sl@0: /* GLIB - Library of useful routines for C programming
sl@0:  * Copyright (C) 2001 Matthias Clasen <matthiasc@poet.de>
sl@0:  * Portion Copyright © 2008-09 Nokia Corporation and/or its subsidiary(-ies). All rights reserved.
sl@0:  * This library is free software; you can redistribute it and/or
sl@0:  * modify it under the terms of the GNU Lesser General Public
sl@0:  * License as published by the Free Software Foundation; either
sl@0:  * version 2 of the License, or (at your option) any later version.
sl@0:  *
sl@0:  * This library is distributed in the hope that it will be useful,
sl@0:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
sl@0:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
sl@0:  * Lesser General Public License for more details.
sl@0:  *
sl@0:  * You should have received a copy of the GNU Lesser General Public
sl@0:  * License along with this library; if not, write to the
sl@0:  * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
sl@0:  * Boston, MA 02111-1307, USA.
sl@0:  */
sl@0: 
sl@0: #include "glib.h"
sl@0: #ifdef __SYMBIAN32__
sl@0: #include "mrt2_glib2_test.h"
sl@0: #endif /*__SYMBIAN32__*/
sl@0: 
sl@0: #define UNICODE_VALID(Char)                   \
sl@0:     ((Char) < 0x110000 &&                     \
sl@0:      (((Char) & 0xFFFFF800) != 0xD800) &&     \
sl@0:      ((Char) < 0xFDD0 || (Char) > 0xFDEF) &&  \
sl@0:      ((Char) & 0xFFFE) != 0xFFFE)
sl@0: 
sl@0: 
sl@0: 
sl@0: static gboolean any_failed = FALSE;
sl@0: 
sl@0: struct {
sl@0:   const gchar *text;
sl@0:   gint max_len;
sl@0:   gint offset;
sl@0:   gboolean valid;
sl@0: } test[] = {  
sl@0:   /* some tests to check max_len handling */
sl@0:   /* length 1 */
sl@0:   { "abcde", -1, 5, TRUE },
sl@0:   { "abcde", 3, 3, TRUE },
sl@0:   { "abcde", 5, 5, TRUE },
sl@0:   { "abcde", 7, 5, FALSE },
sl@0:   /* length 2 */
sl@0:   { "\xc2\xa9\xc2\xa9\xc2\xa9", -1, 6, TRUE }, 
sl@0:   { "\xc2\xa9\xc2\xa9\xc2\xa9",  1, 0, FALSE }, 
sl@0:   { "\xc2\xa9\xc2\xa9\xc2\xa9",  2, 2, TRUE }, 
sl@0:   { "\xc2\xa9\xc2\xa9\xc2\xa9",  3, 2, FALSE }, 
sl@0:   { "\xc2\xa9\xc2\xa9\xc2\xa9",  4, 4, TRUE }, 
sl@0:   { "\xc2\xa9\xc2\xa9\xc2\xa9",  5, 4, FALSE }, 
sl@0:   { "\xc2\xa9\xc2\xa9\xc2\xa9",  6, 6, TRUE }, 
sl@0:   { "\xc2\xa9\xc2\xa9\xc2\xa9",  7, 6, FALSE }, 
sl@0:   /* length 3 */
sl@0:   { "\xe2\x89\xa0\xe2\x89\xa0", -1, 6, TRUE },
sl@0:   { "\xe2\x89\xa0\xe2\x89\xa0",  1, 0, FALSE },
sl@0:   { "\xe2\x89\xa0\xe2\x89\xa0",  2, 0, FALSE },
sl@0:   { "\xe2\x89\xa0\xe2\x89\xa0",  3, 3, TRUE },
sl@0:   { "\xe2\x89\xa0\xe2\x89\xa0",  4, 3, FALSE },
sl@0:   { "\xe2\x89\xa0\xe2\x89\xa0",  5, 3, FALSE },
