[Patch] Resubmit: Timestamp patch to AT2
"Oded Arbel" <[email protected]>
| Newsgroups | gmane.comp.mobile.kannel.devel |
|---|---|
| Message-ID | <[email protected]> |
Hi list. I resubmit the timestamp patch to AT2 against current CVS snapshot - I tested it more throughly and found out that while it works correctly on parsing the time in the PDU (unlike the current implementation), I don't know how to parse the time zone data - on all the handsets I tested, the timestamp contains the handset's clock setting (even if different then the actual time) and the timezone set to GMT (which, since I'm not in GMT timezone, is false). I think that the correct behavior would be to apply the timezone entry in reverse - that is if it says GMT+3, then substruct 3 hours from the hour field. at least, the documentation I've read suggest that - so that's what the code says. please test this throughly. OTOH - even if the timezone parsing is incorrect, this code offers much improved PDU timestamp parsing over the current implimentation (which doesn't do much more then sort of a memcopy to the universtal time struct), and should be applied to the CVS. TIA -- Oded Arbel m-Wise Inc. [email protected] Any clod can have the facts, but having opinions is an art. -- Charles McCabe <<smsc_at2.timestamp.patch>>
smsc_at2.timestamp.patch
(application/octet-stream, 3.3 KB)
Index: gw/smsc_at2.c
===================================================================
RCS file: /home/cvs/gateway/gw/smsc_at2.c,v
retrieving revision 1.34
diff -u -r1.34 smsc_at2.c
--- gw/smsc_at2.c 14 Mar 2002 19:47:32 -0000 1.34
+++ gw/smsc_at2.c 17 Mar 2002 15:53:15 -0000
@@ -130,6 +130,7 @@
int at2_detect_modem_type(PrivAT2data *privdata);
ModemDef *at2_read_modems(PrivAT2data *privdata, Octstr *file, Octstr *id, int idnumber);
void at2_destroy_modem(ModemDef *modem);
+int swap_nibbles(char byte);
/******************************************************************************
** For debugging purposes octstr_destroy might be a macro,
@@ -1236,6 +1237,7 @@
Msg *message = NULL;
struct universaltime mtime; /* time structure */
long stime; /* time in seconds */
+ int timezone; /* timezone in 15 minutes jumps from GMT */
/* Note: some parts of the PDU are not decoded because they are
* not needed for the Msg type. */
@@ -1282,16 +1284,27 @@
pos++;
/* get the timestamp */
- mtime.year = octstr_get_char(pdu, pos) + 1900; pos++;
- mtime.month = octstr_get_char(pdu, pos); pos++;
- mtime.day = octstr_get_char(pdu, pos); pos++;
- mtime.hour = octstr_get_char(pdu, pos); pos++;
- mtime.minute = octstr_get_char(pdu, pos); pos++;
- mtime.second = octstr_get_char(pdu, pos); pos++;
+ mtime.year = swap_nibbles(octstr_get_char(pdu, pos)); pos++;
+ mtime.year += (mtime.year < 70 ? 2000 : 1900);
+ mtime.month = swap_nibbles(octstr_get_char(pdu, pos)) ; pos++;
+ mtime.day = swap_nibbles(octstr_get_char(pdu, pos)); pos++;
+ mtime.hour = swap_nibbles(octstr_get_char(pdu, pos)); pos++;
+ mtime.minute = swap_nibbles(octstr_get_char(pdu, pos)); pos++;
+ mtime.second = swap_nibbles(octstr_get_char(pdu, pos)); pos++;
+
/* time zone: */
- /* XXX handle negative time zones */
- mtime.hour += octstr_get_char(pdu, pos); pos++;
- stime = date_convert_universal(&mtime);
+ /* time zone is "swapped nibble", with the MSB as the sign (1 is negative). */
+ timezone = swap_nibbles(octstr_get_char(pdu, pos)); pos++;
+ timezone = (timezone >> 7) ? -1 : 1) * (timezone & 127);
+ /* ok - that was the time zone as read from the PDU. now how to interpert it ? all the handsets I tested
+ send the timestamp of their local time and the timezone as GMT+0. I assume that the timestamp is
+ the handset's local time, so we need to apply the timezone in reverse to get GM time : */
+
+ /* time in PDU is handset's local time and timezone is handset's time zone difference from GMT */
+ mtime.hour -= timezone / 4;
+ mtime.minute -= 15 * (timezone % 4);
+
+ stime = date_convert_universal(&mtime);
/* get data length */
len = octstr_get_char(pdu, pos);
@@ -2057,5 +2070,14 @@
octstr_destroy(modem->keepalive_cmd);
gw_free(modem);
}
+}
+
+/* this silly thing here will just translate a "swapped nibble" pseodo Hex encoding
+ (from PDU) into something that people can actually understand.
+ implementation completly ripped off Dennis Malmstrom timestamp patches against 1.0.3.
+ thanks Dennis ! */
+int swap_nibbles(char byte)
+{
+ return ( ( byte & 15 ) * 10 ) + ( byte >> 4 );
}