erts: Change erlang:open_port spawn to handle unicode

Previously only 'spawn_executable' handled unicode input.
Also change 'cd' option to always handle unicode.
Update open_port documentation and tests

3 files changed, 34 insertions, 53 deletions

diff --git a/erts/emulator/beam/erl_bif_port.c b/erts/emulator/beam/erl_bif_port.c
index 109c54fd7f..5b26e53c9f 100644
--- a/erts/emulator/beam/erl_bif_port.c
+++ b/erts/emulator/beam/erl_bif_port.c
@@ -784,43 +784,29 @@ open_port(Process* p, Eterm name, Eterm settings, int *err_typep, int *err_nump)
 	    goto badarg;
 	}
     
-	if (*tp == am_spawn || *tp == am_spawn_driver) {	/* A process port */
+	if (*tp == am_spawn || *tp == am_spawn_driver || *tp == am_spawn_executable) {	/* A process port */
+	    int encoding;
 	    if (arity != make_arityval(2)) {
 		goto badarg;
 	    }
 	    name = tp[1];
-	    if (is_atom(name)) {
-		name_buf = (char *) erts_alloc(ERTS_ALC_T_TMP,
-					       atom_tab(atom_val(name))->len+1);
-		sys_memcpy((void *) name_buf,
-			   (void *) atom_tab(atom_val(name))->name, 
-			   atom_tab(atom_val(name))->len);
-		name_buf[atom_tab(atom_val(name))->len] = '\0';
-	    } else if ((i = is_string(name))) {
-		name_buf = (char *) erts_alloc(ERTS_ALC_T_TMP, i + 1);
-		if (intlist_to_buf(name, name_buf, i) != i)
-		    erl_exit(1, "%s:%d: Internal error\n", __FILE__, __LINE__);
-		name_buf[i] = '\0';
-	    } else {
+	    encoding = erts_get_native_filename_encoding();
+	    /* Do not convert the command to utf-16le yet, do that in win32 specific code */
+	    /* since the cmd is used for comparsion with drivers names and copied to port info */
+	    if (encoding == ERL_FILENAME_WIN_WCHAR) {
+		encoding = ERL_FILENAME_UTF8;
+	    }
+	    if ((name_buf = erts_convert_filename_to_encoding(name, NULL, 0, ERTS_ALC_T_TMP,0,1, encoding, NULL))
+		== NULL) {
 		goto badarg;
 	    }
+
 	    if (*tp == am_spawn_driver) {
 		opts.spawn_type = ERTS_SPAWN_DRIVER;
+	    } else if (*tp == am_spawn_executable) {
+		opts.spawn_type = ERTS_SPAWN_EXECUTABLE;
 	    }
-	    driver = &spawn_driver;
-	} else if (*tp == am_spawn_executable) {	/* A program */
-	    /*
-	     * {spawn_executable,Progname}
-	     */
-	    
-	    if (arity != make_arityval(2)) {
-		goto badarg;
-	    }
-	    name = tp[1];
-	    if ((name_buf = erts_convert_filename_to_native(name, NULL, 0, ERTS_ALC_T_TMP,0,1, NULL)) == NULL) {
-		goto badarg;
-	    }
-	    opts.spawn_type = ERTS_SPAWN_EXECUTABLE;
+
 	    driver = &spawn_driver;
 	} else if (*tp == am_fd) { /* An fd port */
 	    int n;
@@ -861,29 +847,8 @@ open_port(Process* p, Eterm name, Eterm settings, int *err_typep, int *err_nump)
     }
 