sl@0:   { "\xe2\x89\xa0\xe2\x89\xa0",  6, 6, TRUE },
sl@0:   { "\xe2\x89\xa0\xe2\x89\xa0",  7, 6, FALSE },
sl@0: 
sl@0:   /* examples from http://www.cl.cam.ac.uk/~mgk25/ucs/examples/UTF-8-test.txt */
sl@0:   /* greek 'kosme' */
sl@0:   { "\xce\xba\xe1\xbd\xb9\xcf\x83\xce\xbc\xce\xb5", -1, 11, TRUE },
sl@0:   /* first sequence of each length */
sl@0:   { "\x00", -1, 0, TRUE },
sl@0:   { "\xc2\x80", -1, 2, TRUE },
sl@0:   { "\xe0\xa0\x80", -1, 3, TRUE },
sl@0:   { "\xf0\x90\x80\x80", -1, 4, TRUE },
sl@0:   { "\xf8\x88\x80\x80\x80", -1, 0, FALSE },
sl@0:   { "\xfc\x84\x80\x80\x80\x80", -1, 0, FALSE },
sl@0:   /* last sequence of each length */
sl@0:   { "\x7f", -1, 1, TRUE },
sl@0:   { "\xdf\xbf", -1, 2, TRUE },
sl@0:   { "\xef\xbf\xbf", -1, 0, FALSE },
sl@0:   { "\xf7\xbf\xbf\xbf", -1, 0, FALSE },
sl@0:   { "\xfb\xbf\xbf\xbf\xbf", -1, 0, FALSE },
sl@0:   { "\xfd\xbf\xbf\xbf\xbf\xbf", -1, 0, FALSE },
sl@0:   /* other boundary conditions */
sl@0:   { "\xed\x9f\xbf", -1, 3, TRUE },
sl@0:   { "\xee\x80\x80", -1, 3, TRUE },
sl@0:   { "\xef\xbf\xbd", -1, 3, TRUE },
sl@0:   { "\xf4\x8f\xbf\xbf", -1, 0, FALSE },
sl@0:   { "\xf4\x90\x80\x80", -1, 0, FALSE },
sl@0:   /* malformed sequences */
sl@0:   /* continuation bytes */
sl@0:   { "\x80", -1, 0, FALSE },
sl@0:   { "\xbf", -1, 0, FALSE },
sl@0:   { "\x80\xbf", -1, 0, FALSE },
sl@0:   { "\x80\xbf\x80", -1, 0, FALSE },
sl@0:   { "\x80\xbf\x80\xbf", -1, 0, FALSE },
sl@0:   { "\x80\xbf\x80\xbf\x80", -1, 0, FALSE },
sl@0:   { "\x80\xbf\x80\xbf\x80\xbf", -1, 0, FALSE },
sl@0:   { "\x80\xbf\x80\xbf\x80\xbf\x80", -1, 0, FALSE },
sl@0: 
sl@0:   /* all possible continuation byte */
sl@0:   { "\x80", -1, 0, FALSE },
sl@0:   { "\x81", -1, 0, FALSE },
sl@0:   { "\x82", -1, 0, FALSE },
sl@0:   { "\x83", -1, 0, FALSE },
sl@0:   { "\x84", -1, 0, FALSE },
sl@0:   { "\x85", -1, 0, FALSE },
sl@0:   { "\x86", -1, 0, FALSE },
sl@0:   { "\x87", -1, 0, FALSE },
sl@0:   { "\x88", -1, 0, FALSE },
sl@0:   { "\x89", -1, 0, FALSE },
sl@0:   { "\x8a", -1, 0, FALSE },
sl@0:   { "\x8b", -1, 0, FALSE },
sl@0:   { "\x8c", -1, 0, FALSE },
sl@0:   { "\x8d", -1, 0, FALSE },
sl@0:   { "\x8e", -1, 0, FALSE },
sl@0:   { "\x8f", -1, 0, FALSE },
sl@0:   { "\x90", -1, 0, FALSE },
sl@0:   { "\x91", -1, 0, FALSE },
sl@0:   { "\x92", -1, 0, FALSE },
sl@0:   { "\x93", -1, 0, FALSE },
sl@0:   { "\x94", -1, 0, FALSE },
sl@0:   { "\x95", -1, 0, FALSE },
sl@0:   { "\x96", -1, 0, FALSE },
sl@0:   { "\x97", -1, 0, FALSE },
sl@0:   { "\x98", -1, 0, FALSE },
sl@0:   { "\x99", -1, 0, FALSE },
sl@0:   { "\x9a", -1, 0, FALSE },
