From dfa8215910b6e1798898c1240767fc2d7d614f6c Mon Sep 17 00:00:00 2001
From: David Shorthouse <davidpshorthouse@gmail.com>
Date: Sat, 29 Jun 2024 09:56:08 -0400
Subject: [PATCH] Strip out more characters

---
 dwc_agent.gemspec             |  2 +-
 lib/dwc_agent/constants.rb    |  4 +++-
 lib/dwc_agent/version.rb      |  2 +-
 spec/dwc_agent/parser_spec.rb | 20 ++++++++++++++++++++
 4 files changed, 25 insertions(+), 3 deletions(-)

diff --git a/dwc_agent.gemspec b/dwc_agent.gemspec
index fd2742e..3902d5f 100644
--- 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        = '2024-06-17'
+  s.date        = '2024-06-30'
   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
index f865ca7..1d339e1 100644
--- a/lib/dwc_agent/constants.rb
+++ b/lib/dwc_agent/constants.rb
@@ -1,6 +1,8 @@
 module DwcAgent
   STRIP_OUT = %r{
     (?i:acc\s?\#)|
+    ["'-]{2,}|
+    \-\.\s|
     [,;]?\s*(?i:1st|2nd|3rd|[4-9]th)|
     \s*?\d+\.\d+|
     \b\d+\(?(?i:[[:alpha:]])\)?\b|
@@ -103,7 +105,7 @@ module DwcAgent
     (?i:no\s+coll\.?(ector)?)|
     (?i:not?)\s+(?i:name|date|details?|specific)?\s*?(?i:given|name|date|noted)|
     (?i:non?)\s+(?i:specificato)|
-    \b[,;]\s+\d+\z|
+    \b[,;]\s+\d+\.?\z|
     [!@?]|
     [,]?\d+|
     \s+\d+?(\/|\.)?(?i:i|ii|iii|iv|v|vi|vii|viii|ix|x)(\/|\.)\d+|
diff --git a/lib/dwc_agent/version.rb b/lib/dwc_agent/version.rb
index 4870105..0f79c78 100644
--- a/lib/dwc_agent/version.rb
+++ b/lib/dwc_agent/version.rb
@@ -4,7 +4,7 @@ class Version
 
     MAJOR = 3
     MINOR = 1
-    PATCH = 2
+    PATCH = 3
     BUILD = 0
 
     def self.version
diff --git a/spec/dwc_agent/parser_spec.rb b/spec/dwc_agent/parser_spec.rb
index fcb0a23..a443e5c 100644
--- a/spec/dwc_agent/parser_spec.rb
+++ b/spec/dwc_agent/parser_spec.rb
@@ -2143,5 +2143,25 @@ module DwcAgent
       expect(parsed[2].values_at(:given, :family)).to eq(["West", nil])
     end
 
+    it "should ignore exclamation marks and period at end" do
+      input = '! L.P. Kvist, 1996.'
+      parsed = parser.parse(input)
+      expect(parsed.size).to eq(1)
+      expect(parsed[0].values_at(:given, :family)).to eq(["L.P.", "Kvist"])
+    end
+
+    it "should strip out multiple quotes" do
+      input = '"""Antonio Bausa"""'
+      parsed = parser.parse(input)
+      expect(parsed.size).to eq(1)
+      expect(parsed[0].values_at(:given, :family)).to eq(["Antonio", "Bausa"])
+    end
+
+    it "should strip out double dashes and extra punctuation" do
+      input = "-- Luke & -. Robertson"
+      parsed = parser.parse(input)
+      expect(parsed[0].values_at(:given, :family)).to eq(["Luke", nil])
+    end
+
   end
 end
\ No newline at end of file