Mercurial > dovecot > original-hg > dovecot-1.2
annotate src/lib-mail/message-parser.c @ 956:26cafa3dc09c HEAD
minor optimization
author | Timo Sirainen <tss@iki.fi> |
---|---|
date | Sun, 12 Jan 2003 01:49:45 +0200 |
parents | 411006be3c66 |
children | 60646878858e |
rev | line source |
---|---|
0 | 1 /* Copyright (C) 2002 Timo Sirainen */ |
2 | |
3 #include "lib.h" | |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
4 #include "istream.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
|
5 #include "strescape.h" |
0 | 6 #include "message-content-parser.h" |
7 #include "message-parser.h" | |
9
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
8 #include "message-size.h" |
0 | 9 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
10 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
|
11 struct message_boundary *next; |
0 | 12 |
903
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_part *part; |
0 | 14 const char *boundary; |
184 | 15 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
|
16 }; |
0 | 17 |
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 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
|
19 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
|
20 struct message_part *part; |
0 | 21 |
22 char *last_boundary; | |
23 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
|
24 struct message_boundary *boundaries; |
0 | 25 |
953
411006be3c66
Naming change for function typedefs.
Timo Sirainen <tss@iki.fi>
parents:
903
diff
changeset
|
26 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
|
27 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
|
28 }; |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
29 |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
30 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
|
31 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
|
32 struct parser_context *parser_ctx); |
0 | 33 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
34 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
|
35 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
|
36 struct message_size *body_size); |
0 | 37 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
38 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
|
39 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
|
40 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
|
41 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
|
42 |
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 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
|
44 struct message_part *part) |
0 | 45 { |
46 dest->physical_size += | |
47 part->header_size.physical_size + | |
48 part->body_size.physical_size; | |
49 dest->virtual_size += | |
50 part->header_size.virtual_size + | |
51 part->body_size.virtual_size; | |
52 dest->lines += part->header_size.lines + part->body_size.lines; | |
53 } | |
54 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
55 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
|
56 message_part_append(pool_t pool, struct message_part *parent) |
0 | 57 { |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
58 struct message_part *part, **list; |
0 | 59 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
60 part = p_new(pool, struct message_part, 1); |
0 | 61 part->parent = parent; |
62 | |
212
4327b1266604
message/rfc822 mime parts weren't parsed correctly
Timo Sirainen <tss@iki.fi>
parents:
184
diff
changeset
|
63 /* set child position */ |
4327b1266604
message/rfc822 mime parts weren't parsed correctly
Timo Sirainen <tss@iki.fi>
parents:
184
diff
changeset
|
64 part->physical_pos = |
4327b1266604
message/rfc822 mime parts weren't parsed correctly
Timo Sirainen <tss@iki.fi>
parents:
184
diff
changeset
|
65 parent->physical_pos + |
4327b1266604
message/rfc822 mime parts weren't parsed correctly
Timo Sirainen <tss@iki.fi>
parents:
184
diff
changeset
|
66 parent->body_size.physical_size + |
4327b1266604
message/rfc822 mime parts weren't parsed correctly
Timo Sirainen <tss@iki.fi>
parents:
184
diff
changeset
|
67 parent->header_size.physical_size; |
4327b1266604
message/rfc822 mime parts weren't parsed correctly
Timo Sirainen <tss@iki.fi>
parents:
184
diff
changeset
|
68 |
0 | 69 list = &part->parent->children; |
70 while (*list != NULL) | |
71 list = &(*list)->next; | |
72 | |
73 *list = part; | |
74 return part; | |
75 } | |
76 | |
898
0d5be52d7131
Use unsigned char* when accessing non-NUL terminating strings. Compiler
Timo Sirainen <tss@iki.fi>
parents:
896
diff
changeset
|
77 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
|
78 void *context) |
0 | 79 { |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
80 struct parser_context *parser_ctx = context; |
0 | 81 const char *str; |
82 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
83 if (parser_ctx->last_content_type != NULL || value_len == 0) |
0 | 84 return; |
85 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
