dateparse/parseany.go

2717 lines
66 KiB
Go
Raw Normal View History

2017-10-04 18:50:25 +08:00
// Package dateparse parses date-strings without knowing the format
// in advance, using a fast lex based approach to eliminate shotgun
// attempts. It leans towards US style dates when there is a conflict.
2014-04-21 10:56:17 +08:00
package dateparse
import (
"fmt"
2014-04-26 07:59:10 +08:00
"strconv"
2018-05-24 12:11:15 +08:00
"strings"
2014-04-21 10:56:17 +08:00
"time"
"unicode"
2017-11-19 03:07:33 +08:00
"unicode/utf8"
2014-04-21 10:56:17 +08:00
)
// func init() {
// gou.SetupLogging("debug")
// gou.SetColorOutput()
// }
var days = []string{
"mon",
"tue",
"wed",
"thu",
"fri",
"sat",
"sun",
"monday",
"tuesday",
"wednesday",
"thursday",
"friday",
"saturday",
"sunday",
}
2018-05-24 12:11:15 +08:00
var months = []string{
"january",
"february",
"march",
"april",
"may",
"june",
"july",
"august",
"september",
"october",
"november",
"december",
}
2018-02-26 08:35:33 +08:00
type dateState uint8
type timeState uint8
2014-05-21 13:08:25 +08:00
2014-04-21 10:56:17 +08:00
const (
dateStart dateState = iota // 0
2018-02-26 08:35:33 +08:00
dateDigit
2021-02-04 14:23:19 +08:00
dateDigitSt
2018-10-10 11:50:09 +08:00
dateYearDash
dateYearDashAlphaDash
dateYearDashDash
dateYearDashDashWs // 6
2018-10-10 11:50:09 +08:00
dateYearDashDashT
dateYearDashDashOffset
2018-02-26 08:35:33 +08:00
dateDigitDash
2018-10-10 11:50:09 +08:00
dateDigitDashAlpha
dateDigitDashAlphaDash // 11
dateDigitDashDigit
dateDigitDashDigitDash
dateDigitDot
2018-02-26 08:35:33 +08:00
dateDigitDotDot
dateDigitDotDotWs
dateDigitDotDotT
dateDigitDotDotOffset
2018-02-26 08:35:33 +08:00
dateDigitSlash
dateDigitYearSlash
dateDigitSlashAlpha // 21
dateDigitColon
2018-02-26 08:35:33 +08:00
dateDigitChineseYear
dateDigitChineseYearWs
dateDigitWs
dateDigitWsMoYear // 26
2018-02-26 08:35:33 +08:00
dateAlpha
dateAlphaWs
dateAlphaWsDigit
dateAlphaWsDigitMore // 30
dateAlphaWsDigitMoreWs
dateAlphaWsDigitMoreWsYear
dateAlphaWsMonth
dateAlphaWsDigitYearmaybe
dateAlphaWsMonthMore
dateAlphaWsMonthSuffix
dateAlphaWsMore
dateAlphaWsAtTime
2018-03-02 10:36:46 +08:00
dateAlphaWsAlpha
dateAlphaWsAlphaYearmaybe // 40
dateAlphaPeriodWsDigit
dateAlphaSlash
dateAlphaSlashDigit
dateAlphaSlashDigitSlash
2018-02-26 08:35:33 +08:00
dateWeekdayComma
dateWeekdayAbbrevComma
dateYearWs
dateYearWsMonthWs
2018-02-26 14:46:45 +08:00
)
const (
// Time state
timeIgnore timeState = iota // 0
2018-02-26 08:35:33 +08:00
timeStart
timeWs
2018-03-12 08:39:59 +08:00
timeWsAlpha
timeWsAlphaRParen
2018-03-12 08:39:59 +08:00
timeWsAlphaWs
timeWsAlphaZoneOffset // 6
2018-03-02 10:36:46 +08:00
timeWsAlphaZoneOffsetWs
timeWsAlphaZoneOffsetWsYear
timeWsAlphaZoneOffsetWsExtra
2018-02-26 08:35:33 +08:00
timeWsAMPMMaybe
timeWsAMPM // 11
2018-03-02 10:36:46 +08:00
timeWsOffset
timeWsOffsetWs // 13
2018-02-26 08:35:33 +08:00
timeWsOffsetColonAlpha
timeWsOffsetColon
timeWsYear // 16
timeWsYearOffset
2018-03-02 10:36:46 +08:00
timeOffset
2018-02-26 08:35:33 +08:00
timeOffsetColon
timeOffsetColonAlpha
2018-02-26 08:35:33 +08:00
timePeriod
timePeriodAMPM
2018-02-26 08:35:33 +08:00
timeZ
2014-04-21 10:56:17 +08:00
)
var (
// ErrAmbiguousMMDD for date formats such as 04/02/2014 the mm/dd vs dd/mm are
// ambiguous, so it is an error for strict parse rules.
ErrAmbiguousMMDD = fmt.Errorf("this date has ambiguous mm/dd vs dd/mm type format")
ErrCouldntFindFormat = fmt.Errorf("could not find format for")
ErrUnexpectedTail = fmt.Errorf("unexpected content after date/time: ")
)
2018-06-09 09:01:07 +08:00
func unknownErr(datestr string) error {
return fmt.Errorf("%w %q", ErrCouldntFindFormat, datestr)
2018-06-09 09:01:07 +08:00
}
func unexpectedTail(tail string) error {
return fmt.Errorf("%w %q", ErrUnexpectedTail, tail)
}
2018-03-26 07:48:11 +08:00
// ParseAny parse an unknown date format, detect the layout.
// Normal parse. Equivalent Timezone rules as time.Parse().
// NOTE: please see readme on mmdd vs ddmm ambiguous dates.
func ParseAny(datestr string, opts ...ParserOption) (time.Time, error) {
p, err := parseTime(datestr, nil, opts...)
2018-03-25 06:56:26 +08:00
if err != nil {
return time.Time{}, err
}
return p.parse()
2017-07-27 07:42:12 +08:00
}
2017-08-13 00:48:28 +08:00
// ParseIn with Location, equivalent to time.ParseInLocation() timezone/offset
2017-07-27 13:37:37 +08:00
// rules. Using location arg, if timezone/offset info exists in the
// datestring, it uses the given location rules for any zone interpretation.
// That is, MST means one thing when using America/Denver and something else
// in other locations.
func ParseIn(datestr string, loc *time.Location, opts ...ParserOption) (time.Time, error) {
p, err := parseTime(datestr, loc, opts...)
2018-03-25 06:56:26 +08:00
if err != nil {
return time.Time{}, err
}
return p.parse()
2017-07-27 07:42:12 +08:00
}
// ParseLocal Given an unknown date format, detect the layout,
// using time.Local, parse.
2017-07-27 08:08:58 +08:00
//
2017-07-27 13:37:37 +08:00
// Set Location to time.Local. Same as ParseIn Location but lazily uses
// the global time.Local variable for Location argument.
//
2023-02-15 12:56:17 +08:00
// denverLoc, _ := time.LoadLocation("America/Denver")
// time.Local = denverLoc
2017-07-27 13:37:37 +08:00
//
2023-02-15 12:56:17 +08:00
// t, err := dateparse.ParseLocal("3/1/2014")
2017-07-27 13:37:37 +08:00
//
// Equivalent to:
//
2023-02-15 12:56:17 +08:00
// t, err := dateparse.ParseIn("3/1/2014", denverLoc)
func ParseLocal(datestr string, opts ...ParserOption) (time.Time, error) {
p, err := parseTime(datestr, time.Local, opts...)
2018-03-25 06:56:26 +08:00
if err != nil {
return time.Time{}, err
}
return p.parse()
2017-07-27 07:42:12 +08:00
}
2017-08-13 00:48:28 +08:00
// MustParse parse a date, and panic if it can't be parsed. Used for testing.
// Not recommended for most use-cases.
func MustParse(datestr string, opts ...ParserOption) time.Time {
p, err := parseTime(datestr, nil, opts...)
2018-03-25 06:56:26 +08:00
if err != nil {
panic(err.Error())
}
t, err := p.parse()
2017-07-27 13:39:44 +08:00
if err != nil {
panic(err.Error())
}
return t
}
2018-03-26 07:48:11 +08:00
// ParseFormat parse's an unknown date-time string and returns a layout
// string that can parse this (and exact same format) other date-time strings.
//
2023-02-15 12:56:17 +08:00
// layout, err := dateparse.ParseFormat("2013-02-01 00:00:00")
// // layout = "2006-01-02 15:04:05"
func ParseFormat(datestr string, opts ...ParserOption) (string, error) {
p, err := parseTime(datestr, nil, opts...)
2018-03-25 06:56:26 +08:00
if err != nil {
return "", err
}
_, err = p.parse()
if err != nil {
return "", err
2017-07-27 07:42:12 +08:00
}
2018-03-25 06:56:26 +08:00
return string(p.format), nil
2017-07-27 07:42:12 +08:00
}
2014-04-21 10:56:17 +08:00
2018-03-25 08:49:27 +08:00
// ParseStrict parse an unknown date format. IF the date is ambigous
2018-03-26 07:48:11 +08:00
// mm/dd vs dd/mm then return an error. These return errors: 3.3.2014 , 8/8/71 etc
func ParseStrict(datestr string, opts ...ParserOption) (time.Time, error) {
p, err := parseTime(datestr, nil, opts...)
2018-03-25 08:49:27 +08:00
if err != nil {
return time.Time{}, err
}
if p.ambiguousMD {
return time.Time{}, ErrAmbiguousMMDD
}
return p.parse()
}
func parseTime(datestr string, loc *time.Location, opts ...ParserOption) (p *parser, err error) {
p, err = newParser(datestr, loc, opts...)
if err != nil {
return
}
// if this string is impossibly long, don't even try. longest date might be something like:
// 'Wednesday, 8 February 2023 19:00:46.999999999 +11:00 (AEDT) m=+0.000000001'
if len(datestr) > 75 {
return p, unknownErr(datestr)
}
if p.retryAmbiguousDateWithSwap {
// month out of range signifies that a day/month swap is the correct solution to an ambiguous date
// this is because it means that a day is being interpreted as a month and overflowing the valid value for that
// by retrying in this case, we can fix a common situation with no assumptions
defer func() {
2021-04-28 23:23:48 +08:00
if p != nil && p.ambiguousMD {
// if it errors out with the following error, swap before we
// get out of this function to reduce scope it needs to be applied on
_, err = p.parse()
if err != nil && strings.Contains(err.Error(), "month out of range") {
// create the option to reverse the preference
preferMonthFirst := PreferMonthFirst(!p.preferMonthFirst)
// turn off the retry to avoid endless recursion
retryAmbiguousDateWithSwap := RetryAmbiguousDateWithSwap(false)
modifiedOpts := append(opts, preferMonthFirst, retryAmbiguousDateWithSwap)
p, err = parseTime(datestr, time.Local, modifiedOpts...)
}
}
}()
}
2018-02-26 08:35:33 +08:00
// IMPORTANT: we may need to modify the datestr while we are parsing (e.g., to
// remove pieces of the string that should be ignored during golang parsing).
// We will iterate over the modified datestr, and whenever we update datestr,
// we need to make sure that i is adjusted accordingly to resume parsing in
// the correct place. In error messages though we'll use the original datestr.
2018-02-26 08:35:33 +08:00
i := 0
2014-04-21 10:56:17 +08:00
// General strategy is to read rune by rune through the date looking for
// certain hints of what type of date we are dealing with.
