annotate src/lib-mail/message-decoder.c @ 6117:9214044ce1f1 HEAD

Removed message-content-parser. Instead added rfc822_parse_content_type() and rfc822_parse_content_param() to help parse such headers.
author Timo Sirainen <tss@iki.fi>
date Fri, 20 Jul 2007 09:42:25 +0300
parents 325667778ae3
children 841b052e269e
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1 /* Copyright (C) 2006 Timo Sirainen */
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"
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
7 #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
8 #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
9 #include "rfc822-parser.h"
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
10 #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
11 #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
12 #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
13
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
14 enum content_type {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
15 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
16 CONTENT_TYPE_BINARY,
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
17 CONTENT_TYPE_QP,
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
18 CONTENT_TYPE_BASE64
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
19 };
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
20
4885
Timo Sirainen <tss@iki.fi>
parents: 4708
diff changeset
21 /* base64 takes max 4 bytes per character, q-p takes max 3. */
Timo Sirainen <tss@iki.fi>
parents: 4708
diff changeset
22 #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
23
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
24 /* 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
25 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
26 #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
27
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
28 struct message_decoder_context {
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
29 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
30
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
31 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
32 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
33
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
34 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
35 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
36 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
37
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
38 char encoding_buf[MAX_ENCODING_BUF_SIZE];
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
39 unsigned int encoding_size;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
40
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
41 char *content_charset;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
42 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
43
6113
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
44 unsigned int ucase:1;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
45 unsigned int charset_utf8:1;
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
6113
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
48 struct message_decoder_context *message_decoder_init(bool ucase)
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 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
51
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
52 ctx = i_new(struct message_decoder_context, 1);
6113
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
53 ctx->ucase = ucase;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
54 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
55 ctx->buf2 = 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
56 return ctx;
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
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
59 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
60 {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
61 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
62
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
63 *_ctx = NULL;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
64
4888
c3e15280e4cb Memory leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4885
diff changeset
65 if (ctx->charset_trans != NULL)
c3e15280e4cb Memory leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4885
diff changeset
66 charset_to_utf8_end(&ctx->charset_trans);
c3e15280e4cb Memory leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4885
diff changeset
67
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
68 buffer_free(ctx->buf);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
69 buffer_free(ctx->buf2);
4888
c3e15280e4cb Memory leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4885
diff changeset
70 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
71 i_free(ctx);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
72 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
73
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
74 static bool
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
75 message_decode_header_callback(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
76 const char *charset, void *context)
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
77 {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
78 struct message_decoder_context *ctx = context;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
79 struct charset_translation *t;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
80 bool unknown_charset;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
81
5511
518a77f3bf49 Renamed message_decoder_init() to message_decoder_init_ucase() and changed
Timo Sirainen <tss@iki.fi>
parents: 5510
diff changeset
82 if (charset == NULL || charset_is_utf8(charset)) {
5522
5dee807e53cf Header parser has now flags parameter to tell it how to handle linefeeds.
