/**************************************************************************/
public void SetPrefsFormControlFields()
{
{ // Configure Display Options
this.checkBoxPauseDisplayDuringScan.Checked = MacroscopePreferencesManager.GetPauseDisplayDuringScan();
this.checkBoxShowProgressDialogues.Checked = MacroscopePreferencesManager.GetShowProgressDialogues();
}
{ //Configure Form Fields
/** Spidering Control ---------------------------------------------- **/
this.numericUpDownDepth.Minimum = -1;
this.numericUpDownDepth.Maximum = 10000;
this.numericUpDownPageLimit.Minimum = -1;
this.numericUpDownPageLimit.Maximum = 10000;
this.numericUpDownCrawlDelay.Minimum = 0;
this.numericUpDownCrawlDelay.Maximum = 60;
this.numericUpDownMaxRetries.Minimum = 0;
this.numericUpDownMaxRetries.Maximum = 10;
}
{
/** WebProxy Options ----------------------------------------------- **/
this.comboBoxProxyType.SelectedIndex = (int)MacroscopePreferencesManager.GetProxyType();
/** Server Certificate Options --------------------------------------- **/
this.checkBoxServerCertificateValidation.Checked = MacroscopePreferencesManager.GetServerCertificateValidation();
/** Spidering Control ---------------------------------------------- **/
this.numericUpDownMaxThreads.Value = MacroscopePreferencesManager.GetMaxThreads();
this.numericUpDownDepth.Value = MacroscopePreferencesManager.GetDepth();
this.numericUpDownPageLimit.Value = MacroscopePreferencesManager.GetPageLimit();
this.numericUpDownCrawlDelay.Value = MacroscopePreferencesManager.GetCrawlDelay();
this.numericUpDownRequestTimeout.Value = (Decimal)MacroscopePreferencesManager.GetRequestTimeout();
this.numericUpDownMaxRetries.Value = (Decimal)MacroscopePreferencesManager.GetMaxRetries();
this.checkBoxCrawlStrictUrlCheck.Checked = MacroscopePreferencesManager.GetCrawlStrictUrlCheck();
this.checkBoxCheckExternalLinks.Checked = MacroscopePreferencesManager.GetCheckExternalLinks();
this.checkBoxFetchExternalLinks.Checked = MacroscopePreferencesManager.GetFetchExternalLinks();
this.checkBoxFollowRobotsProtocol.Checked = MacroscopePreferencesManager.GetFollowRobotsProtocol();
this.checkBoxFollowSitemapLinks.Checked = MacroscopePreferencesManager.GetFollowSitemapLinks();
this.checkBoxProbeHumansText.Checked = MacroscopePreferencesManager.GetProbeHumansText();
this.checkBoxCheckRedirects.Checked = MacroscopePreferencesManager.GetCheckRedirects();
this.checkBoxFollowRedirects.Checked = MacroscopePreferencesManager.GetFollowRedirects();
this.checkBoxFollowNoFollow.Checked = MacroscopePreferencesManager.GetFollowNoFollow();
this.checkBoxIgnoreQueries.Checked = MacroscopePreferencesManager.GetIgnoreQueries();
this.checkBoxIgnoreHashFragments.Checked = MacroscopePreferencesManager.GetIgnoreHashFragments();
this.checkBoxFollowCanonicalLinks.Checked = MacroscopePreferencesManager.GetFollowCanonicalLinks();
this.checkBoxFollowAlternateLinks.Checked = MacroscopePreferencesManager.GetFollowAlternateLinks();
this.checkBoxFollowHrefLangLinks.Checked = MacroscopePreferencesManager.GetFollowHrefLangLinks();
this.checkBoxDowncaseLinks.Checked = MacroscopePreferencesManager.GetDowncaseLinks();
this.checkBoxFetchStylesheets.Checked = MacroscopePreferencesManager.GetFetchStylesheets();
