dimus-biodiversity 0.0.10 → 0.0.11

data/bin/nnparse

@@ -1,6 +1,6 @@
 #!/usr/bin/env ruby
 require 'rubygems'
-gem 'dimus-biodiversity' rescue gem 'biodiversity'
+gem 'dimus-biodiversity' rescue gem 'biodiversity' rescue nil
 
 $LOAD_PATH.unshift(File.expand_path(File.dirname(__FILE__) + "/../lib"))
 require 'biodiversity'
@@ -37,11 +37,15 @@ IO.foreach(ARGV[0]) do |n|
     puts n
     count += 1
   else
-    name_dict[:output] = parsed.value
-    name_dict[:canononical] = parsed.canonical
-    name_dict[:details] = parsed.details
-    name_dict[:parsed => true]
-    last_result =  JSON.generate name_dict
+    begin
+      name_dict[:output] = parsed.value
+      name_dict[:canononical] = parsed.canonical
+      name_dict[:details] = parsed.details
+      name_dict[:parsed => true]
+      last_result =  JSON.generate name_dict
+    rescue
+     puts 'PROBLEM: ' + n 
+    end
   end
 end
 o.write(last_result + "\n") if last_result

data/lib/biodiversity/parser/scientific_name.rb

@@ -741,6 +741,10 @@ module ScientificName
       elements[1]
     end
 
+    def ex_sep
+      elements[2]
+    end
+
     def space
       elements[3]
     end
@@ -821,13 +825,7 @@ module ScientificName
         r7 = _nt_space
         s5 << r7
         if r7
-          if input.index("ex", index) == index
-            r8 = instantiate_node(SyntaxNode,input, index...(index + 2))
-            @index += 2
-          else
-            terminal_parse_failure("ex")
-            r8 = nil
-          end
+          r8 = _nt_ex_sep
           s5 << r8
           if r8
             r9 = _nt_space
@@ -984,9 +982,44 @@ module ScientificName
       elements[3]
     end
 
+    def space
+      elements[5]
+    end
+
+    def space
+      elements[7]
+    end
+
+    def b
+      elements[8]
+    end
   end
 
   module OriginalAuthorsNamesFull1
+    def value
+      "(" + a.value + " " + b.value + ")"
+    end
+    def details
+      {:orig_authors => a.details[:authors], :year => b.details[:year]}
+    end
+  end
+
+  module OriginalAuthorsNamesFull2
+    def space
+      elements[1]
+    end
+
+    def a
+      elements[2]
+    end
+
+    def space
+      elements[3]
+    end
+
+  end
+
+  module OriginalAuthorsNamesFull3
     def value
       "(" + a.value + ")"
     end
@@ -1003,44 +1036,120 @@ module ScientificName
       return cached
     end
 
-    i0, s0 = index, []
+    i0 = index
+    i1, s1 = index, []
     if input.index("(", index) == index
-      r1 = instantiate_node(SyntaxNode,input, index...(index + 1))
+      r2 = instantiate_node(SyntaxNode,input, index...(index + 1))
       @index += 1
     else
       terminal_parse_failure("(")
-      r1 = nil
+      r2 = nil
     end
-    s0 << r1
-    if r1
-      r2 = _nt_space
-      s0 << r2
-      if r2
-        r3 = _nt_authors_names_full
-        s0 << r3
-        if r3
-          r4 = _nt_space
-          s0 << r4
-          if r4
+    s1 << r2
+    if r2
+      r3 = _nt_space
+      s1 << r3
+      if r3
+        r4 = _nt_authors_names
+        s1 << r4
+        if r4
+          r5 = _nt_space
+          s1 << r5
+          if r5
             if input.index(")", index) == index
-              r5 = instantiate_node(SyntaxNode,input, index...(index + 1))
+              r6 = instantiate_node(SyntaxNode,input, index...(index + 1))
               @index += 1
             else
               terminal_parse_failure(")")
-              r5 = nil
+              r6 = nil
+            end
+            s1 << r6
+            if r6
+              r7 = _nt_space
+              s1 << r7
+              if r7
+                if input.index(Regexp.new('[,]'), index) == index
+                  r9 = instantiate_node(SyntaxNode,input, index...(index + 1))
+                  @index += 1
+                else
+                  r9 = nil
+                end
+                if r9
+                  r8 = r9
+                else
+                  r8 = instantiate_node(SyntaxNode,input, index...index)
+                end
+                s1 << r8
+                if r8
+                  r10 = _nt_space
+                  s1 << r10
+                  if r10
+                    r11 = _nt_year
+                    s1 << r11
+                  end
+                end
+              end
             end
-            s0 << r5
           end
         end
       end
     end
-    if s0.last
-      r0 = instantiate_node(SyntaxNode,input, i0...index, s0)
-      r0.extend(OriginalAuthorsNamesFull0)
-      r0.extend(OriginalAuthorsNamesFull1)
+    if s1.last
+      r1 = instantiate_node(SyntaxNode,input, i1...index, s1)
+      r1.extend(OriginalAuthorsNamesFull0)
+      r1.extend(OriginalAuthorsNamesFull1)
     else
-      self.index = i0
-      r0 = nil
+      self.index = i1
+      r1 = nil
+    end
+    if r1
+      r0 = r1
+    else
+      i12, s12 = index, []
+      if input.index("(", index) == index
+        r13 = instantiate_node(SyntaxNode,input, index...(index + 1))
+        @index += 1
+      else
+        terminal_parse_failure("(")
+        r13 = nil
+      end
+      s12 << r13
+      if r13
+        r14 = _nt_space
+        s12 << r14
+        if r14
+          r15 = _nt_authors_names_full
+          s12 << r15
+          if r15
+            r16 = _nt_space
+            s12 << r16
+            if r16
+              if input.index(")", index) == index
+                r17 = instantiate_node(SyntaxNode,input, index...(index + 1))
+                @index += 1
+              else
+                terminal_parse_failure(")")
+                r17 = nil
+              end
+              s12 << r17
+            end
+          end
+        end
+      end
+      if s12.last
+        r12 = instantiate_node(SyntaxNode,input, i12...index, s12)
+        r12.extend(OriginalAuthorsNamesFull2)
+        r12.extend(OriginalAuthorsNamesFull3)
+      else
+        self.index = i12
+        r12 = nil
+      end
+      if r12
+        r0 = r12
+      else
+        self.index = i0
+        r0 = nil
+      end
     end
 
