annotate src/lib-mail/message-decoder.c @ 13121:15082db3225a

lib-mail: Added message_decoder_set_return_binary()
author Timo Sirainen <tss@iki.fi>
date Thu, 28 Jul 2011 17:01:26 +0300
parents c392158f374d
children ba770cba5598
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
12782
447bce266022 Updated copyright notices to include year 2011.
Timo Sirainen <tss@iki.fi>
parents: 11074
diff changeset
1 /* Copyright (c) 2006-2011 Dovecot authors, see the included COPYING file */
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
2
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
3 #include "lib.h"
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
4 #include "buffer.h"
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
5 #include "base64.h"
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
6 #include "str.h"
6131
5f56b2eb32b3 Use uni_utf8_to_decomposed_titlecase() to have proper case-insensitive UTF-8
Timo Sirainen <tss@iki.fi>
parents: 6122
diff changeset
7 #include "unichar.h"
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
8 #include "charset-utf8.h"
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
9 #include "quoted-printable.h"
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
10 #include "rfc822-parser.h"
7950
3412c43d6707 Merge RFC 2231 header continuations in BODY/BODYSTRUCTURE replies. Also use
Timo Sirainen <tss@iki.fi>
parents: 7226
diff changeset
11 #include "rfc2231-parser.h"
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
12 #include "message-parser.h"
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
13 #include "message-header-decode.h"
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
14 #include "message-decoder.h"
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
15
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
16 enum content_type {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
17 CONTENT_TYPE_UNKNOWN = 0,
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
18 CONTENT_TYPE_BINARY,
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
19 CONTENT_TYPE_QP,
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
20 CONTENT_TYPE_BASE64
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
21 };
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
22
4885
Timo Sirainen <tss@iki.fi>
parents: 4708
diff changeset
23 /* base64 takes max 4 bytes per character, q-p takes max 3. */
Timo Sirainen <tss@iki.fi>
parents: 4708
diff changeset
24 #define MAX_ENCODING_BUF_SIZE 3
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
25
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
26 /* UTF-8 takes max 5 bytes per character. Not sure about others, but I'd think
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
27 10 is more than enough for everyone.. */
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
28 #define MAX_TRANSLATION_BUF_SIZE 10
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
29
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
30 struct message_decoder_context {
9322
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
31 enum message_decoder_flags flags;
5410
73032525bca0 Don't try to access output variable's contents. They are undefined at least
Timo Sirainen <tss@iki.fi>
parents: 4903
diff changeset
32 struct message_part *prev_part;
73032525bca0 Don't try to access output variable's contents. They are undefined at least
Timo Sirainen <tss@iki.fi>
parents: 4903
diff changeset
33
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
34 struct message_header_line hdr;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
35 buffer_t *buf, *buf2;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
36
6916
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
37 char *charset_trans_charset;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
38 struct charset_translation *charset_trans;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
39 char translation_buf[MAX_TRANSLATION_BUF_SIZE];
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
40 unsigned int translation_size;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
41
9664
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
42 buffer_t *encoding_buf;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
43
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
44 char *content_charset;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
45 enum content_type content_type;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
46
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
47 unsigned int charset_utf8:1;
9322
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
48 unsigned int binary_input:1;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
49 };
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
50
13121
15082db3225a lib-mail: Added message_decoder_set_return_binary()
Timo Sirainen <tss@iki.fi>
parents: 13015
diff changeset
51 static void
15082db3225a lib-mail: Added message_decoder_set_return_binary()
Timo Sirainen <tss@iki.fi>
parents: 13015
diff changeset
52 message_decode_body_init_charset(struct message_decoder_context *ctx,
15082db3225a lib-mail: Added message_decoder_set_return_binary()
Timo Sirainen <tss@iki.fi>
parents: 13015
diff changeset
53 struct message_part *part);
15082db3225a lib-mail: Added message_decoder_set_return_binary()
Timo Sirainen <tss@iki.fi>
parents: 13015
diff changeset
54
9322
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
55 struct message_decoder_context *
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
56 message_decoder_init(enum message_decoder_flags flags)
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
