From 3ef3037a8bdfca6bb16fdd4455e1c9218352ad3a Mon Sep 17 00:00:00 2001
From: martinahansen <martinahansen@74ccb610-7750-0410-82ae-013aeee3265d>
Date: Wed, 25 Sep 2013 19:29:53 +0000
Subject: [PATCH] added support for gzip/bzip2 in write_fasta_files (and wrap)

git-svn-id: http://biopieces.googlecode.com/svn/trunk@2212 74ccb610-7750-0410-82ae-013aeee3265d
---
 bp_bin/write_fasta_files                     | 20 +++++++++++--------
 bp_test/out/write_fasta_files.out.3/11.fasta |  8 ++++++++
 bp_test/out/write_fasta_files.out.3/9.fasta  |  4 ++++
 bp_test/test/test_write_fasta_files          | 21 ++++++++++++++++++--
 4 files changed, 43 insertions(+), 10 deletions(-)
 create mode 100644 bp_test/out/write_fasta_files.out.3/11.fasta
 create mode 100644 bp_test/out/write_fasta_files.out.3/9.fasta

diff --git a/bp_bin/write_fasta_files b/bp_bin/write_fasta_files
index 36011ed..a0e20a5 100755
--- 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]
diff --git a/bp_test/out/write_fasta_files.out.3/11.fasta b/bp_test/out/write_fasta_files.out.3/11.fasta
new file mode 100644
index 0000000..85f3184
--- /dev/null
+++ b/bp_test/out/write_fasta_files.out.3/11.fasta
@@ -0,0 +1,8 @@
+>test2
+ACGA
+CTAC
+AGT
+>test3
+GCAC
+ACAG
+AGC
diff --git a/bp_test/out/write_fasta_files.out.3/9.fasta b/bp_test/out/write_fasta_files.out.3/9.fasta
new file mode 100644
index 0000000..4e39e70
--- /dev/null
+++ b/bp_test/out/write_fasta_files.out.3/9.fasta
@@ -0,0 +1,4 @@
+>test1
+GACA
+TCGA
+C
diff --git a/bp_test/test/test_write_fasta_files b/bp_test/test/test_write_fasta_files
index d87fb8a..9332864 100755
--- a/bp_test/test/test_write_fasta_files
+++ b/bp_test/test/test_write_fasta_files
@@ -3,16 +3,33 @@
 source "$BP_DIR/bp_test/lib/test.sh"
 
 mkdir $tmp_dir
-
 run "$bp -I $in -d $tmp_dir -k SEQ_NAME -x"
 assert_no_diff_dir $tmp_dir $out.1
 clean
+rm -rf $tmp_dir
 
+mkdir $tmp_dir
+run "$bp -I $in -d $tmp_dir -k SEQ_LEN -x"
+assert_no_diff_dir $tmp_dir $out.2
+clean
 rm -rf $tmp_dir
+
 mkdir $tmp_dir
+run "$bp -I $in -d $tmp_dir -k SEQ_LEN -Z gzip -x"
+gunzip $tmp_dir/*.gz
+assert_no_diff_dir $tmp_dir $out.2
+clean
+rm -rf $tmp_dir
 
-run "$bp -I $in -d $tmp_dir -k SEQ_LEN -x"
+mkdir $tmp_dir
+run "$bp -I $in -d $tmp_dir -k SEQ_LEN -Z bzip2 -x"
+bunzip2 $tmp_dir/*.bz2
 assert_no_diff_dir $tmp_dir $out.2
 clean
+rm -rf $tmp_dir
 
+mkdir $tmp_dir
+run "$bp -I $in -d $tmp_dir -k SEQ_LEN -w 4 -x"
+assert_no_diff_dir $tmp_dir $out.3
+clean
 rm -rf $tmp_dir
-- 
2.39.2