Add utf8 atom encode and decode testcases

5 files changed, 500 insertions, 17 deletions

diff --git a/lib/erl_interface/test/all_SUITE_data/init_tc.erl b/lib/erl_interface/test/all_SUITE_data/init_tc.erl
index 7e370c566a..0f2ac215f9 100644
--- a/lib/erl_interface/test/all_SUITE_data/init_tc.erl
+++ b/lib/erl_interface/test/all_SUITE_data/init_tc.erl
@@ -40,7 +40,7 @@ run([]) ->
 run1(Name) ->
     CFile = Name ++ ".c",
     {ok, Bin} = file:read_file(CFile),
-    RE = "\nTESTCASE\\(([_a-zA-Z]*)\\)",
+    RE = "\nTESTCASE\\(([_a-zA-Z0-9]*)\\)",
     {match, Cases0} = re:run(Bin, RE, [{capture,all_but_first,list},global]),
     Cases = lists:concat(Cases0),
     generate(Name, Cases).
diff --git a/lib/erl_interface/test/ei_decode_SUITE.erl b/lib/erl_interface/test/ei_decode_SUITE.erl
index 2c4b6e5541..5a94994912 100644
--- a/lib/erl_interface/test/ei_decode_SUITE.erl
+++ b/lib/erl_interface/test/ei_decode_SUITE.erl
@@ -27,14 +27,16 @@
 -export(
    [
     all/0, suite/0,groups/0,init_per_suite/1, end_per_suite/1, 
-    init_per_group/2,end_per_group/2,
+    init_per_group/2,end_per_group/2, init_per_testcase/2,
+    end_per_testcase/2,
     test_ei_decode_long/1,
     test_ei_decode_ulong/1,
     test_ei_decode_longlong/1,
     test_ei_decode_ulonglong/1,
     test_ei_decode_char/1,
     test_ei_decode_nonoptimal/1,
-    test_ei_decode_misc/1
+    test_ei_decode_misc/1,
+    test_ei_decode_utf8_atom/1
    ]).
 
 suite() -> [{ct_hooks,[ts_install_cth]}].
@@ -43,7 +45,7 @@ all() ->
     [test_ei_decode_long, test_ei_decode_ulong,
      test_ei_decode_longlong, test_ei_decode_ulonglong,
      test_ei_decode_char, test_ei_decode_nonoptimal,
-     test_ei_decode_misc].
+     test_ei_decode_misc, test_ei_decode_utf8_atom].
 
 groups() -> 
     [].
@@ -60,6 +62,11 @@ init_per_group(_GroupName, Config) ->
 end_per_group(_GroupName, Config) ->
     Config.
 
+init_per_testcase(_TC, Config) ->
+    Config.
+
+end_per_testcase(_RC, Config) ->
+    Config.
 
 %% ---------------------------------------------------------------------------
 
@@ -221,6 +228,29 @@ test_ei_decode_misc(Config) when is_list(Config) ->
     ?line runner:recv_eot(P),
     ok.
 
+%% ######################################################################## %%
+
+test_ei_decode_utf8_atom(Config) ->
+    ?line P = runner:start(?test_ei_decode_utf8_atom),
+
+    send_utf8_atom_as_binary(P,"å"),
+    send_utf8_atom_as_binary(P,"ä"),
+    send_term_as_binary(P,'ö'),
+    send_term_as_binary(P,'õ'),
+    
+    ?line send_utf8_atom_as_binary(P,[1758]),
+    ?line send_utf8_atom_as_binary(P,[1758,1758]),
+    ?line send_utf8_atom_as_binary(P,[1758,1758,1758]),
+    ?line send_utf8_atom_as_binary(P,[1758,1758,1758,1758]),
+
+    send_utf8_atom_as_binary(P,"a"),
+    send_utf8_atom_as_binary(P,"b"),
+    send_term_as_binary(P,'c'),
+    send_term_as_binary(P,'d'),
+
+    ?line runner:recv_eot(P),
+    ok.
+
 
 %% ######################################################################## %%
 
@@ -230,6 +260,8 @@ send_term_as_binary(Port, Term) when is_port(Port) ->
 send_raw(Port, Bin) when is_port(Port) ->
     Port ! {self(), {command, Bin}}.
 