57 {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
58 struct message_decoder_context *ctx;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
59
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
60 ctx = i_new(struct message_decoder_context, 1);
9322
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
61 ctx->flags = flags;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
62 ctx->buf = buffer_create_dynamic(default_pool, 8192);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
63 ctx->buf2 = buffer_create_dynamic(default_pool, 8192);
9664
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
64 ctx->encoding_buf = buffer_create_dynamic(default_pool, 128);
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
65 return ctx;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
66 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
67
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
68 void message_decoder_deinit(struct message_decoder_context **_ctx)
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
69 {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
70 struct message_decoder_context *ctx = *_ctx;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
71
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
72 *_ctx = NULL;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
73
4888
c3e15280e4cb Memory leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4885
diff changeset
74 if (ctx->charset_trans != NULL)
c3e15280e4cb Memory leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4885
diff changeset
75 charset_to_utf8_end(&ctx->charset_trans);
c3e15280e4cb Memory leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4885
diff changeset
76
9664
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
77 buffer_free(&ctx->encoding_buf);
6414
a6a49d5efc59 Changed buffer_free() and buffer_free_without_data() APIs to take ** pointer
Timo Sirainen <tss@iki.fi>
parents: 6133
diff changeset
78 buffer_free(&ctx->buf);
a6a49d5efc59 Changed buffer_free() and buffer_free_without_data() APIs to take ** pointer
Timo Sirainen <tss@iki.fi>
parents: 6133
diff changeset
79 buffer_free(&ctx->buf2);
6916
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
80 i_free(ctx->charset_trans_charset);
4888
c3e15280e4cb Memory leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4885
diff changeset
81 i_free(ctx->content_charset);
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
82 i_free(ctx);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
83 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
84
13121
15082db3225a lib-mail: Added message_decoder_set_return_binary()
Timo Sirainen <tss@iki.fi>
parents: 13015
diff changeset
85 void message_decoder_set_return_binary(struct message_decoder_context *ctx,
15082db3225a lib-mail: Added message_decoder_set_return_binary()
Timo Sirainen <tss@iki.fi>
parents: 13015
diff changeset
86 bool set)
15082db3225a lib-mail: Added message_decoder_set_return_binary()
Timo Sirainen <tss@iki.fi>
parents: 13015
diff changeset
87 {
15082db3225a lib-mail: Added message_decoder_set_return_binary()
Timo Sirainen <tss@iki.fi>
parents: 13015
diff changeset
88 if (set)
15082db3225a lib-mail: Added message_decoder_set_return_binary()
Timo Sirainen <tss@iki.fi>
parents: 13015
diff changeset
89 ctx->flags |= MESSAGE_DECODER_FLAG_RETURN_BINARY;
15082db3225a lib-mail: Added message_decoder_set_return_binary()
Timo Sirainen <tss@iki.fi>
parents: 13015
diff changeset
90 else
15082db3225a lib-mail: Added message_decoder_set_return_binary()
Timo Sirainen <tss@iki.fi>
parents: 13015
diff changeset
91 ctx->flags &= ~MESSAGE_DECODER_FLAG_RETURN_BINARY;
15082db3225a lib-mail: Added message_decoder_set_return_binary()
Timo Sirainen <tss@iki.fi>
parents: 13015
diff changeset
92 message_decode_body_init_charset(ctx, ctx->prev_part);
15082db3225a lib-mail: Added message_decoder_set_return_binary()
Timo Sirainen <tss@iki.fi>
parents: 13015
diff changeset
93 }
15082db3225a lib-mail: Added message_decoder_set_return_binary()
Timo Sirainen <tss@iki.fi>
parents: 13015
diff changeset
94
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
95 static void
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
96 parse_content_transfer_encoding(struct message_decoder_context *ctx,
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
97 struct message_header_line *hdr)
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
98 {
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
99 struct rfc822_parser_context parser;
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
100 string_t *value;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
101
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
102 value = t_str_new(64);
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
103 rfc822_parser_init(&parser, hdr->full_value, hdr->full_value_len, NULL);
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
104
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
105 (void)rfc822_skip_lwsp(&parser);
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
106 (void)rfc822_parse_mime_token(&parser, value);
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
107
7071
5de49bb0a7b3 If Content-Transfer-Encoding type is unknown, skip the body instead of
Timo Sirainen <tss@iki.fi>
parents: 6952
diff changeset
108 ctx->content_type = CONTENT_TYPE_UNKNOWN;
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
109 switch (str_len(value)) {
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
110 case 4:
6423
13b0a3ba5f7b Added i_ prefix to strcmp_p(), strcasecmp_p() and memcasecmp().