this.checkBoxFetchJavascripts.Checked = MacroscopePreferencesManager.GetFetchJavascripts();
this.checkBoxFetchImages.Checked = MacroscopePreferencesManager.GetFetchImages();
this.checkBoxFetchAudio.Checked = MacroscopePreferencesManager.GetFetchAudio();
this.checkBoxFetchVideo.Checked = MacroscopePreferencesManager.GetFetchVideo();
this.checkBoxFetchXml.Checked = MacroscopePreferencesManager.GetFetchXml();
this.checkBoxFetchBinaries.Checked = MacroscopePreferencesManager.GetFetchBinaries();
this.checkBoxScanSitesInList.Checked = MacroscopePreferencesManager.GetScanSitesInList();
this.checkBoxProbeParentFolderUrls.Checked = MacroscopePreferencesManager.GetProbeParentFolderUrls();
this.checkBoxProbeHead404sWithGet.Checked = MacroscopePreferencesManager.GetProbeHead404sWithGet();
/** Analysis Options ----------------------------------------------- **/
this.checkBoxResolveAddresses.Checked = MacroscopePreferencesManager.GetResolveAddresses();
this.checkBoxCheckHreflangs.Checked = MacroscopePreferencesManager.GetCheckHreflangs();
this.checkBoxDetectLanguage.Checked = MacroscopePreferencesManager.GetDetectLanguage();
this.checkBoxProcessStylesheets.Checked = MacroscopePreferencesManager.GetProcessStylesheets();
this.checkBoxProcessJavascripts.Checked = MacroscopePreferencesManager.GetProcessJavascripts();
this.checkBoxProcessImages.Checked = MacroscopePreferencesManager.GetProcessImages();
this.checkBoxProcessPdfs.Checked = MacroscopePreferencesManager.GetProcessPdfs();
this.checkBoxProcessAudio.Checked = MacroscopePreferencesManager.GetProcessAudio();
this.checkBoxProcessVideo.Checked = MacroscopePreferencesManager.GetProcessVideo();
this.checkBoxProcessXml.Checked = MacroscopePreferencesManager.GetProcessXml();
this.checkBoxProcessBinaries.Checked = MacroscopePreferencesManager.GetProcessBinaries();
this.numericUpDownRedirectChainsMaxHops.Value = MacroscopePreferencesManager.GetRedirectChainsMaxHops();
this.checkBoxWarnAboutInsecureLinks.Checked = MacroscopePreferencesManager.GetWarnAboutInsecureLinks();
this.checkBoxEnableTextIndexing.Checked = MacroscopePreferencesManager.GetEnableTextIndexing();
this.checkBoxCaseSensitiveTextIndexing.Checked = MacroscopePreferencesManager.GetCaseSensitiveTextIndexing();
this.checkBoxDisregardHtml5ElementNav.Checked = MacroscopePreferencesManager.GetDisregardHtml5ElementNav();
this.checkBoxDisregardHtml5ElementHeader.Checked = MacroscopePreferencesManager.GetDisregardHtml5ElementHeader();
this.checkBoxDisregardHtml5ElementFooter.Checked = MacroscopePreferencesManager.GetDisregardHtml5ElementFooter();
this.checkBoxDetectQrCodeInImage.Checked = MacroscopePreferencesManager.GetDetectQrCodeInImage();
/** SEO Options ---------------------------------------------------- **/
this.numericUpDownTitleMinLen.Value = MacroscopePreferencesManager.GetTitleMinLen();
this.numericUpDownTitleMaxLen.Value = MacroscopePreferencesManager.GetTitleMaxLen();
this.numericUpDownTitleMinWords.Value = MacroscopePreferencesManager.GetTitleMinWords();
this.numericUpDownTitleMaxWords.Value = MacroscopePreferencesManager.GetTitleMaxWords();
this.numericUpDownTitleMaxPixelWidth.Value = MacroscopePreferencesManager.GetTitleMaxPixelWidth();
this.numericUpDownDescriptionMinLen.Value = MacroscopePreferencesManager.GetDescriptionMinLen();