86 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
|
87 p_strndup(parser_ctx->pool, value, value_len); |
0 | 88 |
89 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
|
90 parser_ctx->part->flags |= MESSAGE_PART_FLAG_MESSAGE_RFC822; |
0 | 91 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
|
92 parser_ctx->part->flags |= MESSAGE_PART_FLAG_TEXT; |
0 | 93 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
|
94 parser_ctx->part->flags |= MESSAGE_PART_FLAG_MULTIPART; |
0 | 95 |
106
5fe3e04ca8d9
Added support for caching of MessagePart data. This is useful for fetching
Timo Sirainen <tss@iki.fi>
parents:
105
diff
changeset
|
96 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
|
97 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
|
98 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
|
99 } |
0 | 100 } |
101 } | |
102 | |
898
0d5be52d7131
Use unsigned char* when accessing non-NUL terminating strings. Compiler
Timo Sirainen <tss@iki.fi>
parents:
896
diff
changeset
|
103 static void |
0d5be52d7131
Use unsigned char* when accessing non-NUL terminating strings. Compiler
Timo Sirainen <tss@iki.fi>
parents:
896
diff
changeset
|
104 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
|
105 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
|
106 int value_quoted, void *context) |
0 | 107 { |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
108 struct parser_context *parser_ctx = context; |
0 | 109 |
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 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
|
111 name_len != 8 || memcasecmp(name, "boundary", 8) != 0) |
0 | 112 return; |
113 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
114 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
|
115 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
|
116 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
|
117 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
|
118 str_unescape(parser_ctx->last_boundary); |
0 | 119 } |
120 } | |
121 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
122 static void parse_header_field(struct message_part *part, |
898
0d5be52d7131
Use unsigned char* when accessing non-NUL terminating strings. Compiler
Timo Sirainen <tss@iki.fi>
parents:
896
diff
changeset
|
123 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
|
124 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
|
125 void *context) |
0 | 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 struct parser_context *parser_ctx = context; |
0 | 128 |
129 /* call the user-defined header parser */ | |
953
411006be3c66
Naming change for function typedefs.
Timo Sirainen <tss@iki.fi>
parents:
903
diff
changeset
|
130 if (parser_ctx->callback != NULL) { |
411006be3c66
Naming change for function typedefs.
Timo Sirainen <tss@iki.fi>
parents:
903
diff
changeset
|
131 parser_ctx->callback(part, name, name_len, value, value_len, |
411006be3c66
Naming change for function typedefs.
Timo Sirainen <tss@iki.fi>
parents:
903
diff
changeset
|
132 parser_ctx->context); |
0 | 133 } |
134 | |
898
0d5be52d7131
Use unsigned char* when accessing non-NUL terminating strings. Compiler
Timo Sirainen <tss@iki.fi>
parents:
896
diff
changeset
|
135 if (name_len == 12 && memcasecmp(name, "Content-Type", 12) == 0) { |
0 | 136 /* we need to know the boundary */ |
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
|
137 message_content_parse_header(value, value_len, |
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
|
138 parse_content_type, |
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
|
139 parse_content_type_param, |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
140 parser_ctx); |
0 | 141 } |
142 } | |
143 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
144 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
|
145 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
|
146 struct parser_context *parser_ctx) |
0 | 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 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
|
149 struct message_boundary *b; |
0 | 150 |
151 /* 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
|
152 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
|
153 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
|
154 b->boundary = parser_ctx->last_boundary; |
0 | 155 b->len = strlen(b->boundary); |
156 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
157 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
|
158 parser_ctx->boundaries = b; |
0 | 159 |
160 /* 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
|
161 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
|
162 parser_ctx->last_content_type = NULL; |
0 | 163 |
164 /* 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
|
165 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
|
166 next_part = message_skip_boundary(input, parser_ctx->boundaries, |
0 | 167 &parent_part->body_size); |
