Hey, it's the overzealous newb again. APR doesn't have any function
for parsing an RFC 822 date. Since these dates show up in e-mail and
http headers, I've attached a patch to trunk that adds
apr_parse_rfc822_date. I've also added it to test_rfcstr, but the test
is just for one date. I'm guessing this needs to be added to time/
win32/timestr.c as well, but I don't have a Windows box for testing.
If you're wondering why the time zone parsing in
apr_parse_rfc822_time_zone is so crazy, take a look at section 5.1 of
the spec (http://www.faqs.org/rfcs/rfc822.html). The UTC offset can be
of the following forms:
1. +/-HHMM
2. A single letter corresponding to an offset (A = UTC-1, B = UTC-2,
but J is skipped so I is UTC-9 and K is UTC-10). Letter order is
reversed for positive UTC offsets (N is UTC+1, O is UTC+2). Also Z is
UTC.
3. A popular abbreviation (GMT and UT are both UTC. MST and PDT are
UTC-7).
Anyways, it works, but it's not written very well. There are too many
hard-coded constants hanging around. Validation is pretty weak as
well. Nonexistent dates (Wed, 17 Aug 2008) can be parsed without
returning APR_EBADDATE.
It's a start, though. Suggestions or comments would be nice.
Geoff
Index: time/unix/timestr.c
===================================================================
--- time/unix/timestr.c (revision 689153)
+++ time/unix/timestr.c (working copy)
@@ -39,6 +39,36 @@
"Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat"
};
+APR_DECLARE_DATA const char *apr_rfc822_time_zones[26][4] =
+{
+ {"M", NULL}, //UTC-12
+ {"L", NULL},
+ {"K", NULL},
+ {"I", NULL},
+ {"H", "PST", NULL}, //UTC-8
+ {"G", "MST", "PDT", NULL}, //UTC-7
+ {"F", "CST", "MDT", NULL}, //UTC-6
+ {"E", "EST", "CDT", NULL}, //UTC-5
+ {"D", "EDT", NULL}, //UTC-4
+ {"C", NULL},
+ {"B", NULL},
+ {"A", NULL},
+ {"Z", "UT", "GMT", NULL}, //UTC
+ {"N", NULL},
+ {"O", NULL},
+ {"P", NULL},
+ {"Q", NULL},
+ {"R", NULL},
+ {"S", NULL},
+ {"T", NULL},
+ {"U", NULL},
+ {"V", NULL},
+ {"W", NULL},
+ {"X", NULL},
+ {"Y", NULL}, //UTC+12
+ {NULL}
+};
+
apr_status_t apr_rfc822_date(char *date_str, apr_time_t t)
{
apr_time_exp_t xt;
@@ -87,6 +117,108 @@
return APR_SUCCESS;
}
+apr_int32_t apr_parse_day_of_week(char *day_of_week) {
+ apr_int32_t i;
+ for(i = 0; i < 7; i++) {
+ if(strcmp(day_of_week, apr_day_snames[i]) == 0) {
+ return i;
+ }
+ }
+
+ return -1;
+}
+
+apr_int32_t apr_parse_month(char *month) {
+ apr_int32_t i;
+ for(i = 0; i < 12; i++) {
+ if(strcmp(month, apr_month_snames[i]) == 0) {
+ return i;
+ }
+ }
+
+ return -1;
+}
+
+apr_status_t apr_parse_rfc822_time_zone(apr_int32_t *offset, const char *s) {
+ apr_int32_t hours = 0, minutes = 0, sign = 1;
+
+ /* +HHMM */
+ if(strlen(s) == 5 && (s[0] == '+' || s[0] == '-')) {
+ if (s[0] == '-')
+ sign = -1;
+
+ /* The pointer is passed by value so we're not actually corrupting the
string here. */
+ s++;
+ hours = atoi(s) / 100;
+ s+=2;
+ minutes = atoi(s);
+
+ *offset = sign * ((hours * 60) + minutes) * 60;
+
+ return APR_SUCCESS;
+ }
+
+ /* UT, GMT, PST, etc */
+ int i;
+ for(i = 0; apr_rfc822_time_zones[i] != NULL; i++) {
+ const char **offset_array = apr_rfc822_time_zones[i];
+
+ int j;
+ for(j = 0; offset_array[j] != NULL; j++) {
+ const char *offset_string = offset_array[j];
+
