mjhjmtu

I am trying to make a check in a shell script on linux which exits if STR contains "only alphabets" or contains "only numbers". it should pass if STR contains both numbers and alphabets. And not contain special characters.

I am trying it with something like this but it works only partially.

#!/bin/sh

STR=$1

if [[ ! $STR =~ ^[[:alnum:]]*[[:alpha:]][[:alnum:]]*$ ]]; then
 echo "The input must contain both digits and alphabets like abc123"
 exit 1
fi

echo "Success"
exit 0

The check works partially, that is if STR contains only numbers. But does not work if STR contains all alphabets. I want exit 1 to not execute only if STR contains both alphabets and numbers like abc123 or ABC123.

The check fails if STR contains abc.

How can I make the check work when STR contains abc only as well as when STR contains 123 only?

The regex ^[[:alnum:]]*[[:alpha:]][[:alnum:]]*$ doesn't check for numbers separately, since [:alnum:] matches also letters. [:alpha:] should be a subset of [:alnum:], so the pattern matches any string that is all alphanumerics, with at least one letter.

(Assuming the usual greedy matching, if the input is abcd, then the abc will be matched by [[:alnum:]]*, the final d will be matched by [[:alpha:]] and the last [[:alnum:]]* will not (need to) match anything.)

If you want to check that the string contains at least one letter and one digit, it's easier to separate that to two tests, i.e. test for [[:alpha:]] and [[:digit:]]:

if [[ $str =~ [[:alpha:]] && $str =~ [[:digit:]] ]]; then
 echo "contains a letter and a digit"

If you also want to reject strings that contain anything else (like punctuation), add a check for that:

if [[ $str =~ [[:alpha:]] && $str =~ [[:digit:]] && ! $str =~ [^[:alnum:]] ]]; then
 echo "contains a letter and a digit, but no non-alphanumerics"

To find the failing cases, invert the whole test with a ! at the start (if ! [[ ...).

_{(To do that with one regex, you could use something like this:
^([[:alnum:]]*[[:alpha:]][[:alnum:]]*[[:digit:]][[:alnum:]]*|[[:alnum:]]*[[:digit:]][[:alnum:]]*[[:alpha:]][[:alnum:]]*)$
but that's a bit horrible)}

You don't need to use complicated regular expressions here. You have two conditions, so use two tests:

if [[ $str == *[[:digit:]]* ]] &&
 [[ $str == *[[:alpha:]]* ]]
then
 printf '"%s" contains both letters and digitsn' "$str"
else
 printf '"%s" lacks either letters or digitsn' "$str"
fi

You also say something about "special characters" but you don't specify what these are. Assuming you mean characters matched by [[:punct:]], and that you don't want these in the string, you could use

if [[ $str == *[[:digit:]]* ]] &&
 [[ $str == *[[:alpha:]]* ]] &&
 [[ $str != *[[:punct:]]* ]]
then
 printf '"%s" contains both letters and digits, and no specialsn' "$str"
else
 printf '"%s" lacks either letters or digits, or contains specialsn' "$str"
fi

[[:punct:]] would match any one of the characters in the string

!"#$%&'()*+,-./:;<=>?@[]^_`~

Try to make it simpler. Separate the question in two:

...
if [[ $STR =~ ([a-zA-Z]+) && $STR =~ ([0-9]+) ]]; then 
...