[SOLVED] Grep Regex to match emails with single top level domains

Issue

I am writing a bash file that matches emails using regex. But I only want to match emails with single top level domain NOT emails with multiple ones.

For example those emails should match:

[email protected]
[email protected]
[email protected]

But this email should NOT match because it has 2 top level domains .co.fr

[email protected]

I tried the following:

grep -E -o '[A-Za-z0-9.]+@[A-Za-z0-9-]+\.[A-Za-z]{2,}(?!\.[A-Za-z])' log.txt > mails.txt

But the (?!\.[A-Za-z]) part is not working with bash, my understanding that it negates the match if it finds a second domain after the first dot.

it's working fine when I try it on online tools: https://regex101.com/r/H4ftC3/1

I also tried use $ at the end: [A-Za-z0-9.]+@[A-Za-z0-9-]+\.[A-Za-z]{2,}$ but this one doesn't match anything.

How can I match only single top level domains?

Thanks

Solution

tl;dr

grep -P -i "[\w+-.]+@[\w+-]+\.[a-z]{2,}$" file.txt 

-P - option for advanced Perl like regex (allows using \w)
-i - ignore case (matches @xyz.com ou @xyz.COM)

For input file: file.txt

    [email protected]
    [email protected]
    [email protected]
    [email protected]
    [email protected]
    [email protected]
    [email protected]
    [email protected]
    [email protected]
    [email protected]
    [email protected]

Resulting:

    [email protected]
    [email protected]
    [email protected]
    [email protected]
    [email protected]
    [email protected]
    [email protected]

No fancy characters, please.

In order to answer your question it's important to make some assumptions.

1. E-mails regex are tricky, and you already read this answer on Stackoverflow (1), as well as this article on Wikipedia (2).
2. Your e-mails local part (a.k.a. user name) only have the following characters: letter from A-Za-z, numbers from 0 to 9, special characters +-_ (a very reduced of the allowed set), and dot . in the middle.
3. No fancy utf-8 or utf-16 characters. Not even latin ones (e.g. ç, ñ)

This assumption represents 99,73% of all e-mail addresses known so far.

Allowed chars

username_allowed_chars = [A-Za-z0-9_+-.]

In fact, I assume you're using gnu grep, therefore you may use grep -P (perl style regex) and the following set \w which is equivalent to [A-Za-z0-9_], thence:

username_allowed_chars = [\w+-.]

As for the domain part, remove + and dot ., thence:

domain_allowed_chars = [\w-]

Finally we will use + for 1 or more repetitions of chars.

grep -P -i "[\w+-.]+@[\w+-]+\.[a-z]{2,}$" file.txt 

I'll break this regex in parts. First the character set \w that is used extensively.

• \w - Translates do [A-Za-z0-9_] word indentifier a.k.a. allowed chars for variable names, in programming parlance. In practice disallows punctuations and other unusual characters in e-mail user name;
• \. - literal dot .;
• [\w+-.]+- One or more of these identifiers, and includes the period or dot in user names. e.g. [email protected].
• @ - literal @ to separate username from domain name.
• [a-z]{2,}$ - No less than two lowercase letters up to the end of the string (marked by $).

References

(1) Stack Overflow

(2) Wikipedia

Answered By - Jayr Magave

Answer Checked By - David Marino (WPSolving Volunteer)