public async Task <IActionResult> GenerateEmailSentence(string messageId, bool isSpam)
{
var(mail, uniqueId) = await _mailKitRepository.GetUnreadEmailByIdAsync(messageId).ConfigureAwait(false);
var headerEmailSentence = new StringBuilder();
var authorizationResultsHeaders = mail.Headers.Where(header => header.Field == "Authentication-Results");
var authorizationResultsHeadersValues = authorizationResultsHeaders.Select(authorizationResultsHeader => authorizationResultsHeader.Value).ToList();
var senderPolicyFrameworkProtocolsResults = _informationExtractorService.GetSenderPolicyFrameworkResults(authorizationResultsHeadersValues);
var domainKeysIdentifiedMailProtocolsResults = _informationExtractorService.GetDomainKeysIdentifiedMailResults(authorizationResultsHeadersValues);
var domainBasedMessageAuthenticationReportingConformanceProtocolsResults = _informationExtractorService.GetDomainBasedMessageAuthenticationReportingConformanceResults(authorizationResultsHeadersValues);
//Add attachments info
foreach (var attachment in mail.Attachments)
{
if (attachment.ContentDisposition.FileName != null)
{
var file = attachment.ContentDisposition.FileName;
var fileSize = attachment.ContentDisposition.Size;
var fileName = Path.GetFileName(file).ToLower();
var fileExtension = Path.GetExtension(file).ToLower();
headerEmailSentence.Append($"attachment filename {fileName} extension {fileExtension} size {fileSize} ");
}
}
//Add protocol result info
foreach (var protocolResult in senderPolicyFrameworkProtocolsResults)
{
headerEmailSentence.Append($"spfProtocol result {protocolResult} ");
}
foreach (var protocolResult in domainKeysIdentifiedMailProtocolsResults)
{
headerEmailSentence.Append($"dkimProtocol result {protocolResult} ");
}
foreach (var protocolResult in domainBasedMessageAuthenticationReportingConformanceProtocolsResults)
{
headerEmailSentence.Append($"dmarcProtocol result {protocolResult} ");
}
//Add body info
var bodyHtmlText = mail.HtmlBody;
var uris = Regex
.Matches(bodyHtmlText, @"((http|ftp|https):\/\/[\w\-_]+(\.[\w\-_]+)+([\w\-\.,@?^=%&:/~\+#]*[\w\-\@?^=%&/~\+#])?)", RegexOptions.IgnoreCase)
.Select(match => match.Value)
.ToList();
var subjectSentence = _textPreparerService.Prepare(mail.Subject).ToList();
var subjectLabel = Convert.ToInt32(isSpam);
var headerSentence = _textPreparerService.Prepare(headerEmailSentence.ToString()).ToList();
var headerLabel = Convert.ToInt32(isSpam);
var bodySentence = _textPreparerService.Prepare(GetBodyText(bodyHtmlText)).ToList();
var bodyLabel = Convert.ToInt32(isSpam);
await _grannyRepository.AddSubjectArray(subjectSentence, subjectLabel).ConfigureAwait(false);
await _grannyRepository.AddHeaderArray(headerSentence, headerLabel).ConfigureAwait(false);
await _grannyRepository.AddBodyArray(bodySentence, bodyLabel).ConfigureAwait(false);
foreach (var uri in uris)
{
await _grannyRepository.AddUri(uri, (int)UriEvaluationCategory.NotEvaluated).ConfigureAwait(false);
}
await _mailKitRepository.MarkEmailAsSeen(uniqueId).ConfigureAwait(false);
var trainedEmailAddedResultViewModel = new TrainedEmailAddedResultViewModel()
{
SubjectSentence = subjectSentence,
HeaderSentence = headerSentence,
BodySentence = bodySentence,
Uris = uris
};
return(View(trainedEmailAddedResultViewModel));
}
