dwc_agent 0.4.2 → 0.4.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: f54434352d90dff97e7c733fbc2fb53117380d5f04d6841d9796f48a5603dc18
-  data.tar.gz: 1ca4ae5b5bf104b51fa030cfe3984e428ab4325df266a6c90de50dff134038c6
+  metadata.gz: c4ea46a2cca2719aebba4a99251aaf02f6d2fb36f21f3e6ea28b76584fc7345a
+  data.tar.gz: 72cde7bbdf5c8f93923710f887f299a4618e32c5d129e8cac0bbcc1a285492fd
 SHA512:
-  metadata.gz: bd1a44b5926abba1a7124493980fcf8e91aebc3bf854bca8c92da973299a345f8eb79cd3a02696e93d1c4426a6fa3bd5ff3414239c884566ed5de0679b0e3760
-  data.tar.gz: 2dcea697b1c7d589227469e5a69a3da40f71886c01e29a714557276bcdc0d3a2755cc0ff89bf36dcd70388fd640917384b776ace8de4bd3352cdea217d08ae3e
+  metadata.gz: 77c1027c302f5b853641266a833d197c1d81045ffad0ca0a2b3f4161d1fc4443fff944e6493d9c1089e710ac5aaff33bd0075698c7b70ff0ddf90d79602c8442
+  data.tar.gz: 9b92955bf9421e4b5f7a0c5cc5fa04265a393f04de6d7f23217cea739732b81bd8f9fb9042c02db20fe13118f2c9f3b75e1a374787b83d8e0ee1350e3e2a6c7b

data/lib/dwc_agent/cleaner.rb

@@ -18,7 +18,7 @@ module DwcAgent
     def clean(parsed_namae)
       blank_name = { given: nil, family: 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
 
@@ -111,7 +111,7 @@ module DwcAgent
         return blank_name
       end
 
-      if !family.nil? && FAMILY_BLACKLIST.include?(family)
+      if !family.nil? && FAMILY_BLACKLIST.any?{ |s| s.casecmp(family) == 0 }
         return blank_name
       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?|
@@ -78,7 +77,7 @@ 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)|
@@ -165,6 +164,7 @@ module DwcAgent
   BLACKLIST = %r{
     (?i:abundant)|
     (?i:adult|juvenile)|
+    (?i:administra(d|t)or)|
     (?i:anon)|
     (?i:australian?)|
     (?i:average)|
@@ -173,9 +173,10 @@ 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))|
@@ -193,7 +194,9 @@ 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)|
@@ -223,8 +226,8 @@ module DwcAgent
     "van",
     "von",
     "the",
-    "The",
-    "Catalog"
+    "of",
+    "curators"
   ]
 
   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/version.rb

@@ -3,7 +3,7 @@ module DwcAgent
 
     MAJOR = 0
     MINOR = 4
-    PATCH = 2
+    PATCH = 3
     BUILD = nil
 
     def self.version

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: dwc_agent
 version: !ruby/object:Gem::Version
-  version: 0.4.2
+  version: 0.4.3
 platform: ruby
 authors:
 - David P. Shorthouse
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-09-26 00:00:00.000000000 Z
+date: 2019-09-27 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: namae