ported write_fasta to ruby enabling bzip2 support

git-svn-id: http://biopieces.googlecode.com/svn/trunk@2210 74ccb610-7750-0410-82ae-013aeee3265d

diff --git a/bp_bin/write_fasta b/bp_bin/write_fasta
index 52f8a7011089f04b3052aaad2bc0d08b53c85ac6..f44d6f4fb64d2188b83cda14d2da7e29e68c2cbe 100755 (executable)
--- a/bp_bin/write_fasta
+++ b/bp_bin/write_fasta
@@ -1,6 +1,6 @@
-#!/usr/bin/env perl
+#!/usr/bin/env ruby
 
-# Copyright (C) 2007-2009 Martin A. Hansen.
+# Copyright (C) 2007-2012 Martin A. Hansen.
 
 # This program is free software; you can redistribute it and/or
 # modify it under the terms of the GNU General Public License
@@ -18,67 +18,46 @@
 
 # http://www.gnu.org/copyleft/gpl.html
 
-
-# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
-
-# Write sequences from stream in FASTA format.
-
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
+# This program is part of the Biopieces framework (www.biopieces.org).
 
-use warnings;
-use strict;
-use Maasha::Fasta;
-use Maasha::Biopieces;
+# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
+# Write sequences from stream in FASTA format.
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
+require 'maasha/biopieces'
+require 'maasha/fasta'
 
-my ( $options, $in, $out, $record, $data_out, $entry );
+casts = []
+casts << {long: 'no_stream', short: 'x', type: 'flag',   mandatory: false, default: nil, allowed: nil,               disallowed: nil}
+casts << {long: 'data_out',  short: 'o', type: 'file',   mandatory: false, default: nil, allowed: nil,               disallowed: nil}
+casts << {long: 'wrap',      short: 'w', type: 'uint',   mandatory: false, default: nil, allowed: nil,               disallowed: "0"}
+casts << {long: 'compress',  short: 'Z', type: 'string', mandatory: false, default: nil, allowed: "gzip,bzip,bzip2", disallowed: nil}
 
-$options = Maasha::Biopieces::parse_options(
-    [
-        { long => 'no_stream', short => 'x', type => 'flag', mandatory => 'no', default => undef, allowed => undef, disallowed => undef },
-        { long => 'data_out',  short => 'o', type => 'file', mandatory => 'no', default => undef, allowed => undef, disallowed => undef },
-        { long => 'wrap',      short => 'w', type => 'uint', mandatory => 'no', default => undef, allowed => undef, disallowed => '0' },
-        { long => 'compress',  short => 'Z', type => 'flag', mandatory => 'no', default => undef, allowed => undef, disallowed => undef },
-    ]   
-);
+options = Biopieces.options_parse(ARGV, casts)
 
-$in  = Maasha::Biopieces::read_stream( $options->{ "stream_in" } );
-$out = Maasha::Biopieces::write_stream( $options->{ "stream_out" } );
+compress = options[:compress] ? options[:compress].to_sym : nil
 
-$data_out = Maasha::Biopieces::write_stream( $options->{ "data_out" }, $options->{ "compress" } );
+raise "--data_out is mandatory for compressed output" if compress and not options[:data_out]
 
-while ( $record = Maasha::Biopieces::get_record( $in ) ) 
-{
-    if ( $entry = Maasha::Fasta::biopiece2fasta( $record ) ) {
-        Maasha::Fasta::put_entry( $entry, $data_out, $options->{ "wrap" } );
-    }
-
-    Maasha::Biopieces::put_record( $record, $out ) if not $options->{ "no_stream" };
-}
-
-close $data_out;
-
-Maasha::Biopieces::close_stream( $in );
-Maasha::Biopieces::close_stream( $out );
-
-
-# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
+Biopieces.open(options[:stream_in], options[:stream_out]) do |input, output|
+  fasta_out = options[:data_out] ? Fasta.open(options[:data_out], 'w', compress: compress) : STDOUT
 
+  input.each do |record|
+    if record[:SEQ_NAME] and record[:SEQ]
+      entry = Seq.new_bp(record)
 
-BEGIN
-{
-    Maasha::Biopieces::status_set();
-}
+      fasta_out.puts entry.to_fasta(options[:wrap])
+    end
 
+    output.puts record unless options[:no_stream]
+  end
 
-END
-{
-    Maasha::Biopieces::status_log();
-}
+  fasta_out.close
+end
 
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<

diff --git a/bp_test/test/test_write_fasta b/bp_test/test/test_write_fasta
index 68f37ce7e5ec796a6a7009a4b9f1a5c01513fd7b..4307e2590cd68261d3c82b0919420bb23fae9b67 100755 (executable)
--- a/bp_test/test/test_write_fasta
+++ b/bp_test/test/test_write_fasta
@@ -10,7 +10,12 @@ run "$bp -I $in -w 4 -o $tmp -x"
 assert_no_diff $tmp $out.2
 clean
 
-run "$bp -I $in -w 4 -Z -o $tmp.gz -x"
+run "$bp -I $in -w 4 -Z gzip -o $tmp.gz -x"
 gunzip $tmp.gz
 assert_no_diff $tmp $out.3
 clean
+
+run "$bp -I $in -w 4 -Z bzip2 -o $tmp.bz2 -x"
+bunzip2 $tmp.bz2
+assert_no_diff $tmp $out.3
+clean