Timo Sirainen <tss@iki.fi>
parents: 6414
diff changeset
111 if (i_memcasecmp(str_data(value), "7bit", 4) == 0 ||
13b0a3ba5f7b Added i_ prefix to strcmp_p(), strcasecmp_p() and memcasecmp().
Timo Sirainen <tss@iki.fi>
parents: 6414
diff changeset
112 i_memcasecmp(str_data(value), "8bit", 4) == 0)
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
113 ctx->content_type = CONTENT_TYPE_BINARY;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
114 break;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
115 case 6:
6423
13b0a3ba5f7b Added i_ prefix to strcmp_p(), strcasecmp_p() and memcasecmp().
Timo Sirainen <tss@iki.fi>
parents: 6414
diff changeset
116 if (i_memcasecmp(str_data(value), "base64", 6) == 0)
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
117 ctx->content_type = CONTENT_TYPE_BASE64;
6423
13b0a3ba5f7b Added i_ prefix to strcmp_p(), strcasecmp_p() and memcasecmp().
Timo Sirainen <tss@iki.fi>
parents: 6414
diff changeset
118 else if (i_memcasecmp(str_data(value), "binary", 6) == 0)
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
119 ctx->content_type = CONTENT_TYPE_BINARY;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
120 break;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
121 case 16:
6423
13b0a3ba5f7b Added i_ prefix to strcmp_p(), strcasecmp_p() and memcasecmp().
Timo Sirainen <tss@iki.fi>
parents: 6414
diff changeset
122 if (i_memcasecmp(str_data(value), "quoted-printable", 16) == 0)
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
123 ctx->content_type = CONTENT_TYPE_QP;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
124 break;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
125 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
126 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
127
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
128 static void
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
129 parse_content_type(struct message_decoder_context *ctx,
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
130 struct message_header_line *hdr)
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
131 {
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
132 struct rfc822_parser_context parser;
7950
3412c43d6707 Merge RFC 2231 header continuations in BODY/BODYSTRUCTURE replies. Also use
Timo Sirainen <tss@iki.fi>
parents: 7226
diff changeset
133 const char *const *results;
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
134 string_t *str;
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
135
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
136 if (ctx->content_charset != NULL)
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
137 return;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
138
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
139 rfc822_parser_init(&parser, hdr->full_value, hdr->full_value_len, NULL);
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
140 (void)rfc822_skip_lwsp(&parser);
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
141 str = t_str_new(64);
6940
414c9d631a81 Replaced t_push/t_pop calls with T_FRAME*() macros.
Timo Sirainen <tss@iki.fi>
parents: 6916
diff changeset
142 if (rfc822_parse_content_type(&parser, str) <= 0)
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
143 return;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
144
7950
3412c43d6707 Merge RFC 2231 header continuations in BODY/BODYSTRUCTURE replies. Also use
Timo Sirainen <tss@iki.fi>
parents: 7226
diff changeset
145 (void)rfc2231_parse(&parser, &results);
3412c43d6707 Merge RFC 2231 header continuations in BODY/BODYSTRUCTURE replies. Also use
Timo Sirainen <tss@iki.fi>
parents: 7226
diff changeset
146 for (; *results != NULL; results += 2) {
3412c43d6707 Merge RFC 2231 header continuations in BODY/BODYSTRUCTURE replies. Also use
Timo Sirainen <tss@iki.fi>
parents: 7226
diff changeset
147 if (strcasecmp(results[0], "charset") == 0) {
3412c43d6707 Merge RFC 2231 header continuations in BODY/BODYSTRUCTURE replies. Also use
Timo Sirainen <tss@iki.fi>
parents: 7226
diff changeset
148 ctx->content_charset = i_strdup(results[1]);
3412c43d6707 Merge RFC 2231 header continuations in BODY/BODYSTRUCTURE replies. Also use
Timo Sirainen <tss@iki.fi>
parents: 7226
diff changeset
149 ctx->charset_utf8 = charset_is_utf8(results[1]);
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
150 break;
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
151 }
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
152 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
153 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
154
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
155 static bool message_decode_header(struct message_decoder_context *ctx,
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
156 struct message_header_line *hdr,
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
157 struct message_block *output)
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
158 {
9322
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
159 bool dtcase = (ctx->flags & MESSAGE_DECODER_FLAG_DTCASE) != 0;
5511
518a77f3bf49 Renamed message_decoder_init() to message_decoder_init_ucase() and changed
Timo Sirainen <tss@iki.fi>
parents: 5510
diff changeset
160 size_t value_len;
518a77f3bf49 Renamed message_decoder_init() to message_decoder_init_ucase() and changed
Timo Sirainen <tss@iki.fi>
parents: 5510
diff changeset
161
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
162 if (hdr->continues) {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
163 hdr->use_full_value = TRUE;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
164 return FALSE;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
165 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
166
7226
e6693a0ec8e1 Renamed T_FRAME_BEGIN/END to T_BEGIN/END. Removed T_FRAME() macro and
Timo Sirainen <tss@iki.fi>
parents: 7086
diff changeset
167 T_BEGIN {
6940
414c9d631a81 Replaced t_push/t_pop calls with T_FRAME*() macros.
