X-Git-Url: https://git.exim.org/exim.git/blobdiff_plain/384152a6b0cce686255894a502e4b88743abd356..3a2ac12bcabf06f73ef372dfa23e6edead237aa7:/src/src/regex.c

diff --git a/src/src/regex.c b/src/src/regex.c
index c811f439f..93422fadc 100644
--- a/src/src/regex.c
+++ b/src/src/regex.c
@@ -1,10 +1,8 @@
-/* $Cambridge: exim/src/src/regex.c,v 1.5 2005/05/24 08:15:02 tom Exp $ */
-
 /*************************************************
 *     Exim - an Internet mail transport agent    *
 *************************************************/
 
-/* Copyright (c) Tom Kistner <tom@duncanthrax.net> 2003-???? */
+/* Copyright (c) Tom Kistner <tom@duncanthrax.net> 2003-2015 */
 /* License: GPL */
 
 /* Code for matching regular expressions against headers and body.
@@ -27,177 +25,150 @@ uschar regex_match_string_buffer[1024];
 extern FILE *mime_stream;
 extern uschar *mime_current_boundary;
 
-int regex(uschar **listptr) {
+static pcre_list *
+compile(const uschar * list)
+{
   int sep = 0;
-  uschar *list = *listptr;
   uschar *regex_string;
   uschar regex_string_buffer[1024];
-  unsigned long mbox_size;
-  FILE *mbox_file;
-  pcre *re;
-  pcre_list *re_list_head = NULL;
-  pcre_list *re_list_item;
   const char *pcre_error;
   int pcre_erroffset;
+  pcre_list *re_list_head = NULL;
+  pcre_list *ri;
+
+  /* precompile our regexes */
+  while ((regex_string = string_nextinlist(&list, &sep,
+                                           regex_string_buffer,
+                                           sizeof(regex_string_buffer))) != NULL) {
+    pcre *re;
+
+    /* parse option */
+    if ( (strcmpic(regex_string,US"false") == 0) ||
+         (Ustrcmp(regex_string,"0") == 0) )
+      continue;				/* explicitly no matching */
+
+    /* compile our regular expression */
+    if (!(re = pcre_compile( CS regex_string,
+                       0, &pcre_error, &pcre_erroffset, NULL ))) {
+      log_write(0, LOG_MAIN,
+           "regex acl condition warning - error in regex '%s': %s at offset %d, skipped.",
+	   regex_string, pcre_error, pcre_erroffset);
+      continue;
+    }
+
+    ri = store_get(sizeof(pcre_list));
+    ri->re = re;
+    ri->pcre_text = string_copy(regex_string);
+    ri->next = re_list_head;
+    re_list_head = ri;
+  }
+  return re_list_head;
+}
+
+static int
+matcher(pcre_list * re_list_head, uschar * linebuffer, int len)
+{
+  pcre_list * ri;
+
+  for(ri = re_list_head; ri; ri = ri->next)
+    {
+    int ovec[3*(REGEX_VARS+1)];
+    int n, nn;
+
+    /* try matcher on the line */
+    n = pcre_exec(ri->re, NULL,
+	 CS linebuffer, len, 0, 0,
+	 ovec, nelem(ovec));
+    if (n > 0)
+      {
+      Ustrncpy(regex_match_string_buffer, ri->pcre_text, 1023);
+      regex_match_string = regex_match_string_buffer;
+
+      for (nn = 1; nn < n; nn++)
+	regex_vars[nn-1] =
+	  string_copyn(linebuffer + ovec[nn*2], ovec[nn*2+1] - ovec[nn*2]);
+
+      return OK;
+      }
+    }
+  return FAIL;
+}
+
+int
+regex(const uschar **listptr)
+{
+  unsigned long mbox_size;
+  FILE *mbox_file;
+  pcre_list *re_list_head;
   uschar *linebuffer;
   long f_pos = 0;
+  int ret = FAIL;
 
   /* reset expansion variable */
   regex_match_string = NULL;
 
