postgresql/src/interfaces/ecpg/pgtypeslib/dt_common.c

3004 lines
73 KiB
C

/* src/interfaces/ecpg/pgtypeslib/dt_common.c */
#include "postgres_fe.h"
#include <time.h>
#include <ctype.h>
#include <math.h>
#include "common/string.h"
#include "dt.h"
#include "pgtypes_timestamp.h"
#include "pgtypeslib_extern.h"
const int day_tab[2][13] = {
{31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31, 0},
{31, 29, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31, 0}};
typedef long AbsoluteTime;
static const datetkn datetktbl[] = {
/* text, token, lexval */
{EARLY, RESERV, DTK_EARLY}, /* "-infinity" reserved for "early time" */
{"acsst", DTZ, 37800}, /* Cent. Australia */
{"acst", DTZ, -14400}, /* Atlantic/Porto Acre */
{"act", TZ, -18000}, /* Atlantic/Porto Acre */
{DA_D, ADBC, AD}, /* "ad" for years >= 0 */
{"adt", DTZ, -10800}, /* Atlantic Daylight Time */
{"aesst", DTZ, 39600}, /* E. Australia */
{"aest", TZ, 36000}, /* Australia Eastern Std Time */
{"aft", TZ, 16200}, /* Kabul */
{"ahst", TZ, -36000}, /* Alaska-Hawaii Std Time */
{"akdt", DTZ, -28800}, /* Alaska Daylight Time */
{"akst", DTZ, -32400}, /* Alaska Standard Time */
{"allballs", RESERV, DTK_ZULU}, /* 00:00:00 */
{"almst", TZ, 25200}, /* Almaty Savings Time */
{"almt", TZ, 21600}, /* Almaty Time */
{"am", AMPM, AM},
{"amst", DTZ, 18000}, /* Armenia Summer Time (Yerevan) */
#if 0
{"amst", DTZ, -10800}, /* Porto Velho */
#endif
{"amt", TZ, 14400}, /* Armenia Time (Yerevan) */
{"anast", DTZ, 46800}, /* Anadyr Summer Time (Russia) */
{"anat", TZ, 43200}, /* Anadyr Time (Russia) */
{"apr", MONTH, 4},
{"april", MONTH, 4},
#if 0
aqtst
aqtt
arst
#endif
{"art", TZ, -10800}, /* Argentina Time */
#if 0
ashst
ast /* Atlantic Standard Time, Arabia Standard
* Time, Acre Standard Time */
#endif
{"ast", TZ, -14400}, /* Atlantic Std Time (Canada) */
{"at", IGNORE_DTF, 0}, /* "at" (throwaway) */
{"aug", MONTH, 8},
{"august", MONTH, 8},
{"awsst", DTZ, 32400}, /* W. Australia */
{"awst", TZ, 28800}, /* W. Australia */
{"awt", DTZ, -10800},
{"azost", DTZ, 0}, /* Azores Summer Time */
{"azot", TZ, -3600}, /* Azores Time */
{"azst", DTZ, 18000}, /* Azerbaijan Summer Time */
{"azt", TZ, 14400}, /* Azerbaijan Time */
{DB_C, ADBC, BC}, /* "bc" for years < 0 */
{"bdst", TZ, 7200}, /* British Double Summer Time */
{"bdt", TZ, 21600}, /* Dacca */
{"bnt", TZ, 28800}, /* Brunei Darussalam Time */
{"bort", TZ, 28800}, /* Borneo Time (Indonesia) */
#if 0
bortst
bost
#endif
{"bot", TZ, -14400}, /* Bolivia Time */
{"bra", TZ, -10800}, /* Brazil Time */
#if 0
brst
brt
#endif
{"bst", DTZ, 3600}, /* British Summer Time */
#if 0
{"bst", TZ, -10800}, /* Brazil Standard Time */
{"bst", DTZ, -39600}, /* Bering Summer Time */
#endif
{"bt", TZ, 10800}, /* Baghdad Time */
{"btt", TZ, 21600}, /* Bhutan Time */
{"cadt", DTZ, 37800}, /* Central Australian DST */
{"cast", TZ, 34200}, /* Central Australian ST */
{"cat", TZ, -36000}, /* Central Alaska Time */
{"cct", TZ, 28800}, /* China Coast Time */
#if 0
{"cct", TZ, 23400}, /* Indian Cocos (Island) Time */
#endif
{"cdt", DTZ, -18000}, /* Central Daylight Time */
{"cest", DTZ, 7200}, /* Central European Dayl.Time */
{"cet", TZ, 3600}, /* Central European Time */
{"cetdst", DTZ, 7200}, /* Central European Dayl.Time */
{"chadt", DTZ, 49500}, /* Chatham Island Daylight Time (13:45) */
{"chast", TZ, 45900}, /* Chatham Island Time (12:45) */
#if 0
ckhst
#endif
{"ckt", TZ, 43200}, /* Cook Islands Time */
{"clst", DTZ, -10800}, /* Chile Summer Time */
{"clt", TZ, -14400}, /* Chile Time */
#if 0
cost
#endif
{"cot", TZ, -18000}, /* Columbia Time */
{"cst", TZ, -21600}, /* Central Standard Time */
#if 0
cvst
#endif
{"cvt", TZ, 25200}, /* Christmas Island Time (Indian Ocean) */
{"cxt", TZ, 25200}, /* Christmas Island Time (Indian Ocean) */
{"d", UNITS, DTK_DAY}, /* "day of month" for ISO input */
{"davt", TZ, 25200}, /* Davis Time (Antarctica) */
{"ddut", TZ, 36000}, /* Dumont-d'Urville Time (Antarctica) */
{"dec", MONTH, 12},
{"december", MONTH, 12},
{"dnt", TZ, 3600}, /* Dansk Normal Tid */
{"dow", UNITS, DTK_DOW}, /* day of week */
{"doy", UNITS, DTK_DOY}, /* day of year */
{"dst", DTZMOD, SECS_PER_HOUR},
#if 0
{"dusst", DTZ, 21600}, /* Dushanbe Summer Time */
#endif
{"easst", DTZ, -18000}, /* Easter Island Summer Time */
{"east", TZ, -21600}, /* Easter Island Time */
{"eat", TZ, 10800}, /* East Africa Time */
#if 0
{"east", DTZ, 14400}, /* Indian Antananarivo Savings Time */
{"eat", TZ, 10800}, /* Indian Antananarivo Time */
{"ect", TZ, -14400}, /* Eastern Caribbean Time */
{"ect", TZ, -18000}, /* Ecuador Time */
#endif
{"edt", DTZ, -14400}, /* Eastern Daylight Time */
{"eest", DTZ, 10800}, /* Eastern Europe Summer Time */
{"eet", TZ, 7200}, /* East. Europe, USSR Zone 1 */
{"eetdst", DTZ, 10800}, /* Eastern Europe Daylight Time */
{"egst", DTZ, 0}, /* East Greenland Summer Time */
{"egt", TZ, -3600}, /* East Greenland Time */
#if 0
ehdt
#endif
{EPOCH, RESERV, DTK_EPOCH}, /* "epoch" reserved for system epoch time */
{"est", TZ, -18000}, /* Eastern Standard Time */
{"feb", MONTH, 2},
{"february", MONTH, 2},
{"fjst", DTZ, -46800}, /* Fiji Summer Time (13 hour offset!) */
{"fjt", TZ, -43200}, /* Fiji Time */
{"fkst", DTZ, -10800}, /* Falkland Islands Summer Time */
{"fkt", TZ, -7200}, /* Falkland Islands Time */
#if 0
fnst
fnt
#endif
{"fri", DOW, 5},
{"friday", DOW, 5},
{"fst", TZ, 3600}, /* French Summer Time */
{"fwt", DTZ, 7200}, /* French Winter Time */
{"galt", TZ, -21600}, /* Galapagos Time */
{"gamt", TZ, -32400}, /* Gambier Time */
{"gest", DTZ, 18000}, /* Georgia Summer Time */
{"get", TZ, 14400}, /* Georgia Time */
{"gft", TZ, -10800}, /* French Guiana Time */
#if 0
ghst
#endif
{"gilt", TZ, 43200}, /* Gilbert Islands Time */
{"gmt", TZ, 0}, /* Greenwich Mean Time */
{"gst", TZ, 36000}, /* Guam Std Time, USSR Zone 9 */
{"gyt", TZ, -14400}, /* Guyana Time */
{"h", UNITS, DTK_HOUR}, /* "hour" */
#if 0
hadt
hast
#endif
{"hdt", DTZ, -32400}, /* Hawaii/Alaska Daylight Time */
#if 0
hkst
#endif
{"hkt", TZ, 28800}, /* Hong Kong Time */
#if 0
{"hmt", TZ, 10800}, /* Hellas ? ? */
hovst
hovt
#endif
{"hst", TZ, -36000}, /* Hawaii Std Time */
#if 0
hwt
#endif
{"ict", TZ, 25200}, /* Indochina Time */
{"idle", TZ, 43200}, /* Intl. Date Line, East */
{"idlw", TZ, -43200}, /* Intl. Date Line, West */
#if 0
idt /* Israeli, Iran, Indian Daylight Time */
#endif
{LATE, RESERV, DTK_LATE}, /* "infinity" reserved for "late time" */
{"iot", TZ, 18000}, /* Indian Chagos Time */
{"irkst", DTZ, 32400}, /* Irkutsk Summer Time */
{"irkt", TZ, 28800}, /* Irkutsk Time */
{"irt", TZ, 12600}, /* Iran Time */
{"isodow", UNITS, DTK_ISODOW}, /* ISO day of week, Sunday == 7 */
#if 0
isst
#endif
{"ist", TZ, 7200}, /* Israel */
{"it", TZ, 12600}, /* Iran Time */
{"j", UNITS, DTK_JULIAN},
{"jan", MONTH, 1},
{"january", MONTH, 1},
{"javt", TZ, 25200}, /* Java Time (07:00? see JT) */
{"jayt", TZ, 32400}, /* Jayapura Time (Indonesia) */
{"jd", UNITS, DTK_JULIAN},
{"jst", TZ, 32400}, /* Japan Std Time,USSR Zone 8 */
{"jt", TZ, 27000}, /* Java Time (07:30? see JAVT) */
{"jul", MONTH, 7},
{"julian", UNITS, DTK_JULIAN},
{"july", MONTH, 7},
{"jun", MONTH, 6},
{"june", MONTH, 6},
{"kdt", DTZ, 36000}, /* Korea Daylight Time */
{"kgst", DTZ, 21600}, /* Kyrgyzstan Summer Time */
{"kgt", TZ, 18000}, /* Kyrgyzstan Time */
{"kost", TZ, 43200}, /* Kosrae Time */
{"krast", DTZ, 25200}, /* Krasnoyarsk Summer Time */
{"krat", TZ, 28800}, /* Krasnoyarsk Standard Time */
{"kst", TZ, 32400}, /* Korea Standard Time */
{"lhdt", DTZ, 39600}, /* Lord Howe Daylight Time, Australia */
{"lhst", TZ, 37800}, /* Lord Howe Standard Time, Australia */
{"ligt", TZ, 36000}, /* From Melbourne, Australia */
{"lint", TZ, 50400}, /* Line Islands Time (Kiribati; +14 hours!) */
{"lkt", TZ, 21600}, /* Lanka Time */
{"m", UNITS, DTK_MONTH}, /* "month" for ISO input */
{"magst", DTZ, 43200}, /* Magadan Summer Time */
{"magt", TZ, 39600}, /* Magadan Time */
{"mar", MONTH, 3},
{"march", MONTH, 3},
{"mart", TZ, -34200}, /* Marquesas Time */
{"mawt", TZ, 21600}, /* Mawson, Antarctica */
{"may", MONTH, 5},
{"mdt", DTZ, -21600}, /* Mountain Daylight Time */
{"mest", DTZ, 7200}, /* Middle Europe Summer Time */
