in progress...

[jalview.git] / forester / ruby / evoruby / lib / evo / io / parser / fasta_parser.rb

diff --git a/forester/ruby/evoruby/lib/evo/io/parser/fasta_parser.rb b/forester/ruby/evoruby/lib/evo/io/parser/fasta_parser.rb
index dfc3a79..f1da2a9 100644 (file)
--- a/forester/ruby/evoruby/lib/evo/io/parser/fasta_parser.rb
+++ b/forester/ruby/evoruby/lib/evo/io/parser/fasta_parser.rb
@@ -1,77 +1,70 @@
 #
 # = lib/evo/io/parser/fasta_parser - FastaParser class
 #
-# Copyright::  Copyright (C) 2006-2007 Christian M. Zmasek
-# License::    GNU Lesser General Public License (LGPL)
+# Copyright::    Copyright (C) 2017 Christian M. Zmasek
+# License::      GNU Lesser General Public License (LGPL)
 #
-# $Id: fasta_parser.rb,v 1.11 2010/10/08 22:04:17 cmzmasek Exp $
-#
-# last modified: 05/17/2007
+# Last modified: 2017/02/07
 
 require 'lib/evo/io/parser/msa_parser'
 require 'lib/evo/msa/msa'
 
-require 'iconv'
-
 module Evoruby
 
-    class FastaParser < MsaParser
+  class FastaParser < MsaParser
 
-        def initialize
-        end
+    def initialize
+    end
 
-        def parse( path )
-            Util.check_file_for_readability( path )
-            msa = Msa.new
-            current_seq = String.new()
-            name        = String.new()
-            saw_first_seq = false
-            ic = Iconv.new( 'UTF-8//IGNORE', 'UTF-8' )
-            File.open( path ) do | file |
-                while line = file.gets
-                    line = ic.iconv( line )
-                    if can_ignore?( line, saw_first_seq )
+    def parse( path )
+      Util.check_file_for_readability( path )
+      msa = Msa.new
+      current_seq = String.new()
+      name        = String.new()
+      saw_first_seq = false
+      File.open( path ) do | file |
+        while line = file.gets
+          
+          line.encode!("UTF-8", :invalid => :replace, :undef => :replace, :replace => "?")          
+          if can_ignore?( line, saw_first_seq )
 
-                    elsif line =~ /^\s*>\s*(.+)/
-                        saw_first_seq = true
-                        add_seq( name, current_seq, msa )
-                        name = $1
-                        current_seq = String.new()
-                    elsif line =~ /^\s*(.+)/
-                        if name.length < 1
-                            error_msg = "format error at: " + line
-                            raise IOError, error_msg
-                        end
-                        # was: seq = $1.rstrip
-                        seq =  $1.gsub(/\s+/, '')
-                        current_seq << seq
-                    else
-                        error_msg = "Unexpected line: " + line
-                        raise IOError, error_msg
-                    end
-                end
-            end
+          elsif line =~ /^\s*>\s*(.+)/
+            saw_first_seq = true
             add_seq( name, current_seq, msa )
-            return msa
+            name = $1
+            current_seq = String.new()
+          elsif line =~ /^\s*(.+)/
+            if name.length < 1
+              error_msg = "format error at: " + line
+              raise IOError, error_msg
+            end
+            # was: seq = $1.rstrip
+            seq =  $1.gsub(/\s+/, '')
+            current_seq << seq
+          else
+            error_msg = "Unexpected line: " + line
+            raise IOError, error_msg
+          end
         end
+      end
+      add_seq( name, current_seq, msa )
+      return msa
+    end
 
-        private
+    private
 
-        def add_seq( name, seq, msa )
-            if name.length > 0 && seq.length > 0
-                msa.add( name, seq )
-            end
-        end
+    def add_seq( name, seq, msa )
+      if name.length > 0 && seq.length > 0
+        msa.add( name, seq )
+      end
+    end
 
-        def can_ignore?( line, saw_first_seq )
-            return ( line !~ /\S/  ||
-                 line =~ /^\s*#/ ||
-                 line =~ /^\s*%/ ||
-                 line =~ /^\s*\/\// ||
-                 line =~ /^\s*!!/ ||
-                 ( !saw_first_seq && line =~/^\s*[^>]/ ) )
-        end
+    def can_ignore?( line, saw_first_seq )
+      return ( line !~ /\S/  ||
+         line =~ /^\s*#/ ||
+         ( !saw_first_seq && line =~/^\s*[^>]/ ) )
+    end
 
-    end # class FastaParser
+  end # class FastaParser
 
 end # module Evoruby