     node_cache[:original_authors_names_full][start_index] = r0
@@ -1135,6 +1244,10 @@ module ScientificName
       elements[1]
     end
 
+    def ex_sep
+      elements[2]
+    end
+
     def space
       elements[3]
     end
@@ -1168,13 +1281,7 @@ module ScientificName
       r2 = _nt_space
       s0 << r2
       if r2
-        if input.index("ex", index) == index
-          r3 = instantiate_node(SyntaxNode,input, index...(index + 2))
-          @index += 2
-        else
-          terminal_parse_failure("ex")
-          r3 = nil
-        end
+        r3 = _nt_ex_sep
         s0 << r3
         if r3
           r4 = _nt_space
@@ -1290,6 +1397,45 @@ module ScientificName
     return r0
   end
 
+  def _nt_ex_sep
+    start_index = index
+    if node_cache[:ex_sep].has_key?(index)
+      cached = node_cache[:ex_sep][index]
+      @index = cached.interval.end if cached
+      return cached
+    end
+
+    i0 = index
+    if input.index("ex", index) == index
+      r1 = instantiate_node(SyntaxNode,input, index...(index + 2))
+      @index += 2
+    else
+      terminal_parse_failure("ex")
+      r1 = nil
+    end
+    if r1
+      r0 = r1
+    else
+      if input.index("in", index) == index
+        r2 = instantiate_node(SyntaxNode,input, index...(index + 2))
+        @index += 2
+      else
+        terminal_parse_failure("in")
+        r2 = nil
+      end
+      if r2
+        r0 = r2
+      else
+        self.index = i0
+        r0 = nil
+      end
+    end
+
+    node_cache[:ex_sep][start_index] = r0
+
+    return r0
+  end
+
   module AuthorsNames0
     def a
       elements[0]
@@ -1378,7 +1524,7 @@ module ScientificName
   module AuthorNameSeparator0
     def apply(a,b)
       sep = text_value.strip
-      sep = " " + sep if sep == "&"
+      sep = " et" if ["&","and","et"].include? sep
       a.value + sep + " " + b.value
     end
 
@@ -1418,8 +1564,32 @@ module ScientificName
         r0 = r2
         r0.extend(AuthorNameSeparator0)
       else
-        self.index = i0
-        r0 = nil
+        if input.index("and", index) == index
+          r3 = instantiate_node(SyntaxNode,input, index...(index + 3))
+          @index += 3
+        else
+          terminal_parse_failure("and")
+          r3 = nil
+        end
+        if r3
+          r0 = r3
+          r0.extend(AuthorNameSeparator0)
+        else
+          if input.index("et", index) == index
+            r4 = instantiate_node(SyntaxNode,input, index...(index + 2))
+            @index += 2
+          else
+            terminal_parse_failure("et")
+            r4 = nil
+          end
+          if r4
+            r0 = r4
+            r0.extend(AuthorNameSeparator0)
+          else
+            self.index = i0
+            r0 = nil
+          end
+        end
       end
     end
 
@@ -2104,17 +2274,17 @@ module ScientificName
               s13 << r17
               if r17
                 i18 = index
-                if input.index(Regexp.new('[^\\.]'), index) == index
+                if input.index(Regexp.new('[\\.]'), index) == index
                   r19 = instantiate_node(SyntaxNode,input, index...(index + 1))
                   @index += 1
                 else
                   r19 = nil
                 end
                 if r19
+                  r18 = nil
+                else
                   self.index = i18
                   r18 = instantiate_node(SyntaxNode,input, index...index)
-                else
-                  r18 = nil
                 end
                 s13 << r18
               end
@@ -2392,6 +2562,9 @@ module ScientificName
     def value
       a.value + b.value
     end
+    def details
+      {:editorial_markup => value, :is_valid => false}
+    end
   end
 
   def _nt_editorials
@@ -3277,25 +3450,40 @@ module ScientificName
     return r0
   end
 