Timo Sirainen <tss@iki.fi>
parents: 6916
diff changeset
168 if (hdr->name_len == 12 &&
414c9d631a81 Replaced t_push/t_pop calls with T_FRAME*() macros.
Timo Sirainen <tss@iki.fi>
parents: 6916
diff changeset
169 strcasecmp(hdr->name, "Content-Type") == 0)
414c9d631a81 Replaced t_push/t_pop calls with T_FRAME*() macros.
Timo Sirainen <tss@iki.fi>
parents: 6916
diff changeset
170 parse_content_type(ctx, hdr);
414c9d631a81 Replaced t_push/t_pop calls with T_FRAME*() macros.
Timo Sirainen <tss@iki.fi>
parents: 6916
diff changeset
171 if (hdr->name_len == 25 &&
414c9d631a81 Replaced t_push/t_pop calls with T_FRAME*() macros.
Timo Sirainen <tss@iki.fi>
parents: 6916
diff changeset
172 strcasecmp(hdr->name, "Content-Transfer-Encoding") == 0)
414c9d631a81 Replaced t_push/t_pop calls with T_FRAME*() macros.
Timo Sirainen <tss@iki.fi>
parents: 6916
diff changeset
173 parse_content_transfer_encoding(ctx, hdr);
7226
e6693a0ec8e1 Renamed T_FRAME_BEGIN/END to T_BEGIN/END. Removed T_FRAME() macro and
Timo Sirainen <tss@iki.fi>
parents: 7086
diff changeset
174 } T_END;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
175
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
176 buffer_set_used_size(ctx->buf, 0);
6118
841b052e269e Added message_header_decode_utf8(). Use it in message decoder instead of
Timo Sirainen <tss@iki.fi>
parents: 6117
diff changeset
177 message_header_decode_utf8(hdr->full_value, hdr->full_value_len,
9322
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
178 ctx->buf, dtcase);
5511
518a77f3bf49 Renamed message_decoder_init() to message_decoder_init_ucase() and changed
Timo Sirainen <tss@iki.fi>
parents: 5510
diff changeset
179 value_len = ctx->buf->used;
518a77f3bf49 Renamed message_decoder_init() to message_decoder_init_ucase() and changed
Timo Sirainen <tss@iki.fi>
parents: 5510
diff changeset
180
9322
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
181 if (dtcase) {
6131
5f56b2eb32b3 Use uni_utf8_to_decomposed_titlecase() to have proper case-insensitive UTF-8
Timo Sirainen <tss@iki.fi>
parents: 6122
diff changeset
182 (void)uni_utf8_to_decomposed_titlecase(hdr->name, hdr->name_len,
5f56b2eb32b3 Use uni_utf8_to_decomposed_titlecase() to have proper case-insensitive UTF-8
Timo Sirainen <tss@iki.fi>
parents: 6122
diff changeset
183 ctx->buf);
6113
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
184 buffer_append_c(ctx->buf, '\0');
13015
c392158f374d message [header] decoder: Output only valid UTF-8 data.
Timo Sirainen <tss@iki.fi>
parents: 12782
diff changeset
185 } else {
c392158f374d message [header] decoder: Output only valid UTF-8 data.
Timo Sirainen <tss@iki.fi>
parents: 12782
diff changeset
186 if (!uni_utf8_get_valid_data((const unsigned char *)hdr->name,
c392158f374d message [header] decoder: Output only valid UTF-8 data.
Timo Sirainen <tss@iki.fi>
parents: 12782
diff changeset
187 hdr->name_len, ctx->buf))
c392158f374d message [header] decoder: Output only valid UTF-8 data.
