File Coverage

blib/lib/EOL.pm

Criterion	Covered	Total	%
statement	56	65	86.1
branch	25	46	54.3
condition	12	15	80.0
subroutine	7	7	100.0
pod	3	3	100.0
total	103	136	75.7

package EOL;

# | PACKAGE | EOL
# | AUTHOR  | Todd Wylie
# | EMAIL   | [e-mail]
# | ID      | $Id: EOL.pm 238 2006-10-31 19:08:15Z twylie $

# ---------------------------------------------------------------------------
# PURPOSE:
# This module aids in the conversion of text file newline characters within
# the context of other perl code. There are several choices a user can
# specifiy; see POD for details on usage.
# ---------------------------------------------------------------------------

use version; $VERSION = qv('0.0.2');
use warnings;
use strict;
use Carp;

# END OF LINE DECLARATIONS
my $CR   = "\015";      # Apple II family, Mac OS thru version 9
my $CRLF = "\015\012";  # CP/M, MP/M, DOS, Microsoft Windows
my $LF   = "\012";      # Unix, Linux, Xenix, Mac OS X, BeOS, Amiga
my $FF   = "\014";      # printer form feed

# --------------------------------------------------------------------------
# E O L  N E W  F I L E  (subroutine)
# ==========================================================================
# USAGE      : eol_new_file( in => $in_fh, out => $out_fh, eol => 'LF' );
# PURPOSE    : Incoming file's newlines converted & saved to new file.
# RETURNS    : none
# PARAMETERS : in  => '' # file handle
#            : out => '' # file handle
#            : eol => '' # [CR; CRLF; LF]
# THROWS     : Croaks if required arguments missing/null or unknown.
# COMMENTS   : By default, all newlines will be converted to standard
#            : Unix-style line feeds (LF or octal \012). A user may change
#            : the end-of-file tag manually by supplying the "eol" argument.
#            : Acceptable eol tags are: CR; CRLF; LF.
# SEE ALSO   : n/a
# --------------------------------------------------------------------------
sub eol_new_file {
    my %args = @_;
    my $EOL;

    # Check arguments:
    foreach my $arg (keys %args) {
        if ( ($arg ne "in") && ($arg ne "out") && ($arg ne "eol") ) {
            croak "EOL reports: not legal arg \"$arg\".";
        }
    }

    # Check for text file input:
    if (!-T $args{in}) {
        croak "EOL reports: $args{in} is not a text file.";
    }

    # Conversion steps:
    open(IN, "$args{in}") or croak "EOL reports: cannot open $args{in}.";
    if (-f $args{out}) { unlink($args{out}) }
    open(OUT, ">$args{out}") or croak "EOL reports: cannot open $args{in}.";
    if ($args{eol}) {
        ($args{eol} eq "CR")     ? ($EOL = $CR  )
        : ($args{eol} eq "CRLF") ? ($EOL = $CRLF)
        : ($args{eol} eq "LF"  ) ? ($EOL = $LF  )
        : croak "EOL reports: not legal EOL tag! Use only: CR; CRLF; LF.";
    }
    else {
        $EOL = $LF;  # Default is Unix line feed.
    }
    while() {
        s/$CRLF$|$CR|$FF$|$LF$/$EOL/g;
        print OUT $_;
    }
    close(IN);
    close(OUT);

    return(1);
}

# --------------------------------------------------------------------------
# E O L  S A M E  F I L E  (subroutine)
# ==========================================================================
# USAGE      : eol_same_file( in => $in_fh, eol => 'LF', backup => '.bak' );
# PURPOSE    : Incoming file's newlines converted & overwrites self.
# RETURNS    : none
# PARAMETERS : in     => '' # file handle
#            : eol    => '' # [CR; CRLF; LF]
#            : backup => '' # scalar
# THROWS     : Croaks if required arguments missing/null or unknown.
# COMMENTS   : By default, all newlines will be converted to standard
#            : Unix-style line feeds (LF or octal \012). A user may change
#            : the end-of-file tag manually by supplying the "eol" argument.
#            : Acceptable eol tags are: CR; CRLF; LF. As already mentioned,
#            : this routine will clobber the in-file. Sending the "backup"
#            : argument with a file suffix will produce a one-time backup of
#            : the original file. !!! USE AT OWN RISK !!!
# SEE ALSO   : n/a
# --------------------------------------------------------------------------
sub eol_same_file {
    my %args = @_;
    my $EOL;

    # Check arguments:
    foreach my $arg (keys %args) {
        if ( ($arg ne "in") && ($arg ne "backup") && ($arg ne "eol") ) {
            croak "EOL reports: not legal arg \"$arg\".";
        }
    }

    # Check for text file input:
    if (!-T $args{in}) {
        croak "EOL reports: $args{in} is not a text file.";
    }

    # Conversion steps:
    if ($args{eol}) {
        ($args{eol} eq "CR")     ? ($EOL = $CR  )
        : ($args{eol} eq "CRLF") ? ($EOL = $CRLF)
        : ($args{eol} eq "LF"  ) ? ($EOL = $LF  )
        : croak "EOL reports: Not legal EOL tag! Use only: CR; CRLF; LF.";
    }
    else {
        $EOL = $LF;  # Default is Unix line feed.
    }
    (exists $args{backup}) ? ($^I = $args{backup}) : ($^I = "");
    while(<>) {
        s/$CRLF$|$CR|$FF$|$LF$/$EOL/g;
        print $_;
    }

    return(1);
}

# --------------------------------------------------------------------------
# E O L  R E T U R N  A R R A Y  (subroutine)
# ==========================================================================
# USAGE      : eol_return_array( in => $in_fh, eol => 'LF' );
# PURPOSE    : Incoming file's newlines converted & array of lines returned.
# RETURNS    : array-reference
# PARAMETERS : in     => '' # file handle
#            : eol    => '' # [CR; CRLF; LF]
# THROWS     : Croaks if required arguments missing/null or unknown.
# COMMENTS   : By default, all newlines will be converted to standard
#            : Unix-style line feeds (LF or octal \012). A user may change
#            : the end-of-file tag manually by supplying the "eol" argument.
#            : Acceptable eol tags are: CR; CRLF; LF. An array reference to
#            : the converted file lines will be returned. Obviously, for
#            : larger files, this may be a considerable drain on memory.
# SEE ALSO   : n/a
# --------------------------------------------------------------------------
sub eol_return_array {
    my %args = @_;
    my $EOL;

    # Check arguments:
    foreach my $arg (keys %args) {
        if ( ($arg ne "in") && ($arg ne "eol") ) {
            croak "EOL reports: not legal arg \"$arg\".";
        }
    }

    # Check for text file input:
    if (!-T $args{in}) {
        croak "EOL reports: $args{in} is not a text file.";
    }

    # Conversion steps:
    my @lines;
    open(IN, "$args{in}") or croak "EOL reports: cannot open $args{in}";
    if ($args{eol}) {
        ($args{eol} eq "CR")     ? ($EOL = $CR  )
        : ($args{eol} eq "CRLF") ? ($EOL = $CRLF)
        : ($args{eol} eq "LF"  ) ? ($EOL = $LF  )
        : croak "EOL reports: not legal EOL tag! Use only: CR; CRLF; LF.\n\n";
    }
    else {
        $EOL = $LF;  # Default is Unix line feed.
    }
    while() {
        my $line = $_;
        $line =~ s/$CRLF$|$CR|$FF$|$LF$/$EOL/g;
        push (@lines, split (/($EOL)/, $line));
    }
    close(IN);

    return(\@lines);
}

1; # End of module.

__END__