Address qualification: ensure maformed addresses are logged. Bug 3061

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

diff --git a/src/src/rewrite.c b/src/src/rewrite.c
index ab5f78020ecdff3dd35947ac558f02daff3f9694..70bd79e262aa38589f5032957ef8703fb10a3501 100644 (file)
--- a/src/src/rewrite.c
+++ b/src/src/rewrite.c
@@ -2,8 +2,10 @@
 *     Exim - an Internet mail transport agent    *
 *************************************************/
 
+/* Copyright (c) The Exim Maintainers 2021 - 2023 */
 /* Copyright (c) University of Cambridge 1995 - 2018 */
 /* See the file NOTICE for conditions of use and distribution. */
+/* SPDX-License-Identifier: GPL-2.0-or-later */
 
 /* Functions concerned with rewriting headers */
 
@@ -117,9 +119,9 @@ for (rewrite_rule * rule = rewrite_rules;
   {
   int start, end, pdomain;
   int count = 0;
-  uschar *save_localpart;
-  const uschar *save_domain;
-  uschar *error, *new;
+  const uschar * save_localpart;
+  const uschar * save_domain;
+  uschar * error, * new;
   const uschar * newparsed;
 
   /* Come back here for a repeat after a successful rewrite. We do this
@@ -135,7 +137,8 @@ for (rewrite_rule * rule = rewrite_rules;
 
   if (flag & rewrite_smtp)
     {
-    uschar *key = expand_string(rule->key);
+    BOOL textonly_re;
+    const uschar * key = expand_string_2(rule->key, &textonly_re);
     if (!key)
       {
       if (!f.expand_string_forcedfail)
@@ -143,7 +146,8 @@ for (rewrite_rule * rule = rewrite_rules;
           "checking for SMTP rewriting: %s", rule->key, expand_string_message);
       continue;
       }
-    if (match_check_string(subject, key, 0, TRUE, FALSE, FALSE, NULL) != OK)
+    if (match_check_string(subject, key, 0,
+      textonly_re ? MCS_CACHEABLE | MCS_PARTIAL : MCS_PARTIAL, NULL) != OK)
       continue;
     new = expand_string(rule->replacement);
     }
@@ -446,14 +450,14 @@ rewrite_one_header(header_line *h, int flag,
   rewrite_rule *rewrite_rules, int existflags, BOOL replace)
 {
 int lastnewline = 0;
-header_line *newh = NULL;
+header_line * newh = NULL;
 rmark function_reset_point = store_mark();
-uschar *s = Ustrchr(h->text, ':') + 1;
+uschar * s = Ustrchr(h->text, ':') + 1;
 
 while (isspace(*s)) s++;
 
-DEBUG(D_rewrite)
-  debug_printf("rewrite_one_header: type=%c:\n  %s", h->type, h->text);
+DEBUG(D_rewrite)       /* The header text includes the trailing newline */
+  debug_printf_indent("rewrite_one_header: type=%c:\n  %s", h->type, h->text);
 
 f.parse_allow_group = TRUE;     /* Allow group syntax */
 
@@ -466,30 +470,49 @@ We want to avoid keeping store for any intermediate versions. */
 
 while (*s)
   {
-  uschar *sprev;
-  uschar *ss = parse_find_address_end(s, FALSE);
-  uschar *recipient, *new, *errmess;
+  uschar * sprev = s;
+  uschar * ss = parse_find_address_end(s, FALSE), * ss1 = ss;
+  uschar * recipient, * new;
   rmark loop_reset_point = store_mark();
+  uschar * errmess = NULL;
   BOOL changed = FALSE;
-  int terminator = *ss;
+  uschar terminator = *ss;
   int start, end, domain;
 
+  /* If we hit the end of the header, trim trailing newline and whitespace */
+
+  if (!terminator)
+    {
+    while (ss1 > s && isspace(ss1[-1])) ss1--;
+    terminator = *ss1;
+    }
+
   /* Temporarily terminate the string at this point, and extract the
   operative address within. Then put back the terminator and prepare for
   the next address, saving the start of the old one. */
 
-  *ss = 0;
+  *ss1 = '\0';
   recipient = parse_extract_address(s, &errmess, &start, &end, &domain, FALSE);
-  *ss = terminator;
-  sprev = s;
-  s = ss + (terminator ? 1 :0);
+  *ss1 = terminator;
+  s = ss + (*ss ? 1 : 0);
   while (isspace(*s)) s++;
 
   /* There isn't much we can do for syntactic disasters at this stage.
-  Pro tem (possibly for ever) ignore them. */
+  Pro tem (possibly for ever) ignore them.
+  If we got nothing, then there was any sort of error: non-parsable address,
+  empty address, overlong addres. Sometimes the result matters, sometimes not.
+  It seems this function is called for *any* header we see. */
 
   if (!recipient)
     {
+    /* Log unparesable addresses in the header. Slightly ugly because a
+    null output from the extract can also result from a header without an
+    address, "To: undisclosed recpients:;" being the classic case. Ignore
+    this one and carry on. */
+
+    if (Ustrcmp(errmess, "empty address") != 0)
+      log_write(0, LOG_MAIN, "qualify/rewrite: %s", errmess);
+
     loop_reset_point = store_reset(loop_reset_point);
     continue;
     }
@@ -571,7 +594,8 @@ while (*s)
   point, because we may have a rewritten line from a previous time round the
   loop. */
 
-  if (!changed) loop_reset_point = store_reset(loop_reset_point);
+  if (!changed)
+    loop_reset_point = store_reset(loop_reset_point);
 
   /* If the address has changed, create a new header containing the
   rewritten address. We do not need to set the chain pointers at this
@@ -589,7 +613,7 @@ while (*s)
     int oldlen = end - start;
 
     header_line * prev = newh ? newh : h;
-    uschar * newt = store_get_perm(prev->slen - oldlen + newlen + 4, TRUE);
+    uschar * newt = store_get_perm(prev->slen - oldlen + newlen + 4, GET_TAINTED);
     uschar * newtstart = newt;
 
     int type = prev->type;
@@ -653,7 +677,7 @@ while (*s)
 
     store_reset(function_reset_point);
     function_reset_point = store_mark();
-    newh = store_get(sizeof(header_line), FALSE);
+    newh = store_get(sizeof(header_line), GET_UNTAINTED);
     newh->type = type;
     newh->slen = slen;
     newh->text = string_copyn(newtstart, slen);