Fix a Unicode filename bug affecting Leex and Yecc

A bug where Unicode filenames combined with Latin-1 encoding could
crash Yecc and Leex has been fixed.

5 files changed, 149 insertions, 18 deletions

diff --git a/lib/parsetools/doc/src/leex.xml b/lib/parsetools/doc/src/leex.xml
index d5c24c303d..b4e2af6857 100644
--- a/lib/parsetools/doc/src/leex.xml
+++ b/lib/parsetools/doc/src/leex.xml
@@ -4,7 +4,7 @@
 <erlref>
   <header>
     <copyright>
-      <year>2009</year><year>2011</year>
+      <year>2009</year><year>2013</year>
       <holder>Ericsson AB. All Rights Reserved.</holder>
     </copyright>
     <legalnotice>
@@ -38,19 +38,21 @@ Token = tuple()</code>
   </section>
   <funcs>
     <func>
-      <name>file(FileName) -> ok | error</name>
-      <name>file(FileName, Options) -> ok | error</name>
+      <name>file(FileName, [, Options]) -> LeexRet</name>
       <fsummary>Generate a lexical analyzer</fsummary>
       <type>
         <v>FileName = filename()</v>
         <v>Options = Option | [Option]</v>
         <v>Option =&nbsp;-&nbsp;see below&nbsp;-</v>
-        <v>FileReturn = {ok, Scannerfile} 
-                      | {ok, Scannerfile, Warnings}
-                      | error
-                      | {error, Warnings, Errors}</v>
+        <v>LeexRet = {ok, Scannerfile}
+                   | {ok, Scannerfile, Warnings}
+                   | error
+                   | {error, Warnings, Errors}</v>
         <v>Scannerfile = filename()</v>
         <v>Warnings = Errors = [{filename(), [ErrorInfo]}]</v>
+        <v>ErrorInfo  = {ErrorLine, module(), Reason}</v>
+        <v>ErrorLine = integer()</v>
+        <v>Reason =&nbsp;-&nbsp;formatable by format_error/1&nbsp;-</v>
       </type>
       <desc>
         <p>Generates a lexical analyzer from the definition in the input
diff --git a/lib/parsetools/src/leex.erl b/lib/parsetools/src/leex.erl
index e531b78a5b..7039aea1ae 100644
--- a/lib/parsetools/src/leex.erl
+++ b/lib/parsetools/src/leex.erl
@@ -1645,10 +1645,14 @@ output_encoding_comment(File, #leex{encoding = Encoding}) ->
 
 output_file_directive(File, Filename, Line) ->
     io:fwrite(File, <<"-file(~ts, ~w).\n">>,
-              [format_filename(Filename), Line]).
+              [format_filename(Filename, File), Line]).
 
-format_filename(Filename) ->
-    io_lib:write_string(filename:flatten(Filename)).
+format_filename(Filename0, File) ->
+    Filename = filename:flatten(Filename0),
+    case lists:keyfind(encoding, 1, io:getopts(File)) of
+        {encoding, unicode} -> io_lib:write_string(Filename);
+        _ ->                   io_lib:write_string_as_latin1(Filename)
+    end.
 
 quote($^)  -> "\\^";
 quote($.)  -> "\\.";
diff --git a/lib/parsetools/src/yecc.erl b/lib/parsetools/src/yecc.erl
index f9207d926e..b698beb558 100644
--- a/lib/parsetools/src/yecc.erl
+++ b/lib/parsetools/src/yecc.erl
@@ -482,7 +482,7 @@ generate(St0) ->
     F = case member(time, St1#yecc.options) of
             true -> 
                 io:fwrite(<<"Generating parser from grammar in ~ts\n">>,
-                          [format_filename(St1#yecc.infile)]),
+                          [format_filename(St1#yecc.infile, St1)]),
                 fun timeit/3;
             false ->
                 fun(_Name, Fn, St) -> Fn(St) end
@@ -2519,7 +2519,7 @@ output_encoding_comment(#yecc{encoding = Encoding}=St) ->
 
 output_file_directive(St, Filename, Line) when St#yecc.file_attrs ->
     fwrite(St, <<"-file(~ts, ~w).\n">>,
-           [format_filename(Filename), Line]);
+           [format_filename(Filename, St), Line]);
 output_file_directive(St, _Filename, _Line) ->
     St.
 
@@ -2547,8 +2547,12 @@ nl(#yecc{outport = Outport, line = Line}=St) ->
     io:nl(Outport),
     St#yecc{line = Line + 1}.
 
