/** -------------------------------------------------------------------- **/
public List <KeyValuePair <string, string> > AnalyzeHtmlDoc(HtmlDocument HtmlDoc)
{
List <KeyValuePair <string, string> > ResultList = new List <KeyValuePair <string, string> > (32);
if (this.IsEnabled())
{
for (int Slot = 0; Slot < this.GetSize(); Slot++)
{
IList <HtmlNode> NodeSet;
string Label = this.ExtractCssSelectors[Slot].Key;
string Expression = this.ExtractCssSelectors[Slot].Value.Expression;
MacroscopeConstants.DataExtractorType ExtractorType = this.ExtractCssSelectors[Slot].Value.ExtractorType;
if (this.GetActiveInactive(Slot).Equals(MacroscopeConstants.ActiveInactive.INACTIVE))
{
continue;
}
else
if (!SyntaxCheckCssSelector(CssSelectorString: Expression))
{
continue;
}
NodeSet = HtmlDoc.QuerySelectorAll(Expression);
if (
(NodeSet != null) &&
(NodeSet.Count > 0))
{
foreach (HtmlNode Node in NodeSet)
{
KeyValuePair <string, string> Pair;
string Text;
switch (ExtractorType)
{
case MacroscopeConstants.DataExtractorType.OUTERHTML:
Text = Node.OuterHtml;
Pair = new KeyValuePair <string, string> (key: Label, value: Text);
ResultList.Add(item: Pair);
break;
case MacroscopeConstants.DataExtractorType.INNERHTML:
Text = Node.InnerHtml;
Pair = new KeyValuePair <string, string> (key: Label, value: Text);
ResultList.Add(item: Pair);
break;
case MacroscopeConstants.DataExtractorType.INNERTEXT:
Text = Node.InnerText;
if (MacroscopePreferencesManager.GetDataExtractorsCleanWhiteSpace())
{
Text = this.CleanWhiteSpace(Text: Text);
}
Pair = new KeyValuePair <string, string> (key: Label, value: Text);
ResultList.Add(item: Pair);
break;
default:
break;
}
}
}
}
}
return(ResultList);
}
/**************************************************************************/
public List <KeyValuePair <string, string> > AnalyzeText(string Text)
{
List <KeyValuePair <string, string> > ResultList = new List <KeyValuePair <string, string> > (8);
if (this.IsEnabled())
{
for (int Slot = 0; Slot < this.GetSize(); Slot++)
{
MatchCollection PatternMatches;
string SlotRegex;
this.DebugMsg(
string.Format(
"SLOT KEY: {0} => {1} => {2}",
Slot,
this.ExtractRegexes[Slot].Key,
this.ExtractRegexes[Slot].Value
)
);
if (string.IsNullOrEmpty(this.ExtractRegexes[Slot].Key))
{
continue;
}
SlotRegex = this.ExtractRegexes[Slot].Value.ToString();
PatternMatches = Regex.Matches(
Text,
SlotRegex,
RegexOptions.Singleline
);
for (int i = 0; i < PatternMatches.Count; i++)
{
Match match = PatternMatches[i];
for (int j = 1; j < match.Groups.Count; j++)
{
Group captured = match.Groups[j];
this.DebugMsg(string.Format("captured: {0} => \"{1}\"", captured.Index, captured.Value));
string FoundString = captured.Value;
if (!string.IsNullOrEmpty(FoundString))
{
KeyValuePair <string, string> MatchedItem;
if (MacroscopePreferencesManager.GetDataExtractorsCleanWhiteSpace())
{
FoundString = this.CleanWhiteSpace(Text: FoundString);
}
MatchedItem = new KeyValuePair <string, string> (
this.ExtractRegexes[Slot].Key,
FoundString
);
ResultList.Add(MatchedItem);
}
}
}
}
}
return(ResultList);
}
/**************************************************************************/
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();
}
}