locutus
Version:
Locutus other languages' standard libraries to JavaScript for fun and educational purposes
374 lines (373 loc) • 16.4 kB
JavaScript
;
Object.defineProperty(exports, "__esModule", { value: true });
exports.strptime = strptime;
const _phpRuntimeState_ts_1 = require("../_helpers/_phpRuntimeState.js");
const setlocale_ts_1 = require("../strings/setlocale.js");
const isStringArray = (value, key) => {
const candidate = value[key];
return Array.isArray(candidate) && candidate.every((item) => typeof item === 'string');
};
const isLcTime = (value) => isStringArray(value, 'a') &&
isStringArray(value, 'A') &&
isStringArray(value, 'b') &&
isStringArray(value, 'B') &&
typeof value.c === 'string' &&
typeof value.r === 'string' &&
typeof value.x === 'string' &&
typeof value.X === 'string';
function strptime(dateStr, format) {
// discuss at: https://locutus.io/php/strptime/
// parity verified: PHP 8.3
// original by: Brett Zamir (https://brett-zamir.me)
// original by: strftime
// example 1: strptime('20091112222135', '%Y%m%d%H%M%S') // Return value will depend on date and locale
// returns 1: {tm_sec: 35, tm_min: 21, tm_hour: 22, tm_mday: 12, tm_mon: 10, tm_year: 109, tm_wday: 4, tm_yday: 315, unparsed: ''}
// example 2: strptime('2009extra', '%Y')
// returns 2: {tm_sec:0, tm_min:0, tm_hour:0, tm_mday:0, tm_mon:0, tm_year:109, tm_wday:3, tm_yday: -1, unparsed: 'extra'}
const retObj = {
tm_sec: 0,
tm_min: 0,
tm_hour: 0,
tm_mday: 0,
tm_mon: 0,
tm_year: 0,
tm_wday: 0,
tm_yday: 0,
unparsed: '',
};
let i = 0;
let j = 0;
let amPmOffset = 0;
let prevHour = false;
const _reset = function (dateObj, realMday) {
// realMday is to allow for a value of 0 in return results (but without
// messing up the Date() object)
const o = retObj;
const d = dateObj;
o.tm_sec = d.getUTCSeconds();
o.tm_min = d.getUTCMinutes();
o.tm_hour = d.getUTCHours();
o.tm_mday = realMday === 0 ? realMday : d.getUTCDate();
o.tm_mon = d.getUTCMonth();
o.tm_year = d.getUTCFullYear() - 1900;
o.tm_wday = realMday === 0 ? (d.getUTCDay() > 0 ? d.getUTCDay() - 1 : 6) : d.getUTCDay();
const jan1 = new Date(Date.UTC(d.getUTCFullYear(), 0, 1));
o.tm_yday = Math.ceil((d.getTime() - jan1.getTime()) / (1000 * 60 * 60 * 24));
};
const _date = function () {
const o = retObj;
// We set date to at least 1 to ensure year or month doesn't go backwards
_reset(new Date(Date.UTC(o.tm_year + 1900, o.tm_mon, o.tm_mday || 1, o.tm_hour, o.tm_min, o.tm_sec)), o.tm_mday);
};
const _NWS = /\S/;
const _WS = /\s/;
const _aggregates = {
c: 'locale',
D: '%m/%d/%y',
F: '%y-%m-%d',
r: 'locale',
R: '%H:%M',
T: '%H:%M:%S',
x: 'locale',
X: 'locale',
};
/* Fix: Locale alternatives are supported though not documented in PHP; see https://linux.die.net/man/3/strptime
Ec
EC
Ex
EX
Ey
EY
Od or Oe
OH
OI
Om
OM
OS
OU
Ow
OW
Oy
*/
const _pregQuote = function (str) {
return str.replace(/([\\.+*?[^\]$(){}=!<>|:])/g, '\\$1');
};
// ensure setup of localization variables takes place
(0, setlocale_ts_1.setlocale)('LC_ALL', 0);
const lcTimeGroup = (0, _phpRuntimeState_ts_1.getPhpLocaleGroup)('LC_TIME', 'LC_TIME');
if (!lcTimeGroup || !isLcTime(lcTimeGroup)) {
return false;
}
const lcTime = lcTimeGroup;
// First replace aggregates (run in a loop because an agg may be made up of other aggs)
while (/%[cDFhnrRtTxX]/.test(format)) {
format = format.replace(/%([cDFhnrRtTxX])/g, function (_m0, m1) {
const f = _aggregates[m1];
if (f === undefined) {
return m1;
}
return f === 'locale' ? String(lcTime[m1] ?? '') : f;
});
}
const _addNext = function (index, regex, cb) {
