public bool ClassifyFile(FileInfo fileInfo)
{
BlockingMq Mq = BlockingMq.GetMq();
// figure out what part we gonna look at
string stringToMatch = null;
switch (ClassifierRule.MatchLocation)
{
case MatchLoc.FileExtension:
stringToMatch = fileInfo.Extension;
// special handling to treat files named like 'thing.kdbx.bak'
if (stringToMatch == ".bak")
{
// strip off .bak
string subName = fileInfo.Name.Replace(".bak", "");
stringToMatch = Path.GetExtension(subName);
// if this results in no file extension, put it back.
if (stringToMatch == "")
{
stringToMatch = ".bak";
}
}
// this is insane that i have to do this but apparently files with no extension return
// this bullshit
if (stringToMatch == "")
{
return(false);
}
break;
case MatchLoc.FileName:
stringToMatch = fileInfo.Name;
break;
case MatchLoc.FilePath:
stringToMatch = fileInfo.FullName;
break;
case MatchLoc.FileLength:
if (!SizeMatch(fileInfo))
{
return(false);
}
else
{
break;
}
default:
Mq.Error("You've got a misconfigured file classifier rule named " + ClassifierRule.RuleName + ".");
return(false);
}
TextResult textResult = null;
if (!String.IsNullOrEmpty(stringToMatch))
{
TextClassifier textClassifier = new TextClassifier(ClassifierRule);
// check if it matches
textResult = textClassifier.TextMatch(stringToMatch);
if (textResult == null)
{
// if it doesn't we just bail now.
return(false);
}
}
FileResult fileResult;
// if it matches, see what we're gonna do with it
switch (ClassifierRule.MatchAction)
{
case MatchAction.Discard:
// chuck it.
return(true);
case MatchAction.Snaffle:
// snaffle that bad boy
fileResult = new FileResult(fileInfo)
{
MatchedRule = ClassifierRule,
TextResult = textResult
};
// if the file was list-only, don't bother sending a result back to the user.
if (!fileResult.RwStatus.CanRead && !fileResult.RwStatus.CanModify && !fileResult.RwStatus.CanWrite)
{
return(false);
}
;
Mq.FileResult(fileResult);
return(false);
case MatchAction.CheckForKeys:
// do a special x509 dance
List <string> x509MatchReason = x509Match(fileInfo);
if (x509MatchReason.Count >= 0)
{
// if there were any matchreasons, cat them together...
string matchContext = String.Join(",", x509MatchReason);
// and sling the results on the queue
fileResult = new FileResult(fileInfo)
{
MatchedRule = ClassifierRule,
TextResult = new TextResult()
{
MatchContext = matchContext,
MatchedStrings = new List <string>()
{
""
}
}
};
if (!fileResult.RwStatus.CanRead && !fileResult.RwStatus.CanModify && !fileResult.RwStatus.CanWrite)
{
return(false);
}
;
Mq.FileResult(fileResult);
}
return(false);
case MatchAction.Relay:
// bounce it on to the next ClassifierRule
try
{
bool fLoggedContentSizeWarning = false;
foreach (string relayTarget in ClassifierRule.RelayTargets)
{
ClassifierRule nextRule =
MyOptions.ClassifierRules.First(thing => thing.RuleName == relayTarget);
if (nextRule.EnumerationScope == EnumerationScope.ContentsEnumeration)
{
if (fileInfo.Length > MyOptions.MaxSizeToGrep)
{
if (!fLoggedContentSizeWarning)
{
// Just log once per relay rule, no need to fill up the log with one for each relay target
Mq.Trace("The following file was bigger than the MaxSizeToGrep config parameter:" + fileInfo.FullName);
fLoggedContentSizeWarning = true;
}
continue;
}
ContentClassifier nextContentClassifier = new ContentClassifier(nextRule);
nextContentClassifier.ClassifyContent(fileInfo);
}
else if (nextRule.EnumerationScope == EnumerationScope.FileEnumeration)
{
FileClassifier nextFileClassifier = new FileClassifier(nextRule);
nextFileClassifier.ClassifyFile(fileInfo);
}
else
{
Mq.Error("You've got a misconfigured file ClassifierRule named " + ClassifierRule.RuleName + ".");
}
}
return(false);
}
catch (IOException e)
{
Mq.Trace(e.ToString());
}
catch (Exception e)
{
Mq.Error("You've got a misconfigured file ClassifierRule named " + ClassifierRule.RuleName + ".");
Mq.Trace(e.ToString());
}
return(false);
case MatchAction.EnterArchive:
// do a special looking inside archive files dance using
// https://github.com/adamhathcock/sharpcompress
// TODO FUUUUUCK
throw new NotImplementedException("Haven't implemented walking dir structures inside archives.");
default:
Mq.Error("You've got a misconfigured file ClassifierRule named " + ClassifierRule.RuleName + ".");
return(false);
}
}
public void ClassifyContent(FileInfo fileInfo)
{
BlockingMq Mq = BlockingMq.GetMq();
FileResult fileResult;
try
{
if (MyOptions.MaxSizeToGrep >= fileInfo.Length)
{
// figure out if we need to look at the content as bytes or as string.