Timo Sirainen <tss@iki.fi>
parents: 12782
diff changeset
188 buffer_append_c(ctx->buf, '\0');
6113
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
189 }
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
190
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
191 ctx->hdr = *hdr;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
192 ctx->hdr.full_value = ctx->buf->data;
5511
518a77f3bf49 Renamed message_decoder_init() to message_decoder_init_ucase() and changed
Timo Sirainen <tss@iki.fi>
parents: 5510
diff changeset
193 ctx->hdr.full_value_len = value_len;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
194 ctx->hdr.value_len = 0;
13015
c392158f374d message [header] decoder: Output only valid UTF-8 data.
Timo Sirainen <tss@iki.fi>
parents: 12782
diff changeset
195 if (ctx->buf->used != value_len) {
6113
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
196 ctx->hdr.name = CONST_PTR_OFFSET(ctx->buf->data,
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
197 ctx->hdr.full_value_len);
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
198 ctx->hdr.name_len = ctx->buf->used - 1 - value_len;
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
199 }
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
200
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
201 output->hdr = &ctx->hdr;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
202 return TRUE;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
203 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
204
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
205 static void translation_buf_decode(struct message_decoder_context *ctx,
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
206 const unsigned char **data, size_t *size)
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
207 {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
208 unsigned char trans_buf[MAX_TRANSLATION_BUF_SIZE+1];
6832
ea4d3778c99b Charset translation fixes
Timo Sirainen <tss@iki.fi>
parents: 6579
diff changeset
209 unsigned int data_wanted, skip;
ea4d3778c99b Charset translation fixes
Timo Sirainen <tss@iki.fi>
parents: 6579
diff changeset
210 size_t trans_size;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
211
6832
ea4d3778c99b Charset translation fixes
Timo Sirainen <tss@iki.fi>
parents: 6579
diff changeset
212 /* @UNSAFE: move the previously untranslated bytes to trans_buf
ea4d3778c99b Charset translation fixes
Timo Sirainen <tss@iki.fi>
parents: 6579
diff changeset
213 and see if we have now enough data to get the next character
ea4d3778c99b Charset translation fixes
Timo Sirainen <tss@iki.fi>
parents: 6579
diff changeset
214 translated */
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
215 memcpy(trans_buf, ctx->translation_buf, ctx->translation_size);
6832
ea4d3778c99b Charset translation fixes
Timo Sirainen <tss@iki.fi>
parents: 6579
diff changeset
216 data_wanted = sizeof(trans_buf) - ctx->translation_size;
ea4d3778c99b Charset translation fixes
Timo Sirainen <tss@iki.fi>
parents: 6579
diff changeset
217 if (data_wanted > *size)
ea4d3778c99b Charset translation fixes
Timo Sirainen <tss@iki.fi>
parents: 6579
diff changeset
218 data_wanted = *size;
ea4d3778c99b Charset translation fixes
Timo Sirainen <tss@iki.fi>
parents: 6579
diff changeset
219 memcpy(trans_buf + ctx->translation_size, *data, data_wanted);
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
220
6832
ea4d3778c99b Charset translation fixes
Timo Sirainen <tss@iki.fi>
parents: 6579
diff changeset
221 trans_size = ctx->translation_size + data_wanted;
ea4d3778c99b Charset translation fixes
Timo Sirainen <tss@iki.fi>
parents: 6579
diff changeset
222 (void)charset_to_utf8(ctx->charset_trans, trans_buf,
ea4d3778c99b Charset translation fixes
Timo Sirainen <tss@iki.fi>
parents: 6579
diff changeset
223 &trans_size, ctx->buf2);
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
224
6832
ea4d3778c99b Charset translation fixes
Timo Sirainen <tss@iki.fi>
parents: 6579
diff changeset
225 i_assert(trans_size > ctx->translation_size);
ea4d3778c99b Charset translation fixes
Timo Sirainen <tss@iki.fi>
parents: 6579
diff changeset
226 skip = trans_size - ctx->translation_size;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
227
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
228 i_assert(*size >= skip);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
229 *data += skip;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
230 *size -= skip;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
231
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
232 ctx->translation_size = 0;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
233 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
234
9322
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
235 static void
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
236 message_decode_body_init_charset(struct message_decoder_context *ctx,
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
237 struct message_part *part)
6916
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
238 {
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
239 enum charset_flags flags;
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
240
9360
fd3c13b9837b Message-decoder: Fixes to handling binary body parts.