this.numericUpDownDescriptionMaxLen.Value = MacroscopePreferencesManager.GetDescriptionMaxLen();
this.numericUpDownDescriptionMinWords.Value = MacroscopePreferencesManager.GetDescriptionMinWords();
this.numericUpDownDescriptionMaxWords.Value = MacroscopePreferencesManager.GetDescriptionMaxWords();
this.numericUpDownMaxHeadingDepth.Value = MacroscopePreferencesManager.GetMaxHeadingDepth();
this.checkBoxAnalyzeKeywordsInText.Checked = MacroscopePreferencesManager.GetAnalyzeKeywordsInText();
this.checkBoxAnalyzeTextReadability.Checked = MacroscopePreferencesManager.GetAnalyzeTextReadability();
this.comboBoxAnalyzeTextReadabilityEnglishAlgorithm.SelectedIndex = (int)MacroscopePreferencesManager.GetAnalyzeTextReadabilityEnglishAlgorithm();
this.checkBoxEnableLevenshteinDeduplication.Checked = MacroscopePreferencesManager.GetEnableLevenshteinDeduplication();
this.comboBoxLevenshteinAnalysisLevel.SelectedIndex = (int)MacroscopePreferencesManager.GetLevenshteinAnalysisLevel();
this.numericUpDownMaxLevenshteinSizeDifference.Value = MacroscopePreferencesManager.GetMaxLevenshteinSizeDifference();
this.numericUpDownMaxLevenshteinDistance.Value = MacroscopePreferencesManager.GetMaxLevenshteinDistance();
this.checkBoxAnalyzeClickPaths.Checked = MacroscopePreferencesManager.GetAnalyzeClickPaths();
// TODO: Finish implementing click path analysis:
#if DEBUG
this.groupBoxPageNavigationAnalysis.Visible = true;
#else
this.groupBoxPageNavigationAnalysis.Visible = false;
#endif
/** Custom Filter Options ------------------------------------------ **/
this.checkBoxCustomFiltersEnable.Checked = MacroscopePreferencesManager.GetCustomFiltersEnable();
this.numericUpDownCustomFiltersMaxItems.Value = MacroscopePreferencesManager.GetCustomFiltersMaxItems();
this.checkBoxCustomFiltersApplyToHtml.Checked = MacroscopePreferencesManager.GetCustomFiltersApplyToHtml();
this.checkBoxCustomFiltersApplyToCss.Checked = MacroscopePreferencesManager.GetCustomFiltersApplyToCss();
this.checkBoxCustomFiltersApplyToJavascripts.Checked = MacroscopePreferencesManager.GetCustomFiltersApplyToJavascripts();
this.checkBoxCustomFiltersApplyToText.Checked = MacroscopePreferencesManager.GetCustomFiltersApplyToText();
this.checkBoxCustomFiltersApplyToXml.Checked = MacroscopePreferencesManager.GetCustomFiltersApplyToXml();
/** Extractor Options ---------------------------------------------- **/
this.checkBoxDataExtractorsEnable.Checked = MacroscopePreferencesManager.GetDataExtractorsEnable();
this.checkBoxDataExtractorsCleanWhiteSpace.Checked = MacroscopePreferencesManager.GetDataExtractorsCleanWhiteSpace();
this.numericUpDownDataExtractorsMaxItemsCssSelectors.Value = MacroscopePreferencesManager.GetDataExtractorsMaxItemsCssSelectors();
this.numericUpDownDataExtractorsMaxItemsRegexes.Value = MacroscopePreferencesManager.GetDataExtractorsMaxItemsRegexes();
this.numericUpDownDataExtractorsMaxItemsXpaths.Value = MacroscopePreferencesManager.GetDataExtractorsMaxItemsXpaths();
this.checkBoxDataExtractorsApplyToHtml.Checked = MacroscopePreferencesManager.GetDataExtractorsApplyToHtml();
this.checkBoxDataExtractorsApplyToCss.Checked = MacroscopePreferencesManager.GetDataExtractorsApplyToCss();
this.checkBoxDataExtractorsApplyToJavascripts.Checked = MacroscopePreferencesManager.GetDataExtractorsApplyToJavascripts();