-  module LatinWord0
+  module CapLatinWord0
+    def a
+      elements[0]
+    end
+
+    def b
+      elements[1]
+    end
   end
 
-  module LatinWord1
-    def value 
-      text_value.strip
+  module CapLatinWord1
+    def value
+      a.text_value + b.value
+    end
+    
+    def canonical 
+      value
+    end
+    
+    def details 
+      {:uninomial => value}
     end
   end
 
-  def _nt_latin_word
+  def _nt_cap_latin_word
     start_index = index
-    if node_cache[:latin_word].has_key?(index)
-      cached = node_cache[:latin_word][index]
+    if node_cache[:cap_latin_word].has_key?(index)
+      cached = node_cache[:cap_latin_word][index]
       @index = cached.interval.end if cached
       return cached
     end
 
     i0, s0 = index, []
-    if input.index(Regexp.new('[a-zë]'), index) == index
+    if input.index(Regexp.new('[A-Z]'), index) == index
       r1 = instantiate_node(SyntaxNode,input, index...(index + 1))
       @index += 1
     else
@@ -3303,117 +3491,303 @@ module ScientificName
     end
     s0 << r1
     if r1
-      s2, i2 = [], index
-      loop do
-        if input.index(Regexp.new('[a-z\\-ëüäöï]'), index) == index
-          r3 = instantiate_node(SyntaxNode,input, index...(index + 1))
-          @index += 1
-        else
-          r3 = nil
-        end
-        if r3
-          s2 << r3
-        else
-          break
-        end
-      end
-      if s2.empty?
-        self.index = i2
-        r2 = nil
-      else
-        r2 = instantiate_node(SyntaxNode,input, i2...index, s2)
-      end
+      r2 = _nt_latin_word
       s0 << r2
     end
     if s0.last
       r0 = instantiate_node(SyntaxNode,input, i0...index, s0)
-      r0.extend(LatinWord0)
-      r0.extend(LatinWord1)
+      r0.extend(CapLatinWord0)
+      r0.extend(CapLatinWord1)
     else
       self.index = i0
       r0 = nil
     end
 
-    node_cache[:latin_word][start_index] = r0
+    node_cache[:cap_latin_word][start_index] = r0
 
     return r0
   end
 
-  module CapLatinWord0
+  module LatinWord0
+    def a
+      elements[0]
+    end
+
+    def b
+      elements[1]
+    end
   end
 
-  module CapLatinWord1
+  module LatinWord1
     def value
-      text_value.strip
-    end
-    
-    def canonical 
-      text_value.strip
-    end
-    
-    def details 
-      {:uninomial => value}
+      a.text_value + b.value
     end
   end
 
-  def _nt_cap_latin_word
-    start_index = index
-    if node_cache[:cap_latin_word].has_key?(index)
-      cached = node_cache[:cap_latin_word][index]
-      @index = cached.interval.end if cached
-      return cached
+  module LatinWord2
+    def a
+      elements[0]
     end
 
