dwc_agent 1.1.0 → 1.4.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: d56e8cb49d4857e324dbdda7d6cfa31e9a7a3079c40a12659ff6077d13308613
-  data.tar.gz: 7652bc849261aa4179a99e64a725a0f82d0b65b4d2e2885042ad35521887e705
+  metadata.gz: 75ac9f77b1b4d5761881cafca45e66c2745eb47bb501ad9fab84cad6324b41ee
+  data.tar.gz: 16057726f330cb033fe7e3af0af58dd4fbb2f6f9817a56c3e86c88da1fbccf61
 SHA512:
-  metadata.gz: 99d384a56c180c8c5db64c5cc8cf072da272b427830875783a22d67279cd070af3bd0bd2b881e4a20018bdeba774d27e8adb8ae048668c28452171017b72c70b
-  data.tar.gz: edc59989136745e0b603039d08420b368ef92e95a0ab17b48bcabf439472c1252ec763f4f9c7a398b735995b340f897dc0f3d85a45c2ea1c9ccec643f050e904
+  metadata.gz: f628d3bfec046b54338eb08ec18cc5352dff7560e5e87e2c949378e66bcb44df84af637ea7c8b4d559504945b5682b6f019f2bde56d31552a1c4537902b2a71e
+  data.tar.gz: 444b15c0d391e12d8ccd6823f8ae97688d9b20ed075323795e22acd7834add4e5068fcc2d4fc3729e36aa3bdc11f29f4a6f8af8c7fc1cfc34d1f294159212328

data/bin/dwcagent

@@ -5,7 +5,7 @@ require 'dwc_agent'
 require 'json'
 
 names = []
-DwcAgent.parse(ARGV[0]).each do |r|
+DwcAgent.parse(ARGV[0].dup).each do |r|
   name = DwcAgent.clean(r)
   if !name[:family].nil? && name[:family].length >= 2
     names << name

data/bin/dwcagent-similarity

@@ -3,4 +3,4 @@
 
 require 'dwc_agent'
 
-puts DwcAgent.similarity_score(ARGV[0],ARGV[1])
+puts DwcAgent.similarity_score(ARGV[0].dup,ARGV[1].dup)

data/lib/dwc_agent/cleaner.rb

@@ -22,7 +22,7 @@ module DwcAgent
         return blank_name
       end
 
-      if parsed_namae.family && parsed_namae.family.length < 2 && parsed_namae.family.count('.') == 0
+      if parsed_namae.given && GIVEN_BLACKLIST.any?{ |s| s.casecmp(parsed_namae.given) == 0 }
         return blank_name
       end
 
@@ -42,9 +42,9 @@ module DwcAgent
         return blank_name
       end
 
-      if parsed_namae.given && 
-         parsed_namae.family && 
-         parsed_namae.family.count(".") > 0 && 
+      if parsed_namae.given &&
+         parsed_namae.family &&
+         parsed_namae.family.count(".") > 0 &&
          parsed_namae.family.length - parsed_namae.family.count(".") <= 3
           given = parsed_namae.given
           family = parsed_namae.family
@@ -52,9 +52,9 @@ module DwcAgent
           parsed_namae.given = family
       end
 
-      if parsed_namae.given && 
-         parsed_namae.family && 
-         parsed_namae.family.length <=3 && 
+      if parsed_namae.given &&
+         parsed_namae.family &&
+         parsed_namae.family.length <=3 &&
          parsed_namae.family == parsed_namae.family.upcase &&
          parsed_namae.given[-1] != "."
           given = parsed_namae.given
@@ -63,9 +63,9 @@ module DwcAgent
           parsed_namae.given = family
       end
 
-      if parsed_namae.given && 
-        (parsed_namae.given == parsed_namae.given.upcase || 
-        parsed_namae.given == parsed_namae.given.downcase) && 
+      if parsed_namae.given &&
+        (parsed_namae.given == parsed_namae.given.upcase ||
+        parsed_namae.given == parsed_namae.given.downcase) &&
         !parsed_namae.given.include?(".") &&
         parsed_namae.given.tr(".","").length >= 4
           parsed_namae.given = NameCase(parsed_namae.given)
@@ -119,8 +119,12 @@ module DwcAgent
         return blank_name
       end
 
+      if !given.nil? && GIVEN_BLACKLIST.any?{ |s| s.casecmp(given) == 0 }
+        return blank_name
+      end
+
       { given: given, family: family, particle: particle }
     end
 
   end
-end
+end

data/lib/dwc_agent/constants.rb

@@ -162,10 +162,6 @@ module DwcAgent
     ^(\S{4,},\s+(?:\S\.\s*){1,})\s+(\S{4,},\s+(?:\S\.\s*){1,})$
   }x
 