{"met", TZ, 3600}, /* Middle Europe Time */
{"metdst", DTZ, 7200}, /* Middle Europe Daylight Time */
{"mewt", TZ, 3600}, /* Middle Europe Winter Time */
{"mez", TZ, 3600}, /* Middle Europe Zone */
{"mht", TZ, 43200}, /* Kwajalein */
{"mm", UNITS, DTK_MINUTE}, /* "minute" for ISO input */
{"mmt", TZ, 23400}, /* Myannar Time */
{"mon", DOW, 1},
{"monday", DOW, 1},
#if 0
most
#endif
{"mpt", TZ, 36000}, /* North Mariana Islands Time */
{"msd", DTZ, 14400}, /* Moscow Summer Time */
{"msk", TZ, 10800}, /* Moscow Time */
{"mst", TZ, -25200}, /* Mountain Standard Time */
{"mt", TZ, 30600}, /* Moluccas Time */
{"mut", TZ, 14400}, /* Mauritius Island Time */
{"mvt", TZ, 18000}, /* Maldives Island Time */
{"myt", TZ, 28800}, /* Malaysia Time */
#if 0
ncst
#endif
{"nct", TZ, 39600}, /* New Caledonia Time */
{"ndt", DTZ, -9000}, /* Nfld. Daylight Time */
{"nft", TZ, -12600}, /* Newfoundland Standard Time */
{"nor", TZ, 3600}, /* Norway Standard Time */
{"nov", MONTH, 11},
{"november", MONTH, 11},
{"novst", DTZ, 25200}, /* Novosibirsk Summer Time */
{"novt", TZ, 21600}, /* Novosibirsk Standard Time */
{NOW, RESERV, DTK_NOW}, /* current transaction time */
{"npt", TZ, 20700}, /* Nepal Standard Time (GMT-5:45) */
{"nst", TZ, -12600}, /* Nfld. Standard Time */
{"nt", TZ, -39600}, /* Nome Time */
{"nut", TZ, -39600}, /* Niue Time */
{"nzdt", DTZ, 46800}, /* New Zealand Daylight Time */
{"nzst", TZ, 43200}, /* New Zealand Standard Time */
{"nzt", TZ, 43200}, /* New Zealand Time */
{"oct", MONTH, 10},
{"october", MONTH, 10},
{"omsst", DTZ, 25200}, /* Omsk Summer Time */
{"omst", TZ, 21600}, /* Omsk Time */
{"on", IGNORE_DTF, 0}, /* "on" (throwaway) */
{"pdt", DTZ, -25200}, /* Pacific Daylight Time */
#if 0
pest
#endif
{"pet", TZ, -18000}, /* Peru Time */
{"petst", DTZ, 46800}, /* Petropavlovsk-Kamchatski Summer Time */
{"pett", TZ, 43200}, /* Petropavlovsk-Kamchatski Time */
{"pgt", TZ, 36000}, /* Papua New Guinea Time */
{"phot", TZ, 46800}, /* Phoenix Islands (Kiribati) Time */
#if 0
phst
#endif
{"pht", TZ, 28800}, /* Philippine Time */
{"pkt", TZ, 18000}, /* Pakistan Time */
{"pm", AMPM, PM},
{"pmdt", DTZ, -7200}, /* Pierre & Miquelon Daylight Time */
#if 0
pmst
#endif
{"pont", TZ, 39600}, /* Ponape Time (Micronesia) */
{"pst", TZ, -28800}, /* Pacific Standard Time */
{"pwt", TZ, 32400}, /* Palau Time */
{"pyst", DTZ, -10800}, /* Paraguay Summer Time */
{"pyt", TZ, -14400}, /* Paraguay Time */
{"ret", DTZ, 14400}, /* Reunion Island Time */
{"s", UNITS, DTK_SECOND}, /* "seconds" for ISO input */
{"sadt", DTZ, 37800}, /* S. Australian Dayl. Time */
#if 0
samst
samt
#endif
{"sast", TZ, 34200}, /* South Australian Std Time */
{"sat", DOW, 6},
{"saturday", DOW, 6},
#if 0
sbt
#endif
{"sct", DTZ, 14400}, /* Mahe Island Time */
{"sep", MONTH, 9},
{"sept", MONTH, 9},
{"september", MONTH, 9},
{"set", TZ, -3600}, /* Seychelles Time ?? */
#if 0
sgt
#endif
{"sst", DTZ, 7200}, /* Swedish Summer Time */
{"sun", DOW, 0},
{"sunday", DOW, 0},
{"swt", TZ, 3600}, /* Swedish Winter Time */
#if 0
syot
#endif
{"t", ISOTIME, DTK_TIME}, /* Filler for ISO time fields */
{"tft", TZ, 18000}, /* Kerguelen Time */
{"that", TZ, -36000}, /* Tahiti Time */
{"thu", DOW, 4},
{"thur", DOW, 4},
{"thurs", DOW, 4},
{"thursday", DOW, 4},
{"tjt", TZ, 18000}, /* Tajikistan Time */
{"tkt", TZ, -36000}, /* Tokelau Time */
{"tmt", TZ, 18000}, /* Turkmenistan Time */
{TODAY, RESERV, DTK_TODAY}, /* midnight */
{TOMORROW, RESERV, DTK_TOMORROW}, /* tomorrow midnight */
#if 0
tost
#endif
{"tot", TZ, 46800}, /* Tonga Time */
#if 0
tpt
#endif
{"truk", TZ, 36000}, /* Truk Time */
{"tue", DOW, 2},
{"tues", DOW, 2},
{"tuesday", DOW, 2},
{"tvt", TZ, 43200}, /* Tuvalu Time */
#if 0
uct
#endif
{"ulast", DTZ, 32400}, /* Ulan Bator Summer Time */
{"ulat", TZ, 28800}, /* Ulan Bator Time */
{"ut", TZ, 0},
{"utc", TZ, 0},
{"uyst", DTZ, -7200}, /* Uruguay Summer Time */
{"uyt", TZ, -10800}, /* Uruguay Time */
{"uzst", DTZ, 21600}, /* Uzbekistan Summer Time */
{"uzt", TZ, 18000}, /* Uzbekistan Time */
{"vet", TZ, -14400}, /* Venezuela Time */
{"vlast", DTZ, 39600}, /* Vladivostok Summer Time */
{"vlat", TZ, 36000}, /* Vladivostok Time */
#if 0
vust
#endif
{"vut", TZ, 39600}, /* Vanuata Time */
{"wadt", DTZ, 28800}, /* West Australian DST */
{"wakt", TZ, 43200}, /* Wake Time */
#if 0
warst
#endif
{"wast", TZ, 25200}, /* West Australian Std Time */
{"wat", TZ, -3600}, /* West Africa Time */
{"wdt", DTZ, 32400}, /* West Australian DST */
{"wed", DOW, 3},
{"wednesday", DOW, 3},
{"weds", DOW, 3},
{"west", DTZ, 3600}, /* Western Europe Summer Time */
{"wet", TZ, 0}, /* Western Europe */
{"wetdst", DTZ, 3600}, /* Western Europe Daylight Savings Time */
{"wft", TZ, 43200}, /* Wallis and Futuna Time */
{"wgst", DTZ, -7200}, /* West Greenland Summer Time */
{"wgt", TZ, -10800}, /* West Greenland Time */
{"wst", TZ, 28800}, /* West Australian Standard Time */
{"y", UNITS, DTK_YEAR}, /* "year" for ISO input */
{"yakst", DTZ, 36000}, /* Yakutsk Summer Time */
{"yakt", TZ, 32400}, /* Yakutsk Time */
{"yapt", TZ, 36000}, /* Yap Time (Micronesia) */
{"ydt", DTZ, -28800}, /* Yukon Daylight Time */
{"yekst", DTZ, 21600}, /* Yekaterinburg Summer Time */
{"yekt", TZ, 18000}, /* Yekaterinburg Time */
{YESTERDAY, RESERV, DTK_YESTERDAY}, /* yesterday midnight */
{"yst", TZ, -32400}, /* Yukon Standard Time */
{"z", TZ, 0}, /* time zone tag per ISO-8601 */
{"zp4", TZ, -14400}, /* UTC +4 hours. */
{"zp5", TZ, -18000}, /* UTC +5 hours. */
{"zp6", TZ, -21600}, /* UTC +6 hours. */
{ZULU, TZ, 0}, /* UTC */
};
static const datetkn deltatktbl[] = {
/* text, token, lexval */
{"@", IGNORE_DTF, 0}, /* postgres relative prefix */
{DAGO, AGO, 0}, /* "ago" indicates negative time offset */
{"c", UNITS, DTK_CENTURY}, /* "century" relative */
{"cent", UNITS, DTK_CENTURY}, /* "century" relative */
{"centuries", UNITS, DTK_CENTURY}, /* "centuries" relative */
{DCENTURY, UNITS, DTK_CENTURY}, /* "century" relative */
{"d", UNITS, DTK_DAY}, /* "day" relative */
{DDAY, UNITS, DTK_DAY}, /* "day" relative */
{"days", UNITS, DTK_DAY}, /* "days" relative */
{"dec", UNITS, DTK_DECADE}, /* "decade" relative */
{DDECADE, UNITS, DTK_DECADE}, /* "decade" relative */
{"decades", UNITS, DTK_DECADE}, /* "decades" relative */
{"decs", UNITS, DTK_DECADE}, /* "decades" relative */
{"h", UNITS, DTK_HOUR}, /* "hour" relative */
{DHOUR, UNITS, DTK_HOUR}, /* "hour" relative */
{"hours", UNITS, DTK_HOUR}, /* "hours" relative */
{"hr", UNITS, DTK_HOUR}, /* "hour" relative */
{"hrs", UNITS, DTK_HOUR}, /* "hours" relative */
{"m", UNITS, DTK_MINUTE}, /* "minute" relative */
{"microsecon", UNITS, DTK_MICROSEC}, /* "microsecond" relative */
{"mil", UNITS, DTK_MILLENNIUM}, /* "millennium" relative */
{"millennia", UNITS, DTK_MILLENNIUM}, /* "millennia" relative */
{DMILLENNIUM, UNITS, DTK_MILLENNIUM}, /* "millennium" relative */
{"millisecon", UNITS, DTK_MILLISEC}, /* relative */
{"mils", UNITS, DTK_MILLENNIUM}, /* "millennia" relative */
{"min", UNITS, DTK_MINUTE}, /* "minute" relative */
{"mins", UNITS, DTK_MINUTE}, /* "minutes" relative */
{DMINUTE, UNITS, DTK_MINUTE}, /* "minute" relative */
{"minutes", UNITS, DTK_MINUTE}, /* "minutes" relative */
{"mon", UNITS, DTK_MONTH}, /* "months" relative */
{"mons", UNITS, DTK_MONTH}, /* "months" relative */
{DMONTH, UNITS, DTK_MONTH}, /* "month" relative */
{"months", UNITS, DTK_MONTH},
{"ms", UNITS, DTK_MILLISEC},
{"msec", UNITS, DTK_MILLISEC},
{DMILLISEC, UNITS, DTK_MILLISEC},
{"mseconds", UNITS, DTK_MILLISEC},
{"msecs", UNITS, DTK_MILLISEC},
{"qtr", UNITS, DTK_QUARTER}, /* "quarter" relative */
{DQUARTER, UNITS, DTK_QUARTER}, /* "quarter" relative */
{"s", UNITS, DTK_SECOND},
{"sec", UNITS, DTK_SECOND},
{DSECOND, UNITS, DTK_SECOND},
{"seconds", UNITS, DTK_SECOND},
{"secs", UNITS, DTK_SECOND},
{DTIMEZONE, UNITS, DTK_TZ}, /* "timezone" time offset */
{"timezone_h", UNITS, DTK_TZ_HOUR}, /* timezone hour units */
{"timezone_m", UNITS, DTK_TZ_MINUTE}, /* timezone minutes units */
{"us", UNITS, DTK_MICROSEC}, /* "microsecond" relative */
{"usec", UNITS, DTK_MICROSEC}, /* "microsecond" relative */
{DMICROSEC, UNITS, DTK_MICROSEC}, /* "microsecond" relative */
{"useconds", UNITS, DTK_MICROSEC}, /* "microseconds" relative */