168 | |
169 /* now, parse the parts */ | |
170 while (next_part == parent_part) { | |
171 /* 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
|
172 part = message_part_append(parser_ctx->pool, parent_part); |
0 | 173 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
174 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
|
175 next_part = message_parse_part(input, parser_ctx); |
0 | 176 |
177 /* update our size */ | |
178 message_size_add_part(&parent_part->body_size, part); | |
179 | |
180 if (next_part != parent_part) | |
181 break; | |
182 | |
183 /* 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
|
184 next_part = message_skip_boundary(input, parser_ctx->boundaries, |
0 | 185 &parent_part->body_size); |
186 } | |
187 | |
188 /* 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
|
189 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
|
190 parser_ctx->boundaries = b->next; |
0 | 191 return next_part; |
192 } | |
193 | |
874 | 194 #define MUTEX_FLAGS \ |
195 (MESSAGE_PART_FLAG_MESSAGE_RFC822 | MESSAGE_PART_FLAG_MULTIPART) | |
196 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
197 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
|
198 message_parse_part(struct istream *input, struct parser_context *parser_ctx) |
0 | 199 { |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
200 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
|
201 uoff_t hdr_size; |
0 | 202 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
203 message_parse_header(parser_ctx->part, input, |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
204 &parser_ctx->part->header_size, |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
205 parse_header_field, parser_ctx); |
0 | 206 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
207 i_assert((parser_ctx->part->flags & MUTEX_FLAGS) != MUTEX_FLAGS); |
874 | 208 |
0 | 209 /* 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
|
210 hdr_size = parser_ctx->part->header_size.physical_size; |
0 | 211 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
212 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
|
213 return message_parse_multipart(input, parser_ctx); |
0 | 214 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
215 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
|
216 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
|
217 (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
|
218 MESSAGE_PART_FLAG_MULTIPART_DIGEST)) { |
0 | 219 /* when there's no content-type specified and we're |
220 below multipart/digest, the assume message/rfc822 | |
221 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
|
222 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
|
223 MESSAGE_PART_FLAG_MESSAGE_RFC822; |
0 | 224 } else { |
225 /* 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
|
226 parser_ctx->part->flags |= MESSAGE_PART_FLAG_TEXT; |
0 | 227 } |
228 } | |
229 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
230 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
|
231 parser_ctx->last_content_type = NULL; |
0 | 232 |
903
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->flags & MESSAGE_PART_FLAG_MESSAGE_RFC822) { |
0 | 234 /* message/rfc822 part - the message body begins with |
235 headers again, this works pretty much the same as | |
236 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
|
237 part = message_part_append(parser_ctx->pool, parser_ctx->part); |
0 | 238 |
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 = part; |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
240 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
|
241 parser_ctx->part = part->parent; |
0 | 242 |
243 /* our body size is the size of header+body in message/rfc822 */ | |
244 message_size_add_part(&part->parent->body_size, part); | |
245 } else { | |
246 /* 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
|
247 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
|
248 next_part = message_parse_body(input, parser_ctx->boundaries, |
0 | 249 &part->body_size); |
250 } | |
251 | |
252 return next_part; | |
253 } | |
254 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
255 struct message_part *message_parse(pool_t pool, struct istream *input, |
953
411006be3c66
Naming change for function typedefs.
Timo Sirainen <tss@iki.fi>
parents:
903
diff
changeset
|
256 message_header_callback_t callback, |
411006be3c66
Naming change for function typedefs.
Timo Sirainen <tss@iki.fi>
parents:
903
diff
changeset
|
257 void *context) |
0 | 258 { |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
259 struct message_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
|
260 struct parser_context parser_ctx; |
0 | 261 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
262 memset(&parser_ctx, 0, sizeof(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
|
263 parser_ctx.pool = pool; |
953
411006be3c66
Naming change for function typedefs.