// Hopefully we only need to read about 5 or 6 bytes before
// we figure it out and then attempt a parse
2014-05-07 12:15:43 +08:00
iterRunes:
for ; i < len(p.datestr); i++ {
r, bytesConsumed := utf8.DecodeRuneInString(p.datestr[i:])
2017-11-19 03:07:33 +08:00
if bytesConsumed > 1 {
2023-02-15 13:10:45 +08:00
i += bytesConsumed - 1
2017-11-19 03:07:33 +08:00
}
2014-05-07 12:15:43 +08:00
// gou.Debugf("i=%d r=%s state=%d %s", i, string(r), p.stateDate, p.datestr)
2018-02-26 14:46:45 +08:00
switch p.stateDate {
2018-02-26 08:35:33 +08:00
case dateStart:
2014-05-07 12:15:43 +08:00
if unicode.IsDigit(r) {
2018-02-26 14:46:45 +08:00
p.stateDate = dateDigit
2017-02-06 02:04:03 +08:00
} else if unicode.IsLetter(r) {
2018-02-26 14:46:45 +08:00
p.stateDate = dateAlpha
2018-03-25 06:56:26 +08:00
} else {
return p, unknownErr(datestr)
2014-04-21 10:56:17 +08:00
}
2018-02-26 14:46:45 +08:00
case dateDigit:
2014-05-07 12:15:43 +08:00
switch r {
2017-07-14 10:57:15 +08:00
case '-', '\u2212':
2018-02-26 14:46:45 +08:00
// 2006-01-02
// 2013-Feb-03
2018-10-10 11:50:09 +08:00
// 13-Feb-03
// 29-Jun-2016
2018-02-26 14:46:45 +08:00
if i == 4 {
2018-10-10 11:50:09 +08:00
p.stateDate = dateYearDash
p.yeari = 0
p.yearlen = i
p.moi = i + 1
2018-02-26 14:46:45 +08:00
p.set(0, "2006")
2018-10-10 11:50:09 +08:00
} else {
p.stateDate = dateDigitDash
2018-02-26 14:46:45 +08:00
}
2014-05-07 12:15:43 +08:00
case '/':
2021-02-04 15:51:24 +08:00
// 08/May/2005
// 03/31/2005
// 2014/02/24
2018-02-26 14:46:45 +08:00
p.stateDate = dateDigitSlash
if i == 4 {
// 2014/02/24 - Year first /
p.yearlen = i // since it was start of datestr, i=len
p.moi = i + 1
if !p.setYear() {
return p, unknownErr(datestr)
}
p.stateDate = dateDigitYearSlash
} else {
// Either Ambiguous dd/mm vs mm/dd OR dd/month/yy
// 08/May/2005
// 03/31/2005
// 31/03/2005
if i+2 < len(p.datestr) && unicode.IsLetter(rune(p.datestr[i+1])) {
// 08/May/2005
p.stateDate = dateDigitSlashAlpha
p.moi = i + 1
p.daylen = 2
p.dayi = 0
if !p.setDay() {
return p, unknownErr(datestr)
}
continue
}
// Ambiguous dd/mm vs mm/dd the bane of date-parsing
// 03/31/2005
// 31/03/2005
2018-03-25 06:56:26 +08:00
p.ambiguousMD = true
if p.preferMonthFirst {
if p.molen == 0 {
// 03/31/2005
p.molen = i
if !p.setMonth() {
return p, unknownErr(datestr)
}
p.dayi = i + 1
}
} else {
if p.daylen == 0 {
p.daylen = i
if !p.setDay() {
return p, unknownErr(datestr)
}
p.moi = i + 1
}
}
}
case ':':
// 03/31/2005
// 2014/02/24
p.stateDate = dateDigitColon
if i == 4 {
p.yearlen = i
p.moi = i + 1
if !p.setYear() {
return p, unknownErr(datestr)
}
} else {
p.ambiguousMD = true
if p.preferMonthFirst {
if p.molen == 0 {
p.molen = i
if !p.setMonth() {
return p, unknownErr(datestr)
}
p.dayi = i + 1
}
}
}
2018-01-12 05:21:04 +08:00
case '.':
2018-02-28 12:05:14 +08:00
// 3.31.2014
2018-05-04 09:06:52 +08:00
// 08.21.71
// 2014.05
2018-02-26 14:46:45 +08:00
p.stateDate = dateDigitDot
2018-05-04 09:06:52 +08:00
if i == 4 {
p.yearlen = i
p.moi = i + 1
if !p.setYear() {
return p, unknownErr(datestr)
}
} else if i <= 2 {
2018-05-04 09:06:52 +08:00
p.ambiguousMD = true
if p.preferMonthFirst {
if p.molen == 0 {
// 03.31.2005
p.molen = i
if !p.setMonth() {
return p, unknownErr(datestr)
}
p.dayi = i + 1
}
} else {
if p.daylen == 0 {
p.daylen = i
if !p.setDay() {
return p, unknownErr(datestr)
}
p.moi = i + 1
}
}
2018-05-04 09:06:52 +08:00
}
// else this might be a unixy combined datetime of the form:
// yyyyMMddhhmmss.SSS
2018-05-04 09:06:52 +08:00
2018-01-25 09:59:25 +08:00
case ' ':
2018-02-28 12:05:14 +08:00
// 18 January 2018
// 8 January 2018
2018-03-11 05:43:16 +08:00
// 8 jan 2018
2018-02-28 12:05:14 +08:00
// 02 Jan 2018 23:59
// 02 Jan 2018 23:59:34
// 12 Feb 2006, 19:17
// 12 Feb 2006, 19:17:22
// 2013 Jan 06 15:04:05
if i == 4 {
p.yearlen = i
p.moi = i + 1
if !p.setYear() {
return p, unknownErr(datestr)
}
p.stateDate = dateYearWs
} else if i == 6 {
2021-02-04 14:23:19 +08:00
p.stateDate = dateDigitSt
} else {
p.stateDate = dateDigitWs
p.dayi = 0
p.daylen = i
}
2018-02-26 14:46:45 +08:00
case '年':
// Chinese Year
p.stateDate = dateDigitChineseYear
p.yearlen = i - 2
p.moi = i + 1
if !p.setYear() {
return p, unknownErr(datestr)
}
2018-04-19 08:36:22 +08:00
case ',':
return p, unknownErr(datestr)
case 's', 'S', 'r', 'R', 't', 'T', 'n', 'N':
// 1st January 2018
// 2nd Jan 2018 23:59
// st, rd, nd, st
p.stateDate = dateAlphaWsMonthSuffix
i--
2018-02-26 14:46:45 +08:00
default:
continue
2014-04-25 08:51:56 +08:00
}
2018-02-26 14:46:45 +08:00
p.part1Len = i
2018-01-25 09:59:25 +08:00
2021-02-04 14:23:19 +08:00
case dateDigitSt:
p.set(0, "060102")
2021-02-04 15:51:24 +08:00
i = i - 1
2021-02-04 14:23:19 +08:00
p.stateTime = timeStart
break iterRunes
2018-10-10 11:50:09 +08:00
case dateYearDash:
// dateYearDashDashT
2018-02-14 12:44:26 +08:00
// 2006-01-02T15:04:05Z07:00
2021-02-06 07:19:04 +08:00
// 2020-08-17T17:00:00:000+0100
2018-10-10 11:50:09 +08:00
// dateYearDashDashWs
2018-02-14 12:44:26 +08:00
// 2013-04-01 22:43:22
2018-10-10 11:50:09 +08:00
// dateYearDashAlphaDash
// 2013-Feb-03
2018-02-26 14:46:45 +08:00
switch r {
case '-':
p.molen = i - p.moi
p.dayi = i + 1
2018-10-10 11:50:09 +08:00
p.stateDate = dateYearDashDash
if !p.setMonth() {
return p, unknownErr(datestr)
}
2014-12-03 06:57:00 +08:00
default:
2018-10-10 11:50:09 +08:00
if unicode.IsLetter(r) {
p.stateDate = dateYearDashAlphaDash
2014-12-03 06:57:00 +08:00
}
2014-05-07 12:15:43 +08:00
}
2018-10-10 11:50:09 +08:00
case dateYearDashDash:
// dateYearDashDashT
2018-02-26 14:46:45 +08:00
// 2006-01-02T15:04:05Z07:00
2018-10-10 11:50:09 +08:00
// dateYearDashDashWs
2018-02-26 14:46:45 +08:00
// 2013-04-01 22:43:22
// dateYearDashDashOffset
// 2020-07-20+00:00
2018-02-26 14:46:45 +08:00
switch r {
case '+', '-':
p.offseti = i
p.daylen = i - p.dayi
p.stateDate = dateYearDashDashOffset
if !p.setDay() {
return p, unknownErr(datestr)
}
2018-02-26 14:46:45 +08:00
case ' ':
p.daylen = i - p.dayi
2018-10-10 11:50:09 +08:00
p.stateDate = dateYearDashDashWs
2018-02-26 14:46:45 +08:00
p.stateTime = timeStart
if !p.setDay() {
return p, unknownErr(datestr)
}
2018-02-26 14:46:45 +08:00
break iterRunes
case 'T':
p.daylen = i - p.dayi
2018-10-10 11:50:09 +08:00
p.stateDate = dateYearDashDashT
2018-02-26 14:46:45 +08:00
p.stateTime = timeStart
if !p.setDay() {
return p, unknownErr(datestr)
}
2018-02-26 14:46:45 +08:00
break iterRunes
}
2021-02-06 07:19:04 +08:00
case dateYearDashDashT:
// dateYearDashDashT
// 2006-01-02T15:04:05Z07:00
// 2020-08-17T17:00:00:000+0100
case dateYearDashDashOffset:
// 2020-07-20+00:00
switch r {
case ':':
p.set(p.offseti, "-07:00")
}
2018-10-10 11:50:09 +08:00
case dateYearDashAlphaDash:
2018-02-26 14:46:45 +08:00
// 2013-Feb-03
switch r {
case '-':
2018-02-28 12:05:14 +08:00
p.molen = i - p.moi
2018-02-26 14:46:45 +08:00
p.set(p.moi, "Jan")
p.dayi = i + 1
}
2018-10-10 11:50:09 +08:00
case dateDigitDash:
// 13-Feb-03
// 29-Jun-2016
if unicode.IsLetter(r) {
p.stateDate = dateDigitDashAlpha
p.moi = i
} else if unicode.IsDigit(r) {
p.stateDate = dateDigitDashDigit
p.moi = i
2018-10-10 11:50:09 +08:00
} else {
return p, unknownErr(datestr)
2018-10-10 11:50:09 +08:00
}
case dateDigitDashAlpha:
// 13-Feb-03
// 28-Feb-03
// 29-Jun-2016
switch r {
case '-':
p.molen = i - p.moi
p.set(p.moi, "Jan")
p.yeari = i + 1
p.stateDate = dateDigitDashAlphaDash
}
case dateDigitDashDigit:
// 29-06-2026
switch r {
case '-':
// X
// 29-06-2026
p.molen = i - p.moi
if p.molen == 2 {
p.set(p.moi, "01")
p.yeari = i + 1
p.stateDate = dateDigitDashDigitDash
} else {
return p, unknownErr(datestr)
}
}
case dateDigitDashAlphaDash, dateDigitDashDigitDash:
// dateDigitDashAlphaDash:
// 13-Feb-03 ambiguous
// 28-Feb-03 ambiguous
// 29-Jun-2016 dd-month(alpha)-yyyy
// dateDigitDashDigitDash:
// 29-06-2026
2018-10-10 11:50:09 +08:00
switch r {
case ' ':
// we need to find if this was 4 digits, aka year
// or 2 digits which makes it ambiguous year/day
length := i - (p.moi + p.molen + 1)
if length == 4 {
p.yearlen = 4
p.set(p.yeari, "2006")
// We now also know that part1 was the day
p.dayi = 0
p.daylen = p.part1Len
if !p.setDay() {
return p, unknownErr(datestr)
}
2018-10-10 11:50:09 +08:00
} else if length == 2 {
// We have no idea if this is
// yy-mon-dd OR dd-mon-yy
// (or for dateDigitDashDigitDash, yy-mm-dd OR dd-mm-yy)
2018-10-10 11:50:09 +08:00
//
// We are going to ASSUME (bad, bad) that it is dd-mon-yy (dd-mm-yy),
// which is a horrible assumption, but seems to be the convention for
// dates that are formatted in this way.
