view sort-header @ 0:5314e5d6f040 draft

Imported from capsule None

#!/usr/bin/env perl
##
## Sort-header - wrapper for GNU sort with header-line support
##
## Copyright(C) A. Gordon
## license AGPLv3+
##
use strict;
use warnings;
use Data::Dumper;
use IO::Handle;
use Getopt::Long qw(:config bundling no_ignore_case_always);

## Forward declarations
sub add_standard_sort_param(@);
sub add_standard_sort_param_value(@);
sub forbidden_sort_param(@);
sub show_help();
sub show_version();
sub show_examples();
sub parse_commandline_options();
sub reassign_input_output();
sub process_header_lines();
sub run_sort();
sub read_line_non_buffered();


##
## Runtime options
##
my $PROGRAM="sort-header";
my $VERSION=0.4;

my $check_only=undef;
my $input_file=undef;
my $output_file=undef;
my $field_separator=undef;
my $header_lines =1 ;
my $debug=undef;
my $sort_exit_code=1; #by default, assume some error

my @sort_options;

##
## Program Start
##
parse_commandline_options();
reassign_input_output();
process_header_lines();
run_sort();
exit($sort_exit_code);
##
## Program End
##

sub show_examples()
{
print<<EOF;
Sorting a file with a header line:

\$ cat input.txt
Fruit	Color	Price
Banana	Yellow	4.1
Avocado	Green	8.0
Apple	Red	3.0
Melon	Green	6.1

# By default, 'sort-header' assumes 1 header line
# (no need to use --header in this case).

\$ sort-header -k3,3nr input.txt
Fruit	Color	Price
Avocado	Green	8.0
Melon	Green	6.1
Banana	Yellow	4.1
Apple	Red	3.0

EOF
	exit(0);
}

sub show_help()
{
print<<EOF;
${PROGRAM}: Wrapper for GNU sort, allowing sorting files with header lines.

Usage: $PROGRAM [HEADER-OPTIONS] [GNU sort Options] [INPUT-FILE]

HEADER-OPTIONS: the following options are supported by '${PROGRAM}':

   --header N    =  Treat the first N lines as header lines.
                    These line will NOT be sorted. They will be passed
		    directly to the output file. (default: 1)

   --version     =  Print ${PROGRAM}'s version.

   --debugheader =  Print debug messages (relating to ${PROGRAM}'s operation).

   --help        =  Show this help screen.

   --examples    =  Show usage examples.

GNU sort options:
   Most of the standard GNU sort options are supported and passed to GNU sort.
   The following options can not be used with '${PROGRAM}':

   -m --merge           => ${PROGRAM} can only sort one file, not merge multiple files.
   -c -C --check        => Currently not supported
   --files0-from        => Currently not supported
   -z --zero-terminated => Currently not supported

INPUT-FILE:
   If INPUT-FILE is not specified, $PROGRAM will use STDIN (just like GNU sort).

EOF
	exit(0);
}

sub show_version()
{
print<<EOF;
$PROGRAM $VERSION
Copyright (C) 2010 A. Gordon (gordon\@cshl.edu)
License AGPLv3+: Affero GPL version 3 or later (http://www.gnu.org/licenses/agpl.html)

To see the GNU's sort version, run:
	sort --version
EOF
	exit(0);
}

sub parse_commandline_options()
{
	my $rc = GetOptions(
		"ignore-leading-blanks|b" => \&add_standard_sort_param,
		"dictionary-order|d" => \&add_standard_sort_param,
		"ignore-case|f" => \&add_standard_sort_param,
		"general-numeric-sort|g" => \&add_standard_sort_param,
		"ignore-nonprinting|i" => \&add_standard_sort_param,
		"month-sort|M" => \&add_standard_sort_param,
		"human-numeric-sort|h" => \&add_standard_sort_param,
		"numeric-sort|n" => \&add_standard_sort_param,
		"random-source=s" => \&add_standard_sort_param_value,
		"random-sort|R" => \&add_standard_sort_param,
		"reverse|r" => \&add_standard_sort_param,
		"sort=s" => \&add_standard_sort_param_value,
		"version-sort|V" => \&add_standard_sort_param,

		"check|c" => \&forbidden_sort_param,
		"C" => \&forbidden_sort_param,
		"compress-program=s" => \&add_standard_sort_param_value,
		"debug" => \&add_standard_sort_param,

