lib/bio/db/embl/uniprotkb.rb in bio-2.0.4 vs lib/bio/db/embl/uniprotkb.rb in bio-2.0.5

@@ -528,26 +528,47 @@
   #
   # OH   NCBI_TaxID=TaxID; HostName.
   # http://br.expasy.org/sprot/userman.html#OH_line
   def oh
     unless @data['OH']
-      @data['OH'] = fetch('OH').split("\. ").map {|x|
-        if x =~ /NCBI_TaxID=(\d+);/
-          taxid = $1
-        else
-          raise ArgumentError, ["Error: Invalid OH line format (#{self.entry_id}):",
-                                $!, "\n", get('OH'), "\n"].join
-          
+      oh = []
+      a = fetch('OH').split(/(NCBI\_TaxID\=)(\d+)(\;)/)
+      t = catch :error do
+        taxid = nil
+        host_name = nil
+        while x = a.shift
+          x = x.to_s.strip
+          case x
+          when ''
+            next
+          when 'NCBI_TaxID='
+            if taxid then
+              oh.push({'NCBI_TaxID' => taxid, 'HostName' => host_name})
+              taxid = nil
+              host_name = nil
+            end
+            taxid = a.shift
+            throw :error, :missing_semicolon if a.shift != ';'
+          else
+            throw :error, :missing_taxid if host_name
+            host_name = x
+            host_name.sub!(/\.\z/, '')
+          end
+        end #while x...
+        if taxid then
+          oh.push({'NCBI_TaxID' => taxid, 'HostName' => host_name})
+        elsif host_name then
+          throw :error, :missing_taxid_last
         end
-        if x =~ /NCBI_TaxID=\d+; (.+)/ 
-          host_name = $1
-          host_name.sub!(/\.$/, '')
-        else
-          host_name = nil
-        end
-        {'NCBI_TaxID' => taxid, 'HostName' => host_name}
-      }
+        nil
+      end #t = catch...
+      if t then
+        raise ArgumentError,
+              ["Error: Invalid OH line format (#{self.entry_id}):",
+               $!, "\n", get('OH'), "\n"].join
+      end
+      @data['OH'] = oh
     end
     @data['OH']
   end
 
 
@@ -920,10 +941,11 @@
     end
   end
 
 
   def cc_alternative_products(data)
+    return nil unless data
     ap = data.join('')
     return ap unless ap
 
     # Event, Named isoforms, Comment, [Name, Synonyms, IsoId, Sequnce]+
     tmp = {'Event' => "", 'Named isoforms' => "", 'Comment' => "", 
@@ -958,10 +980,11 @@
   end
   private :cc_alternative_products_variants
 
 
   def cc_biophysiochemical_properties(data)
+    return nil unless data
     data = data[0]
 
     hash = {'Absorption' => {}, 
             'Kinetic parameters' => {},
             'pH dependence' => "",
@@ -993,19 +1016,21 @@
   end
   private :cc_biophysiochemical_properties
 
 
   def cc_caution(data)
+    return nil unless data
     data.join('')
   end
   private :cc_caution
 
 
   # returns conteins in a line of the CC INTERACTION section.
   #
   #   CC       P46527:CDKN1B; NbExp=1; IntAct=EBI-359815, EBI-519280;
   def cc_interaction(data)
+    return nil unless data
     str = data.join('')
     it = str.scan(/(.+?); NbExp=(.+?); IntAct=(.+?);/)
     it.map {|ent|
       ent.map! {|x| x.strip }
       if ent[0] =~ /^(.+):(.+)/
@@ -1057,18 +1082,20 @@
   end
   private :cc_mass_spectrometry
 
 
   def cc_pathway(data)
+    return nil unless data
     data.map {|x| x.sub(/\.$/, '') }.map {|x|
       x.split(/; | and |: /)
     }[0]
   end
   private :cc_pathway
 
 
   def cc_rna_editing(data)
+    return nil unless data
     data = data.join('')
     entry = {'Modified_positions' => [], 'Note' => ""}
     if data =~ /Modified_positions=(.+?)(\.|;)/
       entry['Modified_positions'] = $1.sub(/\.$/, '').split(', ')
     else
@@ -1081,10 +1108,11 @@
   end
   private :cc_rna_editing
 
 
   def cc_subcellular_location(data)
+    return nil unless data
     data.map {|x| 
       x.split('. ').map {|y| 
         y.split('; ').map {|z| 
           z.sub(/\.$/, '') 
         } 
@@ -1099,10 +1127,11 @@
   # CC   -!- WEB RESOURCE: Name=ResourceName[; Note=FreeText][; URL=WWWAddress].  # Old format:
   # CC   -!- WEB RESOURCE: NAME=ResourceName[; NOTE=FreeText][; URL=WWWAddress].
   #++
 
   def cc_web_resource(data)
+    return nil unless data
     data.map {|x|
       entry = {'Name' => nil, 'Note' => nil, 'URL' => nil}
       x.split(';').each do |y|
         case y
         when /(Name|Note)\=(.+)/
@@ -1233,13 +1262,17 @@
     cur_ft = nil
     cont = false
     begin
       ftlines.each do |line|
         if /^FT +([^\s]+) +(([^\s]+)\:)?([\<\?]?[0-9]+|\?)(?:\.\.([\>\?]?[0-9]+|\?))?\s*$/ =~ line
-          cur_ft = [$1.to_s,        # Feature Name
-                    "#{$2}#{$4}",   # From
-                    $5.to_s,        # To
-                    []              # Qualifiers
+          f_name = $1.to_s
+          f_from = "#{$2}#{$4}"
+          f_to = $5.to_s
+          f_to = f_from if f_to.empty?
+          cur_ft = [f_name,   # Feature Name
+                    f_from,   # From
+                    f_to,     # To
+                    []        # Qualifiers
                    ]
           table.push cur_ft
           cont = false
         elsif cont && /^FT {19}/ =~ line
           str = $'