perl version oddity

[exim.git] / src / src / regex.c

diff --git a/src/src/regex.c b/src/src/regex.c
index 757243e7f19207b07bb446ea67ff6c6f1c8a6fae..285ab8dbd2582a3416f91a8cf0a0337a295b7be4 100644 (file)
--- a/src/src/regex.c
+++ b/src/src/regex.c
@@ -3,7 +3,7 @@
 *************************************************/
 
 /*
- * Copyright (c) The Exim Maintainers 2016 - 2022
+ * Copyright (c) The Exim Maintainers 2016 - 2023
  * Copyright (c) Tom Kistner <tom@duncanthrax.net> 2003-2015
  * License: GPL
  * SPDX-License-Identifier: GPL-2.0-or-later
@@ -24,19 +24,16 @@ typedef struct pcre_list {
   struct pcre_list *   next;
 } pcre_list;
 
-uschar regex_match_string_buffer[1024];
-
 extern FILE *mime_stream;
 extern uschar *mime_current_boundary;
 
 
 static pcre_list *
-compile(const uschar * list, BOOL cacheable)
+compile(const uschar * list, BOOL cacheable, int * cntp)
 {
-int sep = 0;
+int sep = 0, cnt = 0;
 uschar * regex_string;
-pcre_list * re_list_head = NULL;
-pcre_list * ri;
+pcre_list * re_list_head = NULL, * ri;
 
 /* precompile our regexes */
 while ((regex_string = string_nextinlist(&list, &sep, NULL, 0)))
@@ -58,10 +55,19 @@ while ((regex_string = string_nextinlist(&list, &sep, NULL, 0)))
     ri->pcre_text = regex_string;
     ri->next = re_list_head;
     re_list_head = ri;
+    cnt++;
     }
+if (cntp) *cntp = cnt;
 return re_list_head;
 }
 
+
+/* Check list of REs against buffer, returning OK for (first) match,
+else FAIL.  On match return allocated result strings in regex_vars[]. 
+
+We use the perm-pool for that, so that our caller can release
+other allocations.
+*/
 static int
 matcher(pcre_list * re_list_head, uschar * linebuffer, int len)
 {
@@ -74,9 +80,10 @@ for (pcre_list * ri = re_list_head; ri; ri = ri->next)
   /* 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;
+    int save_pool = store_pool;
+    store_pool = POOL_PERM;
+
+    regex_match_string = string_copy(ri->pcre_text);
 
     for (int nn = 1; nn < n; nn++)
       {
@@ -86,6 +93,7 @@ for (pcre_list * ri = re_list_head; ri; ri = ri->next)
       regex_vars[nn-1] = string_copyn(linebuffer + ovec[off], len);
       }
 
+    store_pool = save_pool;
     return OK;
     }
   }
@@ -110,9 +118,8 @@ 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;
+int ret = FAIL, cnt, lcount = REGEX_LOOPCOUNT_STORE_RESET;
 
 regex_vars_clear();
 
@@ -136,26 +143,33 @@ else
   mbox_file = mime_stream;
   }
 
-/* precompile our regexes */
-if (!(re_list_head = compile(*listptr, cacheable)))
-  return FAIL;                 /* no regexes -> nothing to do */
+  /* precompile our regexes */
+  if ((re_list_head = compile(*listptr, cacheable, &cnt)))
+    {
+    rmark reset_point = store_mark();
 
-/* 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 ... */
+    /* match each line against all regexes */
+    while (fgets(CS big_buffer, big_buffer_size, mbox_file))
+      {
+      if (  mime_stream && mime_current_boundary               /* check boundary */
+        && Ustrncmp(big_buffer, "--", 2) == 0
+        && Ustrncmp((big_buffer+2), mime_current_boundary,
+                     Ustrlen(mime_current_boundary)) == 0)
+       break;                                          /* found boundary */
+
+      if ((ret = matcher(re_list_head, big_buffer, (int)Ustrlen(big_buffer))) == OK)
+       break;
+
+      if ((lcount -= cnt) <= 0)
+       {
+       store_reset(reset_point); reset_point = store_mark();
+       lcount = REGEX_LOOPCOUNT_STORE_RESET;
+       }
+      }
+
+    store_reset(reset_point);
+    }
 
-done:
 if (!mime_stream)
   (void)fclose(mbox_file);
 else
@@ -180,14 +194,11 @@ pcre_list * re_list_head = NULL;
 FILE * f;
 uschar * mime_subject = NULL;
 int mime_subject_len = 0;
-int ret;
+int ret = FAIL;
+rmark reset_point;
 
 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 */
@@ -210,12 +221,20 @@ if (!(f = fopen(CS mime_decoded_filename, "rb")))
   return DEFER;
   }
 
-/* get 32k memory, tainted */
-mime_subject = store_get(32767, GET_TAINTED);
+reset_point = store_mark();
+  {
+  /* precompile our regexes */
+  if ((re_list_head = compile(*listptr, cacheable, NULL)))
+    {
+    /* get 32k memory, tainted */
+    mime_subject = store_get(32767, GET_TAINTED);
 
-mime_subject_len = fread(mime_subject, 1, 32766, f);
+    mime_subject_len = fread(mime_subject, 1, 32766, f);
 
-ret = matcher(re_list_head, mime_subject, mime_subject_len);
+    ret = matcher(re_list_head, mime_subject, mime_subject_len);
+    }
+  }
+store_reset(reset_point);
 (void)fclose(f);
 return ret;
 }