+send_utf8_atom_as_binary(Port, String) ->
+    Port ! {self(), {command, term_to_binary(uc_atup(String))}}.
 
 send_integers(P) ->
     ?line send_term_as_binary(P,0),		% SMALL_INTEGER_EXT smallest
@@ -304,3 +336,43 @@ send_integers2(P) ->
     ?line send_term_as_binary(P, 16#ffffffffffffffff), % largest  u64
     ?line send_term_as_binary(P, []), % illegal type
     ok.
+
+uc_atup(ATxt) ->
+    string_to_atom(ATxt).
+
+string_to_atom(String) ->
+    Utf8List = string_to_utf8_list(String),
+    Len = length(Utf8List),
+    TagLen = case Len < 256 of
+		 true -> [119, Len];
+		 false -> [118, Len bsr 8, Len band 16#ff]
+	     end,
+    binary_to_term(list_to_binary([131, TagLen, Utf8List])).
+
+string_to_utf8_list([]) ->
+    [];
+string_to_utf8_list([CP|CPs]) when is_integer(CP),
+				   0 =< CP,
+				   CP =< 16#7F ->
+    [CP | string_to_utf8_list(CPs)];
+string_to_utf8_list([CP|CPs]) when is_integer(CP),
+				   16#80 =< CP,
+				   CP =< 16#7FF ->
+    [16#C0 bor (CP bsr 6),
+     16#80 bor (16#3F band CP)
+     | string_to_utf8_list(CPs)];
+string_to_utf8_list([CP|CPs]) when is_integer(CP),
+				   16#800 =< CP,
+				   CP =< 16#FFFF ->
+    [16#E0 bor (CP bsr 12),
+     16#80 bor (16#3F band (CP bsr 6)),
+     16#80 bor (16#3F band CP)
+     | string_to_utf8_list(CPs)];
+string_to_utf8_list([CP|CPs]) when is_integer(CP),
+				   16#10000 =< CP,
+				   CP =< 16#10FFFF ->
+    [16#F0 bor (CP bsr 18),
+     16#80 bor (16#3F band (CP bsr 12)),
+     16#80 bor (16#3F band (CP bsr 6)),
+     16#80 bor (16#3F band CP)
+     | string_to_utf8_list(CPs)].
diff --git a/lib/erl_interface/test/ei_decode_SUITE_data/ei_decode_test.c b/lib/erl_interface/test/ei_decode_SUITE_data/ei_decode_test.c
index b349138ae9..6db04aa676 100644
--- a/lib/erl_interface/test/ei_decode_SUITE_data/ei_decode_test.c
+++ b/lib/erl_interface/test/ei_decode_SUITE_data/ei_decode_test.c
@@ -17,6 +17,8 @@
  * %CopyrightEnd%
  */
 
+#include <string.h>
+
 #ifdef VXWORKS
 #include "reclaim.h"
 #endif
@@ -38,6 +40,22 @@
              err, size1, SIZE, (EI_LONGLONG)p); 
 #endif
 
+struct my_atom {
+  enum erlang_char_encoding from;
+  enum erlang_char_encoding was_check;
+  enum erlang_char_encoding result_check;
+};
+
+/* Allow arrays constants to be part of macro arguments */
+#define P99(...) __VA_ARGS__ 
+
+int ei_decode_my_atom_as(const char *buf, int *index, char *to,
+			 struct my_atom *atom);
+int ei_decode_my_atom(const char *buf, int *index, char *to,
+		      struct my_atom *atom);
+int ei_decode_my_string(const char *buf, int *index, char *to,
+			struct my_atom *atom);
+
 #define EI_DECODE_2(FUNC,SIZE,TYPE,VAL) \
   { \
     TYPE p; \
@@ -129,17 +147,26 @@
     } \
   } \
 
