dwc_agent 1.4.3 → 1.4.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: ee3ede60926c849a6db7094f63a9d03d07f3e9d13fd74428841812f8970aa681
-  data.tar.gz: 7075c82cd35834f8dad00d57922e46f7ab0eb4f086eb4c0ad40b02c5726ca001
+  metadata.gz: fa4fb87ea91fd1f0e67278590192a55bfc7f1e8d6f4b8dc92c1f9f5eb508e44c
+  data.tar.gz: a89b51ea705885713ef8615c67e1ea10798abfe593b5646b4de9fb8e1b478762
 SHA512:
-  metadata.gz: 9e7b655e50ec2d744d74ad44a30b35d75e76b7e0160a35ee7e6b295a03dbc343a012a702e3614c10516bf1282617ec00a0f472e2cc1aefc9d3decc9b4494a946
-  data.tar.gz: a28bd967f7df5afaf5e71a01f3902c0bab33dc735bc0861ceb137bad2d81981ec9ae6768497afff267b986bf653f9ba1fb8091221720964866b3352f8d8e2aae
+  metadata.gz: d676b64441d0097bd6272e2cd694c5754c4bdaed8fd0f523ecbe28748c8ccedffd9dd1c0430f5ad25cf48c02705b8131a9ce1021c07965da0791ae5f62e36c8a
+  data.tar.gz: 00effae1b438e6d97ef8da8383aa407985876d148b5b30a51ca98d2befa0dc8ac4a8c69bb389f75dd08c147853522490a406470dd8c4aa20d976238cf2cb4d82

data/lib/dwc_agent/constants.rb

@@ -1,7 +1,7 @@
 module DwcAgent
   STRIP_OUT = %r{
     ^[\[{(]|
-    [\]})]$|
+    [\]})]\??$|
     \s*?\d+\.\d+|
     \b\d+\(?(?i:[[:alpha:]])\)?\b|
     \b[,;]?\s*(?i:et\.?\s+al)\.?|
@@ -13,7 +13,7 @@ module DwcAgent
     \b[,;]?\s*(?i:unkn?own)\b|
     \b[,;]?\s*(?i:n/a)\b|
     \b[,;]?\s*(?i:ann?onymous)\b|
-    \b[,;]?\s*\(?(?i:undetermined|indeterminable|dummy|interim|accession|illegible|scripsit)\)?\b|
+    \b[,;]?\s*\(?(?i:undetermined|indeterminable|dummy|interim|accession|ill(eg|is)ible|scripsit)\)?\b|
     \b[,;]?\s*(?i:importer|gift)\:?\b|
     \b[,;]?\s*(?i:frère|frere|père|pere|soeur|sister|bro)\.?(\b|\z)|
     \b[,;]?\s*(?i:string)\b|
@@ -80,11 +80,12 @@ module DwcAgent
     \b\s*\(?(?i:(fe)?male)\)?\s*\b|
     \b(?i:to\s+(sub)?spp?)\.?|
     (?i:nom\.?\s+rev\.?)|
-    FNA|DAO|HUH|FDNMB|MNHN|PNI|USNM|
+    FNA|DAO|HUH|FDNMB|MNHN|PNI|USNM|ZMUC|
     AFSC\/POLISH\s+SORTING\s+CTR\.?|
     (?i:university|museum|exhibits?)|
     (?i:uqam)|
     (?i:sem\s+(colec?tor|data))|
+    (?i:no\s+coll\.?(ector)?)|
     \b[,;]\s+\d+\z|
     ["!@?]|
     [,]?\d+|
@@ -136,8 +137,6 @@ module DwcAgent
     '|' => ' | ',
     '(' => ' ',
     ')' => ' ',
-    '[' => ' ',
-    ']' => ' ',
     '?' => '',
     '!' => '',
     '=' => '',
@@ -196,7 +195,7 @@ module DwcAgent
     (?i:geographic)|
     (?i:mus(eum|ée)|universit(y|é|e|at)|college|institute?|acad(e|é)m|school|écol(e|iers?)|laboratoi?r|projec?t|polytech|dep(t|art?ment)|research|clinic|hospital|cientifica|sanctuary|safari)|
     (?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:graduate|student|élèves?|éleveur|étudiants|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))|
@@ -211,7 +210,7 @@ module DwcAgent
     (?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)|
+    (?i:survey|assessment|station|monitor|stn\.|index|project|bureau|engine|(e|é)x?chang(e|é)s?|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)|
     (?i:submersible)|
     (?i:synonymy?)|(topo|syn|holo)type|
     (?i:systematic|perspective)|
@@ -230,21 +229,28 @@ module DwcAgent
   }x
 
   FAMILY_BLACKLIST = [
+    "da",
+    "de'",
+    "del",
     "der",
+    "du",
+    "el",
     "van",
     "von",
     "the",
     "of",
-    "curators",
-    "nomenclatural",
     "adjustment",
     "available",
-    "data",
-    "orig",
-    "science",
     "catalogue",
+    "curators",
+    "data",
+    "determination",
+    "dissection",
     "entered",
-    "registration"
+    "nomenclatural",
+    "orig",
+    "registration",
+    "science"
   ]
 
   GIVEN_BLACKLIST = [

data/lib/dwc_agent/parser.rb

@@ -17,7 +17,7 @@ module DwcAgent
 
       @strip_out_regex = Regexp.new STRIP_OUT.to_s
       @residual_terminators_regex = Regexp.new SPLIT_BY.to_s + %r{\s*\z}.to_s
-      @char_subs_regex = Regexp.new [CHAR_SUBS.keys.join('\\')].to_s
+      @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{(\S{1}\.)([[:alpha:]]{2,})}.to_s
@@ -30,6 +30,7 @@ module DwcAgent
     def parse(name)
       return [] if name.nil? || name == ""
       name.gsub!(@strip_out_regex, ' ')
+      name.gsub!(/\[|\]/, '')
       name.gsub!(@char_subs_regex, CHAR_SUBS)
       name.gsub!(@phrase_subs_regex, PHRASE_SUBS)
       name.gsub!(@add_separators_regex, '\1 \2')

data/lib/dwc_agent/version.rb

@@ -3,7 +3,7 @@ module DwcAgent
 
     MAJOR = 1
     MINOR = 4
-    PATCH = 3
+    PATCH = 4
     BUILD = nil
 
     def self.version

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: dwc_agent
 version: !ruby/object:Gem::Version
-  version: 1.4.3
+  version: 1.4.4
 platform: ruby
 authors:
 - David P. Shorthouse
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2020-01-03 00:00:00.000000000 Z
+date: 2020-01-15 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: namae