{"usecs", UNITS, DTK_MICROSEC}, /* "microseconds" relative */
{"w", UNITS, DTK_WEEK}, /* "week" relative */
{DWEEK, UNITS, DTK_WEEK}, /* "week" relative */
{"weeks", UNITS, DTK_WEEK}, /* "weeks" relative */
{"y", UNITS, DTK_YEAR}, /* "year" relative */
{DYEAR, UNITS, DTK_YEAR}, /* "year" relative */
{"years", UNITS, DTK_YEAR}, /* "years" relative */
{"yr", UNITS, DTK_YEAR}, /* "year" relative */
{"yrs", UNITS, DTK_YEAR}, /* "years" relative */
};
static const unsigned int szdatetktbl = lengthof(datetktbl);
static const unsigned int szdeltatktbl = lengthof(deltatktbl);
static const datetkn *datecache[MAXDATEFIELDS] = {NULL};
static const datetkn *deltacache[MAXDATEFIELDS] = {NULL};
char *months[] = {"Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec", NULL};
char *days[] = {"Sunday", "Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", NULL};
char *pgtypes_date_weekdays_short[] = {"Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat", NULL};
char *pgtypes_date_months[] = {"January", "February", "March", "April", "May", "June", "July", "August", "September", "October", "November", "December", NULL};
static const datetkn *
datebsearch(const char *key, const datetkn *base, unsigned int nel)
{
if (nel > 0)
{
const datetkn *last = base + nel - 1,
*position;
int result;
while (last >= base)
{
position = base + ((last - base) >> 1);
/* precheck the first character for a bit of extra speed */
result = (int) key[0] - (int) position->token[0];
if (result == 0)
{
/* use strncmp so that we match truncated tokens */
result = strncmp(key, position->token, TOKMAXLEN);
if (result == 0)
return position;
}
if (result < 0)
last = position - 1;
else
base = position + 1;
}
}
return NULL;
}
/* DecodeUnits()
* Decode text string using lookup table.
* This routine supports time interval decoding.
*/
int
DecodeUnits(int field, char *lowtoken, int *val)
{
int type;
const datetkn *tp;
/* use strncmp so that we match truncated tokens */
if (deltacache[field] != NULL &&
strncmp(lowtoken, deltacache[field]->token, TOKMAXLEN) == 0)
tp = deltacache[field];
else
tp = datebsearch(lowtoken, deltatktbl, szdeltatktbl);
deltacache[field] = tp;
if (tp == NULL)
{
type = UNKNOWN_FIELD;
*val = 0;
}
else
{
type = tp->type;
*val = tp->value;
}
return type;
} /* DecodeUnits() */
/*
* Calendar time to Julian date conversions.
* Julian date is commonly used in astronomical applications,
* since it is numerically accurate and computationally simple.
* The algorithms here will accurately convert between Julian day
* and calendar date for all non-negative Julian days
* (i.e. from Nov 24, -4713 on).
*
* These routines will be used by other date/time packages
* - thomas 97/02/25
*
* Rewritten to eliminate overflow problems. This now allows the
* routines to work correctly for all Julian day counts from
* 0 to 2147483647 (Nov 24, -4713 to Jun 3, 5874898) assuming
* a 32-bit integer. Longer types should also work to the limits
* of their precision.
*/
int
date2j(int y, int m, int d)
{
int julian;
int century;
if (m > 2)
{
m += 1;
y += 4800;
}
else
{
m += 13;
y += 4799;
}
century = y / 100;
julian = y * 365 - 32167;
julian += y / 4 - century + century / 4;
julian += 7834 * m / 256 + d;
return julian;
} /* date2j() */
void
j2date(int jd, int *year, int *month, int *day)
{
unsigned int julian;
unsigned int quad;
unsigned int extra;
int y;
julian = jd;
julian += 32044;
quad = julian / 146097;
extra = (julian - quad * 146097) * 4 + 3;
julian += 60 + quad * 3 + extra / 146097;
quad = julian / 1461;
julian -= quad * 1461;
y = julian * 4 / 1461;
julian = ((y != 0) ? (julian + 305) % 365 : (julian + 306) % 366) + 123;
y += quad * 4;
*year = y - 4800;
quad = julian * 2141 / 65536;
*day = julian - 7834 * quad / 256;
*month = (quad + 10) % 12 + 1;
return;
} /* j2date() */
/* DecodeSpecial()
* Decode text string using lookup table.
* Implement a cache lookup since it is likely that dates
* will be related in format.
*/
static int
DecodeSpecial(int field, char *lowtoken, int *val)
{
int type;
const datetkn *tp;
/* use strncmp so that we match truncated tokens */
if (datecache[field] != NULL &&
strncmp(lowtoken, datecache[field]->token, TOKMAXLEN) == 0)
tp = datecache[field];
else
{
tp = NULL;
if (!tp)
tp = datebsearch(lowtoken, datetktbl, szdatetktbl);
}
datecache[field] = tp;
if (tp == NULL)
{
type = UNKNOWN_FIELD;
*val = 0;
}
else
{
type = tp->type;
*val = tp->value;
}
return type;
} /* DecodeSpecial() */
/* EncodeDateOnly()
* Encode date as local time.
*/
void
EncodeDateOnly(struct tm *tm, int style, char *str, bool EuroDates)
{
Assert(tm->tm_mon >= 1 && tm->tm_mon <= MONTHS_PER_YEAR);
switch (style)
{
case USE_ISO_DATES:
/* compatible with ISO date formats */
if (tm->tm_year > 0)
sprintf(str, "%04d-%02d-%02d",
tm->tm_year, tm->tm_mon, tm->tm_mday);
else
sprintf(str, "%04d-%02d-%02d %s",
-(tm->tm_year - 1), tm->tm_mon, tm->tm_mday, "BC");
break;
case USE_SQL_DATES:
/* compatible with Oracle/Ingres date formats */
if (EuroDates)
sprintf(str, "%02d/%02d", tm->tm_mday, tm->tm_mon);
else
sprintf(str, "%02d/%02d", tm->tm_mon, tm->tm_mday);
if (tm->tm_year > 0)
sprintf(str + 5, "/%04d", tm->tm_year);
else
sprintf(str + 5, "/%04d %s", -(tm->tm_year - 1), "BC");
break;
case USE_GERMAN_DATES:
/* German-style date format */
sprintf(str, "%02d.%02d", tm->tm_mday, tm->tm_mon);
if (tm->tm_year > 0)
sprintf(str + 5, ".%04d", tm->tm_year);
else
sprintf(str + 5, ".%04d %s", -(tm->tm_year - 1), "BC");
break;
case USE_POSTGRES_DATES:
default:
/* traditional date-only style for Postgres */
if (EuroDates)
sprintf(str, "%02d-%02d", tm->tm_mday, tm->tm_mon);
else
sprintf(str, "%02d-%02d", tm->tm_mon, tm->tm_mday);
if (tm->tm_year > 0)
sprintf(str + 5, "-%04d", tm->tm_year);
else
sprintf(str + 5, "-%04d %s", -(tm->tm_year - 1), "BC");
break;
}
}
void
TrimTrailingZeros(char *str)
{
int len = strlen(str);
/* chop off trailing zeros... but leave at least 2 fractional digits */
while (*(str + len - 1) == '0' && *(str + len - 3) != '.')
{
len--;
*(str + len) = '\0';
}
}
/* EncodeDateTime()
* Encode date and time interpreted as local time.
*
* tm and fsec are the value to encode, print_tz determines whether to include
* a time zone (the difference between timestamp and timestamptz types), tz is
* the numeric time zone offset, tzn is the textual time zone, which if
* specified will be used instead of tz by some styles, style is the date
* style, str is where to write the output.
*
* Supported date styles:
* Postgres - day mon hh:mm:ss yyyy tz
* SQL - mm/dd/yyyy hh:mm:ss.ss tz
* ISO - yyyy-mm-dd hh:mm:ss+/-tz
* German - dd.mm.yyyy hh:mm:ss tz
* Variants (affects order of month and day for Postgres and SQL styles):
* US - mm/dd/yyyy
* European - dd/mm/yyyy
*/
void
EncodeDateTime(struct tm *tm, fsec_t fsec, bool print_tz, int tz, const char *tzn, int style, char *str, bool EuroDates)
{
int day,
hour,
min;
/*
* Negative tm_isdst means we have no valid time zone translation.
*/
if (tm->tm_isdst < 0)
print_tz = false;
switch (style)
{
case USE_ISO_DATES:
/* Compatible with ISO-8601 date formats */
sprintf(str, "%04d-%02d-%02d %02d:%02d",
(tm->tm_year > 0) ? tm->tm_year : -(tm->tm_year - 1),
tm->tm_mon, tm->tm_mday, tm->tm_hour, tm->tm_min);
/*
* Print fractional seconds if any. The field widths here should
* be at least equal to MAX_TIMESTAMP_PRECISION.
*/
if (fsec != 0)
{
sprintf(str + strlen(str), ":%02d.%06d", tm->tm_sec, fsec);
TrimTrailingZeros(str);
}
else
sprintf(str + strlen(str), ":%02d", tm->tm_sec);
if (tm->tm_year <= 0)
sprintf(str + strlen(str), " BC");
if (print_tz)
{
hour = -(tz / SECS_PER_HOUR);
min = (abs(tz) / MINS_PER_HOUR) % MINS_PER_HOUR;
if (min != 0)
sprintf(str + strlen(str), "%+03d:%02d", hour, min);
else
sprintf(str + strlen(str), "%+03d", hour);
}
break;
case USE_SQL_DATES:
/* Compatible with Oracle/Ingres date formats */
if (EuroDates)
sprintf(str, "%02d/%02d", tm->tm_mday, tm->tm_mon);
else
sprintf(str, "%02d/%02d", tm->tm_mon, tm->tm_mday);
sprintf(str + 5, "/%04d %02d:%02d",
(tm->tm_year > 0) ? tm->tm_year : -(tm->tm_year - 1),
tm->tm_hour, tm->tm_min);
/*
* Print fractional seconds if any. The field widths here should
* be at least equal to MAX_TIMESTAMP_PRECISION.
