go-pkg/dateparse
2
0
Fork 0
mirror of https://github.com/araddon/dateparse.git synced 2024-11-10 11:51:54 +08:00
Code Issues Projects Releases Wiki Activity
a45d593447
dateparse/parseany.go
Klondike Dragon a45d593447 Optimize checks for day of week and full month 
Reduces CPU usage on large benchmarks by ~2%-3% and prepares for future with international month names in future.
2023-12-16 23:40:14 -07:00

3090 lines
77 KiB
Go
Raw Blame History

// 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.
package dateparse
import (
"fmt"
"strconv"
"strings"
"sync"
"time"
"unicode"
"unicode/utf8"
"unsafe"
)
// func init() {
// gou.SetupLogging("debug")
// gou.SetColorOutput()
// }
var knownDays = map[string]struct{}{
"mon": {},
"tue": {},
"wed": {},
"thu": {},
"fri": {},
"sat": {},
"sun": {},
"monday": {},
"tuesday": {},
"wednesday": {},
"thursday": {},
"friday": {},
"saturday": {},
"sunday": {},
}
var knownMonths = map[string]struct{}{
"january": {},
"february": {},
"march": {},
"april": {},
"may": {},
"june": {},
"july": {},
"august": {},
"september": {},
"october": {},
"november": {},
"december": {},
}
type dateState uint8
type timeState uint8
const (
dateStart dateState = iota // 0
dateDigit
dateDigitSt
dateYearDash
dateYearDashAlpha
dateYearDashDash
dateYearDashDashWs // 6
dateYearDashDashT
dateYearDashDashOffset
dateDigitDash
dateDigitDashAlpha
dateDigitDashAlphaDash // 11
dateDigitDashDigit
dateDigitDashDigitDash
dateDigitDot
dateDigitDotDot
dateDigitDotDotWs
dateDigitDotDotT
dateDigitDotDotOffset
dateDigitSlash
dateDigitYearSlash
dateDigitSlashAlpha // 21
dateDigitSlashAlphaSlash
dateDigitColon
dateDigitChineseYear
dateDigitChineseYearWs
dateDigitWs
dateDigitWsMoYear // 27
dateAlpha
dateAlphaWs
dateAlphaWsDigit
dateAlphaWsDigitMore // 31
dateAlphaWsDigitMoreWs
dateAlphaWsDigitMoreWsYear
dateAlphaWsDigitYearMaybe
dateVariousDaySuffix
dateAlphaFullMonthWs
dateAlphaFullMonthWsDayWs
dateAlphaWsAlpha
dateAlphaPeriodWsDigit
dateAlphaSlash
dateAlphaSlashDigit
dateAlphaSlashDigitSlash
dateWeekdayComma
dateWeekdayAbbrevComma
dateYearWs
dateYearWsMonthWs
)
const (
// Time state
timeIgnore timeState = iota // 0
timeStart
timeWs
timeWsAlpha
timeWsAlphaRParen
timeWsAlphaWs
timeWsAlphaZoneOffset // 6
timeWsAlphaZoneOffsetWs
timeWsAlphaZoneOffsetWsYear
timeWsAlphaZoneOffsetWsExtra
timeWsAMPMMaybe
timeWsAMPM // 11
timeWsOffset
timeWsOffsetWs // 13
timeWsOffsetColonAlpha
timeWsOffsetColon
timeWsYear // 16
timeWsYearOffset
timeOffset
timeOffsetColon
timeOffsetColonAlpha
timePeriod
timePeriodAMPM
timeZ
)
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: ")
ErrUnknownTZOffset = fmt.Errorf("TZ offset not recognized")
ErrUnknownTimeZone = fmt.Errorf("timezone not recognized")
ErrFracSecTooLong = fmt.Errorf("fractional seconds too long")
)
func (p *parser) unknownErr(datestr string) error {
if p == nil || !p.simpleErrorMessages {
return fmt.Errorf("%w %q", ErrCouldntFindFormat, datestr)
} else {
return ErrCouldntFindFormat
}
}
func (p *parser) unexpectedTail(tailStart int) error {
if p != nil && !p.simpleErrorMessages {
return fmt.Errorf("%w %q", ErrUnexpectedTail, p.datestr[tailStart:])
} else {
return ErrUnexpectedTail
}
}
// go 1.20 allows us to convert a byte slice to a string without a memory allocation.
// See https://github.com/golang/go/issues/53003#issuecomment-1140276077.
func bytesToString(b []byte) string {
if b == nil || len(b) <= 0 {
return ""
} else {
return unsafe.String(&b[0], len(b))
}
}
// 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...)
defer putBackParser(p)
if err != nil {
return time.Time{}, err
}
return p.parse(nil, opts...)
}
// ParseIn with Location, equivalent to time.ParseInLocation() timezone/offset
// 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...)
defer putBackParser(p)
if err != nil {
return time.Time{}, err
}
return p.parse(loc, opts...)
}
// ParseLocal Given an unknown date format, detect the layout,
// using time.Local, parse.
//
// Set Location to time.Local. Same as ParseIn Location but lazily uses
// the global time.Local variable for Location argument.
//
// denverLoc, _ := time.LoadLocation("America/Denver")
// time.Local = denverLoc
//
// t, err := dateparse.ParseLocal("3/1/2014")
//
// Equivalent to:
//
// t, err := dateparse.ParseIn("3/1/2014", denverLoc)
func ParseLocal(datestr string, opts ...ParserOption) (time.Time, error) {
p, err := parseTime(datestr, time.Local, opts...)
defer putBackParser(p)
if err != nil {
return time.Time{}, err
}
return p.parse(time.Local, opts...)
}
// 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...)
defer putBackParser(p)
if err != nil {
panic(err.Error())
}
t, err := p.parse(nil, opts...)
if err != nil {
panic(err.Error())
}
return t
}
// 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.
//
// 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...)
defer putBackParser(p)
if err != nil {
return "", err
}
_, err = p.parse(nil, opts...)
if err != nil {
return "", err
}
return string(p.format), nil
}
// ParseStrict parse an unknown date format. IF the date is ambigous
// 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...)
defer putBackParser(p)
if err != nil {
return time.Time{}, err
}
if p.ambiguousMD {
return time.Time{}, ErrAmbiguousMMDD
}
return p.parse(nil, opts...)
}
// Creates a new parser and parses the given datestr in the given loc with the given options.
