tests for base64 encoding and line-ending canonicalization

lib/as2.rb

@@ -25,8 +25,8 @@ def self.generate_message_id(server_info)
 
   def self.valid_base64_schemes
     [
-      'rfc2045', # https://www.rfc-editor.org/rfc/rfc2045#section-6.8
-      'rfc4648'  # https://www.rfc-editor.org/rfc/rfc4648#section-4
+      'rfc2045',
+      'rfc4648'
     ]
   end
 
@@ -36,17 +36,19 @@ def self.valid_base64_schemes
   # @param [String] scheme one of As2.valid_base64_schemes
   # @return [String]
   def self.base64_encode(content, scheme: 'rfc4648')
-    case scheme
+    case scheme.to_s
     when 'rfc2045'
       # "This method complies with RFC 2045."
       # https://ruby-doc.org/stdlib-3.0.4/libdoc/base64/rdoc/Base64.html#method-i-encode64
+      # https://www.rfc-editor.org/rfc/rfc2045#section-6.8
       then Base64.encode64(content)
     when 'rfc4648'
       # "This method complies with RFC 4648."
       # https://ruby-doc.org/stdlib-3.0.4/libdoc/base64/rdoc/Base64.html#method-i-strict_encode64
+      # https://www.rfc-editor.org/rfc/rfc4648#section-4
       then Base64.strict_encode64(content)
     else
-      raise "unsupported base64_scheme '#{@partner.base64_scheme}'"
+      raise ArgumentError, "unsupported scheme '#{scheme}'. choose one of: #{valid_base64_schemes}"
     end
   end
 

test/as2_test.rb

@@ -17,15 +17,104 @@
   end
 
   describe '.base64_encode' do
-    it 'can encode according to RFC-2045'
-    it 'can encode according to RFC-4648'
-    it 'raises if the given encoding scheme is not recognized'
-    it 'defaults to RFC-4648 for backwards-compatibility'
+    before(:all) do
+      @ascii = (32..126).map(&:chr).join.freeze
+      @binary = File.open('test/fixtures/white-box.png', 'rb').read.freeze
+      @utf8 = "こんにちは".freeze
+    end
+
+    describe 'with rfc2045 rules' do
+      it 'can encode all printable ascii characters' do
+        encoded = As2.base64_encode(@ascii, scheme: 'rfc2045')
+        decoded = Base64.decode64(encoded)
+        assert_equal @ascii, decoded
+      end
+
+      it 'can encode arbitrary binary data' do
+        encoded = As2.base64_encode(@binary, scheme: 'rfc2045')
+        decoded = Base64.decode64(encoded)
+        assert_equal @binary, decoded
+      end
+
+      it 'can encode UTF-8' do
+        encoded = As2.base64_encode(@utf8, scheme: 'rfc2045')
+        decoded = Base64.decode64(encoded)
+        decoded.force_encoding('UTF-8')
+        assert_equal @utf8, decoded
+      end
+    end
+
+    describe 'with rfc4648 rules' do
+      it 'can encode all printable ascii characters' do
+        encoded = As2.base64_encode(@ascii, scheme: 'rfc4648')
+        decoded = Base64.strict_decode64(encoded)
+        assert_equal @ascii, decoded
+      end
+
+      it 'can encode arbitrary binary data' do
+        encoded = As2.base64_encode(@binary, scheme: 'rfc4648')
+        decoded = Base64.strict_decode64(encoded)
+        assert_equal @binary, decoded
+      end
+
+      it 'can encode UTF-8' do
+        encoded = As2.base64_encode(@utf8, scheme: 'rfc4648')
+        decoded = Base64.decode64(encoded)
+        decoded.force_encoding('UTF-8')
+        assert_equal @utf8, decoded
+      end
+    end
+
+    it 'raises if the given encoding scheme is not recognized' do
+      error = assert_raises(ArgumentError) do
+                As2.base64_encode(@binary, scheme: 'blah')
+              end
+      assert_equal "unsupported scheme 'blah'. choose one of: [\"rfc2045\", \"rfc4648\"]", error.message
+    end
+
+    it 'defaults to RFC-4648 for backwards-compatibility' do
+      expected = As2.base64_encode(@ascii, scheme: 'rfc4648')
+      assert_equal expected, As2.base64_encode(@ascii)
+    end
   end
 
   describe '.canonicalize_line_endings' do
-    it 'replaces \n with \r\n'
-    it 'does not alter existing \r\n sequences'
+    it 'replaces \n with \r\n' do
+      input = "a\nb\nc\n"
+      expected = "a\r\nb\r\nc\r\n"
+      assert_equal expected, As2.canonicalize_line_endings(input)
+    end
+
+    it 'does not alter existing \r\n sequences' do
+      input = "a\r\nb\nc\n"
+      expected = "a\r\nb\r\nc\r\n"
+      assert_equal expected, As2.canonicalize_line_endings(input)
+    end
+
+    it 'does not add trailing newlines if string does not end with a newline' do
+      input = "a"
+      expected = "a"
+      assert_equal expected, As2.canonicalize_line_endings(input)
+    end
+
+    it 'is compatible with all base64_encode schemes' do
+      ascii = (32..126).map(&:chr).join.freeze
+      input = ascii * 10 # long enough to be split onto multiple lines in rfc2045
+
+      # if a new scheme is added, this assertion will remind us to test it here also.
+      valid_schemes = As2.valid_base64_schemes
+      assert_equal(['rfc2045', 'rfc4648'], valid_schemes)
+
+      encoded = As2.base64_encode(input, scheme: 'rfc2045')
+      canonicalized = As2.canonicalize_line_endings(encoded)
+      decoded = Base64.decode64(encoded)
+      assert_equal input, decoded
+
+      encoded = As2.base64_encode(input, scheme: 'rfc4648')
+      canonicalized = As2.canonicalize_line_endings(encoded)
+      decoded = Base64.strict_decode64(encoded)
+      assert_equal input, decoded
+    end
   end
 
   describe '.choose_mic_algorithm' do