sl@0:   { "\x9b", -1, 0, FALSE },
sl@0:   { "\x9c", -1, 0, FALSE },
sl@0:   { "\x9d", -1, 0, FALSE },
sl@0:   { "\x9e", -1, 0, FALSE },
sl@0:   { "\x9f", -1, 0, FALSE },
sl@0:   { "\xa0", -1, 0, FALSE },
sl@0:   { "\xa1", -1, 0, FALSE },
sl@0:   { "\xa2", -1, 0, FALSE },
sl@0:   { "\xa3", -1, 0, FALSE },
sl@0:   { "\xa4", -1, 0, FALSE },
sl@0:   { "\xa5", -1, 0, FALSE },
sl@0:   { "\xa6", -1, 0, FALSE },
sl@0:   { "\xa7", -1, 0, FALSE },
sl@0:   { "\xa8", -1, 0, FALSE },
sl@0:   { "\xa9", -1, 0, FALSE },
sl@0:   { "\xaa", -1, 0, FALSE },
sl@0:   { "\xab", -1, 0, FALSE },
sl@0:   { "\xac", -1, 0, FALSE },
sl@0:   { "\xad", -1, 0, FALSE },
sl@0:   { "\xae", -1, 0, FALSE },
sl@0:   { "\xaf", -1, 0, FALSE },
sl@0:   { "\xb0", -1, 0, FALSE },
sl@0:   { "\xb1", -1, 0, FALSE },
sl@0:   { "\xb2", -1, 0, FALSE },
sl@0:   { "\xb3", -1, 0, FALSE },
sl@0:   { "\xb4", -1, 0, FALSE },
sl@0:   { "\xb5", -1, 0, FALSE },
sl@0:   { "\xb6", -1, 0, FALSE },
sl@0:   { "\xb7", -1, 0, FALSE },
sl@0:   { "\xb8", -1, 0, FALSE },
sl@0:   { "\xb9", -1, 0, FALSE },
sl@0:   { "\xba", -1, 0, FALSE },
sl@0:   { "\xbb", -1, 0, FALSE },
sl@0:   { "\xbc", -1, 0, FALSE },
sl@0:   { "\xbd", -1, 0, FALSE },
sl@0:   { "\xbe", -1, 0, FALSE },
sl@0:   { "\xbf", -1, 0, FALSE },
sl@0:   /* lone start characters */
sl@0:   { "\xc0\x20", -1, 0, FALSE },
sl@0:   { "\xc1\x20", -1, 0, FALSE },
sl@0:   { "\xc2\x20", -1, 0, FALSE },
sl@0:   { "\xc3\x20", -1, 0, FALSE },
sl@0:   { "\xc4\x20", -1, 0, FALSE },
sl@0:   { "\xc5\x20", -1, 0, FALSE },
sl@0:   { "\xc6\x20", -1, 0, FALSE },
sl@0:   { "\xc7\x20", -1, 0, FALSE },
sl@0:   { "\xc8\x20", -1, 0, FALSE },
sl@0:   { "\xc9\x20", -1, 0, FALSE },
sl@0:   { "\xca\x20", -1, 0, FALSE },
sl@0:   { "\xcb\x20", -1, 0, FALSE },
sl@0:   { "\xcc\x20", -1, 0, FALSE },
sl@0:   { "\xcd\x20", -1, 0, FALSE },
sl@0:   { "\xce\x20", -1, 0, FALSE },
sl@0:   { "\xcf\x20", -1, 0, FALSE },
sl@0:   { "\xd0\x20", -1, 0, FALSE },
sl@0:   { "\xd1\x20", -1, 0, FALSE },
sl@0:   { "\xd2\x20", -1, 0, FALSE },
sl@0:   { "\xd3\x20", -1, 0, FALSE },
sl@0:   { "\xd4\x20", -1, 0, FALSE },
sl@0:   { "\xd5\x20", -1, 0, FALSE },
sl@0:   { "\xd6\x20", -1, 0, FALSE },
sl@0:   { "\xd7\x20", -1, 0, FALSE },
sl@0:   { "\xd8\x20", -1, 0, FALSE },
sl@0:   { "\xd9\x20", -1, 0, FALSE },
sl@0:   { "\xda\x20", -1, 0, FALSE },
sl@0:   { "\xdb\x20", -1, 0, FALSE },
sl@0:   { "\xdc\x20", -1, 0, FALSE },
sl@0:   { "\xdd\x20", -1, 0, FALSE },
sl@0:   { "\xde\x20", -1, 0, FALSE },
sl@0:   { "\xdf\x20", -1, 0, FALSE },
sl@0:   { "\xe0\x20", -1, 0, FALSE },
sl@0:   { "\xe1\x20", -1, 0, FALSE },
sl@0:   { "\xe2\x20", -1, 0, FALSE },
