Support diacritics

lib/dry/inflector.rb

@@ -155,7 +155,7 @@ def humanize(input)
       result = inflections.humans.apply_to(input)
       result.delete_suffix!("_id")
       result.tr!("_", " ")
-      match = /(\W)/.match(result)
+      match = /([^[:alnum:]])/.match(result)
       separator = match ? match[0] : DEFAULT_SEPARATOR
       result.split(separator).map.with_index { |word, index|
         inflections.acronyms.apply_to(word, capitalize: index.zero?)
@@ -283,8 +283,8 @@ def underscore(input)
         m2 = Regexp.last_match(2)
         "#{m1 ? "_" : ""}#{m2.downcase}"
       end
-      input.gsub!(/([A-Z\d]+)([A-Z][a-z])/, '\1_\2')
-      input.gsub!(/([a-z\d])([A-Z])/, '\1_\2')
+      input.gsub!(/([[:upper:][:digit:]]+)([[:upper:]][[:lower:]])/, '\1_\2')
+      input.gsub!(/([[:lower:][:digit:]])([[:upper:]])/, '\1_\2')
       input.tr!("-", "_")
       input.downcase!
       input
@@ -328,8 +328,10 @@ def to_s
     # @api private
     def internal_camelize(input, upper)
       input = input.to_s.dup
-      input.sub!(/^[a-z\d]*/) { |match| inflections.acronyms.apply_to(match, capitalize: upper) }
-      input.gsub!(%r{(?:[_-]|(/))([a-z\d]*)}i) do
+      input.sub!(/^[[:lower:][:digit:]]*/) do |match|
+        inflections.acronyms.apply_to(match, capitalize: upper)
+      end
+      input.gsub!(%r{(?:[_-]|(/))([[:lower:][:digit:]]*)}i) do
         m1 = Regexp.last_match(1)
         m2 = Regexp.last_match(2)
         "#{m1}#{inflections.acronyms.apply_to(m2)}"

spec/unit/dry/inflector/camelize_lower_spec.rb

@@ -26,7 +26,7 @@
     context "custom acronyms" do
       subject do
         Dry::Inflector.new do |inflect|
-          inflect.acronym("IP", "HTML", "XML", "BSD")
+          inflect.acronym("IP", "HTML", "XML", "BSD", "ÉUA", "4K")
         end
       end
 
@@ -38,6 +38,13 @@
         expect(subject.camelize_lower(i("force_xml_controller"))).to eql("forceXMLController")
         expect(subject.camelize_lower(i("free_bsd"))).to eql("freeBSD")
       end
+
+      it "handles diacritics" do
+        expect(subject.camelize_lower(i("éclair_fest"))).to eql("éclairFest")
+        expect(subject.camelize_lower(i("éua-trip"))).to eql("ÉUATrip")
+        expect(subject.camelize_lower(i("festival_des_éclairs"))).to eql("festivalDesÉclairs")
+        expect(subject.camelize_lower(i("festival_des_4k"))).to eql("festivalDes4K")
+      end
     end
   end
 end

spec/unit/dry/inflector/camelize_upper_spec.rb

@@ -30,7 +30,7 @@
     context "custom acronyms" do
       subject do
         Dry::Inflector.new do |inflect|
-          inflect.acronym("IP", "HTML", "XML", "BSD")
+          inflect.acronym("IP", "HTML", "XML", "BSD", "ÉUA", "4K")
         end
       end
 
@@ -41,6 +41,14 @@
         expect(subject.camelize_upper(i("html_tidy_generator"))).to eql("HTMLTidyGenerator")
         expect(subject.camelize_upper(i("force_xml_controller"))).to eql("ForceXMLController")
         expect(subject.camelize_upper(i("free_bsd"))).to eql("FreeBSD")
+        expect(subject.camelize_upper(i("4k-television"))).to eql("4KTelevision")
+      end
+
+      it "handles diacritics" do
+        expect(subject.camelize_upper(i("éclair_fest"))).to eql("ÉclairFest")
+        expect(subject.camelize_upper(i("éua-trip"))).to eql("ÉUATrip")
+        expect(subject.camelize_upper(i("festival_des_éclairs"))).to eql("FestivalDesÉclairs")
+        expect(subject.camelize_upper(i("festival_des_4k"))).to eql("FestivalDes4K")
       end
     end
   end

spec/unit/dry/inflector/humanize_spec.rb

@@ -29,5 +29,9 @@
       expect(subject.humanize(i("openssl/hmac"))).to eql("OpenSSL/HMAC")
       expect(subject.humanize(i("openssl/digest"))).to eql("OpenSSL/digest")
     end
+
+    it "handles diacritics" do
+      expect(subject.humanize(i("éclair"))).to eql("Éclair")
+    end
   end
 end

spec/unit/dry/inflector/underscore_spec.rb

@@ -34,6 +34,10 @@
       expect(subject.underscore(i("CLIRunner"))).to eq("cli_runner")
     end
 
+    it "underscores aCli as a_cli" do
+      expect(subject.underscore(i("aCLI"))).to eq("a_cli")
+    end
+
     it "accepts symbols" do
       expect(subject.underscore(:DataMapper)).to eq("data_mapper")
     end
@@ -44,5 +48,41 @@
       expect(subject.underscore(i("OpenSSL::HMAC"))).to eql("openssl/hmac")
       expect(subject.underscore(i("OpenSSL::Digest"))).to eql("openssl/digest")
     end
+
+    it "handles diacritics" do
+      expect(subject.underscore(i("éclairFest"))).to eql("éclair_fest")
+      expect(subject.underscore(i("éuaTrip"))).to eql("éua_trip")
+      expect(subject.underscore(i("festival-des-éclairs"))).to eql("festival_des_éclairs")
+      expect(subject.underscore(i("ÉBellisimo"))).to eql("é_bellisimo")
+      expect(subject.underscore(i("éBellisimo"))).to eql("é_bellisimo")
+    end
+
+    it "handles number followed by uppercase letter" do
+      expect(subject.underscore(i("9Lives"))).to eql("9_lives")
+    end
+
+    pending "handles number followed by lowercase letter" do
+      pending "this should be `7_ate_9`"
+      expect(subject.underscore(i("7Ate9"))).to eql("7_ate_9")
+    end
+
+    it "handles number followed by lowercase letter" do
+      pending "this should be 9_lives"
+      expect(subject.underscore(i("9lives"))).to eql("9_lives")
+    end
+
+    it "handle leading diacritic" do
+      expect(subject.underscore(i("éBellisimo"))).to eql("é_bellisimo")
+    end
+
+    it "handles leading number " do
+      pending "this should be 1_is_one"
+      expect(subject.underscore(i("1isOne"))).to eql("1_is_one")
+    end
+
+    it "handles leading number with diacritic" do
+      pending "this should be 1_é_bellisimo"
+      expect(subject.underscore(i("1éBellisimo"))).to eql("1_é_bellisimo")
+    end
   end
 end