*/
if (fsec != 0)
{
sprintf(str + strlen(str), ":%02d.%06d", tm->tm_sec, fsec);
TrimTrailingZeros(str);
}
else
sprintf(str + strlen(str), ":%02d", tm->tm_sec);
if (tm->tm_year <= 0)
sprintf(str + strlen(str), " BC");
/*
* Note: the uses of %.*s in this function would be risky if the
* timezone names ever contain non-ASCII characters. However, all
* TZ abbreviations in the IANA database are plain ASCII.
*/
if (print_tz)
{
if (tzn)
sprintf(str + strlen(str), " %.*s", MAXTZLEN, tzn);
else
{
hour = -(tz / SECS_PER_HOUR);
min = (abs(tz) / MINS_PER_HOUR) % MINS_PER_HOUR;
if (min != 0)
sprintf(str + strlen(str), "%+03d:%02d", hour, min);
else
sprintf(str + strlen(str), "%+03d", hour);
}
}
break;
case USE_GERMAN_DATES:
/* German variant on European style */
sprintf(str, "%02d.%02d", tm->tm_mday, tm->tm_mon);
sprintf(str + 5, ".%04d %02d:%02d",
(tm->tm_year > 0) ? tm->tm_year : -(tm->tm_year - 1),
tm->tm_hour, tm->tm_min);
/*
* Print fractional seconds if any. The field widths here should
* be at least equal to MAX_TIMESTAMP_PRECISION.
*/
if (fsec != 0)
{
sprintf(str + strlen(str), ":%02d.%06d", tm->tm_sec, fsec);
TrimTrailingZeros(str);
}
else
sprintf(str + strlen(str), ":%02d", tm->tm_sec);
if (tm->tm_year <= 0)
sprintf(str + strlen(str), " BC");
if (print_tz)
{
if (tzn)
sprintf(str + strlen(str), " %.*s", MAXTZLEN, tzn);
else
{
hour = -(tz / SECS_PER_HOUR);
min = (abs(tz) / MINS_PER_HOUR) % MINS_PER_HOUR;
if (min != 0)
sprintf(str + strlen(str), "%+03d:%02d", hour, min);
else
sprintf(str + strlen(str), "%+03d", hour);
}
}
break;
case USE_POSTGRES_DATES:
default:
/* Backward-compatible with traditional Postgres abstime dates */
day = date2j(tm->tm_year, tm->tm_mon, tm->tm_mday);
tm->tm_wday = (int) ((day + date2j(2000, 1, 1) + 1) % 7);
memcpy(str, days[tm->tm_wday], 3);
strcpy(str + 3, " ");
if (EuroDates)
sprintf(str + 4, "%02d %3s", tm->tm_mday, months[tm->tm_mon - 1]);
else
sprintf(str + 4, "%3s %02d", months[tm->tm_mon - 1], tm->tm_mday);
sprintf(str + 10, " %02d:%02d", tm->tm_hour, tm->tm_min);
/*
* Print fractional seconds if any. The field widths here should
* be at least equal to MAX_TIMESTAMP_PRECISION.
*/
if (fsec != 0)
{
sprintf(str + strlen(str), ":%02d.%06d", tm->tm_sec, fsec);
TrimTrailingZeros(str);
}
else
sprintf(str + strlen(str), ":%02d", tm->tm_sec);
sprintf(str + strlen(str), " %04d",
(tm->tm_year > 0) ? tm->tm_year : -(tm->tm_year - 1));
if (tm->tm_year <= 0)
sprintf(str + strlen(str), " BC");
if (print_tz)
{
if (tzn)
sprintf(str + strlen(str), " %.*s", MAXTZLEN, tzn);
else
{
/*
* We have a time zone, but no string version. Use the
* numeric form, but be sure to include a leading space to
* avoid formatting something which would be rejected by
* the date/time parser later. - thomas 2001-10-19
*/
hour = -(tz / SECS_PER_HOUR);
min = (abs(tz) / MINS_PER_HOUR) % MINS_PER_HOUR;
if (min != 0)
sprintf(str + strlen(str), " %+03d:%02d", hour, min);
else
sprintf(str + strlen(str), " %+03d", hour);
}
}
break;
}
}
int
GetEpochTime(struct tm *tm)
{
struct tm *t0;
time_t epoch = 0;
t0 = gmtime(&epoch);
if (t0)
{
tm->tm_year = t0->tm_year + 1900;
tm->tm_mon = t0->tm_mon + 1;
tm->tm_mday = t0->tm_mday;
tm->tm_hour = t0->tm_hour;
tm->tm_min = t0->tm_min;
tm->tm_sec = t0->tm_sec;
return 0;
}
return -1;
} /* GetEpochTime() */
static void
abstime2tm(AbsoluteTime _time, int *tzp, struct tm *tm, char **tzn)
{
time_t time = (time_t) _time;
struct tm *tx;
errno = 0;
if (tzp != NULL)
tx = localtime((time_t *) &time);
else
tx = gmtime((time_t *) &time);
if (!tx)
{
errno = PGTYPES_TS_BAD_TIMESTAMP;
return;
}
tm->tm_year = tx->tm_year + 1900;
tm->tm_mon = tx->tm_mon + 1;
tm->tm_mday = tx->tm_mday;
tm->tm_hour = tx->tm_hour;
tm->tm_min = tx->tm_min;
tm->tm_sec = tx->tm_sec;
tm->tm_isdst = tx->tm_isdst;
#if defined(HAVE_STRUCT_TM_TM_ZONE)
tm->tm_gmtoff = tx->tm_gmtoff;
tm->tm_zone = tx->tm_zone;
if (tzp != NULL)
{
/*
* We have a brute force time zone per SQL99? Then use it without
* change since we have already rotated to the time zone.
*/
*tzp = -tm->tm_gmtoff; /* tm_gmtoff is Sun/DEC-ism */
/*
* FreeBSD man pages indicate that this should work - tgl 97/04/23
*/
if (tzn != NULL)
{
/*
* Copy no more than MAXTZLEN bytes of timezone to tzn, in case it
* contains an error message, which doesn't fit in the buffer
*/
StrNCpy(*tzn, tm->tm_zone, MAXTZLEN + 1);
if (strlen(tm->tm_zone) > MAXTZLEN)
tm->tm_isdst = -1;
}
}
else
tm->tm_isdst = -1;
#elif defined(HAVE_INT_TIMEZONE)
if (tzp != NULL)
{
*tzp = (tm->tm_isdst > 0) ? TIMEZONE_GLOBAL - SECS_PER_HOUR : TIMEZONE_GLOBAL;
if (tzn != NULL)
{
/*
* Copy no more than MAXTZLEN bytes of timezone to tzn, in case it
* contains an error message, which doesn't fit in the buffer
*/
StrNCpy(*tzn, TZNAME_GLOBAL[tm->tm_isdst], MAXTZLEN + 1);
if (strlen(TZNAME_GLOBAL[tm->tm_isdst]) > MAXTZLEN)
tm->tm_isdst = -1;
}
}
else
tm->tm_isdst = -1;
#else /* not (HAVE_STRUCT_TM_TM_ZONE || HAVE_INT_TIMEZONE) */
if (tzp != NULL)
{
/* default to UTC */
*tzp = 0;
if (tzn != NULL)
*tzn = NULL;
}
else
tm->tm_isdst = -1;
#endif
}
void
GetCurrentDateTime(struct tm *tm)
{
int tz;
abstime2tm(time(NULL), &tz, tm, NULL);
}
void
dt2time(double jd, int *hour, int *min, int *sec, fsec_t *fsec)
{
int64 time;
time = jd;
*hour = time / USECS_PER_HOUR;
time -= (*hour) * USECS_PER_HOUR;
*min = time / USECS_PER_MINUTE;
time -= (*min) * USECS_PER_MINUTE;
*sec = time / USECS_PER_SEC;
*fsec = time - (*sec * USECS_PER_SEC);
} /* dt2time() */
/* DecodeNumberField()
* Interpret numeric string as a concatenated date or time field.
* Use the context of previously decoded fields to help with
* the interpretation.
*/
static int
DecodeNumberField(int len, char *str, int fmask,
int *tmask, struct tm *tm, fsec_t *fsec, bool *is2digits)
{
char *cp;
/*
* Have a decimal point? Then this is a date or something with a seconds
* field...
*/
if ((cp = strchr(str, '.')) != NULL)
{
char fstr[7];
int i;
cp++;
/*
* OK, we have at most six digits to care about. Let's construct a
* string with those digits, zero-padded on the right, and then do the
* conversion to an integer.
*
* XXX This truncates the seventh digit, unlike rounding it as the
* backend does.
*/
for (i = 0; i < 6; i++)
fstr[i] = *cp != '\0' ? *cp++ : '0';
fstr[i] = '\0';
*fsec = strtoint(fstr, NULL, 10);
*cp = '\0';
len = strlen(str);
}
/* No decimal point and no complete date yet? */
else if ((fmask & DTK_DATE_M) != DTK_DATE_M)
{
/* yyyymmdd? */
if (len == 8)
{
*tmask = DTK_DATE_M;
tm->tm_mday = atoi(str + 6);
*(str + 6) = '\0';
tm->tm_mon = atoi(str + 4);
*(str + 4) = '\0';
tm->tm_year = atoi(str + 0);
return DTK_DATE;
}
/* yymmdd? */
else if (len == 6)
{
*tmask = DTK_DATE_M;
tm->tm_mday = atoi(str + 4);
*(str + 4) = '\0';
tm->tm_mon = atoi(str + 2);
*(str + 2) = '\0';
tm->tm_year = atoi(str + 0);
*is2digits = true;
return DTK_DATE;
}
/* yyddd? */
else if (len == 5)
{
*tmask = DTK_DATE_M;
tm->tm_mday = atoi(str + 2);
*(str + 2) = '\0';
tm->tm_mon = 1;
tm->tm_year = atoi(str + 0);
*is2digits = true;
return DTK_DATE;
}
}
/* not all time fields are specified? */
if ((fmask & DTK_TIME_M) != DTK_TIME_M)
{
/* hhmmss */
if (len == 6)
{
*tmask = DTK_TIME_M;
tm->tm_sec = atoi(str + 4);
*(str + 4) = '\0';
tm->tm_min = atoi(str + 2);
*(str + 2) = '\0';
tm->tm_hour = atoi(str + 0);
return DTK_TIME;
}
/* hhmm? */
else if (len == 4)
{
*tmask = DTK_TIME_M;
tm->tm_sec = 0;
tm->tm_min = atoi(str + 2);
*(str + 2) = '\0';
tm->tm_hour = atoi(str + 0);
return DTK_TIME;
}
}
return -1;
} /* DecodeNumberField() */
/* DecodeNumber()
* Interpret plain numeric field as a date value in context.