+ if(strcmp(s, offset_string) == 0) {
+ *offset = (i - 12) * 60 * 60;
+ return APR_SUCCESS;
+ }
+ }
+ }
+
+ return APR_EBADDATE;
+}
+
+apr_status_t apr_parse_rfc822_date(apr_time_exp_t *xt, char *date_str)
+{
+ apr_int32_t year = 0, month = 0, day = 0, hour = 0, minute = 0, second =
0, day_of_week = 0, offset = 0;
+ char day_of_week_str[4];
+ char month_str[4];
+ char zone[6];
+ apr_status_t err = APR_SUCCESS;
+
+ //Day of week, seconds
+ if (sscanf(date_str, "%3s%*c %u %3s %u %u:%u:%u %5s", &day_of_week_str,
&day, &month_str, &year, &hour, &minute, &second, &zone) == 8) {}
+ //Day of week, no seconds
+ else if (sscanf(date_str, "%3s%*c %u %3s %u %u:%u %5s", &day_of_week_str,
&day, month_str, &year, &hour, &minute, &zone) == 7) {}
+ //No day of week, seconds
+ else if (sscanf(date_str, "%u %3s %u %u:%u:%u %5s", &day, &month_str,
&year, &hour, &minute, &second, &zone) == 7) {}
+ //No day of week, no seconds
+ else if (sscanf(date_str, "%u %3s %u %u:%u %5s", &day, &month_str, &year,
&hour, &minute, &zone) == 6) {}
+ else {
+ return APR_EBADDATE;
+ }
+
+ day_of_week = apr_parse_day_of_week(day_of_week_str);
+ month = apr_parse_month(month_str);
+ err = apr_parse_rfc822_time_zone(&offset, zone);
+
+ if(day_of_week == -1 || month == -1 || err != APR_SUCCESS) {
+ return APR_EBADDATE;
+ }
+
+ xt->tm_year = year - 1900;
+ xt->tm_mon = month;
+ xt->tm_mday = day;
+ xt->tm_hour = hour;
+ xt->tm_min = minute;
+ xt->tm_sec = second;
+ xt->tm_wday = day_of_week;
+ xt->tm_gmtoff = offset;
+ xt->tm_usec = 0;
+ /* XXXX: set xt->tm_isdst if time zone string is EDT, CDT, etc */
+
+ return APR_SUCCESS;
+}
+
apr_status_t apr_ctime(char *date_str, apr_time_t t)
{
apr_time_exp_t xt;
Index: test/testtime.c
===================================================================
--- test/testtime.c (revision 689153)
+++ test/testtime.c (working copy)
@@ -166,9 +166,20 @@
{
apr_status_t rv;
char str[STR_SIZE];
+ apr_time_exp_t xt;
+ apr_time_t imp;
- rv = apr_rfc822_date(str, now);
+ rv = apr_parse_rfc822_date(&xt, "Sat, 14 Sep 2002 19:05:36 GMT");
if (rv == APR_ENOTIMPL) {
+ ABTS_NOT_IMPL(tc, "apr_parse_rfc822_date");
+ }
+ ABTS_TRUE(tc, rv == APR_SUCCESS);
+
+ rv = apr_time_exp_get(&imp, &xt);
+ ABTS_TRUE(tc, rv == APR_SUCCESS);
+
+ rv = apr_rfc822_date(str, imp);
+ if (rv == APR_ENOTIMPL) {
ABTS_NOT_IMPL(tc, "apr_rfc822_date");
}
ABTS_TRUE(tc, rv == APR_SUCCESS);
Index: include/apr_time.h
===================================================================
--- include/apr_time.h (revision 689153)
+++ include/apr_time.h (working copy)
@@ -189,6 +189,21 @@
*/
APR_DECLARE(apr_status_t) apr_rfc822_date(char *date_str, apr_time_t t);
+/**
+ * apr_parse_rfc822_time_zone parses RFC822 time zone for use in
apr_time_exp_t.tm_gmtoff.
+ * This is a helper for apr_parse_rfc822_date.
+ * @param offset tm_gmtoff seconds east of UTC
+ * @param s String to parse
+ */
+APR_DECLARE(apr_status_t) apr_parse_rfc822_time_zone(apr_int32_t *offset,
const char *s);
+
+/**
+ * apr_parse_rfc822_date parses RFC822 dates into an apr_time_exp_t
+ * @param xt time struct to write to
+ * @param date_str String to parse
+ */
+APR_DECLARE(apr_status_t) apr_parse_rfc822_date(apr_time_exp_t *xt, char
*date_str);
+
/** length of a CTIME date */
#define APR_CTIME_LEN (25)
/**
