dimus-biodiversity 0.5.1 → 0.5.2

data/lib/biodiversity/parser/scientific_name_dirty.treetop

@@ -5,6 +5,68 @@ grammar ScientificNameDirty
   rule root
     super
   end
+
+  rule scientific_name_5
+    a:scientific_name_4 garbage {
+      def value
+        a.value
+      end
+      
+      def canonical
+        a.canonical
+      end
+      
+      def pos
+        a.pos
+      end
+      
+      def details
+        a.details
+      end
+    }
+    /
+    super
+  end
+  
+  rule infraspecies
+    a:infraspecies_epitheton space b:year {
+      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
+    }
+    /
+    a:infraspecies_epitheton space epitheton_authorship_inconsistencies space b:authorship {
+      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
+    }
+    /
+    super
+  end
   
   rule species
     a:species_epitheton space b:year {
@@ -33,8 +95,6 @@ grammar ScientificNameDirty
    /
    super
   end
-    
-  
   
   rule left_paren
    "(" space "("
@@ -71,6 +131,8 @@ grammar ScientificNameDirty
       end
     }
     /
+    year_number_with_punctuation
+    /
     approximate_year
     /
     double_year
@@ -111,6 +173,23 @@ grammar ScientificNameDirty
     }
   end
   
+  rule year_number_with_punctuation
+    a:year_number "." {
+      def value
+        a.text_value
+      end
+
+      def pos
+        {interval.begin => ['year', interval.end]}
+      end
+
+      def details
+        {:year => value}
+      end
+    }
+  end
+
+  
   rule page_number
     ":" space [\d]+
     {
@@ -119,5 +198,14 @@ grammar ScientificNameDirty
     }
   end
   
+  rule epitheton_authorship_inconsistencies
+    ("corrig.")
+  end
+  
+  rule garbage
+    space (["',.]) space [^щ]*
+    /
+    space_hard [^ш]+
+  end
   
 end

data/spec/parser/scientific_name.spec.rb

@@ -1,39 +1,11 @@
-require 'rubygems'
-require 'spec'
-require 'treetop'
-require 'yaml'
-
 #NOTE: this spec needs compiled treetop files.
 dir = File.dirname("__FILE__")
+require File.expand_path(dir + '../../spec/parser/spec_helper')
 require File.expand_path(dir + '../../lib/biodiversity/parser')
 
 describe ScientificNameClean do
   before(:all) do
-    @parser = ScientificNameParser.new 
-  end
-  
-  def parse(input)
-    @parser.parse(input)
-  end
-  
-  def value(input)
-    parse(input).value
-  end
-  
-  def canonical(input)
-    parse(input).canonical
-  end
-  
-  def details(input)
-    parse(input).details
-  end
-  
-  def pos(input)
-    parse(input).pos
-  end
-  
-  def json(input)
-    parse(input).to_json
+    set_parser(ScientificNameParser.new)
   end
   
   it 'should generate standardized json' do
@@ -44,5 +16,10 @@ describe ScientificNameClean do
       JSON.load(json(name)).should == JSON.load(jsn)
     end
   end
+  
+  it 'should generate reasonable output if parser failed' do
+    sn = 'ddd sljlkj 3223452432'
+    json(sn).should == '{"scientificName":{"parsed":false,"verbatim":"ddd sljlkj 3223452432"}}'
+  end
 
 end

data/spec/parser/scientific_name_canonical.spec.rb

@@ -1,38 +1,12 @@
+# encoding: UTF-8
 dir = File.dirname("__FILE__")
-require 'rubygems'
-require 'spec'
-require 'treetop'
-require 'yaml'
-
-Treetop.load(File.expand_path(dir + '../../lib/biodiversity/parser/scientific_name_clean'))
-Treetop.load(File.expand_path(dir + '../../lib/biodiversity/parser/scientific_name_dirty'))
-Treetop.load(File.expand_path(dir + '../../lib/biodiversity/parser/scientific_name_canonical'))
-
+require File.expand_path(dir + '../../spec/parser/spec_helper')
 
 describe ScientificNameCanonical do
   before(:all) do
-    @parser = ScientificNameCanonicalParser.new 
-  end
-  
-  def parse(input)
-    @parser.parse(input)
-  end
-  
-  def value(input)
-    parse(input).value
+    set_parser(ScientificNameCanonicalParser.new)
   end
   
