NAME

File::CountLines - efficiently count the number of line breaks in a file.

SYNOPSIS

use File::CountLines qw(count_lines);
my $no_of_lines = count_lines('/etc/passwd');

# other uses
my $carriage_returns = count_lines( 
        'path/to/file.txt', 
        style   => 'cr',
    );
# possible styles are 'native' (the default), 'cr', 'lf'

DESCRIPTION

perlfaq5 answers the question on how to count the number of lines in a file. This module is a convenient wrapper around that method, with additional options.

More specifically, it counts the number of line breaks rather than lines. On Unix systems nearlly all text files end with a newline (by convention), so usually the number of lines and number of line breaks is equal.

Since different operating systems have different ideas of what a newline is, you can specifiy a style option, which can be one of the following values:

native

This takes Perl's \n as the line separator, which should be the right thing in most cases. See perlport for details. This is the default.

cr

Take a carriage return as line separator (MacOS style)

lf

Take a line feed as line separator (Unix style)

crlf

Take a carriage return followed by a line feed as separator (Microsoft Windows style)

Alternatively you can specify an arbitrary separator like this:

my $lists = count_lines($file, separator => '\end{itemize}');

It is taken verbatim and searched for in the file.

The file is read in equally sized blocks. The size of the blocks can be supplied with the blocksize option. The default is 4096, and can be changed by setting $File::CountLines::BlockSize.

Do not use a block size smaller than the length of the separator, that might produce wrong results. (In general there's no reason to chose a smaller block size at all. Depending on your size a larger block size might speed up things a bit.)

Character Encodings

If you supply a separator yourself, it should not be a decoded string.

The file is read in binary mode, which implies that this module works fine for text files in ASCII-compatible encodings, including ASCII itself, UTF-8 and all the ISO-8859-* encodings (aka Latin-1, Latin-2, ...).

Note that the multi byte encodings like UTF-32, UTF-16le, UTF-16be and UCS-2 encode a line feed character in a way that the 0x0A byte is a substring of the encoded character, but if you search blindly for that byte you will get false positives. For example the LATIN CAPITAL LETTER C WITH DOT ABOVE, U+010A has the byte sequence 0x0A 0x01 when encoded as UTF-16le, so it would be counted as a newline. Even search for 0x0A 0x00 might give false positives.

So the summary is that for now you can't use this module in a meaningful way to count lines of text files in encodings that are not ASCII-compatible. If there's demand for, I can implement that though.

Extending

You can add your own EOL styles by adding them to the %File::CountLines::StyleMap hash, with the name of the style as hash key and the separator as the value.

AUTHOR

Moritz Lenz http://perlgeek.de, mailto:moritz@faui2k3.org

COPYRIGHT AND LICENSE

Copyright (C) 2008 by Moritz A. Lenz. This module is free software. You may use, redistribute and modify it under the same terms as perl itself.

Example code included in this package may be used as if it were in the Public Domain.

DEVELOPMENT

You can obtain the latest development version from http://github.com/moritz/File-CountLines:

git clone git://github.com/moritz/File-CountLines.git