// The caller must call putBackParser on the returned parser when done with it.
func parseTime(datestr string, loc *time.Location, opts ...ParserOption) (p *parser, err error) {
p, err = newParser(datestr, loc, opts...)
if err != nil {
return
}
// 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.
i := 0
// 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
iterRunes:
for ; i < len(p.datestr); i++ {
r, bytesConsumed := utf8.DecodeRuneInString(p.datestr[i:])
if bytesConsumed > 1 {
i += bytesConsumed - 1
}
// gou.Debugf("i=%d r=%s state=%d %s", i, string(r), p.stateDate, p.datestr)
switch p.stateDate {
case dateStart:
// NOTE: don't use unicode.IsDigit and unicode.IsLetter here because
// we don't expect non-ANSI chars to start a valid date/time format.
// This will let us quickly reject strings that begin with any non-ANSI char.
if '0' <= r && r <= '9' {
p.stateDate = dateDigit
} else if ('a' <= r && r <= 'z') || ('A' <= r && r <= 'Z') {
p.stateDate = dateAlpha
} else {
return p, p.unknownErr(datestr)
}
case dateDigit:
switch r {
case '-', '\u2212':
// 2006-01-02
// 2013-Feb-03
// 13-Feb-03
// 29-Jun-2016
if i == 4 {
p.stateDate = dateYearDash
p.yeari = 0
p.yearlen = i
p.moi = i + 1
p.set(0, "2006")
} else {
p.stateDate = dateDigitDash
}
case '/':
// 08/May/2005
// 03/31/2005
// 2014/02/24
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, 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, p.unknownErr(datestr)
}
continue
}
// Ambiguous dd/mm vs mm/dd the bane of date-parsing
// 03/31/2005
// 31/03/2005
p.ambiguousMD = true
p.ambiguousRetryable = true
if p.preferMonthFirst {
if p.molen == 0 {
// 03/31/2005
p.molen = i
if !p.setMonth() {
return p, p.unknownErr(datestr)
}
p.dayi = i + 1
} else {
return p, p.unknownErr(datestr)
}
} else {
if p.daylen == 0 {
p.daylen = i
if !p.setDay() {
return p, p.unknownErr(datestr)
}
p.moi = i + 1
} else {
return p, p.unknownErr(datestr)
}
}
}
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, p.unknownErr(datestr)
}
} else {
p.ambiguousMD = true
p.ambiguousRetryable = true
if p.preferMonthFirst {
if p.molen == 0 {
p.molen = i
if !p.setMonth() {
return p, p.unknownErr(datestr)
}
p.dayi = i + 1
} else {
return p, p.unknownErr(datestr)
}
} else {
if p.daylen == 0 {
p.daylen = i
if !p.setDay() {
return p, p.unknownErr(datestr)
}
p.moi = i + 1
} else {
return p, p.unknownErr(datestr)
}
}
}
case '.':
// 3.31.2014
// 08.21.71
// 2014.05
p.stateDate = dateDigitDot
if i == 4 {
p.yearlen = i
p.moi = i + 1
if !p.setYear() {
return p, p.unknownErr(datestr)
}
} else if i <= 2 {
p.ambiguousMD = true
p.ambiguousRetryable = true
if p.preferMonthFirst {
if p.molen == 0 {
// 03.31.2005
p.molen = i
if !p.setMonth() {
return p, p.unknownErr(datestr)
}
p.dayi = i + 1
} else {
return p, p.unknownErr(datestr)
}
} else {
if p.daylen == 0 {
p.daylen = i
if !p.setDay() {
return p, p.unknownErr(datestr)
}
p.moi = i + 1
} else {
return p, p.unknownErr(datestr)
}
}
}
// else this might be a unixy combined datetime of the form:
// yyyyMMddhhmmss.SSS
case ' ':
// 18 January 2018
// 8 January 2018
// 8 jan 2018
// 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, p.unknownErr(datestr)
}
p.stateDate = dateYearWs
} else if i == 6 {
p.stateDate = dateDigitSt
} else {
p.stateDate = dateDigitWs
p.dayi = 0
p.daylen = i
}
case '年':
// Chinese Year
p.stateDate = dateDigitChineseYear
p.yearlen = i - 2
p.moi = i + 1
if !p.setYear() {
return p, p.unknownErr(datestr)
}
case ',':
return p, p.unknownErr(datestr)
case 's', 'S', 'r', 'R', 't', 'T', 'n', 'N':
// 1st January 2018
// 2nd Jan 2018 23:59
// st, rd, nd, th
p.stateDate = dateVariousDaySuffix
i--
default:
if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
continue
}
p.part1Len = i
case dateDigitSt:
p.set(0, "060102")
i = i - 1
p.stateTime = timeStart
break iterRunes
case dateYearDash:
// dateYearDashDashT
// 2006-01-02T15:04:05Z07:00
// 2020-08-17T17:00:00:000+0100
// dateYearDashDashWs
// 2013-04-01 22:43:22
// dateYearDashAlpha
// 2013-Feb-03
// 2013-February-03
switch r {
case '-', '\u2212':
p.molen = i - p.moi
p.dayi = i + 1
p.stateDate = dateYearDashDash
if !p.setMonth() {
return p, p.unknownErr(datestr)
}
default:
if unicode.IsLetter(r) {
p.stateDate = dateYearDashAlpha
} else if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
}
case dateYearDashDash:
// dateYearDashDashT
// 2006-01-02T15:04:05Z07:00
// dateYearDashDashWs
// 2013-04-01 22:43:22
// dateYearDashDashOffset
// 2020-07-20+00:00
// (these states are also reused after dateYearDashAlpha, like 2020-July-20...)
