added support for gzip/bzip2 in write_fasta_files (and wrap)

[biopieces.git] / bp_bin / write_fasta_files

diff --git a/bp_bin/write_fasta_files b/bp_bin/write_fasta_files
index 36011ed411fad74d891f2f12f7108bfc02eb91a2..a0e20a50545643b2183ee2fdf229ba08ab2d68df 100755 (executable)
--- a/bp_bin/write_fasta_files
+++ b/bp_bin/write_fasta_files
@@ -33,15 +33,17 @@ require 'maasha/fasta'
 require 'pp'
 
 casts = []
-casts << {:long=>'key',       :short=>'k', :type=>'string', :mandatory=>true,  :default=>nil, :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'dir',       :short=>'d', :type=>'dir!',   :mandatory=>true,  :default=>nil, :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'no_stream', :short=>'x', :type=>'flag',   :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
+casts << {long: 'key',       short: 'k', type: 'string', mandatory: true,  default: nil, allowed: nil,               disallowed: nil}
+casts << {long: 'dir',       short: 'd', type: 'dir!',   mandatory: true,  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}
+casts << {long: 'no_stream', short: 'x', type: 'flag',   mandatory: false, default: nil, allowed: nil,               disallowed: nil}
 
 options = Biopieces.options_parse(ARGV, casts)
 
-key = options[:key].to_sym
-
-fh_hash = {}
+key      = options[:key].to_sym
+compress = options[:compress] ? options[:compress].to_sym : nil
+fh_hash  = {}
 
 Biopieces.open(options[:stream_in], options[:stream_out]) do |input, output|
   input.each_record do |record|
@@ -52,11 +54,13 @@ Biopieces.open(options[:stream_in], options[:stream_out]) do |input, output|
         fasta_io = fh_hash[record[key].to_sym]
       else
         fasta_file = File.join(options[:dir], record[key] + ".fasta")
-        fasta_io   = Fasta.open(fasta_file, "w")
+        fasta_file << ".gz"  if compress == :gzip
+        fasta_file << ".bz2" if compress == :bzip or compress == :bzip2
+        fasta_io   = Fasta.open(fasta_file, "w", compress: compress)
         fh_hash[record[key].to_sym] = fasta_io
       end
 
-      fasta_io.puts seq.to_fasta
+      fasta_io.puts seq.to_fasta(options[:wrap])
     end
 
     output.puts record unless options[:no_stream]