biodiversity 3.5.0 → 3.5.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 39237a14ed21c0cdf5ea9f7687240b3af4477e47
-  data.tar.gz: 7d1d770ce5748b48f86cb0bebadb89307ef0a3bc
+  metadata.gz: 832687083196a98a5923318b0d778fde31c852d0
+  data.tar.gz: 2dc45e6f8d201ee8c9abcd5c2cbb41ad3e0c20fa
 SHA512:
-  metadata.gz: d2eb0fa48343641f9c05ec8320834ddef60dc5794c058b91d813eed1db40991c0c865496fe6c39c485aabbe8d200a20aac71d67e3be58535e5972bf8b4d04697
-  data.tar.gz: eda4fb5b701f8c1606285d2e90d64a82f59958e1cf0a56247be8e2f2cc1c30b6ff04ace786ba6e0e6e7e36ce4ce3b113ed1e83ad28fdbdc02df6a11c39938841
+  metadata.gz: af478fe58904ec048ef4b67f0671a1c7c8dc59144303af60129b117ce7ec775a1fa73ceb154f3e10f2a2d05cac173b72ff539231acb0b6cc7dab2a4baacc230f
+  data.tar.gz: e6494923af734096b435e27b50ef8239477d58110c6b9f59a0f36324cc7110b6461eb81f0218c00766d6919a3d7eaa7f03dfc688195c0456ea50a5141e5d44ac

data/CHANGELOG

@@ -1,3 +1,5 @@
+3.5.1 -- allow comma before 'and' or '&' in authorship
+
 3.5.0 -- add the tail cut by preprocessing to the results. The tail usually
          contains annotations about the name.
 

data/lib/biodiversity/parser/scientific_name_clean.rb

@@ -5567,7 +5567,8 @@ module ScientificNameClean
   module AuthorSeparator0
     def apply(a,b)
       sep = text_value.strip
-      sep = " &" if ["&", "&","and","et"].include? sep
+      sep = " &" if ["&", "&", "and", "et",
+                     ",&", ", &", ",and", ", and"].include? sep
       sep = " apud" if sep == "apud"
       a.value + sep + " " + b.value
     end
@@ -5619,11 +5620,11 @@ module ScientificNameClean
         r0.extend(AuthorSeparator0)
         r0.extend(AuthorSeparator0)
       else
-        if (match_len = has_terminal?(",", false, index))
-          r3 = true
+        if (match_len = has_terminal?(",&", false, index))
+          r3 = instantiate_node(SyntaxNode,input, index...(index + match_len))
           @index += match_len
         else
-          terminal_parse_failure('","')
+          terminal_parse_failure('",&"')
           r3 = nil
         end
         if r3
@@ -5632,11 +5633,11 @@ module ScientificNameClean
           r0.extend(AuthorSeparator0)
           r0.extend(AuthorSeparator0)
         else
-          if (match_len = has_terminal?("apud", false, index))
+          if (match_len = has_terminal?(", &", false, index))
             r4 = instantiate_node(SyntaxNode,input, index...(index + match_len))
             @index += match_len
           else
-            terminal_parse_failure('"apud"')
+            terminal_parse_failure('", &"')
             r4 = nil
           end
           if r4
@@ -5645,11 +5646,11 @@ module ScientificNameClean
             r0.extend(AuthorSeparator0)
             r0.extend(AuthorSeparator0)
           else
-            if (match_len = has_terminal?("and", false, index))
+            if (match_len = has_terminal?(",and", false, index))
               r5 = instantiate_node(SyntaxNode,input, index...(index + match_len))
               @index += match_len
             else
-              terminal_parse_failure('"and"')
+              terminal_parse_failure('",and"')
               r5 = nil
             end
             if r5
@@ -5658,11 +5659,11 @@ module ScientificNameClean
               r0.extend(AuthorSeparator0)
               r0.extend(AuthorSeparator0)
             else
-              if (match_len = has_terminal?("et", false, index))
+              if (match_len = has_terminal?(", and", false, index))
                 r6 = instantiate_node(SyntaxNode,input, index...(index + match_len))
                 @index += match_len
               else
-                terminal_parse_failure('"et"')
+                terminal_parse_failure('", and"')
                 r6 = nil
               end
               if r6
@@ -5671,8 +5672,64 @@ module ScientificNameClean
                 r0.extend(AuthorSeparator0)
                 r0.extend(AuthorSeparator0)
               else
-                @index = i0
-                r0 = nil
+                if (match_len = has_terminal?("apud", false, index))
+                  r7 = instantiate_node(SyntaxNode,input, index...(index + match_len))
+                  @index += match_len
+                else
+                  terminal_parse_failure('"apud"')
+                  r7 = nil
+                end
+                if r7
+                  r7 = SyntaxNode.new(input, (index-1)...index) if r7 == true
+                  r0 = r7
+                  r0.extend(AuthorSeparator0)
+                  r0.extend(AuthorSeparator0)
+                else
+                  if (match_len = has_terminal?("and", false, index))
+                    r8 = instantiate_node(SyntaxNode,input, index...(index + match_len))
+                    @index += match_len
+                  else
+                    terminal_parse_failure('"and"')
+                    r8 = nil
+                  end
+                  if r8
+                    r8 = SyntaxNode.new(input, (index-1)...index) if r8 == true
+                    r0 = r8
+                    r0.extend(AuthorSeparator0)
+                    r0.extend(AuthorSeparator0)
+                  else
+                    if (match_len = has_terminal?("et", false, index))
+                      r9 = instantiate_node(SyntaxNode,input, index...(index + match_len))
+                      @index += match_len
+                    else
+                      terminal_parse_failure('"et"')
+                      r9 = nil
+                    end
+                    if r9
+                      r9 = SyntaxNode.new(input, (index-1)...index) if r9 == true
+                      r0 = r9
+                      r0.extend(AuthorSeparator0)
+                      r0.extend(AuthorSeparator0)
+                    else
+                      if (match_len = has_terminal?(",", false, index))
+                        r10 = true
+                        @index += match_len
+                      else
+                        terminal_parse_failure('","')
+                        r10 = nil
+                      end
+                      if r10
+                        r10 = SyntaxNode.new(input, (index-1)...index) if r10 == true
+                        r0 = r10
+                        r0.extend(AuthorSeparator0)
+                        r0.extend(AuthorSeparator0)
+                      else
+                        @index = i0
+                        r0 = nil
+                      end
+                    end
+                  end
+                end
               end
             end
           end

