More work on issues uncovered in #18

bionomia · Oct 13, 2023 · 99a6595 · 99a6595
1 parent a24c2bb
commit 99a6595
Show file tree

Hide file tree

Showing 5 changed files with 18 additions and 12 deletions.
diff --git a/dwc_agent.gemspec b/dwc_agent.gemspec
@@ -7,7 +7,7 @@ Gem::Specification.new do |s|
   s.name        = 'dwc_agent'
   s.version     = DwcAgent::Version.version
   s.license     = 'MIT'
-  s.date        = '2023-10-09'
+  s.date        = '2023-10-12'
   s.summary     = "Parse Darwin Core agent terms such as recordedBy and identifiedBy"
   s.description = "Parses the typically messy content in Darwin Core terms that contain people names"
   s.authors     = ["David P. Shorthouse"]

diff --git a/lib/dwc_agent/constants.rb b/lib/dwc_agent/constants.rb
@@ -1,7 +1,5 @@
 module DwcAgent
   STRIP_OUT = %r{
-    ^[\[{(]|
-    [\]})]\??$|
     (?i:acc\s?\#)|
     [,;]?\s*(?i:1st|2nd|3rd|[4-9]th)|
     \s*?\d+\.\d+|
@@ -14,7 +12,7 @@ module DwcAgent
     \b[,;]?\s*(?i:etc)\.?|
     \b[,;]?\s*(?i:exp)\.?\s*(\b|\z)|
     \b[,;]?\s*(?i:aboard)[^$]+|
-    \b[,;]?\s*(?i:on)\b|
+    \b[,;]?\s+(?i:on)\b|
     \b[,;]?\s*(?i:unkn?own)\b|
     \b[,;]?\s*(?i:n/a)\b|
     \b[,;]?\s*(?i:ann?onymous)\b|
@@ -113,8 +111,9 @@ module DwcAgent
     (?i:annot)\.?\s*?\b|
     \s+(?i:stet)\s*!?\s*\z|
     \s+(?i:prep)\.?\s*\z|
-    \W([({\[].*?[)}\]])|
-    \W[\(\[\{][A-Za-z]{1,3}$|
+    ([({].*?[)}])|
+    \s+\[([[:word:]]|[[:space:]]|[-\?\.]){10,}\]|
+    [\(\{][A-Za-z]{1,3}$|
     \b(?i:leg)[\.:]?\s*\b|
     (?:[Dd](ed|on))[\.:]|
     \d*[A-Za-z]*\d*-\d*\z|

diff --git a/lib/dwc_agent/version.rb b/lib/dwc_agent/version.rb
@@ -4,7 +4,7 @@ class Version
 
     MAJOR = 3
     MINOR = 0
-    PATCH = 14
+    PATCH = 15
     BUILD = 0
 
     def self.version

diff --git a/spec/dwc_agent/parser_spec.rb b/spec/dwc_agent/parser_spec.rb
@@ -470,7 +470,7 @@ module DwcAgent
       it "should remove extraneous material" do
         input = "Unknown [J. S. Erskine?]"
         parsed = parser.parse(input)
-        expect(parsed.size).to eq(0)
+        expect(parsed).to eq([])
       end
 
       it "should parse name with many given initials" do
@@ -491,7 +491,7 @@ module DwcAgent
         input = "Jack [John] Smith12345"
         parsed = parser.parse(input)
         expect(parsed.size).to eq(1)
-        expect(parsed[0].values_at(:given, :family)).to eq(['Jack', 'Smith'])
+        expect(parsed[0].values_at(:given, :family)).to eq(['Jack John', 'Smith'])
       end
 
       it "should explode names with '/'" do
@@ -1348,8 +1348,8 @@ module DwcAgent
         input = "Holm, E (operator).; Ng, J.(netter); Litwiller, S. (netter); Lee, C. (data recorder)"
         parsed = parser.parse(input)
         expect(parsed.size).to eq(4)
-        expect(parsed[0].values_at(:given, :family)).to eq(["E.", "Holm"])
-        expect(parsed[1].values_at(:given, :family)).to eq(["J", "Ng"])
+        expect(parsed[0].values_at(:given, :family)).to eq(["E", "Holm"])
+        expect(parsed[1].values_at(:given, :family)).to eq(["J.", "Ng"])
         expect(parsed[2].values_at(:given, :family)).to eq(["S.", "Litwiller"])
         expect(parsed[3].values_at(:given, :family)).to eq(["C.", "Lee"])
       end
@@ -2073,5 +2073,12 @@ module DwcAgent
       expect(parsed[0].values_at(:given, :family)).to eq(["A.", "Breuckner"])
     end
 
+    it "should remove multiple square braclets within parts of name" do
+      input = "Attila Meste[r]há[zy]"
+      parsed = parser.parse(input)
+      expect(parsed.size).to eq(1)
+      expect(parsed[0].values_at(:given, :family)).to eq(["Attila", "Mesterházy"])
+    end
+
   end
 end
diff --git a/spec/resources/test_data.txt b/spec/resources/test_data.txt
@@ -119,7 +119,7 @@ FAH Sperling || [{"family":"Sperling", "given":"F.A.H.", "particle":null, "appel
 
 Chris MacQuarrie || [{"family":"MacQuarrie", "given":"Chris", "particle":null, "appellation": null, "suffix": null, "title": null, "dropping_particle": null, "nick": null}]
 
-Jack [John] Smith12345 || [{"family":"Smith", "given":"Jack", "particle":null, "appellation": null, "suffix": null, "title": null, "dropping_particle": null, "nick": null}]
+Jack [John] Smith12345 || [{"family":"Smith", "given":"Jack John", "particle":null, "appellation": null, "suffix": null, "title": null, "dropping_particle": null, "nick": null}]
 
 O.Bennedict/G.J. Spencer || [{"family":"Bennedict", "given":"O.", "particle":null, "appellation": null, "suffix": null, "title": null, "dropping_particle": null, "nick": null}, {"family":"Spencer", "given":"G.J.", "particle":null, "appellation": null, "suffix": null, "title": null, "dropping_particle": null, "nick": null}]