[asterisk-dev] SIPPEER() mistake getting IPAddress in peer

[Walter Doekes]

On 12/03/13 02:06, keiichiro kawasaki wrote:
> I tried processing text messages on asterisk. I have used version of
> asterisk is SVN-trunk-r382764M.
>
> extensions.conf is:
> exten = 161616,1,NoOp(recv messages)
> exten = 161616,n,NoOp(${EXTEN}: ${MESSAGE(body)})
> exten = 161616,n,MessageSend(sip:${EXTEN}@$[SIPPEER(${EXTEN},ip)]:$[SIPPEER(${EXTEN},port)],
> ${MESSAGE(from)})

You might be right that expr2 ($[...]) wrongly treats your value as a 
double and that should probably be fixed (*).

The reason why no one has caught that, is that you're using an odd 
construct.

You do:

   sip:${EXTEN}@$[SIPPEER(${EXTEN},ip)]:$[SIPPEER(${EXTEN},port)]

While we do:

   sip:${EXTEN}@${SIPPEER(${EXTEN},ip)}:${SIPPEER(${EXTEN},port)}

The $[...] construct is generally used for arithmetic and comparisons, 
so treating values as numbers isn't at all odd.

Regards,
Walter

(*) Fixing that could be invasive and have unexpected side-effects for 
people who have gotten used to the automatic casting.



> SIPPEER function should get registered ip address and port of peer.
> But I sent MESSAGE from 171717 to 161616, SIPPEER func got "192.168"
> and "43967". Its correct ip address is "192.168.1.195".
>
> this is full log then.
> [Mar 11 20:28:30] VERBOSE[7630][C-00000000] pbx.c:     -- Executing
> [161616 at messages:1] NoOp("Message/ast_msg_queue", "recv messages
> 161616:aaa") in new stack
> [Mar 11 20:28:30] VERBOSE[7679] chan_sip.c: Scheduling destruction of
> SIP dialog 'btLtuPlbap44Vrc3D5oAGZrXrSj9uXZn' in 32000 ms (Method:
> MESSAGE)
> [Mar 11 20:28:30] DEBUG[7630][C-00000000] pbx.c: Result of 'EXTEN' is '161616'
> [Mar 11 20:28:30] DEBUG[7630][C-00000000] pbx.c: Result of 'EXTEN' is '161616'
> [Mar 11 20:28:30] NOTICE[7630][C-00000000] ast_expr2.y: argbuf
> allocated 10 bytes;
> [Mar 11 20:28:30] NOTICE[7630][C-00000000] ast_expr2.y: argbuf uses 9 bytes;
> [Mar 11 20:28:30] DEBUG[7630][C-00000000] pbx.c: Expression result is '192.168'
> [Mar 11 20:28:30] DEBUG[7630][C-00000000] pbx.c: Result of 'EXTEN' is '161616'
> [Mar 11 20:28:30] NOTICE[7630][C-00000000] ast_expr2.y: argbuf
> allocated 12 bytes;
> [Mar 11 20:28:30] NOTICE[7630][C-00000000] ast_expr2.y: argbuf uses 11 bytes;
> [Mar 11 20:28:30] DEBUG[7630][C-00000000] pbx.c: Expression result is '43967'
> [Mar 11 20:28:30] DEBUG[7630][C-00000000] pbx.c: Function result is
> '<sip:171717 at 192.168.1.242>'
> [Mar 11 20:28:30] DEBUG[7630][C-00000000] pbx.c: Launching 'MessageSend'
> [Mar 11 20:28:30] VERBOSE[7630][C-00000000] pbx.c:     -- Executing
> [161616 at messages:2] MessageSend("Message/ast_msg_queue",
> "sip:161616 at 192.168:43967, <sip:171717 at 192.168.1.242>") in new stack
>
> I found issue in main/ast_expr2.c. the gotten value is checked in the
> number or the string in line about 3066 of op_func(). If is_really_num() has
> one or more periods, the func will assess that long double format even
> if IP address format.
>
> I fixed:
> Index: main/ast_expr2.c
> ===================================================================
> --- main/ast_expr2.c    (revision 382782)
> +++ main/ast_expr2.c    (working copy)
> @@ -2792,7 +2792,7 @@
>
>   static int is_really_num(char *str)
>   {
> -       if ( strspn(str,"-0123456789.   ") == strlen(str))
> +       if ((strspn(str,"-0123456789.   ") == strlen(str)) &&
> ((strchr(str, '.') == strrchr(str, '.'))) )
>                  return 1;
>          else
>                  return 0;
>
> It's fine this situation.
> Is this known Issues? If I have worng, please tell me fixed points.
>
> I thank you for reading it through.
> --
> Keiichiro Kawasaki