data/lib/biodiversity/parser/scientific_name_clean.treetop

@@ -1000,7 +1000,7 @@ grammar ScientificNameClean
     }
   end
 
-  rule emend_authorship  
+  rule emend_authorship
     emend space b:simple_authorship {
       def value
         " emend. " + b.value
@@ -1125,10 +1125,11 @@ grammar ScientificNameClean
   end
 
   rule author_separator
-    ("&"/"&"/","/"apud"/"and"/"et") {
+    ("&"/"&"/",&"/", &"/",and"/", and"/"apud"/"and"/"et"/",") {
       def apply(a,b)
         sep = text_value.strip
-        sep = " &" if ["&", "&","and","et"].include? sep
+        sep = " &" if ["&", "&", "and", "et",
+                       ",&", ", &", ",and", ", and"].include? sep
         sep = " apud" if sep == "apud"
         a.value + sep + " " + b.value
       end

data/lib/biodiversity/version.rb

@@ -1,3 +1,3 @@
 module Biodiversity
-  VERSION = "3.5.0"
+  VERSION = "3.5.1"
 end

data/spec/files/test_data.txt

@@ -80,6 +80,10 @@ M. alpium|{"scientificName":{"id":"9001ffb5-eac2-5bb4-8f78-d7b7e3e02bd8", "parse
 Mo. alpium (Osbeck, 1778)|{"scientificName":{"id":"1e9437b7-bf45-5b12-8da0-8966c6ea1c5c", "parsed":true, "parser_version":"test_version", "verbatim":"Mo. alpium (Osbeck, 1778)", "normalized":"Mo. alpium (Osbeck 1778)", "canonical":"Mo. alpium", "hybrid":false, "details":[{"genus":{"string":"Mo."}, "species":{"string":"alpium", "authorship":"(Osbeck, 1778)", "basionymAuthorTeam":{"authorTeam":"Osbeck", "author":["Osbeck"], "year":"1778"}}}], "parser_run":1, "positions":{"0":["genus", 4], "4":["species", 10], "12":["author_word", 18], "20":["year", 24]}}}
 Mom.alpium (Osbeck, 1778)|{"scientificName":{"id":"f1452bcf-b779-5d98-bfc8-56455105e3f5", "parsed":true, "parser_version":"test_version", "verbatim":"Mom.alpium (Osbeck, 1778)", "normalized":"Mom. alpium (Osbeck 1778)", "canonical":"Mom. alpium", "hybrid":false, "details":[{"genus":{"string":"Mom."}, "species":{"string":"alpium", "authorship":"(Osbeck, 1778)", "basionymAuthorTeam":{"authorTeam":"Osbeck", "author":["Osbeck"], "year":"1778"}}}], "parser_run":1, "positions":{"0":["genus", 4], "4":["species", 10], "12":["author_word", 18], "20":["year", 24]}}}
 