-#define EI_DECODE_STRING(FUNC,SIZE,VAL) \
+#define dump(arr, num) {	    \
+    int i;		    \
+    message("Dumping " #arr ": ");			\
+    for (i = 0; i < num; i++) message("%u, ",(unsigned char)arr[i]);	\
+    message("\n");					\
+  }
+  
+#define EI_DECODE_STRING_4(FUNC,SIZE,VAL,ATOM)	\
   { \
     char p[1024]; \
     char *buf; \
+    unsigned char val[] = VAL; \
     int size1 = 0; \
     int size2 = 0; \
     int err; \
-    message("ei_" #FUNC " should be " #VAL); \
+    struct my_atom atom = ATOM; \
+    message("ei_" #FUNC " should be " #VAL "\n"); \
     buf = read_packet(NULL); \
 \
-    err = ei_ ## FUNC(buf+1, &size1, NULL); \
+    err = ei_ ## FUNC(buf+1, &size1, NULL, &atom); \
     message("err = %d, size = %d, expected size = %d\n",err,size1,SIZE); \
     if (err != 0) { \
       if (err != -1) { \
@@ -150,7 +177,7 @@
       return; \
     } \
 \
-    err = ei_ ## FUNC(buf+1, &size2, p); \
+    err = ei_ ## FUNC(buf+1, &size2, p, &atom);	\
     message("err = %d, size = %d, expected size = %d\n",err,size2,SIZE); \
     if (err != 0) { \
       if (err != -1) { \
@@ -161,7 +188,7 @@
       return; \
     } \
 \
-    if (strcmp(p,VAL) != 0) { \
+    if (strcmp(p,val) != 0) { \
       fail("value is not correct"); \
       return; \
     } \
@@ -177,6 +204,51 @@
     } \
   } \
 
+#define EI_DECODE_STRING(FUNC,SIZE,VAL) \
+  EI_DECODE_STRING_4(FUNC,SIZE,VAL, \
+		     P99({ERLANG_ANY,ERLANG_ANY,ERLANG_ANY}))
+
+#define EI_DECODE_STRING_FAIL(FUNC,ATOM) \
+  { \
+    char p[1024]; \
+    char *buf; \
+    int size1 = 0; \
+    int size2 = 0; \
+    int err;					  \
+    struct my_atom atom = ATOM;\
+    message("ei_" #FUNC " should fail\n"); \
+    p[0] = 0;				   \
+    message("p[0] is %d\n",p[0]);		   \
+    buf = read_packet(NULL); \
+\
+    err = ei_ ## FUNC(buf+1, &size1, NULL, &atom);			\
+    if (err != -1) {				\
+      fail("should return -1 if NULL pointer"); \
+      return; \
+    } \
+\
+    err = ei_ ## FUNC(buf+1, &size2, p, &atom);				\
+    if (err != -1) {							\
+      fail("should return -1"); \
+      return; \
+    } \
+    if (p[0] != 0) { \
+      message("p[0] argument was modified to %u\n",(unsigned char)p[0]);	\
+    } \
+\
+    if (size1 != 0) { \
+      fail("size of encoded data should be 0 if NULL"); \
+      return; \
+    } \
+\
+    if (size2 != 0) { \
+      fail("size of encoded data should be 0"); \
+      return; \
+    } \
+  } \
+
+//#define EI_DECODE_UTF8_STRING(FUNC,SIZE,VAL) 
+
 #define EI_DECODE_BIN(FUNC,SIZE,VAL,LEN) \
   { \
     char p[1024]; \
@@ -536,13 +608,13 @@ TESTCASE(test_ei_decode_misc)
     EI_DECODE_2(decode_boolean, 8, int, 0);
     EI_DECODE_2(decode_boolean, 7, int, 1);
 