switch r {
case '+', '-':
p.offseti = i
p.daylen = i - p.dayi
p.stateDate = dateYearDashDashOffset
if !p.setDay() {
return p, p.unknownErr(datestr)
}
case ' ':
p.daylen = i - p.dayi
p.stateDate = dateYearDashDashWs
p.stateTime = timeStart
if !p.setDay() {
return p, p.unknownErr(datestr)
}
break iterRunes
case 'T':
p.daylen = i - p.dayi
p.stateDate = dateYearDashDashT
p.stateTime = timeStart
if !p.setDay() {
return p, p.unknownErr(datestr)
}
break iterRunes
default:
if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
}
case dateYearDashDashT:
// dateYearDashDashT
// 2006-01-02T15:04:05Z07:00
// 2020-08-17T17:00:00:000+0100
// (this state should never be reached, we break out when in this state)
return p, p.unknownErr(datestr)
case dateYearDashDashOffset:
// 2020-07-20+00:00
switch r {
case ':':
p.set(p.offseti, "-07:00")
default:
if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
}
case dateYearDashAlpha:
// dateYearDashAlpha
// 2013-Feb-03
// 2013-February-03
switch r {
case '-', '\u2212':
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 = dateYearDashDash
} 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 = dateYearDashDash
} else {
return p, p.unknownErr(datestr)
}
}
default:
if !unicode.IsLetter(r) {
return p, p.unknownErr(datestr)
}
}
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
} else {
return p, p.unknownErr(datestr)
}
case dateDigitDashAlpha:
// 13-Feb-03
// 28-Feb-03
// 29-Jun-2016
switch r {
case '-', '\u2212':
p.molen = i - p.moi
p.set(p.moi, "Jan")
p.yeari = i + 1
p.stateDate = dateDigitDashAlphaDash
default:
if !unicode.IsLetter(r) {
return p, p.unknownErr(datestr)
}
}
case dateDigitDashDigit:
// 29-06-2026
switch r {
case '-', '\u2212':
// 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, p.unknownErr(datestr)
}
default:
if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
}
case dateDigitDashAlphaDash, dateDigitDashDigitDash:
// dateDigitDashAlphaDash:
// 13-Feb-03 ambiguous
// 28-Feb-03 ambiguous
// 29-Jun-2016 dd-month(alpha)-yyyy
// 8-Mar-2018::
// dateDigitDashDigitDash:
// 29-06-2026
// 08-03-18:: ambiguous (dd-mm-yy or yy-mm-dd)
switch r {
case ' ', ':':
doubleColonTimeConnector := false
if r == ':' {
p.link++
if p.link == 2 {
if i+1 < len(p.datestr) {
// only legitimate content to follow "::" is the start of the time
nextChar, _ := utf8.DecodeRuneInString(p.datestr[i+1:])
if unicode.IsDigit(nextChar) {
doubleColonTimeConnector = true
}
}
if !doubleColonTimeConnector {
return p, p.unknownErr(datestr)
}
}
} else if p.link > 0 {
return p, p.unknownErr(datestr)
}
if r == ' ' || doubleColonTimeConnector {
// we need to find if this was 4 digits, aka year
// or 2 digits which makes it ambiguous year/day
var sepLen int
if doubleColonTimeConnector {
sepLen = 2
} else {
sepLen = 1
}
length := i - (p.moi + p.molen + sepLen)
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, 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 // not retryable
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, p.unknownErr(datestr)
}
} else {
return p, p.unknownErr(datestr)
}
p.stateTime = timeStart
break iterRunes
}
default:
if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
}
case dateDigitYearSlash:
// 2014/07/10 06:55:38.156283
// I honestly don't know if this format ever shows up as yyyy/
switch r {
case ' ':
fallthrough
case ':':
p.stateTime = timeStart
if p.daylen == 0 {
p.daylen = i - p.dayi
if !p.setDay() {
return p, p.unknownErr(datestr)
}
}
break iterRunes
case '/':
if p.molen == 0 {
p.molen = i - p.moi
if !p.setMonth() {
return p, p.unknownErr(datestr)
}
p.dayi = i + 1
}
default:
if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
}
case dateDigitSlashAlpha:
// 06/May/2008
// 06/September/2008
switch r {
case '/':
// |
// 06/May/2008
if p.molen == 0 {
p.molen = i - p.moi
if p.molen == 3 {
p.set(p.moi, "Jan")
p.yeari = i + 1
p.stateDate = dateDigitSlashAlphaSlash
} else {
possibleFullMonth := strings.ToLower(p.datestr[p.moi:(p.moi + p.molen)])
if i > 3 && isMonthFull(possibleFullMonth) {
p.fullMonth = possibleFullMonth
p.yeari = i + 1
p.stateDate = dateDigitSlashAlphaSlash
} else {
return p, p.unknownErr(datestr)
}
}
} else {
return p, p.unknownErr(datestr)
}
default:
if !unicode.IsLetter(r) {
return p, p.unknownErr(datestr)
}
}
case dateDigitSlashAlphaSlash:
switch r {
case ' ':
fallthrough
case ':':
p.stateTime = timeStart
if p.yearlen == 0 {
p.yearlen = i - p.yeari
if !p.setYear() {
return p, p.unknownErr(datestr)
}
}
break iterRunes
default:
if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
}
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 {
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, p.unknownErr(datestr)
}
p.yeari = i + 1
}
} else {
if p.molen == 0 {
p.molen = i - p.moi
if !p.setMonth() {
return p, 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, p.unknownErr(datestr)
}
}
break iterRunes
default:
if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
}
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, p.unknownErr(datestr)
}
} else if p.daylen == 0 {
p.daylen = i - p.dayi
if !p.setDay() {
return p, p.unknownErr(datestr)
}
} else if p.molen == 0 {
p.molen = i - p.moi
if !p.setMonth() {
return p, 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, 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, p.unknownErr(datestr)
}
p.yeari = i + 1
}
} else {
if p.molen == 0 {
p.molen = i - p.moi
if !p.setMonth() {
return p, p.unknownErr(datestr)
}
p.yeari = i + 1
}
}
default:
if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
}
case dateDigitWs:
// 18 January 2018
// 8 January 2018
// 8 jan 2018
// 1 jan 18
// 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
//p.yearlen = 4
p.dayi = 0
p.daylen = p.part1Len
if !p.setDay() {
return p, p.unknownErr(datestr)
}
p.stateTime = timeStart
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, p.unknownErr(datestr)
}
} 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
p.molen = i - p.moi
p.set(p.moi, "Jan")
p.stateDate = dateDigitWsMoYear
}
default:
if !unicode.IsLetter(r) {
return p, p.unknownErr(datestr)
}
}
case dateDigitWsMoYear:
// 8 jan 2018
// 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.yearlen = i - p.yeari
if !p.setYear() {
return p, p.unknownErr(datestr)
}
i++
break iterRunes
case ' ':
p.yearlen = i - p.yeari
if !p.setYear() {
return p, p.unknownErr(datestr)
}
break iterRunes
default:
if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
}
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, p.unknownErr(datestr)