Timo Sirainen <tss@iki.fi>
parents:
903
diff
changeset
|
264 parser_ctx.callback = callback; |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
265 parser_ctx.context = context; |
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
266 parser_ctx.part = part = p_new(pool, struct message_part, 1); |
0 | 267 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
268 message_parse_part(input, &parser_ctx); |
0 | 269 return part; |
270 } | |
271 | |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
272 /* skip over to next line increasing message 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
|
273 static void message_skip_line(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
|
274 struct message_size *msg_size) |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
275 { |
410
1f0e7229ee58
Split IOBuffer into mmaped IBuffer, file IBuffer, memory data IBuffer and
Timo Sirainen <tss@iki.fi>
parents:
369
diff
changeset
|
276 const unsigned char *msg; |
184 | 277 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
|
278 |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
279 startpos = 0; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
280 |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
281 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
|
282 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
|
283 if (msg[i] == '\n') { |
153
9660df1ca44f
message_parse_header() was buggy with big headers, and io_buffer_read_data()
Timo Sirainen <tss@iki.fi>
parents:
112
diff
changeset
|
284 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
|
285 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
|
286 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
|
287 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
|
288 } |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
289 break; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
290 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
291 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
292 |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
293 if (i < size) { |
9
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
294 startpos = 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
|
295 break; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
296 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
297 |
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
|
298 /* 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
|
299 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
|
300 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
|
301 |
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 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
|
303 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
|
304 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
|
305 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
306 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
307 |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
308 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
|
309 |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
310 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
|
311 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
|
312 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
|
313 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
314 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
315 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
316 void message_parse_header(struct message_part *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
|
317 struct message_size *hdr_size, |
953
411006be3c66
Naming change for function typedefs.
Timo Sirainen <tss@iki.fi>
parents:
903
diff
changeset
|
318 message_header_callback_t callback, void *context) |
0 | 319 { |
410
1f0e7229ee58
Split IOBuffer into mmaped IBuffer, file IBuffer, memory data IBuffer and
Timo Sirainen <tss@iki.fi>
parents:
369
diff
changeset
|
320 const unsigned char *msg; |
513
01e94a80235d
Header parsing should always keep one character lookahead, so it can see if
Timo Sirainen <tss@iki.fi>
parents:
505
diff
changeset
|
321 size_t i, size, parse_size, startpos, missing_cr_count; |
184 | 322 size_t line_start, colon_pos, end_pos, name_len, value_len; |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
323 int ret; |
0 | 324 |
325 if (hdr_size != NULL) | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
326 memset(hdr_size, 0, sizeof(struct message_size)); |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
327 |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
328 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
|
329 colon_pos = UINT_MAX; |
253
62058959536a
message_parse_header() works now properly if there's no message body at all,
Timo Sirainen <tss@iki.fi>
parents:
232
diff
changeset
|
330 for (;;) { |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
331 ret = i_stream_read_data(input, &msg, &size, 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
|
332 if (ret == -2) { |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
333 /* overflow, line is too long. just skip it. */ |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
334 i_assert(size > 2); |
0 | 335 |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
336 message_skip_line(input, hdr_size); |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
337 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
|
338 colon_pos = UINT_MAX; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
339 continue; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
340 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
341 |
517 | 342 if (ret < 0 || (ret <= 0 && size == startpos)) { |
343 /* EOF and nothing in buffer. the later check is | |
344 needed only when there's no message body */ | |
345 break; | |
253
62058959536a
message_parse_header() works now properly if there's no message body at all,
Timo Sirainen <tss@iki.fi>
parents:
232
diff
changeset
|
346 } |
62058959536a
message_parse_header() works now properly if there's no message body at all,
Timo Sirainen <tss@iki.fi>
parents:
232
diff
changeset
|
347 |
517 | 348 parse_size = size <= startpos+1 ? size : size-1; |
513
01e94a80235d
Header parsing should always keep one character lookahead, so it can see if
Timo Sirainen <tss@iki.fi>
parents:
505
diff
changeset
|
349 for (i = startpos; i < parse_size; i++) { |
9
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
350 if (msg[i] == ':' && colon_pos == UINT_MAX) { |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
351 colon_pos = i; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
352 continue; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
353 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
354 |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
355 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
|
356 continue; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
357 |
0 | 358 if (hdr_size != NULL) |
359 hdr_size->lines++; | |
360 | |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
361 if (i == 0 || msg[i-1] != '\r') { |
0 | 362 /* missing CR */ |
363 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
|
364 } |
0 | 365 |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
366 if (i == 0 || (i == 1 && msg[i-1] == '\r')) { |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
367 /* no headers at all */ |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
368 break; |
0 | 369 } |
370 | |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
371 if ((i > 0 && msg[i-1] == '\n') || |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
372 (i > 1 && msg[i-2] == '\n' && msg[i-1] == '\r')) { |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
373 /* \n\n or \n\r\n - end of headers */ |
0 | 374 break; |
375 } | |
376 | |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
377 /* make sure the header doesn't continue to next line */ |
253
62058959536a
message_parse_header() works now properly if there's no message body at all,
Timo Sirainen <tss@iki.fi>
parents:
232
diff
changeset
|
378 if (i+1 == size || !IS_LWSP(msg[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
|
379 if (colon_pos != UINT_MAX && |
953
411006be3c66
Naming change for function typedefs.