-  if (mime_stream == NULL) {
-    /* We are in the DATA ACL */
-    mbox_file = spool_mbox(&mbox_size);
-    if (mbox_file == NULL) {
-      /* error while spooling */
+  if (mime_stream == NULL) {			/* We are in the DATA ACL */
+    mbox_file = spool_mbox(&mbox_size, NULL);
+    if (mbox_file == NULL) {			/* error while spooling */
       log_write(0, LOG_MAIN|LOG_PANIC,
              "regex acl condition: error while creating mbox spool file");
       return DEFER;
-    };
+    }
   }
   else {
     f_pos = ftell(mime_stream);
     mbox_file = mime_stream;
-  };
+  }
 
   /* precompile our regexes */
-  while ((regex_string = string_nextinlist(&list, &sep,
-                                           regex_string_buffer,
-                                           sizeof(regex_string_buffer))) != NULL) {
-
-    /* parse option */
-    if ( (strcmpic(regex_string,US"false") == 0) ||
-         (Ustrcmp(regex_string,"0") == 0) ) {
-      /* explicitly no matching */
-      continue;
-    };
-
-    /* compile our regular expression */
-    re = pcre_compile( CS regex_string,
-                       0,
-                       &pcre_error,
-                       &pcre_erroffset,
-                       NULL );
-
-    if (re == NULL) {
-      log_write(0, LOG_MAIN,
-           "regex acl condition warning - error in regex '%s': %s at offset %d, skipped.", regex_string, pcre_error, pcre_erroffset);
-      continue;
-    }
-    else {
-      re_list_item = store_get(sizeof(pcre_list));
-      re_list_item->re = re;
-      re_list_item->pcre_text = string_copy(regex_string);
-      re_list_item->next = re_list_head;
-      re_list_head = re_list_item;
-    };
-  };
-
-  /* no regexes -> nothing to do */
-  if (re_list_head == NULL) {
-    return FAIL;
-  };
+  if (!(re_list_head = compile(*listptr)))
+    return FAIL;			/* no regexes -> nothing to do */
 
   /* match each line against all regexes */
   linebuffer = store_get(32767);
   while (fgets(CS linebuffer, 32767, mbox_file) != NULL) {
-    if ( (mime_stream != NULL) && (mime_current_boundary != NULL) ) {
-      /* check boundary */
-      if (Ustrncmp(linebuffer,"--",2) == 0) {
-        if (Ustrncmp((linebuffer+2),mime_current_boundary,Ustrlen(mime_current_boundary)) == 0)
-          /* found boundary */
-          break;
-      };
-    };
-    re_list_item = re_list_head;
-    do {
-      /* try matcher on the line */
-      if (pcre_exec(re_list_item->re, NULL, CS linebuffer,
-      (int)Ustrlen(linebuffer), 0, 0, NULL, 0) >= 0) {
-        Ustrncpy(regex_match_string_buffer, re_list_item->pcre_text, 1023);
-        regex_match_string = regex_match_string_buffer;
-        if (mime_stream == NULL)
-          fclose(mbox_file);
-        else {
-          clearerr(mime_stream);
-          fseek(mime_stream,f_pos,SEEK_SET);
-        };
-        return OK;
-      };
-      re_list_item = re_list_item->next;
-    } while (re_list_item != NULL);
-  };
 
+    if (  mime_stream && mime_current_boundary		/* check boundary */
+       && Ustrncmp(linebuffer,"--",2) == 0
+       && Ustrncmp((linebuffer+2),mime_current_boundary,Ustrlen(mime_current_boundary)) == 0)
+	break;						/* found boundary */
+
+    if ((ret = matcher(re_list_head, linebuffer, (int)Ustrlen(linebuffer))) == OK)
+      goto done;
+  }
+  /* no matches ... */
+
+done:
   if (mime_stream == NULL)
-    fclose(mbox_file);
+    (void)fclose(mbox_file);
   else {
     clearerr(mime_stream);
     fseek(mime_stream,f_pos,SEEK_SET);
   };
 
