From: martinahansen <martinahansen@74ccb610-7750-0410-82ae-013aeee3265d>
Date: Tue, 15 Oct 2013 17:45:28 +0000 (+0000)
Subject: rewrote write_tab
X-Git-Url: https://git.donarmstrong.com/?p=biopieces.git;a=commitdiff_plain;h=041e3704a12bb721c4dd3ee3f28bf286c5654e33

rewrote write_tab

git-svn-id: http://biopieces.googlecode.com/svn/trunk@2231 74ccb610-7750-0410-82ae-013aeee3265d
---

diff --git a/bp_bin/write_tab b/bp_bin/write_tab
index b512d94..c852939 100755
--- a/bp_bin/write_tab
+++ b/bp_bin/write_tab
@@ -1,6 +1,6 @@
-#!/usr/bin/env perl
+#!/usr/bin/env ruby
 
-# Copyright (C) 2007-2009 Martin A. Hansen.
+# Copyright (C) 2007-2013 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
@@ -18,129 +18,123 @@
 
 # http://www.gnu.org/copyleft/gpl.html
 
-
-# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
-
-# Write tabular output from the stream.
-
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
+# This program is part of the Biopieces framework (www.biopieces.org).
 
-use warnings;
-use strict;
-use Maasha::Fasta;
-use Maasha::Biopieces;
+# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
+# Write tabular output from the stream.
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
-
-my ( $options, $in, $out, $record, $data_out, @vals, $ok, $key, @keys, $A, $B, %no_keys, $sort_keys, $check_sort );
-
-$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 => 'comment',   short => 'c', type => 'flag',   mandatory => 'no', default => undef, allowed => undef, disallowed => undef },
-        { long => 'delimit',   short => 'd', type => 'string', mandatory => 'no', default => "\t",  allowed => undef, disallowed => undef },
-        { long => 'keys',      short => 'k', type => 'list',   mandatory => 'no', default => undef, allowed => undef, disallowed => undef },
-        { long => 'no_keys',   short => 'K', type => 'list',   mandatory => 'no', default => undef, allowed => undef, disallowed => undef },
-        { long => 'compress',  short => 'Z', type => 'flag',   mandatory => 'no', default => undef, allowed => undef, disallowed => undef },
-    ]   
-);
-
-$in       = Maasha::Biopieces::read_stream( $options->{ "stream_in" } );
-$out      = Maasha::Biopieces::write_stream( $options->{ "stream_out" } );
-$data_out = Maasha::Biopieces::write_stream( $options->{ "data_out" }, $options->{ "compress" } );
-
-map { $no_keys{ $_ } = 1 } @{ $options->{ "no_keys" } };
-
-while ( $record = Maasha::Biopieces::get_record( $in ) ) 
-{
-    undef @vals;
-    $ok = 1;
-    
-    if ( $options->{ "keys" } )
-    {
-        map { $ok = 0 if not exists $record->{ $_ } } @{ $options->{ "keys" } };
-
-        if ( $ok )
-        {
-            foreach $key ( @{ $options->{ "keys" }  } )
-            {
-                if ( exists $record->{ $key } )
-                {
-                    push @keys, $key if $options->{ "comment" };
-                    push @vals, $record->{ $key };
-                }
-            }
-         }
-    }
+class Numeric
+  def commify
+    self.to_s.gsub(/(^[-+]?\d+?(?=(?>(?:\d{3})+)(?!\d))|\G\d{3}(?=\d))/, '\1,')
+  end
+end
+
+require 'terminal-table'
+require 'maasha/biopieces'
+require 'maasha/filesys'
+
+casts = []
+casts << {long: 'no_stream', short: 'x', type: 'flag',   mandatory: false, default: nil,  allowed: nil,               disallowed: nil}
+casts << {long: 'pretty',    short: 'p', type: 'flag',   mandatory: false, default: nil,  allowed: nil,               disallowed: nil}
+casts << {long: 'comment',   short: 'c', type: 'flag',   mandatory: false, default: nil,  allowed: nil,               disallowed: nil}
+casts << {long: 'commify',   short: 'C', type: 'flag',   mandatory: false, default: nil,  allowed: nil,               disallowed: nil}
+casts << {long: 'delimit',   short: 'd', type: 'string', mandatory: false, default: "\t", allowed: nil,               disallowed: nil}
+casts << {long: 'keys',      short: 'k', type: 'list',   mandatory: false, default: nil,  allowed: nil,               disallowed: nil}
