Add tests for invalid utf-8

test/004-strings.t

@@ -4,9 +4,12 @@ main([]) ->
     code:add_pathz("ebin"),
     code:add_pathz("test"),
     
-    etap:plan(21),
+    etap:plan(75),
     util:test_good(good()),
     util:test_errors(errors()),
+    
+    test_utf8(utf8_cases()),
+    
     etap:end_tests().
 
 good() ->
@@ -32,5 +35,65 @@ errors() ->
         <<"\"\\uFFFF\"">>,
         <<"\"\\uD834foo\\uDD1E\"">>,
         % CouchDB-345
-        <<"\"",78,69,73,77,69,78,32,70,216,82,82,32,70,65,69,78,33,"\"">>
+        <<34,78,69,73,77,69,78,32,70,216,82,82,32,70,65,69,78,33,34>>
+    ].
+
+test_utf8([]) ->
+    ok;
+test_utf8([Case | Rest]) ->
+    etap:fun_is(
+        fun({error, invalid_string}) -> true; (Else) -> Else end,
+        jiffy:encode(Case),
+        lists:flatten(io_lib:format("Invalid utf-8: ~p", [Case]))
+    ),
+    Case2 = <<34, Case/binary, 34>>,
+    etap:fun_is(
+        fun({error, {_, invalid_string}}) -> true; (Else) -> Else end,
+        jiffy:decode(Case2),
+        lists:flatten(io_lib:format("Invalid utf-8: ~p", [Case2]))
+    ),
+    test_utf8(Rest).
+
+utf8_cases() ->
+    [
+        % Stray continuation byte
+        <<16#C2, 16#81, 16#80>>,
+        <<"foo", 16#80, "bar">>,
+        
+        % Not enough extension bytes
+        <<16#C0>>,
+        
+        <<16#E0>>,
+        <<16#E0, 16#80>>,
+        
+        <<16#F0>>,
+        <<16#F0, 16#80>>,
+        <<16#F0, 16#80, 16#80>>,
+
+        <<16#F8>>,
+        <<16#F8, 16#80>>,
+        <<16#F8, 16#80, 16#80>>,
+        <<16#F8, 16#80, 16#80, 16#80>>,
+        
+        <<16#FC>>,
+        <<16#FC, 16#80>>,
+        <<16#FC, 16#80, 16#80>>,
+        <<16#FC, 16#80, 16#80, 16#80>>,
+        <<16#FC, 16#80, 16#80, 16#80, 16#80>>,
+
+        % No data in high bits.
+        <<16#C0, 16#80>>,
+        <<16#C1, 16#80>>,
+        
+        <<16#E0, 16#80, 16#80>>,
+        <<16#E0, 16#90, 16#80>>,
+        
+        <<16#F0, 16#80, 16#80, 16#80>>,
+        <<16#F0, 16#88, 16#80, 16#80>>,
+
+        <<16#F8, 16#80, 16#80, 16#80, 16#80>>,
+        <<16#F8, 16#84, 16#80, 16#80, 16#80>>,
+        
+        <<16#FC, 16#80, 16#80, 16#80, 16#80, 16#80>>,
+        <<16#FC, 16#82, 16#80, 16#80, 16#80, 16#80>>
     ].

test/util.erl

@@ -27,9 +27,16 @@ check_good({J, E, J2}) ->
     etap:is(jiffy:decode(J), {ok, E}, ok_dec(J, E)),
     etap:is(do_encode(E), {ok, J2}, ok_enc(E, J2)).
 
+check_error({J, E}) ->
+    etap:fun_is(
+        fun({error, E1}) when E1 == E -> true; (E1) -> E1 end,
+        jiffy:decode(J),
+        error_mesg(J)
+    );
 check_error(J) ->
     etap:fun_is(
-        fun({error, _}) -> true; (_) -> false end,
+        fun({error, _}) -> true; (Else) -> Else end,
         jiffy:decode(J),
         error_mesg(J)
     ).
+