Macs formerly used CR (\x0D) as the line separator. Perl for the older Macs responded by redefining \n as CR (\x0D) and leaving the input record separator ($/) as \n.

Newer Macs use LF (\x0A) as the line separator like other unix platforms. Perl builds for modern Macs should leave \n as LF (\x0A).

I believe your Word is using the old standard (using CR as the input record separator) and your Perl is expecting LF. TextEdit silently(?) converts the line endings. You can do so programmatically using the following tool:

#!/usr/bin/perl

#
# Converts LF, CR and CRLF line endings to the local line ending.
#
# Usage:
#    fix_line_endings infile > outfile
#    fix_line_endings < infile > outfile
#    perl -i.bak fix_line_endings file   (In place)
#

use strict;
use warnings;

my $file = do { local $/; <> };
$file =~ s/\x0A|\x0D\x0A?/\n/g;
print $file;
[download]

Update: Simplified code.

\n is never redefined, io layer translates

Sorry, that's not true. It's the case for Windows, but it was NOT the case for MacPerl. The string literals "\x0A" and "\n" produced different strings (chr(10) and chr(13)) respectively). That's why it was best practice to use "\x0D\x0A" for internet protocol modules instead of "\r\n". The latter wasn't portable.

Update: Elaborated.

-> od -c word.txt 
0000000    1  \r   2  \r   3  \r   4  \r   5  \r
0000012
-> od -c textedit.txt
0000000    1  \n   2  \n   3  \n   4  \n   5  \n
0000012
-> od -c emacs.txt
0000000    1  \n   2  \n   3  \n   4  \n   5  \n
0000012
-> od -c nano.txt 
0000000    1  \n   2  \n   3  \n   4  \n   5  \n
0000012
[download]

Use ikegami's code from above reply to convert existing old-style files and use one of the non-Office text editors for editing plain text (choose the one you are most comfortable with; other options available on Mac OS X include vi or vim).

Phil

Word for Mac seems to use a line ending character not recognized by perl. Unix/Linux and probably MacOS uses \n (LF, ascii 10) for line endings, Windows uses \r\n (CR/LF, ascii 13,10). Your Word for Mac seems to use something without any \n at all, maybe a relict from old MacOS9 ??

Anyway, if you really want to use Word for Mac and it doesn't have a switch to change its behaviour, you could change perls notion of what a line ending is by changing $/ before reading this file.

Just to be sure, you really did save as a text file and didn't only change the filename from bla.doc to bla.txt ?