-    i0, s0 = index, []
-    if input.index(Regexp.new('[A-Z]'), index) == index
-      r1 = instantiate_node(SyntaxNode,input, index...(index + 1))
+    def b
+      elements[1]
+    end
+  end
+
+  module LatinWord3
+    def value
+      a.value + b.value
+    end
+  end
+
+  def _nt_latin_word
+    start_index = index
+    if node_cache[:latin_word].has_key?(index)
+      cached = node_cache[:latin_word][index]
+      @index = cached.interval.end if cached
+      return cached
+    end
+
+    i0 = index
+    i1, s1 = index, []
+    if input.index(Regexp.new('[a-zë]'), index) == index
+      r2 = instantiate_node(SyntaxNode,input, index...(index + 1))
       @index += 1
     else
+      r2 = nil
+    end
+    s1 << r2
+    if r2
+      r3 = _nt_full_name_letters
+      s1 << r3
+    end
+    if s1.last
+      r1 = instantiate_node(SyntaxNode,input, i1...index, s1)
+      r1.extend(LatinWord0)
+      r1.extend(LatinWord1)
+    else
+      self.index = i1
       r1 = nil
     end
-    s0 << r1
     if r1
-      if input.index(Regexp.new('[a-zë]'), index) == index
-        r2 = instantiate_node(SyntaxNode,input, index...(index + 1))
-        @index += 1
+      r0 = r1
+    else
+      i4, s4 = index, []
+      r5 = _nt_digraph
+      s4 << r5
+      if r5
+        r6 = _nt_full_name_letters
+        s4 << r6
+      end
+      if s4.last
+        r4 = instantiate_node(SyntaxNode,input, i4...index, s4)
+        r4.extend(LatinWord2)
+        r4.extend(LatinWord3)
       else
-        r2 = nil
+        self.index = i4
+        r4 = nil
       end
-      s0 << r2
-      if r2
-        s3, i3 = [], index
-        loop do
-          if input.index(Regexp.new('[a-z\\-ëüäöï]'), index) == index
-            r4 = instantiate_node(SyntaxNode,input, index...(index + 1))
-            @index += 1
-          else
-            r4 = nil
-          end
-          if r4
-            s3 << r4
-          else
-            break
-          end
+      if r4
+        r0 = r4
+      else
+        self.index = i0
+        r0 = nil
+      end
+    end
+
+    node_cache[:latin_word][start_index] = r0
+
+    return r0
+  end
+
+  module FullNameLetters0
+    def a
+      elements[0]
+    end
+
+    def b
+      elements[1]
+    end
+  end
+
+  module FullNameLetters1
+    def value
+      a.value + b.value
+    end
+  end
+
+  module FullNameLetters2
+    def a
+      elements[0]
+    end
+
+    def b
+      elements[1]
+    end
+
+    def c
+      elements[2]
+    end
+  end
+
+  module FullNameLetters3
+    def value
+      a.value + b.value + c.value
+    end
+  end
+
+  def _nt_full_name_letters
+    start_index = index
+    if node_cache[:full_name_letters].has_key?(index)
+      cached = node_cache[:full_name_letters][index]
+      @index = cached.interval.end if cached
+      return cached
+    end
+
+    i0 = index
+    i1, s1 = index, []
+    r2 = _nt_digraph
+    s1 << r2
+    if r2
+      r3 = _nt_full_name_letters
+      s1 << r3
+    end
+    if s1.last
+      r1 = instantiate_node(SyntaxNode,input, i1...index, s1)
+      r1.extend(FullNameLetters0)
+      r1.extend(FullNameLetters1)
+    else
+      self.index = i1
+      r1 = nil
+    end
+    if r1
+      r0 = r1
+    else
+      i4, s4 = index, []
+      r5 = _nt_valid_name_letters
+      s4 << r5
+      if r5
+        r6 = _nt_digraph
+        s4 << r6
+        if r6
+          r7 = _nt_full_name_letters
+          s4 << r7
         end
-        if s3.empty?
-          self.index = i3
-          r3 = nil
+      end
+      if s4.last
+        r4 = instantiate_node(SyntaxNode,input, i4...index, s4)
+        r4.extend(FullNameLetters2)
+        r4.extend(FullNameLetters3)
+      else
+        self.index = i4
+        r4 = nil
+      end
+      if r4
+        r0 = r4
+      else
+        r8 = _nt_valid_name_letters
+        if r8
+          r0 = r8
         else
-          r3 = instantiate_node(SyntaxNode,input, i3...index, s3)
+          self.index = i0
+          r0 = nil
         end
-        s0 << r3
       end
     end
-    if s0.last
-      r0 = instantiate_node(SyntaxNode,input, i0...index, s0)
-      r0.extend(CapLatinWord0)
-      r0.extend(CapLatinWord1)
-    else
+
+    node_cache[:full_name_letters][start_index] = r0
+
+    return r0
+  end
+
+  module ValidNameLetters0
+    def value
+      text_value
+    end
+  end
+
+  def _nt_valid_name_letters
+    start_index = index
+    if node_cache[:valid_name_letters].has_key?(index)
+      cached = node_cache[:valid_name_letters][index]
+      @index = cached.interval.end if cached
+      return cached
+    end
+
+    s0, i0 = [], index
+    loop do
+      if input.index(Regexp.new('[a-z\\-ëüäöï]'), index) == index
+        r1 = instantiate_node(SyntaxNode,input, index...(index + 1))
+        @index += 1
+      else
+        r1 = nil
+      end
+      if r1
+        s0 << r1
+      else
+        break
+      end
+    end
+    if s0.empty?
       self.index = i0
       r0 = nil
+    else
+      r0 = instantiate_node(SyntaxNode,input, i0...index, s0)
+      r0.extend(ValidNameLetters0)
     end
 
-    node_cache[:cap_latin_word][start_index] = r0
+    node_cache[:valid_name_letters][start_index] = r0
+
+    return r0
+  end
+
+  module Digraph0
+    def value
+      'ae'
+    end
+  end
+
+  module Digraph1
+    def value
+      'oe'
+    end
+  end
+
+  def _nt_digraph
+    start_index = index
+    if node_cache[:digraph].has_key?(index)
+      cached = node_cache[:digraph][index]
+      @index = cached.interval.end if cached
+      return cached
+    end
+
+    i0 = index
+    if input.index(Regexp.new('[æ]'), index) == index
+      r1 = instantiate_node(SyntaxNode,input, index...(index + 1))
+      r1.extend(Digraph0)
+      @index += 1
+    else
+      r1 = nil
+    end
+    if r1
+      r0 = r1
+    else
+      if input.index(Regexp.new('[œ]'), index) == index
+        r2 = instantiate_node(SyntaxNode,input, index...(index + 1))
+        r2.extend(Digraph1)
+        @index += 1
+      else
+        r2 = nil
+      end
+      if r2
+        r0 = r2
+      else
+        self.index = i0
+        r0 = nil
+      end
+    end
+
+    node_cache[:digraph][start_index] = r0
 
     return r0
   end
@@ -3494,29 +3868,110 @@ module ScientificName
       return cached
     end
 
