Perl - Relative path of a file inside a module - perl

I am building up a website using Perl. I organized my files as follow:
/index.cgi
/perl/modules/databaseFunctions.pm
/perl/indexCheck.cgi
/database/database.xml
Inside databaseFunctions.pm I have a function X() that reads and writes on database.xml. In X() I have specified relative path of the database as follow:
sub X{
my $db_path='../../database/database.xml';
my $parser=XML::LibXML->new();
my $doc=$parser->parse_file($db_path);
....
....
}
Here is the problem:
I have to call X() from index.cgi and indexCheck.cgi but I get an error the following error:
Could not create file parser context for file "../../database/database.xml": No such file or directory at perl/modules/databaseFunctions.pm line 21.
I think the problem is that when I call X() inside index.cgi or inside /perl/indexCheck.cgi the relative path of the database is different but I don't know how to set a path that works for index.cgi and /perl/indexCheck.cgi.

I think the problem boils down to "How to find out the path of the current script (*.pl)?" and
"How to find out the path of the current module (*.pm)?".
Scripts
For scripts, there is a very convenient module, FindBin, that offers 4 variables for the current script's name and path with
either symlinks resolved or not. Usually $FindBin::Bin is what you are looking for. It's the path of the current script.
I often use it to enhance the #INC path so that my scripts find additional (own) modules like so:
use FindBin;
use lib "$FindBin::Bin/my_mod_path";
use MyModule;
In this case MyModule.pm is searched for in the directory my_mod_path below the current script's path. Very convenient.
The module is part of the core distribution, i.e. no further installation is neccessary.
Modules
FindBin may not safely be used from inside modules because then it depends who (script or module) makes the first use FindBin;.
So if you don't want to care about the order, don't use FindBin; in modules, only in scripts.
For modules, there is some trick. Use the perl function caller().
Depending on the context called in, it returns the $filename of the file where it actually was called.
Thus, in modules you can safely use the following to get the module's path:
use File::Basename;
my $path_of_this_module = File::Basename::dirname( eval { ( caller() )[1] } );
Given that path you can navigate relative to it in order to find the other files you need, e.g. "$path_of_this_module/../.." and so on.
EDIT
index.cgi:
#!/usr/bin/env perl
use strict;
use warnings;
use FindBin;
use lib "$FindBin::Bin/perl/modules";
use databaseFunctions;
databaseFunctions::X( "called from index.cgi\n" );
perl/indexCheck.cgi:
#!/usr/bin/env perl
use strict;
use warnings;
use FindBin;
use lib "$FindBin::Bin/modules";
use databaseFunctions;
databaseFunctions::X( "called from indexCheck.cgi\n" );
perl/modules/databaseFunctions.pm:
package databaseFunctions;
use File::Basename;
my $path_of_this_module = File::Basename::dirname( eval { ( caller() )[1] } );
sub X {
my $arg = shift;
my $db_path="$path_of_this_module/../../database/database.xml";
open(my $fh, '>>', $db_path) or die "cannot open $db_path: $!\n";
print $fh $arg;
close($fh);
}
1;
When I now call ./index.cgi and then ./perl/indexCheck.cgi, then I get the following:
database/database.xml:
called from index.cgi
called from indexCheck.cgi
Exactly, what I thought you were looking for.

Related

Perl, cannot require a module dynamicaly from #INC

So, i've got a problem with loading a module via require.
We have a working directory. The program loads a PACKAGE (bolded not to make you confused) (which is ok - thanks to correct and local namespaces), but it has to load another module from very different directory. So, as i've heard, it should be unshifted into #INC in BEGIN block. But....This begin should take a param (currently from initial programm), containing some path to configuration file, which containing parameter i need (path to module, which i need to unshift).
BEGIN inited, i check #INC - unshift seems to be succeed. Then, in PACKAGE methods we need to load this module, but when i try to do something like:
eval{
print STDERR "Trying...\n";
my $path = "path/to/module"; # contains "DIR" dir and "Module.pm",
# also tried to write this path as "path/to/module/DIR/Module.pm"
require $path;
DIR::Module->import();
print STDERR "Success\n";
1;
} or {print STDERR "Failed\n";}
my $module = DIR::Module->new();
And I got "Trying.." and "Failed". Tried use lib with fullpath - got nothing. What am i doing wrong?
You say $path is the path to the module, but you also say it's the path to the directory containing DIR/Module.pm. I'm assuming it's the latter because it needs to be the former.
my $path = "/path/to/lib";
require "$path/DIR/Module.pm";
Remember to use $RealBin if the path is relative to your script.
use FindBin qw( $RealBin );
my $path = "$RealBin/../lib";
require "$path/DIR/Module.pm";
Note that it rarely makes sense to import from a dynamically loaded module.
If there's no reason to avoid loading this module at compile-time, I'd go with
use lib qw( /path/to/lib );
use DIR::Module;
Again, remember to use $RealBin if the path is relative to your script.
use FindBin qw( $RealBin );
use lib "$RealBin/../lib";
use DIR::Module;

How can I mock builtin `require` function?

