Ensure PLS files are valid

lexicons/Latin00.pls

@@ -1796,3 +1796,4 @@
     <grapheme>pedis</grapheme>
     <phoneme>pɛdɪs</phoneme>
   </lexeme>
+  <lexicon/>

lexicons/Latin01.pls

@@ -1891,3 +1891,4 @@
   <grapheme>parcere</grapheme>
   <phoneme>parkɛre</phoneme>
 </lexeme>
+<lexicon/>

lexicons/Latin02.pls

@@ -1851,3 +1851,4 @@
   <grapheme>desitum</grapheme>
   <phoneme>desitumː</phoneme>
 </lexeme>
+<lexicon/>

lexicons/Latin03.pls

@@ -1888,3 +1888,4 @@
   <grapheme>insulto</grapheme>
   <phoneme>inˈsul.to</phoneme>
 </lexeme>
+<lexicon/>

lexicons/Latin04.pls

@@ -192,3 +192,4 @@
   <grapheme>vulnero</grapheme>
   <phoneme>ˈvul.ne.ro</phoneme>
 </lexeme>
+<lexicon/>

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

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