Timo Sirainen <tss@iki.fi>
parents: 9322
diff changeset
241 ctx->binary_input = ctx->content_charset == NULL &&
fd3c13b9837b Message-decoder: Fixes to handling binary body parts.
Timo Sirainen <tss@iki.fi>
parents: 9322
diff changeset
242 (ctx->flags & MESSAGE_DECODER_FLAG_RETURN_BINARY) != 0 &&
fd3c13b9837b Message-decoder: Fixes to handling binary body parts.
Timo Sirainen <tss@iki.fi>
parents: 9322
diff changeset
243 (part->flags & (MESSAGE_PART_FLAG_TEXT |
fd3c13b9837b Message-decoder: Fixes to handling binary body parts.
Timo Sirainen <tss@iki.fi>
parents: 9322
diff changeset
244 MESSAGE_PART_FLAG_MESSAGE_RFC822)) == 0;
fd3c13b9837b Message-decoder: Fixes to handling binary body parts.
Timo Sirainen <tss@iki.fi>
parents: 9322
diff changeset
245
fd3c13b9837b Message-decoder: Fixes to handling binary body parts.
Timo Sirainen <tss@iki.fi>
parents: 9322
diff changeset
246 if (ctx->charset_utf8 || ctx->binary_input)
6916
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
247 return;
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
248
9360
fd3c13b9837b Message-decoder: Fixes to handling binary body parts.
Timo Sirainen <tss@iki.fi>
parents: 9322
diff changeset
249 if (ctx->charset_trans != NULL && ctx->content_charset != NULL &&
6916
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
250 strcasecmp(ctx->content_charset, ctx->charset_trans_charset) == 0) {
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
251 /* already have the correct translation selected */
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
252 return;
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
253 }
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
254
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
255 if (ctx->charset_trans != NULL)
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
256 charset_to_utf8_end(&ctx->charset_trans);
9322
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
257 i_free_and_null(ctx->charset_trans_charset);
6916
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
258
9322
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
259 flags = (ctx->flags & MESSAGE_DECODER_FLAG_DTCASE) != 0 ?
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
260 CHARSET_FLAG_DECOMP_TITLECASE : 0;
6916
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
261 ctx->charset_trans_charset = i_strdup(ctx->content_charset != NULL ?
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
262 ctx->content_charset : "UTF-8");
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
263 if (charset_to_utf8_begin(ctx->charset_trans_charset,
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
264 flags, &ctx->charset_trans) < 0)
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
265 ctx->charset_trans = NULL;
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
266 }
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
267
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
268 static bool message_decode_body(struct message_decoder_context *ctx,
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
269 struct message_block *input,
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
270 struct message_block *output)
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
271 {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
272 const unsigned char *data = NULL;
11074
c32f55615055 Small code changes to make static analyzer happier.
Timo Sirainen <tss@iki.fi>
parents: 10582
diff changeset
273 size_t pos = 0, size = 0;
5510
93e9e081855a Fix for base64 handling
Timo Sirainen <tss@iki.fi>
parents: 5422
diff changeset
274 int ret;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
275
9664
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
276 if (ctx->encoding_buf->used != 0) {
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
277 /* @UNSAFE */
9664
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
278 buffer_append(ctx->encoding_buf, input->data, input->size);
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
279 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
280
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
281 switch (ctx->content_type) {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
282 case CONTENT_TYPE_UNKNOWN:
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
283 /* just skip this body */
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
284 return FALSE;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
285
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
286 case CONTENT_TYPE_BINARY:
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
287 data = input->data;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
288 size = pos = input->size;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
289 break;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
290 case CONTENT_TYPE_QP:
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
291 buffer_set_used_size(ctx->buf, 0);
9664
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
292 if (ctx->encoding_buf->used != 0) {
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
293 quoted_printable_decode(ctx->encoding_buf->data,
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
294 ctx->encoding_buf->used,
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
295 &pos, ctx->buf);
9664
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
296 } else {
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
297 quoted_printable_decode(input->data, input->size,
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
298 &pos, ctx->buf);
