regex pattern issue

Question

Hello I have the below awk in my script . The regex pattern is not working correctly for me .I wanted to validate the email address which can have characters [a-z],[0-9] ,[.] ,@

code

here are the sample email patterns in the input file
1.abc@gmail.com
2.abc123@hotmail.com
3.abc.xyz@yahoo.com
4.a1@gmail.net
5.a2@xcom.in

the pattern is extracted from a metadata file and passed as a script paramter .here is the metadata line defines the pattern for email id validation

1~4~~~char~Y~\"\@\.com\"~100

sh -x run for the script code

val=$(
     awk -F , 
         -v n=4
         -v 'm="*@*.com"'
         -v count=0 
         'NR!=1 && $n !~ "^" m "$"
                      {
                         printf "%s:%s:%s\n", FILENAME, FNR, $0 > "/dev/stderr"
                         count++
                       }
                       END {print count}' BNC.csv

vi of the script code

val=$(awk -F "$sep"
        -v n="$col_pos" 
        -v m="$col_patt" 
        -v count=0 
        'NR!=1 && $n !~ "^" m "$" 
                       {
                         printf "%s:%s:%s\n", FILENAME, FNR, $0 > "/dev/stderr"
                         count++
                       }
                       END {print count}' $input_file

Although *@*.com is a valid regex, it probably doesn't do what you expect. See How do regular expressions differ from wildcards used to filter files — steeldriver, May 20 '21 at 18:29
you now plus this question have asked 5 questions for the same thing (validating mail) each with some small changes (or add a bit new requirement). what is your final goal?! — αғsнιη, May 20 '21 at 18:41
As well, note that the task is much harder than it at first appears - see for example How to validate an email address using a regular expression? and the previous discussion on this site Why this regex pattern for email is so popular when it does not even take in to consideration for lower case letters? — steeldriver, May 20 '21 at 19:42
You can edit questions if you feel that they need to be improved. — FairOPShotgun, May 20 '21 at 20:46
@αғsнιη Sorry for the multiple question :) . My final goal is to develop a generic flat file validation script , which has diff requirements . the challenges i had given was related to column datatype validations . i have a metadata file which defines the details of columns in a file. i need to read the metadata file and validate it against incoming files eg if i define a 4th column of file file1.txt as email text then it should match with that regex, if i define it with number it should match with number. i have completed my coding hopefully no more questions — daturm girl, May 20 '21 at 20:59
There you go with the weird code layout again but this time it's actually breaking your code as newlines between a condition (NR!=1 && $n !~ "^" m "$") and associated action ({ printf ... }) matter. Please see the example I provided for you previously for one of the few common, legible ways to format your code or, again, just run it through gawk -o- and gawk will format it for you. Also, copy/paste your code into http://shellcheck.net and it'll tell you about the shell errors in it. — Ed Morton, May 20 '21 at 23:08
Does this answer your question? Why this regex pattern for email is so popular when it does not even take in to consideration for lower case letters? — AdminBee, May 21 '21 at 09:31
@EdMorton Thanks for the Format tip . I used heavily in script ur tip. while copy pasted the same in the chat it looked very weird and hence i ended up in manually formatting . But i will follow what you said in future listings — daturm girl, May 21 '21 at 13:36

score 1 · Answer 1 · answered May 21 '21 at 13:42

1

If you're looking for a way to validate email addresses, FWIW this is what I have in an old awk script I have lying around that does that:

    # valid addrs regexp from http://www.regular-expressions.info/email.html
    # Specifically do NOT want to use [:alpha:] to drop Asian characters etc
    # Added a check that we have at least 2 consecutive alphabetic characters
    # both before and after the "@" to get rid of x@y.co etc. garbage
    (addr ~ /^[0-9a-zA-Z._%+-]+@[0-9a-zA-Z.-]+\.[a-zA-Z]{2,}$/) &&
    (addr ~ /^.*[a-zA-Z]{2}.*@.*[a-zA-Z]{2}.*\.[a-zA-Z]{2,}$/)

I'm sure that could be consolidated into 1 regexp but I don't care enough to do it and the end result would probably be less clear anyway.

answered May 21 '21 at 13:42

Ed Morton

31,617

I was trying to understand why the pattern @.com didnt worked in the awk . Then steel driver showed me a god example of file globbing and regex pattern example . But surprise for me is email id validation is far beyond what it is expected :) – daturm girl May 21 '21 at 18:14
Well, I don't know what other issues you experienced but a couple of obvious things are that . is a regexp metachar meaning any character so @.com would match @xcom, not just @.com, but more importantly valid email addresses don't contain @.com, they contain @<domain>.com. If you still have a question, though, then please update your question to contain concise, testable sample input and expected output that demonstrates the problem and reasonably formatted code plus a clear statement of what exactly the problem is, not just "The regex pattern is not working correctly for me". – Ed Morton May 21 '21 at 18:33
And yes, the set of valid email addresses is hard to validate (see https://stackoverflow.com/a/201378/1745001) but what I have in my answer is adequate for most purposes for people using the Roman alphabet. – Ed Morton May 21 '21 at 18:38
Also, when you talk about a regexp that matches an email address there's a couple of use-cases: a) finding email addresses in a block of arbitrary text, or b) validating that a specific string that should be an email address actually is one. You would write different code with slightly different regexps for each case. – Ed Morton May 21 '21 at 18:49
And there's the additional consideration of the text matching a regexp but it not actually being a valid email address, e.g. "k8e7klo9@a562nd71jd651j5l.foo" should match an email regexp but chances are there is no ".foo" TLD and if there is "a562nd71jd651j5l.foo" probably isn't a real domain at that TLD and if it is "k8e7klo9@" probably isn't a real username at that domain. – Ed Morton May 21 '21 at 18:51

regex pattern issue

1 Answers1