static EmailFilter()
{
//This cannot be run in the constructor and needs to run in a separate thread
//not our fault, see here: https://blogs.msdn.microsoft.com/pfxteam/2011/05/03/static-constructor-deadlocks/
new Thread((() =>
{
//Create a set of IP addresses for known bad domains
//used to reverse filter any future MX lookups for a match
//this will catch aliases for temporary email address services
BlockedMxAddresses = new HashSet <IPAddress>();
Parallel.ForEach(MxBlackList, () => new HashSet <IPAddress>(), (domain, state, local) =>
{
try
{
var results = DnsLookup.GetMXRecords(domain, out bool found);
foreach (var result in results)
{
DnsLookup.GetIpAddresses(result, out var addresses);
if (addresses == null)
{
continue;
}
foreach (var ip in addresses)
{
local.Add(ip);
}
}
}
catch
{
}
return(local);
}, (local) =>
{
foreach (var ip in local)
{
BlockedMxAddresses.Add(ip);
}
});
ReverseDnsCompleteEvent.Set();
})).Start();
}
static public bool IsProbablyFakeEmail(string email, int meanness, bool validateMx = false)
{
if (validateMx && !ReverseDnsComplete)
{
ReverseDnsCompleteEvent.Wait();
ReverseDnsComplete = true;
}
if (string.IsNullOrWhiteSpace(email))
{
return(true);
}
//Instead of making all the regex rules case-insensitive
email = email.ToLower();
if (!IsValidFormat(email))
{
return(true);
}
var mailAddress = new MailAddress(email);
if (meanness >= 0)
{
if (DomainMinimumPrefix.TryGetValue(mailAddress.Host, out var minimumPrefix) &&
minimumPrefix > mailAddress.User.Length)
{
return(true);
}
if (MistypedDomains.Contains(mailAddress.Host))
{
return(true);
}
if (MistypedTldRegex.IsMatch(mailAddress.Host))
{
return(true);
}
}
if (meanness >= 1)
{
if (BlockedDomains.Contains(mailAddress.Host))
{
return(true);
}
}
if (meanness >= 2)
{
if (ExpressionRegex.IsMatch(mailAddress.User))
{
return(true);
}
}
if (meanness >= 4)
{
if (RepeatedCharsRegex.IsMatch(mailAddress.User) ||
RepeatedCharsRegex.IsMatch(mailAddress.Host))
{
return(true);
}
}
if (meanness >= 5)
{
if (NumericEmailRegex.IsMatch(mailAddress.User))
{
return(true);
}
if (ExpressionRegex.IsMatch(email))
{
return(true);
}
}
if (meanness >= 6)
{
if (QwertyRegex.IsMatch(mailAddress.User))
{
return(true);
}
if (QwertyDomainRegex.IsMatch(mailAddress.Host))
{
return(true);
}
if (NumericDomainRegex.IsMatch(mailAddress.Host))
{
return(true);
}
}
if (meanness >= 7)
{
//this is including the tld, so 3 is insanely generous
//2 letters + period + 3 tld = 6
if (mailAddress.Host.Length < 6)
{
return(true);
}
}
if (meanness >= 8)
{
if (mailAddress.User.Length < 3)
{
return(true);
}
}
if (meanness >= 9)
{
if (mailAddress.User.Length < 5)
{
return(true);
}
}
if (meanness >= 10)
{
if (TldRegex.IsMatch(mailAddress.Host))
{
return(true);
}
}
//Do this last because it's the most expensive
if (validateMx)
{
if (!ValidDomainCache.Contains(mailAddress.Host))
{
bool mxFound;
var mxRecords = DnsLookup.GetMXRecords(mailAddress.Host, out mxFound);
if (!mxFound || !mxRecords.Any())
{
//no MX record associated with this address or timeout
return(true);
}
//compare against blacklist
foreach (var record in mxRecords)
{
DnsLookup.GetIpAddresses(record, out var addresses);
if (addresses != null && addresses.Any(BlockedMxAddresses.Contains))
{
//this mx record points to the same IP as a blacklisted MX record or timeout
return(true);
}
}
lock (ValidDomainCache)
{
ValidDomainCache.Add(mailAddress.Host);
}
}
}
return(false);
}
