dwc_agent 0.4.0 → 1.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 988e79daa81edb97377cdaf65ad2e6aa79b6e524cd3c9ab940d7b4cc1c1c9885
-  data.tar.gz: 7b6969b8da5e772858e68313e24954bd352c78f62e552d0b5b16a328b3d20e0c
+  metadata.gz: d56e8cb49d4857e324dbdda7d6cfa31e9a7a3079c40a12659ff6077d13308613
+  data.tar.gz: 7652bc849261aa4179a99e64a725a0f82d0b65b4d2e2885042ad35521887e705
 SHA512:
-  metadata.gz: 5794e0e8b1d779a72e6d8ba599e85b196f870f16494e663830f3c265c3ee39101bf729e6670f0f8f7d5762dadd438b3843aab5d7bcd450bd1212d4c95f0c6183
-  data.tar.gz: 91fca40528d0f20a698ea66602de134ba6c442e6ab7e51118003ba0afdf49eca1247caa33fcf025f9cced537dc6f2ecaad8a440f07d7d34232d1326bb2eaa213
+  metadata.gz: 99d384a56c180c8c5db64c5cc8cf072da272b427830875783a22d67279cd070af3bd0bd2b881e4a20018bdeba774d27e8adb8ae048668c28452171017b72c70b
+  data.tar.gz: edc59989136745e0b603039d08420b368ef92e95a0ab17b48bcabf439472c1252ec763f4f9c7a398b735995b340f897dc0f3d85a45c2ea1c9ccec643f050e904

data/lib/dwc_agent/cleaner.rb

@@ -16,9 +16,9 @@ module DwcAgent
     # @param parsed_namae [Object] the namae object
     # @return [Hash] the given, family hash
     def clean(parsed_namae)
-      blank_name = { given: nil, family: nil }
+      blank_name = { given: nil, family: nil, particle: nil }
 
-      if parsed_namae.family && FAMILY_BLACKLIST.include?(parsed_namae.family)
+      if parsed_namae.family && FAMILY_BLACKLIST.any?{ |s| s.casecmp(parsed_namae.family) == 0 }
         return blank_name
       end
 
@@ -99,6 +99,10 @@ module DwcAgent
         particle = nil
       end
 
+      if !particle.nil? && particle.include?(".")
+        particle = nil
+      end
+
       if !family.nil? && (family == family.upcase || family == family.downcase)
         family = NameCase(family)
       end
@@ -111,7 +115,11 @@ module DwcAgent
         return blank_name
       end
 
-      { given: given, family: family }
+      if !family.nil? && FAMILY_BLACKLIST.any?{ |s| s.casecmp(family) == 0 }
+        return blank_name
+      end
+
+      { given: given, family: family, particle: particle }
     end
 
   end

data/lib/dwc_agent/constants.rb

@@ -20,7 +20,6 @@ module DwcAgent
     \b[,;]?\s*(?i:person\s*string)\b|
     \b[,;]?\s*(?i:colls)\.(\b|\z)|
     \b[,;]?\s*(?i:colln?)[:.]?(\b|\z)|
-    (?i:no\s+(data|disponible))|
     \b[,;]?\s*(?i:stet)[,!]?\s*\d*\z|
     [,;]?\s*\d+[-/\s+](?i:\d+|Jan|Feb|Mar|Apr|
       May|Jun|Jul|Aug|Sept?|
@@ -78,10 +77,11 @@ module DwcAgent
     \b\s*\(?(?i:(fe)?male)\)?\s*\b|
     \b(?i:to\s+(sub)?spp?)\.?|
     (?i:nom\.?\s+rev\.?)|
-    FNA|DAO|HUH|FDNMB|
+    FNA|DAO|HUH|FDNMB|MNHN|PNI|USNM|
     AFSC\/POLISH\s+SORTING\s+CTR\.?|
     (?i:university|museum|exhibits?)|
     (?i:uqam)|
+    (?i:sem\s+(colec?tor|data))|
     \b[,;]\s+\d+\z|
     ["!@?]|
     [,]?\d+|
@@ -147,7 +147,8 @@ module DwcAgent
     '{' => '',
     '}' => '',
     '@' => '',
-    '%' => ''
+    '%' => '',
+    '\\' => ''
   }
 
   PHRASE_SUBS = {
@@ -158,12 +159,17 @@ module DwcAgent
   }
 
   COMPLEX_SEPARATORS = %r{
-    ^([A-Za-z]{4,},\s+(?:[A-Z]\.\s*){1,})\s+([A-Za-z]{4,},\s+(?:[A-Z]\.\s*){1,})$
+    ^(\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)|
+    (?i:administra(d|t)or)|
     (?i:anon)|
     (?i:australian?)|
     (?i:average)|
@@ -172,9 +178,10 @@ module DwcAgent
     (?i:biolog|botan|zoo|ecolog|mycol|(in)?vertebrate|fisheries|genetic|animal|mushroom|wildlife|plumage|flower|agriculture)|
     (?i:bris?tish|canadi?an?|chinese|arctic|japan|russian|north\s+america)|
     (?i:carex|salix)|
-    (?:catalog)|
+    (?i:catalog(ue)?)|
     (?i:herbarium|herbier|collection|collected|publication|specimen|species|describe|an(a|o)morph|isolated|recorded|inspection|define|status|lighthouse)|
     \b\s*(?i:help)\s*\b|
+    (?i:data\s+not\s+captured)|
     (?i:description|drawing|identification|remark|original|illustration|checklist|intermedia|measurement|indisting|series|imperfect)|
     (?i:desconocido)|
     (?i:exc?s?icc?at(a|i))|
@@ -192,6 +199,9 @@ module DwcAgent
     (?i:univ\.)|
     (?i:graduate|student|estudi?antes?|labo\.|storekeep|supervisor|superint|rcmp|coordinator|minority|fishermen|police|taxonomist|consultant|participante?s?|team|(é|e)quipe|memb(er|re)|crew|group|staff|personnel|family|captain|friends|assistant|worker)|
     (?i:non\s+pr(é|e)cis(é|e))|
+    (?i:no\s+(agent)?\s?(data|disponible)(\s+available)?)|
+    (?i:not?\s+(entered|stated))|
+    (?i:nomenclatur(e|al)\s+adjustment)|
     (?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)|
@@ -219,7 +229,12 @@ module DwcAgent
   FAMILY_BLACKLIST = [
     "der",
     "van",
-    "von"
+    "von",
+    "the",
+    "of",
+    "curators",
+    "nomenclatural",
+    "adjustment"
   ]
 
   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

@@ -20,7 +20,8 @@ module DwcAgent
       @char_subs_regex = Regexp.new [CHAR_SUBS.keys.join('\\')].to_s
       @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{([A-Z]{1}\.)([[:alpha:]]{2,})}.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.
@@ -34,6 +35,7 @@ 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

@@ -1,8 +1,8 @@
 module DwcAgent
   class Version
 
-    MAJOR = 0
-    MINOR = 4
+    MAJOR = 1
+    MINOR = 1
     PATCH = 0
     BUILD = nil
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: dwc_agent
 version: !ruby/object:Gem::Version
-  version: 0.4.0
+  version: 1.1.0
 platform: ruby
 authors:
 - David P. Shorthouse
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-09-07 00:00:00.000000000 Z
+date: 2019-10-11 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: namae