-    EI_DECODE_STRING(decode_atom, 6, "foo");
-    EI_DECODE_STRING(decode_atom, 3, "");
-    EI_DECODE_STRING(decode_atom, 9, "������");
+    EI_DECODE_STRING(decode_my_atom, 6, "foo");
+    EI_DECODE_STRING(decode_my_atom, 3, "");
+    EI_DECODE_STRING(decode_my_atom, 9, "������");
 
-    EI_DECODE_STRING(decode_string, 6, "foo");
-    EI_DECODE_STRING(decode_string, 1, "");
-    EI_DECODE_STRING(decode_string, 9, "������");
+    EI_DECODE_STRING(decode_my_string, 6, "foo");
+    EI_DECODE_STRING(decode_my_string, 1, "");
+    EI_DECODE_STRING(decode_my_string, 9, "������");
 
     EI_DECODE_BIN(decode_binary,  8, "foo", 3);
     EI_DECODE_BIN(decode_binary,  5, "", 0);
@@ -559,3 +631,63 @@ TESTCASE(test_ei_decode_misc)
 
 /* ******************************************************************** */
 
+TESTCASE(test_ei_decode_utf8_atom)
+{
+
+  EI_DECODE_STRING_4(decode_my_atom_as, 4, P99({229,0}), /* LATIN1 "�" */
+		   P99({ERLANG_ANY,ERLANG_LATIN1,ERLANG_LATIN1}));
+  EI_DECODE_STRING_4(decode_my_atom_as, 4, P99({195,164,0}), /* UTF8 "�" */
+		     P99({ERLANG_UTF8,ERLANG_LATIN1,ERLANG_UTF8}));
+  EI_DECODE_STRING_4(decode_my_atom_as, 4, P99({246,0}), /* LATIN1 "�" */
+		     P99({ERLANG_LATIN1,ERLANG_LATIN1,ERLANG_LATIN1}));
+  EI_DECODE_STRING_FAIL(decode_my_atom_as, 
+			P99({ERLANG_ASCII,ERLANG_ANY,ERLANG_ANY}));
+
+  EI_DECODE_STRING_4(decode_my_atom_as, 4, P99({219,158,0}),
+		     P99({ERLANG_ANY,ERLANG_UTF8,ERLANG_UTF8}));
+  EI_DECODE_STRING_4(decode_my_atom_as, 6, P99({219,158,219,158,0}),
+		     P99({ERLANG_UTF8,ERLANG_UTF8,ERLANG_UTF8}));
+  EI_DECODE_STRING_FAIL(decode_my_atom_as,
+			P99({ERLANG_LATIN1,ERLANG_ANY,ERLANG_ANY}));
+  EI_DECODE_STRING_FAIL(decode_my_atom_as,
+			P99({ERLANG_ASCII,ERLANG_ANY,ERLANG_ANY}));
+
+  EI_DECODE_STRING_4(decode_my_atom_as, 4, "a",
+		   P99({ERLANG_ANY,ERLANG_LATIN1,ERLANG_ASCII}));
+  EI_DECODE_STRING_4(decode_my_atom_as, 4, "b",
+		     P99({ERLANG_UTF8,ERLANG_LATIN1,ERLANG_ASCII}));
+  EI_DECODE_STRING_4(decode_my_atom_as, 4, "c",
+		     P99({ERLANG_LATIN1,ERLANG_LATIN1,ERLANG_ASCII}));
+  EI_DECODE_STRING_4(decode_my_atom_as, 4, "d",
+		     P99({ERLANG_ASCII,ERLANG_LATIN1,ERLANG_ASCII}));
+
+  report(1);
+}
+
+/* ******************************************************************** */
+
+int ei_decode_my_atom_as(const char *buf, int *index, char *to,
+			 struct my_atom *atom) {
+  enum erlang_char_encoding was,result;
+  int res = ei_decode_atom_as(buf,index,to,1024,atom->from,&was,&result);
+  if (res != 0)
+    return res;
+  if (!(was & atom->was_check)) {
+    message("Original encoding was %d not %d\n",was,atom->was_check);
+    return -1;
+  } else if (!(result & atom->result_check)) {
+    message("Result encoding was %d not %d\n",result,atom->result_check);
+    return -1;
+  }
+  return res;
+}
+
+int ei_decode_my_atom(const char *buf, int *index, char *to, 
+		      struct my_atom *atom) {
+  return ei_decode_atom(buf, index, to);
+}
+
+int ei_decode_my_string(const char *buf, int *index, char *to, 
+			struct my_atom *atom) {
+  return ei_decode_string(buf, index, to);
+}
diff --git a/lib/erl_interface/test/ei_encode_SUITE.erl b/lib/erl_interface/test/ei_encode_SUITE.erl
index 537e9cb01c..e0ac4c0637 100644
--- a/lib/erl_interface/test/ei_encode_SUITE.erl
+++ b/lib/erl_interface/test/ei_encode_SUITE.erl
@@ -34,7 +34,9 @@
     test_ei_encode_ulonglong/1,
     test_ei_encode_char/1,
     test_ei_encode_misc/1,
-    test_ei_encode_fails/1
+    test_ei_encode_fails/1,
+    test_ei_encode_utf8_atom/1,
+    test_ei_encode_utf8_atom_len/1
    ]).
 
