OTP-11660: make eunit unicode safe

All output from eunit is unicode, including the surefire XML files.

1 files changed, 21 insertions, 6 deletions

diff --git a/lib/eunit/src/eunit_surefire.erl b/lib/eunit/src/eunit_surefire.erl
index 2d1f0b1497..d6684f33cb 100644
--- a/lib/eunit/src/eunit_surefire.erl
+++ b/lib/eunit/src/eunit_surefire.erl
@@ -206,6 +206,7 @@ handle_cancel(test, Data, St) ->
 format_name({Module, Function, Arity}, Line) ->
     lists:flatten([atom_to_list(Module), ":", atom_to_list(Function), "/",
 		   integer_to_list(Arity), "_", integer_to_list(Line)]).
+
 format_desc(undefined) ->
     "";
 format_desc(Desc) when is_binary(Desc) ->
@@ -279,7 +280,7 @@ write_report_to(TestSuite, FileDescriptor) ->
 %% Write the XML header.
 %% ----------------------------------------------------------------------------
 write_header(FileDescriptor) ->
-    file:write(FileDescriptor, [<<"<?xml version=\"1.0\" encoding=\"UTF-8\" ?>">>, ?NEWLINE]).
+    io:format(FileDescriptor, "~ts~ts", [<<"<?xml version=\"1.0\" encoding=\"UTF-8\" ?>">>, ?NEWLINE]).
 
 %% ----------------------------------------------------------------------------
 %% Write the testsuite start tag, with attributes describing the statistics
@@ -303,7 +304,7 @@ write_start_tag(
         <<"\" time=\"">>, format_time(Time),
         <<"\" name=\"">>, escape_attr(Name),
         <<"\">">>, ?NEWLINE],
-    file:write(FileDescriptor, StartTag).
+    io:format(FileDescriptor, "~ts", [StartTag]).
 
 %% ----------------------------------------------------------------------------
 %% Recursive function to write the test cases.
@@ -317,7 +318,7 @@ write_testcases([TestCase| Tail], FileDescriptor) ->
 %% Write the testsuite end tag.
 %% ----------------------------------------------------------------------------
 write_end_tag(FileDescriptor) ->
-    file:write(FileDescriptor, [<<"</testsuite>">>, ?NEWLINE]).
+    io:format(FileDescriptor, "~ts~ts", [<<"</testsuite>">>, ?NEWLINE]).
 
 %% ----------------------------------------------------------------------------
 %% Write a test case, as a testcase tag.
@@ -344,7 +345,7 @@ write_testcase(
         {ok, <<>>} -> [<<"/>">>, ?NEWLINE];
         _ -> [<<">">>, ?NEWLINE, format_testcase_result(Result), format_testcase_output(Output), ?INDENT, <<"</testcase>">>, ?NEWLINE]
     end,
-    file:write(FileDescriptor, [StartTag, ContentAndEndTag]).
+    io:format(FileDescriptor, "~ts~ts", [StartTag, ContentAndEndTag]).
 
 %% ----------------------------------------------------------------------------
 %% Format the result of the test.
@@ -427,7 +428,7 @@ escape_suitename([Char | Tail], Acc) -> escape_suitename(Tail, [Char | Acc]).
 %% Replace < with &lt;, > with &gt; and & with &amp;
 %% ----------------------------------------------------------------------------
 escape_text(Text) when is_binary(Text) -> escape_text(binary_to_list(Text));
-escape_text(Text) -> escape_xml(lists:flatten(Text), [], false).
+escape_text(Text) -> escape_xml(to_utf8(lists:flatten(Text)), [], false).
 
 
 %% ----------------------------------------------------------------------------
@@ -435,7 +436,7 @@ escape_text(Text) -> escape_xml(lists:flatten(Text), [], false).
 %% Replace < with &lt;, > with &gt; and & with &amp;
 %% ----------------------------------------------------------------------------
 escape_attr(Text) when is_binary(Text) -> escape_attr(binary_to_list(Text));
-escape_attr(Text) -> escape_xml(lists:flatten(Text), [], true).
+escape_attr(Text) -> escape_xml(to_utf8(lists:flatten(Text)), [], true).
 
 escape_xml([], Acc, _ForAttr) -> lists:reverse(Acc);
 escape_xml([$< | Tail], Acc, ForAttr) -> escape_xml(Tail, [$;, $t, $l, $& | Acc], ForAttr);
@@ -443,3 +444,17 @@ escape_xml([$> | Tail], Acc, ForAttr) -> escape_xml(Tail, [$;, $t, $g, $& | Acc]
 escape_xml([$& | Tail], Acc, ForAttr) -> escape_xml(Tail, [$;, $p, $m, $a, $& | Acc], ForAttr);
 escape_xml([$" | Tail], Acc, true) -> escape_xml(Tail, [$;, $t, $o, $u, $q, $& | Acc], true); % "
 escape_xml([Char | Tail], Acc, ForAttr) when is_integer(Char) -> escape_xml(Tail, [Char | Acc], ForAttr).
+
+%% the input may be utf8 or latin1; the resulting list is unicode
+to_utf8(Desc) when is_binary(Desc) ->
+	case unicode:characters_to_list(Desc) of
+		{_,_,_} -> unicode:characters_to_list(Desc, latin1);
+		X -> X
+	end;
+to_utf8(Desc) when is_list(Desc) ->
+	try
+		to_utf8(list_to_binary(Desc))
+	catch
+		_:_ ->
+			Desc
+	end.