Timo Sirainen <tss@iki.fi>
parents:
903
diff
changeset
|
380 colon_pos != line_start && |
411006be3c66
Naming change for function typedefs.
Timo Sirainen <tss@iki.fi>
parents:
903
diff
changeset
|
381 callback != NULL && |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
382 !IS_LWSP(msg[line_start])) { |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
383 /* we have a valid header line */ |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
384 |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
385 /* get length of name-field */ |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
386 end_pos = colon_pos-1; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
387 while (end_pos > line_start && |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
388 IS_LWSP(msg[end_pos])) |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
389 end_pos--; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
390 name_len = end_pos - line_start + 1; |
0 | 391 |
169
53e8e405266b
Don't strip extra spaces after "field: ".
Timo Sirainen <tss@iki.fi>
parents:
153
diff
changeset
|
392 /* get length of value field. skip |
53e8e405266b
Don't strip extra spaces after "field: ".
Timo Sirainen <tss@iki.fi>
parents:
153
diff
changeset
|
393 only the initial LWSP after ':'. |
53e8e405266b
Don't strip extra spaces after "field: ".
Timo Sirainen <tss@iki.fi>
parents:
153
diff
changeset
|
394 some fields may want to keep |
53e8e405266b
Don't strip extra spaces after "field: ".
Timo Sirainen <tss@iki.fi>
parents:
153
diff
changeset
|
395 the extra spaces.. */ |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
396 colon_pos++; |
169
53e8e405266b
Don't strip extra spaces after "field: ".
Timo Sirainen <tss@iki.fi>
parents:
153
diff
changeset
|
397 if (colon_pos < i && |
53e8e405266b
Don't strip extra spaces after "field: ".
Timo Sirainen <tss@iki.fi>
parents:
153
diff
changeset
|
398 IS_LWSP(msg[colon_pos])) |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
399 colon_pos++; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
400 value_len = i - colon_pos; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
401 if (msg[i-1] == '\r') value_len--; |
0 | 402 |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
403 /* and finally call the function */ |
953
411006be3c66
Naming change for function typedefs.
Timo Sirainen <tss@iki.fi>
parents:
903
diff
changeset
|
404 callback(part, |
411006be3c66
Naming change for function typedefs.
Timo Sirainen <tss@iki.fi>
parents:
903
diff
changeset
|
405 msg + line_start, name_len, |
411006be3c66
Naming change for function typedefs.
Timo Sirainen <tss@iki.fi>
parents:
903
diff
changeset
|
406 msg + colon_pos, value_len, |
411006be3c66
Naming change for function typedefs.
Timo Sirainen <tss@iki.fi>
parents:
903
diff
changeset
|
407 context); |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
408 } |
0 | 409 |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
410 colon_pos = UINT_MAX; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
411 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
|
412 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
413 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
414 |
516
fc13fbca2fff
Header parser was still a bit wrong.
