Mercurial > dovecot > original-hg > dovecot-1.2

annotate src/lib-charset/charset-iconv.c @ 3863:55df57c028d4 HEAD

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Added "bool" type and changed all ints that were used as booleans to bool.
author Timo Sirainen <tss@iki.fi>
date Fri, 13 Jan 2006 22:25:57 +0200
parents 689f791b480f
children 928229f8b3e6
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
568
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1 /* Copyright (C) 2002 Timo Sirainen */
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
2
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
3 #include "lib.h"
765
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
4 #include "buffer.h"
579
e524da896d92 Several minor fixes: signess, casting away const, missing static, etc.
Timo Sirainen <tss@iki.fi>
parents: 568
diff changeset
5 #include "charset-utf8.h"
568
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
6
1300
952bf533c2ea Better iconv() checking.
Timo Sirainen <tss@iki.fi>
parents: 903
diff changeset
7 #ifdef HAVE_ICONV
568
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
8
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
9 #include <iconv.h>
608
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
10 #include <ctype.h>
568
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
11
903
fd8888f6f037 Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents: 898
diff changeset
12 struct charset_translation {
608
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
13 iconv_t cd;
3863
55df57c028d4 Added "bool" type and changed all ints that were used as booleans to bool.
Timo Sirainen <tss@iki.fi>
parents: 1991
diff changeset
14 bool ascii;
608
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
15 };
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
16
903
fd8888f6f037 Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents: 898
diff changeset
17 struct charset_translation *charset_to_utf8_begin(const char *charset,
3863
55df57c028d4 Added "bool" type and changed all ints that were used as booleans to bool.
Timo Sirainen <tss@iki.fi>
parents: 1991
diff changeset
18 bool *unknown_charset)
608
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
19 {
903
fd8888f6f037 Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents: 898
diff changeset
20 struct charset_translation *t;
608
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
21 iconv_t cd;
3863
55df57c028d4 Added "bool" type and changed all ints that were used as booleans to bool.
Timo Sirainen <tss@iki.fi>
parents: 1991
diff changeset
22 bool ascii;
608
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
23
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
24 if (unknown_charset != NULL)
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
25 *unknown_charset = FALSE;
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
26
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
27 if (strcasecmp(charset, "us-ascii") == 0 ||
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
28 strcasecmp(charset, "ascii") == 0) {
1991
689f791b480f iconv_t isn't necessarily pointer.
Timo Sirainen <tss@iki.fi>
parents: 1471
diff changeset
29 cd = (iconv_t)-1;
609
5470c0cb13a7 We can support UTF-8 charset too without any translations.
Timo Sirainen <tss@iki.fi>
parents: 608
diff changeset
30 ascii = TRUE;
5470c0cb13a7 We can support UTF-8 charset too without any translations.
Timo Sirainen <tss@iki.fi>
parents: 608
diff changeset
31 } else if (strcasecmp(charset, "UTF-8") == 0 ||
5470c0cb13a7 We can support UTF-8 charset too without any translations.
Timo Sirainen <tss@iki.fi>
parents: 608
diff changeset
32 strcasecmp(charset, "UTF8") == 0) {
1991
689f791b480f iconv_t isn't necessarily pointer.
Timo Sirainen <tss@iki.fi>
parents: 1471
diff changeset
33 cd = (iconv_t)-1;
609
5470c0cb13a7 We can support UTF-8 charset too without any translations.
Timo Sirainen <tss@iki.fi>
parents: 608
diff changeset
34 ascii = FALSE;
608
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
35 } else {
609
5470c0cb13a7 We can support UTF-8 charset too without any translations.
Timo Sirainen <tss@iki.fi>
parents: 608
diff changeset
36 ascii = FALSE;
5470c0cb13a7 We can support UTF-8 charset too without any translations.
