How to Detect Mixed File Line-Endings

paul-s- has asked for the wisdom of the Perl Monks concerning the following question:

Greetings Brothers, I use a code repository that works on UNIX, so that the code may be accessed in Windows. We want Windows line-endings, most of the files are, but some have developed UNIX line-endings (which we don't want). We don't want to change all of the files - only the ones that are incorrect. So, I want a simple algorithm to run on Windows, that will read the first line of a file and tell me what the line-ending is. The following code seems to report UNIX line-endings in all cases - even when they are Windows. Please help.

my $cr = chr(0x0d);
my $lf = chr(0x0a);
my $infile = "test.txt";
open (IN, "<$infile") or die "Couldn't open input file: $!";
my $first_line = <IN>;
my ($line_text, $line_ending) = ($first_line =~ /(.*)(\015|\012|\015\0
+12)/);
# detect line ending type
if ($line_ending eq "$cr$lf") {
  print "Windows line ending\n";
} elsif ($line_ending eq "$lf") {
  print "UNIX line ending\n";
} elsif ($line_ending eq "$cr") {
  print "Mac line ending\n";
} else {
  print "Unknown line ending\n";
  }
close (IN);
[download]

Comment on How to Detect Mixed File Line-Endings Download Code

Replies are listed 'Best First'.
Re: How to Detect Mixed File Line-Endings by Corion (Patriarch) on Jan 17, 2014 at 12:28 UTC
Your regular expression does not check whether the `\015` or the `\012` or the `\015\012` is really the last thing on the line. This can be fixed by by anchoring your regular expression to the end of the string: `/(\015\|\012\|\015\012)\z/` [download]	[reply] [d/l] [select]
Re^2: How to Detect Mixed File Line-Endings by robby_dobby (Hermit) on Jan 18, 2014 at 13:15 UTC
In the interest of a longer match first, I think OP should change the regex to test for CRLF first followed by either of CR or LF. `m!\015\012\|\012\|\015!`	[reply] [d/l]
Re: How to Detect Mixed File Line-Endings by kcott (Archbishop) on Jan 17, 2014 at 20:49 UTC
G'day paul-s-, Welcome to the monastery. This solution, which requires Perl 5.10.0, will skip your Windows files and allow you to process the Unix files. Given these files: `$ cat -vet pm_1070943_CRLF.txt text^M$ $ cat -vet pm_1070943_LF.txt text$` [download] [Note: "`cat -vet`" shows special characters in files: '`^M`' = carriage return; '`$`' = newline] This script: `#!/usr/bin/env perl use 5.010; use strict; use warnings; use autodie; for my $file (qw{pm_1070943_CRLF.txt pm_1070943_LF.txt}) { open my $fh, '<', $file; my $first_line = <$fh>; $first_line =~ /(\R)\z/; if ($1 eq "\x0D\x0A") { say "Skipping: $file"; } else { say "Processing: $file"; } }` [download] [Note: '`\R`' is described in "perlrebackslash: Misc"] Produces this output: `Skipping: pm_1070943_CRLF.txt Processing: pm_1070943_LF.txt` [download] -- Ken	[reply] [d/l] [select]
Re: How to Detect Mixed File Line-Endings by Anonymous Monk on Jan 17, 2014 at 13:34 UTC
You need to amend your regular expression with a negative look-behind in the Unix case, so that it is not accidentally matched. Or just not use a regular expression at all. (substr is a better hammer for this kind of nail.)	[reply]