Timo Sirainen <tss@iki.fi>
parents:
513
diff
changeset
|
415 if (i < parse_size) { |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
416 /* end of header */ |
9
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
417 startpos = i+1; |
0 | 418 break; |
419 } | |
420 | |
253
62058959536a
message_parse_header() works now properly if there's no message body at all,
Timo Sirainen <tss@iki.fi>
parents:
232
diff
changeset
|
421 /* leave the last line to buffer */ |
62058959536a
message_parse_header() works now properly if there's no message body at all,
Timo Sirainen <tss@iki.fi>
parents:
232
diff
changeset
|
422 if (colon_pos != UINT_MAX) |
62058959536a
message_parse_header() works now properly if there's no message body at all,
Timo Sirainen <tss@iki.fi>
parents:
232
diff
changeset
|
423 colon_pos -= line_start; |
62058959536a
message_parse_header() works now properly if there's no message body at all,
Timo Sirainen <tss@iki.fi>
parents:
232
diff
changeset
|
424 if (hdr_size != NULL) |
62058959536a
message_parse_header() works now properly if there's no message body at all,
Timo Sirainen <tss@iki.fi>
parents:
232
diff
changeset
|
425 hdr_size->physical_size += line_start; |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
426 i_stream_skip(input, 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
|
427 |
253
62058959536a
message_parse_header() works now properly if there's no message body at all,
Timo Sirainen <tss@iki.fi>
parents:
232
diff
changeset
|
428 startpos = i-line_start; |
62058959536a
message_parse_header() works now properly if there's no message body at all,
Timo Sirainen <tss@iki.fi>
parents:
232
diff
changeset
|
429 line_start = 0; |
0 | 430 } |
253
62058959536a
message_parse_header() works now properly if there's no message body at all,
Timo Sirainen <tss@iki.fi>
parents:
232
diff
changeset
|
431 |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
432 i_stream_skip(input, startpos); |
0 | 433 |
434 if (hdr_size != NULL) { | |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
435 hdr_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
|
436 hdr_size->virtual_size += |
0 | 437 hdr_size->physical_size + 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
|
438 i_assert(hdr_size->virtual_size >= hdr_size->physical_size); |
0 | 439 } |
369
d037915978ca
message_parse_header() now calls the function with empty name/value at end
Timo Sirainen <tss@iki.fi>
parents:
348
diff
changeset
|
440 |
953
411006be3c66
Naming change for function typedefs.
Timo Sirainen <tss@iki.fi>
parents:
903
diff
changeset
|
441 if (callback != NULL) { |
369
d037915978ca
message_parse_header() now calls the function with empty name/value at end
Timo Sirainen <tss@iki.fi>
parents:
348
diff
changeset
|
442 /* "end of headers" notify */ |
953
411006be3c66
Naming change for function typedefs.
Timo Sirainen <tss@iki.fi>
parents:
903
diff
changeset
|
443 callback(part, NULL, 0, NULL, 0, context); |
369
d037915978ca
message_parse_header() now calls the function with empty name/value at end
Timo Sirainen <tss@iki.fi>
parents:
348
diff
changeset
|
444 } |
0 | 445 } |
446 | |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
447 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
|
448 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
|
449 const unsigned char *msg, size_t len) |
0 | 450 { |
451 while (boundaries != NULL) { | |
452 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
|
453 memcmp(boundaries->boundary, msg, boundaries->len) == 0) |
0 | 454 return boundaries; |
455 | |
456 boundaries = boundaries->next; | |
457 } | |
458 | |
459 return NULL; | |
460 } | |
461 | |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
462 /* 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
|
463 [\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
|
464 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
|
465 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
|
466 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
|
467 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
|
468 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
|
469 { |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
470 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
|
471 const unsigned char *msg; |
184 | 472 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
|
473 |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
474 boundary = NULL; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
475 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
|
476 |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
477 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
|
478 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
|
479 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
|
480 continue; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
481 |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
482 if (i > line_start+2 && msg[line_start] == '-' && |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
483 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
|
484 /* possible boundary */ |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
485 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
|
486 msg + line_start + 2, |
0d5be52d7131
Use unsigned char* when accessing non-NUL terminating strings. Compiler
Timo Sirainen <tss@iki.fi>
parents:
896
diff
changeset
|
487 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
|
488 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
|
489 break; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
490 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
491 |
9
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
492 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
|
493 /* missing CR */ |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
494 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
|
495 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
496 |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
497 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
|
498 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
|
499 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
500 |
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
|
501 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
|
502 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
|
503 |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
504 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
|
505 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
|
506 /* 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