-    s0, i0 = [], index
+    i0 = index
+    r1 = _nt_year_with_character
+    if r1
+      r0 = r1
+    else
+      s2, i2 = [], index
+      loop do
+        if input.index(Regexp.new('[0-9\\?]'), index) == index
+          r3 = instantiate_node(SyntaxNode,input, index...(index + 1))
+          @index += 1
+        else
+          r3 = nil
+        end
+        if r3
+          s2 << r3
+        else
+          break
+        end
+      end
+      if s2.empty?
+        self.index = i2
+        r2 = nil
+      else
+        r2 = instantiate_node(SyntaxNode,input, i2...index, s2)
+        r2.extend(Year0)
+      end
+      if r2
+        r0 = r2
+      else
+        self.index = i0
+        r0 = nil
+      end
+    end
+
+    node_cache[:year][start_index] = r0
+
+    return r0
+  end
+
+  module YearWithCharacter0
+    def a
+      elements[0]
+    end
+
+  end
+
+  module YearWithCharacter1
+    def value
+      a.text_value
+    end
+    def details
+      {:year => value}
+    end
+  end
+
+  def _nt_year_with_character
+    start_index = index
+    if node_cache[:year_with_character].has_key?(index)
+      cached = node_cache[:year_with_character][index]
+      @index = cached.interval.end if cached
+      return cached
+    end
+
+    i0, s0 = index, []
+    s1, i1 = [], index
     loop do
       if input.index(Regexp.new('[0-9\\?]'), index) == index
-        r1 = instantiate_node(SyntaxNode,input, index...(index + 1))
+        r2 = instantiate_node(SyntaxNode,input, index...(index + 1))
         @index += 1
       else
-        r1 = nil
+        r2 = nil
       end
-      if r1
-        s0 << r1
+      if r2
+        s1 << r2
       else
         break
       end
     end
-    if s0.empty?
-      self.index = i0
-      r0 = nil
+    if s1.empty?
+      self.index = i1
+      r1 = nil
     else
+      r1 = instantiate_node(SyntaxNode,input, i1...index, s1)
+    end
+    s0 << r1
+    if r1
+      if input.index(Regexp.new('[a-zA-Z]'), index) == index
+        r3 = instantiate_node(SyntaxNode,input, index...(index + 1))
+        @index += 1
+      else
+        r3 = nil
+      end
+      s0 << r3
+    end
+    if s0.last
       r0 = instantiate_node(SyntaxNode,input, i0...index, s0)
-      r0.extend(Year0)
+      r0.extend(YearWithCharacter0)
+      r0.extend(YearWithCharacter1)
+    else
+      self.index = i0
+      r0 = nil
     end
 
-    node_cache[:year][start_index] = r0
+    node_cache[:year_with_character][start_index] = r0
 
     return r0
   end

data/lib/biodiversity/parser/scientific_name.treetop

@@ -1,3 +1,4 @@
+# encoding: UTF-8
 grammar ScientificName
 
   rule composite_scientific_name
@@ -27,7 +28,7 @@ grammar ScientificName
       end
     }
     /
-    scientific_name
+    scientific_name 
   end
   
   rule scientific_name
@@ -137,7 +138,7 @@ grammar ScientificName
       end
     }
     /
-    a:simple_authors_part space "ex" space b:simple_authors_part {
+    a:simple_authors_part space ex_sep space b:simple_authors_part {
       def value 
         a.value + " ex " + b.value
       end
@@ -179,6 +180,15 @@ grammar ScientificName
   end
   
   rule original_authors_names_full
+    "(" space a:authors_names space ")" space [,]? space b:year {
+      def value
+        "(" + a.value + " " + b.value + ")"
+      end
+      def details
+        {:orig_authors => a.details[:authors], :year => b.details[:year]}
+      end
+    }
+    /
     "(" space a:authors_names_full space ")" {
       def value
         "(" + a.value + ")"
@@ -202,7 +212,7 @@ grammar ScientificName
   end
   
   rule authors_revised_name
-    a:authors_names_full space "ex" space b:authors_names_full {
+    a:authors_names_full space ex_sep space b:authors_names_full {
       def value
         a.value + " ex " + b.value
       end
@@ -224,6 +234,10 @@ grammar ScientificName
     /
     authors_names
   end
+  
+  rule ex_sep
+    ("ex"/"in")
+  end
       
   rule authors_names
     a:author_name space sep:author_name_separator space b:authors_names {
@@ -240,10 +254,10 @@ grammar ScientificName
   end  
   
   rule author_name_separator
-    ("&"/",") {
+    ("&"/","/"and"/"et") {
       def apply(a,b)
         sep = text_value.strip
-        sep = " " + sep if sep == "&"
+        sep = " et" if ["&","and","et"].include? sep
         a.value + sep + " " + b.value
       end
 
