hi,

i have a text file with alphabets and numbers, as follows:

TestString123
AnotherString456

i'm using awk's "sub" function to substitute all alphabets with a "" so that i'm left with just the numbers, as follows:

cat filename | gawk '{sub(/TestString/,"")} ; { print $1 }'

This works just fine for the first line of the file and i've the output:

123
AnotherString456

My problem is that if i try to use character class [:alpha:] as follows:

cat filename | gawk '{sub(/[[:alpha:]]/,"")} ; { print $1 }'

it does not work properly and it removes only a part of the alphabets and not all, like the following output:

estString123
notherString456

if i try:
cat filename | gawk '{sub(/[A-Za-z]/,"")} ; { print $1 }'

that also does not work properly:

ing123
ing456

i can't figure out why the character classes or A-Za-z are only partially working whereas if i spell out the word explicitly, it works.

any guidance will be really helpful. thanks!

You want to use gsub. sub only substitutes the first occurrence, gsub (the g stands for global) replaces all occurrences.

Yes! That was it!
Thanks a lot!!

Works as defined. You're asking for ONE occurrence to be
removed, and that's what it's doing.

If you want ALL alphas to disappear, use a +
Same with [A-Za-z] ... w/o a + (or *) it will match ONE
(exactly one) character...


Cheers,
Tink

thanks for that too!
i didn't know that... gud thing to learn!

Actually in this context sub does exactly what you
want. gsub would come in handy if the strings were
something like:

and your expected result should be 123456... I'd
say that using gsub in the way described above is
abuse :}


Cheers,
Tink

I find sed easier for substitutions:

echo st1906ghFU22 > teststring
sed 's/[[:alpha:]]//g' teststring
190622