Relative path in test file

[Dirk80]

I have my test files in a directory "t", e.g. the file ..../t/mytest.t. In mytest.t I want to open a file by using a relative path. If I want that make test is successful, I have to use a relative path like this ./t/data/test.txt. But if would call e.g. perl mytest.t it would fail, because the relative path should be ./data/test.txt.

Is there an elegant way to open a file with a relative path independent if I call it from the "t" directory or its parent directory?

Thank you!

[Corion]

$0 contains the name of the program as it was invoked. In your test cases, $0 will be:

perl t/mytest.t
# t/mytest.t

perl mytest.t
# mytest.t

So, if you want to open the file relative to your main program, use a path relative to $0. The FindBin module does that already, so you can just use that:

use FindBin;
my $datafile = "$FindBin::Bin/data/test.txt";
open my $fh, $datafile
    or die "Couldn't open '$datafile': $!";

[ikegami]

Whever it makes a difference, $FindBin::RealBin is better than $FindBin::Bin. So might as well always use it.

[Anonymous Monk]

Whever it makes a difference, $FindBin::RealBin is better than $FindBin::Bin.

why?

[haukex]

From your problem description it's a bit unclear to me whether you mean that you simply want to locate a file relative to the location of your test script, or you mean that the filename string you work with in your script should be expressed as a relative path (e.g. absolute "/path/to/data/test.txt" vs. relative "../data/test.txt"). The former is more common and what Corion's reply is doing - though I would personally recommend using catfile from File::Spec instead of manually concatenating the paths for maximum portability. If you need the latter, which is sometimes useful e.g. when you have to test an API and make sure it handles relative filenames correctly, then you can take the previously mentioned approach and additionally apply abs2rel from File::Spec on the filename.

[Dirk80]

Thank you all so much. This community here is so great.

Corion's solution already worked great. But I extended it now by using $FindBin::RealBin and doing the concatenation by using File::Spec->catfile(...).

[JayBee]

Consider these from CGI::Simple

$q = CGI::Simple->new;

$full_url      = $q->url();
$full_url      = $q->url(-full=>1);
$relative_url  = $q->url(-relative=>1);
$absolute_url  = $q->url(-absolute=>1);
$url_with_path = $q->url(-path_info=>1);
$url_with_path_and_query = $q->url(-path_info=>1,-query=>1);
$netloc        = $q->url(-base => 1);

[haukex]

Sorry, but the question has nothing to do with CGI, and URL paths are not the same as file system paths.