     if (edir != NIL) {
-	/* A working directory is expressed differently if spawn_executable, i.e. Unicode is handles 
-	   for spawn_executable... */
-	if (opts.spawn_type != ERTS_SPAWN_EXECUTABLE) {
-	    Eterm iolist;
-	    DeclareTmpHeap(heap,4,p);
-	    int r;
-	    
-	    UseTmpHeap(4,p);
-	    heap[0] = edir;
-	    heap[1] = make_list(heap+2);
-	    heap[2] = make_small(0);
-	    heap[3] = NIL;
-	    iolist = make_list(heap);
-	    r = erts_iolist_to_buf(iolist, (char*) dir, MAXPATHLEN);
-	    UnUseTmpHeap(4,p);
-	    if (ERTS_IOLIST_TO_BUF_FAILED(r)) {
-		goto badarg;
-	    }
-	    opts.wd = (char *) dir;
-	} else {
-	    if ((opts.wd = erts_convert_filename_to_native(edir, NULL, 0, ERTS_ALC_T_TMP,0,1,NULL)) == NULL) {
-		goto badarg;
-	    }
+	if ((opts.wd = erts_convert_filename_to_native(edir, NULL, 0, ERTS_ALC_T_TMP,0,1,NULL)) == NULL) {
+	    goto badarg;
 	}
     }
 
@@ -961,11 +926,11 @@ static char **convert_args(Eterm l)
     int n;
     int i = 0;
     Eterm str;
-    /* We require at least one element in list (argv[0]) */
     if (is_not_list(l) && is_not_nil(l)) {
 	return NULL;
     }
     n = list_length(l);
+    /* We require at least one element in argv[0] + NULL at end */
     pp = erts_alloc(ERTS_ALC_T_TMP, (n + 2) * sizeof(char **));
     pp[i++] = erts_default_arg0;
     while (is_list(l)) {
diff --git a/erts/emulator/beam/erl_unicode.c b/erts/emulator/beam/erl_unicode.c
index ad6f8b993a..71794e1101 100644
--- a/erts/emulator/beam/erl_unicode.c
+++ b/erts/emulator/beam/erl_unicode.c
@@ -1981,9 +1981,19 @@ BIF_RETTYPE binary_to_existing_atom_2(BIF_ALIST_2)
  * string routines, that will certainly fail on some OS.
  */
 
-char *erts_convert_filename_to_native(Eterm name, char *statbuf, size_t statbuf_size, ErtsAlcType_t alloc_type, int allow_empty, int allow_atom, Sint *used)
+char *erts_convert_filename_to_native(Eterm name, char *statbuf, size_t statbuf_size,
+				      ErtsAlcType_t alloc_type, int allow_empty,
+				      int allow_atom, Sint *used)
 {
     int encoding = erts_get_native_filename_encoding();
+    return erts_convert_filename_to_encoding(name, statbuf, statbuf_size, alloc_type,
+					     allow_empty, allow_atom, encoding, used);
+}
+
+char *erts_convert_filename_to_encoding(Eterm name, char *statbuf, size_t statbuf_size,
+					ErtsAlcType_t alloc_type, int allow_empty,
+					int allow_atom, int encoding, Sint *used)
+{
     char* name_buf = NULL;
 
     if ((allow_atom && is_atom(name)) || 
diff --git a/erts/emulator/beam/global.h b/erts/emulator/beam/global.h
index 26ed5f82c1..f7e4c81181 100755
--- a/erts/emulator/beam/global.h
+++ b/erts/emulator/beam/global.h
@@ -788,6 +788,12 @@ char *erts_convert_filename_to_native(Eterm name, char *statbuf,
 				      ErtsAlcType_t alloc_type, 
 				      int allow_empty, int allow_atom,
 				      Sint *used /* out */);
+char *erts_convert_filename_to_encoding(Eterm name, char *statbuf,
+					size_t statbuf_size,
+					ErtsAlcType_t alloc_type,
+					int allow_empty, int allow_atom,
+					int encoding,
+					Sint *used /* out */);
 Eterm erts_convert_native_to_filename(Process *p, byte *bytes);
 Eterm erts_utf8_to_list(Process *p, Uint num, byte *bytes, Uint sz, Uint left,
 			Uint *num_built, Uint *num_eaten, Eterm tail);