Timo Sirainen <tss@iki.fi>
parents: 608
diff changeset
37 cd = iconv_open("UTF-8", charset);
608
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
38 if (cd == (iconv_t)-1) {
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
39 if (unknown_charset != NULL)
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
40 *unknown_charset = TRUE;
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
41 return NULL;
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
42 }
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
43 }
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
44
903
fd8888f6f037 Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents: 898
diff changeset
45 t = i_new(struct charset_translation, 1);
608
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
46 t->cd = cd;
609
5470c0cb13a7 We can support UTF-8 charset too without any translations.
Timo Sirainen <tss@iki.fi>
parents: 608
diff changeset
47 t->ascii = ascii;
608
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
48 return t;
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
49 }
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
50
903
fd8888f6f037 Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents: 898
diff changeset
51 void charset_to_utf8_end(struct charset_translation *t)
608
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
52 {
1991
689f791b480f iconv_t isn't necessarily pointer.
Timo Sirainen <tss@iki.fi>
parents: 1471
diff changeset
53 if (t->cd != (iconv_t)-1)
608
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
54 iconv_close(t->cd);
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
55 i_free(t);
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
56 }
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
57
903
fd8888f6f037 Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents: 898
diff changeset
58 void charset_to_utf8_reset(struct charset_translation *t)
608
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
59 {
1991
689f791b480f iconv_t isn't necessarily pointer.
Timo Sirainen <tss@iki.fi>
parents: 1471
diff changeset
60 if (t->cd != (iconv_t)-1)
608
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
61 (void)iconv(t->cd, NULL, NULL, NULL, NULL);
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
62 }
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
63
903
fd8888f6f037 Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents: 898
diff changeset
64 enum charset_result
fd8888f6f037 Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents: 898
diff changeset
65 charset_to_ucase_utf8(struct charset_translation *t,
fd8888f6f037 Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents: 898
diff changeset
66 const unsigned char *src, size_t *src_size,
fd8888f6f037 Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents: 898
diff changeset
67 buffer_t *dest)
765
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
68 {
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
69 ICONV_CONST char *ic_srcbuf;
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
70 char *ic_destbuf;
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
71 size_t srcleft, destpos, destleft, size;
903
fd8888f6f037 Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents: 898
diff changeset
72 enum charset_result ret;
765
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
73
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
74 destpos = buffer_get_used_size(dest);
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
75 destleft = buffer_get_size(dest) - destpos;
608
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
76
1991
689f791b480f iconv_t isn't necessarily pointer.
Timo Sirainen <tss@iki.fi>
parents: 1471
diff changeset
77 if (t->cd == (iconv_t)-1) {
609
5470c0cb13a7 We can support UTF-8 charset too without any translations.
Timo Sirainen <tss@iki.fi>
parents: 608
diff changeset
78 /* no translation needed - just copy it to outbuf uppercased */
785
d96cbba73a8b Don't use Buffers with read-only data, just makes it more difficult without
Timo Sirainen <tss@iki.fi>
parents: 766
diff changeset
79 if (*src_size > destleft)
d96cbba73a8b Don't use Buffers with read-only data, just makes it more difficult without
Timo Sirainen <tss@iki.fi>
parents: 766
diff changeset
80 *src_size = destleft;
d96cbba73a8b Don't use Buffers with read-only data, just makes it more difficult without
Timo Sirainen <tss@iki.fi>
parents: 766
diff changeset
81 _charset_utf8_ucase(src, *src_size, dest, destpos);
765
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
82 return CHARSET_RET_OK;
608
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
83 }
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
84
765
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
85 size = destleft;
785
d96cbba73a8b Don't use Buffers with read-only data, just makes it more difficult without
Timo Sirainen <tss@iki.fi>
parents: 766
diff changeset
86 srcleft = *src_size;
d96cbba73a8b Don't use Buffers with read-only data, just makes it more difficult without
Timo Sirainen <tss@iki.fi>
parents: 766
diff changeset
87 ic_srcbuf = (ICONV_CONST char *) src;
1471
8f56379c3917 Renamed buffer_*_space() to buffer_*_space_unsafe() and added several