-format_filename(Filename) ->
-    io_lib:write_string(filename:flatten(Filename)).
+format_filename(Filename0, St) ->
+    Filename = filename:flatten(Filename0),
+    case lists:keyfind(encoding, 1, io:getopts(St#yecc.outport)) of
+        {encoding, unicode} -> io_lib:write_string(Filename);
+        _ ->                   io_lib:write_string_as_latin1(Filename)
+    end.
 
 format_assoc(left) ->
     "Left";
diff --git a/lib/parsetools/test/leex_SUITE.erl b/lib/parsetools/test/leex_SUITE.erl
index afedd79a4e..7cbc72accb 100644
--- a/lib/parsetools/test/leex_SUITE.erl
+++ b/lib/parsetools/test/leex_SUITE.erl
@@ -45,7 +45,7 @@
 	 
 	 pt/1, man/1, ex/1, ex2/1, not_yet/1,
 
-         otp_10302/1]).
+         otp_10302/1, otp_11286/1]).
 
 % Default timetrap timeout (set in init_per_testcase).
 -define(default_timeout, ?t:minutes(1)).
@@ -67,7 +67,7 @@ all() ->
 groups() -> 
     [{checks, [], [file, compile, syntax]},
      {examples, [], [pt, man, ex, ex2, not_yet]},
-     {tickets, [], [otp_10302]}].
+     {tickets, [], [otp_10302, otp_11286]}].
 
 init_per_suite(Config) ->
     Config.
@@ -983,6 +983,68 @@ otp_10302(Config) when is_list(Config) ->
 
     ok.
 
+otp_11286(doc) ->
+    "OTP-11286. A Unicode filename bug; both Leex and Yecc.";
+otp_11286(suite) -> [];
+otp_11286(Config) when is_list(Config) ->
+    Node = start_node(otp_11286, "+fnu"),
+    Dir = ?privdir,
+    UName = [1024] ++ "u",
+    UDir = filename:join(Dir, UName),
+    ok = rpc:call(Node, file, make_dir, [UDir]),
+
+    %% Note: Cannot use UName as filename since the filename is used
+    %% as module name. To be fixed in R18.
+    Filename = filename:join(UDir, 'OTP-11286.xrl'),
+    Scannerfile = filename:join(UDir, 'OTP-11286.erl'),
+    Options = [return, {scannerfile, Scannerfile}],
+
+    Mini1 = <<"%% coding: utf-8\n"
+              "Definitions.\n"
+              "D  = [0-9]\n"
+              "Rules.\n"
+              "{L}+  : {token,{word,TokenLine,TokenChars}}.\n"
+              "Erlang code.\n">>,
+    ok = rpc:call(Node, file, write_file, [Filename, Mini1]),
+    {ok, _, []} = rpc:call(Node, leex, file, [Filename, Options]),
+    {ok,_,_} = rpc:call(Node, compile, file,
+                  [Scannerfile,[basic_validation,return]]),
+
+    Mini2 = <<"Definitions.\n"
+              "D  = [0-9]\n"
+              "Rules.\n"
+              "{L}+  : {token,{word,TokenLine,TokenChars}}.\n"
+              "Erlang code.\n">>,
+    ok = rpc:call(Node, file, write_file, [Filename, Mini2]),
+    {ok, _, []} = rpc:call(Node, leex, file, [Filename, Options]),
+    {ok,_,_} = rpc:call(Node, compile, file,
+                  [Scannerfile,[basic_validation,return]]),
+
+    Mini3 = <<"%% coding: latin-1\n"
+              "Definitions.\n"
+              "D  = [0-9]\n"
+              "Rules.\n"
+              "{L}+  : {token,{word,TokenLine,TokenChars}}.\n"
+              "Erlang code.\n">>,
+    ok = rpc:call(Node, file, write_file, [Filename, Mini3]),
+    {ok, _, []} = rpc:call(Node, leex, file, [Filename, Options]),
+    {ok,_,_} = rpc:call(Node, compile, file,
+                  [Scannerfile,[basic_validation,return]]),
+
+    true = test_server:stop_node(Node),
+    ok.
+
+start_node(Name, Args) ->
+    [_,Host] = string:tokens(atom_to_list(node()), "@"),
+    ct:log("Trying to start ~w@~s~n", [Name,Host]),
+    case test_server:start_node(Name, peer, [{args,Args}]) of
+	{error,Reason} ->
+	    test_server:fail(Reason);
+	{ok,Node} ->
+	    ct:log("Node ~p started~n", [Node]),
+	    Node
+    end.
+
 unwritable(Fname) ->
     {ok, Info} = file:read_file_info(Fname),
     Mode = Info#file_info.mode - 8#00200,
diff --git a/lib/parsetools/test/yecc_SUITE.erl b/lib/parsetools/test/yecc_SUITE.erl
index 9c865a1ec6..c7ac9fd232 100644
--- a/lib/parsetools/test/yecc_SUITE.erl
+++ b/lib/parsetools/test/yecc_SUITE.erl
@@ -49,7 +49,8 @@
 	 