@@ -315,7 +329,7 @@ grammar ScientificName
       end
     }
     /
-    space a:species_name space b:latin_word &[^\.] {
+    space a:species_name space b:latin_word ![\.] {
       def value
         a.value + " " + b.value
       end
@@ -385,6 +399,9 @@ grammar ScientificName
     space a:rank space [&]? space b:editorials {
       def value
         a.value + b.value
+      end
+      def details
+        {:editorial_markup => value, :is_valid => false}
       end 
     }
     /
@@ -483,23 +500,15 @@ grammar ScientificName
       end
     }
   end
-
-  rule latin_word
-    [a-zë] [a-z\-ëüäöï]+ {
-      def value 
-        text_value.strip
-      end
-    }
-  end
   
   rule cap_latin_word
-    [A-Z] [a-zë] [a-z\-ëüäöï]+ {
+    a:[A-Z] b:latin_word {
       def value
-        text_value.strip
+        a.text_value + b.value
       end
       
       def canonical 
-        text_value.strip
+        value
       end
       
       def details 
@@ -507,7 +516,59 @@ grammar ScientificName
       end
     }
   end
+
+  rule latin_word
+    a:[a-zë] b:full_name_letters {
+      def value
+        a.text_value + b.value
+      end
+     }
+     /
+     a:digraph b:full_name_letters {
+      def value
+        a.value + b.value
+      end
+     }
+  end
   
+  rule full_name_letters
+    a:digraph b:full_name_letters {
+      def value
+        a.value + b.value
+      end
+    }
+    /
+    a:valid_name_letters b:digraph c:full_name_letters {
+      def value
+        a.value + b.value + c.value
+      end
+    }
+    /
+    valid_name_letters
+  end
+  
+  rule valid_name_letters
+    [a-z\-ëüäöï]+ {
+      def value
+        text_value
+      end
+    }
+  end 
+  
+  rule digraph
+    [æ] {
+   def value
+     'ae'
+   end
+   }
+      /
+   [œ] {
+   def value
+     'oe'
+   end
+   }
+  end
+    
   rule hybrid_separator
     ("x"/"X"/"×") {
       def value
@@ -517,6 +578,8 @@ grammar ScientificName
   end
   
   rule year
+    year_with_character
+    /
     [0-9\?]+ {
       def value
         text_value.strip
@@ -526,7 +589,18 @@ grammar ScientificName
       end
     }
   end
-    
+  
+  rule year_with_character
+    a:[0-9\?]+ [a-zA-Z] {
+      def value
+        a.text_value
+      end
+      def details
+        {:year => value}
+      end
+    }
+  end
+  
   rule space
     [\s]*
   end

data/spec/parser/scientific_name.spec.rb

@@ -51,6 +51,31 @@ describe ScientificName do
     details(sn).should == {:subgenus=>"Amerigo", :authors=>{:year=>"1999", :names=>["Author"]}, :species=>"pealeii", :genus=>"Doriteuthis"}
   end
   
+  it 'should parse æ in the name' do
+    names = [
+      ["Læptura laetifica Dow, 1913", "Laeptura laetifica Dow 1913"],
+      ["Leptura lætifica Dow, 1913", "Leptura laetifica Dow 1913"],
+      ["Leptura leætifica Dow, 1913", "Leptura leaetifica Dow 1913"],
+      ["Leæptura laetifica Dow, 1913", "Leaeptura laetifica Dow 1913"],
+      ["Leœptura laetifica Dow, 1913", "Leoeptura laetifica Dow 1913"]
+    ]
+    names.each do |name_pair|
+      parse(name_pair[0]).should_not be_nil
+      value(name_pair[0]).should == name_pair[1]
+    end
+  end
+
+  it 'should parse year' do
+    sn = "Platypus bicaudatulus Schedl 1935"
+    parse(sn).should_not be_nil
+    value(sn).should == "Platypus bicaudatulus Schedl 1935"
+    sn = "Platypus bicaudatulus Schedl, 1935h"
+    parse(sn).should_not be_nil
+    value(sn).should == "Platypus bicaudatulus Schedl 1935"
+    details(sn).should == {:genus=>"Platypus", :species=>"bicaudatulus", :authors=>{:names=>["Schedl"], :year=>"1935"}}
+    parse("Platypus bicaudatulus Schedl, 1935B").should_not be_nil
+  end
+  
   it 'should parse species autonym for complex subspecies authorships' do
     parse("Aus bus Linn. var. bus").should_not be_nil
     details("Aus bus Linn. var. bus").should == {:species=>"bus", :species_authors=>{:authors=>{:names=>["Linn."]}}, :genus=>"Aus", :subspecies=>[{:rank=>"var.", :value=>"bus"}]}
@@ -61,18 +86,24 @@ describe ScientificName do
   it 'should parse several authors' do
     sn = "Pseudocercospora dendrobii U. Braun & Crous"
     parse(sn).should_not be_nil
-    value(sn).should == "Pseudocercospora dendrobii U. Braun & Crous"
+    value(sn).should == "Pseudocercospora dendrobii U. Braun et Crous"
     canonical(sn).should == "Pseudocercospora dendrobii"
     details(sn).should == {
         :authors=>{:names=>["U. Braun","Crous"]},
         :species=>"dendrobii", 
         :genus=>"Pseudocercospora"}
+    sn = "Pseudocercospora dendrobii U. Braun and Crous"
+    parse(sn).should_not be_nil
+    value(sn).should == "Pseudocercospora dendrobii U. Braun et Crous"
+    sn = "Pseudocercospora dendrobii U. Braun et Crous"
+    parse(sn).should_not be_nil
+    value(sn).should == "Pseudocercospora dendrobii U. Braun et Crous"    
   end
 