}
}
} else if !unicode.IsLetter(r) {
return p, 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
default:
if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
}
case dateDigitChineseYear:
// dateDigitChineseYear
// 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, p.unknownErr(datestr)
}
case '日':
// day
p.daylen = i - p.dayi - 2
if !p.setDay() {
return p, p.unknownErr(datestr)
}
case ' ':
if p.daylen <= 0 {
return p, p.unknownErr(datestr)
}
p.stateDate = dateDigitChineseYearWs
p.stateTime = timeStart
break iterRunes
default:
if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
}
case dateDigitDot:
// 3.31.2014
// 08.21.71
// 2014.05
// 2018.09.30
// This is the 2nd period
if r == '.' {
if p.moi == 0 {
// 3.31.2014
p.daylen = i - p.dayi
p.yeari = i + 1
if !p.setDay() {
return p, p.unknownErr(datestr)
}
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, p.unknownErr(datestr)
}
p.stateDate = dateDigitDotDot
} else {
// 2018.09.30
// p.molen = 2
p.molen = i - p.moi
p.dayi = i + 1
if !p.setMonth() {
return p, p.unknownErr(datestr)
}
p.stateDate = dateDigitDotDot
}
} else if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
case dateDigitDotDot:
// dateDigitDotDotT
// 2006.01.02T15:04:05Z07:00
// dateDigitDotDotWs
// 2013.04.01 22:43:22
// dateDigitDotDotOffset
// 2020.07.20+00:00
switch r {
case '+', '-':
p.offseti = i
p.daylen = i - p.dayi
p.stateDate = dateDigitDotDotOffset
if !p.setDay() {
return p, p.unknownErr(datestr)
}
case ' ':
p.daylen = i - p.dayi
p.stateDate = dateDigitDotDotWs
p.stateTime = timeStart
if !p.setDay() {
return p, p.unknownErr(datestr)
}
break iterRunes
case 'T':
p.daylen = i - p.dayi
p.stateDate = dateDigitDotDotT
p.stateTime = timeStart
if !p.setDay() {
return p, p.unknownErr(datestr)
}
break iterRunes
default:
if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
}
case dateDigitDotDotT:
// dateDigitDotDotT
// 2006-01-02T15:04:05Z07:00
// 2020-08-17T17:00:00:000+0100
// (should be unreachable, we break in this state)
return p, p.unknownErr(datestr)
case dateDigitDotDotOffset:
// 2020-07-20+00:00
switch r {
case ':':
p.set(p.offseti, "-07:00")
default:
if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
}
case dateAlpha:
// dateAlphaWs
// 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
// Mon Aug 10 15:44:11 UTC+0100 2015
// Fri Jul 03 2015 18:04:07 GMT+0100 (GMT Daylight Time)
// dateAlphaWsDigit
// May 8, 2009 5:57:51 PM
// oct 1, 1970
// dateAlphaFullMonthWs
// January 02, 2006 3:04pm
// January 02, 2006 3:04pm MST-07
// January 02, 2006 at 3:04pm MST-07
//
// dateAlphaPeriodWsDigit
// oct. 1, 1970
// dateAlphaSlash
// dateAlphaSlashDigit
// dateAlphaSlashDigitSlash
// Oct/ 7/1970
// Oct/07/1970
// February/ 7/1970
// February/07/1970
//
// 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
// 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
// Tue, 11 Jul 2017 16:28:13 +0200 (CEST)
// Mon, 02-Jan-06 15:04:05 MST
switch {
case r == ' ':
// This could be a weekday or a month, detect and parse both cases.
// skip & return to dateStart
// Tue 05 May 2020, 05:05:05
// Tuesday 05 May 2020, 05:05:05
// Mon Jan 2 15:04:05 2006
// Monday Jan 2 15:04:05 2006
maybeDayOrMonth := strings.ToLower(p.datestr[0:i])
if isDay(maybeDayOrMonth) {
// using skip throws off indices used by other code; saner to restart
newDateStr := p.datestr[i+1:]
putBackParser(p)
return parseTime(newDateStr, loc)
}
// X
// April 8, 2009
if i > 3 {
// Expecting a full month name at this point
if isMonthFull(maybeDayOrMonth) {
p.moi = 0
p.molen = i
p.fullMonth = maybeDayOrMonth
p.stateDate = dateAlphaFullMonthWs
p.dayi = i + 1
break
} else {
return p, p.unknownErr(datestr)
}
} else if i == 3 {
// 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
p.stateDate = dateAlphaWs
} else {
return p, p.unknownErr(datestr)
}
case r == ',':
// Mon, 02 Jan 2006
if i == 3 {
p.stateDate = dateWeekdayAbbrevComma
p.set(0, "Mon")
} else {
maybeDay := strings.ToLower(p.datestr[0:i])
if isDay(maybeDay) {
p.stateDate = dateWeekdayComma
// Just skip past the weekday, it contains no valuable info
p.skip = i + 2
i++
} else {
return p, p.unknownErr(datestr)
}
}
case r == '.':
// sept. 28, 2017
// jan. 28, 2017
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:]
putBackParser(p)
return parseTime(newDateStr, loc, opts...)
} else {
return p, 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, p.unknownErr(datestr)
}
}
default:
if !unicode.IsLetter(r) {
return p, p.unknownErr(datestr)
}
}
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
// 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
switch {
case unicode.IsLetter(r):
p.set(0, "Mon")
p.stateDate = dateAlphaWsAlpha
p.set(i, "Jan")
case unicode.IsDigit(r):
p.set(0, "Jan")
p.stateDate = dateAlphaWsDigit
p.dayi = i
case r == ' ':
// continue
default:
return p, p.unknownErr(datestr)
}
case dateAlphaWsDigit:
// May 8, 2009 5:57:51 PM
// May 8 2009 5:57:51 PM
// oct 1, 1970
// oct 7, '70
// oct. 7, 1970
// May 8 17:57:51 2009
// May 8 17:57:51 2009
// May 08 17:57:51 2009
if r == ',' {
p.daylen = i - p.dayi
if !p.setDay() {
return p, p.unknownErr(datestr)
}
p.stateDate = dateAlphaWsDigitMore
} else if r == ' ' {
p.daylen = i - p.dayi
if !p.setDay() {
return p, p.unknownErr(datestr)
}
p.yeari = i + 1
p.stateDate = dateAlphaWsDigitYearMaybe
p.stateTime = timeStart
} else if unicode.IsLetter(r) {
p.stateDate = dateVariousDaySuffix
i--
} else if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
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, p.unknownErr(datestr)
}
break iterRunes
} else if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
case dateAlphaWsDigitMore:
// x
// May 8, 2009 5:57:51 PM
// May 05, 2005, 05:05:05
// May 05 2005, 05:05:05
// oct 1, 1970
// oct 7, '70
if r == ' ' {
p.yeari = i + 1
p.stateDate = dateAlphaWsDigitMoreWs
} else {
return p, p.unknownErr(datestr)
}
case dateAlphaWsDigitMoreWs:
// x
// May 8, 2009 5:57:51 PM
// May 05, 2005, 05:05:05
// 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
p.yearlen = i - p.yeari
if !p.setYear() {
return p, p.unknownErr(datestr)
}
p.stateTime = timeStart
break iterRunes
default:
if r != '\'' && !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
}
case dateVariousDaySuffix:
// x
// April 8th, 2009
// April 8th 2009
switch r {
case 't', 'T':
if p.nextIs(i, 'h') || p.nextIs(i, 'H') {
if len(p.datestr) > i+2 {
newDateStr := p.datestr[0:i] + p.datestr[i+2:]
putBackParser(p)
return parseTime(newDateStr, loc, opts...)