Timo Sirainen <tss@iki.fi>
parents: 1300
diff changeset
88 ic_destbuf = buffer_append_space_unsafe(dest, destleft);
765
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
89
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
90 if (iconv(t->cd, &ic_srcbuf, &srcleft,
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
91 &ic_destbuf, &destleft) != (size_t)-1)
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
92 ret = CHARSET_RET_OK;
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
93 else if (errno == E2BIG)
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
94 ret = CHARSET_RET_OUTPUT_FULL;
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
95 else if (errno == EINVAL)
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
96 ret = CHARSET_RET_INCOMPLETE_INPUT;
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
97 else {
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
98 /* should be EILSEQ */
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
99 return CHARSET_RET_INVALID_INPUT;
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
100 }
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
101 size -= destleft;
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
102
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
103 /* give back the memory we didn't use */
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
104 buffer_set_used_size(dest, buffer_get_used_size(dest) - destleft);
608
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
105
785
d96cbba73a8b Don't use Buffers with read-only data, just makes it more difficult without
Timo Sirainen <tss@iki.fi>
parents: 766
diff changeset
106 *src_size -= srcleft;
766
03832c7f389b Compiles again without iconv()
Timo Sirainen <tss@iki.fi>
parents: 765
diff changeset
107 _charset_utf8_ucase((unsigned char *) ic_destbuf - size, size,
03832c7f389b Compiles again without iconv()
Timo Sirainen <tss@iki.fi>
parents: 765
diff changeset
108 dest, destpos);
765
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
109 return ret;
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
110 }
608
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
111
792
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
112 static const char *
3863
55df57c028d4 Added "bool" type and changed all ints that were used as booleans to bool.
Timo Sirainen <tss@iki.fi>
parents: 1991
diff changeset
113 charset_to_utf8_string_int(const char *charset, bool *unknown_charset,
792
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
114 const unsigned char *data, size_t size,
3863
55df57c028d4 Added "bool" type and changed all ints that were used as booleans to bool.
Timo Sirainen <tss@iki.fi>
parents: 1991
diff changeset
115 size_t *utf8_size_r, bool ucase)
568
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
116 {
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
117 iconv_t cd;
611
9373933b1be1 Removed warnings with Solaris' iconv()
Timo Sirainen <tss@iki.fi>
parents: 609
diff changeset
118 ICONV_CONST char *inbuf;
9373933b1be1 Removed warnings with Solaris' iconv()
Timo Sirainen <tss@iki.fi>
parents: 609
diff changeset
119 char *outbuf, *outpos;
568
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
120 size_t inleft, outleft, outsize, pos;
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
121
608
debb8468514e SEARCH CHARSET now works properly with message bodies, and in general body
Timo Sirainen <tss@iki.fi>
parents: 579
diff changeset
122 if (charset == NULL || strcasecmp(charset, "us-ascii") == 0 ||
765
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
123 strcasecmp(charset, "ascii") == 0 ||
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
124 strcasecmp(charset, "UTF-8") == 0 ||
766
03832c7f389b Compiles again without iconv()
Timo Sirainen <tss@iki.fi>
parents: 765
diff changeset
125 strcasecmp(charset, "UTF8") == 0) {
03832c7f389b Compiles again without iconv()
Timo Sirainen <tss@iki.fi>
parents: 765
diff changeset
126 if (unknown_charset != NULL)
03832c7f389b Compiles again without iconv()
Timo Sirainen <tss@iki.fi>
parents: 765
diff changeset
127 *unknown_charset = FALSE;
792
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
128
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
129 if (!ucase) {
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
130 if (utf8_size_r != NULL)
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
131 *utf8_size_r = size;
898
0d5be52d7131 Use unsigned char* when accessing non-NUL terminating strings. Compiler
Timo Sirainen <tss@iki.fi>
parents: 792
diff changeset
132 return t_strndup(data, size);
792
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
133 }
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
134
785
d96cbba73a8b Don't use Buffers with read-only data, just makes it more difficult without
Timo Sirainen <tss@iki.fi>
parents: 766
diff changeset
135 return _charset_utf8_ucase_strdup(data, size, utf8_size_r);
766
03832c7f389b Compiles again without iconv()
Timo Sirainen <tss@iki.fi>
parents: 765
diff changeset
136 }
568
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
137
609
5470c0cb13a7 We can support UTF-8 charset too without any translations.
