X-Git-Url: https://git.exim.org/exim.git/blobdiff_plain/2339c66a83ad30ebc59a8bf065a11c11f4158e7c..a85c067ba6c6940512cf57ec213277a370d87e70:/src/src/regex.c

diff --git a/src/src/regex.c b/src/src/regex.c
index 94a867c65..eefba8ecf 100644
--- a/src/src/regex.c
+++ b/src/src/regex.c
@@ -2,8 +2,12 @@
 *     Exim - an Internet mail transport agent    *
 *************************************************/
 
-/* Copyright (c) Tom Kistner <tom@duncanthrax.net> 2003-???? */
-/* License: GPL */
+/*
+ * Copyright (c) The Exim Maintainers 2016 - 2022
+ * Copyright (c) Tom Kistner <tom@duncanthrax.net> 2003-2015
+ * License: GPL
+ * SPDX-License-Identifier: GPL-2.0-only
+ */
 
 /* Code for matching regular expressions against headers and body.
  Called from acl.c. */
@@ -15,9 +19,9 @@
 
 /* Structure to hold a list of Regular expressions */
 typedef struct pcre_list {
-  pcre *re;
-  uschar *pcre_text;
-  struct pcre_list *next;
+  const pcre2_code *	re;
+  uschar *		pcre_text;
+  struct pcre_list *	next;
 } pcre_list;
 
 uschar regex_match_string_buffer[1024];
@@ -25,221 +29,195 @@ uschar regex_match_string_buffer[1024];
 extern FILE *mime_stream;
 extern uschar *mime_current_boundary;
 
-int regex(uschar **listptr) {
-  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;
-  uschar *linebuffer;
-  long f_pos = 0;
-
-  /* reset expansion variable */
-  regex_match_string = NULL;
-
-  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;
-    };
 
+static pcre_list *
+compile(const uschar * list, BOOL cacheable)
+{
+int sep = 0;
+uschar * regex_string;
+pcre_list * re_list_head = NULL;
+pcre_list * ri;
+
+/* precompile our regexes */
+while ((regex_string = string_nextinlist(&list, &sep, NULL, 0)))
+  if (strcmpic(regex_string, US"false") != 0 && Ustrcmp(regex_string, "0") != 0)
+    {
     /* 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);
+    uschar * errstr;
+    const pcre2_code * re = regex_compile(regex_string,
+      cacheable ? MCS_CACHEABLE : MCS_NOFLAGS, &errstr, pcre_gen_cmp_ctx);
+
+    if (!re)
+      {
+      log_write(0, LOG_MAIN, "regex acl condition warning - %s, skipped", errstr);
       continue;
+      }
+
+    ri = store_get(sizeof(pcre_list), GET_UNTAINTED);
+    ri->re = re;
+    ri->pcre_text = regex_string;
+    ri->next = re_list_head;
+    re_list_head = ri;
     }
-    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;
-  };
-
-  /* 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)
-          (void)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 == NULL)
-    (void)fclose(mbox_file);
-  else {
-    clearerr(mime_stream);
-    fseek(mime_stream,f_pos,SEEK_SET);
-  };
+return re_list_head;
+}
 
-  /* no matches ... */
-  return FAIL;
+static int
+matcher(pcre_list * re_list_head, uschar * linebuffer, int len)
+{
+pcre2_match_data * md = pcre2_match_data_create(REGEX_VARS + 1, pcre_gen_ctx);
+
+for (pcre_list * ri = re_list_head; ri; ri = ri->next)
+  {
+  int n;
+
+  /* try matcher on the line */
+  if ((n = pcre2_match(ri->re, (PCRE2_SPTR)linebuffer, len, 0, 0, md, pcre_gen_mtc_ctx)) > 0)
+    {
+    Ustrncpy(regex_match_string_buffer, ri->pcre_text,
+	      sizeof(regex_match_string_buffer)-1);
+    regex_match_string = regex_match_string_buffer;
+
+    for (int nn = 1; nn < n; nn++)
+      {
+      PCRE2_SIZE * ovec = pcre2_get_ovector_pointer(md);
+      int off = nn * 2;
+      int len = ovec[off + 1] - ovec[off];
+      regex_vars[nn-1] = string_copyn(linebuffer + ovec[off], len);
+      }
+
+    return OK;
+    }
+  }
+/* pcre2_match_data_free(md);	gen ctx needs no free */
+return FAIL;
 }
 
