Equivalent regex in grep and sed of using not operator (!) in awk

Question

Suppose I want to find a regular expression of this form:

awk '!/[^TAP]/ {print $1}' sample.txt

which gives matches for ATA, but not for ATAU, TAR and TR.

On the other hand this gives matches for all four of them:

awk '/[TAP]/ {print $1}' sample.txt

What is the correct regex syntax in grep and sed for the former case?

Mock input file is:

ATA
ATAU
TAR
TR

Output should be the line composed by a string containing only characters A and/or T and/or P, but not any other character.

Answer 1

You may use this awk :

awk '/^[ATP]+$/' file

ATA

Or this grep would also work:

grep '^[ATP]\+$' file

ATA

Regex pattern ^[ATP]+$ will match one or more of A , T or P letters and due to anchors ^ and $ , it won't allow any other character.

Answer 2

This might work for you (GNU sed):

sed '/[^TAP]/d' file

If there in neither a T or an A or a P delete the line.

or:

sed -n '/[^TAP]/!p' file

Turn of implicit printing and if the regexp is false, print the line.