dwc_agent 3.0.17.0 → 3.0.19.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: fbf89a09b7566139a99cd8cd3a41daa5866a0546fb8ddbfefe8fa75d304f9760
-  data.tar.gz: 2f2a4799fd33449ecfe00055250c9dbdbf82199e78fbcdfb43510098868cdfbb
+  metadata.gz: 232c24dde9ae5ecd8221049ab24283c2480c7d09df539f0eeef6b03729295360
+  data.tar.gz: 3fa7afa2113b78326938849744cf2e6a5a1f4949ba8567c0dac200872a72c62f
 SHA512:
-  metadata.gz: 604501af44237d83613eee5e5922e6d3d1fade4095d22af0060cd6c3374a78d4267abe06146929b1e313e5ce659cbdedc6d230227a72f792e80e8f9ba85efc46
-  data.tar.gz: cacc2b1e925ab4024ffd32a1af5f6ba773e006c7ad1b0d0d512994f74bd15ec583b9742e8e92e174be180c122a491d215bce6832534180585bd370dbd5f4717f
+  metadata.gz: 4a5778f3d32830ff95a2790c320d838df3c2e08937d8ca66cd1d3bf2a84560ce46b286f259bb213b7cf29c298d817c5f33d8ec09f958d65fd255825dc9dc1860
+  data.tar.gz: 8c01720ea2044073e3c850956e74b0a01f723350abb8b6d98183f9f8a6952f17dcbb5f64ab19311d72d57ff191634c783bc2b9e93cd40fd6c1819fa6c8de83fe

data/lib/dwc_agent/cleaner.rb

@@ -78,7 +78,6 @@ module DwcAgent
          parsed_namae.family &&
          parsed_namae.family.length <=3 &&
          parsed_namae.family == parsed_namae.family.upcase &&
-         #parsed_namae.family != NameCase(parsed_namae.family) &&
          parsed_namae.given[-1] != "."
           given = parsed_namae.given
           family = parsed_namae.family
@@ -168,6 +167,11 @@ module DwcAgent
         return default
       end
 
+      if !family.nil? && family.downcase.count(VOWELS) == 0 &&
+         !FAMILY_GREENLIST.any?{ |s| s.casecmp(family) == 0 }
+       return default
+      end
+
       name = {
         title: title,
         appellation: appellation,

data/lib/dwc_agent/constants.rb

@@ -208,7 +208,8 @@ module DwcAgent
     '-Jr' => ' Jr.',
     'Dr.' => 'Dr. ',
     'prof.' => 'Prof. ',
-    ' .;' => '. ;'
+    ' .;' => '. ;',
+    ', &' => ' &'
   }
 
   SEPARATORS = {
@@ -219,9 +220,10 @@ module DwcAgent
     "(\\S{1}\\.)([[:alpha:]]{2,})" => "\\1 \\2",
     "^([[:alpha:]]{2,})(?:\\s+)((?:\\S{1}\\.\\s?){1,})$" => "\\1, \\2",
     "([[:alpha:]]*),?\\s*(.*)\\s+(van|von|v\\.|v(a|o)n\\s+der?)$" => "\\3 \\1, \\2",
-    "^([A-Z.\\s]+)\\s+(?:and|&|et|e)\\s+([A-Z.\\s]+)\\s+([[:alpha:]’`'-]{2,})\\s+([[:alpha:]’`'-]{2,})$" => "\\1 \\4 | \\2 \\3 \\4",
-    "^([A-Z.\\s]+)\\s+(?:and|&|et|e)\\s+([A-Z.\\s]+)\\s+([[:alpha:]’`'-]{2,})(.*)$" => "\\1 \\3 | \\2 \\3 | \\4",
-    "^([A-Z.\\s]+),\\s+([A-Z.\\s]+)\\s+(?:and|&|et|e)\\s+([A-Z.\\s]+)\\s+([[:alpha:]’`'-]{2,})(.*)$" => "\\1 \\4 | \\2 \\4 | \\3 \\4 | \\5",
+    "^((?i:[A-Z]\\.\\s?){1,})\\s?(?:and|&|et|e)\\s+((?i:[A-Z]\\.\\s?){1,})\\s+([[:alpha:]’`'-]{2,})\\s+([[:alpha:]’`'-]{2,})$" => "\\1 \\4 | \\2 \\3 \\4",
+    "^((?i:[A-Z]\\.\\s?){1,})\\s?(?:and|&|et|e)\\s+((?i:[A-Z]\\.\\s?){1,})\\s+([[:alpha:]’`'-]{2,})(.*)$" => "\\1 \\3 | \\2 \\3 | \\4",
+    "^([A-Z]{1,3})\\s+(?:and|&|et|e)\\s+([A-Z]{1,3})\\s+([[:alpha:]’`'-]{2,})(.*)$" => "\\1 \\3 | \\2 \\3 | \\4",
+    "^((?i:[A-Z]\\.\\s?){1,}),\\s+([A-Z.\\s]+)\\s+(?:and|&|et|e)\\s+((?i:[A-Z]\\.\\s?){1,})\\s+([[:alpha:]’`'-]{2,})(.*)$" => "\\1 \\4 | \\2 \\4 | \\3 \\4 | \\5",
     "^([A-Z][[:alpha:]]{2,}),\\s*?([A-Z][[:alpha:]]{2,})\\s*?(?i:and|&|et|e|,)\\s+([A-Z][[:alpha:]]{2,})$" => "\\1 | \\2 | \\3",
     "^([A-Z][[:alpha:]]{2,}),\\s*?([A-Z][[:alpha:]]{2,}),\\s*?([A-Z][[:alpha:]]{2,})\\s*?(?i:and|&|et|e|,)\\s+([A-Z][[:alpha:]]{3,})$" => "\\1 | \\2 | \\3 | \\4",
     "^([A-Z][[:alpha:]]{2,}),\\s*?([A-Z][[:alpha:]]{2,}),\\s*?([A-Z][[:alpha:]]{2,}),\\s*?([A-Z][[:alpha:]]{2,})\\s*?(?i:and|&|et|e|,)\\s+([A-Z][[:alpha:]]{3,})$" => "\\1 | \\2 | \\3 | \\4 | \\5"
@@ -300,6 +302,18 @@ module DwcAgent
     ^\s*?de\s*?$
   }x
 
+  FAMILY_GREENLIST = [
+    "Ng",
+    "Srb",
+    "Srp",
+    "Vlk",
+    "Smrz",
+    "Smrž",
+    "Smrt",
+    "Krc",
+    "Krč"
+  ]
+
   FAMILY_BLACKLIST = [
     "a b",
     "a e",
@@ -402,4 +416,6 @@ module DwcAgent
     "von der"
   ]
 
+  VOWELS = "aeiouäèéêëìíôöü"
+
 end

data/lib/dwc_agent/version.rb

@@ -4,7 +4,7 @@ module DwcAgent
 
     MAJOR = 3
     MINOR = 0
-    PATCH = 17
+    PATCH = 19
     BUILD = 0
 
     def self.version

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: dwc_agent
 version: !ruby/object:Gem::Version
-  version: 3.0.17.0
+  version: 3.0.19.0
 platform: ruby
 authors:
 - David P. Shorthouse
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2023-11-23 00:00:00.000000000 Z
+date: 2024-01-05 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: namae