-  def canonical(input)
-    parse(input).canonical
-  end
-  
-  def details(input)
-    parse(input).details
-  end
-  
-  def pos(input)
-    parse(input).pos
-  end
     
   it 'should parse names with valid name part and unparseable rest' do
     [
@@ -48,4 +22,5 @@ describe ScientificNameCanonical do
       pos(n[0]).should == n[3]
     end
   end  
+  
 end

data/spec/parser/scientific_name_clean.spec.rb

@@ -1,35 +1,11 @@
 # encoding: UTF-8
 dir = File.dirname("__FILE__")
-require 'rubygems'
-require 'spec'
-require 'yaml'
-require 'treetop'
+require File.expand_path(dir + '../../spec/parser/spec_helper')
 
-Treetop.load(File.expand_path(dir + '../../lib/biodiversity/parser/scientific_name_clean'))
 
 describe ScientificNameClean do
   before(:all) do
-    @parser = ScientificNameCleanParser.new 
-  end
-  
-  def parse(input)
-    @parser.parse(input)
-  end
-  
-  def value(input)
-    parse(input).value
-  end
-  
-  def canonical(input)
-    parse(input).canonical
-  end
-  
-  def details(input)
-    parse(input).details
-  end
-  
-  def pos(input)
-    parse(input).pos
+    set_parser(ScientificNameCleanParser.new)
   end
   
   it 'should parse uninomial' do
@@ -312,7 +288,7 @@ describe ScientificNameClean do
     sn = "Gastrosericus eremorum von Beaumont 1955"
     canonical(sn).should == 'Gastrosericus eremorum'
     sn = "Cypraeovula (Luponia) amphithales perdentata"
-    canonical(sn).should == 'Cypraeovula amphithales perdentata'
+    canonical(sn).should == 'Cypraeovula Luponia amphithales perdentata'
     details(sn).should == {:genus=>{:epitheton=>"Cypraeovula"}, :subgenus=>{:epitheton=>"Luponia"}, :species=>{:epitheton=>"amphithales"}, :infraspecies=>{:epitheton=>"perdentata", :rank=>"n/a"}}
     sn = "Polyrhachis orsyllus nat musculus Forel 1901"
     canonical(sn).should == "Polyrhachis orsyllus musculus"
@@ -490,5 +466,33 @@ describe ScientificNameClean do
     parse("Trematоsphaeria phaeáapora").should be_nil #cyrillic o
   end
 
