Grep not working with variables, even with quotes

Question

name=$(echo "$FILENAME" | grep -E '*\.(eng|por|pt-BR)\.*')

Why this line doesn't work?

if I do

echo "Test (2013).1080p.por.mkv" | grep -E "*\.(eng|por)\.*"

It works, I tried with "" without " and with ' in my script and it seems like it's not getting piped to grep. I don't know why? I'm using bash.

FILENAME="Test (2013).1080p.por.mkv"

I want to grab only .por. or .eng in a file

Answer 1

grep is the tool to print (whole) lines matching a pattern. It's not appropriate to extract parts of a string (though some grep implementations have -o<n> or -o options that can be used in some cases for that).

Here, you can use expr:

name=$(expr " $FILENAME" : '.*\.\(eng\)\.' '|' \
            " $FILENAME" : '.*\.\(por\)\.' '|' \
            " $FILENAME" : '.*\.\(pt-PT\)\.')

(for foo.por.eng.bar, gives priority to eng over por over pt-PT).

Some expr implementations like GNU expr also support:

name=$(expr " $FILENAME" : '.*\.\(eng\|por\|pt-PT\)\.')

(here returns the rightmost occurrence if there are several in the filename)

With GNU grep or compatible:

name=$(
  printf '%s\n' "$FILENAME" |
    grep -Po '\.\K(eng|por|pt-PT)(?=\.)' |
    head -n 1
)

(returns first occurrence, replace head with tail for last)

Or you could use your shell's case construct and not run any command at all:

case $FILENAME in
  (*.por.*) name=por;;
  (*.eng.*) name=eng;;
  (*.pt-PT.*) name=pt-PT;;
esac

With bash:

re='\.(por|eng|pt-PT)\.'
[[ $FILENAME =~ $re ]]
name=${BASH_REMATCH[1]}

(first occurrence. With zsh, replace BASH_REMATCH with match)

Answer 2

Assuming the file names are like Some name.por.mkv, with the final extension fixed, you could use the POSIX shell string operations, too:

$ echo "$FILENAME"
Test (2013).1080p.por.mkv
$ x=${FILENAME%.mkv}; echo "${x##*.}"
por

Here, ${var%.mkv} removes the trailing .mkv, and ${var##*.} removes everything up to the (now) last dot. You could change the first expansion to handle other extensions by changing .mkv to just .*.