this.checkBoxDataExtractorsApplyToText.Checked = MacroscopePreferencesManager.GetDataExtractorsApplyToText();
this.checkBoxDataExtractorsApplyToPdf.Checked = MacroscopePreferencesManager.GetDataExtractorsApplyToPdf();
this.checkBoxDataExtractorsApplyToXml.Checked = MacroscopePreferencesManager.GetDataExtractorsApplyToXml();
/** Export Options ------------------------------------------------- **/
this.checkBoxSitemapIncludeLinkedPdfs.Checked = MacroscopePreferencesManager.GetSitemapIncludeLinkedPdfs();
/** Ignore Errors Settings ----------------------------------------- **/
this.checkBoxIgnoreErrors410.Checked = MacroscopePreferencesManager.GetIgnoreErrors410();
this.checkBoxIgnoreErrors451.Checked = MacroscopePreferencesManager.GetIgnoreErrors451();
}
}
/**************************************************************************/
private async Task <MacroscopeConstants.FetchStatus> Fetch(string Url, string RedirectedFromUrl = null)
{
MacroscopeDocument msDoc = null;
MacroscopeConstants.FetchStatus FetchStatus = MacroscopeConstants.FetchStatus.VOID;
bool BlockedByRobotsRule;
if (MacroscopePreferencesManager.GetPageLimit() > -1)
{
int PagesFound = this.JobMaster.GetPagesFound();
int PageLimit = MacroscopePreferencesManager.GetPageLimit();
if (PagesFound >= PageLimit)
{
this.DebugMsg(string.Format("PAGE LIMIT REACHED: {0} :: {1}", PageLimit, PagesFound));
return(FetchStatus);
}
}
if (this.DocCollection.ContainsDocument(Url: Url))
{
msDoc = this.DocCollection.GetDocumentByUrl(Url: Url);
if (msDoc.GetAuthenticationRealm() != null)
{
if (msDoc.GetAuthenticationType() == MacroscopeConstants.AuthenticationType.BASIC)
{
MacroscopeCredential Credential;
Credential = this.JobMaster.GetCredentialsHttp().GetCredential(
msDoc.GetHostAndPort(),
msDoc.GetAuthenticationRealm()
);
if (Credential != null)
{
msDoc = this.DocCollection.CreateDocument(
Credential: Credential,
Url: Url
);
}
}
}
}
else
{
msDoc = this.DocCollection.CreateDocument(Url: Url);
}
if (!string.IsNullOrEmpty(RedirectedFromUrl))
{
msDoc.SetUrlRedirectFrom(Url: RedirectedFromUrl);
}
msDoc.SetFetchStatus(MacroscopeConstants.FetchStatus.OK);
if (!MacroscopeDnsTools.CheckValidHostname(Url: Url))
{
this.DebugMsg(string.Format("Fetch :: CheckValidHostname: {0}", "NOT OK"));
msDoc.SetStatusCode(HttpStatusCode.BadGateway);
FetchStatus = MacroscopeConstants.FetchStatus.NETWORK_ERROR;
msDoc.SetFetchStatus(FetchStatus);
}
if (await this.JobMaster.GetRobots().CheckRobotRule(Url: Url))
{
msDoc.SetAllowedByRobots(true);
}
else
{
msDoc.SetAllowedByRobots(false);
}
BlockedByRobotsRule = await this.JobMaster.GetRobots().ApplyRobotRule(Url: Url);
if (!BlockedByRobotsRule)
{
this.DebugMsg(string.Format("Disallowed by robots.txt: {0}", Url));
this.JobMaster.AddToBlockedByRobots(Url);
FetchStatus = MacroscopeConstants.FetchStatus.ROBOTS_DISALLOWED;
msDoc.SetFetchStatus(FetchStatus);
JobHistory.VisitedHistoryItem(Url: msDoc.GetUrl());
}
else
{
this.JobMaster.RemoveFromBlockedByRobots(Url);
}
if (this.AllowedHosts.IsExternalUrl(Url: Url))
{
this.DebugMsg(string.Format("IsExternalUrl: {0}", Url));
msDoc.SetIsExternal(State: true);
}
if (this.DocCollection.ContainsDocument(Url: Url))
{
if (!this.DocCollection.GetDocumentByUrl(Url: Url).GetIsDirty())
{
FetchStatus = MacroscopeConstants.FetchStatus.ALREADY_SEEN;