 
-int mime_regex(uschar **listptr) {
-  int sep = 0;
-  uschar *list = *listptr;
-  uschar *regex_string;
-  uschar regex_string_buffer[1024];
-  pcre *re;
-  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;
-
-  /* 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;
-    };
+/* reset expansion variables */
+void
+regex_vars_clear(void)
+{
+regex_match_string = NULL;
+for (int i = 0; i < REGEX_VARS; i++) regex_vars[i] = NULL;
+}
 
-    /* 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;
+
+
+int
+regex(const uschar ** listptr, BOOL cacheable)
+{
+unsigned long mbox_size;
+FILE * mbox_file;
+pcre_list * re_list_head;
+uschar * linebuffer;
+long f_pos = 0;
+int ret = FAIL;
+
+regex_vars_clear();
+
+if (!mime_stream)				/* We are in the DATA ACL */
+  {
+  if (!(mbox_file = spool_mbox(&mbox_size, NULL, NULL)))
+    {						/* error while spooling */
+    log_write(0, LOG_MAIN|LOG_PANIC,
+	   "regex acl condition: error while creating mbox spool file");
+    return DEFER;
+    }
+  }
+else
+  {
+  if ((f_pos = ftell(mime_stream)) < 0)
+    {
+    log_write(0, LOG_MAIN|LOG_PANIC,
+	   "regex acl condition: mime_stream: %s", strerror(errno));
+    return DEFER;
+    }
+  mbox_file = mime_stream;
+  }
+
+/* precompile our regexes */
+if (!(re_list_head = compile(*listptr, cacheable)))
+  return FAIL;			/* no regexes -> nothing to do */
+
+/* match each line against all regexes */
+linebuffer = store_get(32767, GET_TAINTED);
+while (fgets(CS linebuffer, 32767, mbox_file))
+  {
+  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)
+  (void)fclose(mbox_file);
+else
+  {
+  clearerr(mime_stream);
+  if (fseek(mime_stream, f_pos, SEEK_SET) == -1)
+    {
+    log_write(0, LOG_MAIN|LOG_PANIC,
+	   "regex acl condition: mime_stream: %s", strerror(errno));
+    clearerr(mime_stream);
     }
-    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;
-  };
-
-  /* check if the file is already decoded */
-  if (mime_decoded_filename == NULL) {
-    uschar *empty = US"";
-    /* no, decode it first */
-    mime_decode(&empty);
-    if (mime_decoded_filename == NULL) {
-      /* decoding failed */
-      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, "rb");
-  if (f == NULL) {
-    /* open failed */
+  }
+
+return ret;
+}
+
+
+int
+mime_regex(const uschar **listptr, BOOL cacheable)
+{
+pcre_list * re_list_head = NULL;
+FILE * f;
+uschar * mime_subject = NULL;
+int mime_subject_len = 0;
+int ret;
+
+regex_vars_clear();
+
+/* precompile our regexes */
+if (!(re_list_head = compile(*listptr, cacheable)))
+  return FAIL;			/* no regexes -> nothing to do */
+
+/* check if the file is already decoded */
+if (!mime_decoded_filename)
+  {				/* no, decode it first */
+  const uschar *empty = US"";
+  mime_decode(&empty);
+  if (!mime_decoded_filename)
+    {				/* decoding failed */
     log_write(0, LOG_MAIN,
-         "mime_regex acl condition warning - can't open '%s' for reading.", mime_decoded_filename);
+       "mime_regex acl condition warning - could not decode MIME part to file");
     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;
-      (void)fclose(f);
-      return OK;
-    };
-    re_list_item = re_list_item->next;
-  } while (re_list_item != NULL);
-
-  (void)fclose(f);
-
-  /* no matches ... */
-  return FAIL;
+    }
+  }
+
+/* open file */
+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);
+  return DEFER;
+  }
+
+/* get 32k memory, tainted */
+mime_subject = store_get(32767, GET_TAINTED);
+
+mime_subject_len = fread(mime_subject, 1, 32766, f);
+
+ret = matcher(re_list_head, mime_subject, mime_subject_len);
+(void)fclose(f);
+return ret;
 }
 
 #endif /* WITH_CONTENT_SCAN */