}
}
return p, p.unknownErr(datestr)
case 'n', 'N':
if p.nextIs(i, 'd') || p.nextIs(i, 'D') {
if len(p.datestr) > i+2 {
newDateStr := p.datestr[0:i] + p.datestr[i+2:]
putBackParser(p)
return parseTime(newDateStr, loc, opts...)
}
}
return p, p.unknownErr(datestr)
case 's', 'S':
if p.nextIs(i, 't') || p.nextIs(i, 'T') {
if len(p.datestr) > i+2 {
newDateStr := p.datestr[0:i] + p.datestr[i+2:]
putBackParser(p)
return parseTime(newDateStr, loc, opts...)
}
}
return p, p.unknownErr(datestr)
case 'r', 'R':
if p.nextIs(i, 'd') || p.nextIs(i, 'D') {
if len(p.datestr) > i+2 {
newDateStr := p.datestr[0:i] + p.datestr[i+2:]
putBackParser(p)
return parseTime(newDateStr, loc, opts...)
}
}
return p, p.unknownErr(datestr)
default:
return p, p.unknownErr(datestr)
}
case dateAlphaFullMonthWs:
// 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, p.unknownErr(datestr)
}
p.yeari = i + 2
p.stateDate = dateAlphaFullMonthWsDayWs
i++
} else {
return p, p.unknownErr(datestr)
}
case r == ' ':
// x
// January 02 2006, 15:04:05
p.daylen = i - p.dayi
if !p.setDay() {
return p, p.unknownErr(datestr)
}
p.yeari = i + 1
p.stateDate = dateAlphaFullMonthWsDayWs
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, p.unknownErr(datestr)
}
p.stateDate = dateVariousDaySuffix
i--
default:
return p, p.unknownErr(datestr)
}
case dateAlphaFullMonthWsDayWs:
// 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, p.unknownErr(datestr)
}
p.stateTime = timeStart
i++
break iterRunes
case ' ':
p.yearlen = i - p.yeari
if !p.setYear() {
return p, p.unknownErr(datestr)
}
p.stateTime = timeStart
break iterRunes
default:
if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
}
case dateAlphaPeriodWsDigit:
// oct. 7, '70
switch {
case r == ' ':
// continue
case unicode.IsDigit(r):
p.stateDate = dateAlphaWsDigit
p.dayi = i
default:
return p, 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, 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, p.unknownErr(datestr)
}
p.stateDate = dateAlphaSlashDigitSlash
case unicode.IsDigit(r):
// continue
default:
return p, p.unknownErr(datestr)
}
case dateAlphaSlashDigitSlash:
switch {
case unicode.IsDigit(r):
// continue
case r == ' ':
p.stateTime = timeStart
break iterRunes
default:
return p, p.unknownErr(datestr)
}
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 '-', '\u2212':
if p.moi == 0 {
p.moi = i + 1
p.daylen = i - p.dayi
if !p.setDay() {
return p, p.unknownErr(datestr)
}
} else if p.yeari == 0 {
p.yeari = i + 1
p.molen = i - p.moi
if p.molen == 3 {
p.set(p.moi, "Jan")
} else {
return p, p.unknownErr(datestr)
}
} else {
p.stateTime = timeStart
break iterRunes
}
default:
if !unicode.IsDigit(r) && !unicode.IsLetter(r) {
return p, p.unknownErr(datestr)
}
}
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
var offset int
switch r {
case ' ':
for i+1 < len(p.datestr) && p.datestr[i+1] == ' ' {
i++
offset++
}
fallthrough
case '-', '\u2212':
if p.dayi == 0 {
p.dayi = i + 1
} else if p.moi == 0 {
p.daylen = i - p.dayi
if !p.setDay() {
return p, p.unknownErr(datestr)
}
p.moi = i + 1
} else if p.yeari == 0 {
p.molen = i - p.moi - offset
if p.molen == 3 {
p.set(p.moi, "Jan")
} else {
return p, p.unknownErr(datestr)
}
p.yeari = i + 1
} else {
p.yearlen = i - p.yeari - offset
if !p.setYear() {
return p, p.unknownErr(datestr)
}
p.stateTime = timeStart
break iterRunes
}
default:
if !unicode.IsDigit(r) && !unicode.IsLetter(r) {
return p, p.unknownErr(datestr)
}
}
default:
// Reaching an unhandled state unexpectedly should always fail parsing
return p, p.unknownErr(datestr)
}
}
if !p.coalesceDate(i) {
return p, p.unknownErr(datestr)
}
if p.stateTime == timeStart {
// increment first one, since the i++ occurs at end of loop
if i < len(p.datestr) {
i++
}
// ensure we skip any whitespace prefix
for ; i < len(p.datestr); i++ {
r := rune(p.datestr[i])
if r != ' ' {
break
}
}
iterTimeRunes:
for ; i < len(p.datestr); i++ {
r := rune(p.datestr[i])
// gou.Debugf("i=%d r=%s state=%d iterTimeRunes %s %s", i, string(r), p.stateTime, p.ds(), p.ts())
switch p.stateTime {
case timeStart:
// 22:43:22
// 22:43
// timeComma
// 08:20:13,787
// timeWs
// 05:24:37 PM
// 06:20:00 UTC
// 06:20:00 UTC-05
// 00:12:00 +0000 UTC
// 22:18:00 +0000 UTC m=+0.000000001
// 15:04:05 -0700
