dwc_agent 3.2.0.0 → 3.3.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 817c66f4671d6f4b7adf698f743e2bd9f3af0b8c0acdcaa011a0a7aaa4e26b81
-  data.tar.gz: 64e55fb97847322db8b35d974c5f1643887bf0fdc186f6befeba464c5d600fd1
+  metadata.gz: 40a0a33de602297e5e87e3059edbd0c88e0ad36bbb90f0803362698120f9e3b1
+  data.tar.gz: 94705d02d7cb7a3ac1647f903d15db7140ece849008ba84f6d49f1bf678abf14
 SHA512:
-  metadata.gz: aa52bf9395a8df4678ec1a5cbf3dc3e523daa366eebb89f2e63918d94fbbe784c648a5a0f7906a76b6c5778752a98a74bfe61b6831c5157fe7aa74dd885b116a
-  data.tar.gz: f285c2eabd308dba87a03d7a92487d55e3b224f9dfbad1c37d9e43fff9ae2457dd16015bc47805e18c3a69c794e8a29b64ee6d8474321c94de3952bd96959937
+  metadata.gz: 1baeb5c1367e570139fca30a1181395030b12d09b1d4f12a026ad48e28ebaecee70b20d81b96b02f2600da269f2342281c8929facc70f0b9c571cafcfb273501
+  data.tar.gz: bf7676d6ed221258ab6efe217a7b695cb619c54be8a8d14f16bb23a3f1bc9e0fa0a33474211e7f20e5a08b058e2ab4b57e4c5e1367be74d9719ad2502a035d77

data/lib/dwc_agent/cleaner.rb

@@ -2,17 +2,25 @@ module DwcAgent
 
   class Cleaner
 
+    @defaults = {
+      blacklist: BLACKLIST,
+      given_blacklist: GIVEN_BLACKLIST,
+      family_blacklist: FAMILY_BLACKLIST,
+      particles: PARTICLES
+    }
+
     class << self
+      attr_reader :defaults
+
       def instance
         Thread.current[:dwc_agent_cleaner] ||= new
       end
     end
 
-    def initialize
-      @blacklist = BLACKLIST
-      @given_blacklist = GIVEN_BLACKLIST
-      @family_blacklist = FAMILY_BLACKLIST
-      @particles = PARTICLES
+    attr_reader :options
+
+    def initialize(options = {})
+      @options = self.class.defaults.merge(options)
     end
 
     def default
@@ -26,6 +34,8 @@ module DwcAgent
     # @return Namae::Name [Object] a new Namae object
     def clean(parsed_namae)
 
+      return default if !parsed_namae.instance_of?(Namae::Name)
+
       if parsed_namae.family &&
          parsed_namae.family == NameCase(parsed_namae.family) &&
          parsed_namae.display_order.split.join == parsed_namae.initials
@@ -33,7 +43,7 @@ module DwcAgent
       end
 
       if parsed_namae.given &&
-         @given_blacklist.any?{ |s| s.casecmp(parsed_namae.given) == 0 }
+         options[:given_blacklist].any?{ |s| s.casecmp(parsed_namae.given) == 0 }
         return
       end
 
@@ -53,7 +63,7 @@ module DwcAgent
         return default
       end
 
-      if parsed_namae.display_order =~ @blacklist
+      if parsed_namae.display_order =~ options[:blacklist]
         return default
       end
 
@@ -111,7 +121,7 @@ module DwcAgent
       end
 
       if parsed_namae.family &&
-         @family_blacklist.any?{ |s| s.casecmp(parsed_namae.family) == 0 }
+         options[:family_blacklist].any?{ |s| s.casecmp(parsed_namae.family) == 0 }
         return default
       end
 
@@ -138,7 +148,7 @@ module DwcAgent
       if !family.nil? &&
          given.nil? &&
          !particle.nil? &&
-         !@particles.include?(particle.downcase)
+         !options[:particles].include?(particle.downcase)
         given = particle.sub(/[a-z]\./, &:upcase).sub(/^(.)/) { $1.capitalize }
         particle = nil
       end
@@ -159,11 +169,11 @@ module DwcAgent
         return default
       end
 
-      if !family.nil? && @family_blacklist.any?{ |s| s.casecmp(family) == 0 }
+      if !family.nil? && options[:family_blacklist].any?{ |s| s.casecmp(family) == 0 }
         return default
       end
 