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
299 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
300 data = ctx->buf->data;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
301 size = ctx->buf->used;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
302 break;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
303 case CONTENT_TYPE_BASE64:
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
304 buffer_set_used_size(ctx->buf, 0);
9664
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
305 if (ctx->encoding_buf->used != 0) {
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
306 ret = base64_decode(ctx->encoding_buf->data,
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
307 ctx->encoding_buf->used,
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
308 &pos, ctx->buf);
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
309 } else {
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
310 ret = base64_decode(input->data, input->size,
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
311 &pos, ctx->buf);
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
312 }
5510
93e9e081855a Fix for base64 handling
Timo Sirainen <tss@iki.fi>
parents: 5422
diff changeset
313 if (ret < 0) {
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
314 /* corrupted base64 data, don't bother with
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
315 the rest of it */
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
316 return FALSE;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
317 }
5510
93e9e081855a Fix for base64 handling
Timo Sirainen <tss@iki.fi>
parents: 5422
diff changeset
318 if (ret == 0) {
4885
Timo Sirainen <tss@iki.fi>
parents: 4708
diff changeset
319 /* end of base64 input */
9664
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
320 pos = input->size;
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
321 buffer_set_used_size(ctx->encoding_buf, 0);
4885
Timo Sirainen <tss@iki.fi>
parents: 4708
diff changeset
322 }
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
323 data = ctx->buf->data;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
324 size = ctx->buf->used;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
325 break;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
326 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
327
9664
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
328 if (ctx->encoding_buf->used != 0)
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
329 buffer_delete(ctx->encoding_buf, 0, pos);
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
330 else if (pos != input->size) {
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
331 buffer_append(ctx->encoding_buf,
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
332 input->data + pos, input->size - pos);
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
333 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
334
9322
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
335 if (ctx->binary_input) {
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
336 output->data = data;
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
337 output->size = size;
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
338 } else if (ctx->charset_utf8) {
6952
08e4d7efcd6a uni_utf8_get_valid_data() API changed.
Timo Sirainen <tss@iki.fi>
parents: 6951
diff changeset
339 buffer_set_used_size(ctx->buf2, 0);
9322
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
340 if ((ctx->flags & MESSAGE_DECODER_FLAG_DTCASE) != 0) {
6131
5f56b2eb32b3 Use uni_utf8_to_decomposed_titlecase() to have proper case-insensitive UTF-8
Timo Sirainen <tss@iki.fi>
parents: 6122
diff changeset
341 (void)uni_utf8_to_decomposed_titlecase(data, size,
6579
825621c540b8 Body decoding was broken.
Timo Sirainen <tss@iki.fi>
parents: 6429
diff changeset
342 ctx->buf2);
6113
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
343 output->data = ctx->buf2->data;
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
344 output->size = ctx->buf2->used;
6952
08e4d7efcd6a uni_utf8_get_valid_data() API changed.
Timo Sirainen <tss@iki.fi>
parents: 6951
diff changeset
345 } else if (uni_utf8_get_valid_data(data, size, ctx->buf2)) {
08e4d7efcd6a uni_utf8_get_valid_data() API changed.
Timo Sirainen <tss@iki.fi>
parents: 6951
diff changeset
346 output->data = data;
08e4d7efcd6a uni_utf8_get_valid_data() API changed.
Timo Sirainen <tss@iki.fi>
parents: 6951
diff changeset
347 output->size = size;
6113
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
348 } else {
6952
08e4d7efcd6a uni_utf8_get_valid_data() API changed.
Timo Sirainen <tss@iki.fi>
parents: 6951
diff changeset
349 output->data = ctx->buf2->data;
08e4d7efcd6a uni_utf8_get_valid_data() API changed.
Timo Sirainen <tss@iki.fi>
parents: 6951
diff changeset
350 output->size = ctx->buf2->used;
6113
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
351 }
5519
7b6511e67476 Some data still wasn't uppercased
Timo Sirainen <tss@iki.fi>
parents: 5511
diff changeset
352 } else if (ctx->charset_trans == NULL) {
6915
671c2eb25f3d Remove illegal UTF-8 sequences from output.
Timo Sirainen <tss@iki.fi>
parents: 6910
diff changeset
353 /* unknown charset */
6952
08e4d7efcd6a uni_utf8_get_valid_data() API changed.
Timo Sirainen <tss@iki.fi>
parents: 6951
diff changeset
354 buffer_set_used_size(ctx->buf2, 0);
08e4d7efcd6a uni_utf8_get_valid_data() API changed.
Timo Sirainen <tss@iki.fi>
parents: 6951
diff changeset
355 if (uni_utf8_get_valid_data(data, size, ctx->buf2)) {
08e4d7efcd6a uni_utf8_get_valid_data() API changed.
