From 3d8f0cc6b70b55aad20ab4d86642f0e6d605c921 Mon Sep 17 00:00:00 2001 From: Louise Crow Date: Thu, 11 Jun 2015 17:11:26 +0100 Subject: convert_string_to_utf8 returns struct of string and scrubbing status. --- spec/lib/basic_encoding_spec.rb | 23 +++++++++++++++-------- 1 file changed, 15 insertions(+), 8 deletions(-) (limited to 'spec/lib/basic_encoding_spec.rb') diff --git a/spec/lib/basic_encoding_spec.rb b/spec/lib/basic_encoding_spec.rb index d802da892..6758d60a3 100644 --- a/spec/lib/basic_encoding_spec.rb +++ b/spec/lib/basic_encoding_spec.rb @@ -160,20 +160,24 @@ describe "convert_string_to_utf8" do describe "when passed uninterpretable character data" do - it "should return it as a valid utf8 string with non-utf8 characters removed" do + it "should return it as a valid utf8 string with non-utf8 characters removed + and mark it as scrubbed" do converted = convert_string_to_utf8 random_string if String.method_defined?(:encode) - converted.encoding.to_s.should == 'UTF-8' - converted.valid_encoding?.should == true + converted.string.encoding.to_s.should == 'UTF-8' + converted.string.valid_encoding?.should == true end + converted.scrubbed?.should == true converted = convert_string_to_utf8 random_string,'UTF-8' if String.method_defined?(:encode) - converted.encoding.to_s.should == 'UTF-8' + converted.string.encoding.to_s.should == 'UTF-8' + converted.string.valid_encoding?.should == true end + converted.scrubbed?.should == true end end @@ -184,11 +188,13 @@ describe "convert_string_to_utf8" do converted = convert_string_to_utf8 windows_1252_string - converted.should == "DASH – DASH" + converted.string.should == "DASH – DASH" if String.method_defined?(:encode) - converted.encoding.to_s.should == 'UTF-8' + converted.string.encoding.to_s.should == 'UTF-8' end + converted.scrubbed?.should == false + end end @@ -199,11 +205,12 @@ describe "convert_string_to_utf8" do converted = convert_string_to_utf8 gb_18030_spam_string - converted.should start_with("贵公司负责人") + converted.string.should start_with("贵公司负责人") if String.method_defined?(:encode) - converted.encoding.to_s.should == 'UTF-8' + converted.string.encoding.to_s.should == 'UTF-8' end + converted.scrubbed?.should == false end end -- cgit v1.2.3