dimus-biodiversity 0.5.1 → 0.5.2

data/lib/biodiversity/parser/scientific_name_clean.treetop

@@ -162,13 +162,13 @@ grammar ScientificNameClean
   
   
   rule multinomial_name
-    a:genus space b:subgenus space c:species space_hard d:infraspecies_mult {
+    a:genus space b:subgenus space species_prefix? space c:species space_hard d:infraspecies_mult {
       def value
         a.value + " " + b.value + " " + c.value + " " + d.value
       end
     
       def canonical
-        a.canonical + " " + c.canonical + " " + d.canonical
+        a.canonical + " " + b.canonical + " " + c.canonical + " " + d.canonical
       end
     
       def pos
@@ -180,7 +180,7 @@ grammar ScientificNameClean
       end
     }
     /  
-    a:genus space b:subgenus space c:species {
+    a:genus space b:subgenus space species_prefix? space c:species {
       def value
         a.value + " " + b.value + " " + c.value
       end
@@ -198,7 +198,7 @@ grammar ScientificNameClean
       end
     }
     /
-    a:genus space_hard b:species space_hard c:infraspecies_mult {
+    a:genus space species_prefix? space b:species space_hard c:infraspecies_mult {
       def value
         a.value + " " + b.value + " " + c.value 
       end
@@ -216,7 +216,7 @@ grammar ScientificNameClean
       end
     }
     /
-    a:genus space_hard b:species {
+    a:genus space species_prefix? space b:species {
       def value
         a.value + " " + b.value 
       end
@@ -250,7 +250,6 @@ grammar ScientificNameClean
       end
       
       def details
-        #{:infraspecies => a.details[:infraspceies] << b.details[:infraspecies]}
         a_array =  a.details[:infraspecies].class == Array ? a.details[:infraspecies] : [a.details[:infraspecies]] 
         b_array = b.details[:infraspecies].class == Array ? b.details[:infraspecies] : [b.details[:infraspecies]]
         a.details.merge({:infraspecies => a_array + b_array})
@@ -320,7 +319,7 @@ grammar ScientificNameClean
   end
   
   rule taxon_concept_rank
-    "sec." {
+    ("sec."/"sensu.") {
       def value
         "sec."
       end
@@ -454,17 +453,21 @@ grammar ScientificNameClean
   end
   
   rule genus
-    cap_latin_word {
+    a:cap_latin_word !(space_hard author_prefix_word space_hard author_word) {
+      def value
+        a.value
+      end
+      
       def pos
-        {interval.begin => ['genus', interval.end]}
+        {a.interval.begin => ['genus', a.interval.end]}
       end
       
       def canonical
-        value
+        a.value
       end
           
       def details
-        {:genus => {:epitheton => value}}
+        {:genus => {:epitheton => a.value}}
       end
     }
   end
@@ -508,7 +511,7 @@ grammar ScientificNameClean
   end
   
   rule authorship
-    a:basionym_authorship_with_parenthesis space b:simple_authorship space c:ex_authorship {
+    a:basionym_authorship_with_parenthesis space b:simple_authorship ","? space c:ex_authorship {
       def value
         a.value + " " + b.value + " " + c.value
       end
@@ -540,7 +543,7 @@ grammar ScientificNameClean
     /
     basionym_authorship_with_parenthesis
     /
-    a:simple_authorship space b:ex_authorship {
+    a:simple_authorship ","? space b:ex_authorship {
       def value
         a.value + " " + b.value
       end
@@ -578,7 +581,7 @@ grammar ScientificNameClean
       end
     }
     /
-    left_paren space a:simple_authorship space b:ex_authorship space right_paren {
+    left_paren space a:simple_authorship ","? space b:ex_authorship space right_paren {
       def value
         "(" + a.value + " " + b.value + ")"
       end
@@ -645,6 +648,28 @@ grammar ScientificNameClean
   