I'm going to refactoring large number of old perl scripts. (over 150k lines, no tests, no strict, no packages, no commit logs, no comments)
When I started to write some tests under t directory. I found almost all files require each others with absolute paths. So I tried mocking builtin require function to make them portable, but no luck:
t/001-require.t
use strict;
use warnings;
use Test::More;
use FindBin;
BEGIN {
my $root = "$FindBin::RealBin/../";
sub require {
$_[0] = $root . $_[0];
CORE::require(#_);
}
}
require_ok "foo.pl";
done_testing();
The above script gives me: Error: Can't locate foo.pl in #INC...
How can I prepend a root path before Perl requires them?
update
Thank you Axeman, I modified absolute paths using following hook.
my $root = "$RealBin/../";
unshift #INC, sub {
my ($coderef, $filename) = #_;
$filename =~ s/^\///;
open(FILE, '<', $root . $filename);
return *FILE;
};
Aren't you simply looking for
use FindBin qw( $RealBin );
use lib "$RealBin/..";
You don't have to.
When a module is required, either by require or use, perl looks through a list to see if it can locate the module by the list of (usually) directories set up as libraries for the installation. This list is stored in a variable called #INC.
However, #INC takes more than directories, it also takes "hooks", which are subroutines which can change the loading behavior for Perl modules. If you insert a hook (a subroutine) as the first entry into #INC, require will call your behavior.
You can find a more complete treatment at the perldoc on require. I just wanted to give a quick profile of a hook:
sub inc_hook {
my ( $ref_to_this_sub, $relative_module_path ) = #_;
# $relative_module_path will be in directory form: Root/Package.pm
# return nothing to pass it to standard behavior
return unless _i_want_to_handle( $relative_module_path );
# most commonly, return handle to source
return my $io_handle_to_source = handle_module( $relative_module_path );
}
unshift #INC, inc_hook;

Using FindBin in more than one module

Here is the scenario, I have two files:
1. dir/A.pm
2. dir/new_dir/a.t
This is how A.pm looks like:
package A;
use FindBin;
use Test::More;
is (FindBin->again, 'dir', 'got dir');
1;
This is how a.t looks like:
use FindBin;
use Test::More qw(no_plan);
use A;
is (FindBin->again, 'dir/new_dir', 'got dir/new_dir');
So I ran file a.t with perl new_dir/a.t and expect my tests to pass. But this is my test result:
not ok 1 - got dir
# Failed test 'got fir'
# at A.pm line 6.
# got: 'dir/new_dir'
# expected: 'dir'
ok 2 - got dir/new_dir
1..2
Am I doing anything wrong? I am very new to perl. Please help!!
As Dave Sherohman notes, FindBin is for finding the location of the main script, not individual modules. From the documentation:
NAME
FindBin - Locate directory of original perl script
(Admittedly, the documentation does, somewhat confusingly, refer to "modules" in the "KNOWN ISSUES" section, but it doesn't really mean what you think it means by that.)
Anyway, if you look at the source with perldoc -m FindBin, you'll see that FindBin obtains the path to the script from the $0 variable. If you're interested in finding the location of a module included via use (or require), you should look under %INC instead, something like this:
package Foo::Bar;
use File::Spec;
my ($vol, $dir, $file) = File::Spec->splitpath( $INC{'Foo/Bar.pm'} );
FindBin finds the location of the file Perl launched, not of file currently executing.
I don't see why you'd need the path to a module — File::ShareDir can be used to access your module's data files — but the following will find it:
use Cwd qw( realpath );
use File::Basename qw( dirname );
my $module_dir = dirname(realpath(__FILE__));
The same caveat as Find::Bin applies: This only works if chdir hasn't been changed.
If I understand the question correctly, a.t is in the directory dir/new_dir/ and you're using new_dir/a.t to run it from dir/, right?
If so, then it is doing the right thing. Since a.t is in dir/new_dir, you should always get dir/new_dir from FindBin - its job is to Find the Binary (program), not to find the file it's being called from, so the result will be the same in A.pm as it is in a.t.
The ->again function is for running instances of completely different programs from within the same perl interpreter, such as what mod_perl does, not for just using different modules within a single program.