*/
static int
DecodeNumber(int flen, char *str, int fmask,
int *tmask, struct tm *tm, fsec_t *fsec, bool *is2digits, bool EuroDates)
{
int val;
char *cp;
*tmask = 0;
val = strtoint(str, &cp, 10);
if (cp == str)
return -1;
if (*cp == '.')
{
/*
* More than two digits? Then could be a date or a run-together time:
* 2001.360 20011225 040506.789
*/
if (cp - str > 2)
return DecodeNumberField(flen, str, (fmask | DTK_DATE_M),
tmask, tm, fsec, is2digits);
*fsec = strtod(cp, &cp);
if (*cp != '\0')
return -1;
}
else if (*cp != '\0')
return -1;
/* Special case day of year? */
if (flen == 3 && (fmask & DTK_M(YEAR)) && val >= 1 && val <= 366)
{
*tmask = (DTK_M(DOY) | DTK_M(MONTH) | DTK_M(DAY));
tm->tm_yday = val;
j2date(date2j(tm->tm_year, 1, 1) + tm->tm_yday - 1,
&tm->tm_year, &tm->tm_mon, &tm->tm_mday);
}
/***
* Enough digits to be unequivocal year? Used to test for 4 digits or
* more, but we now test first for a three-digit doy so anything
* bigger than two digits had better be an explicit year.
* - thomas 1999-01-09
* Back to requiring a 4 digit year. We accept a two digit
* year farther down. - thomas 2000-03-28
***/
else if (flen >= 4)
{
*tmask = DTK_M(YEAR);
/* already have a year? then see if we can substitute... */
if ((fmask & DTK_M(YEAR)) && !(fmask & DTK_M(DAY)) &&
tm->tm_year >= 1 && tm->tm_year <= 31)
{
tm->tm_mday = tm->tm_year;
*tmask = DTK_M(DAY);
}
tm->tm_year = val;
}
/* already have year? then could be month */
else if ((fmask & DTK_M(YEAR)) && !(fmask & DTK_M(MONTH)) && val >= 1 && val <= MONTHS_PER_YEAR)
{
*tmask = DTK_M(MONTH);
tm->tm_mon = val;
}
/* no year and EuroDates enabled? then could be day */
else if ((EuroDates || (fmask & DTK_M(MONTH))) &&
!(fmask & DTK_M(YEAR)) && !(fmask & DTK_M(DAY)) &&
val >= 1 && val <= 31)
{
*tmask = DTK_M(DAY);
tm->tm_mday = val;
}
else if (!(fmask & DTK_M(MONTH)) && val >= 1 && val <= MONTHS_PER_YEAR)
{
*tmask = DTK_M(MONTH);
tm->tm_mon = val;
}
else if (!(fmask & DTK_M(DAY)) && val >= 1 && val <= 31)
{
*tmask = DTK_M(DAY);
tm->tm_mday = val;
}
/*
* Check for 2 or 4 or more digits, but currently we reach here only if
* two digits. - thomas 2000-03-28
*/
else if (!(fmask & DTK_M(YEAR)) && (flen >= 4 || flen == 2))
{
*tmask = DTK_M(YEAR);
tm->tm_year = val;
/* adjust ONLY if exactly two digits... */
*is2digits = (flen == 2);
}
else
return -1;
return 0;
} /* DecodeNumber() */
/* DecodeDate()
* Decode date string which includes delimiters.
* Insist on a complete set of fields.
*/
static int
DecodeDate(char *str, int fmask, int *tmask, struct tm *tm, bool EuroDates)
{
fsec_t fsec;
int nf = 0;
int i,
len;
bool bc = false;
bool is2digits = false;
int type,
val,
dmask = 0;
char *field[MAXDATEFIELDS];
/* parse this string... */
while (*str != '\0' && nf < MAXDATEFIELDS)
{
/* skip field separators */
while (!isalnum((unsigned char) *str))
str++;
field[nf] = str;
if (isdigit((unsigned char) *str))
{
while (isdigit((unsigned char) *str))
str++;
}
else if (isalpha((unsigned char) *str))
{
while (isalpha((unsigned char) *str))
str++;
}
/* Just get rid of any non-digit, non-alpha characters... */
if (*str != '\0')
*str++ = '\0';
nf++;
}
#if 0
/* don't allow too many fields */
if (nf > 3)
return -1;
#endif
*tmask = 0;
/* look first for text fields, since that will be unambiguous month */
for (i = 0; i < nf; i++)
{
if (isalpha((unsigned char) *field[i]))
{
type = DecodeSpecial(i, field[i], &val);
if (type == IGNORE_DTF)
continue;
dmask = DTK_M(type);
switch (type)
{
case MONTH:
tm->tm_mon = val;
break;
case ADBC:
bc = (val == BC);
break;
default:
return -1;
}
if (fmask & dmask)
return -1;
fmask |= dmask;
*tmask |= dmask;
/* mark this field as being completed */
field[i] = NULL;
}
}
/* now pick up remaining numeric fields */
for (i = 0; i < nf; i++)
{
if (field[i] == NULL)
continue;
if ((len = strlen(field[i])) <= 0)
return -1;
if (DecodeNumber(len, field[i], fmask, &dmask, tm, &fsec, &is2digits, EuroDates) != 0)
return -1;
if (fmask & dmask)
return -1;
fmask |= dmask;
*tmask |= dmask;
}
if ((fmask & ~(DTK_M(DOY) | DTK_M(TZ))) != DTK_DATE_M)
return -1;
/* there is no year zero in AD/BC notation; i.e. "1 BC" == year 0 */
if (bc)
{
if (tm->tm_year > 0)
tm->tm_year = -(tm->tm_year - 1);
else
return -1;
}
else if (is2digits)
{
if (tm->tm_year < 70)
tm->tm_year += 2000;
else if (tm->tm_year < 100)
tm->tm_year += 1900;
}
return 0;
} /* DecodeDate() */
/* DecodeTime()
* Decode time string which includes delimiters.
* Only check the lower limit on hours, since this same code
* can be used to represent time spans.
*/
int
DecodeTime(char *str, int *tmask, struct tm *tm, fsec_t *fsec)
{
char *cp;
*tmask = DTK_TIME_M;
tm->tm_hour = strtoint(str, &cp, 10);
if (*cp != ':')
return -1;
str = cp + 1;
tm->tm_min = strtoint(str, &cp, 10);
if (*cp == '\0')
{
tm->tm_sec = 0;
*fsec = 0;
}
else if (*cp != ':')
return -1;
else
{
str = cp + 1;
tm->tm_sec = strtoint(str, &cp, 10);
if (*cp == '\0')
*fsec = 0;
else if (*cp == '.')
{
char fstr[7];
int i;
cp++;
/*
* OK, we have at most six digits to care about. Let's construct a
* string with those digits, zero-padded on the right, and then do
* the conversion to an integer.
*
* XXX This truncates the seventh digit, unlike rounding it as the
* backend does.
*/
for (i = 0; i < 6; i++)
fstr[i] = *cp != '\0' ? *cp++ : '0';
fstr[i] = '\0';
*fsec = strtoint(fstr, &cp, 10);
if (*cp != '\0')
return -1;
}
else
return -1;
}
/* do a sanity check */
if (tm->tm_hour < 0 || tm->tm_min < 0 || tm->tm_min > 59 ||
tm->tm_sec < 0 || tm->tm_sec > 59 || *fsec >= USECS_PER_SEC)
return -1;
return 0;
} /* DecodeTime() */
/* DecodeTimezone()
* Interpret string as a numeric timezone.
*
* Note: we allow timezone offsets up to 13:59. There are places that
* use +1300 summer time.
*/
static int
DecodeTimezone(char *str, int *tzp)
{
int tz;
int hr,
min;
char *cp;
int len;
/* assume leading character is "+" or "-" */
hr = strtoint(str + 1, &cp, 10);
/* explicit delimiter? */
if (*cp == ':')
min = strtoint(cp + 1, &cp, 10);
/* otherwise, might have run things together... */
else if (*cp == '\0' && (len = strlen(str)) > 3)
{
min = strtoint(str + len - 2, &cp, 10);
if (min < 0 || min >= 60)
return -1;
*(str + len - 2) = '\0';
hr = strtoint(str + 1, &cp, 10);
if (hr < 0 || hr > 13)
return -1;
}
else
min = 0;
tz = (hr * MINS_PER_HOUR + min) * SECS_PER_MINUTE;
if (*str == '-')
tz = -tz;
*tzp = -tz;
return *cp != '\0';
} /* DecodeTimezone() */
/* DecodePosixTimezone()
* Interpret string as a POSIX-compatible timezone:
* PST-hh:mm
* PST+h
* - thomas 2000-03-15
*/
static int
DecodePosixTimezone(char *str, int *tzp)
{
int val,
tz;
int type;
char *cp;
char delim;
cp = str;
while (*cp != '\0' && isalpha((unsigned char) *cp))
cp++;
if (DecodeTimezone(cp, &tz) != 0)
return -1;
delim = *cp;
*cp = '\0';
type = DecodeSpecial(MAXDATEFIELDS - 1, str, &val);
*cp = delim;
switch (type)
{
case DTZ:
case TZ:
*tzp = -(val + tz);
break;
default:
return -1;
}
return 0;
} /* DecodePosixTimezone() */
/* ParseDateTime()
* Break string into tokens based on a date/time context.
* Several field types are assigned:
* DTK_NUMBER - digits and (possibly) a decimal point
* DTK_DATE - digits and two delimiters, or digits and text
* DTK_TIME - digits, colon delimiters, and possibly a decimal point
* DTK_STRING - text (no digits)
* DTK_SPECIAL - leading "+" or "-" followed by text
* DTK_TZ - leading "+" or "-" followed by digits
* Note that some field types can hold unexpected items:
* DTK_NUMBER can hold date fields (yy.ddd)
* DTK_STRING can hold months (January) and time zones (PST)
* DTK_DATE can hold Posix time zones (GMT-8)
*
* The "lowstr" work buffer must have at least strlen(timestr) + MAXDATEFIELDS
* bytes of space. On output, field[] entries will point into it.
* The field[] and ftype[] arrays must have at least MAXDATEFIELDS entries.
*/
int
ParseDateTime(char *timestr, char *lowstr,
char **field, int *ftype, int *numfields, char **endstr)
{
int nf = 0;
char *lp = lowstr;
*endstr = timestr;
/* outer loop through fields */
while (*(*endstr) != '\0')
{
/* Record start of current field */
if (nf >= MAXDATEFIELDS)
return -1;
field[nf] = lp;
/* leading digit? then date or time */
if (isdigit((unsigned char) *(*endstr)))
{
*lp++ = *(*endstr)++;
while (isdigit((unsigned char) *(*endstr)))
*lp++ = *(*endstr)++;
/* time field? */
if (*(*endstr) == ':')
{
ftype[nf] = DTK_TIME;
*lp++ = *(*endstr)++;
while (isdigit((unsigned char) *(*endstr)) ||
(*(*endstr) == ':') || (*(*endstr) == '.'))
*lp++ = *(*endstr)++;
}
/* date field? allow embedded text month */
else if (*(*endstr) == '-' || *(*endstr) == '/' || *(*endstr) == '.')
{
/* save delimiting character to use later */
char *dp = (*endstr);
*lp++ = *(*endstr)++;
/* second field is all digits? then no embedded text month */
if (isdigit((unsigned char) *(*endstr)))
{
ftype[nf] = (*dp == '.') ? DTK_NUMBER : DTK_DATE;
while (isdigit((unsigned char) *(*endstr)))
*lp++ = *(*endstr)++;
/*
* insist that the delimiters match to get a three-field
* date.
