annotate src/lib-mail/message-header-decode.c @ 6954:aee731c09668 HEAD

If we encounter an unknown charset, get the ASCII (or UTF-8 actually) characters instead of ignoring it completely.
author Timo Sirainen <tss@iki.fi>
date Sat, 08 Dec 2007 15:59:48 +0200
parents 63e225ab7361
children 7ed926ed7aa4
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
6429
65c69a53a7be Replaced my Copyright notices. The year range always ends with 2007 now.
Timo Sirainen <tss@iki.fi>
parents: 6133
diff changeset
1 /* Copyright (c) 2002-2007 Dovecot authors, see the included COPYING file */
783
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
2
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
3 #include "lib.h"
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
4 #include "base64.h"
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
5 #include "buffer.h"
6131
5f56b2eb32b3 Use uni_utf8_to_decomposed_titlecase() to have proper case-insensitive UTF-8
Timo Sirainen <tss@iki.fi>
parents: 6123
diff changeset
6 #include "unichar.h"
6118
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
7 #include "charset-utf8.h"
783
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
8 #include "quoted-printable.h"
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
9 #include "message-header-decode.h"
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
10
5517
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
11 static size_t
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
12 message_header_decode_encoded(const unsigned char *data, size_t size,
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
13 buffer_t *decodebuf, unsigned int *charsetlen_r)
783
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
14 {
5517
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
15 #define QCOUNT 3
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
16 unsigned int num = 0;
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
17 size_t i, start_pos[QCOUNT];
783
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
18
5517
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
19 /* data should contain "charset?encoding?text?=" */
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
20 for (i = 0; i < size; i++) {
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
21 if (data[i] == '?') {
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
22 start_pos[num++] = i;
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
23 if (num == QCOUNT)
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
24 break;
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
25 }
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
26 }
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
27 if (i == size || data[i+1] != '=') {
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
28 /* invalid block */
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
29 return 0;
783
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
30 }
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
31
5517
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
32 buffer_append(decodebuf, data, start_pos[0]);
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
33 buffer_append_c(decodebuf, '\0');
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
34 *charsetlen_r = decodebuf->used;
783
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
35
5517
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
36 switch (data[start_pos[0]+1]) {
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
37 case 'q':
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
38 case 'Q':
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
39 quoted_printable_decode(data + start_pos[1] + 1,
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
40 start_pos[2] - start_pos[1] - 1,
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
41 NULL, decodebuf);
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
42 break;
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
43 case 'b':
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
44 case 'B':
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
45 if (base64_decode(data + start_pos[1] + 1,
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
46 start_pos[2] - start_pos[1] - 1,
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
47 NULL, decodebuf) < 0) {
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
48 /* contains invalid data. show what we got so far. */
783
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
49 }
5517
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
50 break;
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
51 default:
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
52 /* unknown encoding */
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
53 return 0;
783
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
54 }
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
55
5517
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
56 return start_pos[2] + 2;
783
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
57 }
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
58
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
59 void message_header_decode(const unsigned char *data, size_t size,
1038
60646878858e Function typedefs now define them as functions, not function pointers.
Timo Sirainen <tss@iki.fi>
parents: 953
diff changeset
60 message_header_decode_callback_t *callback,
953
411006be3c66 Naming change for function typedefs.
Timo Sirainen <tss@iki.fi>
parents: 903
diff changeset
61 void *context)
783
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
62 {
5517
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
63 buffer_t *decodebuf = NULL;
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
64 unsigned int charsetlen = 0;
6950
63e225ab7361 If header contains invalid MIME input, preserve the =? in output.
Timo Sirainen <tss@iki.fi>
parents: 6910
diff changeset
65 size_t pos, start_pos, ret;
783
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
66
5517
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
67 /* =?charset?Q|B?text?= */
783
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
68 start_pos = pos = 0;
5517
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
69 for (pos = 0; pos + 1 < size; ) {
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
70 if (data[pos] != '=' || data[pos+1] != '?') {
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
71 pos++;
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
72 continue;
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
73 }
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
74
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
75 /* encoded string beginning */
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
76 if (pos != start_pos) {
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
77 /* send the unencoded data so far */
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
78 if (!callback(data + start_pos, pos - start_pos,
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
79 NULL, context)) {
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
80 start_pos = size;
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
81 break;
783
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
82 }
5517
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
83 }
783
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
84
5517
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
85 if (decodebuf == NULL) {
6537
58d9f94b9919 message_header_decode(): Don't use t_push/t_pop. The callback may want to
Timo Sirainen <tss@iki.fi>
parents: 6429
diff changeset
86 decodebuf = buffer_create_dynamic(default_pool,
58d9f94b9919 message_header_decode(): Don't use t_push/t_pop. The callback may want to
Timo Sirainen <tss@iki.fi>
parents: 6429
diff changeset
87 size - pos);
5517
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
88 } else {
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
89 buffer_set_used_size(decodebuf, 0);
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
90 }
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
91
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
92 pos += 2;
6950
63e225ab7361 If header contains invalid MIME input, preserve the =? in output.