Perl application installation

I have written a small perl application for deployment on a few servers. It consists of some scripts, some modules and some data files. It will be used by multiple users. I'd like to keep all these files together in one directory rather than moving the app modules to the site_perl directory.
For example, lets say the application consists of jump.pl swim.pl Common.pm and messages.txt
The application (either of jump.pl or swim.pl) will be invoked from another directory. The installation location of the application may not be on $PATH so the user (or another app) might invoke it by an absolute pathname such as /some/path/swim.pl or a relative pathname such as ../path/swim.pl
I'd like relocating the application to be as simple as possible, this means having the path to the application in as few places as possible.
At the moment swim.pl and jump.pl start like this
#!/usr/bin/perl
use strict;
use warnings;
use lib '/some/path';
use Common;
#
my $basedir = '/some/path';
...
open my $fh, '<', "$basedir/messages.txt" or die ... ;
So I have /some/path in several locations in several scripts and modules.
I've considered the following as possible means to minimise the number of different places to store /some/path
Set environment variable PERL5LIB=/some/path and do
open my $fh, '<', "$ENV{PERL5LIB}/messages.txt"
Set the shebang line to #!/usr/bin/perl -I/some/path - but then how find messages.txt?
Try use Cwd qw(abs_path);
and find a way to use basename abs_path($0) for use lib in BEGIN?
Any suggestions to maximise flexibility of location, keep everything together, minimise multiple hard-coding of installation location?
FindBin:
use FindBin qw( $Bin );
use File::Spec::Functions qw( catfile updir );
use lib catfile( updir($Bin), 'lib');

How do I load libraries relative to the script location in Perl?

How can you get current script directory in Perl?
This has to work even if the script is imported from another script (require).
This is not the current directory
Example:
#/aaa/foo.pl
require "../bbb/foo.pl"
#/bbb/bar.pl
# I want to obtain my directory (`/bbb/`)
print($mydir)
The script foo.pl could be executed in any ways and from any directory, like perl /aaa/foo.pl, or ./foo.pl.
What people usually do is
use FindBin '$Bin';
and then use $Bin as the base-directory of the running script. However, this won't work if you do things like
do '/some/other/file.pl';
and then expect $Bin to contain /some/other/ within that. I'm sure someone thought of something incredibly clever to work this around and you'll find it on CPAN somewhere, but a better approach might be to not include a program within a program, but to use Perl's wonderful ways of code-reuse that are much nicer than do and similar constructs. Modules, for example.
Those generally shouldn't care about what directory they were loaded from. If they really need to operate on some path, you can just pass that path to them.
See Dir::Self CPAN module. This adds pseudo-constant __DIR__ to compliment __FILE__ & __LINE__.
use Dir::Self;
use lib __DIR__ . '/lib';
I use this snippet very often:
use Cwd qw(realpath);
use File::Basename;
my $cwd = dirname(realpath($0));
This will give you the real path to the directory containing the currently running script. "real path" means all symlinks, "." and ".." resolved.
Sorry for the other 4 responses but none of them worked, here is a solution that really works.
In below example that adds the lib directory to include path the $dirname will contain the path to the current script. This will work even if this script is included using require from another directory.
BEGIN {
use File::Spec;
use File::Basename;
$dirname = dirname(File::Spec->rel2abs( __FILE__ )) . "/lib/";
}
use lib $dirname;
From perlfaq8's answer to How do I add the directory my program lives in to the module/library search path?
(contributed by brian d foy)
If you know the directory already, you can add it to #INC as you would for any other directory. You might if you know the directory at compile time:
use lib $directory;
The trick in this task is to find the directory. Before your script does anything else (such as a chdir), you can get the current working directory with the Cwd module, which comes with Perl:
BEGIN {
use Cwd;
our $directory = cwd;
}
use lib $directory;
You can do a similar thing with the value of $0, which holds the script name. That might hold a relative path, but rel2abs can turn it into an absolute path. Once you have the
BEGIN {
use File::Spec::Functions qw(rel2abs);
use File::Basename qw(dirname);
my $path = rel2abs( $0 );
our $directory = dirname( $path );
}
use lib $directory;
The FindBin module, which comes with Perl, might work. It finds the directory of the currently running script and puts it in $Bin, which you can then use to construct the right library path:
use FindBin qw($Bin);
You can also use local::lib to do much of the same thing. Install modules using local::lib's settings then use the module in your program:
use local::lib; # sets up a local lib at ~/perl5
See the local::lib documentation for more details.
Let's say you're looking for script.pl. You may be running it, or you may have included it. You don't know. So it either lies in the %INC table in the first case or as $PROGRAM_NAME (aka $0) in the second.
use strict;
use warnings;
use English qw<$PROGRAM_NAME>;
use File::Basename qw<dirname>;
use File::Spec;
use List::Util qw<first>;
# Here we get the first entry that ends with 'script.pl'
my $key = first { defined && m/\bscript\.pl$/ } keys %INC, $PROGRAM_NAME;
die "Could not find script.pl!" unless $key;
# Here we get the absolute path of the indicated path.
print File::Spec->rel2abs( dirname( $INC{ $key } || $key )), "\n";
Link to File::Basename, File::Spec, and List::Util