Timo Sirainen <tss@iki.fi>
parents: 5519
diff changeset
83 /* ASCII / UTF-8 */
6113
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
84 if (ctx->ucase) {
6114
325667778ae3 _charset_utf8_ucase() -> charset_utf8_ucase_write(),
Timo Sirainen <tss@iki.fi>
parents: 6113
diff changeset
85 charset_utf8_ucase_write(ctx->buf, ctx->buf->used,
325667778ae3 _charset_utf8_ucase() -> charset_utf8_ucase_write(),
Timo Sirainen <tss@iki.fi>
parents: 6113
diff changeset
86 data, size);
6113
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
87 } else {
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
88 buffer_append(ctx->buf, data, size);
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
89 }
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
90 return TRUE;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
91 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
92
6113
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
93 t = charset_to_utf8_begin(charset, ctx->ucase, &unknown_charset);
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
94 if (unknown_charset) {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
95 /* let's just ignore this part */
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
96 return TRUE;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
97 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
98
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
99 /* ignore any errors */
6112
e5451501ff2f charset_to_utf8_begin() now takes bool ucase parameter. Changed
Timo Sirainen <tss@iki.fi>
parents: 5524
diff changeset
100 (void)charset_to_utf8_full(t, data, &size, ctx->buf);
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
101 charset_to_utf8_end(&t);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
102 return TRUE;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
103 }
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 static void
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
106 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
107 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
108 {
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
109 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
110 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
111
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
112 t_push();
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
113 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
114 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
115
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
116 (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
117 (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
118
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
119 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
120 case 4:
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
121 if (memcasecmp(str_data(value), "7bit", 4) == 0 ||
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
122 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
123 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
124 break;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
125 case 6:
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
126 if (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
127 ctx->content_type = CONTENT_TYPE_BASE64;
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
128 else if (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
129 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
130 break;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
131 case 16:
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
132 if (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
133 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
134 break;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
135 }
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
136 t_pop();
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
137 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
138
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
139 static void
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
140 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
141 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
142 {
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
143 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
144 const char *key, *value;
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
145 string_t *str;
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
146
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
147 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
148 return;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
149
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
150 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
151 (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
152 t_push();
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
153 str = t_str_new(64);
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
154 if (rfc822_parse_content_type(&parser, str) <= 0) {
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
155 t_pop();
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
156 return;
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
157 }
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
158
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
159 while (rfc822_parse_content_param(&parser, &key, &value) > 0) {
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
160 if (strcasecmp(key, "charset") == 0) {
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
161 ctx->content_charset = i_strdup(value);
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
162 ctx->charset_utf8 = charset_is_utf8(value);
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
163 break;
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
164 }
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
165 }
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
166 t_pop();
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
167 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
168
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
169 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
170 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
171 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
172 {
5511
518a77f3bf49 Renamed message_decoder_init() to message_decoder_init_ucase() and changed
Timo Sirainen <tss@iki.fi>
parents: 5510
diff changeset
173 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
174
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
175 if (hdr->continues) {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
176 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
177 return FALSE;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
178 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
179
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
180 if (hdr->name_len == 12 &&
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
181 strcasecmp(hdr->name, "Content-Type") == 0)
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
182 parse_content_type(ctx, hdr);
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
183 if (hdr->name_len == 25 &&
6117
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
184 strcasecmp(hdr->name, "Content-Transfer-Encoding") == 0)
9214044ce1f1 Removed message-content-parser. Instead added rfc822_parse_content_type()
Timo Sirainen <tss@iki.fi>
parents: 6114
diff changeset
185 parse_content_transfer_encoding(ctx, hdr);
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
186
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
187 buffer_set_used_size(ctx->buf, 0);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
188 message_header_decode(hdr->full_value, hdr->full_value_len,