Timo Sirainen <tss@iki.fi>
parents: 608
diff changeset
138 cd = iconv_open("UTF-8", charset);
568
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
139 if (cd == (iconv_t)-1) {
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
140 if (unknown_charset != NULL)
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
141 *unknown_charset = TRUE;
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
142 return NULL;
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
143 }
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
144
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
145 if (unknown_charset != NULL)
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
146 *unknown_charset = FALSE;
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
147
785
d96cbba73a8b Don't use Buffers with read-only data, just makes it more difficult without
Timo Sirainen <tss@iki.fi>
parents: 766
diff changeset
148 inbuf = (ICONV_CONST char *) data;
d96cbba73a8b Don't use Buffers with read-only data, just makes it more difficult without
Timo Sirainen <tss@iki.fi>
parents: 766
diff changeset
149 inleft = size;
568
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
150
765
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
151 outsize = outleft = inleft * 2;
568
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
152 outbuf = outpos = t_buffer_get(outsize + 1);
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
153
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
154 while (iconv(cd, &inbuf, &inleft, &outpos, &outleft) == (size_t)-1) {
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
155 if (errno != E2BIG) {
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
156 /* invalid data */
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
157 iconv_close(cd);
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
158 return NULL;
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
159 }
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
160
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
161 /* output buffer too small, grow it */
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
162 pos = outsize - outleft;
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
163 outsize *= 2;
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
164 outleft = outsize - pos;
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
165
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
166 outbuf = t_buffer_reget(outbuf, outsize + 1);
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
167 outpos = outbuf + pos;
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
168 }
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
169
785
d96cbba73a8b Don't use Buffers with read-only data, just makes it more difficult without
Timo Sirainen <tss@iki.fi>
parents: 766
diff changeset
170 if (utf8_size_r != NULL)
d96cbba73a8b Don't use Buffers with read-only data, just makes it more difficult without
Timo Sirainen <tss@iki.fi>
parents: 766
diff changeset
171 *utf8_size_r = (size_t) (outpos - outbuf);
568
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
172 *outpos++ = '\0';
765
553f050c8313 Added buffer API. Point is to hide all buffer writing behind this API which
Timo Sirainen <tss@iki.fi>
parents: 753
diff changeset
173 t_buffer_alloc((size_t) (outpos - outbuf));
568
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
174
792
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
175 if (ucase)
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
176 str_ucase(outbuf); /* FIXME: utf8 */
568
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
177
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
178 iconv_close(cd);
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
179 return outbuf;
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
180 }
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
181
792
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
182 const char *
3863
55df57c028d4 Added "bool" type and changed all ints that were used as booleans to bool.
Timo Sirainen <tss@iki.fi>
parents: 1991
diff changeset
183 charset_to_utf8_string(const char *charset, bool *unknown_charset,
792
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
184 const unsigned char *data, size_t size,
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
185 size_t *utf8_size_r)
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
186 {
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
187 return charset_to_utf8_string_int(charset, unknown_charset,
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
188 data, size, utf8_size_r, FALSE);
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
189 }
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
190
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
191 const char *
3863
55df57c028d4 Added "bool" type and changed all ints that were used as booleans to bool.
Timo Sirainen <tss@iki.fi>
parents: 1991
diff changeset
192 charset_to_ucase_utf8_string(const char *charset, bool *unknown_charset,
792
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
193 const unsigned char *data, size_t size,
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
194 size_t *utf8_size_r)
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
195 {
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
196 return charset_to_utf8_string_int(charset, unknown_charset,
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
197 data, size, utf8_size_r, TRUE);
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
198 }
d573c53946ac Full not-too-well-tested support for SORT extension. Required a few
Timo Sirainen <tss@iki.fi>
parents: 785
diff changeset
199
568
f2aa58c2afd0 SEARCH CHARSET support. Currently we do it through iconv() and only ASCII
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
200 #endif