Commit | Line | Data |
---|---|---|
8523533c TK |
1 | /************************************************* |
2 | * Exim - an Internet mail transport agent * | |
3 | *************************************************/ | |
4 | ||
3386088d | 5 | /* Copyright (c) Tom Kistner <tom@duncanthrax.net> 2003-2015 */ |
8523533c TK |
6 | /* License: GPL */ |
7 | ||
8 | /* Code for matching regular expressions against headers and body. | |
9 | Called from acl.c. */ | |
10 | ||
11 | #include "exim.h" | |
12 | #ifdef WITH_CONTENT_SCAN | |
13 | #include <unistd.h> | |
14 | #include <sys/mman.h> | |
15 | ||
16 | /* Structure to hold a list of Regular expressions */ | |
17 | typedef struct pcre_list { | |
18 | pcre *re; | |
19 | uschar *pcre_text; | |
20 | struct pcre_list *next; | |
21 | } pcre_list; | |
22 | ||
23 | uschar regex_match_string_buffer[1024]; | |
24 | ||
25 | extern FILE *mime_stream; | |
26 | extern uschar *mime_current_boundary; | |
27 | ||
f38917cc JH |
28 | static pcre_list * |
29 | compile(const uschar * list) | |
55414b25 | 30 | { |
8523533c | 31 | int sep = 0; |
8523533c TK |
32 | uschar *regex_string; |
33 | uschar regex_string_buffer[1024]; | |
8523533c TK |
34 | const char *pcre_error; |
35 | int pcre_erroffset; | |
f38917cc JH |
36 | pcre_list *re_list_head = NULL; |
37 | pcre_list *ri; | |
38 | ||
39 | /* precompile our regexes */ | |
40 | while ((regex_string = string_nextinlist(&list, &sep, | |
41 | regex_string_buffer, | |
42 | sizeof(regex_string_buffer))) != NULL) { | |
43 | pcre *re; | |
44 | ||
45 | /* parse option */ | |
46 | if ( (strcmpic(regex_string,US"false") == 0) || | |
47 | (Ustrcmp(regex_string,"0") == 0) ) | |
48 | continue; /* explicitly no matching */ | |
49 | ||
50 | /* compile our regular expression */ | |
51 | if (!(re = pcre_compile( CS regex_string, | |
52 | 0, &pcre_error, &pcre_erroffset, NULL ))) { | |
53 | log_write(0, LOG_MAIN, | |
54 | "regex acl condition warning - error in regex '%s': %s at offset %d, skipped.", | |
55 | regex_string, pcre_error, pcre_erroffset); | |
56 | continue; | |
57 | } | |
58 | ||
59 | ri = store_get(sizeof(pcre_list)); | |
60 | ri->re = re; | |
61 | ri->pcre_text = string_copy(regex_string); | |
62 | ri->next = re_list_head; | |
63 | re_list_head = ri; | |
64 | } | |
65 | return re_list_head; | |
66 | } | |
67 | ||
68 | static int | |
69 | matcher(pcre_list * re_list_head, uschar * linebuffer, int len) | |
70 | { | |
71 | pcre_list * ri; | |
72 | ||
73 | for(ri = re_list_head; ri; ri = ri->next) | |
74 | { | |
75 | int ovec[3*(REGEX_VARS+1)]; | |
76 | int n, nn; | |
77 | ||
78 | /* try matcher on the line */ | |
79 | n = pcre_exec(ri->re, NULL, | |
80 | CS linebuffer, len, 0, 0, | |
81 | ovec, nelem(ovec)); | |
82 | if (n > 0) | |
83 | { | |
84 | Ustrncpy(regex_match_string_buffer, ri->pcre_text, 1023); | |
85 | regex_match_string = regex_match_string_buffer; | |
86 | ||
87 | for (nn = 1; nn < n; nn++) | |
88 | regex_vars[nn-1] = | |
89 | string_copyn(linebuffer + ovec[nn*2], ovec[nn*2+1] - ovec[nn*2]); | |
90 | ||
91 | return OK; | |
92 | } | |
93 | } | |
94 | return FAIL; | |
95 | } | |
96 | ||
97 | int | |
98 | regex(const uschar **listptr) | |
99 | { | |
100 | unsigned long mbox_size; | |
101 | FILE *mbox_file; | |
102 | pcre_list *re_list_head; | |
8523533c TK |
103 | uschar *linebuffer; |
104 | long f_pos = 0; | |
f38917cc | 105 | int ret = FAIL; |
8e669ac1 | 106 | |
8523533c TK |
107 | /* reset expansion variable */ |
108 | regex_match_string = NULL; | |
8e669ac1 | 109 | |