 suite() -> [{ct_hooks,[ts_install_cth]}].
@@ -43,7 +45,8 @@ all() ->
     [test_ei_encode_long, test_ei_encode_ulong,
      test_ei_encode_longlong, test_ei_encode_ulonglong,
      test_ei_encode_char, test_ei_encode_misc,
-     test_ei_encode_fails].
+     test_ei_encode_fails, test_ei_encode_utf8_atom,
+     test_ei_encode_utf8_atom_len].
 
 groups() -> 
     [].
@@ -257,6 +260,38 @@ test_ei_encode_fails(Config) when is_list(Config) ->
 
 %% ######################################################################## %%
 
+test_ei_encode_utf8_atom(Config) ->
+    ?line P = runner:start(?test_ei_encode_utf8_atom),
+    
+    ?line {<<119,2,195,133>>,'Å'} = get_buf_and_term(P),
+    ?line {<<100,0,1,197>>,'Å'} = get_buf_and_term(P),
+    ?line {<<100,0,1,197>>,'Å'} = get_buf_and_term(P),
+    ?line {<<119,2,195,133>>,'Å'} = get_buf_and_term(P),
+
+    ?line {<<119,1,$A>>,'A'} = get_buf_and_term(P),
+    ?line {<<100,0,1,$A>>,'A'} = get_buf_and_term(P),
+
+    ?line runner:recv_eot(P),
+    ok.
+
+%% ######################################################################## %%
+test_ei_encode_utf8_atom_len(Config) ->
+    ?line P = runner:start(?test_ei_encode_utf8_atom_len),
+    
+    ?line {<<119,2,195,133>>,'Å'} = get_buf_and_term(P),
+    ?line {<<100,0,2,197,196>>,'ÅÄ'} = get_buf_and_term(P),
+    ?line {<<100,0,1,197>>,'Å'} = get_buf_and_term(P),
+    ?line {<<119,4,195,133,195,132>>,'ÅÄ'} = get_buf_and_term(P),
+
+    ?line {<<119,1,$A>>,'A'} = get_buf_and_term(P),
+    ?line {<<100,0,2,$A,$B>>,'AB'} = get_buf_and_term(P),
+    ?line {<<100,0,255,_:(255*8)>>,_} = get_buf_and_term(P),
+
+    ?line runner:recv_eot(P),
+    ok.
+
+%% ######################################################################## %%
+
 % We read two packets for each test, the ei_encode and ei_x_encode version....
 