switch (ClassifierRule.MatchLocation)
{
case MatchLoc.FileContentAsBytes:
byte[] fileBytes = File.ReadAllBytes(fileInfo.FullName);
if (ByteMatch(fileBytes))
{
fileResult = new FileResult(fileInfo)
{
MatchedRule = ClassifierRule
};
Mq.FileResult(fileResult);
}
return;
case MatchLoc.FileContentAsString:
string fileString = File.ReadAllText(fileInfo.FullName);
TextClassifier textClassifier = new TextClassifier(ClassifierRule);
TextResult textResult = textClassifier.TextMatch(fileString);
if (textResult != null)
{
fileResult = new FileResult(fileInfo)
{
MatchedRule = ClassifierRule,
TextResult = textResult
};
Mq.FileResult(fileResult);
}
return;
case MatchLoc.FileLength:
bool lengthResult = SizeMatch(fileInfo);
if (lengthResult)
{
fileResult = new FileResult(fileInfo)
{
MatchedRule = ClassifierRule
};
Mq.FileResult(fileResult);
}
return;
case MatchLoc.FileMD5:
bool Md5Result = MD5Match(fileInfo);
if (Md5Result)
{
fileResult = new FileResult(fileInfo)
{
MatchedRule = ClassifierRule
};
Mq.FileResult(fileResult);
}
return;
default:
Mq.Error("You've got a misconfigured file ClassifierRule named " + ClassifierRule.RuleName + ".");
return;
}
}
else
{
Mq.Trace("The following file was bigger than the MaxSizeToGrep config parameter:" + fileInfo.FullName);
}
}
catch (UnauthorizedAccessException)
{
Mq.Error($"Not authorized to access file: {fileInfo.FullName}");
return;
}
catch (IOException e)
{
Mq.Error($"IO Exception on file: {fileInfo.FullName}. {e.Message}");
return;
}
catch (Exception e)
{
Mq.Error(e.ToString());
return;
}
}
public bool ClassifyFile(FileInfo fileInfo)
{
BlockingMq Mq = BlockingMq.GetMq();
// figure out what part we gonna look at
string stringToMatch = null;
switch (ClassifierRule.MatchLocation)
{
case MatchLoc.FileExtension:
stringToMatch = fileInfo.Extension;
// special handling to treat files named like 'thing.kdbx.bak'
if (stringToMatch == ".bak")
{
// strip off .bak
string subName = fileInfo.Name.Replace(".bak", "");
stringToMatch = Path.GetExtension(subName);
// if this results in no file extension, put it back.
if (stringToMatch == "")
{
stringToMatch = ".bak";
}
}
// this is insane that i have to do this but apparently files with no extension return
// this bullshit
if (stringToMatch == "")
{
return(false);
}
break;
case MatchLoc.FileName:
stringToMatch = fileInfo.Name;
break;
case MatchLoc.FilePath:
stringToMatch = fileInfo.FullName;
break;
case MatchLoc.FileLength:
if (!SizeMatch(fileInfo))
{
return(false);
}
else
{
break;
}
default:
Mq.Error("You've got a misconfigured file classifier rule named " + ClassifierRule.RuleName + ".");
return(false);
}
TextResult textResult = null;
if (!String.IsNullOrEmpty(stringToMatch))
{
TextClassifier textClassifier = new TextClassifier(ClassifierRule);
// check if it matches
textResult = textClassifier.TextMatch(stringToMatch);
if (textResult == null)
{
// if it doesn't we just bail now.
return(false);
}
}
FileResult fileResult;
// if it matches, see what we're gonna do with it
switch (ClassifierRule.MatchAction)
{
case MatchAction.Discard:
// chuck it.
