Perl cannot stat $_ - perl

Question

I want the last modified time for each file in the directory. To make sure my loop is working I print $_ and I see the file names of the directory:
for ( #Files ) {
opendir( D, $path . '\/' . $_ ) or die "$!";
my #textfiles = grep { ! /^\.{1,2}$/ } readdir( D );
for ( #textfiles ) {
# print "$_\n"; <----the file names.
my $epoch_timestamp = ( stat( $_ ) )[9];
print "$epoch_timestamp\n";
}
I get this error
Use of uninitialized value $epoch_timestamp in concatenation (.) or string
What am I doing wrong?

Answer 1

readdir returns only the names of the files. If your current working directory is different then you must build the full path as you did with the parameter to opendir. The easiest way is to use map in the list for the for loop
I'm concerned about your statement
opendir( D, $path . '\/' . $_ ) or die "$!";
which will put, literally, \/ between $path and $_. I think you need just /, but it is simplest to interpolate the variables with
opendir( D, "$path/$_" ) or die "$!";
But $_ comes from the array #Files. If these are indeed file names then your opendir will fail. They need to be directory names
In my solution I've built the variable $dir as
my $dir = "$path/$_"
so that it can be used in the call to opendir as well as to build the full path to the files in the following for loop
Note that I have also used a lexical directory handle my $dh, which are far superior to global handles D
for ( #Files ) {
my $dir = "$path/$_";
opendir my $dh, $dir or die $!;
my #textfiles = grep { ! /^\.{1,2}$/ } readdir $dh;
for ( map { "$dir/$_" } #textfiles ) {
# print "$_\n"; <----the file names.
my $epoch_timestamp = ( stat( $_ ) )[9];
print "$epoch_timestamp\n";
}

Answer 2

Or alternatively to above perfect answers, you could use some modules and make your life more easy. :) Like: Path::Tiny[1]
use 5.014;
use warnings;
use Path::Tiny;
my $path = path('/etc');
my #Files = qw(defaults cups ssl);
for my $dir (#Files) {
my #textfiles = $path->child($dir)->children;
for my $file (#textfiles) {
say "$file: ", $file->stat->mtime;
}
}
Of course, the above the nested loop could be written as
for my $dir (#Files) {
my #textfiles = $path->child($dir)->children;
say "$_: ", $_->stat->mtime for (#textfiles);
}
and also storing the list of files into #textfiles isn't necessary, so it could be reduced to:
for my $dir (#Files) {
say "$_: ", $_->stat->mtime for ( $path->child($dir)->children );
}
Path::Tiny conveniently throws a clean exception message on error.

Answer 3

readdir only returns the name of the file in the directory. You need to provide a qualified path to the file to stat.
my $dir_qfn = ...;
opendir(my $dh, $dir_qfn)
or do {
warn("Can't read dir \"$dir_qfn\": $!\n");
next;
};
while (defined( my $fn = readdir($dh) )) {
next if $fn =~ /^\.\.?\z/;
my $qfn = "$dir_qfn/$fn";
my $mtime = ( stat($qfn) )[9];
defined($mtime)
or do {
warn("Can't stat file \"$file_qfn\": $!\n");
next;
};
...
}

Answer 4

Using glob instead
my $dir = ...;
my %ts =
map { $_ => (stat $_)[9] }
grep { !m{/\.\.?\z} } #/
glob "\Q$dir\E/{*,.*}";
say "ts{$_} => $_" for sort keys %ts;
I use a hash name => timestamp to collect both in a data structure. The pattern $dir/{*,.*} is there to catch dot files as well, or it would be just $dir/*.
The grep filters out . and .. filenames, found in path by m{..} match. Its pattern needs \Q..\E to prevent an injection bug with particular directory names. It also escapes spaces so File::Glob with its :bsd_globoption isn't needed. Thanks to ikegami for comments.
If you'd rather process files one at a time, retrieve the list with glob and then iterate through it.