Timo Sirainen <tss@iki.fi>
parents: 6910
diff changeset
93 ret = message_header_decode_encoded(data + pos, size - pos,
63e225ab7361 If header contains invalid MIME input, preserve the =? in output.
Timo Sirainen <tss@iki.fi>
parents: 6910
diff changeset
94 decodebuf, &charsetlen);
63e225ab7361 If header contains invalid MIME input, preserve the =? in output.
Timo Sirainen <tss@iki.fi>
parents: 6910
diff changeset
95 if (ret == 0) {
63e225ab7361 If header contains invalid MIME input, preserve the =? in output.
Timo Sirainen <tss@iki.fi>
parents: 6910
diff changeset
96 start_pos = pos-2;
63e225ab7361 If header contains invalid MIME input, preserve the =? in output.
Timo Sirainen <tss@iki.fi>
parents: 6910
diff changeset
97 continue;
63e225ab7361 If header contains invalid MIME input, preserve the =? in output.
Timo Sirainen <tss@iki.fi>
parents: 6910
diff changeset
98 }
63e225ab7361 If header contains invalid MIME input, preserve the =? in output.
Timo Sirainen <tss@iki.fi>
parents: 6910
diff changeset
99 pos += ret;
783
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
100
5517
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
101 if (decodebuf->used > charsetlen) {
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
102 /* decodebuf contains <charset> NUL <text> */
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
103 if (!callback(CONST_PTR_OFFSET(decodebuf->data,
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
104 charsetlen),
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
105 decodebuf->used - charsetlen,
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
106 decodebuf->data, context)) {
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
107 start_pos = size;
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
108 break;
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
109 }
783
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
110 }
5517
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
111
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
112 start_pos = pos;
783
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
113 }
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
114
5517
2608b9e18bbb Rewrite/cleanup
Timo Sirainen <tss@iki.fi>
parents: 5503
diff changeset
115 if (size != start_pos) {
5503
b7573b83f999 Minor optimization.
Timo Sirainen <tss@iki.fi>
parents: 4906
diff changeset
116 (void)callback(data + start_pos, size - start_pos,
b7573b83f999 Minor optimization.
Timo Sirainen <tss@iki.fi>
parents: 4906
diff changeset
117 NULL, context);
b7573b83f999 Minor optimization.
Timo Sirainen <tss@iki.fi>
parents: 4906
diff changeset
118 }
6538
1b7d91f7b151 Fix for last change: don't crash if temp buffer wasn't created..
Timo Sirainen <tss@iki.fi>
parents: 6537
diff changeset
119 if (decodebuf != NULL)
1b7d91f7b151 Fix for last change: don't crash if temp buffer wasn't created..
Timo Sirainen <tss@iki.fi>
parents: 6537
diff changeset
120 buffer_free(&decodebuf);
783
d826ab8c8d62 Quoted-Printable decoding was buggy. Extracted RFC2047 header decoding into
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
121 }
6118
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
122
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
123 struct decode_utf8_context {
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
124 buffer_t *dest;
6121
157e31742eeb message_header_decode_utf8() returns not TRUE if it actually changed
Timo Sirainen <tss@iki.fi>
parents: 6118
diff changeset
125 unsigned int changed:1;
157e31742eeb message_header_decode_utf8() returns not TRUE if it actually changed
Timo Sirainen <tss@iki.fi>
parents: 6118
diff changeset
126 unsigned int called:1;
6133
fd9f6b4d846e Renamed ucase parameters to dtcase which is more correct.
Timo Sirainen <tss@iki.fi>
parents: 6132
diff changeset
127 unsigned int dtcase:1;
6118
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
128 };
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
129
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
130 static bool
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
131 decode_utf8_callback(const unsigned char *data, size_t size,
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
132 const char *charset, void *context)
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
133 {
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
134 struct decode_utf8_context *ctx = context;
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
135 struct charset_translation *t;
6910
c68564884bae charset_to_utf8_begin() was called with bad flags parameter (although it
Timo Sirainen <tss@iki.fi>
parents: 6538
diff changeset
136 enum charset_flags flags;
6118
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
137
6121
157e31742eeb message_header_decode_utf8() returns not TRUE if it actually changed
Timo Sirainen <tss@iki.fi>
parents: 6118
diff changeset
138 /* one call with charset=NULL means nothing changed */
6123
d3299ed1332a Fixed message_header_decode_utf8() to return FALSE also for single
Timo Sirainen <tss@iki.fi>
parents: 6122
diff changeset
139 if (!ctx->called && charset == NULL)
6121
157e31742eeb message_header_decode_utf8() returns not TRUE if it actually changed
Timo Sirainen <tss@iki.fi>
parents: 6118
diff changeset
140 ctx->called = TRUE;
157e31742eeb message_header_decode_utf8() returns not TRUE if it actually changed
Timo Sirainen <tss@iki.fi>
parents: 6118
diff changeset
141 else
157e31742eeb message_header_decode_utf8() returns not TRUE if it actually changed
Timo Sirainen <tss@iki.fi>
parents: 6118
diff changeset
142 ctx->changed = TRUE;
157e31742eeb message_header_decode_utf8() returns not TRUE if it actually changed
Timo Sirainen <tss@iki.fi>
parents: 6118
diff changeset
143
6118
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
144 if (charset == NULL || charset_is_utf8(charset)) {
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
145 /* ASCII / UTF-8 */
6133
fd9f6b4d846e Renamed ucase parameters to dtcase which is more correct.