2018-10-10 11:50:09 +08:00
p.ambiguousMD = true
p.yearlen = 2
p.set(p.yeari, "06")
// We now also know that part1 was the day
p.dayi = 0
p.daylen = p.part1Len
if !p.setDay() {
return p, unknownErr(datestr)
}
} else {
return p, unknownErr(datestr)
2018-10-10 11:50:09 +08:00
}
p.stateTime = timeStart
break iterRunes
}
case dateDigitYearSlash:
2014-07-11 06:25:23 +08:00
// 2014/07/10 06:55:38.156283
// I honestly don't know if this format ever shows up as yyyy/
2018-02-26 14:46:45 +08:00
2014-05-07 12:15:43 +08:00
switch r {
case ' ':
fallthrough
case ':':
p.stateTime = timeStart
if p.daylen == 0 {
p.daylen = i - p.dayi
if !p.setDay() {
return p, unknownErr(datestr)
}
}
break iterRunes
case '/':
if p.molen == 0 {
p.molen = i - p.moi
if !p.setMonth() {
return p, unknownErr(datestr)
}
p.dayi = i + 1
}
}
case dateDigitSlashAlpha:
// 06/May/2008
switch r {
case '/':
// |
// 06/May/2008
if p.molen == 0 {
p.set(p.moi, "Jan")
p.yeari = i + 1
}
// We aren't breaking because we are going to re-use this case
// to find where the date starts, and possible time begins
case ' ':
fallthrough
case ':':
2021-02-04 14:23:19 +08:00
p.stateTime = timeStart
if p.yearlen == 0 {
p.yearlen = i - p.yeari
if !p.setYear() {
return p, unknownErr(datestr)
}
2021-02-04 14:23:19 +08:00
}
break iterRunes
}
case dateDigitSlash:
// 03/19/2012 10:11:59
// 04/2/2014 03:00:37
// 04/2/2014, 03:00:37
// 3/1/2012 10:11:59
// 4/8/2014 22:05
// 3/1/2014
// 10/13/2014
// 01/02/2006
// 1/2/06
switch r {
2018-02-26 14:46:45 +08:00
case '/':
// This is the 2nd / so now we should know start pts of all of the dd, mm, yy
if p.preferMonthFirst {
if p.daylen == 0 {
p.daylen = i - p.dayi
if !p.setDay() {
return p, unknownErr(datestr)
}
p.yeari = i + 1
}
} else {
if p.molen == 0 {
p.molen = i - p.moi
if !p.setMonth() {
return p, unknownErr(datestr)
}
p.yeari = i + 1
}
}
// Note no break, we are going to pass by and re-enter this dateDigitSlash
// and look for ending (space) or not (just date)
case ' ', ',':
p.stateTime = timeStart
if p.yearlen == 0 {
p.yearlen = i - p.yeari
if r == ',' {
// skip the comma
i++
}
if !p.setYear() {
return p, unknownErr(datestr)
}
2021-02-04 15:51:24 +08:00
}
break iterRunes
2014-05-07 12:15:43 +08:00
}
2018-01-25 09:59:25 +08:00
case dateDigitColon:
// 2014:07:10 06:55:38.156283
// 03:19:2012 10:11:59
// 04:2:2014 03:00:37
// 3:1:2012 10:11:59
// 4:8:2014 22:05
// 3:1:2014
// 10:13:2014
// 01:02:2006
// 1:2:06
switch r {
case ' ':
p.stateTime = timeStart
if p.yearlen == 0 {
p.yearlen = i - p.yeari
if !p.setYear() {
return p, unknownErr(datestr)
}
} else if p.daylen == 0 {
p.daylen = i - p.dayi
if !p.setDay() {
return p, unknownErr(datestr)
}
}
break iterRunes
case ':':
if p.yearlen > 0 {
// 2014:07:10 06:55:38.156283
if p.molen == 0 {
p.molen = i - p.moi
if !p.setMonth() {
return p, unknownErr(datestr)
}
p.dayi = i + 1
}
} else if p.preferMonthFirst {
if p.daylen == 0 {
p.daylen = i - p.dayi
if !p.setDay() {
return p, unknownErr(datestr)
}
p.yeari = i + 1
}
}
}
2018-02-26 08:35:33 +08:00
case dateDigitWs:
2018-01-25 09:59:25 +08:00
// 18 January 2018
// 8 January 2018
2018-03-11 05:43:16 +08:00
// 8 jan 2018
// 1 jan 18
2018-02-28 12:05:14 +08:00
// 02 Jan 2018 23:59
// 02 Jan 2018 23:59:34
// 12 Feb 2006, 19:17
// 12 Feb 2006, 19:17:22
switch r {
case ' ':
p.yeari = i + 1
2018-03-11 05:43:16 +08:00
//p.yearlen = 4
2018-02-28 12:05:14 +08:00
p.dayi = 0
p.daylen = p.part1Len
if !p.setDay() {
return p, unknownErr(datestr)
}
2018-02-28 12:05:14 +08:00
p.stateTime = timeStart
2018-10-21 03:01:51 +08:00
if i > p.daylen+len(" Sep") { // November etc
// If this is a legit full month, then change the string we're parsing
// to compensate for the longest month, and do the same with the format string. We
// must maintain a corresponding length/content and this is the easiest
// way to do this.
possibleFullMonth := strings.ToLower(p.datestr[(p.dayi + (p.daylen + 1)):i])
if isMonthFull(possibleFullMonth) {
p.moi = p.dayi + p.daylen + 1
p.molen = i - p.moi
p.fullMonth = possibleFullMonth
p.stateDate = dateDigitWsMoYear
} else {
return p, unknownErr(datestr)
}
2018-10-21 03:01:51 +08:00
} else {
// If len=3, the might be Feb or May? Ie ambigous abbreviated but
// we can parse may with either. BUT, that means the
// format may not be correct?
// mo := strings.ToLower(p.datestr[p.daylen+1 : i])
p.moi = p.dayi + p.daylen + 1
2018-10-21 03:01:51 +08:00
p.molen = i - p.moi
2018-02-28 12:05:14 +08:00
p.set(p.moi, "Jan")
2018-03-02 10:36:46 +08:00
p.stateDate = dateDigitWsMoYear
}
}
2018-02-28 12:05:14 +08:00
case dateDigitWsMoYear:
2018-03-11 05:43:16 +08:00
// 8 jan 2018
2018-02-28 12:05:14 +08:00
// 02 Jan 2018 23:59
// 02 Jan 2018 23:59:34
// 12 Feb 2006, 19:17
// 12 Feb 2006, 19:17:22
switch r {
2018-02-28 12:05:14 +08:00
case ',':
2018-03-12 08:39:59 +08:00
p.yearlen = i - p.yeari
if !p.setYear() {
return p, unknownErr(datestr)
}
2018-02-28 12:05:14 +08:00
i++
break iterRunes
case ' ':
p.yearlen = i - p.yeari
if !p.setYear() {
return p, unknownErr(datestr)
}
break iterRunes
2018-01-25 09:59:25 +08:00
}
case dateYearWs:
// 2013 Jan 06 15:04:05
// 2013 January 06 15:04:05
if r == ' ' {
p.molen = i - p.moi
// Must be a valid short or long month
if p.molen == 3 {
p.set(p.moi, "Jan")
p.dayi = i + 1
p.stateDate = dateYearWsMonthWs
} else {
possibleFullMonth := strings.ToLower(p.datestr[p.moi:(p.moi + p.molen)])
if i > 3 && isMonthFull(possibleFullMonth) {
p.fullMonth = possibleFullMonth
p.dayi = i + 1
p.stateDate = dateYearWsMonthWs
} else {
return p, unknownErr(datestr)
}
}
}
case dateYearWsMonthWs:
// 2013 Jan 06 15:04:05
// 2013 January 06 15:04:05
switch r {
case ',':
p.daylen = i - p.dayi
p.setDay()
i++
p.stateTime = timeStart
break iterRunes
case ' ':
p.daylen = i - p.dayi
p.setDay()
p.stateTime = timeStart
break iterRunes
}
2018-02-26 08:35:33 +08:00
case dateDigitChineseYear:
// dateDigitChineseYear
2017-11-19 03:07:33 +08:00
// 2014年04月08日
// weekday %Y年%m月%e日 %A %I:%M %p
// 2013年07月18日 星期四 10:27 上午
switch r {
case '月':
// month
p.molen = i - p.moi - 2
p.dayi = i + 1
if !p.setMonth() {
return p, unknownErr(datestr)
}
case '日':
// day
p.daylen = i - p.dayi - 2
if !p.setDay() {
return p, unknownErr(datestr)
}
case ' ':
if p.daylen <= 0 {
return p, unknownErr(datestr)
}
2018-02-26 14:46:45 +08:00
p.stateDate = dateDigitChineseYearWs
p.stateTime = timeStart
2023-12-09 09:31:28 +08:00
break iterRunes
2017-11-19 03:07:33 +08:00
}
2018-02-26 08:35:33 +08:00
case dateDigitDot:
2019-02-23 08:50:43 +08:00
// This is the 2nd period
2018-01-12 05:21:04 +08:00
// 3.31.2014
2018-05-04 09:06:52 +08:00
// 08.21.71
// 2014.05
2019-02-23 08:50:43 +08:00
// 2018.09.30
2018-01-12 05:21:04 +08:00
if r == '.' {
2019-02-23 08:50:43 +08:00
if p.moi == 0 {
// 3.31.2014
p.daylen = i - p.dayi
p.yeari = i + 1
if !p.setDay() {
return p, unknownErr(datestr)
}
2019-02-23 08:50:43 +08:00
p.stateDate = dateDigitDotDot
} else if p.dayi == 0 && p.yearlen == 0 {
// 23.07.2002
p.molen = i - p.moi
p.yeari = i + 1
if !p.setMonth() {
return p, unknownErr(datestr)
}
p.stateDate = dateDigitDotDot
2019-02-23 08:50:43 +08:00
} else {
// 2018.09.30
// p.molen = 2
2019-02-23 08:50:43 +08:00
p.molen = i - p.moi
p.dayi = i + 1
if !p.setMonth() {
return p, unknownErr(datestr)
}
2019-02-23 08:50:43 +08:00
p.stateDate = dateDigitDotDot
}
2018-01-12 05:21:04 +08:00
}
2018-02-26 08:35:33 +08:00
case dateDigitDotDot:
// dateYearDashDashT
// 2006.01.02T15:04:05Z07:00
// dateYearDashDashWs
// 2013.04.01 22:43:22
// dateYearDashDashOffset
// 2020.07.20+00:00
switch r {
case '+', '-':
p.offseti = i
p.daylen = i - p.dayi
p.stateDate = dateDigitDotDotOffset
if !p.setDay() {
return p, unknownErr(datestr)
}
case ' ':
p.daylen = i - p.dayi
p.stateDate = dateDigitDotDotWs
p.stateTime = timeStart
if !p.setDay() {
return p, unknownErr(datestr)
}
break iterRunes
case 'T':
p.daylen = i - p.dayi
p.stateDate = dateDigitDotDotT
p.stateTime = timeStart
if !p.setDay() {
return p, unknownErr(datestr)
}
break iterRunes
}
case dateDigitDotDotT:
// dateYearDashDashT
// 2006-01-02T15:04:05Z07:00
// 2020-08-17T17:00:00:000+0100
case dateDigitDotDotOffset:
// 2020-07-20+00:00
switch r {
case ':':
p.set(p.offseti, "-07:00")
}
2018-02-26 08:35:33 +08:00
case dateAlpha:
// dateAlphaWs
2017-07-14 00:11:41 +08:00
// Mon Jan _2 15:04:05 2006
// Mon Jan _2 15:04:05 MST 2006
// Mon Jan 02 15:04:05 -0700 2006
// Mon Jan 02 15:04:05 2006 -0700
2017-07-14 00:11:41 +08:00
// Mon Aug 10 15:44:11 UTC+0100 2015
// Fri Jul 03 2015 18:04:07 GMT+0100 (GMT Daylight Time)
// dateAlphaWsDigit
2017-07-14 00:11:41 +08:00
// May 8, 2009 5:57:51 PM
2018-03-11 05:43:16 +08:00
// oct 1, 1970
2018-05-24 12:11:15 +08:00
// dateAlphaWsMonth
// April 8, 2009
// dateAlphaWsMore
// dateAlphaWsAtTime
// January 02, 2006 at 3:04pm MST-07
//
2018-06-09 09:01:07 +08:00
// dateAlphaPeriodWsDigit
// oct. 1, 1970
// dateAlphaSlash
// dateAlphaSlashDigit
// dateAlphaSlashDigitSlash
// Oct/ 7/1970
// Oct/07/1970
// February/ 7/1970
// February/07/1970
//
2018-02-26 08:35:33 +08:00
// dateWeekdayComma
// Monday, 02 Jan 2006 15:04:05 MST
// Monday, 02-Jan-06 15:04:05 MST
// Monday, 02 Jan 2006 15:04:05 -0700
// Monday, 02 Jan 2006 15:04:05 +0100
2018-02-26 08:35:33 +08:00
// dateWeekdayAbbrevComma
2017-07-14 00:11:41 +08:00
// Mon, 02 Jan 2006 15:04:05 MST
// Mon, 02 Jan 2006 15:04:05 -0700
// Thu, 13 Jul 2017 08:58:40 +0100
// Tue, 11 Jul 2017 16:28:13 +0200 (CEST)
2018-03-12 08:39:59 +08:00
// Mon, 02-Jan-06 15:04:05 MST
2014-05-07 12:15:43 +08:00
switch {
case r == ' ':
// X
// April 8, 2009
if i > 3 {
// Check to see if the alpha is name of month? or Day?
month := strings.ToLower(p.datestr[0:i])
if isMonthFull(month) {
p.moi = 0
p.molen = i
p.fullMonth = month
// len(" 31, 2018") = 9
if len(p.datestr[i:]) < 10 {
// April 8, 2009
p.stateDate = dateAlphaWsMonth
} else {
p.stateDate = dateAlphaWsMore
2018-05-24 12:11:15 +08:00
}
p.dayi = i + 1
break
2018-05-24 12:11:15 +08:00
}
2018-05-24 12:11:15 +08:00
} else {
// This is possibly ambiguous? May will parse as either though.
