1 /* $Cambridge: exim/src/src/regex.c,v 1.1.2.1 2004/11/26 09:24:05 tom Exp $ */
3 /*************************************************
4 * Exim - an Internet mail transport agent *
5 *************************************************/
7 /* This file is part of the exiscan-acl content scanner
8 patch. It is NOT part of the standard exim distribution. */
10 /* Copyright (c) Tom Kistner <tom@duncanthrax.net> 2003-???? */
13 /* Code for matching regular expressions against headers and body.
20 /* Structure to hold a list of Regular expressions */
21 typedef struct pcre_list {
24 struct pcre_list *next;
27 uschar regex_match_string_buffer[1024];
29 extern FILE *mime_stream;
30 extern uschar *mime_current_boundary;
32 int regex(uschar **listptr) {
34 uschar *list = *listptr;
36 uschar regex_string_buffer[1024];
37 unsigned long long mbox_size;
40 pcre_list *re_list_head = NULL;
41 pcre_list *re_list_item;
42 const char *pcre_error;
47 /* reset expansion variable */
48 regex_match_string = NULL;
50 if (mime_stream == NULL) {
51 /* We are in the DATA ACL */
52 mbox_file = spool_mbox(&mbox_size);
53 if (mbox_file == NULL) {
54 /* error while spooling */
55 log_write(0, LOG_MAIN|LOG_PANIC,
56 "regex acl condition: error while creating mbox spool file");
61 f_pos = ftell(mime_stream);
62 mbox_file = mime_stream;
65 /* precompile our regexes */
66 while ((regex_string = string_nextinlist(&list, &sep,
68 sizeof(regex_string_buffer))) != NULL) {
71 if ( (strcmpic(regex_string,US"false") == 0) ||
72 (Ustrcmp(regex_string,"0") == 0) ) {
73 /* explicitly no matching */
77 /* compile our regular expression */
78 re = pcre_compile( CS regex_string,
85 log_write(0, LOG_MAIN,
86 "regex acl condition warning - error in regex '%s': %s at offset %d, skipped.", regex_string, pcre_error, pcre_erroffset);
90 re_list_item = store_get(sizeof(pcre_list));
91 re_list_item->re = re;
92 re_list_item->pcre_text = string_copy(regex_string);
93 re_list_item->next = re_list_head;
94 re_list_head = re_list_item;
98 /* no regexes -> nothing to do */
99 if (re_list_head == NULL) {
103 /* match each line against all regexes */
104 linebuffer = store_get(32767);
105 while (fgets(CS linebuffer, 32767, mbox_file) != NULL) {
106 if ( (mime_stream != NULL) && (mime_current_boundary != NULL) ) {
108 if (Ustrncmp(linebuffer,"--",2) == 0) {
109 if (Ustrncmp((linebuffer+2),mime_current_boundary,Ustrlen(mime_current_boundary)) == 0)
114 re_list_item = re_list_head;
116 /* try matcher on the line */
117 if (pcre_exec(re_list_item->re, NULL, CS linebuffer,
118 (int)Ustrlen(linebuffer), 0, 0, NULL, 0) >= 0) {
119 Ustrncpy(regex_match_string_buffer, re_list_item->pcre_text, 1023);
120 regex_match_string = regex_match_string_buffer;
121 if (mime_stream == NULL)
124 clearerr(mime_stream);
125 fseek(mime_stream,f_pos,SEEK_SET);
129 re_list_item = re_list_item->next;
130 } while (re_list_item != NULL);
133 if (mime_stream == NULL)
136 clearerr(mime_stream);
137 fseek(mime_stream,f_pos,SEEK_SET);
145 int mime_regex(uschar **listptr) {
147 uschar *list = *listptr;
148 uschar *regex_string;
149 uschar regex_string_buffer[1024];
151 pcre_list *re_list_head = NULL;
152 pcre_list *re_list_item;
153 const char *pcre_error;
156 uschar *mime_subject = NULL;
157 int mime_subject_len = 0;
159 /* reset expansion variable */
160 regex_match_string = NULL;
162 /* precompile our regexes */
163 while ((regex_string = string_nextinlist(&list, &sep,
165 sizeof(regex_string_buffer))) != NULL) {
168 if ( (strcmpic(regex_string,US"false") == 0) ||
169 (Ustrcmp(regex_string,"0") == 0) ) {
170 /* explicitly no matching */
174 /* compile our regular expression */
175 re = pcre_compile( CS regex_string,
182 log_write(0, LOG_MAIN,
183 "regex acl condition warning - error in regex '%s': %s at offset %d, skipped.", regex_string, pcre_error, pcre_erroffset);
187 re_list_item = store_get(sizeof(pcre_list));
188 re_list_item->re = re;
189 re_list_item->pcre_text = string_copy(regex_string);
190 re_list_item->next = re_list_head;
191 re_list_head = re_list_item;
195 /* no regexes -> nothing to do */
196 if (re_list_head == NULL) {
200 /* check if the file is already decoded */
201 if (mime_decoded_filename == NULL) {
202 uschar *empty = US"";
203 /* no, decode it first */
205 if (mime_decoded_filename == NULL) {
206 /* decoding failed */
207 log_write(0, LOG_MAIN,
208 "mime_regex acl condition warning - could not decode MIME part to file.");
215 f = fopen(CS mime_decoded_filename, "r");
218 log_write(0, LOG_MAIN,
219 "mime_regex acl condition warning - can't open '%s' for reading.", mime_decoded_filename);
224 mime_subject = (uschar *)store_get(32767);
226 /* read max 32k chars from file */
227 mime_subject_len = fread(mime_subject, 1, 32766, f);
229 re_list_item = re_list_head;
231 /* try matcher on the mmapped file */
232 debug_printf("Matching '%s'\n", re_list_item->pcre_text);
233 if (pcre_exec(re_list_item->re, NULL, CS mime_subject,
234 mime_subject_len, 0, 0, NULL, 0) >= 0) {
235 Ustrncpy(regex_match_string_buffer, re_list_item->pcre_text, 1023);
236 regex_match_string = regex_match_string_buffer;
240 re_list_item = re_list_item->next;
241 } while (re_list_item != NULL);