sl@0:   { "\xe3\x20", -1, 0, FALSE },
sl@0:   { "\xe4\x20", -1, 0, FALSE },
sl@0:   { "\xe5\x20", -1, 0, FALSE },
sl@0:   { "\xe6\x20", -1, 0, FALSE },
sl@0:   { "\xe7\x20", -1, 0, FALSE },
sl@0:   { "\xe8\x20", -1, 0, FALSE },
sl@0:   { "\xe9\x20", -1, 0, FALSE },
sl@0:   { "\xea\x20", -1, 0, FALSE },
sl@0:   { "\xeb\x20", -1, 0, FALSE },
sl@0:   { "\xec\x20", -1, 0, FALSE },
sl@0:   { "\xed\x20", -1, 0, FALSE },
sl@0:   { "\xee\x20", -1, 0, FALSE },
sl@0:   { "\xef\x20", -1, 0, FALSE },
sl@0:   { "\xf0\x20", -1, 0, FALSE },
sl@0:   { "\xf1\x20", -1, 0, FALSE },
sl@0:   { "\xf2\x20", -1, 0, FALSE },
sl@0:   { "\xf3\x20", -1, 0, FALSE },
sl@0:   { "\xf4\x20", -1, 0, FALSE },
sl@0:   { "\xf5\x20", -1, 0, FALSE },
sl@0:   { "\xf6\x20", -1, 0, FALSE },
sl@0:   { "\xf7\x20", -1, 0, FALSE },
sl@0:   { "\xf8\x20", -1, 0, FALSE },
sl@0:   { "\xf9\x20", -1, 0, FALSE },
sl@0:   { "\xfa\x20", -1, 0, FALSE },
sl@0:   { "\xfb\x20", -1, 0, FALSE },
sl@0:   { "\xfc\x20", -1, 0, FALSE },
sl@0:   { "\xfd\x20", -1, 0, FALSE },
sl@0:   /* missing continuation bytes */
sl@0:   { "\x20\xc0", -1, 1, FALSE },
sl@0:   { "\x20\xe0\x80", -1, 1, FALSE },
sl@0:   { "\x20\xf0\x80\x80", -1, 1, FALSE },
sl@0:   { "\x20\xf8\x80\x80\x80", -1, 1, FALSE },
sl@0:   { "\x20\xfc\x80\x80\x80\x80", -1, 1, FALSE },
sl@0:   { "\x20\xdf", -1, 1, FALSE },
sl@0:   { "\x20\xef\xbf", -1, 1, FALSE },
sl@0:   { "\x20\xf7\xbf\xbf", -1, 1, FALSE },
sl@0:   { "\x20\xfb\xbf\xbf\xbf", -1, 1, FALSE },
sl@0:   { "\x20\xfd\xbf\xbf\xbf\xbf", -1, 1, FALSE },
sl@0:   /* impossible bytes */
sl@0:   { "\x20\xfe\x20", -1, 1, FALSE },
sl@0:   { "\x20\xff\x20", -1, 1, FALSE },
sl@0:   /* overlong sequences */
sl@0:   { "\x20\xc0\xaf\x20", -1, 1, FALSE },
sl@0:   { "\x20\xe0\x80\xaf\x20", -1, 1, FALSE },
sl@0:   { "\x20\xf0\x80\x80\xaf\x20", -1, 1, FALSE },
sl@0:   { "\x20\xf8\x80\x80\x80\xaf\x20", -1, 1, FALSE },
sl@0:   { "\x20\xfc\x80\x80\x80\x80\xaf\x20", -1, 1, FALSE },
sl@0:   { "\x20\xc1\xbf\x20", -1, 1, FALSE },
sl@0:   { "\x20\xe0\x9f\xbf\x20", -1, 1, FALSE },
sl@0:   { "\x20\xf0\x8f\xbf\xbf\x20", -1, 1, FALSE },
sl@0:   { "\x20\xf8\x87\xbf\xbf\xbf\x20", -1, 1, FALSE },
sl@0:   { "\x20\xfc\x83\xbf\xbf\xbf\xbf\x20", -1, 1, FALSE },
sl@0:   { "\x20\xc0\x80\x20", -1, 1, FALSE },
sl@0:   { "\x20\xe0\x80\x80\x20", -1, 1, FALSE },
sl@0:   { "\x20\xf0\x80\x80\x80\x20", -1, 1, FALSE },
sl@0:   { "\x20\xf8\x80\x80\x80\x80\x20", -1, 1, FALSE },
sl@0:   { "\x20\xfc\x80\x80\x80\x80\x80\x20", -1, 1, FALSE },
sl@0:   /* illegal code positions */
sl@0:   { "\x20\xed\xa0\x80\x20", -1, 1, FALSE },