*/
if (*(*endstr) == *dp)
{
ftype[nf] = DTK_DATE;
*lp++ = *(*endstr)++;
while (isdigit((unsigned char) *(*endstr)) || (*(*endstr) == *dp))
*lp++ = *(*endstr)++;
}
}
else
{
ftype[nf] = DTK_DATE;
while (isalnum((unsigned char) *(*endstr)) || (*(*endstr) == *dp))
*lp++ = pg_tolower((unsigned char) *(*endstr)++);
}
}
/*
* otherwise, number only and will determine year, month, day, or
* concatenated fields later...
*/
else
ftype[nf] = DTK_NUMBER;
}
/* Leading decimal point? Then fractional seconds... */
else if (*(*endstr) == '.')
{
*lp++ = *(*endstr)++;
while (isdigit((unsigned char) *(*endstr)))
*lp++ = *(*endstr)++;
ftype[nf] = DTK_NUMBER;
}
/*
* text? then date string, month, day of week, special, or timezone
*/
else if (isalpha((unsigned char) *(*endstr)))
{
ftype[nf] = DTK_STRING;
*lp++ = pg_tolower((unsigned char) *(*endstr)++);
while (isalpha((unsigned char) *(*endstr)))
*lp++ = pg_tolower((unsigned char) *(*endstr)++);
/*
* Full date string with leading text month? Could also be a POSIX
* time zone...
*/
if (*(*endstr) == '-' || *(*endstr) == '/' || *(*endstr) == '.')
{
char *dp = (*endstr);
ftype[nf] = DTK_DATE;
*lp++ = *(*endstr)++;
while (isdigit((unsigned char) *(*endstr)) || *(*endstr) == *dp)
*lp++ = *(*endstr)++;
}
}
/* skip leading spaces */
else if (isspace((unsigned char) *(*endstr)))
{
(*endstr)++;
continue;
}
/* sign? then special or numeric timezone */
else if (*(*endstr) == '+' || *(*endstr) == '-')
{
*lp++ = *(*endstr)++;
/* soak up leading whitespace */
while (isspace((unsigned char) *(*endstr)))
(*endstr)++;
/* numeric timezone? */
if (isdigit((unsigned char) *(*endstr)))
{
ftype[nf] = DTK_TZ;
*lp++ = *(*endstr)++;
while (isdigit((unsigned char) *(*endstr)) ||
(*(*endstr) == ':') || (*(*endstr) == '.'))
*lp++ = *(*endstr)++;
}
/* special? */
else if (isalpha((unsigned char) *(*endstr)))
{
ftype[nf] = DTK_SPECIAL;
*lp++ = pg_tolower((unsigned char) *(*endstr)++);
while (isalpha((unsigned char) *(*endstr)))
*lp++ = pg_tolower((unsigned char) *(*endstr)++);
}
/* otherwise something wrong... */
else
return -1;
}
/* ignore punctuation but use as delimiter */
else if (ispunct((unsigned char) *(*endstr)))
{
(*endstr)++;
continue;
}
/* otherwise, something is not right... */
else
return -1;
/* force in a delimiter after each field */
*lp++ = '\0';
nf++;
}
*numfields = nf;
return 0;
} /* ParseDateTime() */
/* DecodeDateTime()
* Interpret previously parsed fields for general date and time.
* Return 0 if full date, 1 if only time, and -1 if problems.
* External format(s):
* "<weekday> <month>-<day>-<year> <hour>:<minute>:<second>"
* "Fri Feb-7-1997 15:23:27"
* "Feb-7-1997 15:23:27"
* "2-7-1997 15:23:27"
* "1997-2-7 15:23:27"
* "1997.038 15:23:27" (day of year 1-366)
* Also supports input in compact time:
* "970207 152327"
* "97038 152327"
* "20011225T040506.789-07"
*
* Use the system-provided functions to get the current time zone
* if not specified in the input string.
* If the date is outside the time_t system-supported time range,
* then assume UTC time zone. - thomas 1997-05-27
*/
int
DecodeDateTime(char **field, int *ftype, int nf,
int *dtype, struct tm *tm, fsec_t *fsec, bool EuroDates)
{
int fmask = 0,
tmask,
type;
int ptype = 0; /* "prefix type" for ISO y2001m02d04 format */
int i;
int val;
int mer = HR24;
bool haveTextMonth = false;
bool is2digits = false;
bool bc = false;
int t = 0;
int *tzp = &t;
/***
* We'll insist on at least all of the date fields, but initialize the
* remaining fields in case they are not set later...
***/
*dtype = DTK_DATE;
tm->tm_hour = 0;
tm->tm_min = 0;
tm->tm_sec = 0;
*fsec = 0;
/* don't know daylight savings time status apriori */
tm->tm_isdst = -1;
if (tzp != NULL)
*tzp = 0;
for (i = 0; i < nf; i++)
{
switch (ftype[i])
{
case DTK_DATE:
/***
* Integral julian day with attached time zone?
* All other forms with JD will be separated into
* distinct fields, so we handle just this case here.
***/
if (ptype == DTK_JULIAN)
{
char *cp;
int val;
if (tzp == NULL)
return -1;
val = strtoint(field[i], &cp, 10);
if (*cp != '-')
return -1;
j2date(val, &tm->tm_year, &tm->tm_mon, &tm->tm_mday);
/* Get the time zone from the end of the string */
if (DecodeTimezone(cp, tzp) != 0)
return -1;
tmask = DTK_DATE_M | DTK_TIME_M | DTK_M(TZ);
ptype = 0;
break;
}
/***
* Already have a date? Then this might be a POSIX time
* zone with an embedded dash (e.g. "PST-3" == "EST") or
* a run-together time with trailing time zone (e.g. hhmmss-zz).
* - thomas 2001-12-25
***/
else if (((fmask & DTK_DATE_M) == DTK_DATE_M)
|| (ptype != 0))
{
/* No time zone accepted? Then quit... */
if (tzp == NULL)
return -1;
if (isdigit((unsigned char) *field[i]) || ptype != 0)
{
char *cp;
if (ptype != 0)
{
/* Sanity check; should not fail this test */
if (ptype != DTK_TIME)
return -1;
ptype = 0;
}
/*
* Starts with a digit but we already have a time
* field? Then we are in trouble with a date and time
* already...
*/
if ((fmask & DTK_TIME_M) == DTK_TIME_M)
return -1;
if ((cp = strchr(field[i], '-')) == NULL)
return -1;
/* Get the time zone from the end of the string */
if (DecodeTimezone(cp, tzp) != 0)
return -1;
*cp = '\0';
/*
* Then read the rest of the field as a concatenated
* time
*/
if ((ftype[i] = DecodeNumberField(strlen(field[i]), field[i], fmask,
&tmask, tm, fsec, &is2digits)) < 0)
return -1;
/*
* modify tmask after returning from
* DecodeNumberField()
*/
tmask |= DTK_M(TZ);
}
else
{
if (DecodePosixTimezone(field[i], tzp) != 0)
return -1;
ftype[i] = DTK_TZ;
tmask = DTK_M(TZ);
}
}
else if (DecodeDate(field[i], fmask, &tmask, tm, EuroDates) != 0)
return -1;
break;
case DTK_TIME:
if (DecodeTime(field[i], &tmask, tm, fsec) != 0)
return -1;
/*
* Check upper limit on hours; other limits checked in
* DecodeTime()
*/
/* test for > 24:00:00 */
if (tm->tm_hour > 24 ||
(tm->tm_hour == 24 && (tm->tm_min > 0 || tm->tm_sec > 0)))
return -1;
break;
case DTK_TZ:
{
int tz;
if (tzp == NULL)
return -1;
if (DecodeTimezone(field[i], &tz) != 0)
return -1;
/*
* Already have a time zone? Then maybe this is the second
* field of a POSIX time: EST+3 (equivalent to PST)
*/
if (i > 0 && (fmask & DTK_M(TZ)) != 0 &&
ftype[i - 1] == DTK_TZ &&
isalpha((unsigned char) *field[i - 1]))
{
*tzp -= tz;
tmask = 0;
}
else
{
*tzp = tz;
tmask = DTK_M(TZ);
}
}
break;
case DTK_NUMBER:
/*
* Was this an "ISO date" with embedded field labels? An
* example is "y2001m02d04" - thomas 2001-02-04
*/
if (ptype != 0)
{
char *cp;
int val;
val = strtoint(field[i], &cp, 10);
/*
* only a few kinds are allowed to have an embedded
* decimal
*/
if (*cp == '.')
switch (ptype)
{
case DTK_JULIAN:
case DTK_TIME:
case DTK_SECOND:
break;
default:
return 1;
break;
}
else if (*cp != '\0')
return -1;
switch (ptype)
{
case DTK_YEAR:
tm->tm_year = val;
tmask = DTK_M(YEAR);
break;
case DTK_MONTH:
/*
* already have a month and hour? then assume
* minutes
*/
if ((fmask & DTK_M(MONTH)) != 0 &&
(fmask & DTK_M(HOUR)) != 0)
{
tm->tm_min = val;
tmask = DTK_M(MINUTE);
}
else
{
tm->tm_mon = val;
tmask = DTK_M(MONTH);
}
break;
case DTK_DAY:
tm->tm_mday = val;
tmask = DTK_M(DAY);
break;
case DTK_HOUR:
tm->tm_hour = val;
tmask = DTK_M(HOUR);
break;
case DTK_MINUTE:
tm->tm_min = val;
tmask = DTK_M(MINUTE);
break;
case DTK_SECOND:
tm->tm_sec = val;
tmask = DTK_M(SECOND);
if (*cp == '.')
{
double frac;
frac = strtod(cp, &cp);
if (*cp != '\0')
return -1;
*fsec = frac * 1000000;
}
break;
case DTK_TZ:
tmask = DTK_M(TZ);
if (DecodeTimezone(field[i], tzp) != 0)
return -1;
break;
case DTK_JULIAN:
/***
* previous field was a label for "julian date"?
***/
tmask = DTK_DATE_M;
j2date(val, &tm->tm_year, &tm->tm_mon, &tm->tm_mday);
/* fractional Julian Day? */
if (*cp == '.')
{
double time;
time = strtod(cp, &cp);
if (*cp != '\0')
return -1;
tmask |= DTK_TIME_M;
dt2time((time * USECS_PER_DAY), &tm->tm_hour, &tm->tm_min, &tm->tm_sec, fsec);
}
break;
case DTK_TIME:
/* previous field was "t" for ISO time */
if ((ftype[i] = DecodeNumberField(strlen(field[i]), field[i], (fmask | DTK_DATE_M),
&tmask, tm, fsec, &is2digits)) < 0)
return -1;
if (tmask != DTK_TIME_M)
return -1;
break;
default:
return -1;
break;
}
ptype = 0;
*dtype = DTK_DATE;
}
else
{
char *cp;
int flen;
flen = strlen(field[i]);
cp = strchr(field[i], '.');
/* Embedded decimal and no date yet? */
if (cp != NULL && !(fmask & DTK_DATE_M))
{
if (DecodeDate(field[i], fmask, &tmask, tm, EuroDates) != 0)
return -1;
}
/* embedded decimal and several digits before? */
else if (cp != NULL && flen - strlen(cp) > 2)
{
/*
* Interpret as a concatenated date or time Set the
* type field to allow decoding other fields later.