const pattern = typeof regex === 'string' ? new RegExp('^' + regex, 'i') : regex;
const check = dateStr.slice(index);
const match = pattern.exec(check);
if (!match) {
throw new Error('No match in string');
}
// Even if the callback returns null after assigning to the
// return object, the object won't be saved anyways
const testNull = cb(...match);
if (testNull === null) {
throw new Error('No match in string');
}
return index + match[0].length;
};
const _addLocalized = function (index, formatChar, category) {
// Could make each parenthesized instead and pass index to callback:
const localized = lcTime[formatChar];
if (!Array.isArray(localized)) {
throw new Error('No match in string');
}
return _addNext(index, localized.map((entry) => _pregQuote(entry)).join('|'), function (m) {
const matchIndex = localized.findIndex((entry) => new RegExp('^' + _pregQuote(m) + '$', 'i').test(entry));
if (matchIndex === -1) {
return null;
}
retObj[category] = matchIndex;
return matchIndex;
});
};
// BEGIN PROCESSING CHARACTERS
for (i = 0, j = 0; i < format.length; i++) {
if (format.charAt(i) === '%') {
const literalPos = ['%', 'n', 't'].indexOf(format.charAt(i + 1));
if (literalPos !== -1) {
if (['%', '\n', '\t'].indexOf(dateStr.charAt(j)) === literalPos) {
// a matched literal
++i;
// skip beyond
++j;
continue;
}
// Format indicated a percent literal, but not actually present
return false;
}
const formatChar = format.charAt(i + 1);
try {
switch (formatChar) {
case 'a':
case 'A':
// Sunday-Saturday
// Changes nothing else
j = _addLocalized(j, formatChar, 'tm_wday');
break;
case 'h':
case 'b':
// Jan-Dec
j = _addLocalized(j, 'b', 'tm_mon');
// Also changes wday, yday
_date();
break;
case 'B':
// January-December
j = _addLocalized(j, formatChar, 'tm_mon');
// Also changes wday, yday
_date();
break;
case 'C':
// 0+; century (19 for 20th)
// PHP docs say two-digit, but accepts one-digit (two-digit max):
j = _addNext(j, /^\d?\d/, function (d) {
const year = (Number.parseInt(d, 10) - 19) * 100;
retObj.tm_year = year;
_date();
if (!retObj.tm_yday) {
retObj.tm_yday = -1;
}
// Also changes wday; and sets yday to -1 (always?)
});
break;
case 'd':
case 'e':
// 1-31 day
j = _addNext(j, formatChar === 'd' ? /^(0[1-9]|[1-2]\d|3[0-1])/ : /^([1-2]\d|3[0-1]|[1-9])/, function (d) {
const dayMonth = Number.parseInt(d, 10);
retObj.tm_mday = dayMonth;
// Also changes w_day, y_day
_date();
});
break;
case 'g':
// No apparent effect; 2-digit year (see 'V')
break;
case 'G':
// No apparent effect; 4-digit year (see 'V')'
break;
case 'H':
// 00-23 hours
j = _addNext(j, /^([0-1]\d|2[0-3])/, function (d) {
const hour = Number.parseInt(d, 10);
retObj.tm_hour = hour;
// Changes nothing else
});
break;
case 'l':
case 'I':
// 01-12 hours
j = _addNext(j, formatChar === 'l' ? /^([1-9]|1[0-2])/ : /^(0[1-9]|1[0-2])/, function (d) {
const hour = Number.parseInt(d, 10) - 1 + amPmOffset;
retObj.tm_hour = hour;
// Used for coordinating with am-pm
prevHour = true;
// Changes nothing else, but affected by prior 'p/P'
});
break;
case 'j':
// 001-366 day of year
j = _addNext(j, /^(00[1-9]|0[1-9]\d|[1-2]\d\d|3[0-6][0-6])/, function (d) {
const dayYear = Number.parseInt(d, 10) - 1;
retObj.tm_yday = dayYear;
// Changes nothing else
// (oddly, since if original by a given year, could calculate other fields)
});
break;
case 'm':
// 01-12 month
j = _addNext(j, /^(0[1-9]|1[0-2])/, function (d) {
const month = Number.parseInt(d, 10) - 1;
retObj.tm_mon = month;
// Also sets wday and yday
_date();
});