+casts << {long: 'no_keys',   short: 'K', type: 'list',   mandatory: false, default: nil,  allowed: nil,               disallowed: nil}
+casts << {long: 'data_out',  short: 'o', type: 'file',   mandatory: false, default: nil,  allowed: nil,               disallowed: nil}
+casts << {long: 'compress',  short: 'Z', type: 'string', mandatory: false, default: nil,  allowed: "gzip,bzip,bzip2", disallowed: nil}
+
+options = Biopieces.options_parse(ARGV, casts)
+
+compress = options[:compress] ? options[:compress].to_sym : nil
+
+raise "--data_out is mandatory for compressed output" if compress and not options[:data_out]
+
+rows     = []
+headings = nil
+comment  = true if options[:comment]
+no_keys  = options[:no_keys].each_with_object({}) { |i, h| h[i.to_sym] = true } if options[:no_keys]
+
+def columns_order(record)
+  record.keys.each do |key|
+    unless key.match(/^V\d+$/)
+      return record.keys
+    end
+  end
+
+  record.keys.sort { |a, b| a.to_s[1 .. a.to_s.size].to_i <=> b.to_s[1 .. a.to_s.size].to_i }
+end
+
+tab_out = options[:data_out] ? Filesys.open(options[:data_out], 'w', compress: compress) : STDOUT
+
+Biopieces.open(options[:stream_in], options[:stream_out]) do |input, output|
+  input.each do |record|
+    unless headings
+      if options[:keys]
+        headings = options[:keys].map { |k| k.to_sym }
+      else
+        headings = columns_order(record)
+      end
+
+      headings.reject! {|r| no_keys[r] } if options[:no_keys]
+    end
+
+    row = record.values_at(*headings)
+
+    if options[:pretty]
+      rows << row
     else
-    {
-        if ( not $check_sort )
-        {
-            $sort_keys = 1;
-
-            map { $sort_keys = 0 if $_ !~ /^V(\d+)$/ } keys %{ $record };
-
-            $check_sort = 1;
-        }
-
-        if ( $sort_keys )
-        {
-            foreach $key ( sort { $A = $a; $B = $b; $A =~ s/^V(\d+)$/$1/; $B =~ s/^V(\d+)$/$1/; $A <=> $B } keys %{ $record } )
-            {
-                next if exists $no_keys{ $key };
-
-                push @keys, $key if $options->{ "comment" };
-                push @vals, $record->{ $key };
-            }
-        }
-        else
-        {
-            foreach $key ( keys %{ $record } )
-            {
-                next if exists $no_keys{ $key };
-
-                push @keys, $key if $options->{ "comment" };
-                push @vals, $record->{ $key };
-            }
-        
-        }
-    }
-
-    if ( @keys and $options->{ "comment" } )
-    {
-        print $data_out "#", join( $options->{ "delimit" }, @keys ), "\n";
-
-        delete $options->{ "comment" };
-    }
-
-    print $data_out join( $options->{ "delimit" }, @vals ), "\n" if @vals;
-
-    Maasha::Biopieces::put_record( $record, $out ) if not $options->{ "no_stream" };
-}
-
-close $data_out;
-
-Maasha::Biopieces::close_stream( $in );
-Maasha::Biopieces::close_stream( $out );
-
-
-# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
-
-
-BEGIN
-{
-    Maasha::Biopieces::status_set();
-}
-
-
-END
-{
-    Maasha::Biopieces::status_log();
-}
+      if comment
+        tab_out.puts "#" + headings.join(options[:delimit]) unless headings.empty?
+        comment = false
+      end
+
+      tab_out.puts row.join(options[:delimit]) unless row.empty?
+    end
+
+    output.puts record unless options[:no_stream]
+  end
+end
+
+if options[:pretty]
+  table = Terminal::Table.new
+  table.headings = headings if options[:comment]
+
+  first_row = rows.first.dup
+
+  if options[:commify]
+    rows.each do |row|
+      row.each_with_index do |cell, i|
+        begin Integer(cell)
+          row[i] = cell.to_i.commify
+        rescue
+          begin Float(cell)
+            row[i] = cell.to_f.commify
+          rescue
+          end
+        end
+      end
+    end
+  end
+
+  table.rows = rows
+  
+  first_row.each_with_index do |cell, i|
+    begin Float(cell)
+      table.align_column(i, :right)
+    rescue
+    end
+  end
+
+  tab_out.puts table
+end
+
+tab_out.close
 