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
189 message_decode_header_callback, ctx);
5511
518a77f3bf49 Renamed message_decoder_init() to message_decoder_init_ucase() and changed
Timo Sirainen <tss@iki.fi>
parents: 5510
diff changeset
190 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
191
6113
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
192 if (ctx->ucase) {
6114
325667778ae3 _charset_utf8_ucase() -> charset_utf8_ucase_write(),
Timo Sirainen <tss@iki.fi>
parents: 6113
diff changeset
193 charset_utf8_ucase_write(ctx->buf, ctx->buf->used,
325667778ae3 _charset_utf8_ucase() -> charset_utf8_ucase_write(),
Timo Sirainen <tss@iki.fi>
parents: 6113
diff changeset
194 (const unsigned char *)hdr->name,
325667778ae3 _charset_utf8_ucase() -> charset_utf8_ucase_write(),
Timo Sirainen <tss@iki.fi>
parents: 6113
diff changeset
195 hdr->name_len);
6113
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
196 buffer_append_c(ctx->buf, '\0');
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
197 }
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
198
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
199 ctx->hdr = *hdr;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
200 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
201 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
202 ctx->hdr.value_len = 0;
6113
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
203 if (ctx->ucase) {
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
204 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
205 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
206 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
207 }
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
208
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
209 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
210 return TRUE;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
211 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
212
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
213 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
214 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
215 {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
216 unsigned char trans_buf[MAX_TRANSLATION_BUF_SIZE+1];
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
217 size_t pos, skip;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
218
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
219 /* @UNSAFE */
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
220 memcpy(trans_buf, ctx->translation_buf, ctx->translation_size);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
221 skip = sizeof(trans_buf) - ctx->translation_size;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
222 if (skip > *size)
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
223 skip = *size;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
224 memcpy(trans_buf + ctx->translation_size, data, skip);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
225
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
226 pos = *size;
6112
e5451501ff2f charset_to_utf8_begin() now takes bool ucase parameter. Changed
Timo Sirainen <tss@iki.fi>
parents: 5524
diff changeset
227 (void)charset_to_utf8_full(ctx->charset_trans,
e5451501ff2f charset_to_utf8_begin() now takes bool ucase parameter. Changed
Timo Sirainen <tss@iki.fi>
parents: 5524
diff changeset
228 *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
229
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
230 i_assert(pos > ctx->translation_size);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
231 skip = (ctx->translation_size + skip) - pos;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
232
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
233 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
234 *data += skip;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
235 *size -= skip;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
236
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
237 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
238 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
239
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
240 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
241 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
242 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
243 {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
244 unsigned char new_buf[MAX_ENCODING_BUF_SIZE+1];
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
245 const unsigned char *data = NULL;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
246 size_t pos, size = 0, skip = 0;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
247 bool unknown_charset;
5510
93e9e081855a Fix for base64 handling
Timo Sirainen <tss@iki.fi>
parents: 5422
diff changeset
248 int ret;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
249
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
250 if (ctx->charset_trans == NULL && !ctx->charset_utf8) {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
251 ctx->charset_trans =
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
252 charset_to_utf8_begin(ctx->content_charset != NULL ?
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
253 ctx->content_charset : "UTF-8",
6113
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
254 ctx->ucase,
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
255 &unknown_charset);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
256 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
257
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
258 if (ctx->encoding_size != 0) {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
259 /* @UNSAFE */
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
260 memcpy(new_buf, ctx->encoding_buf, ctx->encoding_size);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
261 skip = sizeof(new_buf) - ctx->encoding_size;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
262 if (skip > input->size)
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
263 skip = input->size;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
264 memcpy(new_buf + ctx->encoding_size, input->data, skip);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
265 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
266
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
267 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
268 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
269 /* 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
270 return FALSE;
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 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
273 data = input->data;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
274 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
275 break;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
276 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
277 buffer_set_used_size(ctx->buf, 0);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
278 if (ctx->encoding_size != 0) {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
279 quoted_printable_decode(new_buf,
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
280 ctx->encoding_size + skip,
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
281 &pos, ctx->buf);
5422
Timo Sirainen <tss@iki.fi>
parents: 5410
diff changeset
282 i_assert(pos >= ctx->encoding_size);
4885
Timo Sirainen <tss@iki.fi>
parents: 4708
diff changeset
283 skip = pos - ctx->encoding_size;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
284 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
285
4885
Timo Sirainen <tss@iki.fi>
parents: 4708
diff changeset
286 quoted_printable_decode(input->data + skip, input->size - skip,
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
287 &pos, ctx->buf);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
288 pos += skip;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
289 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
290 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
291 break;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
292 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
293 buffer_set_used_size(ctx->buf, 0);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
294 if (ctx->encoding_size != 0) {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
295 if (base64_decode(new_buf, ctx->encoding_size + skip,
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
296 &pos, ctx->buf) < 0) {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
297 /* 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
298 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
299 return FALSE;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
300 }
5422
Timo Sirainen <tss@iki.fi>
parents: 5410
diff changeset
301 i_assert(pos >= ctx->encoding_size);
4885
Timo Sirainen <tss@iki.fi>
parents: 4708
diff changeset
302 skip = pos - ctx->encoding_size;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
303 }
5510
93e9e081855a Fix for base64 handling
Timo Sirainen <tss@iki.fi>
parents: 5422
diff changeset
304 ret = base64_decode(input->data + skip, input->size - skip,
93e9e081855a Fix for base64 handling
Timo Sirainen <tss@iki.fi>
parents: 5422
diff changeset
305 &pos, ctx->buf);
93e9e081855a Fix for base64 handling
Timo Sirainen <tss@iki.fi>
parents: 5422
diff changeset
306 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
307 /* 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
308 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
309 return FALSE;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
310 }
5510
93e9e081855a Fix for base64 handling
Timo Sirainen <tss@iki.fi>
parents: 5422
diff changeset
311 if (ret == 0) {
4885
Timo Sirainen <tss@iki.fi>
parents: 4708
diff changeset
312 /* end of base64 input */
Timo Sirainen <tss@iki.fi>
parents: 4708
diff changeset
313 pos = input->size - skip;
Timo Sirainen <tss@iki.fi>
parents: 4708
diff changeset
314 }
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
315 pos += skip;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
316 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
317 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
318 break;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
319 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
320
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
321 if (pos != input->size) {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
322 /* @UNSAFE */
4885
Timo Sirainen <tss@iki.fi>
parents: 4708
diff changeset
323 i_assert(pos < input->size);
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
324 ctx->encoding_size = input->size - pos;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
325 i_assert(ctx->encoding_size <= sizeof(ctx->encoding_buf));
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
326 memcpy(ctx->encoding_buf, input->data + pos,
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
327 ctx->encoding_size);
5422
Timo Sirainen <tss@iki.fi>
parents: 5410
diff changeset
328 } else {
Timo Sirainen <tss@iki.fi>
parents: 5410
diff changeset
329 ctx->encoding_size = 0;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
330 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
331
5519
7b6511e67476 Some data still wasn't uppercased
Timo Sirainen <tss@iki.fi>
parents: 5511
diff changeset
332 if (ctx->charset_utf8) {
6113
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
333 if (ctx->ucase) {
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
334 buffer_set_used_size(ctx->buf2, 0);
6114
325667778ae3 _charset_utf8_ucase() -> charset_utf8_ucase_write(),
Timo Sirainen <tss@iki.fi>
parents: 6113
diff changeset
335 charset_utf8_ucase_write(ctx->buf2, 0, data, size);
6113
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
336 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
337 output->size = ctx->buf2->used;
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
338 } else {
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
339 output->data = data;
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
340 output->size = size;
adbfce5bf7d9 Changed message_decoder_init_ucase() to message_decoder_init(bool ucase)
Timo Sirainen <tss@iki.fi>
parents: 6112
diff changeset
341 }
5519
7b6511e67476 Some data still wasn't uppercased
Timo Sirainen <tss@iki.fi>
parents: 5511
diff changeset
342 } else if (ctx->charset_trans == NULL) {
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
343 output->data = data;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
344 output->size = size;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
345 } else {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
346 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
347 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
348 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
349
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
350 pos = size;
6112
e5451501ff2f charset_to_utf8_begin() now takes bool ucase parameter. Changed
Timo Sirainen <tss@iki.fi>
parents: 5524
diff changeset
351 (void)charset_to_utf8_full(ctx->charset_trans,
e5451501ff2f charset_to_utf8_begin() now takes bool ucase parameter. Changed
Timo Sirainen <tss@iki.fi>
parents: 5524
diff changeset
352 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
353 if (pos != size) {
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
354 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
355 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
356 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
357 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
358 ctx->translation_size);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
359 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
360 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
361 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
362 }
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
363
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
364 output->hdr = NULL;
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
365 return TRUE;
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
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
368 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
369 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
370 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
371 {
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
372 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
373 /* MIME part changed. */
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
374 i_free_and_null(ctx->content_charset);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
375 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
376 ctx->charset_utf8 = TRUE;
4885
Timo Sirainen <tss@iki.fi>
parents: 4708
diff changeset
377 ctx->encoding_size = 0;
4608
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
378 }
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 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
381 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
382
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
383 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
384 return message_decode_header(ctx, input->hdr, output);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
385 else
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
386 return message_decode_body(ctx, input, output);
61e0fe257a83 Added message-decoder to decode message's contents into readable UTF-8.
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
387 }