// 15:04:05 -07:00
// 15:04:05 2008
// timeOffset
// 03:21:51+00:00
// 19:55:00+0100
// 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)
if p.houri == 0 {
p.houri = i
}
switch r {
case '-', '+':
// 03:21:51+00:00
p.stateTime = timeOffset
if p.seci == 0 {
// 22:18+0530
p.minlen = i - p.mini
} else {
if p.seclen == 0 {
p.seclen = i - p.seci
}
if p.msi > 0 && p.mslen == 0 {
p.mslen = i - p.msi
}
}
p.offseti = i
case '.', ',':
// NOTE: go 1.20 can now parse a string that has a comma delimiter properly
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
}
// (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, p.unexpectedTail(i)
}
}
case ' ':
p.coalesceTime(i)
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
} 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
}
}
case timeOffset:
// 19:55:00+0100
// timeOffsetColon
// 15:04:05+07:00
// 15:04:05-07:00
if r == ':' {
p.stateTime = timeOffsetColon
}
case timeWs:
// timeWsAlpha
// 06:20:00 UTC
// 06:20:00 UTC-05
// 15:44:11 UTC+0100 2015
// 18:04:07 GMT+0100 (GMT Daylight Time)
// 17:57:51 MST 2009
// timeWsAMPMMaybe
// 05:24:37 PM
// timeWsOffset
// 15:04:05 -0700
// 00:12:00 +0000 UTC
// timeWsOffsetColon
// 15:04:05 -07:00
// 17:57:51 -0700 2009
// timeWsOffsetColonAlpha
// 00:12:00 +00:00 UTC
// timeWsYear
// 00:12:00 2008
// timeWsYearOffset
// 00:12:00 2008 -0700
// timeZ
// 15:04:05.99Z
switch r {
case 'a', 'p', 'A', 'P':
// Could be AM/PM or could be PST or similar
p.tzi = i
p.stateTime = timeWsAMPMMaybe
case '+', '-':
p.offseti = i
p.stateTime = timeWsOffset
default:
if unicode.IsLetter(r) {
// 06:20:00 UTC
// 06:20:00 UTC-05
// 15:44:11 UTC+0100 2015
// 17:57:51 MST 2009
p.tzi = i
p.stateTime = timeWsAlpha
} else if unicode.IsDigit(r) {
// 00:12:00 2008
p.stateTime = timeWsYear
p.yeari = i
}
}
case timeWsYear:
// timeWsYearOffset
// 00:12:00 2008 -0700
switch r {
case ' ':
p.yearlen = i - p.yeari
if !p.setYear() {
return p, p.unknownErr(datestr)
}
case '+', '-':
p.offseti = i
p.stateTime = timeWsYearOffset
default:
if !unicode.IsDigit(r) {
return p, p.unknownErr(datestr)
}
}
case timeWsAlpha:
// 06:20:00 UTC
// 06:20:00 UTC-05
// 06:20:00 (EST)
// timeWsAlphaWs
// 17:57:51 MST 2009
// 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
p.tzi = 0
p.tzlen = 0
} else {
p.tzlen = i - p.tzi
}
if p.tzlen == 4 {
p.set(p.tzi, " MST")
} else if p.tzlen == 3 {
p.set(p.tzi, "MST")
}
p.stateTime = timeWsAlphaZoneOffset
p.offseti = i
case ' ', ')':
// 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, p.unknownErr(datestr)
}
}
}
case timeWsAlphaWs:
// 17:57:51 MST 2009
case timeWsAlphaZoneOffset:
// 06:20:00 UTC-05
// 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
}
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, p.unknownErr(datestr)
}
}
}
case timeWsAMPMMaybe:
// timeWsAMPMMaybe
// timeWsAMPM
// 05:24:37 PM
// timeWsAlpha
// 00:12:00 PST
// 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, p.unexpectedTail(i)
}
// This isn't a time zone after all...
p.tzi = 0
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")
}
} 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, p.unexpectedTail(i)
}
case timeWsOffset:
// timeWsOffset
// 15:04:05 -0700
// timeWsOffsetWsOffset
// 17:57:51 -0700 -07
// timeWsOffsetWs
// 17:57:51 -0700 2009
// 00:12:00 +0000 UTC
// timeWsOffsetColon
// 15:04:05 -07:00
// timeWsOffsetColonAlpha
// 00:12:00 +00:00 UTC
switch r {
case ':':
p.stateTime = timeWsOffsetColon
case ' ':
p.set(p.offseti, "-0700")
p.yeari = i + 1
p.stateTime = timeWsOffsetWs
}
case timeWsOffsetWs:
// 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)
}
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, p.unknownErr(datestr)
}
}
case unicode.IsLetter(r):
// 15:04:05 -0700 MST
if p.tzi == 0 {
p.tzi = i
}
}
}
case timeOffsetColon, timeWsOffsetColon:
// timeOffsetColon
// 15:04:05-07:00
// timeOffsetColonAlpha
// 2015-02-18 00:12:00+00:00 UTC
// 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?
// 2015-02-18 00:12:00 +00:00 UTC
if p.stateTime == timeWsOffsetColon {
p.stateTime = timeWsOffsetColonAlpha
} else {
p.stateTime = timeOffsetColonAlpha
}
p.tzi = i
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
// timePeriod
// 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...)