return(FetchStatus);
}
}
if (MacroscopePreferencesManager.GetDepth() >= 0)
{
int Depth = MacroscopeHttpUrlUtils.FindUrlDepth(Url: Url);
if (Depth > MacroscopePreferencesManager.GetDepth())
{
this.DebugMsg(string.Format("URL Too Deep: {0}", Depth));
FetchStatus = MacroscopeConstants.FetchStatus.SKIPPED;
return(FetchStatus);
}
}
/** ------------------------------------------------------------------ **/
if (!await msDoc.Execute())
{
this.DebugMsg(string.Format("EXECUTE FAILED: {0}", Url));
FetchStatus = MacroscopeConstants.FetchStatus.ERROR;
}
/** ------------------------------------------------------------------ **/
/** ------------------------------------------------------------------ **/
{
if (msDoc.GetStatusCode() == HttpStatusCode.Unauthorized)
{
if (msDoc.GetAuthenticationType() == MacroscopeConstants.AuthenticationType.BASIC)
{
MacroscopeCredentialsHttp CredentialsHttp = this.JobMaster.GetCredentialsHttp();
CredentialsHttp.EnqueueCredentialRequest(
Domain: msDoc.GetHostAndPort(),
Realm: msDoc.GetAuthenticationRealm(),
Url: msDoc.GetUrl()
);
this.JobMaster.AddUrlQueueItem(Url: msDoc.GetUrl());
}
}
if (msDoc.GetIsRedirect())
{
this.DebugMsg(string.Format("REDIRECTION DETECTED GetUrl: {0}", msDoc.GetUrl()));
this.DebugMsg(string.Format("REDIRECTION DETECTED From: {0}", msDoc.GetUrlRedirectFrom()));
if (MacroscopePreferencesManager.GetCheckRedirects())
{
string Hostname = msDoc.GetHostAndPort();
string HostnameFrom = MacroscopeAllowedHosts.ParseHostnameFromUrl(msDoc.GetUrlRedirectFrom());
string UrlRedirectTo = msDoc.GetUrlRedirectTo();
string HostnameTo = MacroscopeAllowedHosts.ParseHostnameFromUrl(UrlRedirectTo);
this.DebugMsg(string.Format("REDIRECTION DETECTED UrlRedirectTo: {0}", UrlRedirectTo));
this.DebugMsg(string.Format("REDIRECTION DETECTED HostnameTo: {0}", HostnameTo));
if (MacroscopePreferencesManager.GetFollowRedirects())
{
if (MacroscopePreferencesManager.GetCheckExternalLinks())
{
this.AllowedHosts.AddFromUrl(Url: UrlRedirectTo);
}
else
{
if (this.AllowedHosts.IsInternalUrl(Url: UrlRedirectTo))
{
this.AllowedHosts.AddFromUrl(Url: UrlRedirectTo);
}
}
}
}
this.JobMaster.AddUrlQueueItem(Url: msDoc.GetUrlRedirectTo());
}
else
{
this.ProcessHrefLangLanguages(msDoc); // Process Languages from HrefLang
this.JobMaster.ProcessOutlinks(msDoc: msDoc); // Process Outlinks from document
}
FetchStatus = MacroscopeConstants.FetchStatus.SUCCESS;
}
/** ------------------------------------------------------------------ **/
if (DocCollection.ContainsDocument(msDoc: msDoc))
{
JobHistory.VisitedHistoryItem(Url: Url);
}
else
{
this.DebugMsg(string.Format("OOPS: {0}", Url));
}
/** ------------------------------------------------------------------ **/
return(FetchStatus);
}
/**************************************************************************/
public async void Execute()
{
int MaxFetches = MacroscopePreferencesManager.GetMaxFetchesPerWorker();
while (MaxFetches > 0)
{
if (this.JobMaster.GetThreadsStop())
{
this.DebugMsg(string.Format("JobMaster.GetThreadsStop: {0}", this.JobMaster.GetThreadsStop()));
break;
}
else
{
MacroscopeJobItem JobItem = this.JobMaster.GetUrlQueueItem();
string Url = null;
string RedirectedFromUrl = null;
if (JobItem != null)
{
Url = JobItem.GetItemUrl();