 get_buf_and_term(P) ->
diff --git a/lib/erl_interface/test/ei_encode_SUITE_data/ei_encode_test.c b/lib/erl_interface/test/ei_encode_SUITE_data/ei_encode_test.c
index c373658152..e904375e9e 100644
--- a/lib/erl_interface/test/ei_encode_SUITE_data/ei_encode_test.c
+++ b/lib/erl_interface/test/ei_encode_SUITE_data/ei_encode_test.c
@@ -208,6 +208,196 @@
     ei_x_free(&arg); \
   }
 
+#define EI_ENCODE_3(FUNC,ARG1,ARG2,ARG3)	\
+  { \
+    char buf[1024]; \
+    int size1 = 0; \
+    int size2 = 0; \
+    int err; \
+    message("ei_" #FUNC " " #ARG1 " " #ARG2 " " #ARG3 " encoded as "); \
+    err = ei_ ## FUNC(NULL, &size1, ARG1, ARG2, ARG3);	     \
+    if (err != 0) { \
+      if (err != -1) { \
+	fail("size calculation returned non zero but not -1"); \
+	return; \
+      } else { \
+	fail("size calculation returned non zero"); \
+	return; \
+      } \
+    } \
+    err = ei_ ## FUNC(buf, &size2, ARG1, ARG2, ARG3); \
+    if (err != 0) { \
+      if (err != -1) { \
+	fail("returned non zero but not -1"); \
+      } else { \
+	fail("returned non zero"); \
+      } \
+      return; \
+    } \
+    if (size1 != size2) { \
+      fail("size differs when arg is NULL or buf"); \
+      return; \
+    } \
+    if (size1 < 1) { \
+      fail("size is < 1"); \
+      return; \
+    } \
+    send_buffer(buf, size1); \
+  } \
+  { \
+    ei_x_buff arg; \
+    int err; \
+    message("ei_x_" #FUNC " " #ARG1 " " #ARG2 " " #ARG3 " encoded as "); \
+    ei_x_new(&arg); \
+    err = ei_x_ ## FUNC(&arg, ARG1, ARG2, ARG3); \
+    if (err != 0) { \
+      if (err != -1) { \
+	fail("returned non zero but not -1"); \
+      } else { \
+	fail("returned non zero"); \
+      } \
+      ei_x_free(&arg); \
+      return; \
+    } \
+    if (arg.index < 1) { \
+      fail("size is < 1"); \
+      ei_x_free(&arg); \
+      return; \
+    } \
+    send_buffer(arg.buff, arg.index); \
+    ei_x_free(&arg); \
+  }
+
+#define EI_ENCODE_3_FAIL(FUNC,ARG1,ARG2,ARG3)	\
+  { \
+    char buf[1024]; \
+    int size1 = 0; \
+    int size2 = 0; \
+    int err; \
+    message("ei_" #FUNC " " #ARG1 " " #ARG2 " " #ARG3 " encoded as "); \
+    err = ei_ ## FUNC(NULL, &size1, ARG1, ARG2, ARG3);	     \
+    if (err != -1) { \
+      fail("size calculation returned non -1");	\
+      return;					\
+    } \
+    err = ei_ ## FUNC(buf, &size2, ARG1, ARG2, ARG3); \
+    if (err != -1) {					    \
+      fail("returned non -1");				    \
+    }							    \
+    if (size1 != size2) { \
+      fail("size differs when arg is NULL or buf"); \
+      return; \
+    } \
+  } \
+  { \
+    ei_x_buff arg; \
+    int err; \
+    message("ei_x_" #FUNC " " #ARG1 " " #ARG2 " " #ARG3 " encoded as "); \
+    ei_x_new(&arg); \
+    err = ei_x_ ## FUNC(&arg, ARG1, ARG2, ARG3); \
+    if (err != -1) {				       \
+      fail("returned non -1");			       \