+  it "should parse new stuff" do
+    sn = 'Nesticus quelpartensis Paik & Namkung, in Paik, Yaginuma & Namkung, 1969'
+    details(sn).should == {:genus=>{:epitheton=>"Nesticus"}, :species=>{:epitheton=>"quelpartensis", :authorship=>"Paik & Namkung, in Paik, Yaginuma & Namkung, 1969", :basionymAuthorTeam=>{:authorTeam=>"Paik & Namkung", :author=>["Paik", "Namkung"], :exAuthorTeam=>{:authorTeam=>"Paik, Yaginuma & Namkung", :author=>["Paik", "Yaginuma", "Namkung"], :year=>"1969"}}}}
+    parse('Dipoena yoshidai Ono, in Ono et al., 1991').should_not be_nil
+    sn = 'Choriozopella trägårdhi Lawrence, 1947'
+    details(sn).should == {:genus=>{:epitheton=>"Choriozopella"}, :species=>{:epitheton=>"trägårdhi", :authorship=>"Lawrence, 1947", :basionymAuthorTeam=>{:authorTeam=>"Lawrence", :author=>["Lawrence"], :year=>"1947"}}}
+    sn = 'Latrodectus mactans bishopi Kaston, 1938'
+    details(sn).should == {:genus=>{:epitheton=>"Latrodectus"}, :species=>{:epitheton=>"mactans"}, :infraspecies=>{:epitheton=>"bishopi", :rank=>"n/a", :authorship=>"Kaston, 1938", :basionymAuthorTeam=>{:authorTeam=>"Kaston", :author=>["Kaston"], :year=>"1938"}}}
+    sn = 'Diplocephalus aff. procerus Thaler, 1972'
+    details(sn).should == {:genus=>{:epitheton=>"Diplocephalus"}, :species=>{:epitheton=>"procerus", :authorship=>"Thaler, 1972", :basionymAuthorTeam=>{:authorTeam=>"Thaler", :author=>["Thaler"], :year=>"1972"}}}
+    sn = 'Dyarcyops birói Kulczynski, 1908'
+    details(sn).should == {:genus=>{:epitheton=>"Dyarcyops"}, :species=>{:epitheton=>"birói", :authorship=>"Kulczynski, 1908", :basionymAuthorTeam=>{:authorTeam=>"Kulczynski", :author=>["Kulczynski"], :year=>"1908"}}}
+    sn = 'Sparassus françoisi Simon, 1898'
+    details(sn).should == {:genus=>{:epitheton=>"Sparassus"}, :species=>{:epitheton=>"françoisi", :authorship=>"Simon, 1898", :basionymAuthorTeam=>{:authorTeam=>"Simon", :author=>["Simon"], :year=>"1898"}}}
+    sn = 'Thiobacillus x Parker and Prisk 1953' #have to figure out black lists for this one
+    sn = 'Bacille de Plaut, Kritchevsky and Séguin 1921'
+    details(sn).should == {:uninomial=>{:epitheton=>"Bacille", :authorship=>"de Plaut, Kritchevsky and Séguin 1921", :basionymAuthorTeam=>{:authorTeam=>"de Plaut, Kritchevsky and Séguin", :author=>["de Plaut", "Kritchevsky", "Séguin"], :year=>"1921"}}}
+    sn = 'Araneus van bruysseli Petrunkevitch, 1911'
+    details(sn).should == {:genus=>{:epitheton=>"Araneus"}, :species=>{:epitheton=>"van"}, :infraspecies=>{:epitheton=>"bruysseli", :rank=>"n/a", :authorship=>"Petrunkevitch, 1911", :basionymAuthorTeam=>{:authorTeam=>"Petrunkevitch", :author=>["Petrunkevitch"], :year=>"1911"}}}
+    sn = 'Sapromyces laidlawi ab Sabin 1941'
+    details(sn).should == {:genus=>{:epitheton=>"Sapromyces"}, :species=>{:epitheton=>"laidlawi", :authorship=>"ab Sabin 1941", :basionymAuthorTeam=>{:authorTeam=>"ab Sabin", :author=>["ab Sabin"], :year=>"1941"}}}
+    sn = 'Nocardia rugosa di Marco and Spalla 1957'
+    details(sn).should == {:genus=>{:epitheton=>"Nocardia"}, :species=>{:epitheton=>"rugosa", :authorship=>"di Marco and Spalla 1957", :basionymAuthorTeam=>{:authorTeam=>"di Marco and Spalla", :author=>["di Marco", "Spalla"], :year=>"1957"}}}
+    sn = 'Flexibacter elegans Lewin 1969 non Soriano 1945'
+    details(sn).should == {:genus=>{:epitheton=>"Flexibacter"}, :species=>{:epitheton=>"elegans", :authorship=>"Lewin 1969 non Soriano 1945", :basionymAuthorTeam=>{:authorTeam=>"Lewin", :author=>["Lewin"], :year=>"1969"}}}
+    sn = 'Flexibacter elegans Soriano 1945, non Lewin 1969'
+    details(sn).should == {:genus=>{:epitheton=>"Flexibacter"}, :species=>{:epitheton=>"elegans", :authorship=>"Soriano 1945, non Lewin 1969", :basionymAuthorTeam=>{:authorTeam=>"Soriano", :author=>["Soriano"], :year=>"1945"}}}
+  end
   
 end

data/spec/parser/scientific_name_dirty.spec.rb

@@ -1,55 +1,13 @@
 # encoding: UTF-8
 dir = File.dirname("__FILE__")
-require 'rubygems'
-require 'spec'
-require 'yaml'
-require 'treetop'
-
-Treetop.load(File.expand_path(dir + '../../lib/biodiversity/parser/scientific_name_clean'))
-Treetop.load(File.expand_path(dir + '../../lib/biodiversity/parser/scientific_name_dirty'))
+require File.expand_path(dir + '../../spec/parser/spec_helper')
 