 
   rule simple_authorship
+    a:authors_names space [,]? space b:year? [,]? space "non" space authors_names space [,]? space year {
+      def value
+        a.value + " " + b.value
+      end
+      
+      def pos
+        a.pos.merge(b.pos)
+      end
+      
+      def details
+        details_with_arg(:basionymAuthorTeam)
+      end
+      
+      def details_with_arg(authorTeamType = 'basionymAuthorTeam')
+        { :authorship => text_value, 
+          authorTeamType.to_sym => {
+            :authorTeam => a.text_value.strip
+          }.merge(a.details).merge(b.details)
+        }
+      end
+    }
+    /
     a:authors_names space [,]? space b:year {
       def value
         a.value + " " + b.value
@@ -790,7 +815,7 @@ grammar ScientificNameClean
       end
     }
     /
-    ("bis"/"arg."/"et al.\{\?\}"/"et al.") {
+    ("arg."/"et al.\{\?\}"/"et al.") {
       def value
         text_value.strip
       end
@@ -837,7 +862,7 @@ grammar ScientificNameClean
   end
   
   rule author_prefix_word
-    space ("da"/"der"/"den"/"de"/"du"/"la"/"ter"/"van"/"von") &space_hard {
+    space ("ab"/"bis"/"da"/"der"/"den"/"della"/"dela"/"de"/"di"/"du"/"la"/"ter"/"van"/"von") &space_hard {
       def value
         text_value
       end
@@ -924,14 +949,10 @@ grammar ScientificNameClean
       end
     }
   end
-  
-  #rule species_word
-  ##  (!"sensu" a:species_word_0) {
-   #   def value
-  #      a.value
-  #    end
-  #  }
-  #end
+
+  rule species_prefix
+    ("aff."/"corrig."/"?") &space_hard
+  end
   
   rule species_word
     a:[0-9]+ "-"? b:latin_word {
@@ -944,7 +965,7 @@ grammar ScientificNameClean
   end
 
   rule latin_word
-    a:[a-zëüäöïé] b:full_name_letters {
+    a:[a-zëüäöïéåóç] b:full_name_letters {
       def value
         a.text_value + b.value
       end
@@ -974,7 +995,7 @@ grammar ScientificNameClean
   end
 
   rule valid_name_letters
-    [a-z\-ëüäöïé]+ {
+    [a-z\-ëüäöïéåóç]+ {
       def value
         text_value
       end