// So, it could return in-correct format.
// dateAlphaWs
// May 05, 2005, 05:05:05
// May 05 2005, 05:05:05
// Jul 05, 2005, 05:05:05
// May 8 17:57:51 2009
// May 8 17:57:51 2009
// skip & return to dateStart
// Tue 05 May 2020, 05:05:05
// Mon Jan 2 15:04:05 2006
maybeDay := strings.ToLower(p.datestr[0:i])
if isDay(maybeDay) {
// using skip throws off indices used by other code; saner to restart
return parseTime(p.datestr[i+1:], loc)
}
2018-05-24 12:11:15 +08:00
p.stateDate = dateAlphaWs
}
2018-05-24 12:11:15 +08:00
2014-05-21 13:53:52 +08:00
case r == ',':
// Mon, 02 Jan 2006
2021-02-06 05:44:33 +08:00
2017-07-14 00:11:41 +08:00
if i == 3 {
2018-02-26 14:46:45 +08:00
p.stateDate = dateWeekdayAbbrevComma
p.set(0, "Mon")
2017-07-14 00:11:41 +08:00
} else {
2018-02-26 14:46:45 +08:00
p.stateDate = dateWeekdayComma
2018-03-02 10:36:46 +08:00
p.skip = i + 2
i++
2018-03-02 10:36:46 +08:00
// TODO: lets just make this "skip" as we don't need
// the mon, monday, they are all superfelous and not needed
// just lay down the skip, no need to fill and then skip
}
2018-06-09 09:01:07 +08:00
case r == '.':
// sept. 28, 2017
// jan. 28, 2017
2018-06-09 09:01:07 +08:00
p.stateDate = dateAlphaPeriodWsDigit
if i == 3 {
p.molen = i
p.set(0, "Jan")
} else if i == 4 {
// gross
newDatestr := p.datestr[0:i-1] + p.datestr[i:]
return parseTime(newDatestr, loc, opts...)
} else {
return p, unknownErr(datestr)
}
case r == '/':
// X
// Oct/ 7/1970
// Oct/07/1970
// X
// February/ 7/1970
// February/07/1970
// Must be a valid short or long month
if i == 3 {
p.moi = 0
p.molen = i - p.moi
p.set(p.moi, "Jan")
p.stateDate = dateAlphaSlash
} else {
possibleFullMonth := strings.ToLower(p.datestr[:i])
if i > 3 && isMonthFull(possibleFullMonth) {
p.moi = 0
p.molen = i - p.moi
p.fullMonth = possibleFullMonth
p.stateDate = dateAlphaSlash
} else {
return p, unknownErr(datestr)
}
}
2017-07-14 22:02:22 +08:00
}
2018-03-02 10:36:46 +08:00
case dateAlphaWs:
// dateAlphaWsAlpha
// Mon Jan _2 15:04:05 2006
// Mon Jan _2 15:04:05 MST 2006
// Mon Jan 02 15:04:05 -0700 2006
// Mon Jan 02 15:04:05 2006 -0700
2018-03-02 10:36:46 +08:00
// Fri Jul 03 2015 18:04:07 GMT+0100 (GMT Daylight Time)
// Mon Aug 10 15:44:11 UTC+0100 2015
// dateAlphaWsDigit
// May 8, 2009 5:57:51 PM
// May 8 2009 5:57:51 PM
// May 8 17:57:51 2009
// May 8 17:57:51 2009
// May 08 17:57:51 2009
// oct 1, 1970
// oct 7, '70
2014-05-07 12:15:43 +08:00
switch {
case unicode.IsLetter(r):
2018-03-02 10:36:46 +08:00
p.set(0, "Mon")
p.stateDate = dateAlphaWsAlpha
p.set(i, "Jan")
2017-07-14 00:11:41 +08:00
case unicode.IsDigit(r):
2018-03-02 10:36:46 +08:00
p.set(0, "Jan")
p.stateDate = dateAlphaWsDigit
p.dayi = i
2014-05-21 13:53:52 +08:00
}
2018-03-02 10:36:46 +08:00
case dateAlphaWsDigit:
2018-03-11 05:43:16 +08:00
// May 8, 2009 5:57:51 PM
// May 8 2009 5:57:51 PM
2018-03-11 05:43:16 +08:00
// oct 1, 1970
// oct 7, '70
2018-06-09 09:01:07 +08:00
// oct. 7, 1970
// May 8 17:57:51 2009
// May 8 17:57:51 2009
// May 08 17:57:51 2009
if r == ',' {
2018-03-02 10:36:46 +08:00
p.daylen = i - p.dayi
if !p.setDay() {
return p, unknownErr(datestr)
}
p.stateDate = dateAlphaWsDigitMore
} else if r == ' ' {
p.daylen = i - p.dayi
if !p.setDay() {
return p, unknownErr(datestr)
}
p.yeari = i + 1
p.stateDate = dateAlphaWsDigitYearmaybe
p.stateTime = timeStart
} else if unicode.IsLetter(r) {
p.stateDate = dateAlphaWsMonthSuffix
i--
2018-02-26 08:35:33 +08:00
}
case dateAlphaWsDigitYearmaybe:
// x
// May 8 2009 5:57:51 PM
// May 8 17:57:51 2009
// May 8 17:57:51 2009
// May 08 17:57:51 2009
// Jul 03 2015 18:04:07 GMT+0100 (GMT Daylight Time)
if r == ':' {
// Guessed wrong; was not a year
i = i - 3
p.stateDate = dateAlphaWsDigit
p.yeari = 0
break iterRunes
} else if r == ' ' {
// must be year format, not 15:04
p.yearlen = i - p.yeari
if !p.setYear() {
return p, unknownErr(datestr)
}
break iterRunes
}
case dateAlphaWsDigitMore:
2018-03-11 05:43:16 +08:00
// x
// May 8, 2009 5:57:51 PM
// May 05, 2005, 05:05:05
// May 05 2005, 05:05:05
2018-03-11 05:43:16 +08:00
// oct 1, 1970
// oct 7, '70
if r == ' ' {
2018-03-02 10:36:46 +08:00
p.yeari = i + 1
p.stateDate = dateAlphaWsDigitMoreWs
2018-02-26 08:35:33 +08:00
}
case dateAlphaWsDigitMoreWs:
2018-03-11 05:43:16 +08:00
// x
// May 8, 2009 5:57:51 PM
// May 05, 2005, 05:05:05
2018-03-11 05:43:16 +08:00
// oct 1, 1970
// oct 7, '70
switch r {
case '\'':
p.yeari = i + 1
case ' ':
fallthrough
case ',':
// x
// May 8, 2009 5:57:51 PM
// x
// May 8, 2009, 5:57:51 PM
p.stateDate = dateAlphaWsDigitMoreWsYear
2018-03-02 10:36:46 +08:00
p.yearlen = i - p.yeari
if !p.setYear() {
return p, unknownErr(datestr)
}
2018-03-02 10:36:46 +08:00
p.stateTime = timeStart
2018-02-26 08:35:33 +08:00
break iterRunes
2018-02-08 23:03:45 +08:00
}
2018-06-09 09:01:07 +08:00
case dateAlphaWsMonth:
// April 8, 2009
// April 8 2009
switch r {
case ' ':
fallthrough
case ',':
// x
// June 8, 2009
// x
// June 8 2009
if p.daylen == 0 {
p.daylen = i - p.dayi
if !p.setDay() {
return p, unknownErr(datestr)
}
2018-06-09 09:01:07 +08:00
}
case 's', 'S', 'r', 'R', 't', 'T', 'n', 'N':
// st, rd, nd, st
i--
p.stateDate = dateAlphaWsMonthSuffix
default:
if p.daylen > 0 && p.yeari == 0 {
p.yeari = i
}
}
case dateAlphaWsMonthMore:
// X
// January 02, 2006, 15:04:05
// January 02 2006, 15:04:05
// January 02, 2006 15:04:05
// January 02 2006 15:04:05
switch r {
case ',':
p.yearlen = i - p.yeari
if !p.setYear() {
return p, unknownErr(datestr)
}
p.stateTime = timeStart
i++
break iterRunes
case ' ':
p.yearlen = i - p.yeari
if !p.setYear() {
return p, unknownErr(datestr)
}
p.stateTime = timeStart
break iterRunes
}
case dateAlphaWsMonthSuffix:
// x
// April 8th, 2009
// April 8th 2009
switch r {
case 't', 'T':
2019-04-26 07:42:47 +08:00
if p.nextIs(i, 'h') || p.nextIs(i, 'H') {
if len(p.datestr) > i+2 {
return parseTime(fmt.Sprintf("%s%s", p.datestr[0:i], p.datestr[i+2:]), loc, opts...)
}
}
case 'n', 'N':
if p.nextIs(i, 'd') || p.nextIs(i, 'D') {
if len(p.datestr) > i+2 {
return parseTime(fmt.Sprintf("%s%s", p.datestr[0:i], p.datestr[i+2:]), loc, opts...)
}
}
case 's', 'S':
if p.nextIs(i, 't') || p.nextIs(i, 'T') {
if len(p.datestr) > i+2 {
return parseTime(fmt.Sprintf("%s%s", p.datestr[0:i], p.datestr[i+2:]), loc, opts...)
}
}
case 'r', 'R':
if p.nextIs(i, 'd') || p.nextIs(i, 'D') {
if len(p.datestr) > i+2 {
return parseTime(fmt.Sprintf("%s%s", p.datestr[0:i], p.datestr[i+2:]), loc, opts...)
