.Empty;
bo.Web.WebPageContent content = new bo.Web.WebPageContent();
string fileName = "tyre-shopper.co.uk.home.html";
string path = GetUnitTestDataFilePath(fileName);
page = GetFileContents(fileName);
Uri uri = new Uri(path);
bo.Web.PageCrawlDetail pageCrawlDetail = new bo.Web.PageCrawlDetail(uri);
pageCrawlDetail.PageUri = new Uri("http://www.tyre-shopper.co.uk");
pageCrawlDetail.LoadContent(page, HttpStatusCode.OK);
pageCrawlDetail.LoadUris();
Assert.IsTrue(pageCrawlDetail.AllLinks.Count == 37);
content = new bo.Web.WebPageContent();
fileName = "tyre-shopper.co.uk.aboutUs.html";
path = GetUnitTestDataFilePath(fileName);
page = GetFileContents(fileName);
uri = new Uri(path);
pageCrawlDetail = new bo.Web.PageCrawlDetail(uri);
pageCrawlDetail.PageUri = new Uri("http://www.tyre-shopper.co.uk/about-us/");
pageCrawlDetail.LoadContent(page, HttpStatusCode.OK);
pageCrawlDetail.LoadUris();
Assert.IsTrue(pageCrawlDetail.AllLinks.Count == 22);
}
public void InvalidHttpResponse()
{
Moq.Mock <HttpWebResponse> fakeResponse = new Moq.Mock <HttpWebResponse>();
fakeResponse.Setup(response => response.StatusCode).Returns(HttpStatusCode.NotFound);
Moq.Mock <HttpWebRequest> fakeRequest = new Moq.Mock <HttpWebRequest>();
fakeRequest.Setup(request => request.GetResponse()).Returns(fakeResponse.Object);
bo.Web.WebPageContent content = new bo.Web.WebPageContent();
content.request = fakeRequest.Object;
Assert.IsFalse(content.TryGetResponse());
}
private static void ProcessUrl(string url)
{
content = new bo.Web.WebPageContent();
content.SetRequestFromURL(url);
if (bo.Helpers.WebHelper.IsUrlValid(url) && content.TryGetResponse())
{
// Now we can get the robots file
Uri baseUri = new Uri(url);
Uri robotsUri = new Uri(bo.Helpers.WebHelper.CombineUrl(baseUri, "/robots.txt"));
content = new bo.Web.WebPageContent();
content.SetRequestFromURL(robotsUri.AbsoluteUri);
string responseString = string.Empty;
if (content.TryGetResponse())
{
responseString = bo.Helpers.WebHelper.ResponseToString(content.response);
}
bo.Web.RobotsTxt robots = bo.Helpers.RobotsHelper.ParseRobotsTxt(responseString);
bo.Web.PageCrawlDetail pageDetails = new bo.Web.PageCrawlDetail(baseUri);
try
{
//Start the Crawling
bo.Crawler.CrawlerProcessing crw = new bo.Crawler.CrawlerProcessing(NoOfThreads, SleepTime, baseUri, robots);
crw.Crawl(pageDetails).Wait();
bo.Helpers.FileHelper.PageDetailToCSV(crw.CrawlList.UrlsCompleted, outputFilePath);
}
catch (AggregateException e)
{
foreach (var ex in e.InnerExceptions)
{
Console.WriteLine(ex.InnerException);
}
Console.ReadLine();
}
}
else
{
string info = "Invalid URL: " + url;
Console.WriteLine(url);
log.Info(url);
}
}
public void ValidHttpResponse()
{
Moq.Mock <HttpWebResponse> fakeResponse = new Moq.Mock <HttpWebResponse>();
fakeResponse.Setup(response => response.StatusCode).Returns(HttpStatusCode.OK);
Moq.Mock <HttpWebRequest> fakeRequest = new Moq.Mock <HttpWebRequest>();
fakeRequest.Setup(request => request.GetResponse()).Returns(fakeResponse.Object);
bo.Web.WebPageContent content = new bo.Web.WebPageContent();
content.request = fakeRequest.Object;
Assert.IsTrue(content.TryGetResponse());
content = new bo.Web.WebPageContent();
//content.SetRequestFromURL("http://www.tyre-shopper.co.uk/robots.txt
content.SetRequestFromURL("http://localhost:4174/");
Assert.IsTrue(content.TryGetResponse());
}