break;
case 'M':
// 00-59 minutes
j = _addNext(j, /^[0-5]\d/, function (d) {
const minute = Number.parseInt(d, 10);
retObj.tm_min = minute;
// Changes nothing else
});
break;
case 'P':
// Seems not to work; AM-PM
// Could make fall-through instead since supposed to be a synonym despite PHP docs
return false;
case 'p':
// am-pm
j = _addNext(j, /^(am|pm)/i, function (d) {
// No effect on 'H' since already 24 hours but
// works before or after setting of l/I hour
amPmOffset = /a/.test(d) ? 0 : 12;
if (prevHour) {
retObj.tm_hour += amPmOffset;
}
});
break;
case 's':
// Unix timestamp (in seconds)
j = _addNext(j, /^\d+/, function (d) {
const timestamp = Number.parseInt(d, 10);
const date = new Date(Date.UTC(timestamp * 1000));
_reset(date, retObj.tm_mday);
// Affects all fields, but can't be negative (and initial + not allowed)
});
break;
case 'S':
// 00-59 seconds
j = _addNext(j, /^[0-5]\d/, // strptime also accepts 60-61 for some reason
function (d) {
const second = Number.parseInt(d, 10);
retObj.tm_sec = second;
// Changes nothing else
});
break;
case 'u':
case 'w':
// 0 (Sunday)-6(Saturday)
j = _addNext(j, /^\d/, function (d) {
retObj.tm_wday = Number(d) - (formatChar === 'u' ? 1 : 0);
// Changes nothing else apparently
});
break;
case 'U':
case 'V':
case 'W':
// Apparently ignored (week of year, from 1st Monday)
break;
case 'y':
// 69 (or higher) for 1969+, 68 (or lower) for 2068-
// PHP docs say two-digit, but accepts one-digit (two-digit max):
j = _addNext(j, /^\d?\d/, function (d) {
const parsed = Number.parseInt(d, 10);
const year = parsed >= 69 ? parsed : parsed + 100;
retObj.tm_year = year;
_date();
if (!retObj.tm_yday) {
retObj.tm_yday = -1;
}
// Also changes wday; and sets yday to -1 (always?)
});
break;
case 'Y':
// 2010 (4-digit year)
// PHP docs say four-digit, but accepts one-digit (four-digit max):
j = _addNext(j, /^\d{1,4}/, function (d) {
const year = Number.parseInt(d, 10) - 1900;
retObj.tm_year = year;
_date();
if (!retObj.tm_yday) {
retObj.tm_yday = -1;
}
// Also changes wday; and sets yday to -1 (always?)
});
break;
case 'z':
// Timezone; on my system, strftime gives -0800,
// but strptime seems not to alter hour setting
break;
case 'Z':
// Timezone; on my system, strftime gives PST, but strptime treats text as unparsed
break;
default:
throw new Error('Unrecognized formatting character in strptime()');
}
}
catch (error) {
if (error instanceof Error && error.message === 'No match in string') {
// Allow us to exit
// There was supposed to be a matching format but there wasn't
return false;
}
// Calculate skipping beyond initial percent too
}
++i;
}
else if (format.charAt(i) !== dateStr.charAt(j)) {
// If extra whitespace at beginning or end of either, or between formats, no problem
// (just a problem when between % and format specifier)
// If the string has white-space, it is ok to ignore
if (_WS.test(dateStr.charAt(j))) {
j++;
// Let the next iteration try again with the same format character
i--;
}
else if (_NWS.test(format.charAt(i))) {
// Any extra formatting characters besides white-space causes
// problems (do check after WS though, as may just be WS in string before next character)
return false;
}
// Extra WS in format
// Adjust strings when encounter non-matching whitespace, so they align in future checks above
// Will check on next iteration (against same (non-WS) string character)
}
else {
j++;
}
}
// POST-PROCESSING
// Will also get extra whitespace; empty string if none
retObj.unparsed = dateStr.slice(j);
return retObj;
}