return(true);
case MatchAction.Snaffle:
// snaffle that bad boy
fileResult = new FileResult(fileInfo)
{
MatchedRule = ClassifierRule,
TextResult = textResult
};
Mq.FileResult(fileResult);
return(true);
case MatchAction.CheckForKeys:
// do a special x509 dance
if (x509PrivKeyMatch(fileInfo))
{
fileResult = new FileResult(fileInfo)
{
MatchedRule = ClassifierRule
};
Mq.FileResult(fileResult);
}
return(true);
case MatchAction.Relay:
// bounce it on to the next ClassifierRule
// TODO concurrency uplift make this a new task on the poolq
try
{
ClassifierRule nextRule =
MyOptions.ClassifierRules.First(thing => thing.RuleName == ClassifierRule.RelayTarget);
if (nextRule.EnumerationScope == EnumerationScope.ContentsEnumeration)
{
ContentClassifier nextContentClassifier = new ContentClassifier(nextRule);
nextContentClassifier.ClassifyContent(fileInfo);
return(true);
}
else if (nextRule.EnumerationScope == EnumerationScope.FileEnumeration)
{
FileClassifier nextFileClassifier = new FileClassifier(nextRule);
nextFileClassifier.ClassifyFile(fileInfo);
return(true);
}
else
{
Mq.Error("You've got a misconfigured file ClassifierRule named " + ClassifierRule.RuleName + ".");
return(false);
}
}
catch (IOException e)
{
Mq.Trace(e.ToString());
}
catch (Exception e)
{
Mq.Error("You've got a misconfigured file ClassifierRule named " + ClassifierRule.RuleName + ".");
Mq.Trace(e.ToString());
}
return(false);
case MatchAction.EnterArchive:
// do a special looking inside archive files dance using
// https://github.com/adamhathcock/sharpcompress
// TODO FUUUUUCK
throw new NotImplementedException("Haven't implemented walking dir structures inside archives. Prob needs pool queue.");
default:
Mq.Error("You've got a misconfigured file ClassifierRule named " + ClassifierRule.RuleName + ".");
return(false);
}
}
public void ClassifyContent(FileInfo fileInfo)
{
BlockingMq Mq = BlockingMq.GetMq();
FileResult fileResult;
try
{
if (MyOptions.MaxSizeToGrep >= fileInfo.Length)
{
// figure out if we need to look at the content as bytes or as string.
switch (ClassifierRule.MatchLocation)
{
case MatchLoc.FileContentAsBytes:
byte[] fileBytes = File.ReadAllBytes(fileInfo.FullName);
if (ByteMatch(fileBytes))
{
fileResult = new FileResult(fileInfo)
{
MatchedRule = ClassifierRule
};
// if the file was list-only, don't bother sending a result back to the user.
if (!fileResult.RwStatus.CanRead && !fileResult.RwStatus.CanModify && !fileResult.RwStatus.CanWrite)
{
return;
}
;
Mq.FileResult(fileResult);
}
return;
case MatchLoc.FileContentAsString:
try
{
string fileString;
#if ULTRASNAFFLER
// if it's an office doc or a PDF or something, parse it to a string first i guess?
List <string> parsedExtensions = new List <string>()
{
".doc", ".docx", ".xls", ".xlsx", ".eml", ".msg", ".pdf", ".ppt", ".pptx", ".rtf", ".docm", ".xlsm", ".pptm", ".dot", ".dotx", ".dotm", ".xlt", ".xlsm", ".xltm"
};
if (parsedExtensions.Contains(fileInfo.Extension))
{
fileString = ParseFileToString(fileInfo);
}
else
{
fileString = File.ReadAllText(fileInfo.FullName);
}
#else
fileString = File.ReadAllText(fileInfo.FullName);
#endif
TextClassifier textClassifier = new TextClassifier(ClassifierRule);
TextResult textResult = textClassifier.TextMatch(fileString);
if (textResult != null)
{
fileResult = new FileResult(fileInfo)
{
MatchedRule = ClassifierRule,
TextResult = textResult
};
// if the file was list-only, don't bother sending a result back to the user.
if (!fileResult.RwStatus.CanRead && !fileResult.RwStatus.CanModify && !fileResult.RwStatus.CanWrite)
{
return;
}
;
Mq.FileResult(fileResult);
}
}
catch (UnauthorizedAccessException)
{
return;
}
catch (IOException)
{
return;
}
return;
case MatchLoc.FileLength:
bool lengthResult = SizeMatch(fileInfo);
if (lengthResult)
{
fileResult = new FileResult(fileInfo)
{
MatchedRule = ClassifierRule
};
// if the file was list-only, don't bother sending a result back to the user.
if (!fileResult.RwStatus.CanRead && !fileResult.RwStatus.CanModify && !fileResult.RwStatus.CanWrite)
{
return;
}
;
Mq.FileResult(fileResult);
}
return;
case MatchLoc.FileMD5:
bool Md5Result = MD5Match(fileInfo);
if (Md5Result)
{
fileResult = new FileResult(fileInfo)
{
MatchedRule = ClassifierRule
};
// if the file was list-only, don't bother sending a result back to the user.
if (!fileResult.RwStatus.CanRead && !fileResult.RwStatus.CanModify && !fileResult.RwStatus.CanWrite)
{
return;
}
;
Mq.FileResult(fileResult);
}
return;
default:
Mq.Error("You've got a misconfigured file ClassifierRule named " + ClassifierRule.RuleName + ".");
return;
}
}
else
{
Mq.Trace("The following file was bigger than the MaxSizeToGrep config parameter:" + fileInfo.FullName);
}
}
catch (UnauthorizedAccessException)
{
Mq.Error($"Not authorized to access file: {fileInfo.FullName}");
return;
}
catch (IOException e)
{
Mq.Error($"IO Exception on file: {fileInfo.FullName}. {e.Message}");
return;
}
catch (Exception e)
{
Mq.Error(e.ToString());
return;
}
}