sl@0:   { "\x20\xed\xad\xbf\x20", -1, 1, FALSE },
sl@0:   { "\x20\xed\xae\x80\x20", -1, 1, FALSE },
sl@0:   { "\x20\xed\xaf\xbf\x20", -1, 1, FALSE },
sl@0:   { "\x20\xed\xb0\x80\x20", -1, 1, FALSE },
sl@0:   { "\x20\xed\xbe\x80\x20", -1, 1, FALSE },
sl@0:   { "\x20\xed\xbf\xbf\x20", -1, 1, FALSE },
sl@0:   { "\x20\xed\xa0\x80\xed\xb0\x80\x20", -1, 1, FALSE },
sl@0:   { "\x20\xed\xa0\x80\xed\xbf\xbf\x20", -1, 1, FALSE },
sl@0:   { "\x20\xed\xad\xbf\xed\xb0\x80\x20", -1, 1, FALSE },
sl@0:   { "\x20\xed\xad\xbf\xed\xbf\xbf\x20", -1, 1, FALSE },
sl@0:   { "\x20\xed\xae\x80\xed\xb0\x80\x20", -1, 1, FALSE },
sl@0:   { "\x20\xed\xae\x80\xed\xbf\xbf\x20", -1, 1, FALSE },
sl@0:   { "\x20\xed\xaf\xbf\xed\xb0\x80\x20", -1, 1, FALSE },
sl@0:   { "\x20\xed\xaf\xbf\xed\xbf\xbf\x20", -1, 1, FALSE },
sl@0:   { "\x20\xef\xbf\xbe\x20", -1, 1, FALSE },
sl@0:   { "\x20\xef\xbf\xbf\x20", -1, 1, FALSE },
sl@0: 
sl@0:   { NULL, }
sl@0: };
sl@0: 
sl@0: static void 
sl@0: do_test (gint         index,
sl@0: 	 const gchar *text, 
sl@0: 	 gint         max_len,
sl@0: 	 gint         offset,
sl@0: 	 gboolean     valid)
sl@0: {
sl@0:   const gchar *end;
sl@0:   gboolean result;
sl@0:   
sl@0:   result = g_utf8_validate (text, max_len, &end);
sl@0: 
sl@0:   if (result != valid || end - text != offset)
sl@0:     {
sl@0:       GString *str;
sl@0:       const gchar *p;
sl@0: 
sl@0:       any_failed = TRUE;
sl@0:       
sl@0:       str = g_string_new (0);
sl@0:       for (p = text; *p; p++)
sl@0: 	g_string_append_printf (str, "\\x%02hhx", *p);
sl@0:       g_print ("%d: g_utf8_validate (\"%s\", %d) failed, "
sl@0: 	       "expected %s %d, got %s %d\n",
sl@0: 	       index,
sl@0: 	       str->str, max_len, 
sl@0: 	       valid ? "TRUE" : "FALSE", offset,
sl@0: 	       result ? "TRUE" : "FALSE", (gint) (end - text));
sl@0:       g_string_free (str, FALSE);
sl@0:     }
sl@0: }
sl@0: 
sl@0: int
sl@0: main (int argc, char *argv[])
sl@0: {
sl@0:   gint i;
sl@0: 
sl@0:   #ifdef __SYMBIAN32__
sl@0:   g_log_set_handler (NULL,  G_LOG_FLAG_FATAL| G_LOG_FLAG_RECURSION | G_LOG_LEVEL_CRITICAL | G_LOG_LEVEL_WARNING | G_LOG_LEVEL_MESSAGE | G_LOG_LEVEL_INFO | G_LOG_LEVEL_DEBUG, &mrtLogHandler, NULL);
sl@0:   g_set_print_handler(mrtPrintHandler);
sl@0:   #endif /*__SYMBIAN32__*/
sl@0: 	  
sl@0: 
sl@0:   for (i = 0; test[i].text; i++)
sl@0:     do_test (i, test[i].text, test[i].max_len, 
sl@0: 	     test[i].offset, test[i].valid);
sl@0: 
sl@0:   #ifdef __SYMBIAN32__
sl@0:   assert_failed = any_failed;
sl@0:   testResultXml("utf8-validate");
sl@0:   #endif /* EMULATOR */
sl@0:   
sl@0:   return any_failed ? 1 : 0;
sl@0: }