Merge txt files in Perl, but modify them before, leaving original files untouched - perl

Question

I've already posted a question and fixed the problem in my code, but now my "specification has changed" so to say, and now I need to change some things about it.
Here's a code that takes all .txt files from the current directory, cuts off the last line of the first file, the first and the last line of every following file and the first line of the last file and writes everything in a new file (in other words: merge all files, deleting header and footer so that the new file has only one header and one footer).
#!/usr/bin/perl
use warnings;
use Cwd;
use Tie::File;
use Tie::Array;
my $cwd = getcwd();
my $buff = '';
# Get all files in cwd.
my #files = grep ( -f ,<*.txt>);
# Cut off header and footer of $files [1] to $files[$#files-1],
# but only footer of $files[0] and header of $#files[$#files]
for (my $i = 0; $i <= $#files; $i++) {
print 'Opening ' . $files[$i] . "\n";
tie (#lines, Tie::File, $files[$i]) or die "can't update $file: $!";
splice #lines, 0, 1 unless $i == 0;
splice #lines, -1, 1 unless $i == $#files;
untie #lines;
open (file, "<", $files[$i]) or die "can't update $file: $!";
while (my $line =<file>) {
$buff .= $line;
}
close file;
}
# Write the buffer to a new file.
my $allfilename = $cwd.'/Trace.txt';
print 'Writing all files into new file: ' . $allfilename . "\n";
open $outputfile, ">".$allfilename or die "can't write to new file $outputfile: $!";
# Write the buffer into the output file.
print $outputfile $buff;
close $outputfile;
My problem: I don't want to change the original files, but my code does exactly that and I'm having trouble coming up with a solution. The simplest way (simple meaning not having to change too much code) would now be, to just copy all the files to a tmp directory, messing around with them and leaving the original files untouched. Problem: a simple use of dircopy doesn't do it for me, since you have to give a new tmp dir to the dircopy function, making the code only usable for Windows or UNIX systems (but I need portability).
The next approach would be to make use of the File::Temp module but I'm really having trouble with the docs on this one.
Does anybody have a good idea on this one?

Answer 1

I suspected that you didn't really want your original files modified when I answered your previous question.
I don't understand why you've gone back to accumulating all the text in a buffer before printing it, or why you've removed use strict, which is essential to any well-written Perl code.
Here's my previous solution modified to leave the input data untouched.
use strict;
use warnings;
use Tie::File;
my #files = grep -f, glob '*.txt';
my $all_filename = 'Trace.txt';
open my $out_fh, '>', $all_filename or die qq{Unable to open "$all_filename" for output: $!};
for my $i ( 0 .. $#files ) {
my $file = $files[$i];
next if $file eq $all_filename;
print "Opening $file\n";
tie my #lines, 'Tie::File', $file or die qq{Can't open "$file": $!};
my ($start, $end) = (0, $#lines);
++$start unless $i == 0;
--$end unless $i == $#files;
print $out_fh "$_\n" for #lines[$start..$end];
}
close $out_fh;

Answer 2

#!/usr/bin/env perl
use strict;
use warnings;
use autodie;
my $outfile = 'Trace.txt';
# Get all files in cwd.
my #files = grep { -f && $_ ne $outfile } <*.txt>;
open my $outfh, '>', $outfile;
for my $file (#files) {
my #lines = do { local #ARGV = $file; <> };
shift #lines unless $file eq $files[0];
pop #lines unless $file eq $files[-1];
print $outfh #lines;
}

Answer 3

Just do not use Tie::File. Or is there a reason you do this, for example all your files together do not fit your memory or something?
A version very close to your current implementation would be something like the following (untested) code. It just skips the part where you update the file, just to reopen and read it afterwards. (Note that this is certainly not a very effective or overly elegant way to do this, it just sticks to your implementation as close as possible)
#!/usr/bin/perl
use warnings;
use Cwd;
# use Tie::File;
# use Tie::Array;
my $cwd = getcwd();
my $buff = '';
# Get all files in cwd.
my #files = grep ( -f ,<*.txt>);
# Cut off header and footer of $files [1] to $files[$#files-1],
# but only footer of $files[0] and header of $#files[$#files]
for (my $i = 0; $i <= $#files; $i++) {
print 'Opening ' . $files[$i] . "\n";
open (my $fh, "<", $files[$i]) or die "can't open $file for reading: $!";
my #lines = <$fh>;
splice #lines, 0, 1 unless $i == 0;
splice #lines, -1, 1 unless $i == $#files;
foreach my $line (#lines) {
$buff .= $line;
}
}
# Write the buffer to a new file.
my $allfilename = $cwd.'/Trace.txt';
print 'Writing all files into new file: ' . $allfilename . "\n";
open $outputfile, ">".$allfilename or die "can't write to new file $outputfile: $!";
# Write the buffer into the output file.
print $outputfile $buff;
close $outputfile;

Answer 4

Based on Miller's answer, but most suitable for large files.
#!/usr/bin/env perl
use strict;
use warnings;
use autodie;
my $outfile = 'Trace.txt';
# Get all files in cwd.
my #files = grep { -f && $_ ne $outfile } <*.txt>;
open my $outfh, '>', $outfile;
my $counter = 0;
for my $file (#files) {
open my $fh, '<', $file;
my ($line, $prev) = ('', '');
my $l = 0;
while ($line = <$fh>) {
print $outfh $prev unless $l++ == 1 and $counter > 0;
$prev = $line;
}
$counter++;
print $outfh $prev if $counter == #files and $l > 0;
close $fh;
}