+      ei_x_free(&arg);				       \
+      return;					       \
+    }						       \
+    ei_x_free(&arg); \
+  }
+
+#define EI_ENCODE_4(FUNC,ARG1,ARG2,ARG3,ARG4)	\
+  { \
+    char buf[1024]; \
+    int size1 = 0; \
+    int size2 = 0; \
+    int err; \
+    message("ei_" #FUNC " " #ARG1 " " #ARG2 " " #ARG3 " " #ARG4 " encoded as "); \
+    err = ei_ ## FUNC(NULL, &size1, ARG1, ARG2, ARG3, ARG4);	     \
+    if (err != 0) { \
+      if (err != -1) { \
+	fail("size calculation returned non zero but not -1"); \
+	return; \
+      } else { \
+	fail("size calculation returned non zero"); \
+	return; \
+      } \
+    } \
+    err = ei_ ## FUNC(buf, &size2, ARG1, ARG2, ARG3, ARG4); \
+    if (err != 0) { \
+      if (err != -1) { \
+	fail("returned non zero but not -1"); \
+      } else { \
+	fail("returned non zero"); \
+      } \
+      return; \
+    } \
+    if (size1 != size2) { \
+      fail("size differs when arg is NULL or buf"); \
+      return; \
+    } \
+    if (size1 < 1) { \
+      fail("size is < 1"); \
+      return; \
+    } \
+    send_buffer(buf, size1); \
+  } \
+  { \
+    ei_x_buff arg; \
+    int err; \
+    message("ei_x_" #FUNC " " #ARG1 " " #ARG2 " " #ARG3 " " #ARG4 " encoded as "); \
+    ei_x_new(&arg); \
+    err = ei_x_ ## FUNC(&arg, ARG1, ARG2, ARG3, ARG4); \
+    if (err != 0) { \
+      if (err != -1) { \
+	fail("returned non zero but not -1"); \
+      } else { \
+	fail("returned non zero"); \
+      } \
+      ei_x_free(&arg); \
+      return; \
+    } \
+    if (arg.index < 1) { \
+      fail("size is < 1"); \
+      ei_x_free(&arg); \
+      return; \
+    } \
+    send_buffer(arg.buff, arg.index); \
+    ei_x_free(&arg); \
+  }
+
+#define EI_ENCODE_4_FAIL(FUNC,ARG1,ARG2,ARG3,ARG4)	\
+  { \
+    char buf[1024]; \
+    int size1 = 0; \
+    int size2 = 0; \
+    int err; \
+    message("ei_" #FUNC " " #ARG1 " " #ARG2 " " #ARG3 " " #ARG4 " encoded as "); \
+    err = ei_ ## FUNC(NULL, &size1, ARG1, ARG2, ARG3, ARG4);	     \
+    if (err != -1) { \
+      fail("size calculation returned non -1");	\
+      return;					\
+    } \
+    err = ei_ ## FUNC(buf, &size2, ARG1, ARG2, ARG3, ARG4); \
+    if (err != -1) {					    \
+      fail("returned non -1");				    \
+    }							    \
+    if (size1 != size2) { \
+      fail("size differs when arg is NULL or buf"); \
+      return; \
+    } \
+  } \
+  { \
+    ei_x_buff arg; \
+    int err; \
+    message("ei_x_" #FUNC " " #ARG1 " " #ARG2 " " #ARG3 " " #ARG4 " encoded as "); \
+    ei_x_new(&arg); \
+    err = ei_x_ ## FUNC(&arg, ARG1, ARG2, ARG3, ARG4); \
+    if (err != -1) {				       \
+      fail("returned non -1");			       \
+      ei_x_free(&arg);				       \
+      return;					       \
+    }						       \
+    ei_x_free(&arg); \
+  }
+
 /* ******************************************************************** */
 