data/lib/biodiversity/parser/scientific_name_dirty.rb

@@ -23,6 +23,223 @@ module ScientificNameDirty
     return r0
   end
 
+  module ScientificName50
+    def a
+      elements[0]
+    end
+
+    def garbage
+      elements[1]
+    end
+  end
+
+  module ScientificName51
+    def value
+      a.value
+    end
+    
+    def canonical
+      a.canonical
+    end
+    
+    def pos
+      a.pos
+    end
+    
+    def details
+      a.details
+    end
+  end
+
+  def _nt_scientific_name_5
+    start_index = index
+    if node_cache[:scientific_name_5].has_key?(index)
+      cached = node_cache[:scientific_name_5][index]
+      @index = cached.interval.end if cached
+      return cached
+    end
+
+    i0 = index
+    i1, s1 = index, []
+    r2 = _nt_scientific_name_4
+    s1 << r2
+    if r2
+      r3 = _nt_garbage
+      s1 << r3
+    end
+    if s1.last
+      r1 = instantiate_node(SyntaxNode,input, i1...index, s1)
+      r1.extend(ScientificName50)
+      r1.extend(ScientificName51)
+    else
+      self.index = i1
+      r1 = nil
+    end
+    if r1
+      r0 = r1
+    else
+      r4 = super
+      if r4
+        r0 = r4
+      else
+        self.index = i0
+        r0 = nil
+      end
+    end
+
+    node_cache[:scientific_name_5][start_index] = r0
+
+    return r0
+  end
+
+  module Infraspecies0
+    def a
+      elements[0]
+    end
+
+    def space
+      elements[1]
+    end
+
+    def b
+      elements[2]
+    end
+  end
+
+  module Infraspecies1
+    def value
+      a.value + " " + b.value
+    end
+
+    def canonical
+      a.canonical
+    end
+
+    def pos
+      a.pos.merge(b.pos)
+    end
+
+    def details
+      {:infraspecies => a.details[:infraspecies].merge(b.details)}
+    end
+  end
+
+  module Infraspecies2
+    def a
+      elements[0]
+    end
+
+    def space
+      elements[1]
+    end
+
+    def epitheton_authorship_inconsistencies
+      elements[2]
+    end
+
+    def space
+      elements[3]
+    end
+
+    def b
+      elements[4]
+    end
+  end
+
+  module Infraspecies3
+    def value
+      a.value + " " + b.value
+    end
+
+    def canonical
+      a.canonical
+    end
+
+    def pos
+      a.pos.merge(b.pos)
+    end
+
+    def details
+      {:infraspecies => a.details[:infraspecies].merge(b.details)}
+    end
+  end
+
+  def _nt_infraspecies
+    start_index = index
+    if node_cache[:infraspecies].has_key?(index)
+      cached = node_cache[:infraspecies][index]
+      @index = cached.interval.end if cached
+      return cached
+    end
+
+    i0 = index
+    i1, s1 = index, []
+    r2 = _nt_infraspecies_epitheton
+    s1 << r2
+    if r2
+      r3 = _nt_space
+      s1 << r3
+      if r3
+        r4 = _nt_year
+        s1 << r4
+      end
+    end
+    if s1.last
+      r1 = instantiate_node(SyntaxNode,input, i1...index, s1)
+      r1.extend(Infraspecies0)
+      r1.extend(Infraspecies1)
+    else
+      self.index = i1
+      r1 = nil
+    end
+    if r1
+      r0 = r1
+    else
+      i5, s5 = index, []
+      r6 = _nt_infraspecies_epitheton
+      s5 << r6
+      if r6
+        r7 = _nt_space
+        s5 << r7
+        if r7
+          r8 = _nt_epitheton_authorship_inconsistencies
+          s5 << r8
+          if r8
+            r9 = _nt_space
+            s5 << r9
+            if r9
+              r10 = _nt_authorship
+              s5 << r10
+            end
+          end
+        end
+      end
+      if s5.last
+        r5 = instantiate_node(SyntaxNode,input, i5...index, s5)
+        r5.extend(Infraspecies2)
+        r5.extend(Infraspecies3)
+      else
+        self.index = i5
+        r5 = nil
+      end
+      if r5
+        r0 = r5
+      else
+        r11 = super
+        if r11
+          r0 = r11
+        else
+          self.index = i0
+          r0 = nil
+        end
+      end
+    end
+
+    node_cache[:infraspecies][start_index] = r0
+
+    return r0
+  end
+
   module Species0
     def a
       elements[0]
@@ -335,20 +552,25 @@ module ScientificNameDirty
       if r5
         r0 = r5
       else
-        r9 = _nt_approximate_year
+        r9 = _nt_year_number_with_punctuation
         if r9
           r0 = r9
         else
-          r10 = _nt_double_year
+          r10 = _nt_approximate_year
           if r10
             r0 = r10
           else
-            r11 = super
+            r11 = _nt_double_year
             if r11
               r0 = r11
             else
-              self.index = i0
-              r0 = nil
+              r12 = super
+              if r12
+                r0 = r12
+              else
+                self.index = i0
+                r0 = nil
+              end
             end
           end
         end
@@ -563,6 +785,62 @@ module ScientificNameDirty
     return r0
   end
 
+  module YearNumberWithPunctuation0
+    def a
+      elements[0]
+    end
+
+  end
+
+  module YearNumberWithPunctuation1
+    def value
+      a.text_value
+    end
+
+    def pos
+      {interval.begin => ['year', interval.end]}
+    end
+
+    def details
+      {:year => value}
+    end
+  end
+
+  def _nt_year_number_with_punctuation
+    start_index = index
+    if node_cache[:year_number_with_punctuation].has_key?(index)
+      cached = node_cache[:year_number_with_punctuation][index]
+      @index = cached.interval.end if cached
+      return cached
+    end
+
+    i0, s0 = index, []
+    r1 = _nt_year_number
+    s0 << r1
+    if r1
+      if input.index(".", index) == index
+        r2 = instantiate_node(SyntaxNode,input, index...(index + 1))
+        @index += 1
+      else
+        terminal_parse_failure(".")
+        r2 = nil
+      end
+      s0 << r2
+    end
+    if s0.last
+      r0 = instantiate_node(SyntaxNode,input, i0...index, s0)
+      r0.extend(YearNumberWithPunctuation0)
+      r0.extend(YearNumberWithPunctuation1)
+    else
+      self.index = i0
+      r0 = nil
+    end
+
+    node_cache[:year_number_with_punctuation][start_index] = r0
+
+    return r0
+  end
+
   module PageNumber0
     def space
       elements[1]
@@ -633,6 +911,144 @@ module ScientificNameDirty
     return r0
   end
 
