private string PerformRedaction(string originalReport)
{
var words = new List <string>();
var wordBuilder = new StringBuilder();
var newReportBuilder = new StringBuilder();
var newReportProcessing = originalReport;
foreach (var regex in RedactionDictionary.GetInstance().RegexCache)
{
newReportProcessing = regex.Replace(newReportProcessing, " [REDACTED]");
}
var characterGap = 0;
foreach (var c in newReportProcessing)
{
if (char.IsLetter(c) || (c == '-' || c == '.') && wordBuilder.Length > 0)
{
wordBuilder.Append(c);
if (c == '-' || c == '.')
{
characterGap++;
}
else
{
characterGap = 0;
}
}
else
{
if (wordBuilder.Length > 0)
{
CompleteWord(wordBuilder, newReportBuilder, words, characterGap);
}
newReportBuilder.Append(c);
}
}
if (wordBuilder.Length > 0)
{
CompleteWord(wordBuilder, newReportBuilder, words, characterGap);
}
var newReport = newReportBuilder.ToString();
for (var i = 0; i < words.Count; i++)
{
if (!DictionaryContains(words[i]))
{
words[i] = "[REDACTED]";
}
newReport = newReport.Replace($"{{{i}}}", words[i]);
}
// Proven init only happens the first time the code is called after its been reloaded from file
return(newReport.Trim());
}
private bool DictionaryContains(string word)
{
var lower = word.ToLowerInvariant();
if (BlacklistContains(lower))
{
return(false);
}
if (RedactionDictionary.GetInstance().BinarySearch(lower) > -1)
{
return(true);
}
if (!word.Contains('-'))
{
return(false);
}
var subwords = lower.Split('-');
foreach (var subword in subwords)
{
if (BlacklistContains(lower))
{
return(false);
}
if (RedactionDictionary.GetInstance().BinarySearch(subword) > -1)
{
return(true);
}
}
return(false);
}
public static string WipeoutReportsMatchingRemovalMarkers(string report)
{
var redactionTool = new Redaction();
// Make sure the database report removal is loaded.
// This will only run once per dictionary instance therefore use the ReloadReportRemovalCache method to refresh from the database
var b = RedactionDictionary.GetInstance().AttemptToLoadReportRemovalRegexFromDatabase();
return(b ?? redactionTool.PerformReportWipeout(report));
}
public static string CleanBoilerPlate(string report)
{
var redactionTool = new Redaction();
// Make sure the database boilerplate is loaded.
// This will only run once per dictionary instance therefore use the ReloadBoilerPlateCache method to refresh from the database
var b = RedactionDictionary.GetInstance().AttemptToLoadBoilerPlateRegexFromDatabase();
return(b ?? redactionTool.CleanOutAllBoilerPlate(report));
}
private string CleanOutAllBoilerPlate(string originalReport)
{
var cleanedReport = originalReport.Trim();
foreach (var regex in RedactionDictionary.GetInstance().BoilerPlateRegexCache)
{
cleanedReport = regex.Replace(cleanedReport, "");
}
return(cleanedReport.Trim());
}
private string PerformReportWipeout(string report)
{
var trimmed = report.Trim();
if (string.IsNullOrEmpty(trimmed))
{
return(null);
}
foreach (var regex in RedactionDictionary.GetInstance().ReportRemovalRegexCache)
{
if (regex.IsMatch(trimmed))
{
return(null);
}
}
return(trimmed);
}
public static void ResetInstance()
{
_instance = null;
}
public static RedactionDictionary GetInstance()
{
return(_instance ?? (_instance = new RedactionDictionary()));
}
public static string ReloadReportRemovalCache()
{
return(RedactionDictionary.GetInstance().ReloadReportRemovalCache());
}
public static string ReloadBoilerPlateCache()
{
return(RedactionDictionary.GetInstance().ReloadBoilerPlateCache());
}