 TESTCASE(test_ei_encode_long)
@@ -464,3 +654,57 @@ TESTCASE(test_ei_encode_fails)
 
     report(1);
 }
+
+/* ******************************************************************** */
+
+TESTCASE(test_ei_encode_utf8_atom)
+{
+
+  EI_ENCODE_3(encode_atom_as, "�", ERLANG_LATIN1, ERLANG_UTF8);
+  EI_ENCODE_3(encode_atom_as, "�", ERLANG_LATIN1, ERLANG_LATIN1);
+  EI_ENCODE_3(encode_atom_as, "\303\205",
+	      ERLANG_UTF8, ERLANG_LATIN1);
+  EI_ENCODE_3(encode_atom_as, "\303\205",
+	      ERLANG_UTF8, ERLANG_UTF8);
+
+  EI_ENCODE_3(encode_atom_as, "A", ERLANG_ASCII, ERLANG_UTF8);
+  EI_ENCODE_3(encode_atom_as, "A", ERLANG_ASCII, ERLANG_LATIN1);
+
+  EI_ENCODE_3_FAIL(encode_atom_as, "�", ERLANG_UTF8, ERLANG_UTF8);
+  EI_ENCODE_3_FAIL(encode_atom_as, "�", ERLANG_UTF8, ERLANG_LATIN1);
+  EI_ENCODE_3_FAIL(encode_atom_as, "�", ERLANG_ASCII, ERLANG_UTF8);
+  EI_ENCODE_3_FAIL(encode_atom_as, "�", ERLANG_ASCII, ERLANG_LATIN1);
+
+  EI_ENCODE_3_FAIL(encode_atom_as, "yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy", ERLANG_ASCII, ERLANG_LATIN1);
+  EI_ENCODE_3_FAIL(encode_atom_as, "yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy", ERLANG_LATIN1, ERLANG_LATIN1);
+
+  EI_ENCODE_3_FAIL(encode_atom_as, "yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy", ERLANG_UTF8, ERLANG_LATIN1);
+
+  report(1);
+}
+
+TESTCASE(test_ei_encode_utf8_atom_len)
+{
+
+  EI_ENCODE_4(encode_atom_len_as, "���", 1, ERLANG_LATIN1, ERLANG_UTF8);
+  EI_ENCODE_4(encode_atom_len_as, "���", 2, ERLANG_LATIN1, ERLANG_LATIN1);
+  EI_ENCODE_4(encode_atom_len_as, "\303\205\303\204\303\226", 2,
+	      ERLANG_UTF8, ERLANG_LATIN1);
+  EI_ENCODE_4(encode_atom_len_as, "\303\205\303\204\303\226", 4,
+	      ERLANG_UTF8, ERLANG_UTF8);
+
+  EI_ENCODE_4(encode_atom_len_as, "ABC", 1, ERLANG_ASCII, ERLANG_UTF8);
+  EI_ENCODE_4(encode_atom_len_as, "ABC", 2, ERLANG_ASCII, ERLANG_LATIN1);
+
+  EI_ENCODE_4_FAIL(encode_atom_len_as, "���", 1, ERLANG_UTF8, ERLANG_UTF8);
+  EI_ENCODE_4_FAIL(encode_atom_len_as, "���", 2, ERLANG_UTF8, ERLANG_LATIN1);
+  EI_ENCODE_4_FAIL(encode_atom_len_as, "���", 3, ERLANG_ASCII, ERLANG_UTF8);
+  EI_ENCODE_4_FAIL(encode_atom_len_as, "���", 4, ERLANG_ASCII, ERLANG_LATIN1);
+
+  EI_ENCODE_4(encode_atom_len_as, "yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy", 255, ERLANG_ASCII, ERLANG_LATIN1);
+
+  EI_ENCODE_4_FAIL(encode_atom_len_as, "\303\205\303\204\303\226", 1,
+		   ERLANG_UTF8, ERLANG_LATIN1);
+
+  report(1);
+}