RedirectedFromUrl = JobItem.GetItemRedirectedFromUrl();
}
if (!string.IsNullOrEmpty(Url))
{
if (!this.CheckIncludeExcludeUrl(Url))
{
Url = null;
}
}
if (!string.IsNullOrEmpty(Url))
{
if (
!MacroscopePreferencesManager.GetCrawlParentDirectories() &&
!MacroscopePreferencesManager.GetCrawlChildDirectories() &&
Url != this.JobMaster.GetStartUrl())
{
Url = null;
}
else if (
!MacroscopePreferencesManager.GetCrawlParentDirectories() ||
!MacroscopePreferencesManager.GetCrawlChildDirectories())
{
this.DebugMsg(string.Format("Running Parent/Child Check: {0}", Url));
if (
MacroscopePreferencesManager.GetCrawlParentDirectories() &&
(!string.IsNullOrEmpty(Url)))
{
if (!MacroscopeHttpUrlUtils.IsWithinParentDirectory(StartUrl: this.JobMaster.GetParentStartingDirectory(), Url: Url))
{
Url = null;
}
}
if (
MacroscopePreferencesManager.GetCrawlChildDirectories() &&
(!string.IsNullOrEmpty(Url)))
{
if (!MacroscopeHttpUrlUtils.IsWithinChildDirectory(StartUrl: this.JobMaster.GetChildStartingDirectory(), Url: Url))
{
Url = null;
}
}
}
else
{
this.DebugMsg(string.Format("Skipping Parent/Child Check: {0}", Url));
}
}
if (!string.IsNullOrEmpty(Url))
{
if (MacroscopePreferencesManager.GetDepth() >= 0)
{
if (MacroscopeHttpUrlUtils.FindUrlDepth(Url: Url) > MacroscopePreferencesManager.GetDepth())
{
this.DebugMsg(string.Format("URL Too Deep: {0}", Url));
Url = null;
}
}
}
if (!string.IsNullOrEmpty(Url))
{
this.DebugMsg(string.Format("Execute: {0}", Url));
int Tries = MacroscopePreferencesManager.GetMaxRetries();
JobHistory.AddHistoryItem(Url: Url);
do
{
this.DebugMsg(string.Format("Trying Fetch: {0} :: {1}", Tries, Url));
MacroscopeConstants.FetchStatus FetchStatus = MacroscopeConstants.FetchStatus.VOID;
try
{
if (!string.IsNullOrEmpty(RedirectedFromUrl))
{
FetchStatus = await this.Fetch(Url, RedirectedFromUrl);
}
else
{
FetchStatus = await this.Fetch(Url);
}
}
catch (Exception ex)
{
this.DebugMsg(string.Format("FetchStatus: {0}", ex.Message));
this.DebugMsg(string.Format("Url: {0}", Url));
this.DebugMsg(string.Format("FetchStatus: {0}", FetchStatus));
}
switch (FetchStatus)
{
case MacroscopeConstants.FetchStatus.ERROR:
this.DebugMsg(string.Format("Fetch Failed: {0} :: {1}", Tries, Url));
Thread.Sleep(25);
break;
case MacroscopeConstants.FetchStatus.NETWORK_ERROR:
this.DebugMsg(string.Format("Fetch Failed: {0} :: {1}", Tries, Url));
Thread.Sleep(25);
break;
default:
this.JobMaster.NotifyWorkersFetched(Url: Url);
Tries = 0;
break;
}
Tries--;
} while(Tries > 0);
if (this.CrawlDelay > 0)
{
this.DebugMsg(string.Format("CRAWL DELAY: Sleeping for {0} seconds...", this.CrawlDelay));
Thread.Sleep(CrawlDelay * 1000);
}
}
}
MaxFetches--;
//Thread.Yield();
}
this.JobMaster.NotifyWorkersDone();
}
/**************************************************************************/
private void InitializeJobMaster(MacroscopeConstants.RunTimeMode JobRunTimeMode)
{
GC.Collect();
/*
* {
* this.JobMasterLog = new EventLog ();
* this.JobMasterLog.Source = MacroscopeConstants.MainEventLogSourceName;
* this.JobGuid = Guid.NewGuid();
* this.LogEntry( string.Format( "Starting Job" ) );
* }
*/
this.RunTimeMode = JobRunTimeMode;
if (this.TaskController != null)
{
this.CredentialsHttp = this.TaskController.IGetCredentialsHttp();
}
this.DocCollection = new MacroscopeDocumentCollection(JobMaster: this);