   it 'should parse several authors with a year' do
     sn = "Pseudocercospora dendrobii U. Braun & Crous 2003"
     parse(sn).should_not be_nil
-    value(sn).should == "Pseudocercospora dendrobii U. Braun & Crous 2003"
+    value(sn).should == "Pseudocercospora dendrobii U. Braun et Crous 2003"
     canonical(sn).should == "Pseudocercospora dendrobii"
     details(sn).should == {
         :authors=>{:names=>["U. Braun","Crous"], :year => "2003"},
@@ -80,21 +111,29 @@ describe ScientificName do
         :genus=>"Pseudocercospora"}
     sn = "Pseudocercospora dendrobii Crous, 2003"
     parse(sn).should_not be_nil
+    parse("Zophosis persis (Chatanay, 1914)").should_not be_nil
+    parse("Zophosis persis (Chatanay 1914)").should_not be_nil
+    parse("Zophosis persis (Chatanay), 1914").should_not be_nil
+    value("Zophosis persis (Chatanay), 1914").should == "Zophosis persis (Chatanay 1914)"
+    details("Zophosis persis (Chatanay), 1914").should == {:genus=>"Zophosis", :species=>"persis", :orig_authors=>{:names=>["Chatanay"]}, :year=>"1914"}
+    
+    parse("Zophosis persis (Chatanay) 1914").should_not be_nil
+    #parse("Zophosis persis Chatanay (1914)").should_not be_nil
   end  
   
   it 'should parse scientific name' do
     parse("Pseudocercospora dendrobii (H.C. Burnett) U. Braun & Crous 2003").should_not be_nil
-    value("Pseudocercospora dendrobii(H.C.     Burnett)U. Braun & Crous     2003").should == "Pseudocercospora dendrobii (H.C. Burnett) U. Braun & Crous 2003"
+    value("Pseudocercospora dendrobii(H.C.     Burnett)U. Braun & Crous     2003").should == "Pseudocercospora dendrobii (H.C. Burnett) U. Braun et Crous 2003"
     canonical("Pseudocercospora dendrobii(H.C.     Burnett)U. Braun & Crous     2003").should == "Pseudocercospora dendrobii"
     {:orig_authors=>{:names=>["H.C. Burnett"]}, :species=>"dendrobii", :authors=>{:year=>"2003", :names=>["U. Braun", "Crous"]}, :genus=>"Pseudocercospora"}
   
     parse("Stagonospora polyspora M.T. Lucas & Sousa da Câmara 1934").should_not be_nil
-    value("Stagonospora polyspora M.T. Lucas & Sousa da Câmara 1934").should == "Stagonospora polyspora M.T. Lucas & Sousa da Câmara 1934"
+    value("Stagonospora polyspora M.T. Lucas & Sousa da Câmara 1934").should == "Stagonospora polyspora M.T. Lucas et Sousa da Câmara 1934"
     details("Stagonospora polyspora M.T. Lucas & Sousa da Câmara 1934").should == {:authors=>{:year=>"1934", :names=>["M.T. Lucas", "Sousa da C\303\242mara"]}, :species=>"polyspora", :genus=>"Stagonospora"}
     
     parse("Cladoniicola staurospora Diederich, van den Boom & Aptroot 2001").should_not be_nil
     parse("Yarrowia lipolytica var. lipolytica (Wick., Kurtzman & E.A. Herrm.) Van der Walt & Arx 1981").should_not be_nil
-    value("Yarrowia lipolytica var. lipolytica (Wick., Kurtzman & E.A. Herrm.) Van der Walt & Arx 1981").should == "Yarrowia lipolytica var. lipolytica (Wick., Kurtzman & E.A. Herrm.) Van der Walt & Arx 1981"
+    value("Yarrowia lipolytica var. lipolytica (Wick., Kurtzman & E.A. Herrm.) Van der Walt & Arx 1981").should == "Yarrowia lipolytica var. lipolytica (Wick., Kurtzman et E.A. Herrm.) Van der Walt et Arx 1981"
     parse("Physalospora rubiginosa (Fr.) anon.").should_not be_nil
     parse("Pleurotus ëous (Berk.) Sacc. 1887").should_not be_nil
     parse("Lecanora wetmorei Śliwa 2004").should_not be_nil
@@ -114,11 +153,12 @@ describe ScientificName do
     parse("Peltula coriacea Büdel, Henssen & Wessels 1986").should_not be_nil
     #had to add no dot rule for trinomials without a rank to make it to work
     parse("Saccharomyces drosophilae anon.").should_not be_nil
+    details("Saccharomyces drosophilae anon.").should == {:genus=>"Saccharomyces", :species=>"drosophilae", :authors=>{:names=>["anon."]}}
   end
   