}
}
}
case dateAlphaWsMore:
// January 02, 2006, 15:04:05
// January 02 2006, 15:04:05
// January 2nd, 2006, 15:04:05
// January 2nd 2006, 15:04:05
// September 17, 2012 at 5:00pm UTC-05
switch {
case r == ',':
// x
// January 02, 2006, 15:04:05
if p.nextIs(i, ' ') {
p.daylen = i - p.dayi
if !p.setDay() {
return p, unknownErr(datestr)
}
p.yeari = i + 2
p.stateDate = dateAlphaWsMonthMore
i++
}
case r == ' ':
// x
// January 02 2006, 15:04:05
p.daylen = i - p.dayi
if !p.setDay() {
return p, unknownErr(datestr)
}
p.yeari = i + 1
p.stateDate = dateAlphaWsMonthMore
case unicode.IsDigit(r):
// XX
// January 02, 2006, 15:04:05
continue
case unicode.IsLetter(r):
// X
// January 2nd, 2006, 15:04:05
p.daylen = i - p.dayi
if !p.setDay() {
return p, unknownErr(datestr)
}
p.stateDate = dateAlphaWsMonthSuffix
i--
2018-06-09 09:01:07 +08:00
}
case dateAlphaPeriodWsDigit:
// oct. 7, '70
switch {
case r == ' ':
// continue
case unicode.IsDigit(r):
p.stateDate = dateAlphaWsDigit
p.dayi = i
default:
return p, unknownErr(datestr)
}
case dateAlphaSlash:
// Oct/ 7/1970
// February/07/1970
switch {
case r == ' ':
// continue
case unicode.IsDigit(r):
p.stateDate = dateAlphaSlashDigit
p.dayi = i
default:
return p, unknownErr(datestr)
}
case dateAlphaSlashDigit:
// dateAlphaSlash:
// dateAlphaSlashDigit:
// dateAlphaSlashDigitSlash:
// Oct/ 7/1970
// Oct/07/1970
// February/ 7/1970
// February/07/1970
switch {
case r == '/':
p.yeari = i + 1
p.daylen = i - p.dayi
if !p.setDay() {
return p, unknownErr(datestr)
}
p.stateDate = dateAlphaSlashDigitSlash
case unicode.IsDigit(r):
// continue
default:
return p, unknownErr(datestr)
}
case dateAlphaSlashDigitSlash:
switch {
case unicode.IsDigit(r):
// continue
case r == ' ':
p.stateTime = timeStart
break iterRunes
default:
return p, unknownErr(datestr)
}
2018-06-09 09:01:07 +08:00
case dateWeekdayComma:
// Monday, 02 Jan 2006 15:04:05 MST
// Monday, 02 Jan 2006 15:04:05 -0700
// Monday, 02 Jan 2006 15:04:05 +0100
// Monday, 02-Jan-06 15:04:05 MST
if p.dayi == 0 {
p.dayi = i
}
switch r {
case ' ':
fallthrough
case '-':
2018-06-09 09:01:07 +08:00
if p.moi == 0 {
p.moi = i + 1
p.daylen = i - p.dayi
if !p.setDay() {
return p, unknownErr(datestr)
}
2018-06-09 09:01:07 +08:00
} else if p.yeari == 0 {
p.yeari = i + 1
p.molen = i - p.moi
p.set(p.moi, "Jan")
} else {
p.stateTime = timeStart
break iterRunes
}
}
case dateWeekdayAbbrevComma:
// Mon, 02 Jan 2006 15:04:05 MST
// Mon, 02 Jan 2006 15:04:05 -0700
// Thu, 13 Jul 2017 08:58:40 +0100
// Thu, 4 Jan 2018 17:53:36 +0000
// Tue, 11 Jul 2017 16:28:13 +0200 (CEST)
// Mon, 02-Jan-06 15:04:05 MST
2023-02-15 20:37:23 +08:00
var offset int
2018-06-09 09:01:07 +08:00
switch r {
case ' ':
for i+1 < len(p.datestr) && p.datestr[i+1] == ' ' {
i++
2023-02-15 20:37:23 +08:00
offset++
}
fallthrough
case '-':
2018-06-09 09:01:07 +08:00
if p.dayi == 0 {
p.dayi = i + 1
} else if p.moi == 0 {
p.daylen = i - p.dayi
if !p.setDay() {
return p, unknownErr(datestr)
}
2018-06-09 09:01:07 +08:00
p.moi = i + 1
} else if p.yeari == 0 {
2023-02-15 20:37:23 +08:00
p.molen = i - p.moi - offset
2018-06-09 09:01:07 +08:00
p.set(p.moi, "Jan")
p.yeari = i + 1
} else {
2023-02-15 20:37:23 +08:00
p.yearlen = i - p.yeari - offset
if !p.setYear() {
return p, unknownErr(datestr)
}
2018-06-09 09:01:07 +08:00
p.stateTime = timeStart
break iterRunes
}
}
2014-05-07 12:15:43 +08:00
default:
break iterRunes
2014-04-21 10:56:17 +08:00
}
2014-05-07 12:15:43 +08:00
}
if !p.coalesceDate(i) {
return p, unknownErr(datestr)
}
2018-02-26 14:46:45 +08:00
if p.stateTime == timeStart {
// increment first one, since the i++ occurs at end of loop
2018-03-11 05:43:16 +08:00
if i < len(p.datestr) {
i++
}
2018-03-12 08:39:59 +08:00
// ensure we skip any whitespace prefix
for ; i < len(p.datestr); i++ {
r := rune(p.datestr[i])
2018-03-12 08:39:59 +08:00
if r != ' ' {
break
}
}
2018-02-26 14:46:45 +08:00
iterTimeRunes:
for ; i < len(p.datestr); i++ {
r := rune(p.datestr[i])
2018-02-26 14:46:45 +08:00
2021-02-06 07:19:04 +08:00
// gou.Debugf("i=%d r=%s state=%d iterTimeRunes %s %s", i, string(r), p.stateTime, p.ds(), p.ts())
2018-03-12 05:08:21 +08:00
2018-02-26 14:46:45 +08:00
switch p.stateTime {
case timeStart:
// 22:43:22
// 22:43
2018-02-26 14:46:45 +08:00
// timeComma
// 08:20:13,787
// timeWs
// 05:24:37 PM
// 06:20:00 UTC
// 06:20:00 UTC-05
2018-02-26 14:46:45 +08:00
// 00:12:00 +0000 UTC
// 22:18:00 +0000 UTC m=+0.000000001
2018-02-26 14:46:45 +08:00
// 15:04:05 -0700
// 15:04:05 -07:00
2018-03-02 10:36:46 +08:00
// 15:04:05 2008
2018-02-26 14:46:45 +08:00
// timeOffset
// 03:21:51+00:00
2018-02-28 12:05:14 +08:00
// 19:55:00+0100
2018-02-26 14:46:45 +08:00
// timePeriod
// 17:24:37.3186369
// 00:07:31.945167
// 18:31:59.257000000
// 00:00:00.000
// (and all variants that can follow the seconds portion of a time format, same as above)
2018-02-26 14:46:45 +08:00
if p.houri == 0 {
p.houri = i
}
switch r {
case ',':
2018-03-11 05:05:55 +08:00
// hm, lets just swap out comma for period. for some reason go
// won't parse it.
// 2014-05-11 08:20:13,787
ds := []byte(p.datestr)
ds[i] = '.'
return parseTime(string(ds), loc, opts...)
2018-02-26 14:46:45 +08:00
case '-', '+':
// 03:21:51+00:00
p.stateTime = timeOffset
2018-04-06 21:57:28 +08:00
if p.seci == 0 {
// 22:18+0530
p.minlen = i - p.mini
} else {
2021-02-06 07:19:04 +08:00
if p.seclen == 0 {
p.seclen = i - p.seci
}
if p.msi > 0 && p.mslen == 0 {
p.mslen = i - p.msi
}
2018-04-06 21:57:28 +08:00
}
2018-02-26 14:46:45 +08:00
p.offseti = i
case '.':
p.stateTime = timePeriod
p.seclen = i - p.seci
p.msi = i + 1
case 'Z':
p.stateTime = timeZ
if p.seci == 0 {
p.minlen = i - p.mini
} else {
p.seclen = i - p.seci
}
2020-07-03 18:21:00 +08:00
// (Z)ulu time
p.loc = time.UTC
endPos := i + 1
if endPos > p.formatSetLen {
p.formatSetLen = endPos
}
case 'a', 'A', 'p', 'P':
if (r == 'a' || r == 'A') && (p.nextIs(i, 't') || p.nextIs(i, 'T')) {
// x
// September 17, 2012 at 5:00pm UTC-05
i++ // skip 't'
if p.nextIs(i, ' ') {
// x
// September 17, 2012 at 5:00pm UTC-05
i++ // skip ' '
p.houri = 0 // reset hour
}
} else {
// Could be AM/PM
isLower := r == 'a' || r == 'p'
isTwoLetterWord := ((i+2) == len(p.datestr) || p.nextIs(i+1, ' '))
switch {
case isLower && p.nextIs(i, 'm') && isTwoLetterWord && !p.parsedAMPM:
p.coalesceTime(i)
p.set(i, "pm")
p.parsedAMPM = true
// skip 'm'
i++
case !isLower && p.nextIs(i, 'M') && isTwoLetterWord && !p.parsedAMPM:
p.coalesceTime(i)
p.set(i, "PM")
p.parsedAMPM = true
// skip 'M'
i++
default:
return p, unexpectedTail(p.datestr[i:])
}
}
2018-02-26 14:46:45 +08:00
case ' ':
p.coalesceTime(i)
2018-02-26 14:46:45 +08:00
p.stateTime = timeWs
case ':':
if p.mini == 0 {
p.mini = i + 1
p.hourlen = i - p.houri
} else if p.seci == 0 {
p.seci = i + 1
p.minlen = i - p.mini
2021-02-06 07:19:04 +08:00
} else if p.seci > 0 {
// 18:31:59:257 ms uses colon, wtf
p.seclen = i - p.seci
p.set(p.seci, "05")
p.msi = i + 1
// gross, gross, gross. manipulating the datestr is horrible.
// https://github.com/araddon/dateparse/issues/117
// Could not get the parsing to work using golang time.Parse() without
// replacing that colon with period.
p.set(i, ".")