f38917cc | 110 | if (mime_stream == NULL) { /* We are in the DATA ACL */ |
8544e77a | 111 | mbox_file = spool_mbox(&mbox_size, NULL); |
f38917cc | 112 | if (mbox_file == NULL) { /* error while spooling */ |
8523533c TK |
113 | log_write(0, LOG_MAIN|LOG_PANIC, |
114 | "regex acl condition: error while creating mbox spool file"); | |
115 | return DEFER; | |
f38917cc | 116 | } |
8523533c TK |
117 | } |
118 | else { | |
119 | f_pos = ftell(mime_stream); | |
120 | mbox_file = mime_stream; | |
f38917cc | 121 | } |
8e669ac1 | 122 | |
8523533c | 123 | /* precompile our regexes */ |
f38917cc JH |
124 | if (!(re_list_head = compile(*listptr))) |
125 | return FAIL; /* no regexes -> nothing to do */ | |
8e669ac1 | 126 | |
8523533c TK |
127 | /* match each line against all regexes */ |
128 | linebuffer = store_get(32767); | |
8e669ac1 | 129 | while (fgets(CS linebuffer, 32767, mbox_file) != NULL) { |
8e669ac1 | 130 | |
f38917cc JH |
131 | if ( mime_stream && mime_current_boundary /* check boundary */ |
132 | && Ustrncmp(linebuffer,"--",2) == 0 | |
133 | && Ustrncmp((linebuffer+2),mime_current_boundary,Ustrlen(mime_current_boundary)) == 0) | |
134 | break; /* found boundary */ | |
135 | ||
136 | if ((ret = matcher(re_list_head, linebuffer, (int)Ustrlen(linebuffer))) == OK) | |
137 | goto done; | |
138 | } | |
139 | /* no matches ... */ | |
140 | ||
141 | done: | |
8523533c | 142 | if (mime_stream == NULL) |
f1e894f3 | 143 | (void)fclose(mbox_file); |
8523533c TK |
144 | else { |
145 | clearerr(mime_stream); | |
146 | fseek(mime_stream,f_pos,SEEK_SET); | |
147 | }; | |
8e669ac1 | 148 | |
f38917cc | 149 | return ret; |
8523533c TK |
150 | } |
151 | ||
152 | ||
55414b25 JH |
153 | int |
154 | mime_regex(const uschar **listptr) | |
155 | { | |
8523533c | 156 | pcre_list *re_list_head = NULL; |
8523533c TK |
157 | FILE *f; |
158 | uschar *mime_subject = NULL; | |
159 | int mime_subject_len = 0; | |
f38917cc | 160 | int ret; |
8523533c TK |
161 | |
162 | /* reset expansion variable */ | |
163 | regex_match_string = NULL; | |
164 | ||
165 | /* precompile our regexes */ | |
f38917cc JH |
166 | if (!(re_list_head = compile(*listptr))) |
167 | return FAIL; /* no regexes -> nothing to do */ | |
8e669ac1 | 168 | |
8523533c TK |
169 | /* check if the file is already decoded */ |
170 | if (mime_decoded_filename == NULL) { | |
55414b25 | 171 | const uschar *empty = US""; |
8523533c TK |
172 | /* no, decode it first */ |
173 | mime_decode(&empty); | |
174 | if (mime_decoded_filename == NULL) { | |
175 | /* decoding failed */ | |
176 | log_write(0, LOG_MAIN, | |
177 | "mime_regex acl condition warning - could not decode MIME part to file."); | |
178 | return DEFER; | |
f38917cc JH |
179 | } |
180 | } | |
8523533c TK |
181 | |
182 | /* open file */ | |
f38917cc | 183 | if (!(f = fopen(CS mime_decoded_filename, "rb"))) { |
8523533c | 184 | log_write(0, LOG_MAIN, |
f38917cc JH |
185 | "mime_regex acl condition warning - can't open '%s' for reading.", |
186 | mime_decoded_filename); | |
8523533c | 187 | return DEFER; |
f38917cc | 188 | } |
8e669ac1 | 189 | |
8523533c TK |
190 | /* get 32k memory */ |
191 | mime_subject = (uschar *)store_get(32767); | |
8e669ac1 | 192 | |
8523533c | 193 | mime_subject_len = fread(mime_subject, 1, 32766, f); |
8e669ac1 | 194 | |
f38917cc | 195 | ret = matcher(re_list_head, mime_subject, mime_subject_len); |
f1e894f3 | 196 | (void)fclose(f); |
f38917cc | 197 | return ret; |
8523533c TK |
198 | } |
199 | ||
476be7e2 | 200 | #endif /* WITH_CONTENT_SCAN */ |