* Example: 20011223 or 040506
*/
if ((ftype[i] = DecodeNumberField(flen, field[i], fmask,
&tmask, tm, fsec, &is2digits)) < 0)
return -1;
}
else if (flen > 4)
{
if ((ftype[i] = DecodeNumberField(flen, field[i], fmask,
&tmask, tm, fsec, &is2digits)) < 0)
return -1;
}
/* otherwise it is a single date/time field... */
else if (DecodeNumber(flen, field[i], fmask,
&tmask, tm, fsec, &is2digits, EuroDates) != 0)
return -1;
}
break;
case DTK_STRING:
case DTK_SPECIAL:
type = DecodeSpecial(i, field[i], &val);
if (type == IGNORE_DTF)
continue;
tmask = DTK_M(type);
switch (type)
{
case RESERV:
switch (val)
{
case DTK_NOW:
tmask = (DTK_DATE_M | DTK_TIME_M | DTK_M(TZ));
*dtype = DTK_DATE;
GetCurrentDateTime(tm);
break;
case DTK_YESTERDAY:
tmask = DTK_DATE_M;
*dtype = DTK_DATE;
GetCurrentDateTime(tm);
j2date(date2j(tm->tm_year, tm->tm_mon, tm->tm_mday) - 1,
&tm->tm_year, &tm->tm_mon, &tm->tm_mday);
tm->tm_hour = 0;
tm->tm_min = 0;
tm->tm_sec = 0;
break;
case DTK_TODAY:
tmask = DTK_DATE_M;
*dtype = DTK_DATE;
GetCurrentDateTime(tm);
tm->tm_hour = 0;
tm->tm_min = 0;
tm->tm_sec = 0;
break;
case DTK_TOMORROW:
tmask = DTK_DATE_M;
*dtype = DTK_DATE;
GetCurrentDateTime(tm);
j2date(date2j(tm->tm_year, tm->tm_mon, tm->tm_mday) + 1,
&tm->tm_year, &tm->tm_mon, &tm->tm_mday);
tm->tm_hour = 0;
tm->tm_min = 0;
tm->tm_sec = 0;
break;
case DTK_ZULU:
tmask = (DTK_TIME_M | DTK_M(TZ));
*dtype = DTK_DATE;
tm->tm_hour = 0;
tm->tm_min = 0;
tm->tm_sec = 0;
if (tzp != NULL)
*tzp = 0;
break;
default:
*dtype = val;
}
break;
case MONTH:
/*
* already have a (numeric) month? then see if we can
* substitute...
*/
if ((fmask & DTK_M(MONTH)) && !haveTextMonth &&
!(fmask & DTK_M(DAY)) && tm->tm_mon >= 1 && tm->tm_mon <= 31)
{
tm->tm_mday = tm->tm_mon;
tmask = DTK_M(DAY);
}
haveTextMonth = true;
tm->tm_mon = val;
break;
case DTZMOD:
/*
* daylight savings time modifier (solves "MET DST"
* syntax)
*/
tmask |= DTK_M(DTZ);
tm->tm_isdst = 1;
if (tzp == NULL)
return -1;
*tzp -= val;
break;
case DTZ:
/*
* set mask for TZ here _or_ check for DTZ later when
* getting default timezone
*/
tmask |= DTK_M(TZ);
tm->tm_isdst = 1;
if (tzp == NULL)
return -1;
*tzp = -val;
ftype[i] = DTK_TZ;
break;
case TZ:
tm->tm_isdst = 0;
if (tzp == NULL)
return -1;
*tzp = -val;
ftype[i] = DTK_TZ;
break;
case IGNORE_DTF:
break;
case AMPM:
mer = val;
break;
case ADBC:
bc = (val == BC);
break;
case DOW:
tm->tm_wday = val;
break;
case UNITS:
tmask = 0;
ptype = val;
break;
case ISOTIME:
/*
* This is a filler field "t" indicating that the next
* field is time. Try to verify that this is sensible.
*/
tmask = 0;
/* No preceding date? Then quit... */
if ((fmask & DTK_DATE_M) != DTK_DATE_M)
return -1;
/***
* We will need one of the following fields:
* DTK_NUMBER should be hhmmss.fff
* DTK_TIME should be hh:mm:ss.fff
* DTK_DATE should be hhmmss-zz
***/
if (i >= nf - 1 ||
(ftype[i + 1] != DTK_NUMBER &&
ftype[i + 1] != DTK_TIME &&
ftype[i + 1] != DTK_DATE))
return -1;
ptype = val;
break;
default:
return -1;
}
break;
default:
return -1;
}
if (tmask & fmask)
return -1;
fmask |= tmask;
}
/* there is no year zero in AD/BC notation; i.e. "1 BC" == year 0 */
if (bc)
{
if (tm->tm_year > 0)
tm->tm_year = -(tm->tm_year - 1);
else
return -1;
}
else if (is2digits)
{
if (tm->tm_year < 70)
tm->tm_year += 2000;
else if (tm->tm_year < 100)
tm->tm_year += 1900;
}
if (mer != HR24 && tm->tm_hour > 12)
return -1;
if (mer == AM && tm->tm_hour == 12)
tm->tm_hour = 0;
else if (mer == PM && tm->tm_hour != 12)
tm->tm_hour += 12;
/* do additional checking for full date specs... */
if (*dtype == DTK_DATE)
{
if ((fmask & DTK_DATE_M) != DTK_DATE_M)
return ((fmask & DTK_TIME_M) == DTK_TIME_M) ? 1 : -1;
/*
* check for valid day of month, now that we know for sure the month
* and year...
*/
if (tm->tm_mday < 1 || tm->tm_mday > day_tab[isleap(tm->tm_year)][tm->tm_mon - 1])
return -1;
/*
* backend tried to find local timezone here but we don't use the
* result afterwards anyway so we only check for this error: daylight
* savings time modifier but no standard timezone?
*/
if ((fmask & DTK_DATE_M) == DTK_DATE_M && tzp != NULL && !(fmask & DTK_M(TZ)) && (fmask & DTK_M(DTZMOD)))
return -1;
}
return 0;
} /* DecodeDateTime() */
/* Function works as follows:
*
*
* */
static char *
find_end_token(char *str, char *fmt)
{
/*
* str: here is28the day12the hour fmt: here is%dthe day%hthe hour
*
* we extract the 28, we read the percent sign and the type "d" then this
* functions gets called as find_end_token("28the day12the hour", "the
* day%hthehour")
*
* fmt points to "the day%hthehour", next_percent points to %hthehour and
* we have to find a match for everything between these positions ("the
* day"). We look for "the day" in str and know that the pattern we are
* about to scan ends where this string starts (right after the "28")
*
* At the end, *fmt is '\0' and *str isn't. end_position then is
* unchanged.
*/
char *end_position = NULL;
char *next_percent,
*subst_location = NULL;
int scan_offset = 0;
char last_char;
/* are we at the end? */
if (!*fmt)
{
end_position = fmt;
return end_position;
}
/* not at the end */
while (fmt[scan_offset] == '%' && fmt[scan_offset + 1])
{
/*
* there is no delimiter, skip to the next delimiter if we're reading
* a number and then something that is not a number "9:15pm", we might
* be able to recover with the strtol end pointer. Go for the next
* percent sign
*/
scan_offset += 2;
}
next_percent = strchr(fmt + scan_offset, '%');
if (next_percent)
{
/*
* we don't want to allocate extra memory, so we temporarily set the
* '%' sign to '\0' and call strstr However since we allow whitespace
* to float around everything, we have to shorten the pattern until we
* reach a non-whitespace character
*/
subst_location = next_percent;
while (*(subst_location - 1) == ' ' && subst_location - 1 > fmt + scan_offset)
subst_location--;
last_char = *subst_location;
*subst_location = '\0';
/*
* the haystack is the str and the needle is the original fmt but it
* ends at the position where the next percent sign would be
*/
/*
* There is one special case. Imagine: str = " 2", fmt = "%d %...",
* since we want to allow blanks as "dynamic" padding we have to
* accept this. Now, we are called with a fmt of " %..." and look for
* " " in str. We find it at the first position and never read the
* 2...
*/
while (*str == ' ')
str++;
end_position = strstr(str, fmt + scan_offset);
*subst_location = last_char;
}
else
{
/*
* there is no other percent sign. So everything up to the end has to
* match.