newDatestr := p.datestr[0:i] + "." + p.datestr[i+1:]
p.datestr = newDatestr
p.stateTime = timePeriod
2018-02-26 14:46:45 +08:00
}
2018-02-26 08:35:33 +08:00
}
2018-02-26 14:46:45 +08:00
case timeOffset:
2018-02-28 12:05:14 +08:00
// 19:55:00+0100
2018-02-26 14:46:45 +08:00
// timeOffsetColon
2018-02-28 12:05:14 +08:00
// 15:04:05+07:00
// 15:04:05-07:00
2018-02-26 14:46:45 +08:00
if r == ':' {
p.stateTime = timeOffsetColon
2018-02-26 08:35:33 +08:00
}
2018-02-26 14:46:45 +08:00
case timeWs:
2018-03-02 10:36:46 +08:00
// timeWsAlpha
2018-02-26 14:46:45 +08:00
// 06:20:00 UTC
// 06:20:00 UTC-05
2018-03-02 10:36:46 +08:00
// 15:44:11 UTC+0100 2015
// 18:04:07 GMT+0100 (GMT Daylight Time)
2018-03-12 08:39:59 +08:00
// 17:57:51 MST 2009
2018-02-26 14:46:45 +08:00
// timeWsAMPMMaybe
// 05:24:37 PM
// timeWsOffset
// 15:04:05 -0700
2018-03-12 08:39:59 +08:00
// 00:12:00 +0000 UTC
2018-02-26 14:46:45 +08:00
// timeWsOffsetColon
// 15:04:05 -07:00
2018-03-12 05:08:21 +08:00
// 17:57:51 -0700 2009
2018-02-26 14:46:45 +08:00
// timeWsOffsetColonAlpha
// 00:12:00 +00:00 UTC
2018-03-02 10:36:46 +08:00
// timeWsYear
// 00:12:00 2008
// timeWsYearOffset
// 00:12:00 2008 -0700
2018-02-26 14:46:45 +08:00
// timeZ
// 15:04:05.99Z
2018-02-26 08:35:33 +08:00
switch r {
case 'a', 'p', 'A', 'P':
2018-02-26 14:46:45 +08:00
// Could be AM/PM or could be PST or similar
2018-03-02 10:36:46 +08:00
p.tzi = i
2018-02-26 14:46:45 +08:00
p.stateTime = timeWsAMPMMaybe
case '+', '-':
2018-02-28 12:05:14 +08:00
p.offseti = i
2018-02-26 14:46:45 +08:00
p.stateTime = timeWsOffset
default:
if unicode.IsLetter(r) {
// 06:20:00 UTC
// 06:20:00 UTC-05
2018-03-02 10:36:46 +08:00
// 15:44:11 UTC+0100 2015
2018-03-12 08:39:59 +08:00
// 17:57:51 MST 2009
2018-03-02 10:36:46 +08:00
p.tzi = i
2018-02-26 14:46:45 +08:00
p.stateTime = timeWsAlpha
2018-03-02 10:36:46 +08:00
} else if unicode.IsDigit(r) {
// 00:12:00 2008
p.stateTime = timeWsYear
p.yeari = i
2018-02-26 14:46:45 +08:00
}
2018-02-26 08:35:33 +08:00
}
case timeWsYear:
// timeWsYearOffset
// 00:12:00 2008 -0700
switch r {
case ' ':
p.yearlen = i - p.yeari
if !p.setYear() {
return p, unknownErr(datestr)
}
case '+', '-':
p.offseti = i
p.stateTime = timeWsYearOffset
default:
if !unicode.IsDigit(r) {
return p, unknownErr(datestr)
}
}
2018-03-02 10:36:46 +08:00
case timeWsAlpha:
// 06:20:00 UTC
// 06:20:00 UTC-05
// 06:20:00 (EST)
2018-03-12 08:39:59 +08:00
// timeWsAlphaWs
// 17:57:51 MST 2009
2018-03-02 10:36:46 +08:00
// timeWsAlphaZoneOffset
// timeWsAlphaZoneOffsetWs
// timeWsAlphaZoneOffsetWsExtra
// 18:04:07 GMT+0100 (GMT Daylight Time)
// timeWsAlphaZoneOffsetWsYear
// 15:44:11 UTC+0100 2015
switch r {
case '+', '-':
tzNameLower := strings.ToLower(p.datestr[p.tzi:i])
if tzNameLower == "gmt" || tzNameLower == "utc" {
// This is a special form where the actual timezone isn't UTC, but is rather
// specifying that the correct offset is a specified numeric offset from UTC:
// 06:20:00 UTC-05
// 06:20:00 GMT+02
2023-02-16 06:39:34 +08:00
p.tzi = 0
p.tzlen = 0
} else {
p.tzlen = i - p.tzi
}
2018-03-02 10:36:46 +08:00
if p.tzlen == 4 {
p.set(p.tzi, " MST")
2018-03-02 10:36:46 +08:00
} else if p.tzlen == 3 {
p.set(p.tzi, "MST")
}
p.stateTime = timeWsAlphaZoneOffset
p.offseti = i
case ' ', ')':
2018-03-12 08:39:59 +08:00
// 17:57:51 MST 2009
// 17:57:51 MST
// 06:20:00 (EST)
p.tzlen = i - p.tzi
if p.tzlen == 4 {
p.set(p.tzi, " MST")
} else if p.tzlen == 3 {
p.set(p.tzi, "MST")
}
if r == ' ' {
p.stateTime = timeWsAlphaWs
p.yeari = i + 1
} else {
// 06:20:00 (EST)
// This must be the end of the datetime or the format is unknown
if i+1 == len(p.datestr) {
p.stateTime = timeWsAlphaRParen
} else {
return p, unknownErr(datestr)
}
}
2018-03-02 10:36:46 +08:00
}
2018-03-12 08:39:59 +08:00
case timeWsAlphaWs:
// 17:57:51 MST 2009
2018-03-02 10:36:46 +08:00
case timeWsAlphaZoneOffset:
// 06:20:00 UTC-05
2018-03-02 10:36:46 +08:00
// timeWsAlphaZoneOffset
// timeWsAlphaZoneOffsetWs
// timeWsAlphaZoneOffsetWsExtra
// 18:04:07 GMT+0100 (GMT Daylight Time)
// timeWsAlphaZoneOffsetWsYear
// 15:44:11 UTC+0100 2015
switch r {
case ' ':
p.set(p.offseti, "-0700")
if p.yeari == 0 {
p.yeari = i + 1
}
2018-03-02 10:36:46 +08:00
p.stateTime = timeWsAlphaZoneOffsetWs
}
case timeWsAlphaZoneOffsetWs:
// timeWsAlphaZoneOffsetWs
// timeWsAlphaZoneOffsetWsExtra
// 18:04:07 GMT+0100 (GMT Daylight Time)
// timeWsAlphaZoneOffsetWsYear
// 15:44:11 UTC+0100 2015
if unicode.IsDigit(r) {
p.stateTime = timeWsAlphaZoneOffsetWsYear
} else {
p.extra = i - 1
p.stateTime = timeWsAlphaZoneOffsetWsExtra
}
case timeWsAlphaZoneOffsetWsYear:
// 15:44:11 UTC+0100 2015
if unicode.IsDigit(r) {
p.yearlen = i - p.yeari + 1
if p.yearlen == 4 {
if !p.setYear() {
return p, unknownErr(datestr)
}
2018-03-02 10:36:46 +08:00
}
}
2018-02-26 14:46:45 +08:00
case timeWsAMPMMaybe:
// timeWsAMPMMaybe
// timeWsAMPM
// 05:24:37 PM
// timeWsAlpha
// 00:12:00 PST
2018-03-02 10:36:46 +08:00
// 15:44:11 UTC+0100 2015
isTwoLetterWord := ((i+1) == len(p.datestr) || p.nextIs(i, ' '))
if (r == 'm' || r == 'M') && isTwoLetterWord {
if p.parsedAMPM {
return p, unexpectedTail(p.datestr[i:])
}
// This isn't a time zone after all...
p.tzi = 0
2018-02-26 14:46:45 +08:00
p.stateTime = timeWsAMPM
if r == 'm' {
p.set(i-1, "pm")
} else {
p.set(i-1, "PM")
}
p.parsedAMPM = true
if p.hourlen == 2 {
p.set(p.houri, "03")
} else if p.hourlen == 1 {
p.set(p.houri, "3")
}
2018-02-26 14:46:45 +08:00
} else {
p.stateTime = timeWsAlpha
}
case timeWsAMPM:
// If we have a continuation after AM/PM indicator, reset parse state back to ws
if r == ' ' {
p.stateTime = timeWs
} else {
// unexpected garbage after AM/PM indicator, fail
return p, unexpectedTail(p.datestr[i:])
}
2018-02-26 14:46:45 +08:00
case timeWsOffset:
// timeWsOffset
// 15:04:05 -0700
// timeWsOffsetWsOffset
// 17:57:51 -0700 -07
2018-03-12 05:08:21 +08:00
// timeWsOffsetWs
// 17:57:51 -0700 2009
2018-03-12 08:39:59 +08:00
// 00:12:00 +0000 UTC
2018-02-26 14:46:45 +08:00
// timeWsOffsetColon
// 15:04:05 -07:00
// timeWsOffsetColonAlpha
// 00:12:00 +00:00 UTC
2018-02-28 12:05:14 +08:00
switch r {
case ':':
2018-02-26 14:46:45 +08:00
p.stateTime = timeWsOffsetColon
2018-02-28 12:05:14 +08:00
case ' ':
p.set(p.offseti, "-0700")
2018-03-12 08:39:59 +08:00
p.yeari = i + 1
2018-03-12 05:08:21 +08:00
p.stateTime = timeWsOffsetWs
2018-02-26 08:35:33 +08:00
}
2018-03-12 05:08:21 +08:00
case timeWsOffsetWs:
2018-03-12 08:39:59 +08:00
// 17:57:51 -0700 2009
// 00:12:00 +0000 UTC
// 22:18:00.001 +0000 UTC m=+0.000000001
// w Extra
// 17:57:51 -0700 -07
switch r {
case '=':
// eff you golang
if p.datestr[i-1] == 'm' {
p.extra = i - 2
p.trimExtra(false)
}
2021-02-06 05:44:33 +08:00
case '+', '-', '(':
// This really doesn't seem valid, but for some reason when round-tripping a go date
// their is an extra +03 printed out. seems like go bug to me, but, parsing anyway.
// 00:00:00 +0300 +03
// 00:00:00 +0300 +0300
p.extra = i - 1
p.stateTime = timeWsOffset
p.trimExtra(false)
default:
switch {
case unicode.IsDigit(r):
p.yearlen = i - p.yeari + 1
if p.yearlen == 4 {
if !p.setYear() {
return p, unknownErr(datestr)
}
}
case unicode.IsLetter(r):
// 15:04:05 -0700 MST
if p.tzi == 0 {
p.tzi = i
}
}
2018-03-12 08:39:59 +08:00
}
case timeOffsetColon, timeWsOffsetColon:
// timeOffsetColon
// 15:04:05-07:00
// timeOffsetColonAlpha
// 2015-02-18 00:12:00+00:00 UTC
2018-02-26 14:46:45 +08:00
// timeWsOffsetColon
// 15:04:05 -07:00
// timeWsOffsetColonAlpha
// 2015-02-18 00:12:00 +00:00 UTC
if unicode.IsLetter(r) {
// TODO: do we need to handle the m=+0.000000001 case?
2018-02-26 14:46:45 +08:00
// 2015-02-18 00:12:00 +00:00 UTC
if p.stateTime == timeWsOffsetColon {
p.stateTime = timeWsOffsetColonAlpha
} else {
p.stateTime = timeOffsetColonAlpha
}
p.tzi = i
2018-02-26 14:46:45 +08:00
break iterTimeRunes
}
case timePeriod:
// 15:04:05.999999999
// 15:04:05.999999999
// 15:04:05.999999
// 15:04:05.999999
// 15:04:05.999
// 15:04:05.999
2018-02-26 14:46:45 +08:00
// timePeriod
2018-02-28 12:05:14 +08:00
// 17:24:37.3186369
// 00:07:31.945167
// 18:31:59.257000000
// 00:00:00.000
// (note: if we have an offset (+/-) or whitespace (Ws) after this state, re-enter the timeWs or timeOffset
// state above so that we do not have to duplicate all of the logic again for this parsing just because we
// have parsed a fractional second...)
