Finding error in my script

Question

I have a messy script which should get the name of the site (like https://google.com/etc):

#!/bin/bash ARTIST=$(echo "$@" | grep -oP 'https:\\/\\/\\K.+?(?=.com)' | sed -e "s/\b\(.\)/\u\1/g") echo $(echo "$@" | grep -oP 'https:\\/\\/\\K.+?(?=.com)' | sed -e "s/\b\(.\)/\u\1/g") echo "$ARTIST" echo "$@" 

And for some reason, $(...) doesn't return anything while running outside of the script works fine.

$ ./test.sh https://nothing.bandcamp.com/music https://nothing.bandcamp.com/music 

Expected behavior:

$ echo "https://nothing.bandcamp.com/music" | grep -oP 'https:\\/\\/\\K.+?(?=.com)' | sed -e "s/\b\(.\)/\u\1/g" Nothing.Bandcamp 

What am I doing wrong?

Answer 1

You are escaping \ itself in your grep regular expression. i.e. your \\/ means a literal backslash followed by a forward slash, and \\K means a literal backslash followed by a capital K.

Also, / doesn't even need to be escaped with grep - that's only necessary when using / as a regex delimiter, as is the default with sed or perl (and, usually, it's better to just use another delimiter - like , or :, or =).

Use just plain un-escaped /, and \K instead. For example:

#!/bin/bash ARTIST=$(echo "$@" | grep -oP 'https://\K.+?(?=.com)' | sed -e 's/\b\(.\)/\u\1/g') echo "$ARTIST" 

Sample output:

$ ./test.sh https://nothing.bandcamp.com/music Nothing.Bandcamp $ echo "https://nothing.bandcamp.com/music" | grep -oP 'https://\K.+?(?=.com)' | sed -e 's/\b\(.\)/\u\1/g' Nothing.Bandcamp 

Note: this is as true when run on the command line as it is when run in a script.

e.g. the following produces no output, same as it would in a script:

$ echo "https://nothing.bandcamp.com/music" | grep -oP 'https:\\/\\/\\K.+?(?=.com)' | sed -e "s/\b\(.\)/\u\1/g"