Fix "Error: invalid byte sequence in UTF-8" error when retrieving catalogues with invalid encoding from PuppetDB on Puppet 3 server

lib/puppet/catalog-diff/comparer.rb

@@ -119,6 +119,14 @@ def do_str_diff(str1, str2)
       diff
     end
 
+    def validate_encoding(str)
+      unless str.valid_encoding?
+        Puppet::debug("Detected that string used in diff had invalid #{str.encoding} encoding. Replacing invalid characters in diff output.")
+        str.encode!('UTF-8', 'UTF-8', :invalid => :replace)
+      end
+      str
+    end 
+
     def str_diff(cont1, cont2)
       return nil unless cont1 && cont2
 
@@ -138,6 +146,9 @@ def str_diff(cont1, cont2)
         sum2 = Digest::MD5.hexdigest(str2)
       end
 
+      str1 = validate_encoding(str1)
+      str2 = validate_encoding(str2)
+
       return nil unless str1 && str2
       return nil if sum1 == sum2
 

spec/unit/puppet/catalog_diff/comparer_spec.rb

@@ -79,6 +79,22 @@
       diffs = compare_resources(res1, res2, show_resource_diff: true)
       expect(diffs[:string_diffs]['file.foo'][3]).to eq("-\t     content => \"foo content\"")
     end
+
+    it 'returns string_diffs with show_resource_diff with content encoded in ISO8859-1 (latin1)' do
+      # Without this the unit test fails when running on LC_ALL=C but works with LC_ALL=en_US.UTF-8
+      # With this it works on both.
+      Encoding.default_internal = 'UTF-8'
+      Encoding.default_external = 'UTF-8' # Needed because diff uses tempfile.
+
+      latin1_string = [246].pack('C*').force_encoding('UTF-8')
+      res1[0][:parameters][:content] = latin1_string 
+      expect{compare_resources(res1, res2, show_resource_diff: true)}.not_to raise_error
+      diffs = compare_resources(res1, res2, show_resource_diff: true)
+
+      ruby_default_replacement_string_for_invalid_characters = '�'
+      expect(diffs[:string_diffs]['file.foo'][3]).to \
+        eq("-\t     content => \"" + ruby_default_replacement_string_for_invalid_characters + "\"")
+    end
 
     it 'returns a diff without path parameter' do
       diffs = compare_resources(res1, res2, ignore_parameters: 'path')