public void ErrorOccurred_ErrorExceptionIsNull_ReturnsFalse()
{
CrawlResult unitUnderTest = new CrawlResult();
unitUnderTest.ErrorException = null;
Assert.IsFalse(unitUnderTest.ErrorOccurred);
Assert.IsNull(unitUnderTest.ErrorException);
}
public void Constructor_ValidUri_CreatesInstance()
{
CrawlResult unitUnderTest = new CrawlResult();
Assert.AreEqual(default(TimeSpan), unitUnderTest.Elapsed);
Assert.AreEqual(null, unitUnderTest.ErrorException);
Assert.AreEqual(false, unitUnderTest.ErrorOccurred);
Assert.AreEqual(null, unitUnderTest.RootUri);
Assert.AreEqual(null, unitUnderTest.CrawlContext);
}
public void ErrorOccurred_ErrorExceptionNotNull_ReturnsTrue()
{
CrawlResult unitUnderTest = new CrawlResult();
Exception ex = new Exception("oh no");
unitUnderTest.ErrorException = ex;
Assert.IsTrue(unitUnderTest.ErrorOccurred);
Assert.AreSame(ex, unitUnderTest.ErrorException);
}
/// <summary>
/// Begins a synchronous crawl using the uri param, subscribe to events to process data as it becomes available
/// </summary>
public virtual CrawlResult Crawl(Uri uri)
{
if (uri == null)
throw new ArgumentNullException("uri");
_crawlContext.RootUri = uri;
_crawlResult = new CrawlResult();
_crawlResult.RootUri = _crawlContext.RootUri;
_crawlResult.CrawlContext = _crawlContext;
_crawlComplete = false;
_logger.InfoFormat("About to crawl site [{0}]", uri.AbsoluteUri);
if (_memoryManager != null)
{
_crawlContext.MemoryUsageBeforeCrawlInMb = _memoryManager.GetCurrentUsageInMb();
_logger.InfoFormat("Starting memory usage for site [{0}] is [{1}mb]", uri.AbsoluteUri, _crawlContext.MemoryUsageBeforeCrawlInMb);
}
PrintConfigValues(_crawlContext.CrawlConfiguration);
_scheduler.Add(new PageToCrawl(uri) { ParentUri = uri, IsInternal = true, IsRoot = true });
_crawlContext.CrawlStartDate = DateTime.Now;
Stopwatch timer = Stopwatch.StartNew();
if (_crawlContext.CrawlConfiguration.CrawlTimeoutSeconds > 0)
{
_timeoutTimer = new Timer(_crawlContext.CrawlConfiguration.CrawlTimeoutSeconds * 1000);
_timeoutTimer.Elapsed += HandleCrawlTimeout;
_timeoutTimer.Start();
}
try
{
VerifyRequiredAvailableMemory();
CrawlSite();
}
catch (Exception e)
{
_crawlResult.ErrorException = e;
_logger.FatalFormat("An error occurred while crawling site [{0}]", uri);
_logger.Fatal(e);
}
finally
{
if(_threadManager != null)
_threadManager.Dispose();
}
if(_timeoutTimer != null)
_timeoutTimer.Stop();
timer.Stop();
if (_memoryManager != null)
{
_crawlContext.MemoryUsageAfterCrawlInMb = _memoryManager.GetCurrentUsageInMb();
_logger.InfoFormat("Ending memory usage for site [{0}] is [{1}mb]", uri.AbsoluteUri, _crawlContext.MemoryUsageAfterCrawlInMb);
}
_crawlResult.Elapsed = timer.Elapsed;
_logger.InfoFormat("Crawl complete for site [{0}]: [{1}]", _crawlResult.RootUri.AbsoluteUri, _crawlResult.Elapsed);
return _crawlResult;
}
static void LogCrawledCompletedStatistics(CrawlResult result)
{
_log.Info("Pages downloaded: " + pagesDownloaded);
_log.Info("Pages downloaded per second: " + pagesDownloaded / result.Elapsed.TotalSeconds);
_log.Info("GB downloaded: " + bytesDownloaded / 1024 / 1024);
_log.Info("GB downloaded per second: " + (float)bytesDownloaded / 1024f / 1024f / (float)result.Elapsed.TotalSeconds);
_log.Info("Time: " + result.Elapsed);
}