2018-02-26 14:46:45 +08:00
switch r {
case ' ':
p.mslen = i - p.msi
p.coalesceTime(i)
p.stateTime = timeWs
2018-02-26 14:46:45 +08:00
case '+', '-':
p.mslen = i - p.msi
2018-02-28 12:05:14 +08:00
p.offseti = i
p.stateTime = timeOffset
case 'Z':
p.stateTime = timeZ
p.mslen = i - p.msi
// (Z)ulu time
p.loc = time.UTC
endPos := i + 1
if endPos > p.formatSetLen {
p.formatSetLen = endPos
}
case 'a', 'A', 'p', 'P':
// Could be AM/PM
isLower := r == 'a' || r == 'p'
isTwoLetterWord := ((i+2) == len(p.datestr) || p.nextIs(i+1, ' '))
switch {
case isLower && p.nextIs(i, 'm') && isTwoLetterWord && !p.parsedAMPM:
p.mslen = i - p.msi
p.coalesceTime(i)
p.set(i, "pm")
p.parsedAMPM = true
// skip 'm'
i++
p.stateTime = timePeriodAMPM
case !isLower && p.nextIs(i, 'M') && isTwoLetterWord && !p.parsedAMPM:
2018-02-26 14:46:45 +08:00
p.mslen = i - p.msi
p.coalesceTime(i)
p.set(i, "PM")
p.parsedAMPM = true
// skip 'M'
i++
p.stateTime = timePeriodAMPM
default:
return p, unexpectedTail(p.datestr[i:])
2018-02-26 14:46:45 +08:00
}
2018-02-28 12:05:14 +08:00
default:
if !unicode.IsDigit(r) {
return p, unexpectedTail(p.datestr[i:])
2018-02-28 12:05:14 +08:00
}
}
case timePeriodAMPM:
2018-02-28 12:05:14 +08:00
switch r {
case ' ':
p.stateTime = timeWs
case '+', '-':
p.offseti = i
p.stateTime = timeOffset
2018-03-12 08:39:59 +08:00
default:
return p, unexpectedTail(p.datestr[i:])
2018-03-12 08:39:59 +08:00
}
2018-02-26 14:46:45 +08:00
case timeZ:
// nothing expected can come after Z
return p, unexpectedTail(p.datestr[i:])
2018-02-26 08:35:33 +08:00
}
2018-02-26 14:46:45 +08:00
}
2018-02-26 08:35:33 +08:00
2018-02-26 14:46:45 +08:00
switch p.stateTime {
case timeOffsetColonAlpha, timeWsOffsetColonAlpha:
// process offset
offsetLen := i - p.offseti
switch offsetLen {
case 6, 7:
// may or may not have a space on the end
if offsetLen == 7 {
if p.datestr[p.offseti+6] != ' ' {
return p, fmt.Errorf("TZ offset not recognized %q near %q (expected offset like -07:00)", datestr, string(p.datestr[p.offseti:p.offseti+offsetLen]))
}
}
p.set(p.offseti, "-07:00")
default:
return p, fmt.Errorf("TZ offset not recognized %q near %q (expected offset like -07:00)", datestr, string(p.datestr[p.offseti:p.offseti+offsetLen]))
}
// process timezone
switch len(p.datestr) - p.tzi {
case 3:
// 13:31:51.999 +01:00 CET
p.set(p.tzi, "MST")
case 4:
p.set(p.tzi, "MST ")
default:
return p, fmt.Errorf("timezone not recognized %q near %q (must be 3 or 4 characters)", datestr, string(p.datestr[p.tzi:]))
}
case timeWsAlpha:
switch len(p.datestr) - p.tzi {
case 3:
// 13:31:51.999 +01:00 CET
p.set(p.tzi, "MST")
case 4:
p.set(p.tzi, "MST ")
default:
return p, fmt.Errorf("timezone not recognized %q near %q (must be 3 or 4 characters)", datestr, string(p.datestr[p.tzi:]))
}
case timeWsAlphaRParen:
// continue
2018-03-12 08:39:59 +08:00
case timeWsAlphaWs:
p.yearlen = i - p.yeari
if !p.setYear() {
return p, unknownErr(datestr)
}
2018-03-02 10:36:46 +08:00
case timeWsYear:
p.yearlen = i - p.yeari
if !p.setYear() {
return p, unknownErr(datestr)
}
2018-03-02 10:36:46 +08:00
case timeWsAlphaZoneOffsetWsExtra:
p.trimExtra(false)
case timeWsAlphaZoneOffset:
// 06:20:00 UTC-05
2023-02-16 06:39:34 +08:00
switch i - p.offseti {
case 2, 3, 4:
p.set(p.offseti, "-07")
2023-02-16 06:39:34 +08:00
case 5:
p.set(p.offseti, "-0700")
2023-02-16 06:39:34 +08:00
case 6:
p.set(p.offseti, "-07:00")
default:
return p, fmt.Errorf("TZ offset not recognized %q near %q (must be 2 or 4 digits optional colon)", datestr, string(p.datestr[p.offseti:i]))
}
2018-02-26 08:35:33 +08:00
case timePeriod:
2018-02-26 14:46:45 +08:00
p.mslen = i - p.msi
if p.mslen >= 10 {
return p, fmt.Errorf("fractional seconds in %q too long near %q", datestr, string(p.datestr[p.msi:p.mslen]))
}
case timeOffset, timeWsOffset, timeWsYearOffset:
switch len(p.datestr) - p.offseti {
case 3:
// 19:55:00+01 (or 19:55:00 +01)
p.set(p.offseti, "-07")
case 5:
// 19:55:00+0100 (or 19:55:00 +0100)
p.set(p.offseti, "-0700")
default:
return p, fmt.Errorf("TZ offset not recognized %q near %q (must be 2 or 4 digits optional colon)", datestr, string(p.datestr[p.offseti:]))
}
2018-03-12 05:08:21 +08:00
case timeWsOffsetWs:
2018-03-12 08:39:59 +08:00
// 17:57:51 -0700 2009
// 00:12:00 +0000 UTC
if p.tzi > 0 {
switch len(p.datestr) - p.tzi {
case 3:
// 13:31:51.999 +01:00 CET
p.set(p.tzi, "MST")
case 4:
// 13:31:51.999 +01:00 CEST
p.set(p.tzi, "MST ")
default:
return p, fmt.Errorf("timezone not recognized %q near %q (must be 3 or 4 characters)", datestr, string(p.datestr[p.tzi:]))
}
}
case timeOffsetColon, timeWsOffsetColon:
// 17:57:51 -07:00 (or 19:55:00.799 +01:00)
// 15:04:05+07:00 (or 19:55:00.799+01:00)
switch len(p.datestr) - p.offseti {
case 6:
p.set(p.offseti, "-07:00")
default:
return p, fmt.Errorf("TZ offset not recognized %q near %q (expected offset like -07:00)", datestr, string(p.datestr[p.offseti:]))
2018-03-12 08:39:59 +08:00
}
2018-02-26 08:35:33 +08:00
}
2018-02-28 12:05:14 +08:00
p.coalesceTime(i)
2018-02-26 08:35:33 +08:00
}
2018-02-26 14:46:45 +08:00
switch p.stateDate {
2018-02-26 08:35:33 +08:00
case dateDigit:
2014-05-07 12:15:43 +08:00
// unixy timestamps ish
// example ct type
// 1499979655583057426 19 nanoseconds
// 1499979795437000 16 micro-seconds
// 20180722105203 14 yyyyMMddhhmmss
// 1499979795437 13 milliseconds
// 1332151919 10 seconds
// 20140601 8 yyyymmdd
// 2014 4 yyyy
t := time.Time{}
if len(p.datestr) == len("1499979655583057426") { // 19
// nano-seconds
if nanoSecs, err := strconv.ParseInt(p.datestr, 10, 64); err == nil {
t = time.Unix(0, nanoSecs)
2014-04-26 07:59:10 +08:00
}
} else if len(p.datestr) == len("1499979795437000") { // 16
// micro-seconds
if microSecs, err := strconv.ParseInt(p.datestr, 10, 64); err == nil {
t = time.Unix(0, microSecs*1000)
2014-05-07 12:15:43 +08:00
}
} else if len(p.datestr) == len("yyyyMMddhhmmss") { // 14
// yyyyMMddhhmmss
p.setEntireFormat([]byte("20060102150405"))
return p, nil
} else if len(p.datestr) == len("1332151919000") { // 13
if miliSecs, err := strconv.ParseInt(p.datestr, 10, 64); err == nil {
t = time.Unix(0, miliSecs*1000*1000)
2014-10-08 09:30:17 +08:00
}
} else if len(p.datestr) == len("1332151919") { //10
if secs, err := strconv.ParseInt(p.datestr, 10, 64); err == nil {
t = time.Unix(secs, 0)
}
} else if len(p.datestr) == len("20140601") {
p.setEntireFormat([]byte("20060102"))
2018-03-25 06:56:26 +08:00
return p, nil
} else if len(p.datestr) == len("2014") {
p.setEntireFormat([]byte("2006"))
2018-03-25 06:56:26 +08:00
return p, nil
} else if len(p.datestr) < 4 {
return p, fmt.Errorf("unrecognized format, too short %v", datestr)
2014-04-26 07:59:10 +08:00
}
if !t.IsZero() {
if loc == nil {
2018-03-25 06:56:26 +08:00
p.t = &t
return p, nil
}
2018-03-25 06:56:26 +08:00
t = t.In(loc)
p.t = &t
return p, nil
2017-08-13 00:48:28 +08:00
}
2021-02-04 14:23:19 +08:00
case dateDigitSt:
// 171113 14:14:20
return p, nil
2018-10-10 11:50:09 +08:00
case dateYearDash:
2018-02-28 12:05:14 +08:00
// 2006-01
2018-03-25 06:56:26 +08:00
return p, nil
2018-02-28 12:05:14 +08:00
2018-10-10 11:50:09 +08:00
case dateYearDashDash:
2014-05-07 12:15:43 +08:00
// 2006-01-02
2018-02-08 23:03:45 +08:00
// 2006-1-02
// 2006-1-2
// 2006-01-2
2018-03-25 06:56:26 +08:00
return p, nil
2018-02-08 23:03:45 +08:00
case dateYearDashDashOffset:
/// 2020-07-20+00:00
switch len(p.datestr) - p.offseti {
case 5:
p.set(p.offseti, "-0700")
case 6:
p.set(p.offseti, "-07:00")
}
return p, nil
2018-10-10 11:50:09 +08:00
case dateYearDashAlphaDash:
2017-07-27 07:55:21 +08:00
// 2013-Feb-03
2018-02-08 23:03:45 +08:00
// 2013-Feb-3
2018-02-26 14:46:45 +08:00
p.daylen = i - p.dayi
if !p.setDay() {
return p, unknownErr(datestr)
}
2018-03-25 06:56:26 +08:00
return p, nil
2018-02-26 14:46:45 +08:00
2018-10-10 11:50:09 +08:00
case dateYearDashDashWs:
// 2013-04-01
return p, nil
case dateYearDashDashT:
2018-03-25 06:56:26 +08:00
return p, nil
2018-02-26 14:46:45 +08:00
case dateDigitDashAlphaDash, dateDigitDashDigitDash:
// This has already been done if we parsed the time already
if p.stateTime == timeIgnore {
// dateDigitDashAlphaDash:
// 13-Feb-03 ambiguous
// 28-Feb-03 ambiguous
// 29-Jun-2016
// dateDigitDashDigitDash:
// 29-06-2026
length := len(p.datestr) - (p.moi + p.molen + 1)
if length == 4 {
p.yearlen = 4
p.set(p.yeari, "2006")
// We now also know that part1 was the day
p.dayi = 0
p.daylen = p.part1Len
if !p.setDay() {
return p, unknownErr(datestr)
}
} else if length == 2 {
// We have no idea if this is
// yy-mon-dd OR dd-mon-yy
// (or for dateDigitDashDigitDash, yy-mm-dd OR dd-mm-yy)
//
// We are going to ASSUME (bad, bad) that it is dd-mon-yy (dd-mm-yy),
// which is a horrible assumption, but seems to be the convention for
// dates that are formatted in this way.
p.ambiguousMD = true
p.yearlen = 2
p.set(p.yeari, "06")
// We now also know that part1 was the day
p.dayi = 0
p.daylen = p.part1Len
if !p.setDay() {
return p, unknownErr(datestr)
}
} else {
return p, unknownErr(datestr)
}
}
2018-03-25 06:56:26 +08:00
return p, nil
2017-07-27 07:55:21 +08:00
2018-05-04 09:06:52 +08:00
case dateDigitDot:
if len(datestr) == len("yyyyMMddhhmmss.SSS") { // 18
p.setEntireFormat([]byte("20060102150405.000"))
return p, nil
} else {
// 2014.05
p.molen = i - p.moi
if !p.setMonth() {
return p, unknownErr(datestr)
}
return p, nil
}
2018-05-04 09:06:52 +08:00
2018-02-26 08:35:33 +08:00
case dateDigitDotDot:
2018-02-28 12:05:14 +08:00
// 03.31.1981
2018-05-04 09:06:52 +08:00
// 3.31.2014
2018-02-28 12:05:14 +08:00
// 3.2.1981
// 3.2.81
2018-05-04 09:06:52 +08:00
// 08.21.71
2019-02-23 08:50:43 +08:00
// 2018.09.30
2018-03-25 06:56:26 +08:00
return p, nil
2018-01-12 05:21:04 +08:00
case dateDigitDotDotWs:
// 2013.04.01
return p, nil
case dateDigitDotDotT:
return p, nil
case dateDigitDotDotOffset:
// 2020.07.20+00:00
switch len(p.datestr) - p.offseti {
case 5:
p.set(p.offseti, "-0700")
case 6:
p.set(p.offseti, "-07:00")
}
return p, nil
2018-02-28 12:05:14 +08:00
case dateDigitWsMoYear:
2018-03-11 05:43:16 +08:00
// 2 Jan 2018
// 2 Jan 18
// 2 Jan 2018 23:59
// 02 Jan 2018 23:59
2018-01-25 09:59:25 +08:00
// 12 Feb 2006, 19:17
2018-03-25 06:56:26 +08:00
return p, nil
2018-03-11 05:05:55 +08:00
case dateAlphaWsMonth:
p.yearlen = i - p.yeari
if !p.setYear() {
return p, unknownErr(datestr)
}
return p, nil
case dateAlphaWsMonthMore:
return p, nil
case dateAlphaWsDigitMoreWs:
2018-03-11 05:43:16 +08:00
// oct 1, 1970
p.yearlen = i - p.yeari
if !p.setYear() {
return p, unknownErr(datestr)
}
2018-03-25 06:56:26 +08:00
return p, nil
2018-03-11 05:43:16 +08:00
case dateAlphaWsDigitMoreWsYear:
2018-02-26 08:35:33 +08:00
// May 8, 2009 5:57:51 PM
// Jun 7, 2005, 05:57:51
2018-03-25 06:56:26 +08:00
return p, nil
2018-03-02 10:36:46 +08:00
case dateAlphaWsAlpha:
2018-03-25 06:56:26 +08:00
return p, nil
2018-03-02 10:36:46 +08:00
case dateAlphaWsDigit:
return p, nil
case dateAlphaWsDigitYearmaybe:
2018-03-25 06:56:26 +08:00
return p, nil
2018-03-02 10:36:46 +08:00
2018-03-12 08:39:59 +08:00
case dateDigitSlash:
2014-05-07 12:15:43 +08:00
// 3/1/2014
// 10/13/2014
// 01/02/2006
return p, nil
case dateDigitSlashAlpha:
// 03/Jun/2014
return p, nil
case dateDigitYearSlash:
2014-07-11 06:25:23 +08:00
// 2014/10/13
2018-03-25 06:56:26 +08:00
return p, nil
2014-05-07 12:15:43 +08:00
case dateDigitColon:
// 3:1:2014
// 10:13:2014
// 01:02:2006
// 2014:10:13
return p, nil
2018-02-26 08:35:33 +08:00
case dateDigitChineseYear:
// dateDigitChineseYear
2017-11-19 03:07:33 +08:00
// 2014年04月08日
// 2014年4月12日
2018-03-25 06:56:26 +08:00
return p, nil
2018-03-02 10:36:46 +08:00
2018-02-26 08:35:33 +08:00
case dateDigitChineseYearWs:
// 2014年04月08日 00:00:00 ...