switch r {
case ' ':
p.mslen = i - p.msi
p.coalesceTime(i)
p.stateTime = timeWs
case '+', '-':
p.mslen = i - p.msi
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:
p.mslen = i - p.msi
p.coalesceTime(i)
p.set(i, "PM")
p.parsedAMPM = true
// skip 'M'
i++
p.stateTime = timePeriodAMPM
default:
return p, p.unexpectedTail(i)
}
default:
if !unicode.IsDigit(r) {
return p, p.unexpectedTail(i)
}
}
case timePeriodAMPM:
switch r {
case ' ':
p.stateTime = timeWs
case '+', '-':
p.offseti = i
p.stateTime = timeOffset
default:
return p, p.unexpectedTail(i)
}
case timeZ:
// nothing expected can come after Z
return p, p.unexpectedTail(i)
}
}
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] != ' ' {
if p.simpleErrorMessages {
return p, ErrUnknownTZOffset
} else {
return p, fmt.Errorf("%w %q near %q (expected offset like -07:00)", ErrUnknownTZOffset, datestr, p.datestr[p.offseti:p.offseti+offsetLen])
}
}
}
p.set(p.offseti, "-07:00")
default:
if p.simpleErrorMessages {
return p, ErrUnknownTZOffset
} else {
return p, fmt.Errorf("%w %q near %q (expected offset like -07:00)", ErrUnknownTZOffset, datestr, 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:
if p.simpleErrorMessages {
return p, ErrUnknownTimeZone
} else {
return p, fmt.Errorf("%w %q near %q (must be 3 or 4 characters)", ErrUnknownTimeZone, datestr, 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:
if p.simpleErrorMessages {
return p, ErrUnknownTimeZone
} else {
return p, fmt.Errorf("%w %q near %q (must be 3 or 4 characters)", ErrUnknownTimeZone, datestr, p.datestr[p.tzi:])
}
}
case timeWsAlphaRParen:
// continue
case timeWsAlphaWs:
p.yearlen = i - p.yeari
if !p.setYear() {
return p, p.unknownErr(datestr)
}
case timeWsYear:
p.yearlen = i - p.yeari
if !p.setYear() {
return p, p.unknownErr(datestr)
}
case timeWsAlphaZoneOffsetWsExtra:
p.trimExtra(false)
case timeWsAlphaZoneOffset:
// 06:20:00 UTC-05
switch i - p.offseti {
case 2, 3, 4:
p.set(p.offseti, "-07")
case 5:
p.set(p.offseti, "-0700")
case 6:
p.set(p.offseti, "-07:00")
default:
if p.simpleErrorMessages {
return p, ErrUnknownTZOffset
} else {
return p, fmt.Errorf("%w %q near %q (must be 2 or 4 digits optional colon)", ErrUnknownTZOffset, datestr, p.datestr[p.offseti:i])
}
}
case timePeriod:
p.mslen = i - p.msi
if p.mslen >= 10 {
if p.simpleErrorMessages {
return p, ErrFracSecTooLong
} else {
return p, fmt.Errorf("%w in %q near %q", ErrFracSecTooLong, datestr, 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:
if p.simpleErrorMessages {
return p, ErrUnknownTZOffset
} else {
return p, fmt.Errorf("%w %q near %q (must be 2 or 4 digits optional colon)", ErrUnknownTZOffset, datestr, p.datestr[p.offseti:])
}
}
case timeWsOffsetWs:
// 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:
if p.simpleErrorMessages {
return p, ErrUnknownTimeZone
} else {
return p, fmt.Errorf("%w %q near %q (must be 3 or 4 characters)", ErrUnknownTimeZone, datestr, 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:
if p.simpleErrorMessages {
return p, ErrUnknownTZOffset
} else {
return p, fmt.Errorf("%w %q near %q (expected offset like -07:00)", ErrUnknownTZOffset, datestr, p.datestr[p.offseti:])
}
}
}
p.coalesceTime(i)
}
switch p.stateDate {
case dateDigit:
// 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)
}
} 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)
}
} 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)
}
} 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"))
return p, nil
} else if len(p.datestr) == len("2014") {
p.setEntireFormat([]byte("2006"))
return p, nil
} else if len(p.datestr) < 4 {
return p, p.unknownErr(datestr)
}
if !t.IsZero() {
if loc == nil {
p.t = &t
return p, nil
}
t = t.In(loc)
p.t = &t
return p, nil
}
case dateDigitSt:
// 171113 14:14:20
return p, nil
case dateYearDash:
// 2006-01
return p, nil
case dateYearDashDash:
// 2006-01-02
// 2006-1-02
// 2006-1-2
// 2006-01-2
return p, nil
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
case dateYearDashAlpha:
// 2013-Feb-03
// 2013-Feb-3
// 2013-February-3
return p, nil
case dateYearDashDashWs:
// 2013-04-01
return p, nil
case dateYearDashDashT:
return p, nil
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, 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 // not retryable
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, p.unknownErr(datestr)
}
} else {
return p, p.unknownErr(datestr)
}
}
return p, nil
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, p.unknownErr(datestr)
}
return p, nil
}
case dateDigitDotDot:
// 03.31.1981
// 3.31.2014
// 3.2.1981
// 3.2.81
// 08.21.71
// 2018.09.30
return p, nil
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
case dateDigitWsMoYear:
// 2 Jan 2018
// 2 Jan 18
// 2 Jan 2018 23:59
// 02 Jan 2018 23:59
// 12 Feb 2006, 19:17
return p, nil
case dateAlphaFullMonthWs:
if p.stateTime == timeIgnore && p.yearlen == 0 {
p.yearlen = i - p.yeari
if !p.setYear() {
return p, p.unknownErr(datestr)
}
}
return p, nil
case dateAlphaFullMonthWsDayWs:
return p, nil
case dateAlphaWsDigitMoreWs:
// oct 1, 1970
p.yearlen = i - p.yeari
if !p.setYear() {
return p, p.unknownErr(datestr)
}
return p, nil
case dateAlphaWsDigitMoreWsYear:
// May 8, 2009 5:57:51 PM
// Jun 7, 2005, 05:57:51
return p, nil
case dateAlphaWsAlpha:
return p, nil
case dateAlphaWsDigit:
return p, nil
case dateAlphaWsDigitYearMaybe:
return p, nil
case dateDigitSlash:
// 3/1/2014
// 10/13/2014
// 01/02/2006
return p, nil
case dateDigitSlashAlphaSlash:
// 03/Jun/2014
return p, nil
case dateDigitYearSlash:
// 2014/10/13
return p, nil
case dateDigitColon:
// 3:1:2014
// 10:13:2014
// 01:02:2006
// 2014:10:13
return p, nil
case dateDigitChineseYear:
// dateDigitChineseYear
// 2014年04月08日
// 2014年4月12日
return p, nil
case dateDigitChineseYearWs:
// 2014年04月08日 00:00:00 ...
