Teach filelib to use re in unicode mode when filenames are not raw

4 files changed, 79 insertions, 39 deletions

diff --git a/erts/emulator/drivers/win32/win_efile.c b/erts/emulator/drivers/win32/win_efile.c
index d0f13437dc..4ec9579529 100755
--- a/erts/emulator/drivers/win32/win_efile.c
+++ b/erts/emulator/drivers/win32/win_efile.c
@@ -1519,9 +1519,6 @@ efile_altname(Efile_error* errInfo, char* orig_name, char* buffer, size_t size)
     return 1;
 }
 
-/*
- * XXX: link and symlink Implemented in pu (?), will need conversion
- */
 
 int
 efile_link(Efile_error* errInfo, char* old, char* new)
diff --git a/erts/test/z_SUITE.erl b/erts/test/z_SUITE.erl
index 8faddeb0d3..9f13a7083d 100644
--- a/erts/test/z_SUITE.erl
+++ b/erts/test/z_SUITE.erl
@@ -253,6 +253,8 @@ core_file_search(#core_search_conf{search_dir = Base,
 				     core_cand(Conf, Core, Cores);
 				 "core." ++ _ ->
 				     core_cand(Conf, Core, Cores);
+				 Bin when is_binary(Bin) -> %Icky filename; ignore
+				     Cores;
 				 BName ->
 				     case lists:suffix(".core", BName) of
 					 true -> core_cand(Conf, Core, Cores);
diff --git a/lib/kernel/test/file_name_SUITE.erl b/lib/kernel/test/file_name_SUITE.erl
index 3faae93bea..d91bc32d11 100644
--- a/lib/kernel/test/file_name_SUITE.erl
+++ b/lib/kernel/test/file_name_SUITE.erl
@@ -140,36 +140,46 @@ icky(suite) ->
 icky(doc) ->
     "Check file operations on normal file names regardless of unicode mode";
 icky(Config) when is_list(Config) ->
-    {ok,Dir} = file:get_cwd(),
-    try
-	Priv = ?config(priv_dir, Config),
-	file:set_cwd(Priv),
-	put(file_module,prim_file),
-	ok = check_icky(prim_file),
-	put(file_module,file),
-	ok = check_icky(file)
-    after
-	file:set_cwd(Dir)
+    case hopeless_darwin() of
+	true ->
+	    {skipped,"This version of darwin does not support icky names at all."};
+	false ->
+	    {ok,Dir} = file:get_cwd(),
+	    try
+		Priv = ?config(priv_dir, Config),
+		file:set_cwd(Priv),
+		put(file_module,prim_file),
+		ok = check_icky(prim_file),
+		put(file_module,file),
+		ok = check_icky(file)
+	    after
+		file:set_cwd(Dir)
+	    end
     end.
 very_icky(suite) ->
     [];
 very_icky(doc) ->
     "Check file operations on normal file names regardless of unicode mode";
 very_icky(Config) when is_list(Config) ->
-    {ok,Dir} = file:get_cwd(),
-    try
-	Priv = ?config(priv_dir, Config),
-	file:set_cwd(Priv),
-	put(file_module,prim_file),
-	case check_very_icky(prim_file) of
-	    need_unicode_mode ->
-		{skipped,"VM needs to be started in Unicode filename mode"};
-	    ok ->
-		put(file_module,file),
-		ok = check_very_icky(file)
-	end
-    after
-	file:set_cwd(Dir)
+    case hopeless_darwin() of
+	true ->
+	    {skipped,"This version of darwin does not support icky names at all."};
+	false ->
+	    {ok,Dir} = file:get_cwd(),
+	    try
+		Priv = ?config(priv_dir, Config),
+		file:set_cwd(Priv),
+		put(file_module,prim_file),
+		case check_very_icky(prim_file) of
+		    need_unicode_mode ->
+			{skipped,"VM needs to be started in Unicode filename mode"};
+		    ok ->
+			put(file_module,file),
+			ok = check_very_icky(file)
+		end
+	    after
+		file:set_cwd(Dir)
+	    end
     end.
     
 
@@ -481,6 +491,21 @@ check_very_icky(Mod) ->
 				       FI#file_info{mode = NewMode2}),
  	?line {ok,#file_info{mode = NewMode2}} = 
 	          Mod:read_file_info([956,965,963,954,959,49]),
+	?line NumOK0 = case has_links() of
+			  true -> 5;
+			  false -> 3
+		      end,
+	?line NumNOK0 = case has_links() of
+			   true -> 4;
+			   false -> 3
+		       end,
+	?line {NumOK,NumNOK} = case is_binary(treat_icky(<<"foo">>)) of
+				   false ->
+				       {NumOK0+NumNOK0,0};
+				   true ->
+				       {NumOK0,NumNOK0}
+			       end,
+	?line {NumOK,NumNOK} = filelib:fold_files(".",".*",true,fun(_F,{N,M}) when is_list(_F) ->  io:format("~ts~n",[_F]),{N+1,M}; (_F,{N,M}) ->  io:format("~p~n",[_F]),{N,M+1} end,{0,0}),
 	ok
     catch
 	throw:need_unicode_mode ->
@@ -548,6 +573,7 @@ rm_r2(Mod,Dir) ->
 chk_cre_dir(_,[]) ->
     ok;
 chk_cre_dir(Mod,[{regular,Name,Content}|T]) ->
+    %io:format("~p~n",[Name]),
     ok = Mod:write_file(Name,Content),
     chk_cre_dir(Mod,T);
 chk_cre_dir(Mod,[{link,Name,Target}|T]) ->
@@ -558,7 +584,9 @@ chk_cre_dir(Mod,[{symlink,Name,Target}|T]) ->
     chk_cre_dir(Mod,T);
 chk_cre_dir(Mod,[{directory,Name,Content}|T]) ->
     ok = Mod:make_dir(Name),
+    %io:format("Content = ~p~n",[Content]),
     Content2 = [{Ty,filename:join(Name,N),case Ty of link -> filename:join(Name,C); _ -> C end} || {Ty,N,C} <- Content ],
+    %io:format("Content2 = ~p~n",[Content2]),
     chk_cre_dir(Mod,Content2),
     chk_cre_dir(Mod,T).
  
