uniyank.rake

desc "yank out the UniGene number from the records downloaded in the ncbi task"
  task :uniyank do
    #require 'open-uri'
    #I_KNOW_I_AM_USING_AN_OLD_AND_BUGGY_VERSION_OF_LIBXML2 = true
    #can't find _why's hpricot. trying nokogiri
    #require 'nokogiri'
    
    #unigene regex
    #regex = /Hs.[0-9]{6}/
    #regex = /Hs.[0-9A-Z]{6}/
    regex = /Hs\./
    # open out file
    
    outfile = File.new("unigene.txt", "w+")
    
    Dir.glob("./ncbigp/*.xml") { |x| 
      myfile = File.new(x, "r")
      myfile.each do |line|
        #puts line.gsub!(/<Object-id_str>|<\/Object-id_str>/, "") if line =~ regex
        outfile << line.gsub!(/<Object-id_str>|<\/Object-id_str>/, "").lstrip if line =~ regex
      end
    }
    
    outfile.close
    
 #Dir.foreach("./ncbigp") { |f|       
 #    
 #    #unigene regex
 #    regex = /Hs.[0-9]{6}/
 #    
 #    if FileTest::exists?(f)
 #      if f =~ /ncbi_outfile_[A-Z]\w{5}.xml/
 #        puts "opening #{f}"
 #        infile = File.new(f, "r")
 #        infile.each do |line|
 #          #Debug
 #          puts line if line =~ regex
 #        end
 #      else
 #        puts "ignoring this #{f}"
 #      end
 #    else
 #      raise "File doesn't exist #{f}"
 #    end
 #  
 #}
    

  end