dwc_agent 0.1.16 → 0.1.17

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 42cdbedccdf7273ea32dba89ec741e732d743b2c
-  data.tar.gz: 3b547d5836201c1f0ac2e63ef48b07abdda15205
+  metadata.gz: 3cec6480c674b8445ed7c7faa22cb9635607ac56
+  data.tar.gz: 6ef18ecc234b816bf0127796ef31e63c4ea8e3be
 SHA512:
-  metadata.gz: 59de6c725b3df8667b4f856da723ae92f228a2e37f238f98a2199a0b846fcfcd2ceced0be1f52f868eee294053370ddb0e19592666d10858e7d80e9f37d2bf77
-  data.tar.gz: 789f96f530c38b47928cf1935c23d058aae3a2214da296dbe0d5cd385f8648109eb6d5c319839af99ef6c440db39807b88580a39e2a8bb70b517d2dae2dfb03f
+  metadata.gz: a6bd424b6a11aa2849424bd1cc8805d1b43505fd296bcab318980d0e76efecf08d12c1ef092baee86d4296c863dd572a169a0332589ec1326e26434b2e801b7c
+  data.tar.gz: a7e7a047ef29b39a54a1be20e39d92c3d4216de2e0cb4a7d4f7d8bfa58f8c7bac39dee64033efa8a2031707c63ee2d910f512d190faf734e3f8588ee0f0f55bf

data/lib/dwc_agent/cleaner.rb

@@ -52,6 +52,10 @@ module DwcAgent
           parsed_namae.given = CapitalizeNames.capitalize(parsed_namae.given)
       end
 
+      if parsed_namae.given && /\.[A-Z]$/.match(parsed_namae.given)
+        parsed_namae.given += "."
+      end
+
       if parsed_namae.given && /[A-Za-z]\./.match(parsed_namae.given)
         parsed_namae.given = CapitalizeNames.capitalize(parsed_namae.given).gsub(/[a-z]\./, &:upcase)
       end

data/lib/dwc_agent/constants.rb

@@ -11,10 +11,12 @@ module DwcAgent
     \b[,;]?\s*(?i:n/a)\b|
     \b[,;]?\s*(?i:ann?onymous)\b|
     \b[,;]?\s*(?i:undetermined|indeterminable|dummy|interim|accession)\b|
-    \b[,;]?\s*(?i:importer)\b|
+    \b[,;]?\s*(?i:importer|gift)\:?\b|
     \b[,;]?\s*(?i:frère|frere|père|pere|soeur|sister|bro)\.?(\b|\z)|
     \b[,;]?\s*(?i:string)\b|
     \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|
@@ -45,7 +47,8 @@ module DwcAgent
     \d+\s+(?i:Nov|Novemb(er|re))\.?\b|
     \d+\s+(?i:Dec|D(e|é)cemb(er|re))\.?\b|
     (?i:autres?\s+de|probab|likely|possibl(e|y)|doubtful)|
-    \b\s*maybe\s*\b|
+    \b\s*(?i:maybe)\s*\b|
+    \b\s*(?i:prob)\.\s*\b|
     \(?(?i:collector|data\s*recorder|netter|(oper|prepar)ator)\(?s?\)?\.?\:?|
     (?i:fide)\:?\s*\b|
     (?i:game\s+dept)\.?\s*\b|
@@ -82,14 +85,15 @@ module DwcAgent
     (?i:ded)\:|
     ^[-,.\s;*\d]+\s?|
     -\d?\z|
-    \s*?-{2,}\s*?
+    \s*?-{2,}\s*?|
+    ^(?i:exc?p?)[:.]\s*
   }x
 
   SPLIT_BY = %r{
     [–|&+/;]|
     \s+-\s+|
     \s+a\.\s+|
-    \b(e|y|en|et|or|per|for)\s*\b|
+    \b(e|y|i|en|et|or|per|for)\s*\b|
     \b(?i:and|with)\s*\b|
     \b(?i:annotated(\s+by)?)\s*\b|
     \b(?i:coll\.)\s*\b|
@@ -156,13 +160,16 @@ module DwcAgent
     \b\s*(?i:help)\s*\b|
     (?i:description|drawing|identification|remark|original|illustration|checklist|intermedia|measurement|indisting|series|imperfect)|
     (?i:desconocido)|
+    (?i:exc?s?icc?at(a|i))|
     (?i:evidence)|
+    (?i:exporter)|
     (?i:inconn?u)|
-    (?i:internation|gou?vern|ministry|unit|district|provincial|na(c|t)ional|military|region|environ|natur(e|al)|naturelles|division|program|direction|national)|
+    (?i:internation|gou?vern|ministry|extension|unit|district|provincial|na(c|t)ional|military|region|environ|natur(e|al)|naturelles|division|program|direction|national)|
     (?i:label)|
     (?i:o?\.?m\.?n\.?r\.?)|
     (?i:measurement)|
     (?i:ent(o|y)mology)|
+    (?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)|
@@ -170,11 +177,11 @@ module DwcAgent
     (?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|alliance|protective|circle)|
+    (?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:size|large|colou?r)\s+|
     (?i:skeleton)|
-    (?i:survey|assessment|station|monitor|stn\.|index|project|bureau|engine|expedi(c|t)ion|festival|generation|inventory|marine|service)|
+    (?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:submersible)|
     (?i:synonymy?)|(topo|syn|holo)type|
     (?i:systematic|perspective)|

data/lib/dwc_agent/version.rb

@@ -3,7 +3,7 @@ module DwcAgent
 
     MAJOR = 0
     MINOR = 1
-    PATCH = 16
+    PATCH = 17
     BUILD = nil
 
     def self.version

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: dwc_agent
 version: !ruby/object:Gem::Version
-  version: 0.1.16
+  version: 0.1.17
 platform: ruby
 authors:
 - David P. Shorthouse
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-01-17 00:00:00.000000000 Z
+date: 2019-01-21 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: namae