@@ -627,6 +655,14 @@ linkify(Passed,[{directory, Name, Content}|T]) ->
 linkify(Passed,[H|T]) ->
     [H|linkify([H|Passed],T)].
 
+hopeless_darwin() ->
+    case {os:type(),os:version()} of
+	{{unix,darwin},{Major,_,_}} when Major < 9 ->
+	    true;
+	_ ->
+	    false
+    end.
+
 icky_dir() ->
     [{regular,"fil1","fil1"},
      {regular,"���2","���2"}] ++
diff --git a/lib/stdlib/src/filelib.erl b/lib/stdlib/src/filelib.erl
index d5ddf9ed7e..eaaca750ab 100644
--- a/lib/stdlib/src/filelib.erl
+++ b/lib/stdlib/src/filelib.erl
@@ -166,36 +166,41 @@ do_is_regular(File, Mod) ->
 %%   If <Recursive> is true all sub-directories to <Dir> are processed
 
 do_fold_files(Dir, RegExp, Recursive, Fun, Acc, Mod) ->
-    {ok, Re1} = re:compile(RegExp),
-    do_fold_files1(Dir, Re1, Recursive, Fun, Acc, Mod).
+    {ok, Re1} = re:compile(RegExp,[unicode]),
+    do_fold_files1(Dir, Re1, RegExp, Recursive, Fun, Acc, Mod).
 
-do_fold_files1(Dir, RegExp, Recursive, Fun, Acc, Mod) ->
+do_fold_files1(Dir, RegExp, OrigRE, Recursive, Fun, Acc, Mod) ->
     case eval_list_dir(Dir, Mod) of
-	{ok, Files} -> do_fold_files2(Files, Dir, RegExp, Recursive, Fun, Acc, Mod);
+	{ok, Files} -> do_fold_files2(Files, Dir, RegExp, OrigRE,
+				      Recursive, Fun, Acc, Mod);
 	{error, _}  -> Acc
     end.
 
-do_fold_files2([], _Dir, _RegExp, _Recursive, _Fun, Acc, _Mod) -> 
+%% OrigRE is not to be compiled as it's for non conforming filenames,
+%% i.e. for filenames that does not comply to the current encoding, which should
+%% be very rare. We use it only in those cases and do not want to precompile.
+do_fold_files2([], _Dir, _RegExp, _OrigRE, _Recursive, _Fun, Acc, _Mod) -> 
     Acc;
-do_fold_files2([File|T], Dir, RegExp, Recursive, Fun, Acc0, Mod) ->
+do_fold_files2([File|T], Dir, RegExp, OrigRE, Recursive, Fun, Acc0, Mod) ->
     FullName = filename:join(Dir, File),
     case do_is_regular(FullName, Mod) of
 	true  ->
-	    case re:run(File, RegExp, [{capture,none}]) of
+	    case re:run(File, if is_binary(File) -> OrigRE; true -> RegExp end, 
+			[{capture,none}]) of
 		match  -> 
 		    Acc = Fun(FullName, Acc0),
-		    do_fold_files2(T, Dir, RegExp, Recursive, Fun, Acc, Mod);
+		    do_fold_files2(T, Dir, RegExp, OrigRE, Recursive, Fun, Acc, Mod);
 		nomatch ->
-		    do_fold_files2(T, Dir, RegExp, Recursive, Fun, Acc0, Mod)
+		    do_fold_files2(T, Dir, RegExp, OrigRE, Recursive, Fun, Acc0, Mod)
 	    end;
 	false ->
 	    case Recursive andalso do_is_dir(FullName, Mod) of
 		true ->
-		    Acc1 = do_fold_files1(FullName, RegExp, Recursive,
+		    Acc1 = do_fold_files1(FullName, RegExp, OrigRE, Recursive,
 					  Fun, Acc0, Mod),
-		    do_fold_files2(T, Dir, RegExp, Recursive, Fun, Acc1, Mod);
+		    do_fold_files2(T, Dir, RegExp, OrigRE, Recursive, Fun, Acc1, Mod);
 		false ->
-		    do_fold_files2(T, Dir, RegExp, Recursive, Fun, Acc0, Mod)
+		    do_fold_files2(T, Dir, RegExp, OrigRE, Recursive, Fun, Acc0, Mod)
 	    end
     end.