For the benefit of those using Exim with Cyrus ([PATCH] New feature: remove ASCII NUL from message data)

[Phil Brutsche]

This is a message I sent to another list earlier today with a patch for 
Exim to strip NUL characters from messages as they are received.

It will probably be of interest to some people.

Phil Brutsche wrote:
> Background:
> 
> Some email clients are quite persistent in putting binary "junk" in 
> message bodies; this junk manifests itself as the ASCII NUL character. 
> It is seen (for me anyway) immediately following base64 data; the email 
> client is almost always Microsoft Outlook in either Corporate mode or 
> Internet mode, although some other email clients have been seen to cause 
> this problem.
> 
> Cyrus, which we use for POP3 and IMAP (Exim is the MTA), isn't very 
> tolerant of this binary data and refuses to deliver the message, causing 
> it to bounce.
> 
> The people I derive my income from tell me bouncing the messages isn't 
> an option.
> 
> The fix:
> 
> I have created a patch, against 4.14 (no idea if it'll work with older 
> versions), that adds a main configuration option called 
> "strip_nul_char".  This option defaults to off, and provides the same 
> behavior as an unpatched Exim.
> 
> Setting "strip_nul_char" to true causes Exim to remove NULs from SMTP 
> and non-SMTP message bodies as they are received.
> 
> At a later date I intend to make this an option for the lmtp and smtp 
> transports.
> 
> The patch has received some testing here and seems to work fine (with 
> and without exiscan), but more testing is welcome.
> 
> Comments and constructive criticism welcome, flames to /dev/null.

-- 

Phil Brutsche
phil at optimumdata.com
-------------- next part --------------
diff -u -r exim-4.14/src/globals.c exim-4.14-modified/src/globals.c
--- exim-4.14/src/globals.c	Tue Mar 11 06:20:20 2003
+++ exim-4.14-modified/src/globals.c	Tue Apr  8 17:22:27 2003
@@ -846,6 +846,7 @@
 			   "\0<--------------Space to patch spool_directory->";
 int     string_datestamp_offset= -1;
 BOOL    strip_excess_angle_brackets = FALSE;
+BOOL    strip_nul_char         = FALSE;
 BOOL    strip_trailing_dot     = FALSE;
 BOOL    synchronous_delivery   = FALSE;
 BOOL    syslog_timestamp       = TRUE;
diff -u -r exim-4.14/src/globals.h exim-4.14-modified/src/globals.h
--- exim-4.14/src/globals.h	Tue Mar 11 06:20:20 2003
+++ exim-4.14-modified/src/globals.h	Tue Apr  8 12:40:16 2003
@@ -511,6 +511,7 @@
 extern uschar *spool_directory;        /* Name of spool directory */
 extern int     string_datestamp_offset;/* After insertion by string_format */
 extern BOOL    strip_excess_angle_brackets; /* Surrounding route-addrs */
+extern BOOL    strip_nul_char;         /* Remove ASCII NUL from messages */
 extern BOOL    strip_trailing_dot;     /* Remove dots at ends of domains */
 extern BOOL    synchronous_delivery;   /* TRUE if -odi is set */
 extern BOOL    syslog_timestamp;       /* TRUE if time on syslogs */
diff -u -r exim-4.14/src/readconf.c exim-4.14-modified/src/readconf.c
--- exim-4.14/src/readconf.c	Tue Mar 11 06:20:22 2003
+++ exim-4.14-modified/src/readconf.c	Mon Apr  7 16:35:39 2003
@@ -270,6 +270,7 @@
   { "split_spool_directory",    opt_bool,        &split_spool_directory },
   { "spool_directory",          opt_stringptr,   &spool_directory },
   { "strip_excess_angle_brackets", opt_bool,     &strip_excess_angle_brackets },
+  { "strip_nul_char",           opt_bool,        &strip_nul_char },
   { "strip_trailing_dot",       opt_bool,        &strip_trailing_dot },
   { "syslog_timestamp",         opt_bool,        &syslog_timestamp },
   { "system_filter",            opt_stringptr,   &system_filter },
diff -u -r exim-4.14/src/receive.c exim-4.14-modified/src/receive.c
--- exim-4.14/src/receive.c	Tue Mar 11 06:20:22 2003
+++ exim-4.14-modified/src/receive.c	Tue Apr  8 15:05:13 2003
@@ -441,6 +441,8 @@
 
   for (; (ch = (receive_getc)()) != EOF; last_ch = ch)
     {
+    if (strip_nul_char && (ch == 0))
+      continue;
     if (drop_cr)
       {
       if (last_ch == '\r' && ch != '\n')
@@ -476,6 +478,8 @@
 
 while ((ch = (receive_getc)()) != EOF)
   {
+  if (strip_nul_char && (ch == 0))
+    continue;
   switch (ch_state)
     {
     case 0:                         /* Normal state */
@@ -572,6 +576,8 @@
 
 while ((ch = (receive_getc)()) != EOF)
   {
+  if (strip_nul_char && (ch == 0))
+    continue;
   switch (ch_state)
     {
     case 0:                             /* After LF or CRLF */