		"files0-from=s" => \&forbidden_sort_param,

		"key|k=s" => \&add_standard_sort_param_value,
		"merge|m" => \&forbidden_sort_param,
		"batch-size=i" => \&forbidden_sort_param,

		"parallel=i" => \&add_standard_sort_param_value,

		"output|o=s" => \$output_file,

		"stable|s" => \&add_standard_sort_param,
		"buffer-size|S=s" => \&add_standard_sort_param_value,

		"field-separator|t=s" => \&add_standard_sort_param_value,
		"temporary-directory|T=s" => \&add_standard_sort_param_value,
		"unique|u" => \&add_standard_sort_param,

		"zero-terminated|z" => \&forbidden_sort_param,

		"help" => \&show_help,
		"version" => \&show_version,
		"examples" => \&show_examples,

		"header=i" => \$header_lines,
		"debugheader" => \$debug,
		);

	exit 1 unless $rc;

	my @INPUT_FILES = @ARGV;

	die "$PROGRAM: error: invalid number of header lines ($header_lines)\n" unless $header_lines>=0;
	die "$PROGRAM: error: Multiple input files specified. This program can sort only a signle file.\n" if (scalar(@INPUT_FILES)>1);
	$input_file = shift @INPUT_FILES if scalar(@INPUT_FILES)==1;

	if ($debug) {
		warn "$PROGRAM: number of header lines = $header_lines\n";
		warn "$PROGRAM: PASS-to-Sort options:\n", Dumper(\@sort_options), "\n";
	}
}

sub reassign_input_output()
{
	if ($output_file) {
		warn "$PROGRAM: Re-assigning STDOUT to '$output_file'\n" if $debug;
		open OUTPUT, '>', $output_file or die "$PROGRAM: Error: failed to create output file '$output_file': $!\n";
		STDOUT->fdopen(\*OUTPUT, 'w') or die "$PROGRAM: Error: failed to reassign STDOUT to '$output_file': $!\n";
	}


	if ($input_file) {
		warn "$PROGRAM: Re-assigning STDIN to '$input_file'\n" if $debug;
		open INPUT, '<', $input_file or die "$PROGRAM: Error: failed to open input file '$input_file': $!\n";
		STDIN->fdopen(\*INPUT, 'r') or die "$PROGRAM: Error: failed to reassign STDIN to '$input_file': $!\n";
	}
}

sub process_header_lines()
{
	warn "$PROGRAM: Reading $header_lines header lines...\n" if $debug;
	for (my $i=0; $i<$header_lines; $i++) {
		my $line = read_line_non_buffered();
		exit unless defined $line;
		print $line;
	}
}

sub run_sort()
{
	warn "$PROGRAM: Running GNU sort...\n" if $debug;
	system('sort', @sort_options);
	if ($? == -1) {
		die "$PROGRAM: Error: failed to execute 'sort': $!\n";
	}
	elsif ($? & 127) {
		my $signal = ($? & 127);
		kill 2, $$ if $signal == 2; ##if sort was interrupted (CTRL-C) - just pass it on and commit suicide
		die "$PROGRAM: Error: 'sort' child-process died with signal $signal\n";
	}
	else {
		$sort_exit_code = ($? >> 8);
	}
}


sub add_standard_sort_param(@)
{
	my ($obj)=  @_;
	add_standard_sort_param_value($obj, undef);
}

sub add_standard_sort_param_value(@)
{
	my ($obj,$value)=  @_;

	my $option = "" . $obj ; #stringify the optino object, get the option name.

	if (length($option)==1) {
		$option = "-" . $option ;
	} else {
		$option = "--" . $option ;
	}
	push @sort_options, $option ;
	push @sort_options, $value if $value;
}

sub forbidden_sort_param(@)
{
	my ($obj,$value)=  @_;
	my $option = "" . $obj ; #stringify the optino object, get the option name.

	die "$PROGRAM: Error: option '$option' can not be used with this program. If you must use it, run GNU sort directly. see --help for more details.\n";
}

sub read_line_non_buffered()
{
	my $line = '';
	while ( 1 ) {
		my $c;
		my $rc = sysread STDIN, $c, 1;
		die "$PROGRAM: STDIN Read error: $!" unless defined $rc;
		return $line if $rc==0 && $line;
		return undef if $rc==0 && (!$line);
		$line .= $c ;
		return $line if ( $c eq "\n");
	}
}