biodiversity 0.6.1 → 0.6.3

data/VERSION

@@ -1 +1 @@
-0.6.1
+0.6.3

data/bin/nnparse

@@ -1,6 +1,12 @@
 #!/usr/bin/env ruby
-require 'rubygems'
-gem_name = RUBY_VERSION.split('.')[0..1].join('').to_i > 18 ? 'biodiversity19' : 'biodiversity'
+# encoding: utf-8
+ruby_min_version = RUBY_VERSION.split(".")[0..1].join('').to_i
+if ruby_min_version < 19
+  require 'rubygems'
+  gem_name = 'biodiversity'
+else
+  gem_name = 'biodiversity19'
+end
 gem gem_name rescue nil
 
 $LOAD_PATH.unshift(File.expand_path(File.dirname(__FILE__) + "/../lib"))
@@ -19,13 +25,13 @@ end
 input = ARGV[0]
 output = ARGV[1] || 'parsed.json'
 
-ruby_min_version = RUBY_VERSION.split(".")[0..1].join('').to_i
 
 p = ScientificNameParser.new
 o = open(output, 'w')
 count = 0
 puts 'Parsing...'
-IO.foreach(input) do |line|
+f = ruby_min_version < 19 ? open(input) : open(input, 'r:utf-8')
+f.each do |line|
   count += 1
   puts("%s lines parsed" % count) if count % 10000 == 0
   name = line.gsub(/^[\d]*\s*/, '').strip

data/spec/parser/scientific_name_clean.spec.rb

@@ -516,6 +516,8 @@ describe ScientificNameClean do
     details(sn).should == [{:genus=>{:string=>"Flexibacter"}, :species=>{:string=>"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=>{:string=>"Flexibacter"}, :species=>{:string=>"elegans", :authorship=>"Soriano 1945, non Lewin 1969", :basionymAuthorTeam=>{:authorTeam=>"Soriano", :author=>["Soriano"], :year=>"1945"}}}]
+    sn = 'Schottera nicaeënsis (J.V. Lamouroux ex Duby) Guiry & Hollenberg'
+    details(sn).should == [{:genus=>{:string=>"Schottera"}, :species=>{:string=>"nicaeënsis", :authorship=>"(J.V. Lamouroux ex Duby) Guiry & Hollenberg", :combinationAuthorTeam=>{:authorTeam=>"Guiry & Hollenberg", :author=>["Guiry", "Hollenberg"]}, :basionymAuthorTeam=>{:authorTeam=>"J.V. Lamouroux", :author=>["J.V. Lamouroux"], :exAuthorTeam=>{:authorTeam=>"Duby", :author=>["Duby"]}}}}]
   end
   
   # Combination genus names should be merged without dash or capital letter

data/spec/parser/test_data.txt

@@ -53,6 +53,7 @@ Platypus bicaudatulus Schedl, 1935h|{"scientificName":{"parsed":true,"parser_run
 Pseudocercospora dendrobii U. Braun & Crous|{"scientificName":{"parsed":true,"parser_run":1,"verbatim":"Pseudocercospora dendrobii U. Braun & Crous","normalized":"Pseudocercospora dendrobii U. Braun et Crous","canonical":"Pseudocercospora dendrobii","hybrid":false,"details":[{"genus":{"string":"Pseudocercospora"},"species":{"string":"dendrobii","authorship":"U. Braun & Crous","basionymAuthorTeam":{"authorTeam":"U. Braun & Crous","author":["U. Braun","Crous"]}}}],"positions":{"0":["genus",16],"17":["species",26],"27":["author_word",29],"30":["author_word",35],"38":["author_word",43]}}}
 Pseudocercospora dendrobii U. Braun et Crous|{"scientificName":{"parsed":true,"parser_run":1,"verbatim":"Pseudocercospora dendrobii U. Braun et Crous","normalized":"Pseudocercospora dendrobii U. Braun et Crous","canonical":"Pseudocercospora dendrobii","hybrid":false,"details":[{"genus":{"string":"Pseudocercospora"},"species":{"string":"dendrobii","authorship":"U. Braun et Crous","basionymAuthorTeam":{"authorTeam":"U. Braun et Crous","author":["U. Braun","Crous"]}}}],"positions":{"0":["genus",16],"17":["species",26],"27":["author_word",29],"30":["author_word",35],"39":["author_word",44]}}}
 Pseudocercospora dendrobii U. Braun and Crous|{"scientificName":{"parsed":true,"parser_run":1,"verbatim":"Pseudocercospora dendrobii U. Braun and Crous","normalized":"Pseudocercospora dendrobii U. Braun et Crous","canonical":"Pseudocercospora dendrobii","hybrid":false,"details":[{"genus":{"string":"Pseudocercospora"},"species":{"string":"dendrobii","authorship":"U. Braun and Crous","basionymAuthorTeam":{"authorTeam":"U. Braun and Crous","author":["U. Braun","Crous"]}}}],"positions":{"0":["genus",16],"17":["species",26],"27":["author_word",29],"30":["author_word",35],"40":["author_word",45]}}}
+Schottera nicaeënsis (J.V. Lamouroux ex Duby) Guiry & Hollenberg|{"scientificName":{"parsed":true, "verbatim":"Schottera nicaeënsis (J.V. Lamouroux ex Duby) Guiry & Hollenberg", "normalized":"Schottera nicaeënsis (J.V. Lamouroux ex Duby) Guiry et Hollenberg", "canonical":"Schottera nicaeënsis", "hybrid":false, "details":[{"genus":{"string":"Schottera"}, "species":{"string":"nicaeënsis", "authorship":"(J.V. Lamouroux ex Duby) Guiry & Hollenberg", "combinationAuthorTeam":{"authorTeam":"Guiry & Hollenberg", "author":["Guiry", "Hollenberg"]}, "basionymAuthorTeam":{"authorTeam":"J.V. Lamouroux", "author":["J.V. Lamouroux"], "exAuthorTeam":{"authorTeam":"Duby", "author":["Duby"]}}}}], "parser_run":1, "positions":{"0":["genus", 9], "10":["species", 20], "22":["author_word", 26], "27":["author_word", 36], "40":["author_word", 44], "46":["author_word", 51], "54":["author_word", 64]}}}
 
 #binomial with several authors and a year
 Pseudocercospora dendrobii U. Braun & Crous 2003|{"scientificName":{"parsed":true,"parser_run":1,"verbatim":"Pseudocercospora dendrobii U. Braun & Crous 2003","normalized":"Pseudocercospora dendrobii U. Braun et Crous 2003","canonical":"Pseudocercospora dendrobii","hybrid":false,"details":[{"genus":{"string":"Pseudocercospora"},"species":{"string":"dendrobii","authorship":"U. Braun & Crous 2003","basionymAuthorTeam":{"authorTeam":"U. Braun & Crous","author":["U. Braun","Crous"],"year":"2003"}}}],"positions":{"0":["genus",16],"17":["species",26],"27":["author_word",29],"30":["author_word",35],"38":["author_word",43],"44":["year",48]}}}

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: biodiversity
 version: !ruby/object:Gem::Version 
-  hash: 5
+  hash: 1
   prerelease: false
   segments: 
   - 0
   - 6
-  - 1
-  version: 0.6.1
+  - 3
+  version: 0.6.3
 platform: ruby
 authors: 
 - Dmitry Mozzherin
@@ -15,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-05-20 00:00:00 -04:00
+date: 2010-05-26 00:00:00 -04:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency