dwc_agent 0.4.1 → 1.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 7bf17473ddabfee14ba9e7a8d451486ef8ebb058c9d1d0214b0ed9cbae996e48
-  data.tar.gz: cb404e507a9f9de35a5ddeebf94a6bdd4e0ad7b245e08909b1d71f50bd7595da
+  metadata.gz: dcdfb4922038d07715bd064b74a516203bcce44ef824683ac9766586764bb98f
+  data.tar.gz: d9113f199abc420d7c608bfbead6e70f1f6192437a82e43541f77eb187752c2a
 SHA512:
-  metadata.gz: a25e5863c6fa384604e399815ec1c4c9befadaf1ebdefb4ef52abdb1e9a296ace16718bab6020a32c3f1ed9c0728c46e76fdfa8c5b3133e98e48b0478c155858
-  data.tar.gz: 1243c70e9d479ca15dfc931e823939216873f536c3441f77e90bc9461e639e2d408e335d31418e0e31616686d24914ceffd43ae97fe9388f2ccfab443b102f1f
+  metadata.gz: 403049856dbfbc83c984b5175615718c738b46312ecced4d3b5853fb4584a080d1e868492ee476187c8cf7238481845c0f77bd181e6ceaeebc22bae8b7aef127
+  data.tar.gz: 01e8bc4beb84140a80763c6c0dc8ff6ee6f967e8382cbacf04a230aa3405f2394ec1a90570471e113d90264c1e202c5452428858eb5aff6bc499f4e795dc3158

data/lib/dwc_agent/cleaner.rb

@@ -16,13 +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)
-        return blank_name
-      end
-
-      if parsed_namae.family && parsed_namae.family.length < 2 && parsed_namae.family.count('.') == 0
+      if parsed_namae.family && FAMILY_BLACKLIST.any?{ |s| s.casecmp(parsed_namae.family) == 0 }
         return blank_name
       end
 
@@ -99,6 +95,10 @@ 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 +111,11 @@ 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 +159,23 @@ 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
 
+# Was used in 1.1.0 but it sunk performance so threw it back to a WIP
+#
+# @contracted_list_regex = Regexp.new CONTRACTED_LIST.to_s
+#
+# name.gsub!(@contracted_list_regex, '\1 \3 | \2 \3')
+#
+#  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 +184,10 @@ 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))|
@@ -192,7 +205,9 @@ 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:not?\s+stated)|
+    (?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)|
@@ -220,7 +235,12 @@ 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 +20,7 @@ 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
     end
       
     # Parses the passed-in string and returns a list of names.

data/lib/dwc_agent/version.rb

@@ -1,9 +1,9 @@
 module DwcAgent
   class Version
 
-    MAJOR = 0
-    MINOR = 4
-    PATCH = 1
+    MAJOR = 1
+    MINOR = 2
+    PATCH = 0
     BUILD = nil
 
     def self.version

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: dwc_agent
 version: !ruby/object:Gem::Version
-  version: 0.4.1
+  version: 1.2.0
 platform: ruby
 authors:
 - David P. Shorthouse
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-09-26 00:00:00.000000000 Z
+date: 2019-10-11 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: namae