annotate test/testiconv.c @ 1502:d403a39389da

UTF-32 by definition can't handle characters higher than 0x10FFFF
author Sam Lantinga <slouken@libsdl.org>
date Mon, 13 Mar 2006 01:16:16 +0000
parents 73dc5d39bbf8
children 4d241ea8a1cd
rev   line source
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
1
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
2 #include <stdio.h>
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
3
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
4 #include "SDL.h"
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
5
1502
d403a39389da UTF-32 by definition can't handle characters higher than 0x10FFFF
Sam Lantinga <slouken@libsdl.org>
parents: 1501
diff changeset
6 static SDL_bool testutf1632(char *data)
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
7 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
8 Uint32 *p = (Uint32 *)data;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
9 while(*p) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
10 if ( *p > 0x10FFFF ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
11 return SDL_FALSE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
12 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
13 ++p;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
14 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
15 return SDL_TRUE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
16 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
17
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
18 static size_t widelen(char *data)
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
19 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
20 size_t len = 0;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
21 Uint32 *p = (Uint32 *)data;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
22 while(*p++) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
23 ++len;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
24 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
25 return len;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
26 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
27
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
28 int main(int argc, char *argv[])
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
29 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
30 const char * formats[] = {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
31 "UTF8",
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
32 "UTF-8",
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
33 "UTF16BE",
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
34 "UTF-16BE",
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
35 "UTF16LE",
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
36 "UTF-16LE",
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
37 "UTF32BE",
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
38 "UTF-32BE",
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
39 "UTF32LE",
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
40 "UTF-32LE",
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
41 "UCS4",
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
42 "UCS-4",
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
43 };
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
44 char buffer[BUFSIZ];
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
45 char *ucs4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
46 char *test[2];
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
47 int i, j, index = 0;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
48 FILE *file;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
49 int errors = 0;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
50
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
51 if ( !argv[1] ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
52 argv[1] = "utf8.txt";
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
53 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
54 file = fopen(argv[1], "rb");
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
55 if ( !file ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
56 fprintf(stderr, "Unable to open %s\n", argv[1]);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
57 return (1);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
58 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
59
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
60 while ( fgets(buffer, sizeof(buffer), file) ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
61 /* Convert to UCS-4 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
62 ucs4 = SDL_iconv_string("UCS-4", "UTF-8", buffer, SDL_strlen(buffer)+1);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
63 size_t len = (widelen(ucs4)+1)*4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
64 for ( i = 0; i < SDL_arraysize(formats); ++i ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
65 if ( (SDL_strncasecmp(formats[i], "UTF16", 5) == 0 ||
1502
d403a39389da UTF-32 by definition can't handle characters higher than 0x10FFFF
Sam Lantinga <slouken@libsdl.org>
parents: 1501
diff changeset
66 SDL_strncasecmp(formats[i], "UTF-16", 6) == 0 ||
d403a39389da UTF-32 by definition can't handle characters higher than 0x10FFFF
Sam Lantinga <slouken@libsdl.org>
parents: 1501
diff changeset
67 SDL_strncasecmp(formats[i], "UTF32", 5) == 0 ||
d403a39389da UTF-32 by definition can't handle characters higher than 0x10FFFF
Sam Lantinga <slouken@libsdl.org>
parents: 1501
diff changeset
68 SDL_strncasecmp(formats[i], "UTF-32", 6) == 0) &&
d403a39389da UTF-32 by definition can't handle characters higher than 0x10FFFF
Sam Lantinga <slouken@libsdl.org>
parents: 1501
diff changeset
69 !testutf1632(ucs4) ) {
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
70 continue;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
71 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
72 test[0] = SDL_iconv_string(formats[i], "UCS-4", ucs4, len);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
73 test[1] = SDL_iconv_string("UCS-4", formats[i], test[0], len);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
74 if ( SDL_memcmp(test[1], ucs4, len) != 0 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
75 fprintf(stderr, "FAIL: %s\n", formats[i]);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
76 ++errors;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
77 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
78 SDL_free(test[0]);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
79 SDL_free(test[1]);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
80 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
81 test[0] = SDL_iconv_string("UTF-8", "UCS-4", ucs4, len);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
82 SDL_free(ucs4);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
83 fputs(test[0], stdout);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
84 SDL_free(test[0]);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
85 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
86 return (errors ? errors + 1 : 0);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
87 }