+  def _nt_epitheton_authorship_inconsistencies
+    start_index = index
+    if node_cache[:epitheton_authorship_inconsistencies].has_key?(index)
+      cached = node_cache[:epitheton_authorship_inconsistencies][index]
+      @index = cached.interval.end if cached
+      return cached
+    end
+
+    if input.index("corrig.", index) == index
+      r0 = instantiate_node(SyntaxNode,input, index...(index + 7))
+      @index += 7
+    else
+      terminal_parse_failure("corrig.")
+      r0 = nil
+    end
+
+    node_cache[:epitheton_authorship_inconsistencies][start_index] = r0
+
+    return r0
+  end
+
+  module Garbage0
+    def space
+      elements[0]
+    end
+
+    def space
+      elements[2]
+    end
+
+  end
+
+  module Garbage1
+    def space_hard
+      elements[0]
+    end
+
+  end
+
+  def _nt_garbage
+    start_index = index
+    if node_cache[:garbage].has_key?(index)
+      cached = node_cache[:garbage][index]
+      @index = cached.interval.end if cached
+      return cached
+    end
+
+    i0 = index
+    i1, s1 = index, []
+    r2 = _nt_space
+    s1 << r2
+    if r2
+      if input.index(Regexp.new('["\',.]'), index) == index
+        r3 = instantiate_node(SyntaxNode,input, index...(index + 1))
+        @index += 1
+      else
+        r3 = nil
+      end
+      s1 << r3
+      if r3
+        r4 = _nt_space
+        s1 << r4
+        if r4
+          s5, i5 = [], index
+          loop do
+            if input.index(Regexp.new('[^щ]'), index) == index
+              r6 = instantiate_node(SyntaxNode,input, index...(index + 1))
+              @index += 1
+            else
+              r6 = nil
+            end
+            if r6
+              s5 << r6
+            else
+              break
+            end
+          end
+          r5 = instantiate_node(SyntaxNode,input, i5...index, s5)
+          s1 << r5
+        end
+      end
+    end
+    if s1.last
+      r1 = instantiate_node(SyntaxNode,input, i1...index, s1)
+      r1.extend(Garbage0)
+    else
+      self.index = i1
+      r1 = nil
+    end
+    if r1
+      r0 = r1
+    else
+      i7, s7 = index, []
+      r8 = _nt_space_hard
+      s7 << r8
+      if r8
+        s9, i9 = [], index
+        loop do
+          if input.index(Regexp.new('[^ш]'), index) == index
+            r10 = instantiate_node(SyntaxNode,input, index...(index + 1))
+            @index += 1
+          else
+            r10 = nil
+          end
+          if r10
+            s9 << r10
+          else
+            break
+          end
+        end
+        if s9.empty?
+          self.index = i9
+          r9 = nil
+        else
+          r9 = instantiate_node(SyntaxNode,input, i9...index, s9)
+        end
+        s7 << r9
+      end
+      if s7.last
+        r7 = instantiate_node(SyntaxNode,input, i7...index, s7)
+        r7.extend(Garbage1)
+      else
+        self.index = i7
+        r7 = nil
+      end
+      if r7
+        r0 = r7
+      else
+        self.index = i0
+        r0 = nil
+      end
+    end
+
+    node_cache[:garbage][start_index] = r0
+
+    return r0
+  end
+
 end
 
 class ScientificNameDirtyParser < Treetop::Runtime::CompiledParser