-      if !given.nil? && @given_blacklist.any?{ |s| s.casecmp(given) == 0 }
+      if !given.nil? && options[:given_blacklist].any?{ |s| s.casecmp(given) == 0 }
         return default
       end
 

data/lib/dwc_agent/constants.rb

@@ -10,7 +10,7 @@ module DwcAgent
     [,]?\s*\#*\s+\d+\-(?i:[A-Z]|\d)+\-?\d*[A-Za-z]*\z|
     \d*[A-Za-z]*\d*-\d*\z|
     \b\d+\(?(?i:[[:alpha:]])\)?\b|
-    \b[,;]?\s*(?:et\.?\s+al|&\s+al)\.?|
+    [,;\s]{1,}(?:et\.?\s+al|&\s+al)l?\.?|
     \b[,;]?\s*(?i:etal)\.?|
     \b[,;]?\s*(?i:et.al)\.?|
     \b\s+(bis|ter)(\b|\z)|
@@ -113,7 +113,7 @@ module DwcAgent
     (?i:not?)\s+(?i:name|date|details?|specific)?\s*?(?i:given|name|date|noted)|
     (?i:non?)\s+(?i:specificato)|
     \b[,;]\s+\d+\.?\z|
-    [!@?]|
+    [!@?]\s*\-?\s*|
     \d{1,4}[\/.]?(?i:i|ii|iii|iv|v|vi|vii|viii|ix|x|xi|xii)[\/.]\d{1,4}|
     [,]?\d+|
     [,;]\z|

data/lib/dwc_agent/parser.rb

@@ -2,27 +2,33 @@ module DwcAgent
 
   class Parser
 
+    @defaults = {
+      prefer_comma_as_separator: true,
+      separator: SPLIT_BY,
+      title: TITLE,
+      appellation: APPELLATION,
+      suffix: SUFFIX,
+      strip_out_regex: Regexp.new(STRIP_OUT.to_s),
+      tidy_remains_regex: Regexp.new(POST_STRIP_TIDY.to_s),
+      char_subs_regex: Regexp.new([CHAR_SUBS.keys.join].to_s),
+      phrase_subs_regex: Regexp.new(PHRASE_SUBS.keys.map{|a| Regexp.escape a }.join('|').to_s),
+      residual_terminators_regex: Regexp.new(SPLIT_BY.to_s + %r{\s*\z}.to_s),
+      separators: SEPARATORS.map{|k,v| [ Regexp.new(k), v] }
+    }
+
     class << self
+      attr_reader :defaults
+
       def instance
         Thread.current[:dwc_agent_parser] ||= new
       end
     end
 
-    def initialize
-      options = {
-        prefer_comma_as_separator: true,
-        separator: SPLIT_BY,
-        title: TITLE,
-        appellation: APPELLATION,
-        suffix: SUFFIX
-      }
-      @namae = Namae::Parser.new(options)
-      @strip_out_regex = Regexp.new STRIP_OUT.to_s
-      @tidy_remains_regex = Regexp.new POST_STRIP_TIDY.to_s
-      @char_subs_regex = Regexp.new [CHAR_SUBS.keys.join].to_s
-      @phrase_subs_regex = Regexp.new PHRASE_SUBS.keys.map{|a| Regexp.escape a }.join('|').to_s
-      @residual_terminators_regex = Regexp.new SPLIT_BY.to_s + %r{\s*\z}.to_s
-      @separators = SEPARATORS.map{|k,v| [ Regexp.new(k), v] }
+    attr_reader :options, :namae
+
+    def initialize(options = {})
+      @options = self.class.defaults.merge(options)
+      @namae = Namae::Parser.new(@options)
     end
 
     # Parses the passed-in string and returns a list of names.
@@ -31,14 +37,14 @@ module DwcAgent
     # @return [Array] the list of parsed names
     def parse(name)
       return [] if name.nil? || name == ""
-      name.gsub!(@strip_out_regex, ' ')
-      name.gsub!(@tidy_remains_regex, '')
-      name.gsub!(Regexp.union(@char_subs_regex, @phrase_subs_regex), CHAR_SUBS.merge(PHRASE_SUBS))
-      @separators.each{|k| name.gsub!(k[0], k[1])}
-      name.gsub!(@residual_terminators_regex, '')
+      name.gsub!(options[:strip_out_regex], ' ')
+      name.gsub!(options[:tidy_remains_regex], '')
+      name.gsub!(Regexp.union(options[:char_subs_regex], options[:phrase_subs_regex]), CHAR_SUBS.merge(PHRASE_SUBS))
+      options[:separators].each{|k| name.gsub!(k[0], k[1])}
+      name.gsub!(options[:residual_terminators_regex], '')
       name.squeeze!(' ')
       name.strip!
-      @namae.parse(name)
+      namae.parse(name)
     end
 
   end

data/lib/dwc_agent/version.rb

@@ -3,7 +3,7 @@ module DwcAgent
   class Version
 
     MAJOR = 3
-    MINOR = 2
+    MINOR = 3
     PATCH = 0
     BUILD = 0
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: dwc_agent
 version: !ruby/object:Gem::Version
-  version: 3.2.0.0
+  version: 3.3.0.0
 platform: ruby
 authors:
 - David P. Shorthouse
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-09-21 00:00:00.000000000 Z
+date: 2024-11-08 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: namae