   it 'should parse several authors with several years' do
     parse("Pseudocercospora dendrobii (H.C. Burnett 1883) U. Braun & Crous 2003").should_not be_nil
-    value("Pseudocercospora dendrobii(H.C.     Burnett1883)U. Braun & Crous     2003").should == "Pseudocercospora dendrobii (H.C. Burnett 1883) U. Braun & Crous 2003"
+    value("Pseudocercospora dendrobii(H.C.     Burnett1883)U. Braun & Crous     2003").should == "Pseudocercospora dendrobii (H.C. Burnett 1883) U. Braun et Crous 2003"
     canonical("Pseudocercospora dendrobii(H.C.     Burnett 1883)U. Braun & Crous     2003").should == "Pseudocercospora dendrobii"
     details("Pseudocercospora dendrobii(H.C.     Burnett 1883)U. Braun & Crous     2003").should == {:orig_authors=>{:year=>"1883", :names=>["H.C. Burnett"]}, :species=>"dendrobii", :authors=>{:year=>"2003", :names=>["U. Braun", "Crous"]}, :genus=>"Pseudocercospora"}
   end
@@ -150,7 +190,7 @@ describe ScientificName do
   
   it "should parse name with several subspecies names NOT BOTANICAL CODE BUT NOT INFREQUENT" do
     parse("Hydnellum scrobiculatum var. zonatum f. parvum (Banker) D. Hall & D.E. Stuntz 1972").should_not be_nil
-    value("Hydnellum scrobiculatum var. zonatum f. parvum (Banker) D. Hall & D.E. Stuntz 1972").should == "Hydnellum scrobiculatum var. zonatum f. parvum (Banker) D. Hall & D.E. Stuntz 1972"
+    value("Hydnellum scrobiculatum var. zonatum f. parvum (Banker) D. Hall & D.E. Stuntz 1972").should == "Hydnellum scrobiculatum var. zonatum f. parvum (Banker) D. Hall et D.E. Stuntz 1972"
     details("Hydnellum scrobiculatum var. zonatum f. parvum (Banker) D. Hall & D.E. Stuntz 1972").should == {:orig_authors=>{:names=>["Banker"]}, :subspecies=>[{:rank=>"var.", :value=>"zonatum"}, {:rank=>"f.", :value=>"parvum"}], :species=>"scrobiculatum", :authors=>{:year=>"1972", :names=>["D. Hall", "D.E. Stuntz"]}, :genus=>"Hydnellum", :is_valid=>false}  
   end
   
@@ -185,6 +225,7 @@ describe ScientificName do
     #invalid but happens
     parse("Mycosphaerella eryngii (Fr. Duby) ex Oudem. 1897").should_not be_nil
     parse("Mycosphaerella eryngii (Fr.ex Duby) ex Oudem. 1897").should_not be_nil
+    parse("Salmonella werahensis (Castellani) Hauduroy and Ehringer in Hauduroy 1937").should_not be_nil
   end
   
   it "should parse multiplication sign" do
@@ -222,9 +263,12 @@ describe ScientificName do
   it "should parse name with subspecies without rank NOT BOTANICAL" do
     name = "Hydnellum scrobiculatum zonatum (Banker) D. Hall & D.E. Stuntz 1972"
     parse(name).should_not be_nil
-    value(name).should == "Hydnellum scrobiculatum zonatum (Banker) D. Hall & D.E. Stuntz 1972"
+    value(name).should == "Hydnellum scrobiculatum zonatum (Banker) D. Hall et D.E. Stuntz 1972"
     canonical(name).should == "Hydnellum scrobiculatum zonatum"
     details(name).should == {:orig_authors=>{:names=>["Banker"]}, :subspecies=>{:rank=>"n/a", :value=>"zonatum"}, :species=>"scrobiculatum", :authors=>{:year=>"1972", :names=>["D. Hall", "D.E. Stuntz"]}, :genus=>"Hydnellum"}
+    sp = "Begonia pingbienensis angustior"
+    parse(sp).should_not be_nil
+    details(sp).should == {:genus=>"Begonia", :species=>"pingbienensis", :subspecies=>{:rank=>"n/a", :value=>"angustior"}}
   end
   
   it "should not parse utf-8 chars in name part" do

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: dimus-biodiversity
 version: !ruby/object:Gem::Version 
-  version: 0.0.10
+  version: 0.0.11
 platform: ruby
 authors: 
 - Dmitry Mozzherin
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2008-10-21 00:00:00 -07:00
+date: 2009-04-11 00:00:00 -07:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency