Mercurial > dovecot > original-hg > dovecot-1.2
annotate src/lib-mail/message-parser.c @ 1322:97f8c00b8d4c HEAD
Better handling for multiline headers. Before we skipped headers larger than
input buffer size (8k with read (default), 256k with mmap). The skipping was
also a bit buggy.
Now we parse the lines one at a time. There's also a way to read the header
fully into memory before parsing it, if really needed.
author | Timo Sirainen <tss@iki.fi> |
---|---|
date | Wed, 26 Mar 2003 19:29:01 +0200 |
parents | 39b899338c99 |
children | baf63e166aeb |
rev | line source |
---|---|
0 | 1 /* Copyright (C) 2002 Timo Sirainen */ |
2 | |
3 #include "lib.h" | |
1322
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
4 #include "buffer.h" |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
5 #include "istream.h" |
1322
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
6 #include "str.h" |
896
21ffcce83c70
Rewrote rfc822-tokenize.c to work one token at a time so it won't uselessly
Timo Sirainen <tss@iki.fi>
parents:
874
diff
changeset
|
7 #include "strescape.h" |
0 | 8 #include "message-content-parser.h" |
9 #include "message-parser.h" | |
9
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
10 #include "message-size.h" |
0 | 11 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
12 struct message_boundary { |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
13 struct message_boundary *next; |
0 | 14 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
15 struct message_part *part; |
0 | 16 const char *boundary; |
184 | 17 size_t len; |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
18 }; |
0 | 19 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
20 struct parser_context { |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
21 pool_t pool; |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
22 struct message_part *part; |
0 | 23 |
24 char *last_boundary; | |
25 char *last_content_type; | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
26 struct message_boundary *boundaries; |
0 | 27 |
1038
60646878858e
Function typedefs now define them as functions, not function pointers.
Timo Sirainen <tss@iki.fi>
parents:
953
diff
changeset
|
28 message_header_callback_t *callback; |
10
82b7de533f98
s/user_data/context/ and some s/Data/Context/
Timo Sirainen <tss@iki.fi>
parents:
9
diff
changeset
|
29 void *context; |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
30 }; |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
31 |
1322
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
32 struct message_header_parser_ctx { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
33 struct message_header_line line; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
34 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
35 struct istream *input; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
36 struct message_size *hdr_size; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
37 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
38 string_t *name; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
39 buffer_t *value_buf; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
40 size_t skip; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
41 }; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
42 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
43 static struct message_part * |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
44 message_parse_part(struct istream *input, |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
45 struct parser_context *parser_ctx); |
0 | 46 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
47 static struct message_part * |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
48 message_parse_body(struct istream *input, struct message_boundary *boundaries, |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
49 struct message_size *body_size); |
0 | 50 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
51 static struct message_part * |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
52 message_skip_boundary(struct istream *input, |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
53 struct message_boundary *boundaries, |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
54 struct message_size *boundary_size); |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
55 |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
56 static void message_size_add_part(struct message_size *dest, |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
57 struct message_part *part) |
0 | 58 { |
59 dest->physical_size += | |
60 part->header_size.physical_size + | |
61 part->body_size.physical_size; | |
62 dest->virtual_size += | |
63 part->header_size.virtual_size + | |
64 part->body_size.virtual_size; | |
65 dest->lines += part->header_size.lines + part->body_size.lines; | |
66 } | |
67 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
68 static struct message_part * |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
69 message_part_append(pool_t pool, struct message_part *parent) |
0 | 70 { |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
71 struct message_part *part, **list; |
0 | 72 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
73 part = p_new(pool, struct message_part, 1); |
0 | 74 part->parent = parent; |
75 | |
212
4327b1266604
message/rfc822 mime parts weren't parsed correctly
Timo Sirainen <tss@iki.fi>
parents:
184
diff
changeset
|
76 /* set child position */ |
4327b1266604
message/rfc822 mime parts weren't parsed correctly
Timo Sirainen <tss@iki.fi>
parents:
184
diff
changeset
|
77 part->physical_pos = |
4327b1266604
message/rfc822 mime parts weren't parsed correctly
Timo Sirainen <tss@iki.fi>
parents:
184
diff
changeset
|
78 parent->physical_pos + |
4327b1266604
message/rfc822 mime parts weren't parsed correctly
Timo Sirainen <tss@iki.fi>
parents:
184
diff
changeset
|
79 parent->body_size.physical_size + |
4327b1266604
message/rfc822 mime parts weren't parsed correctly
Timo Sirainen <tss@iki.fi>
parents:
184
diff
changeset
|
80 parent->header_size.physical_size; |
4327b1266604
message/rfc822 mime parts weren't parsed correctly
Timo Sirainen <tss@iki.fi>
parents:
184
diff
changeset
|
81 |
0 | 82 list = &part->parent->children; |
83 while (*list != NULL) | |
84 list = &(*list)->next; | |
85 | |
86 *list = part; | |
87 return part; | |
88 } | |
89 | |
898
0d5be52d7131
Use unsigned char* when accessing non-NUL terminating strings. Compiler
Timo Sirainen <tss@iki.fi>
parents:
896
diff
changeset
|
90 static void parse_content_type(const unsigned char *value, size_t value_len, |
10
82b7de533f98
s/user_data/context/ and some s/Data/Context/
Timo Sirainen <tss@iki.fi>
parents:
9
diff
changeset
|
91 void *context) |
0 | 92 { |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
93 struct parser_context *parser_ctx = context; |
0 | 94 const char *str; |
95 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
96 if (parser_ctx->last_content_type != NULL || value_len == 0) |
0 | 97 return; |
98 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
99 str = parser_ctx->last_content_type = |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
100 p_strndup(parser_ctx->pool, value, value_len); |
0 | 101 |
102 if (strcasecmp(str, "message/rfc822") == 0) | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
103 parser_ctx->part->flags |= MESSAGE_PART_FLAG_MESSAGE_RFC822; |
0 | 104 else if (strncasecmp(str, "text/", 5) == 0) |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
105 parser_ctx->part->flags |= MESSAGE_PART_FLAG_TEXT; |
0 | 106 else if (strncasecmp(str, "multipart/", 10) == 0) { |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
107 parser_ctx->part->flags |= MESSAGE_PART_FLAG_MULTIPART; |
0 | 108 |
106
5fe3e04ca8d9
Added support for caching of MessagePart data. This is useful for fetching
Timo Sirainen <tss@iki.fi>
parents:
105
diff
changeset
|
109 if (strcasecmp(str+10, "digest") == 0) { |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
110 parser_ctx->part->flags |= |
106
5fe3e04ca8d9
Added support for caching of MessagePart data. This is useful for fetching
Timo Sirainen <tss@iki.fi>
parents:
105
diff
changeset
|
111 MESSAGE_PART_FLAG_MULTIPART_DIGEST; |
5fe3e04ca8d9
Added support for caching of MessagePart data. This is useful for fetching
Timo Sirainen <tss@iki.fi>
parents:
105
diff
changeset
|
112 } |
0 | 113 } |
114 } | |
115 | |
898
0d5be52d7131
Use unsigned char* when accessing non-NUL terminating strings. Compiler
Timo Sirainen <tss@iki.fi>
parents:
896
diff
changeset
|
116 static void |
0d5be52d7131
Use unsigned char* when accessing non-NUL terminating strings. Compiler
Timo Sirainen <tss@iki.fi>
parents:
896
diff
changeset
|
117 parse_content_type_param(const unsigned char *name, size_t name_len, |
0d5be52d7131
Use unsigned char* when accessing non-NUL terminating strings. Compiler
Timo Sirainen <tss@iki.fi>
parents:
896
diff
changeset
|
118 const unsigned char *value, size_t value_len, |
0d5be52d7131
Use unsigned char* when accessing non-NUL terminating strings. Compiler
Timo Sirainen <tss@iki.fi>
parents:
896
diff
changeset
|
119 int value_quoted, void *context) |
0 | 120 { |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
121 struct parser_context *parser_ctx = context; |
0 | 122 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
123 if ((parser_ctx->part->flags & MESSAGE_PART_FLAG_MULTIPART) == 0 || |
898
0d5be52d7131
Use unsigned char* when accessing non-NUL terminating strings. Compiler
Timo Sirainen <tss@iki.fi>
parents:
896
diff
changeset
|
124 name_len != 8 || memcasecmp(name, "boundary", 8) != 0) |
0 | 125 return; |
126 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
127 if (parser_ctx->last_boundary == NULL) { |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
128 parser_ctx->last_boundary = |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
129 p_strndup(parser_ctx->pool, value, value_len); |
896
21ffcce83c70
Rewrote rfc822-tokenize.c to work one token at a time so it won't uselessly
Timo Sirainen <tss@iki.fi>
parents:
874
diff
changeset
|
130 if (value_quoted) |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
131 str_unescape(parser_ctx->last_boundary); |
0 | 132 } |
133 } | |
134 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
135 static struct message_part * |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
136 message_parse_multipart(struct istream *input, |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
137 struct parser_context *parser_ctx) |
0 | 138 { |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
139 struct message_part *parent_part, *next_part, *part; |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
140 struct message_boundary *b; |
0 | 141 |
142 /* multipart message. add new boundary */ | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
143 b = t_new(struct message_boundary, 1); |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
144 b->part = parser_ctx->part; |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
145 b->boundary = parser_ctx->last_boundary; |
0 | 146 b->len = strlen(b->boundary); |
147 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
148 b->next = parser_ctx->boundaries; |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
149 parser_ctx->boundaries = b; |
0 | 150 |
151 /* reset fields */ | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
152 parser_ctx->last_boundary = NULL; |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
153 parser_ctx->last_content_type = NULL; |
0 | 154 |
155 /* skip the data before the first boundary */ | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
156 parent_part = parser_ctx->part; |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
157 next_part = message_skip_boundary(input, parser_ctx->boundaries, |
0 | 158 &parent_part->body_size); |
159 | |
160 /* now, parse the parts */ | |
161 while (next_part == parent_part) { | |
162 /* new child */ | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
163 part = message_part_append(parser_ctx->pool, parent_part); |
0 | 164 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
165 parser_ctx->part = part; |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
166 next_part = message_parse_part(input, parser_ctx); |
0 | 167 |
168 /* update our size */ | |
169 message_size_add_part(&parent_part->body_size, part); | |
170 | |
171 if (next_part != parent_part) | |
172 break; | |
173 | |
174 /* skip the boundary */ | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
175 next_part = message_skip_boundary(input, parser_ctx->boundaries, |
0 | 176 &parent_part->body_size); |
177 } | |
178 | |
179 /* remove boundary */ | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
180 i_assert(parser_ctx->boundaries == b); |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
181 parser_ctx->boundaries = b->next; |
0 | 182 return next_part; |
183 } | |
184 | |
874 | 185 #define MUTEX_FLAGS \ |
186 (MESSAGE_PART_FLAG_MESSAGE_RFC822 | MESSAGE_PART_FLAG_MULTIPART) | |
187 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
188 static struct message_part * |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
189 message_parse_part(struct istream *input, struct parser_context *parser_ctx) |
0 | 190 { |
1322
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
191 struct message_header_parser_ctx *hdr_ctx; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
192 struct message_header_line *hdr; |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
193 struct message_part *next_part, *part; |
50
d493b9cc265e
Introduced uoff_t which is the unsigned-equilevant of off_t. This was needed
Timo Sirainen <tss@iki.fi>
parents:
10
diff
changeset
|
194 uoff_t hdr_size; |
0 | 195 |
1322
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
196 hdr_ctx = message_parse_header_init(input, |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
197 &parser_ctx->part->header_size); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
198 while ((hdr = message_parse_header_next(hdr_ctx)) != NULL) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
199 /* call the user-defined header parser */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
200 if (parser_ctx->callback != NULL) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
201 parser_ctx->callback(parser_ctx->part, hdr, |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
202 parser_ctx->context); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
203 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
204 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
205 if (strcasecmp(hdr->name, "Content-Type") == 0) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
206 if (hdr->continues) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
207 hdr->use_full_value = TRUE; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
208 continue; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
209 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
210 /* we need to know the boundary */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
211 message_content_parse_header(hdr->full_value, |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
212 hdr->full_value_len, |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
213 parse_content_type, |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
214 parse_content_type_param, |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
215 parser_ctx); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
216 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
217 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
218 if (parser_ctx->callback != NULL) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
219 parser_ctx->callback(parser_ctx->part, NULL, |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
220 parser_ctx->context); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
221 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
222 message_parse_header_deinit(hdr_ctx); |
0 | 223 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
224 i_assert((parser_ctx->part->flags & MUTEX_FLAGS) != MUTEX_FLAGS); |
874 | 225 |
0 | 226 /* update message position/size */ |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
227 hdr_size = parser_ctx->part->header_size.physical_size; |
0 | 228 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
229 if (parser_ctx->last_boundary != NULL) |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
230 return message_parse_multipart(input, parser_ctx); |
0 | 231 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
232 if (parser_ctx->last_content_type == NULL) { |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
233 if (parser_ctx->part->parent != NULL && |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
234 (parser_ctx->part->parent->flags & |
106
5fe3e04ca8d9
Added support for caching of MessagePart data. This is useful for fetching
Timo Sirainen <tss@iki.fi>
parents:
105
diff
changeset
|
235 MESSAGE_PART_FLAG_MULTIPART_DIGEST)) { |
0 | 236 /* when there's no content-type specified and we're |
237 below multipart/digest, the assume message/rfc822 | |
238 content-type */ | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
239 parser_ctx->part->flags |= |
106
5fe3e04ca8d9
Added support for caching of MessagePart data. This is useful for fetching
Timo Sirainen <tss@iki.fi>
parents:
105
diff
changeset
|
240 MESSAGE_PART_FLAG_MESSAGE_RFC822; |
0 | 241 } else { |
242 /* otherwise we default to text/plain */ | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
243 parser_ctx->part->flags |= MESSAGE_PART_FLAG_TEXT; |
0 | 244 } |
245 } | |
246 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
247 parser_ctx->last_boundary = NULL; |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
248 parser_ctx->last_content_type = NULL; |
0 | 249 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
250 if (parser_ctx->part->flags & MESSAGE_PART_FLAG_MESSAGE_RFC822) { |
0 | 251 /* message/rfc822 part - the message body begins with |
252 headers again, this works pretty much the same as | |
253 a single multipart/mixed item */ | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
254 part = message_part_append(parser_ctx->pool, parser_ctx->part); |
0 | 255 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
256 parser_ctx->part = part; |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
257 next_part = message_parse_part(input, parser_ctx); |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
258 parser_ctx->part = part->parent; |
0 | 259 |
260 /* our body size is the size of header+body in message/rfc822 */ | |
261 message_size_add_part(&part->parent->body_size, part); | |
262 } else { | |
263 /* normal message, read until the next boundary */ | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
264 part = parser_ctx->part; |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
265 next_part = message_parse_body(input, parser_ctx->boundaries, |
0 | 266 &part->body_size); |
267 } | |
268 | |
269 return next_part; | |
270 } | |
271 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
272 static void message_skip_line(struct istream *input, |
1253
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
273 struct message_size *msg_size, int skip_lf) |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
274 { |
410
1f0e7229ee58
Split IOBuffer into mmaped IBuffer, file IBuffer, memory data IBuffer and
Timo Sirainen <tss@iki.fi>
parents:
369
diff
changeset
|
275 const unsigned char *msg; |
184 | 276 size_t i, size, startpos; |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
277 |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
278 startpos = 0; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
279 |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
280 while (i_stream_read_data(input, &msg, &size, startpos) > 0) { |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
281 for (i = startpos; i < size; i++) { |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
282 if (msg[i] == '\n') { |
1253
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
283 if (!skip_lf) { |
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
284 if (i > 0 && msg[i-1] == '\r') |
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
285 i--; |
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
286 startpos = i; |
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
287 goto __break; |
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
288 } |
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
289 |
153
9660df1ca44f
message_parse_header() was buggy with big headers, and io_buffer_read_data()
Timo Sirainen <tss@iki.fi>
parents:
112
diff
changeset
|
290 if (msg_size != NULL) { |
9660df1ca44f
message_parse_header() was buggy with big headers, and io_buffer_read_data()
Timo Sirainen <tss@iki.fi>
parents:
112
diff
changeset
|
291 if (i == 0 || msg[i-1] != '\r') |
9660df1ca44f
message_parse_header() was buggy with big headers, and io_buffer_read_data()
Timo Sirainen <tss@iki.fi>
parents:
112
diff
changeset
|
292 msg_size->virtual_size++; |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
293 msg_size->lines++; |
153
9660df1ca44f
message_parse_header() was buggy with big headers, and io_buffer_read_data()
Timo Sirainen <tss@iki.fi>
parents:
112
diff
changeset
|
294 } |
1253
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
295 startpos = i+1; |
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
296 goto __break; |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
297 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
298 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
299 |
232
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
300 /* leave the last character, it may be \r */ |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
301 i_stream_skip(input, i - 1); |
232
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
302 startpos = 1; |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
303 |
232
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
304 if (msg_size != NULL) { |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
305 msg_size->physical_size += i - 1; |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
306 msg_size->virtual_size += i - 1; |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
307 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
308 } |
1253
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
309 __break: |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
310 i_stream_skip(input, startpos); |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
311 |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
312 if (msg_size != NULL) { |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
313 msg_size->physical_size += startpos; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
314 msg_size->virtual_size += startpos; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
315 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
316 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
317 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
318 static struct message_boundary * |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
319 boundary_find(struct message_boundary *boundaries, |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
320 const unsigned char *msg, size_t len) |
0 | 321 { |
322 while (boundaries != NULL) { | |
323 if (boundaries->len <= len && | |
898
0d5be52d7131
Use unsigned char* when accessing non-NUL terminating strings. Compiler
Timo Sirainen <tss@iki.fi>
parents:
896
diff
changeset
|
324 memcmp(boundaries->boundary, msg, boundaries->len) == 0) |
0 | 325 return boundaries; |
326 | |
327 boundaries = boundaries->next; | |
328 } | |
329 | |
330 return NULL; | |
331 } | |
332 | |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
333 /* read until next boundary is found. if skip_over = FALSE, stop at the |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
334 [\r]\n before the boundary, otherwise leave it right after the known |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
335 boundary so the ending "--" can be checked. */ |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
336 static struct message_boundary * |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
337 message_find_boundary(struct istream *input, |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
338 struct message_boundary *boundaries, |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
339 struct message_size *msg_size, int skip_over) |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
340 { |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
341 struct message_boundary *boundary; |
410
1f0e7229ee58
Split IOBuffer into mmaped IBuffer, file IBuffer, memory data IBuffer and
Timo Sirainen <tss@iki.fi>
parents:
369
diff
changeset
|
342 const unsigned char *msg; |
184 | 343 size_t i, size, startpos, line_start, missing_cr_count; |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
344 |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
345 boundary = NULL; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
346 missing_cr_count = startpos = line_start = 0; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
347 |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
348 while (i_stream_read_data(input, &msg, &size, startpos) > 0) { |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
349 for (i = startpos; i < size; i++) { |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
350 if (msg[i] != '\n') |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
351 continue; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
352 |
1061 | 353 if (i >= line_start+2 && msg[line_start] == '-' && |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
354 msg[line_start+1] == '-') { |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
355 /* possible boundary */ |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
356 boundary = boundary_find(boundaries, |
898
0d5be52d7131
Use unsigned char* when accessing non-NUL terminating strings. Compiler
Timo Sirainen <tss@iki.fi>
parents:
896
diff
changeset
|
357 msg + line_start + 2, |
0d5be52d7131
Use unsigned char* when accessing non-NUL terminating strings. Compiler
Timo Sirainen <tss@iki.fi>
parents:
896
diff
changeset
|
358 i - line_start - 2); |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
359 if (boundary != NULL) |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
360 break; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
361 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
362 |
9
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
363 if (i == 0 || msg[i-1] != '\r') { |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
364 /* missing CR */ |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
365 missing_cr_count++; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
366 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
367 |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
368 msg_size->lines++; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
369 line_start = i+1; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
370 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
371 |
232
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
372 if (boundary != NULL) |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
373 break; |
232
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
374 |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
375 if (i - line_start > 128 && |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
376 msg[line_start] == '-' && msg[line_start+1] == '-') { |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
377 /* long partial line, see if it's a boundary. |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
378 RFC-2046 says that the boundaries must be |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
379 70 chars without "--" or less. We allow |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
380 a bit larger.. */ |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
381 boundary = boundary_find(boundaries, |
898
0d5be52d7131
Use unsigned char* when accessing non-NUL terminating strings. Compiler
Timo Sirainen <tss@iki.fi>
parents:
896
diff
changeset
|
382 msg + line_start + 2, |
0d5be52d7131
Use unsigned char* when accessing non-NUL terminating strings. Compiler
Timo Sirainen <tss@iki.fi>
parents:
896
diff
changeset
|
383 i - line_start - 2); |
232
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
384 if (boundary != NULL) |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
385 break; |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
386 |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
387 /* nope, we can skip over the line, just |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
388 leave the last char since it may be \r */ |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
389 i--; |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
390 } else { |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
391 /* leave the last line to buffer, it may be |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
392 boundary */ |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
393 i = line_start; |
1253
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
394 if (i > 0) i--; /* leave the \r\n too */ |
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
395 if (i > 0) i--; |
232
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
396 line_start -= i; |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
397 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
398 |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
399 i_stream_skip(input, i); |
232
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
400 msg_size->physical_size += i; |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
401 msg_size->virtual_size += i; |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
402 |
232
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
403 startpos = size - i; |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
404 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
405 |
1253
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
406 if (boundary == NULL && line_start+2 <= size && |
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
407 msg[line_start] == '-' && msg[line_start+1] == '-') { |
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
408 /* possible boundary without line feed at end */ |
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
409 boundary = boundary_find(boundaries, |
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
410 msg + line_start + 2, |
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
411 size - line_start - 2); |
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
412 } |
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
413 |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
414 if (boundary != NULL) { |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
415 if (skip_over) { |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
416 /* leave the pointer right after the boundary */ |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
417 line_start += 2 + boundary->len; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
418 } else if (line_start > 0 && msg[line_start-1] == '\n') { |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
419 /* leave the \r\n before the boundary */ |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
420 line_start--; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
421 msg_size->lines--; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
422 |
9
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
423 if (line_start > 0 && msg[line_start-1] == '\r') |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
424 line_start--; |
9
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
425 else |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
426 missing_cr_count--; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
427 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
428 startpos = line_start; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
429 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
430 |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
431 i_stream_skip(input, startpos); |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
432 msg_size->physical_size += startpos; |
9
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
433 msg_size->virtual_size += startpos + missing_cr_count; |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
434 |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
435 i_assert(msg_size->virtual_size >= msg_size->physical_size); |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
436 |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
437 return boundary; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
438 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
439 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
440 static struct message_part * |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
441 message_parse_body(struct istream *input, struct message_boundary *boundaries, |
1253
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
442 struct message_size *msg_size) |
0 | 443 { |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
444 struct message_boundary *boundary; |
1253
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
445 struct message_size body_size; |
0 | 446 |
9
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
447 if (boundaries == NULL) { |
1253
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
448 message_get_body_size(input, &body_size, (uoff_t)-1, NULL); |
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
449 message_size_add(msg_size, &body_size); |
9
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
450 return NULL; |
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
451 } else { |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
452 boundary = message_find_boundary(input, boundaries, |
1253
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
453 msg_size, FALSE); |
9
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
454 return boundary == NULL ? NULL : boundary->part; |
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
455 } |
0 | 456 } |
457 | |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
458 /* skip data until next boundary is found. if it's end boundary, |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
459 skip the footer as well. */ |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
460 static struct message_part * |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
461 message_skip_boundary(struct istream *input, |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
462 struct message_boundary *boundaries, |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
463 struct message_size *boundary_size) |
0 | 464 { |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
465 struct message_boundary *boundary; |
410
1f0e7229ee58
Split IOBuffer into mmaped IBuffer, file IBuffer, memory data IBuffer and
Timo Sirainen <tss@iki.fi>
parents:
369
diff
changeset
|
466 const unsigned char *msg; |
184 | 467 size_t size; |
0 | 468 int end_boundary; |
469 | |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
470 boundary = message_find_boundary(input, boundaries, |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
471 boundary_size, TRUE); |
0 | 472 if (boundary == NULL) |
473 return NULL; | |
474 | |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
475 /* now, see if it's end boundary */ |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
476 end_boundary = FALSE; |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
477 if (i_stream_read_data(input, &msg, &size, 1) > 0) |
232
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
478 end_boundary = msg[0] == '-' && msg[1] == '-'; |
0 | 479 |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
480 /* skip the rest of the line */ |
1253
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
481 message_skip_line(input, boundary_size, !end_boundary); |
0 | 482 |
483 if (end_boundary) { | |
484 /* skip the footer */ | |
1253
39b899338c99
A few small fixes to MIME and mail address parsers.
Timo Sirainen <tss@iki.fi>
parents:
1061
diff
changeset
|
485 return message_parse_body(input, boundary->next, boundary_size); |
0 | 486 } |
487 | |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
488 return boundary == NULL ? NULL : boundary->part; |
0 | 489 } |
1322
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
490 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
491 struct message_part *message_parse(pool_t pool, struct istream *input, |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
492 message_header_callback_t *callback, |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
493 void *context) |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
494 { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
495 struct message_part *part; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
496 struct parser_context parser_ctx; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
497 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
498 memset(&parser_ctx, 0, sizeof(parser_ctx)); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
499 parser_ctx.pool = pool; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
500 parser_ctx.callback = callback; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
501 parser_ctx.context = context; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
502 parser_ctx.part = part = p_new(pool, struct message_part, 1); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
503 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
504 message_parse_part(input, &parser_ctx); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
505 return part; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
506 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
507 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
508 void message_parse_header(struct message_part *part, struct istream *input, |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
509 struct message_size *hdr_size, |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
510 message_header_callback_t *callback, void *context) |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
511 { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
512 struct message_header_parser_ctx *hdr_ctx; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
513 struct message_header_line *hdr; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
514 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
515 hdr_ctx = message_parse_header_init(input, hdr_size); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
516 while ((hdr = message_parse_header_next(hdr_ctx)) != NULL) |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
517 callback(part, hdr, context); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
518 callback(part, NULL, context); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
519 message_parse_header_deinit(hdr_ctx); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
520 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
521 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
522 struct message_header_parser_ctx * |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
523 message_parse_header_init(struct istream *input, struct message_size *hdr_size) |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
524 { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
525 struct message_header_parser_ctx *ctx; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
526 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
527 ctx = i_new(struct message_header_parser_ctx, 1); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
528 ctx->input = input; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
529 ctx->hdr_size = hdr_size; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
530 ctx->name = str_new(default_pool, 128); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
531 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
532 if (hdr_size != NULL) |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
533 memset(hdr_size, 0, sizeof(*hdr_size)); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
534 return ctx; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
535 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
536 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
537 void message_parse_header_deinit(struct message_header_parser_ctx *ctx) |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
538 { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
539 i_stream_skip(ctx->input, ctx->skip); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
540 if (ctx->value_buf != NULL) |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
541 buffer_free(ctx->value_buf); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
542 str_free(ctx->name); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
543 i_free(ctx); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
544 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
545 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
546 struct message_header_line * |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
547 message_parse_header_next(struct message_header_parser_ctx *ctx) |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
548 { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
549 struct message_header_line *line = &ctx->line; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
550 const unsigned char *msg; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
551 size_t i, size, startpos, colon_pos, parse_size; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
552 int ret; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
553 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
554 if (line->eoh) |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
555 return NULL; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
556 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
557 if (ctx->skip > 0) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
558 i_stream_skip(ctx->input, ctx->skip); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
559 ctx->skip = 0; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
560 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
561 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
562 startpos = 0; colon_pos = UINT_MAX; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
563 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
564 line->no_newline = FALSE; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
565 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
566 if (line->continues) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
567 if (line->use_full_value && !line->continued) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
568 /* save the first line */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
569 if (ctx->value_buf != NULL) |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
570 buffer_set_used_size(ctx->value_buf, 0); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
571 else { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
572 ctx->value_buf = |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
573 buffer_create_dynamic(default_pool, |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
574 4096, (size_t)-1); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
575 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
576 buffer_append(ctx->value_buf, |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
577 line->value, line->value_len); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
578 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
579 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
580 line->continued = TRUE; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
581 line->continues = FALSE; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
582 colon_pos = 0; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
583 } else { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
584 /* new header line */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
585 line->continued = FALSE; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
586 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
587 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
588 for (;;) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
589 ret = i_stream_read_data(ctx->input, &msg, &size, startpos+1); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
590 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
591 if (ret != 0) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
592 /* we want to know one byte in advance to find out |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
593 if it's multiline header */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
594 parse_size = size-1; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
595 } else { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
596 parse_size = size; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
597 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
598 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
599 if (ret <= 0 && (ret != 0 || startpos == size)) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
600 if (ret == -1) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
601 /* error / EOF with no bytes */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
602 return NULL; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
603 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
604 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
605 /* a) line is larger than input buffer |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
606 b) header ended unexpectedly */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
607 if (colon_pos == UINT_MAX) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
608 /* header name is huge. just skip it. */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
609 message_skip_line(ctx->input, ctx->hdr_size, |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
610 TRUE); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
611 continue; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
612 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
613 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
614 /* go back to last LWSP if found. */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
615 for (i = size-1; i > colon_pos; i--) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
616 if (IS_LWSP(msg[i])) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
617 size = i; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
618 break; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
619 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
620 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
621 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
622 line->no_newline = TRUE; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
623 line->continues = TRUE; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
624 ctx->skip = size; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
625 break; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
626 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
627 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
628 /* find ':' */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
629 if (colon_pos == UINT_MAX) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
630 for (i = startpos; i < parse_size; i++) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
631 if (msg[i] <= ':') { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
632 if (msg[i] == ':') { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
633 colon_pos = i; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
634 break; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
635 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
636 if (msg[i] == '\n') { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
637 /* end of headers, or error */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
638 break; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
639 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
640 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
641 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
642 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
643 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
644 /* find '\n' */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
645 for (i = startpos; i < parse_size; i++) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
646 if (msg[i] == '\n') |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
647 break; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
648 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
649 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
650 if (i < parse_size) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
651 /* got a line */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
652 line->continues = i+1 < size && IS_LWSP(msg[i+1]); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
653 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
654 if (ctx->hdr_size != NULL) |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
655 ctx->hdr_size->lines++; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
656 if (i == 0 || msg[i-1] != '\r') { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
657 /* missing CR */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
658 if (ctx->hdr_size != NULL) |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
659 ctx->hdr_size->virtual_size++; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
660 size = i; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
661 } else { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
662 size = i-1; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
663 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
664 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
665 ctx->skip = i+1; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
666 break; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
667 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
668 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
669 startpos = i; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
670 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
671 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
672 if (size == 0 || (size == 1 && msg[0] == '\r')) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
673 /* end of headers */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
674 line->eoh = TRUE; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
675 line->name_len = line->value_len = 0; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
676 } else if (line->continued) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
677 line->value = msg; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
678 line->value_len = size; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
679 } else if (colon_pos == UINT_MAX) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
680 /* missing ':', assume the whole line is name */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
681 line->value = NULL; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
682 line->value_len = 0; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
683 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
684 str_truncate(ctx->name, 0); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
685 str_append_n(ctx->name, msg, size); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
686 line->name = str_c(ctx->name); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
687 line->name_len = str_len(ctx->name); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
688 } else { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
689 /* get value, skip only first LWSP after ':' */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
690 line->value = msg + colon_pos+1; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
691 line->value_len = size - colon_pos - 1; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
692 if (line->value_len > 0 && |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
693 IS_LWSP(line->value[0])) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
694 line->value++; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
695 line->value_len--; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
696 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
697 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
698 /* get name, skip LWSP before ':' */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
699 while (colon_pos > 0 && IS_LWSP(msg[colon_pos-1])) |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
700 colon_pos--; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
701 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
702 str_truncate(ctx->name, 0); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
703 str_append_n(ctx->name, msg, colon_pos); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
704 line->name = str_c(ctx->name); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
705 line->name_len = str_len(ctx->name); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
706 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
707 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
708 if (!line->continued) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
709 /* first header line, set full_value = value */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
710 line->full_value = line->value; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
711 line->full_value_len = line->value_len; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
712 } else if (line->use_full_value) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
713 /* continue saving the full value */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
714 buffer_append(ctx->value_buf, line->value, line->value_len); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
715 line->full_value = buffer_get_data(ctx->value_buf, |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
716 &line->full_value_len); |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
717 } else { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
718 /* we didn't want full_value, and this is a continued line. */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
719 line->full_value = NULL; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
720 line->full_value_len = 0; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
721 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
722 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
723 /* always reset it */ |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
724 line->use_full_value = FALSE; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
725 |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
726 if (ctx->hdr_size != NULL) { |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
727 ctx->hdr_size->physical_size += ctx->skip; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
728 ctx->hdr_size->virtual_size += ctx->skip; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
729 } |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
730 return line; |
97f8c00b8d4c
Better handling for multiline headers. Before we skipped headers larger than
Timo Sirainen <tss@iki.fi>
parents:
1253
diff
changeset
|
731 } |