[SOLVED] bash: regexp in 'if' doesn't evaluate correctly

rrije · 03-01-2011, 07:19 AM

Hi,

I'm using an 'if' statement to check whether one of positional arguments is a word or a construction like "x-y", but something doesn't seem right.

Here is the relevant part of the code:

Code:

if [[ "$3" =~ [a-Z]\-[a-Z] || "$3" =~ [a-Z] ]]; then 
grepRange=$3; else grepRange=$4
fi

And for the input like this:

Code:

bash -x script --list free fornax

it outputs this:

Code:

+ [[ fornax =~ [a-Z]-[a-Z] ]]
+ [[ fornax =~ [a-Z] ]]
+ grepRange=fornax

but shouldn't the $grepRange variable be blank in this case?

Bash version is 4.1.7(1)

Is there something I'm missing? Enclosing regexp in quotes or double quotes doesn't help.

Thanks in advance.

colucix · 03-01-2011, 07:23 AM

Nope. The second regexp is matched:

Code:

[[ fornax =~ [a-Z] ]]

so that according to your code grepRange is $3. Please, can you better explain what you're trying to achieve?

rrije · 03-01-2011, 07:44 AM

Thanks!

I've adjusted it to ^[a-Z]$ and it works okay.

The script itself cats a file and breaks the list down into sections according to the first letter of a line.

grail · 03-01-2011, 08:28 AM

If only interested in the first letter, why not use that as the test:

Code:

if [[ ${3:0:1} =~ [a-Z] ]] ...

crts · 03-01-2011, 09:52 AM

Quote:

Originally Posted by grail

If only interested in the first letter, why not use that as the test:

Code:

if [[ ${3:0:1} =~ [a-Z] ]] ...

If I understood the OP correctly then he only wants the branch to execute if $3 consists of one single letter.

@OP: You probably can also use this RegEx for both your cases:

Code:

if [[ "$3" =~ ^[a-Z](-[a-Z])?$ ]];then

and get rid of the extra OR comparison. Works with my bash version 4.1.5.

rrije · 03-01-2011, 12:44 PM

Thanks for the replies everyone.

Allow me to elaborate a bit on a purpose of the script:
There is a plaintext file containing a number of entries (several files, actually); I need to be able to list these entries by filename (or everything at once) and by letter (or several letters).

Here is a part of the code (screenshot, for syntax highlighting) and exemple output; it's a mess, but as a beginner I figured you have to start somewhere. I'd be grateful for any tips you could give.

grail · 03-01-2011, 11:45 PM

Well on a cursory look, my main thought is:

Try to look at the places where you have multiple pipes, especially when it is multiple pipes to / through the same application:

Code:

ls ... | grep ... | grep ... | sed ... | xargs ... | grep ... | sort ...

I would guarantee this could lose at least 3 of the pipes if not more with a little investigation.