2018-03-25 06:56:26 +08:00
return p, nil
2018-03-02 10:36:46 +08:00
case dateAlphaSlashDigitSlash:
// Oct/ 7/1970
// February/07/1970
return p, nil
2018-03-02 10:36:46 +08:00
case dateWeekdayComma:
2017-07-14 00:11:41 +08:00
// Monday, 02 Jan 2006 15:04:05 -0700
// Monday, 02 Jan 2006 15:04:05 +0100
// Monday, 02-Jan-06 15:04:05 MST
2018-03-25 06:56:26 +08:00
return p, nil
2018-03-02 10:36:46 +08:00
2018-03-12 08:39:59 +08:00
case dateWeekdayAbbrevComma:
2017-07-14 00:11:41 +08:00
// Mon, 02-Jan-06 15:04:05 MST
// Mon, 02 Jan 2006 15:04:05 MST
2018-03-25 06:56:26 +08:00
return p, nil
2018-02-08 23:03:45 +08:00
case dateYearWsMonthWs:
// 2013 May 02 11:37:55
// 2013 December 02 11:37:55
return p, nil
2014-04-21 10:56:17 +08:00
}
return p, unknownErr(datestr)
2014-04-21 10:56:17 +08:00
}
2018-10-10 11:50:09 +08:00
type parser struct {
loc *time.Location
preferMonthFirst bool
retryAmbiguousDateWithSwap bool
ambiguousMD bool
allowPartialStringMatch bool
stateDate dateState
stateTime timeState
format []byte
formatSetLen int
datestr string
fullMonth string
parsedAMPM bool
skip int
extra int
part1Len int
yeari int
yearlen int
moi int
molen int
dayi int
daylen int
houri int
hourlen int
mini int
minlen int
seci int
seclen int
msi int
mslen int
offseti int
tzi int
tzlen int
t *time.Time
2018-10-10 11:50:09 +08:00
}
// ParserOption defines a function signature implemented by options
// Options defined like this accept the parser and operate on the data within
type ParserOption func(*parser) error
// PreferMonthFirst is an option that allows preferMonthFirst to be changed from its default
func PreferMonthFirst(preferMonthFirst bool) ParserOption {
return func(p *parser) error {
p.preferMonthFirst = preferMonthFirst
return nil
}
2018-10-10 11:50:09 +08:00
}
// RetryAmbiguousDateWithSwap is an option that allows retryAmbiguousDateWithSwap to be changed from its default
func RetryAmbiguousDateWithSwap(retryAmbiguousDateWithSwap bool) ParserOption {
return func(p *parser) error {
p.retryAmbiguousDateWithSwap = retryAmbiguousDateWithSwap
return nil
}
}
// AllowPartialStringMatch is an option that allows allowPartialStringMatch to be changed from its default.
// If true, then strings can be attempted to be parsed / matched even if the end of the string might contain
// more than a date/time. This defaults to false.
func AllowPartialStringMatch(allowPartialStringMatch bool) ParserOption {
return func(p *parser) error {
p.allowPartialStringMatch = allowPartialStringMatch
return nil
}
}
func newParser(dateStr string, loc *time.Location, opts ...ParserOption) (*parser, error) {
p := &parser{
stateDate: dateStart,
stateTime: timeIgnore,
datestr: dateStr,
loc: loc,
preferMonthFirst: true,
retryAmbiguousDateWithSwap: false,
2018-10-10 11:50:09 +08:00
}
p.format = []byte(dateStr)
// this tracks how much of the format string has been set, to make sure all of it is set
p.formatSetLen = 0
// allow the options to mutate the parser fields from their defaults
for _, option := range opts {
if err := option(p); err != nil {
2023-02-15 12:56:17 +08:00
return nil, fmt.Errorf("option error: %w", err)
}
}
2023-02-15 12:56:17 +08:00
return p, nil
2018-10-10 11:50:09 +08:00
}
func (p *parser) nextIs(i int, b byte) bool {
if len(p.datestr) > i+1 && p.datestr[i+1] == b {
return true
}
return false
}
func (p *parser) setEntireFormat(format []byte) {
p.format = format
p.formatSetLen = len(format)
}
2018-10-10 11:50:09 +08:00
func (p *parser) set(start int, val string) {
if start < 0 {
return
}
if len(p.format) < start+len(val) {
return
}
for i, r := range val {
p.format[start+i] = byte(r)
}
endingPos := start + len(val)
if endingPos > p.formatSetLen {
p.formatSetLen = endingPos
}
2018-10-10 11:50:09 +08:00
}
func (p *parser) setMonth() bool {
2018-10-10 11:50:09 +08:00
if p.molen == 2 {
p.set(p.moi, "01")
return true
2018-10-10 11:50:09 +08:00
} else if p.molen == 1 {
p.set(p.moi, "1")
return true
} else {
return false
2018-10-10 11:50:09 +08:00
}
}
func (p *parser) setDay() bool {
2018-10-10 11:50:09 +08:00
if p.daylen == 2 {
p.set(p.dayi, "02")
return true
2018-10-10 11:50:09 +08:00
} else if p.daylen == 1 {
p.set(p.dayi, "2")
return true
} else {
return false
2018-10-10 11:50:09 +08:00
}
}
func (p *parser) setYear() bool {
2018-10-10 11:50:09 +08:00
if p.yearlen == 2 {
p.set(p.yeari, "06")
return true
2018-10-10 11:50:09 +08:00
} else if p.yearlen == 4 {
p.set(p.yeari, "2006")
return true
} else {
return false
}
}
// Find the proper end of the current component (scanning chars starting from start and going
// up until the end, and either returning at end or returning the first character that is
// not allowed, as determined by allowNumeric, allowAlpha, and allowOther)
func findProperEnd(s string, start, end int, allowNumeric bool, allowAlpha bool, allowOther bool) int {
for i := start; i < end; i++ {
c := s[i]
if c >= '0' && c <= '9' {
if !allowNumeric {
return i
}
} else if (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') {
if !allowAlpha {
return i
}
} else {
if !allowOther {
return i
}
}
2018-10-10 11:50:09 +08:00
}
return end
2018-10-10 11:50:09 +08:00
}
func (p *parser) coalesceDate(end int) bool {
2018-10-10 11:50:09 +08:00
if p.yeari > 0 {
if p.yearlen == 0 {
p.yearlen = findProperEnd(p.datestr, p.yeari, end, true, false, false) - p.yeari
}
if !p.setYear() {
return false
2018-10-10 11:50:09 +08:00
}
}
if p.moi > 0 && p.molen == 0 {
p.molen = findProperEnd(p.datestr, p.moi, end, true, true, false) - p.moi
// The month may be the name of the month, so don't treat as invalid in this case.
// We can ignore the return value here.
2018-10-10 11:50:09 +08:00
p.setMonth()
}
if p.dayi > 0 && p.daylen == 0 {
p.daylen = findProperEnd(p.datestr, p.dayi, end, true, false, false) - p.dayi
if !p.setDay() {
return false
}
2018-10-10 11:50:09 +08:00
}
return true
2018-10-10 11:50:09 +08:00
}
func (p *parser) ts() string {
return fmt.Sprintf("h:(%d:%d) m:(%d:%d) s:(%d:%d)", p.houri, p.hourlen, p.mini, p.minlen, p.seci, p.seclen)
}
func (p *parser) ds() string {
return fmt.Sprintf("%s d:(%d:%d) m:(%d:%d) y:(%d:%d)", p.datestr, p.dayi, p.daylen, p.moi, p.molen, p.yeari, p.yearlen)
}
func (p *parser) coalesceTime(end int) {
// 03:04:05
// 15:04:05
// 3:04:05
// 3:4:5
// 15:04:05.00
if p.houri > 0 {
if p.hourlen == 2 {
p.set(p.houri, "15")
} else if p.hourlen == 1 {
p.set(p.houri, "3")
}
}
if p.mini > 0 {
if p.minlen == 0 {
p.minlen = end - p.mini
}
if p.minlen == 2 {
p.set(p.mini, "04")
} else {
p.set(p.mini, "4")
}
}
if p.seci > 0 {
if p.seclen == 0 {
p.seclen = end - p.seci
}
if p.seclen == 2 {
p.set(p.seci, "05")
} else {
p.set(p.seci, "5")
}
}
if p.msi > 0 {
for i := 0; i < p.mslen; i++ {
p.format[p.msi+i] = '0'
}
endPos := p.msi + p.mslen
if endPos > p.formatSetLen {
p.formatSetLen = endPos
}
2018-10-10 11:50:09 +08:00
}
}
func (p *parser) setFullMonth(month string) {
oldLen := len(p.format)
const fullMonth = "January"
p.format = []byte(fmt.Sprintf("%s%s%s", p.format[0:p.moi], fullMonth, p.format[p.moi+len(month):]))
newLen := len(p.format)
if newLen > oldLen && p.formatSetLen >= p.moi {
p.formatSetLen += newLen - oldLen
} else if newLen < oldLen && p.formatSetLen >= p.moi {
p.formatSetLen -= oldLen - newLen
}
if p.formatSetLen > len(p.format) {
p.formatSetLen = len(p.format)
} else if p.formatSetLen < len(fullMonth) {
p.formatSetLen = len(fullMonth)
} else if p.formatSetLen < 0 {
p.formatSetLen = 0
}
}
2018-10-10 11:50:09 +08:00
func (p *parser) trimExtra(onlyTrimFormat bool) {
2018-10-10 11:50:09 +08:00
if p.extra > 0 && len(p.format) > p.extra {
p.format = p.format[0:p.extra]
if p.formatSetLen > len(p.format) {
p.formatSetLen = len(p.format)
}
if !onlyTrimFormat {
p.datestr = p.datestr[0:p.extra]
}
2018-10-10 11:50:09 +08:00
}
}
func (p *parser) parse() (time.Time, error) {
if p.t != nil {
return *p.t, nil
}
if len(p.fullMonth) > 0 {
p.setFullMonth(p.fullMonth)
}
// Make sure that the entire string matched to a known format that was detected
if !p.allowPartialStringMatch && p.formatSetLen < len(p.format) {
// We can always ignore punctuation at the end of a date/time, but do not allow
// any numbers or letters in the format string.
validFormatTo := findProperEnd(string(p.format), p.formatSetLen, len(p.format), false, false, true)
if validFormatTo < len(p.format) {
return time.Time{}, unexpectedTail(string(p.format[p.formatSetLen:]))
}
}
2018-10-10 11:50:09 +08:00
if p.skip > 0 && len(p.format) > p.skip {
p.format = p.format[p.skip:]
p.formatSetLen -= p.skip
if p.formatSetLen < 0 {
p.formatSetLen = 0
}
2018-10-10 11:50:09 +08:00
p.datestr = p.datestr[p.skip:]
}
2021-02-04 14:23:19 +08:00
2018-10-10 11:50:09 +08:00
if p.loc == nil {
// gou.Debugf("parse layout=%q input=%q \ntx, err := time.Parse(%q, %q)", string(p.format), p.datestr, string(p.format), p.datestr)
2018-10-10 11:50:09 +08:00
return time.Parse(string(p.format), p.datestr)
}
//gou.Debugf("parse layout=%q input=%q \ntx, err := time.ParseInLocation(%q, %q, %v)", string(p.format), p.datestr, string(p.format), p.datestr, p.loc)
2018-10-10 11:50:09 +08:00
return time.ParseInLocation(string(p.format), p.datestr, p.loc)
}
func isDay(alpha string) bool {
for _, day := range days {
if alpha == day {
return true
}
}
return false
}
func isMonthFull(alpha string) bool {
if len(alpha) > len("september") {
return false
}
for _, month := range months {
if alpha == month {
return true
}
}
return false
}