I have files in format below:
abc_abc_abc
abc_abc_abc.1
abc_bca_bca
abc_bca_bca.1
abc_bca_bca.2
abc_cab
abc
I want to get the files which is having 2 underscores and exclude other files like with . (dot) and single underscore.
The result would look like:
abc_abc_abc
abc_bca_bca
You can use find:
find . -name "*_*_*" -not -name "*.*"
This will search in subdirectories also. If you don't want this, add -maxdepth 1 and maybe -type f if you only want regular files:
find . -maxdepth 1 -type f -name "*_*_*" -not -name "*.*"
Add -printf "%P\n" to get rid of the preceding ./.
With ksh, bash -O extglob or zsh -o kshglob using some double-negations (here assuming you also want to exclude files with 3 or more underscores):
printf '%s\n' !(!(*_*_*)|*_*_*_*|*.*)
With zsh -o extendedglob and its ~ except operator:
printf '%s\n' *_*_*~(*_*_*_*|*.*)
or using zsh's native negation operator:
printf '%s\n' ^(^*_*_*|*_*_*_*|*.*)
If you still want the files with more than 2 underscores, that's just !(!(*_*_*)|*.*) or *_*_*~*.* or ^(^*_*_*|*.*) in place of the above, respectively.
Other approaches could be !(*[._]*)_!(*[._]*)_!(*[._]*) (replace [_.] with . to allow 3 or more underscore).
Simplistically, in bash (or shells that support arrays:
files=(???_???_???)
This uses the ? globbing character to pick up files named as you've shown, with 3 characters, an underscore, 3 characters, an underscore, then 3 characters. This could be foiled if you have a filename such as xyz_qrs_t.1, though.
A little more refined would be:
files=([^._][^._][^._]_[^._][^._][^._]_[^._][^._][^._])
... which uses the [^ ... ] globbing syntax. Each [^._] asks for one character that is not a period or underscore. Mixed in there are two underscores, separating three batches of three [^._].
Loop over them with:
for f in "${files[@]}"; do printf "File: %s\n" "$f"; done
xyz_qrs_t.1?