 	 otp_5369/1, otp_6362/1, otp_7945/1, otp_8483/1, otp_8486/1,
 	 
-	 otp_7292/1, otp_7969/1, otp_8919/1, otp_10302/1, otp_11269/1]).
+	 otp_7292/1, otp_7969/1, otp_8919/1, otp_10302/1, otp_11269/1,
+         otp_11286/1]).
 
 % Default timetrap timeout (set in init_per_testcase).
 -define(default_timeout, ?t:minutes(1)).
@@ -77,7 +78,7 @@ groups() ->
      {bugs, [],
       [otp_5369, otp_6362, otp_7945, otp_8483, otp_8486]},
      {improvements, [], [otp_7292, otp_7969, otp_8919, otp_10302,
-                         otp_11269]}].
+                         otp_11269, otp_11286]}].
 
 init_per_suite(Config) ->
     Config.
@@ -1996,6 +1997,64 @@ otp_11269(Config) when is_list(Config) ->
     {ok,'OTP-11269',_Warnings} = compile:file(ErlFile, Opts),
     ok.
 
+otp_11286(doc) ->
+    "OTP-11286. A Unicode filename bug; both Leex and Yecc.";
+otp_11286(suite) -> [];
+otp_11286(Config) when is_list(Config) ->
+    Node = start_node(otp_11286, "+fnu"),
+    Dir = ?privdir,
+    UName = [1024] ++ "u",
+    UDir = filename:join(Dir, UName),
+    ok = rpc:call(Node, file, make_dir, [UDir]),
+
+    %% Note: Cannot use UName as filename since the filename is used
+    %% as module name. To be fixed in R18.
+    Filename = filename:join(UDir, 'OTP-11286.yrl'),
+    Ret = [return, {report, false}, time],
+
+    Mini1 = <<"%% coding: utf-8
+               Terminals t.
+               Nonterminals nt.
+               Rootsymbol  nt.
+               nt -> t.">>,
+    ok = rpc:call(Node, file, write_file, [Filename, Mini1]),
+    {ok,ErlFile,[]} = rpc:call(Node, yecc, file, [Filename, Ret]),
+    Opts = [return, warn_unused_vars,{outdir,Dir}],
+    {ok,_,_Warnings} = rpc:call(Node, compile, file, [ErlFile, Opts]),
+
+    Mini2 = <<"Terminals t.
+               Nonterminals nt.
+               Rootsymbol  nt.
+               nt -> t.">>,
+    ok = rpc:call(Node, file, write_file, [Filename, Mini2]),
+    {ok,ErlFile,[]} = rpc:call(Node, yecc, file, [Filename, Ret]),
+    Opts = [return, warn_unused_vars,{outdir,Dir}],
+    {ok,_,_Warnings} = rpc:call(Node, compile, file, [ErlFile, Opts]),
+
+    Mini3 = <<"%% coding: latin-1
+               Terminals t.
+               Nonterminals nt.
+               Rootsymbol  nt.
+               nt -> t.">>,
+    ok = rpc:call(Node, file, write_file, [Filename, Mini3]),
+    {ok,ErlFile,[]} = rpc:call(Node, yecc, file, [Filename, Ret]),
+    Opts = [return, warn_unused_vars,{outdir,Dir}],
+    {ok,_,_Warnings} = rpc:call(Node, compile, file, [ErlFile, Opts]),
+
+    true = test_server:stop_node(Node),
+    ok.
+
+start_node(Name, Args) ->
+    [_,Host] = string:tokens(atom_to_list(node()), "@"),
+    ct:log("Trying to start ~w@~s~n", [Name,Host]),
+    case test_server:start_node(Name, peer, [{args,Args}]) of
+	{error,Reason} ->
+	    test_server:fail(Reason);
+	{ok,Node} ->
+	    ct:log("Node ~p started~n", [Node]),
+	    Node
+    end.
+
 yeccpre_size() ->
     yeccpre_size(default_yeccpre()).