|
507 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
|
508 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
|
509 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
|
510 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
|
511 msg + line_start + 2, |
0d5be52d7131
Use unsigned char* when accessing non-NUL terminating strings. Compiler
Timo Sirainen <tss@iki.fi>
parents:
896
diff
changeset
|
512 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
|
513 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
|
514 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
|
515 |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
516 /* 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
|
517 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
|
518 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
|
519 } 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
|
520 /* 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
|
521 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
|
522 i = line_start; |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
523 if (i > 2) i -= 2; /* leave the \r\n too */ |
9277e893304e
s/io_buffer_read_data/io_buffer_read_data_blocking/ and fixed the various
Timo Sirainen <tss@iki.fi>
parents:
212
diff
changeset
|
524 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
|
525 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
526 |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
527 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
|
528 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
|
529 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
|
530 |
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
|
531 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
|
532 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
533 |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
534 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
|
535 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
|
536 /* 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
|
537 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
|
538 } 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
|
539 /* 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
|
540 line_start--; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
541 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
|
542 |
9
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
543 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
|
544 line_start--; |
9
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
545 else |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
546 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
|
547 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
548 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
|
549 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
550 |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
551 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
|
552 msg_size->physical_size += startpos; |
9
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
553 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
|
554 |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
555 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
|
556 |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
557 return boundary; |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
558 } |
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
559 |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
560 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
|
561 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
|
562 struct message_size *body_size) |
0 | 563 { |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
564 struct message_boundary *boundary; |
0 | 565 |
9
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
566 if (boundaries == NULL) { |
856
e49f7397af98
Some bugfixes and speedups for partial fetch handling.
Timo Sirainen <tss@iki.fi>
parents:
765
diff
changeset
|
567 message_get_body_size(input, body_size, (uoff_t)-1, NULL); |
9
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
568 return NULL; |
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
569 } else { |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
570 boundary = message_find_boundary(input, boundaries, |
9
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
571 body_size, FALSE); |
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
572 return boundary == NULL ? NULL : boundary->part; |
21c8e080150d
fixes, seems to be somewhat working now.
Timo Sirainen <tss@iki.fi>
parents:
7
diff
changeset
|
573 } |
0 | 574 } |
575 | |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
576 /* 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
|
577 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
|
578 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
|
579 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
|
580 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
|
581 struct message_size *boundary_size) |
0 | 582 { |
903
fd8888f6f037
Naming style changes, finally got tired of most of the typedefs. Also the
Timo Sirainen <tss@iki.fi>
parents:
898
diff
changeset
|
583 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
|
584 const unsigned char *msg; |
184 | 585 size_t size; |
0 | 586 int end_boundary; |
587 | |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
588 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
|
589 boundary_size, TRUE); |
0 | 590 if (boundary == NULL) |
591 return NULL; | |
592 | |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
593 /* 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
|
594 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
|
595 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
|
596 end_boundary = msg[0] == '-' && msg[1] == '-'; |
0 | 597 |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
598 /* skip the rest of the line */ |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
599 message_skip_line(input, boundary_size); |
0 | 600 |
601 if (end_boundary) { | |
602 /* skip the footer */ | |
764
f57c52738f90
Renamed IBuffer and OBuffer to IStream and OStream which describes their
Timo Sirainen <tss@iki.fi>
parents:
579
diff
changeset
|
603 return message_parse_body(input, boundaries, boundary_size); |
0 | 604 } |
605 | |
5
1b34ec11fff8
Message data is parsed in blocks (no longer entirely mmap()ed). Several
Timo Sirainen <tss@iki.fi>
parents:
0
diff
changeset
|
606 return boundary == NULL ? NULL : boundary->part; |
0 | 607 } |