Updated email address matching pattern

[hiteshbedre]

Handled most of the cases mentioned under "LegacyTests.cs" file.

Sample Email	Correctly Matched
Mary&Jane@example.org	✅
""test\""blah""@example.com	✅
customer/department@example.com	✅
$A12345@example.com	✅
!def!xyz%abc@example.com	✅
_Yosemite.Sam@example.com	✅
Ima.Fool@example.com	✅
foobar@x.com	✅
foobar@c0m.com	✅
foobar@c_m.com	✅
foo@bar_.com	✅
foo@666.com	✅
1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111@example.com	✅

Invalid cases handled:

Invalid Email Address	Regex Invalidating?
char)8 + "ar.com	✅
char)9 + "ar.com	✅
char)127 + "ar.com	✅
.wooly@example.com	✅
pootietang.@example.com	✅
.@example.com	✅
foo@bar	✅
foo@bar.a	✅

Pattern not being handled via Regex:

Email	Description	Handled via Code?
pootietang.@example.com	dot_before_at	✅
wo..oly@example.com	consecutive_dots	✅
foo@bar.1com	tld_starting_with_number	✅
foobar@_.com	domain_with_underscore	✅
1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111@example.com	email_of_256_chars	✅

Fix for issue https://github.com/HaveIBeenPwned/EmailAddressExtractor/issues/5

[KonajuGames]

Regexes are one of these where there is no one correct answer. Everyone has a different take, and that is just how they seem to be designed. And I will admit I am no magician in regex. I have generally tried to avoid them in my professional career.

I've got it down to one test fail (unescaped_double_quote_is_invalid), but it does need some code filtering along with the regex. The regex expression simply cannot do it all by itself.

        public List<string> ExtractAddresses(string content)
        {
            if (string.IsNullOrWhiteSpace(content))
                return new();
            // Expression allows some false positives through. These will be filtered later.
            string addressPattern = @"[\p{L}\d\-_\+\/\&\!\%""\\\.\*]+[^\.]@[\p{L}\d][\p{L}\d\.\-_]*\.[a-z]{2,}\b";
            var matches = Regex.Matches(content, addressPattern, RegexOptions.IgnoreCase);
            var uniqueAddresses = new HashSet<string>();

            foreach (Match match in matches)
            {
                var address = match.Value;
                // Filter out the false positives that the regex expression could not handle
                if (address[0] == '.')
                    continue;
                if (address.Contains('*'))
                    continue;
                if (address.Contains(".."))
                    continue;
                if (address.Length >= 256)
                    continue;
                uniqueAddresses.Add(match.Value.ToLower());
            }

            return uniqueAddresses.OrderBy(a => a).ToList();
        }

[hiteshbedre]

Regexes are one of these where there is no one correct answer.

Completely agree with your statement.

[hiteshbedre]

accented characters

Its a new learning for me. Thanks for sharing, updated the pull request accordingly.

[troyhunt]

Good progress @hiteshbedre, thank you! Still got a few failing legacy tests, want to take a stab at those?

[hiteshbedre]

Yeah sure.