*/
end_position = str + strlen(str);
}
if (!end_position)
{
/*
* maybe we have the following case:
*
* str = "4:15am" fmt = "%M:%S %p"
*
* at this place we could have
*
* str = "15am" fmt = " %p"
*
* and have set fmt to " " because overwrote the % sign with a NULL
*
* In this case where we would have to match a space but can't find
* it, set end_position to the end of the string
*/
if ((fmt + scan_offset)[0] == ' ' && fmt + scan_offset + 1 == subst_location)
end_position = str + strlen(str);
}
return end_position;
}
static int
pgtypes_defmt_scan(union un_fmt_comb *scan_val, int scan_type, char **pstr, char *pfmt)
{
/*
* scan everything between pstr and pstr_end. This is not including the
* last character so we might set it to '\0' for the parsing
*/
char last_char;
int err = 0;
char *pstr_end;
char *strtol_end = NULL;
while (**pstr == ' ')
pstr++;
pstr_end = find_end_token(*pstr, pfmt);
if (!pstr_end)
{
/* there was an error, no match */
return 1;
}
last_char = *pstr_end;
*pstr_end = '\0';
switch (scan_type)
{
case PGTYPES_TYPE_UINT:
/*
* numbers may be blank-padded, this is the only deviation from
* the fmt-string we accept
*/
while (**pstr == ' ')
(*pstr)++;
errno = 0;
scan_val->uint_val = (unsigned int) strtol(*pstr, &strtol_end, 10);
if (errno)
err = 1;
break;
case PGTYPES_TYPE_UINT_LONG:
while (**pstr == ' ')
(*pstr)++;
errno = 0;
scan_val->luint_val = (unsigned long int) strtol(*pstr, &strtol_end, 10);
if (errno)
err = 1;
break;
case PGTYPES_TYPE_STRING_MALLOCED:
scan_val->str_val = pgtypes_strdup(*pstr);
if (scan_val->str_val == NULL)
err = 1;
break;
}
if (strtol_end && *strtol_end)
*pstr = strtol_end;
else
*pstr = pstr_end;
*pstr_end = last_char;
return err;
}
/* XXX range checking */
int
PGTYPEStimestamp_defmt_scan(char **str, char *fmt, timestamp * d,
int *year, int *month, int *day,
int *hour, int *minute, int *second,
int *tz)
{
union un_fmt_comb scan_val;
int scan_type;
char *pstr,
*pfmt,
*tmp;
int err = 1;
unsigned int j;
struct tm tm;
pfmt = fmt;
pstr = *str;
while (*pfmt)
{
err = 0;
while (*pfmt == ' ')
pfmt++;
while (*pstr == ' ')
pstr++;
if (*pfmt != '%')
{
if (*pfmt == *pstr)
{
pfmt++;
pstr++;
}
else
{
/* Error: no match */
err = 1;
return err;
}
continue;
}
/* here *pfmt equals '%' */
pfmt++;
switch (*pfmt)
{
case 'a':
pfmt++;
/*
* we parse the day and see if it is a week day but we do not
* check if the week day really matches the date
*/
err = 1;
j = 0;
while (pgtypes_date_weekdays_short[j])
{
if (strncmp(pgtypes_date_weekdays_short[j], pstr,
strlen(pgtypes_date_weekdays_short[j])) == 0)
{
/* found it */
err = 0;
pstr += strlen(pgtypes_date_weekdays_short[j]);
break;
}
j++;
}
break;
case 'A':
/* see note above */
pfmt++;
err = 1;
j = 0;
while (days[j])
{
if (strncmp(days[j], pstr, strlen(days[j])) == 0)
{
/* found it */
err = 0;
pstr += strlen(days[j]);
break;
}
j++;
}
break;
case 'b':
case 'h':
pfmt++;
err = 1;
j = 0;
while (months[j])
{
if (strncmp(months[j], pstr, strlen(months[j])) == 0)
{
/* found it */
err = 0;
pstr += strlen(months[j]);
*month = j + 1;
break;
}
j++;
}
break;
case 'B':
/* see note above */
pfmt++;
err = 1;
j = 0;
while (pgtypes_date_months[j])
{
if (strncmp(pgtypes_date_months[j], pstr, strlen(pgtypes_date_months[j])) == 0)
{
/* found it */
err = 0;
pstr += strlen(pgtypes_date_months[j]);
*month = j + 1;
break;
}
j++;
}
break;
case 'c':
/* XXX */
break;
case 'C':
pfmt++;
scan_type = PGTYPES_TYPE_UINT;
err = pgtypes_defmt_scan(&scan_val, scan_type, &pstr, pfmt);
*year = scan_val.uint_val * 100;
break;
case 'd':
case 'e':
pfmt++;
scan_type = PGTYPES_TYPE_UINT;
err = pgtypes_defmt_scan(&scan_val, scan_type, &pstr, pfmt);
*day = scan_val.uint_val;
break;
case 'D':
/*
* we have to concatenate the strings in order to be able to
* find the end of the substitution
*/
pfmt++;
tmp = pgtypes_alloc(strlen("%m/%d/%y") + strlen(pstr) + 1);
strcpy(tmp, "%m/%d/%y");
strcat(tmp, pfmt);
err = PGTYPEStimestamp_defmt_scan(&pstr, tmp, d, year, month, day, hour, minute, second, tz);
free(tmp);
return err;
case 'm':
pfmt++;
scan_type = PGTYPES_TYPE_UINT;
err = pgtypes_defmt_scan(&scan_val, scan_type, &pstr, pfmt);
*month = scan_val.uint_val;
break;
case 'y':
case 'g': /* XXX difference to y (ISO) */
pfmt++;
scan_type = PGTYPES_TYPE_UINT;
err = pgtypes_defmt_scan(&scan_val, scan_type, &pstr, pfmt);
if (*year < 0)
{
/* not yet set */
*year = scan_val.uint_val;
}
else
*year += scan_val.uint_val;
if (*year < 100)
*year += 1900;
break;
case 'G':
/* XXX difference to %V (ISO) */
pfmt++;
scan_type = PGTYPES_TYPE_UINT;
err = pgtypes_defmt_scan(&scan_val, scan_type, &pstr, pfmt);
*year = scan_val.uint_val;
break;
case 'H':
case 'I':
case 'k':
case 'l':
pfmt++;
scan_type = PGTYPES_TYPE_UINT;
err = pgtypes_defmt_scan(&scan_val, scan_type, &pstr, pfmt);
*hour += scan_val.uint_val;
break;
case 'j':
pfmt++;
scan_type = PGTYPES_TYPE_UINT;
err = pgtypes_defmt_scan(&scan_val, scan_type, &pstr, pfmt);
/*
* XXX what should we do with that? We could say that it's
* sufficient if we have the year and the day within the year
* to get at least a specific day.
*/
break;
case 'M':
pfmt++;
scan_type = PGTYPES_TYPE_UINT;
err = pgtypes_defmt_scan(&scan_val, scan_type, &pstr, pfmt);
*minute = scan_val.uint_val;
break;
case 'n':
pfmt++;
if (*pstr == '\n')
pstr++;
else
err = 1;
break;
case 'p':
err = 1;
pfmt++;
if (strncmp(pstr, "am", 2) == 0)
{
*hour += 0;
err = 0;
pstr += 2;
}
if (strncmp(pstr, "a.m.", 4) == 0)
{
*hour += 0;
err = 0;
pstr += 4;
}
if (strncmp(pstr, "pm", 2) == 0)
{
*hour += 12;
err = 0;
pstr += 2;
}
if (strncmp(pstr, "p.m.", 4) == 0)
{
*hour += 12;
err = 0;
pstr += 4;
}
break;
case 'P':
err = 1;
pfmt++;
if (strncmp(pstr, "AM", 2) == 0)
{
*hour += 0;
err = 0;
pstr += 2;
}
if (strncmp(pstr, "A.M.", 4) == 0)
{
*hour += 0;
err = 0;
pstr += 4;
}
if (strncmp(pstr, "PM", 2) == 0)
{
*hour += 12;
err = 0;
pstr += 2;
}
if (strncmp(pstr, "P.M.", 4) == 0)
{
*hour += 12;
err = 0;
pstr += 4;
}
break;
case 'r':
pfmt++;
tmp = pgtypes_alloc(strlen("%I:%M:%S %p") + strlen(pstr) + 1);
strcpy(tmp, "%I:%M:%S %p");
strcat(tmp, pfmt);
err = PGTYPEStimestamp_defmt_scan(&pstr, tmp, d, year, month, day, hour, minute, second, tz);
free(tmp);
return err;
case 'R':
pfmt++;
tmp = pgtypes_alloc(strlen("%H:%M") + strlen(pstr) + 1);
strcpy(tmp, "%H:%M");
strcat(tmp, pfmt);
err = PGTYPEStimestamp_defmt_scan(&pstr, tmp, d, year, month, day, hour, minute, second, tz);
free(tmp);
return err;
case 's':
pfmt++;
scan_type = PGTYPES_TYPE_UINT_LONG;
err = pgtypes_defmt_scan(&scan_val, scan_type, &pstr, pfmt);
/* number of seconds in scan_val.luint_val */
{
struct tm *tms;
time_t et = (time_t) scan_val.luint_val;
tms = gmtime(&et);
if (tms)
{
*year = tms->tm_year + 1900;
*month = tms->tm_mon + 1;
*day = tms->tm_mday;
*hour = tms->tm_hour;
*minute = tms->tm_min;
*second = tms->tm_sec;
}
else
err = 1;
}
break;
case 'S':
pfmt++;
scan_type = PGTYPES_TYPE_UINT;
err = pgtypes_defmt_scan(&scan_val, scan_type, &pstr, pfmt);
*second = scan_val.uint_val;
break;
case 't':
pfmt++;
if (*pstr == '\t')
pstr++;
else
err = 1;
break;
case 'T':
pfmt++;
tmp = pgtypes_alloc(strlen("%H:%M:%S") + strlen(pstr) + 1);
strcpy(tmp, "%H:%M:%S");
strcat(tmp, pfmt);
err = PGTYPEStimestamp_defmt_scan(&pstr, tmp, d, year, month, day, hour, minute, second, tz);
free(tmp);
return err;
case 'u':
pfmt++;
scan_type = PGTYPES_TYPE_UINT;
err = pgtypes_defmt_scan(&scan_val, scan_type, &pstr, pfmt);
if (scan_val.uint_val < 1 || scan_val.uint_val > 7)
err = 1;
break;
case 'U':
pfmt++;
scan_type = PGTYPES_TYPE_UINT;
err = pgtypes_defmt_scan(&scan_val, scan_type, &pstr, pfmt);
if (scan_val.uint_val > 53)
err = 1;
break;
case 'V':
pfmt++;
scan_type = PGTYPES_TYPE_UINT;
err = pgtypes_defmt_scan(&scan_val, scan_type, &pstr, pfmt);
if (scan_val.uint_val < 1 || scan_val.uint_val > 53)
err = 1;
break;
case 'w':
pfmt++;
scan_type = PGTYPES_TYPE_UINT;
err = pgtypes_defmt_scan(&scan_val, scan_type, &pstr, pfmt);
if (scan_val.uint_val > 6)
err = 1;
break;
case 'W':
pfmt++;
scan_type = PGTYPES_TYPE_UINT;
err = pgtypes_defmt_scan(&scan_val, scan_type, &pstr, pfmt);
if (scan_val.uint_val > 53)
err = 1;
break;
case 'x':
case 'X':
/* XXX */
break;
case 'Y':
pfmt++;
scan_type = PGTYPES_TYPE_UINT;
err = pgtypes_defmt_scan(&scan_val, scan_type, &pstr, pfmt);
*year = scan_val.uint_val;
break;
case 'z':
pfmt++;
scan_type = PGTYPES_TYPE_STRING_MALLOCED;
err = pgtypes_defmt_scan(&scan_val, scan_type, &pstr, pfmt);
if (!err)
{
err = DecodeTimezone(scan_val.str_val, tz);
free(scan_val.str_val);
}
break;
case 'Z':
pfmt++;
scan_type = PGTYPES_TYPE_STRING_MALLOCED;
err = pgtypes_defmt_scan(&scan_val, scan_type, &pstr, pfmt);
if (!err)
{
/*
* XXX use DecodeSpecial instead? Do we need strcasecmp
* here?
*/
err = 1;
for (j = 0; j < szdatetktbl; j++)
{
if ((datetktbl[j].type == TZ || datetktbl[j].type == DTZ) &&
pg_strcasecmp(datetktbl[j].token,
scan_val.str_val) == 0)
{
*tz = -datetktbl[j].value;
err = 0;
break;
}
}
free(scan_val.str_val);
}
break;
case '+':
/* XXX */
break;
case '%':
pfmt++;
if (*pstr == '%')
pstr++;
else
err = 1;
break;
default:
err = 1;
}
}
if (!err)
{
if (*second < 0)
*second = 0;
if (*minute < 0)
*minute = 0;
if (*hour < 0)
*hour = 0;
if (*day < 0)
{
err = 1;
*day = 1;
}
if (*month < 0)
{
err = 1;
*month = 1;
}
if (*year < 0)
{
err = 1;
*year = 1970;
}
if (*second > 59)
{
err = 1;
*second = 0;
}
if (*minute > 59)
{
err = 1;
*minute = 0;
}
if (*hour > 24 || /* test for > 24:00:00 */
(*hour == 24 && (*minute > 0 || *second > 0)))
{
err = 1;
*hour = 0;
}
if (*month > MONTHS_PER_YEAR)
{
err = 1;
*month = 1;
}
if (*day > day_tab[isleap(*year)][*month - 1])
{
*day = day_tab[isleap(*year)][*month - 1];
err = 1;
}
tm.tm_sec = *second;
tm.tm_min = *minute;
tm.tm_hour = *hour;
tm.tm_mday = *day;
tm.tm_mon = *month;
tm.tm_year = *year;
tm2timestamp(&tm, 0, tz, d);
}
return err;
}
/* XXX: 1900 is compiled in as the base for years */