inprogress

[jalview.git] / forester / ruby / evoruby / lib / evo / tool / taxonomy_processor.rb

diff --git a/forester/ruby/evoruby/lib/evo/tool/taxonomy_processor.rb b/forester/ruby/evoruby/lib/evo/tool/taxonomy_processor.rb
index baaec66..bf194c7 100644 (file)
--- a/forester/ruby/evoruby/lib/evo/tool/taxonomy_processor.rb
+++ b/forester/ruby/evoruby/lib/evo/tool/taxonomy_processor.rb
@@ -22,9 +22,9 @@ module Evoruby
   class TaxonomyProcessor
 
     PRG_NAME       = "tap"
-    PRG_DATE       = "2013.03.20"
+    PRG_DATE       = "130411"
     PRG_DESC       = "replacement of species names in multiple sequence files"
-    PRG_VERSION    = "2.00"
+    PRG_VERSION    = "2.002"
     COPYRIGHT      = "2013 Christian M Zmasek"
     CONTACT        = "phylosoft@gmail.com"
     WWW            = "https://sites.google.com/site/cmzmasek/home/software/forester"
@@ -158,9 +158,6 @@ module Evoruby
         Util.fatal_error( PRG_NAME, "failed to write file: " + e.to_s )
       end
       lf.close()
-      if ( @taxonomies.length > 0 )
-        Util.print_message( PRG_NAME, "number of unique taxonomies: " + @taxonomies.length.to_s )
-      end
       Util.print_message( PRG_NAME, "wrote: " + list_file )
       Util.print_message( PRG_NAME, "wrote: " + output )
       Util.print_message( PRG_NAME, "OK" )
@@ -170,18 +167,19 @@ module Evoruby
 
     def modify_name( desc, counter, file, extract_taxonomy )
       new_desc = nil
-      if desc =~ /^>?\s*\S{1,10}_([A-Z]{3,5})/
-        new_desc = counter.to_s( 16 ) + "_" + $1
-      elsif extract_taxonomy
-        if desc =~/\[([A-Z0-9]{3,5})\]\b/
+      desc.gsub!( /\s+/, ' ' )
+      #if desc =~ /^>?\s*\S{1,10}_(([A-Z9][A-Z]{2}[A-Z0-9]{2})|RAT|PIG|PEA|CAP)/
+      #  new_desc = counter.to_s( 16 ) + "_" + $1
+      if extract_taxonomy
+        if desc =~/\s\[(([A-Z9][A-Z]{2}[A-Z0-9]{2})|RAT|PIG|PEA|CAP)\]/
           new_desc = counter.to_s( 16 ) + "_" + $1
         else
-          Util.fatal_error( PRG_NAME, "illegal format in: " + desc )
+          Util.fatal_error( PRG_NAME, "could not get taxonomy from: " + desc )
         end
       else
         new_desc = counter.to_s( 16 )
       end
-      file.print( new_desc + ": " + desc + "\n" )
+      file.print( new_desc + "\t" + desc + "\n" )
       new_desc
     end