+Cratomastotermitidae Engel, Grimaldi, and Krishna, 2009|{"scientificName":{"canonical":"Cratomastotermitidae", "details":[{"uninomial":{"authorship":"Engel, Grimaldi, and Krishna, 2009", "basionymAuthorTeam":{"author":["Engel", "Grimaldi", "Krishna"], "authorTeam":"Engel, Grimaldi, and Krishna", "year":"2009"}, "string":"Cratomastotermitidae"}}], "hybrid":false, "id":"805c4276-120e-5bef-81fd-e969ab0bccca", "normalized":"Cratomastotermitidae Engel, Grimaldi & Krishna 2009", "parsed":true, "parser_run":1, "parser_version":"test_version", "positions":{"0":["uninomial", 20], "21":["author_word", 26], "28":["author_word", 36], "42":["author_word", 49], "51":["year", 55]}, "verbatim":"Cratomastotermitidae Engel, Grimaldi, and Krishna, 2009"}}
+
+Cratomastotermitidae Engel, Grimaldi,& Krishna, 2009|{"scientificName":{"canonical":"Cratomastotermitidae", "details":[{"uninomial":{"authorship":"Engel, Grimaldi,& Krishna, 2009", "basionymAuthorTeam":{"author":["Engel", "Grimaldi", "Krishna"], "authorTeam":"Engel, Grimaldi,& Krishna", "year":"2009"}, "string":"Cratomastotermitidae"}}], "hybrid":false, "id":"112f3fc9-c6f5-5369-9248-9fd4441bd4f1", "normalized":"Cratomastotermitidae Engel, Grimaldi & Krishna 2009", "parsed":true, "parser_run":1, "parser_version":"test_version", "positions":{"0":["uninomial", 20], "21":["author_word", 26], "28":["author_word", 36], "39":["author_word", 46], "48":["year", 52]}, "verbatim":"Cratomastotermitidae Engel, Grimaldi,& Krishna, 2009"}}
+
 #binomial with apostrophe in species epithet
 Junellia o'donelli Moldenke, 1946|{"scientificName":{"id":"e39a2d98-6ab2-5fb3-9aae-c48aa86c6026", "parsed":true, "parser_version":"test_version", "verbatim":"Junellia o'donelli Moldenke, 1946", "normalized":"Junellia odonelli Moldenke 1946", "canonical":"Junellia odonelli", "hybrid":false, "details":[{"genus":{"string":"Junellia"}, "species":{"string":"odonelli", "authorship":"Moldenke, 1946", "basionymAuthorTeam":{"authorTeam":"Moldenke", "author":["Moldenke"], "year":"1946"}}}], "parser_run":1, "positions":{"0":["genus", 8], "9":["species", 18], "19":["author_word", 27], "29":["year", 33]}}}
 Trophon d'orbignyi Carcelles, 1946|{"scientificName":{"id":"935d4414-05d4-5c16-be30-466f6144b666", "parsed":true, "parser_version":"test_version", "verbatim":"Trophon d'orbignyi Carcelles, 1946", "normalized":"Trophon dorbignyi Carcelles 1946", "canonical":"Trophon dorbignyi", "hybrid":false, "details":[{"genus":{"string":"Trophon"}, "species":{"string":"dorbignyi", "authorship":"Carcelles, 1946", "basionymAuthorTeam":{"authorTeam":"Carcelles", "author":["Carcelles"], "year":"1946"}}}], "parser_run":1, "positions":{"0":["genus", 7], "8":["species", 18], "19":["author_word", 28], "30":["year", 34]}}}

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: biodiversity
 version: !ruby/object:Gem::Version
-  version: 3.5.0
+  version: 3.5.1
 platform: ruby
 authors:
 - Dmitry Mozzherin
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2018-04-16 00:00:00.000000000 Z
+date: 2018-09-14 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: treetop