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
diff --git a/bp_test/out/write_tab.out.7 b/bp_test/out/write_tab.out.7
new file mode 100644
index 0000000..9895e0f
--- /dev/null
+++ b/bp_test/out/write_tab.out.7
@@ -0,0 +1,6 @@
++-------+-------+-----------+
+| Human | 23524 | ATACGTCAG |
+| Dog   |  2442 | AGCATGAC  |
+| Mouse |   234 | GACTG     |
+| Cat   |  2342 | AAATGCA   |
++-------+-------+-----------+
diff --git a/bp_test/out/write_tab.out.8 b/bp_test/out/write_tab.out.8
new file mode 100644
index 0000000..b1b182d
--- /dev/null
+++ b/bp_test/out/write_tab.out.8
@@ -0,0 +1,6 @@
++-------+--------+-----------+
+| Human | 23,524 | ATACGTCAG |
+| Dog   |  2,442 | AGCATGAC  |
+| Mouse |    234 | GACTG     |
+| Cat   |  2,342 | AAATGCA   |
++-------+--------+-----------+
diff --git a/bp_test/out/write_tab.out.9 b/bp_test/out/write_tab.out.9
new file mode 100644
index 0000000..0c4740b
--- /dev/null
+++ b/bp_test/out/write_tab.out.9
@@ -0,0 +1,8 @@
++----------+--------+-----------+
+| Organism | Count  | Sequence  |
++----------+--------+-----------+
+| Human    | 23,524 | ATACGTCAG |
+| Dog      |  2,442 | AGCATGAC  |
+| Mouse    |    234 | GACTG     |
+| Cat      |  2,342 | AAATGCA   |
++----------+--------+-----------+
diff --git a/bp_test/test/test_write_tab b/bp_test/test/test_write_tab
index 7209fd7..6670702 100755
--- a/bp_test/test/test_write_tab
+++ b/bp_test/test/test_write_tab
@@ -14,11 +14,16 @@ run "$bp -I $in -d ',' -o $tmp -x"
 assert_no_diff $tmp $out.3
 clean
 
-run "$bp -I $in -Z -o $tmp.gz -x"
+run "$bp -I $in -Z gzip -o $tmp.gz -x"
 gunzip $tmp.gz
 assert_no_diff $tmp $out.4
 clean
 
+run "$bp -I $in -Z bzip2 -o $tmp.bz2 -x"
+bunzip2 $tmp.bz2
+assert_no_diff $tmp $out.4
+clean
+
 run "$bp -I $in -k 'Count' -o $tmp -x"
 assert_no_diff $tmp $out.5
 clean
@@ -26,3 +31,15 @@ clean
 run "$bp -I $in -K 'Count' -o $tmp -x"
 assert_no_diff $tmp $out.6
 clean
+
+run "$bp -I $in -p -o $tmp -x"
+assert_no_diff $tmp $out.7
+clean
+
+run "$bp -I $in -p -C -o $tmp -x"
+assert_no_diff $tmp $out.8
+clean
+
+run "$bp -I $in -p -c -C -o $tmp -x"
+assert_no_diff $tmp $out.9
+clean
diff --git a/bp_test/test_all b/bp_test/test_all
index 4a1aa13..e94395f 100755
--- a/bp_test/test_all
+++ b/bp_test/test_all
@@ -16,6 +16,7 @@ test_ruby
 test_ruby_gem "gnuplot"
 test_ruby_gem "narray"
 test_ruby_gem "RubyInline"
+test_ruby_gem "terminal-table"
 test_aux_program "blastall"
 test_aux_program "blat"
 test_aux_program "bwa"