Timo Sirainen <tss@iki.fi>
parents: 6132
diff changeset
146 if (ctx->dtcase) {
6131
5f56b2eb32b3 Use uni_utf8_to_decomposed_titlecase() to have proper case-insensitive UTF-8
Timo Sirainen <tss@iki.fi>
parents: 6123
diff changeset
147 (void)uni_utf8_to_decomposed_titlecase(data, size,
5f56b2eb32b3 Use uni_utf8_to_decomposed_titlecase() to have proper case-insensitive UTF-8
Timo Sirainen <tss@iki.fi>
parents: 6123
diff changeset
148 ctx->dest);
6118
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
149 } else {
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
150 buffer_append(ctx->dest, data, size);
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
151 }
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
152 return TRUE;
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
153 }
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
154
6910
c68564884bae charset_to_utf8_begin() was called with bad flags parameter (although it
Timo Sirainen <tss@iki.fi>
parents: 6538
diff changeset
155 flags = ctx->dtcase ? CHARSET_FLAG_DECOMP_TITLECASE : 0;
c68564884bae charset_to_utf8_begin() was called with bad flags parameter (although it
Timo Sirainen <tss@iki.fi>
parents: 6538
diff changeset
156 if (charset_to_utf8_begin(charset, flags, &t) < 0) {
6954
aee731c09668 If we encounter an unknown charset, get the ASCII (or UTF-8 actually)
Timo Sirainen <tss@iki.fi>
parents: 6950
diff changeset
157 /* data probably still contains some valid ASCII characters.
aee731c09668 If we encounter an unknown charset, get the ASCII (or UTF-8 actually)
Timo Sirainen <tss@iki.fi>
parents: 6950
diff changeset
158 append them. */
aee731c09668 If we encounter an unknown charset, get the ASCII (or UTF-8 actually)
Timo Sirainen <tss@iki.fi>
parents: 6950
diff changeset
159 if (uni_utf8_get_valid_data(data, size, ctx->dest))
aee731c09668 If we encounter an unknown charset, get the ASCII (or UTF-8 actually)
Timo Sirainen <tss@iki.fi>
parents: 6950
diff changeset
160 buffer_append(ctx->dest, data, size);
6118
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
161 return TRUE;
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
162 }
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
163
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
164 /* ignore any errors */
6122
d86581f4a0c6 charset_to_utf8() isn't used anymore, so renamed charset_to_utf8_full() to it.
Timo Sirainen <tss@iki.fi>
parents: 6121
diff changeset
165 (void)charset_to_utf8(t, data, &size, ctx->dest);
6118
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
166 charset_to_utf8_end(&t);
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
167 return TRUE;
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
168 }
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
169
6121
157e31742eeb message_header_decode_utf8() returns not TRUE if it actually changed
Timo Sirainen <tss@iki.fi>
parents: 6118
diff changeset
170 bool message_header_decode_utf8(const unsigned char *data, size_t size,
6133
fd9f6b4d846e Renamed ucase parameters to dtcase which is more correct.
Timo Sirainen <tss@iki.fi>
parents: 6132
diff changeset
171 buffer_t *dest, bool dtcase)
6118
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
172 {
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
173 struct decode_utf8_context ctx;
6123
d3299ed1332a Fixed message_header_decode_utf8() to return FALSE also for single
Timo Sirainen <tss@iki.fi>
parents: 6122
diff changeset
174 size_t used = dest->used;
6118
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
175
6121
157e31742eeb message_header_decode_utf8() returns not TRUE if it actually changed
Timo Sirainen <tss@iki.fi>
parents: 6118
diff changeset
176 memset(&ctx, 0, sizeof(ctx));
6118
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
177 ctx.dest = dest;
6133
fd9f6b4d846e Renamed ucase parameters to dtcase which is more correct.
Timo Sirainen <tss@iki.fi>
parents: 6132
diff changeset
178 ctx.dtcase = dtcase;
6118
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
179 message_header_decode(data, size, decode_utf8_callback, &ctx);
6123
d3299ed1332a Fixed message_header_decode_utf8() to return FALSE also for single
Timo Sirainen <tss@iki.fi>
parents: 6122
diff changeset
180 return ctx.changed || (dest->used - used != size);
6118
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 5517
diff changeset
181 }