From 5d3b88f0e1921f7c5af440bf9e28660215cd38c7 Mon Sep 17 00:00:00 2001
From: Jayson Virissimo <jayson.virissimo@asu.edu>
Date: Sat, 27 Apr 2024 19:40:57 -0700
Subject: [PATCH] Ensure PLS files are valid

---
 lexicons/Latin00.pls                   |  1 +
 lexicons/Latin01.pls                   |  1 +
 lexicons/Latin02.pls                   |  1 +
 lexicons/Latin03.pls                   |  1 +
 lexicons/Latin04.pls                   |  1 +
 lib/medieval_latina/lexicon_builder.rb |  4 +++-
 spec/medieval_latina_spec.rb           | 18 ++++++++++++++++++
 7 files changed, 26 insertions(+), 1 deletion(-)

diff --git a/lexicons/Latin00.pls b/lexicons/Latin00.pls
index fb0e65c..3177386 100644
--- a/lexicons/Latin00.pls
+++ b/lexicons/Latin00.pls
@@ -1796,3 +1796,4 @@
     <grapheme>pedis</grapheme>
     <phoneme>pɛdɪs</phoneme>
   </lexeme>
+  <lexicon/>
diff --git a/lexicons/Latin01.pls b/lexicons/Latin01.pls
index e62a56b..8d9e100 100644
--- a/lexicons/Latin01.pls
+++ b/lexicons/Latin01.pls
@@ -1891,3 +1891,4 @@
   <grapheme>parcere</grapheme>
   <phoneme>parkɛre</phoneme>
 </lexeme>
+<lexicon/>
diff --git a/lexicons/Latin02.pls b/lexicons/Latin02.pls
index abc227b..4cec7ab 100644
--- a/lexicons/Latin02.pls
+++ b/lexicons/Latin02.pls
@@ -1851,3 +1851,4 @@
   <grapheme>desitum</grapheme>
   <phoneme>desitumː</phoneme>
 </lexeme>
+<lexicon/>
diff --git a/lexicons/Latin03.pls b/lexicons/Latin03.pls
index 01b0664..1b98575 100644
--- a/lexicons/Latin03.pls
+++ b/lexicons/Latin03.pls
@@ -1888,3 +1888,4 @@
   <grapheme>insulto</grapheme>
   <phoneme>inˈsul.to</phoneme>
 </lexeme>
+<lexicon/>
diff --git a/lexicons/Latin04.pls b/lexicons/Latin04.pls
index 12e12cb..d711bd4 100644
--- a/lexicons/Latin04.pls
+++ b/lexicons/Latin04.pls
@@ -192,3 +192,4 @@
   <grapheme>vulnero</grapheme>
   <phoneme>ˈvul.ne.ro</phoneme>
 </lexeme>
+<lexicon/>
diff --git a/lib/medieval_latina/lexicon_builder.rb b/lib/medieval_latina/lexicon_builder.rb
index a83305d..763809c 100644
--- a/lib/medieval_latina/lexicon_builder.rb
+++ b/lib/medieval_latina/lexicon_builder.rb
@@ -25,17 +25,19 @@ def call
       xml.lexicon(xmlns: URL, version: "1.0") do
         grouped_hash.each do |phonetics, words|
           if xml.target!.length > MAX_SIZE
+            xml.lexicon  # Close the current lexicon tag
             write_file(xml.target!, file_index)
             file_index += 1
             xml = Builder::XmlMarkup.new(indent: 2)  # Reset XML builder
             xml.instruct! :xml, encoding: "UTF-8"
-            xml.lexicon(xmlns: URL, version: "1.0")
+            xml.lexicon(xmlns: URL, version: "1.0")  # Start a new lexicon tag
           end
           xml.lexeme do
             words.each { |word| xml.grapheme word }
             xml.phoneme phonetics
           end
         end
+        xml.lexicon  # Close the final lexicon tag
       end
       write_file(xml.target!, file_index)  # Write the last file
     end
diff --git a/spec/medieval_latina_spec.rb b/spec/medieval_latina_spec.rb
index d6407e1..17c3657 100644
--- a/spec/medieval_latina_spec.rb
+++ b/spec/medieval_latina_spec.rb
@@ -191,5 +191,23 @@
         expect { Nokogiri::XML(content) }.not_to raise_error
       end
     end
+
+    it "includes the XML declaration" do
+      lexicon_files.each do |file|
+        content = File.read(file)
+        expect(content).to match(/\A<\?xml version="1\.0" encoding="UTF-8"\?>\n/)
+      end
+    end
+
+    it "contains valid IPA characters in the <phoneme> elements" do
+      valid_ipa_regex = /\A[\p{L}\p{M}\p{N}\p{P}\p{S}\p{Z}]+\z/
+      lexicon_files.each do |file|
+        content = File.read(file)
+        doc = Nokogiri::XML(content)
+        doc.xpath("//phoneme").each do |phoneme|
+          expect(phoneme.text).to match(valid_ipa_regex)
+        end
+      end
+    end
   end
 end