-  /* no matches ... */
-  return FAIL;
+  return ret;
 }
 
 
-int mime_regex(uschar **listptr) {
-  int sep = 0;
-  uschar *list = *listptr;
-  uschar *regex_string;
-  uschar regex_string_buffer[1024];
-  pcre *re;
+int
+mime_regex(const uschar **listptr)
+{
   pcre_list *re_list_head = NULL;
-  pcre_list *re_list_item;
-  const char *pcre_error;
-  int pcre_erroffset;
   FILE *f;
   uschar *mime_subject = NULL;
   int mime_subject_len = 0;
+  int ret;
 
   /* reset expansion variable */
   regex_match_string = NULL;
 
   /* precompile our regexes */
-  while ((regex_string = string_nextinlist(&list, &sep,
-                                           regex_string_buffer,
-                                           sizeof(regex_string_buffer))) != NULL) {
-
-    /* parse option */
-    if ( (strcmpic(regex_string,US"false") == 0) ||
-         (Ustrcmp(regex_string,"0") == 0) ) {
-      /* explicitly no matching */
-      continue;
-    };
-
-    /* compile our regular expression */
-    re = pcre_compile( CS regex_string,
-                       0,
-                       &pcre_error,
-                       &pcre_erroffset,
-                       NULL );
-
-    if (re == NULL) {
-      log_write(0, LOG_MAIN,
-           "regex acl condition warning - error in regex '%s': %s at offset %d, skipped.", regex_string, pcre_error, pcre_erroffset);
-      continue;
-    }
-    else {
-      re_list_item = store_get(sizeof(pcre_list));
-      re_list_item->re = re;
-      re_list_item->pcre_text = string_copy(regex_string);
-      re_list_item->next = re_list_head;
-      re_list_head = re_list_item;
-    };
-  };
-
-  /* no regexes -> nothing to do */
-  if (re_list_head == NULL) {
-    return FAIL;
-  };
+  if (!(re_list_head = compile(*listptr)))
+    return FAIL;			/* no regexes -> nothing to do */
 
   /* check if the file is already decoded */
   if (mime_decoded_filename == NULL) {
-    uschar *empty = US"";
+    const uschar *empty = US"";
     /* no, decode it first */
     mime_decode(&empty);
     if (mime_decoded_filename == NULL) {
@@ -205,43 +176,25 @@ int mime_regex(uschar **listptr) {
       log_write(0, LOG_MAIN,
            "mime_regex acl condition warning - could not decode MIME part to file.");
       return DEFER;
-    };
-  };
-
+    }
+  }
 
   /* open file */
-  f = fopen(CS mime_decoded_filename, "r");
-  if (f == NULL) {
-    /* open failed */
+  if (!(f = fopen(CS mime_decoded_filename, "rb"))) {
     log_write(0, LOG_MAIN,
-         "mime_regex acl condition warning - can't open '%s' for reading.", mime_decoded_filename);
+         "mime_regex acl condition warning - can't open '%s' for reading.",
+	 mime_decoded_filename);
     return DEFER;
-  };
+  }
 
   /* get 32k memory */
   mime_subject = (uschar *)store_get(32767);
 
-  /* read max 32k chars from file */
   mime_subject_len = fread(mime_subject, 1, 32766, f);
 
-  re_list_item = re_list_head;
-  do {
-    /* try matcher on the mmapped file */
-    debug_printf("Matching '%s'\n", re_list_item->pcre_text);
-    if (pcre_exec(re_list_item->re, NULL, CS mime_subject,
-                  mime_subject_len, 0, 0, NULL, 0) >= 0) {
-      Ustrncpy(regex_match_string_buffer, re_list_item->pcre_text, 1023);
-      regex_match_string = regex_match_string_buffer;
-      fclose(f);
-      return OK;
-    };
-    re_list_item = re_list_item->next;
-  } while (re_list_item != NULL);
-
-  fclose(f);
-
-  /* no matches ... */
-  return FAIL;
+  ret = matcher(re_list_head, mime_subject, mime_subject_len);
+  (void)fclose(f);
+  return ret;
 }
 
-#endif
+#endif /* WITH_CONTENT_SCAN */