this.AllowedHosts = new MacroscopeAllowedHosts();
/** BEGIN: Named Queues *************************************************/
this.NamedQueueJobItems = new MacroscopeNamedQueue <MacroscopeJobItem> ();
this.NamedQueueJobItems.CreateNamedQueue(
Name: MacroscopeConstants.NamedQueueUrlList,
QueueMode: MacroscopeNamedQueue <MacroscopeJobItem> .MODE.USE_HISTORY
);
this.NamedQueue = new MacroscopeNamedQueue <string> ();
{
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayQueue);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayStructure);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayStructureLinkCounts);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayHierarchy);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayCanonicalAnalysis);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayHrefLang);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayErrors);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayHostnames);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayRedirectsAudit);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayLinks);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayHyperlinks);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayUriAnalysis);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayPageTitles);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayPageDescriptions);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayPageKeywords);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayPageHeadings);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayPageText);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayStylesheets);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayImages);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayJavascripts);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayAudios);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayVideos);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplaySitemaps);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayEmailAddresses);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayTelephoneNumbers);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayCustomFilters);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayDataExtractorsCssSelectors);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayDataExtractorsRegexes);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayDataExtractorsXpaths);
this.NamedQueue.CreateNamedQueue(Name: MacroscopeConstants.NamedQueueDisplayRemarks);
}
/** END: Named Queues ***************************************************/
this.CrawlDelay = 0;
this.AdjustThreadsMax();
this.ThreadsRunning = 0;
this.ThreadsStop = false;
this.ThreadsDict = new Dictionary <int, Boolean> ();
this.SemaphoreWorkers = new Semaphore(0, this.ThreadsMax);
this.SemaphoreWorkers.Release(this.ThreadsMax);
this.Depth = MacroscopePreferencesManager.GetDepth();
this.PageLimit = MacroscopePreferencesManager.GetPageLimit();
this.PageLimitCount = 0;
this.PagesFound = 0;
{
this.ParentStartingDirectory = "";
this.ChildStartingDirectory = "";
}
this.JobHistory = new MacroscopeJobHistory();
this.InitProgress();
this.Locales = new Dictionary <string, string> (32);
this.Robots = new MacroscopeRobots();
this.BlockedByRobots = new Dictionary <string, Boolean> ();
}