 describe ScientificNameDirty do
   before(:all) do
-    @parser = ScientificNameDirtyParser.new 
-  end
-  
-  def parse(input)
-    @parser.parse(input)
-  end
-  
-  def value(input)
-    parse(input).value
-  end
-  
-  def canonical(input)
-    parse(input).canonical
-  end
-  
-  def details(input)
-    parse(input).details
-  end
-  
-  def pos(input)
-    parse(input).pos
-  end
-  
-  def debug(input)
-    res = parse(input)
-    puts "<pre>"
-      if res
-        puts 'success!'
-        puts res.inspect
-      else
-        puts input
-        val = @parser.failure_reason.to_s.match(/column [0-9]*/).to_s.gsub(/column /,'').to_i
-        print ("-" * (val - 1))
-        print "^   Computer says 'no'!\n"
-        puts @parser.failure_reason
-        puts @parser.to_yaml
-      end
-    puts "</pre>"
+   set_parser(ScientificNameDirtyParser.new)
   end
   
+
   it 'should parse clean names' do
     parse("Betula verucosa (L.) Bar. 1899").should_not be_nil
   end
@@ -113,4 +71,20 @@ describe ScientificNameDirty do
     pos(sn).should == {0=>["genus", 8], 9=>["species", 18], 19=>["author_word", 24], 26=>["year", 30], 32=>["year", 36]}
   end
   
+  it "should parse new stuff" do
+    sn = 'Zoropsis (TaKeoa) nishimurai Yaginuma, 1971' #skipping for now
+    sn = 'Campylobacter pyloridis Marshall et al.1985.'
+    details(sn).should == {:genus=>{:epitheton=>"Campylobacter"}, :species=>{:epitheton=>"pyloridis", :authorship=>"Marshall et al.1985.", :basionymAuthorTeam=>{:authorTeam=>"Marshall et al.", :author=>["Marshall et al."], :year=>"1985"}}}
+    sn = 'Staphylococcus hyicus chromogenes Devriese et al. 1978 (Approved Lists 1980).'
+    details(sn).should == {:genus=>{:epitheton=>"Staphylococcus"}, :species=>{:epitheton=>"hyicus"}, :infraspecies=>{:epitheton=>"chromogenes", :rank=>"n/a", :authorship=>"Devriese et al. 1978", :basionymAuthorTeam=>{:authorTeam=>"Devriese et al.", :author=>["Devriese et al."], :year=>"1978"}}}
+    sn = 'Kitasatospora corrig. griseola Takahashi et al. 1985.'
+    details(sn).should == {:genus=>{:epitheton=>"Kitasatospora"}, :species=>{:epitheton=>"griseola", :authorship=>"Takahashi et al. 1985.", :basionymAuthorTeam=>{:authorTeam=>"Takahashi et al.", :author=>["Takahashi et al."], :year=>"1985"}}}
+    sn = 'Beijerinckia derxii venezuelae corrig. Thompson and Skerman, 1981'
+    details(sn).should == {:genus=>{:epitheton=>"Beijerinckia"}, :species=>{:epitheton=>"derxii"}, :infraspecies=>{:epitheton=>"venezuelae", :rank=>"n/a", :authorship=>"Thompson and Skerman, 1981", :basionymAuthorTeam=>{:authorTeam=>"Thompson and Skerman", :author=>["Thompson", "Skerman"], :year=>"1981"}}}
+    details('Streptomyces parvisporogenes ignotus 1960').should == {:genus=>{:epitheton=>"Streptomyces"}, :species=>{:epitheton=>"parvisporogenes"}, :infraspecies=>{:epitheton=>"ignotus", :rank=>"n/a", :year=>"1960"}}
+    details('Oscillaria caviae Simons 1920, according to Simons 1922').should == {:genus=>{:epitheton=>"Oscillaria"}, :species=>{:epitheton=>"caviae", :authorship=>"Simons 1920", :basionymAuthorTeam=>{:authorTeam=>"Simons", :author=>["Simons"], :year=>"1920"}}}
+    sn = 'Bacterium monocytogenes hominis"" Nyfeldt 1932'
+    details(sn).should == {:genus=>{:epitheton=>"Bacterium"}, :species=>{:epitheton=>"monocytogenes"}, :infraspecies=>{:epitheton=>"hominis", :rank=>"n/a"}}
+  end
+  
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: dimus-biodiversity
 version: !ruby/object:Gem::Version 
-  version: 0.5.1
+  version: 0.5.2
 platform: ruby
 authors: 
 - Dmitry Mozzherin