-  CONTRACTED_LIST = %r{
-    ^(\S{1,}\.?)+\s+(?i:and|&)\s+(\S{1,}\.?)+\s*(.*)$
-  }x
-
   BLACKLIST = %r{
     (?i:abundant)|
     (?i:adult|juvenile)|
@@ -202,11 +198,13 @@ module DwcAgent
     (?i:no\s+(agent)?\s?(data|disponible)(\s+available)?)|
     (?i:not?\s+(entered|stated))|
     (?i:nomenclatur(e|al)\s+adjustment)|
+    (?i:not\s+available)|
     (?i:ontario|qu(e|é)bec|saskatchewan|new brunswick|sault|newfoundland|assurance|vancouver|u\.?s\.?s\.?r\.?)|
     (?i:recreation|culture)|
     (?i:shaped|dark|pale|areas|phase|spotting|interior|between|closer)|
     (?i:soci(e|é)t(y|é)|cent(er|re)|community|history|conservation|conference|assoc|class|commission|consortium|council|club|exposit|alliance|protective|circle)|
     (?i:commercial|company|control|product)|
+    (?i:sequence\s+data)|
     (?i:size|large|colou?r)\s+|
     (?i:skeleton)|
     (?i:survey|assessment|station|monitor|stn\.|index|project|bureau|engine|exchange|ex(c|k)ursi(e|o|ó)n?|exped\.?|exp(e|i)di(c|t)i(e|o|ó)n?|experiment|explora(d|t)|festival|generation|inventory|marine|service)|
@@ -217,6 +215,7 @@ module DwcAgent
     \s*(?i:too)\s+|\s*(?i:the)\s+|
     (?i:taxiderm(ies|y))|
     (?i:though)|
+    (?i:texas\s+instruments?)\s*?(for)?|
     (?:tropical)|
     (?i:toward|seen at)|
     (?i:unidentified|unspecified|unk?nown|unnamed|unread|unmistak|no agent)|
@@ -234,7 +233,18 @@ module DwcAgent
     "of",
     "curators",
     "nomenclatural",
-    "adjustment"
+    "adjustment",
+    "available",
+    "data",
+    "orig",
+    "science",
+    "catalogue",
+    "entered"
+  ]
+
+  GIVEN_BLACKLIST = [
+    "not any",
+    "has not"
   ]
 
   TITLE = /\s*\b(sir|count(ess)?|(gen|adm|col|maj|capt|cmdr|lt|sgt|cpl|pvt|prof|dr|md|ph\.?d|rev|docteur|mme|abbé|ptre)\.?|frère|frere|père|pere|professor|esq\.?)(\s+|$)/i

data/lib/dwc_agent/parser.rb

@@ -21,7 +21,6 @@ module DwcAgent
       @phrase_subs_regex = Regexp.new (PHRASE_SUBS.keys.join('|')).to_s
       @complex_separators_regex = Regexp.new COMPLEX_SEPARATORS.to_s
       @add_separators_regex = Regexp.new %r{(\S{1}\.)([[:alpha:]]{2,})}.to_s
-      @contracted_list_regex = Regexp.new CONTRACTED_LIST.to_s
     end
       
     # Parses the passed-in string and returns a list of names.
@@ -35,7 +34,6 @@ module DwcAgent
       name.gsub!(@phrase_subs_regex, PHRASE_SUBS)
       name.gsub!(@add_separators_regex, '\1 \2')
       name.gsub!(@complex_separators_regex, '\1 | \2')
-      name.gsub!(@contracted_list_regex, '\1 \3 | \2 \3')
       name.gsub!(@residual_terminators_regex, '')
       name.squeeze!(' ')
       name.strip!

data/lib/dwc_agent/version.rb

@@ -2,8 +2,8 @@ module DwcAgent
   class Version
 
     MAJOR = 1
-    MINOR = 1
-    PATCH = 0
+    MINOR = 4
+    PATCH = 2
     BUILD = nil
 
     def self.version

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: dwc_agent
 version: !ruby/object:Gem::Version
-  version: 1.1.0
+  version: 1.4.2
 platform: ruby
 authors:
 - David P. Shorthouse
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-10-11 00:00:00.000000000 Z
+date: 2019-12-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: namae
@@ -119,7 +119,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.0.3
+rubygems_version: 3.0.6
 signing_key: 
 specification_version: 4
 summary: Parse Darwin Core agent terms such as recordedBy and identifiedBy