src/src/regex.c

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