src/src/regex.c

   1 /*************************************************
   2 *     Exim - an Internet mail transport agent    *
   3 *************************************************/
   4
   5 /* Copyright (c) Tom Kistner <tom@duncanthrax.net> 2003-???? */
   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
  28 int regex(uschar **listptr) {
  29   int sep = 0;
  30   uschar *list = *listptr;
  31   uschar *regex_string;
  32   uschar regex_string_buffer[1024];
  33   unsigned long mbox_size;
  34   FILE *mbox_file;
  35   pcre *re;
  36   pcre_list *re_list_head = NULL;
  37   pcre_list *re_list_item;
  38   const char *pcre_error;
  39   int pcre_erroffset;
  40   uschar *linebuffer;
  41   long f_pos = 0;
  42
  43   /* reset expansion variable */
  44   regex_match_string = NULL;
  45
  46   if (mime_stream == NULL) {
  47     /* We are in the DATA ACL */
  48     mbox_file = spool_mbox(&mbox_size, NULL);
  49     if (mbox_file == NULL) {
  50       /* error while spooling */
  51       log_write(0, LOG_MAIN|LOG_PANIC,
  52              "regex acl condition: error while creating mbox spool file");
  53       return DEFER;
  54     };
  55   }
  56   else {
  57     f_pos = ftell(mime_stream);
  58     mbox_file = mime_stream;
  59   };
  60
  61   /* precompile our regexes */
  62   while ((regex_string = string_nextinlist(&list, &sep,
  63                                            regex_string_buffer,
  64                                            sizeof(regex_string_buffer))) != NULL) {
  65
  66     /* parse option */
  67     if ( (strcmpic(regex_string,US"false") == 0) ||
  68          (Ustrcmp(regex_string,"0") == 0) ) {
  69       /* explicitly no matching */
  70       continue;
  71     };
  72
  73     /* compile our regular expression */
  74     re = pcre_compile( CS regex_string,
  75                        0,
  76                        &pcre_error,
  77                        &pcre_erroffset,
  78                        NULL );
  79
  80     if (re == NULL) {
  81       log_write(0, LOG_MAIN,
  82            "regex acl condition warning - error in regex '%s': %s at offset %d, skipped.", regex_string, pcre_error, pcre_erroffset);
  83       continue;
  84     }
  85     else {
  86       re_list_item = store_get(sizeof(pcre_list));
  87       re_list_item->re = re;
  88       re_list_item->pcre_text = string_copy(regex_string);
  89       re_list_item->next = re_list_head;
  90       re_list_head = re_list_item;
  91     };
  92   };
  93
  94   /* no regexes -> nothing to do */
  95   if (re_list_head == NULL) {
  96     return FAIL;
  97   };
  98
  99   /* match each line against all regexes */
 100   linebuffer = store_get(32767);
 101   while (fgets(CS linebuffer, 32767, mbox_file) != NULL) {
 102     if ( (mime_stream != NULL) && (mime_current_boundary != NULL) ) {
 103       /* check boundary */
 104       if (Ustrncmp(linebuffer,"--",2) == 0) {
 105         if (Ustrncmp((linebuffer+2),mime_current_boundary,Ustrlen(mime_current_boundary)) == 0)
 106           /* found boundary */
 107           break;
 108       };
 109     };
 110     re_list_item = re_list_head;
 111     do {
 112       /* try matcher on the line */
 113       if (pcre_exec(re_list_item->re, NULL, CS linebuffer,
 114       (int)Ustrlen(linebuffer), 0, 0, NULL, 0) >= 0) {
 115         Ustrncpy(regex_match_string_buffer, re_list_item->pcre_text, 1023);
 116         regex_match_string = regex_match_string_buffer;
 117         if (mime_stream == NULL)
 118           (void)fclose(mbox_file);
 119         else {
 120           clearerr(mime_stream);
 121           fseek(mime_stream,f_pos,SEEK_SET);
 122         };
 123         return OK;
 124       };
 125       re_list_item = re_list_item->next;