Timo Sirainen <tss@iki.fi>
parents: 6951
diff changeset
356 output->data = data;
08e4d7efcd6a uni_utf8_get_valid_data() API changed.
Timo Sirainen <tss@iki.fi>
parents: 6951
diff changeset
357 output->size = size;
08e4d7efcd6a uni_utf8_get_valid_data() API changed.
Timo Sirainen <tss@iki.fi>
parents: 6951
diff changeset
358 } else {
08e4d7efcd6a uni_utf8_get_valid_data() API changed.
Timo Sirainen <tss@iki.fi>
parents: 6951
diff changeset
359 output->data = ctx->buf2->data;
08e4d7efcd6a uni_utf8_get_valid_data() API changed.
Timo Sirainen <tss@iki.fi>
parents: 6951
diff changeset
360 output->size = ctx->buf2->used;
08e4d7efcd6a uni_utf8_get_valid_data() API changed.
Timo Sirainen <tss@iki.fi>
parents: 6951
diff changeset
361 }
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
362 } else {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
363 buffer_set_used_size(ctx->buf2, 0);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
364 if (ctx->translation_size != 0)
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
365 translation_buf_decode(ctx, &data, &size);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
366
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
367 pos = size;
6122
d86581f4a0c6 charset_to_utf8() isn't used anymore, so renamed charset_to_utf8_full() to it.
Timo Sirainen <tss@iki.fi>
parents: 6118
diff changeset
368 (void)charset_to_utf8(ctx->charset_trans,
d86581f4a0c6 charset_to_utf8() isn't used anymore, so renamed charset_to_utf8_full() to it.
Timo Sirainen <tss@iki.fi>
parents: 6118
diff changeset
369 data, &pos, ctx->buf2);
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
370 if (pos != size) {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
371 ctx->translation_size = size - pos;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
372 i_assert(ctx->translation_size <=
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
373 sizeof(ctx->translation_buf));
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
374 memcpy(ctx->translation_buf, data + pos,
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
375 ctx->translation_size);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
376 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
377 output->data = ctx->buf2->data;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
378 output->size = ctx->buf2->used;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
379 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
380
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
381 output->hdr = NULL;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
382 return TRUE;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
383 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
384
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
385 bool message_decoder_decode_next_block(struct message_decoder_context *ctx,
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
386 struct message_block *input,
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
387 struct message_block *output)
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
388 {
5410
73032525bca0 Don't try to access output variable's contents. They are undefined at least
Timo Sirainen <tss@iki.fi>
parents: 4903
diff changeset
389 if (input->part != ctx->prev_part) {
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
390 /* MIME part changed. */
6916
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
391 message_decoder_decode_reset(ctx);
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
392 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
393
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
394 output->part = input->part;
5410
73032525bca0 Don't try to access output variable's contents. They are undefined at least
Timo Sirainen <tss@iki.fi>
parents: 4903
diff changeset
395 ctx->prev_part = input->part;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
396
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
397 if (input->hdr != NULL)
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
398 return message_decode_header(ctx, input->hdr, output);
6916
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
399 else if (input->size != 0)
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
400 return message_decode_body(ctx, input, output);
6916
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
401 else {
6942
3af80c378959 Crashfix
Timo Sirainen <tss@iki.fi>
parents: 6940
diff changeset
402 output->hdr = NULL;
3af80c378959 Crashfix
Timo Sirainen <tss@iki.fi>
parents: 6940
diff changeset
403 output->size = 0;
9322
ce612bdafc84 message-decoder now supports optionally returning non-text bodyparts as-is.
Timo Sirainen <tss@iki.fi>
parents: 8590
diff changeset
404 message_decode_body_init_charset(ctx, input->part);
6916
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
405 return TRUE;
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
406 }
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
407 }
6916
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
408
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
409 void message_decoder_decode_reset(struct message_decoder_context *ctx)
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
410 {
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
411 i_free_and_null(ctx->content_charset);
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
412 ctx->content_type = CONTENT_TYPE_BINARY;
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
413 ctx->charset_utf8 = TRUE;
9664
4c255c7b6245 message-decoder: Fixed assert-crashes caused by the previous quoted-printable change.
Timo Sirainen <tss@iki.fi>
parents: 9360
diff changeset
414 buffer_set_used_size(ctx->encoding_buf, 0);
6916
0b8a78914db7 Message decoding wasn't reset properly between messages when searching.
Timo Sirainen <tss@iki.fi>
parents: 6915
diff changeset
415 }