 126     } while (re_list_item != NULL);
 127   };
 128
 129   if (mime_stream == NULL)
 130     (void)fclose(mbox_file);
 131   else {
 132     clearerr(mime_stream);
 133     fseek(mime_stream,f_pos,SEEK_SET);
 134   };
 135
 136   /* no matches ... */
 137   return FAIL;
 138 }
 139
 140
 141 int mime_regex(uschar **listptr) {
 142   int sep = 0;
 143   uschar *list = *listptr;
 144   uschar *regex_string;
 145   uschar regex_string_buffer[1024];
 146   pcre *re;
 147   pcre_list *re_list_head = NULL;
 148   pcre_list *re_list_item;
 149   const char *pcre_error;
 150   int pcre_erroffset;
 151   FILE *f;
 152   uschar *mime_subject = NULL;
 153   int mime_subject_len = 0;
 154
 155   /* reset expansion variable */
 156   regex_match_string = NULL;
 157
 158   /* precompile our regexes */
 159   while ((regex_string = string_nextinlist(&list, &sep,
 160                                            regex_string_buffer,
 161                                            sizeof(regex_string_buffer))) != NULL) {
 162
 163     /* parse option */
 164     if ( (strcmpic(regex_string,US"false") == 0) ||
 165          (Ustrcmp(regex_string,"0") == 0) ) {
 166       /* explicitly no matching */
 167       continue;
 168     };
 169
 170     /* compile our regular expression */
 171     re = pcre_compile( CS regex_string,
 172                        0,
 173                        &pcre_error,
 174                        &pcre_erroffset,
 175                        NULL );
 176
 177     if (re == NULL) {
 178       log_write(0, LOG_MAIN,
 179            "regex acl condition warning - error in regex '%s': %s at offset %d, skipped.", regex_string, pcre_error, pcre_erroffset);
 180       continue;
 181     }
 182     else {
 183       re_list_item = store_get(sizeof(pcre_list));
 184       re_list_item->re = re;
 185       re_list_item->pcre_text = string_copy(regex_string);
 186       re_list_item->next = re_list_head;
 187       re_list_head = re_list_item;
 188     };
 189   };
 190
 191   /* no regexes -> nothing to do */
 192   if (re_list_head == NULL) {
 193     return FAIL;
 194   };
 195
 196   /* check if the file is already decoded */
 197   if (mime_decoded_filename == NULL) {
 198     uschar *empty = US"";
 199     /* no, decode it first */
 200     mime_decode(&empty);
 201     if (mime_decoded_filename == NULL) {
 202       /* decoding failed */
 203       log_write(0, LOG_MAIN,
 204            "mime_regex acl condition warning - could not decode MIME part to file.");
 205       return DEFER;
 206     };
 207   };
 208
 209
 210   /* open file */
 211   f = fopen(CS mime_decoded_filename, "rb");
 212   if (f == NULL) {
 213     /* open failed */
 214     log_write(0, LOG_MAIN,
 215          "mime_regex acl condition warning - can't open '%s' for reading.", mime_decoded_filename);
 216     return DEFER;
 217   };
 218
 219   /* get 32k memory */
 220   mime_subject = (uschar *)store_get(32767);
 221
 222   /* read max 32k chars from file */
 223   mime_subject_len = fread(mime_subject, 1, 32766, f);
 224
 225   re_list_item = re_list_head;
 226   do {
 227     /* try matcher on the mmapped file */
 228     debug_printf("Matching '%s'\n", re_list_item->pcre_text);
 229     if (pcre_exec(re_list_item->re, NULL, CS mime_subject,
 230                   mime_subject_len, 0, 0, NULL, 0) >= 0) {
 231       Ustrncpy(regex_match_string_buffer, re_list_item->pcre_text, 1023);
 232       regex_match_string = regex_match_string_buffer;
 233       (void)fclose(f);
 234       return OK;
 235     };
 236     re_list_item = re_list_item->next;
 237   } while (re_list_item != NULL);
 238
 239   (void)fclose(f);
 240
 241   /* no matches ... */
 242   return FAIL;
 243 }
 244
 245 #endif