return p, nil
case dateAlphaSlashDigitSlash:
// Oct/ 7/1970
// February/07/1970
return p, nil
case dateWeekdayComma:
// Monday, 02 Jan 2006 15:04:05 -0700
// Monday, 02 Jan 2006 15:04:05 +0100
// Monday, 02-Jan-06 15:04:05 MST
return p, nil
case dateWeekdayAbbrevComma:
// Mon, 02-Jan-06 15:04:05 MST
// Mon, 02 Jan 2006 15:04:05 MST
return p, nil
case dateYearWsMonthWs:
// 2013 May 02 11:37:55
// 2013 December 02 11:37:55
return p, nil
}
return p, p.unknownErr(datestr)
}
type parser struct {
loc *time.Location
preferMonthFirst bool
retryAmbiguousDateWithSwap bool
ambiguousMD bool
ambiguousRetryable bool
allowPartialStringMatch bool
simpleErrorMessages bool
stateDate dateState
stateTime timeState
format []byte
formatSetLen int
datestr string
fullMonth string
parsedAMPM bool
skip int
link 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
}
// something like: "Wednesday, 8 February 2023 19:00:46.999999999 +11:00 (AEDT) m=+0.000000001"
const longestPossibleDateStr = 78
// the format byte slice is always a little larger, in case we need to expand it to contain a full month
const formatExtraBufferBytes = 16
const formatBufferCapacity = longestPossibleDateStr + formatExtraBufferBytes
var parserPool = sync.Pool{
New: func() interface{} {
// allocate a max-sized fixed-capacity format byte slice
// that will be re-used with this parser struct
return &parser{
format: make([]byte, 0, formatBufferCapacity),
}
},
}
var emptyString = ""
// Use to put a parser back into the pool in the right way
func putBackParser(p *parser) {
if p == nil {
return
}
// we'll be reusing the backing memory for the format byte slice, put it back
// to maximum capacity
if cap(p.format) == formatBufferCapacity {
p.format = p.format[:formatBufferCapacity]
} else {
// the parsing improperly process replaced this, get back a new one with the right cap
p.format = make([]byte, 0, formatBufferCapacity)
}
// clear out pointers so we don't leak memory we don't need any longer
p.loc = nil
p.datestr = emptyString
p.fullMonth = emptyString
p.t = nil
parserPool.Put(p)
}
// 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
}
}
// 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
}
}
// SimpleErrorMessages is an option that will cause returned error messages to contain less detail,
// but it will avoid allocating any memory for the custom error message. If you expect to attempt
// to parse a lot of text that is not valid, this could help reduce GC pressure.
func SimpleErrorMessages(simpleErrorMessages bool) ParserOption {
return func(p *parser) error {
p.simpleErrorMessages = simpleErrorMessages
return nil
}
}
// Creates a new parser. The caller must call putBackParser on the returned parser when done with it.
func newParser(dateStr string, loc *time.Location, opts ...ParserOption) (*parser, error) {
dateStrLen := len(dateStr)
if dateStrLen > longestPossibleDateStr {
var nilParser *parser
return nil, nilParser.unknownErr(dateStr)
}
// Make sure to re-use the format byte slice from the pooled parser struct
p := parserPool.Get().(*parser)
// This re-slicing is guaranteed to work because of the length check above
startingFormat := p.format[:dateStrLen]
copy(startingFormat, dateStr)
*p = parser{
stateDate: dateStart,
stateTime: timeIgnore,
datestr: dateStr,
loc: loc,
preferMonthFirst: true,
retryAmbiguousDateWithSwap: false,
format: startingFormat,
// this tracks how much of the format string has been set, to make sure all of it is set
formatSetLen: 0,
}
// allow the options to mutate the parser fields from their defaults
for _, option := range opts {
if err := option(p); err != nil {
return nil, fmt.Errorf("option error: %w", err)
}
}
return p, nil
}
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) {
// Copy so that we don't lose this pooled format byte slice
oldLen := len(p.format)
newLen := len(format)
if oldLen != newLen {
// guaranteed to work because of the allocated capacity for format buffers
p.format = p.format[:newLen]
}
copy(p.format, format)
p.formatSetLen = len(format)
}
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
}
}
func (p *parser) setMonth() bool {
if p.molen == 2 {
p.set(p.moi, "01")
return true
} else if p.molen == 1 {
p.set(p.moi, "1")
return true
} else {
return false
}
}
func (p *parser) setDay() bool {
if p.daylen == 2 {
p.set(p.dayi, "02")
return true
} else if p.daylen == 1 {
p.set(p.dayi, "2")
return true
} else {
return false
}
}
func (p *parser) setYear() bool {
if p.yearlen == 2 {
p.set(p.yeari, "06")
return true
} 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
}
}
}
return end
}
func (p *parser) coalesceDate(end int) bool {
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
}
}
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.
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
}
}
return true
}
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
}
}
}
func (p *parser) setFullMonth(month string) {
oldLen := len(p.format)
const fullMonth = "January"
// Do an overlapping copy so we don't lose the pooled format buffer
part1Len := p.moi
part3 := p.format[p.moi+len(month):]
newLen := part1Len + len(fullMonth) + len(part3)
if newLen > oldLen {
// We can re-slice this, because the capacity is guaranteed to be a little longer than any possible datestr
p.format = p.format[:newLen]
}
// first part will not change, we need to shift the third part
copy(p.format[part1Len+len(fullMonth):], part3)
copy(p.format[part1Len:], fullMonth)
// shorten the format slice now if needed
if newLen < oldLen {
p.format = p.format[:newLen]
}
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
}
}
func (p *parser) trimExtra(onlyTrimFormat bool) {
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]
}
}
}
func (p *parser) parse(originalLoc *time.Location, originalOpts ...ParserOption) (t time.Time, err error) {
if p == nil {
var nilParser *parser
return time.Time{}, nilParser.unknownErr("")
}
if p.t != nil {
return *p.t, nil
}
if len(p.fullMonth) > 0 {
p.setFullMonth(p.fullMonth)
}
if p.retryAmbiguousDateWithSwap && p.ambiguousMD && p.ambiguousRetryable {
// 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() {
// if actual time parsing errors out with the following error, swap before we
// get out of this function to reduce scope it needs to be applied on
if err != nil && strings.Contains(err.Error(), "month out of range") {
// simple optimized case where mm and dd can be swapped directly
if p.molen == 2 && p.daylen == 2 {
moi := p.moi
p.moi = p.dayi
p.dayi = moi
if !p.setDay() || !p.setMonth() {
err = p.unknownErr(p.datestr)
} else {
if p.loc == nil {
t, err = time.Parse(bytesToString(p.format), p.datestr)
} else {
t, err = time.ParseInLocation(bytesToString(p.format), p.datestr, p.loc)
}
}
} else {
// create the option to reverse the preference
preferMonthFirst := PreferMonthFirst(!p.preferMonthFirst)
// turn off the retry to avoid endless recursion
retryAmbiguousDateWithSwap := RetryAmbiguousDateWithSwap(false)
modifiedOpts := append(originalOpts, preferMonthFirst, retryAmbiguousDateWithSwap)
var newParser *parser
newParser, err = parseTime(p.datestr, originalLoc, modifiedOpts...)
defer putBackParser(newParser)
if err == nil {
t, err = newParser.parse(originalLoc, modifiedOpts...)
// The caller might use the format and datestr, so copy that back to the original parser
p.setEntireFormat(newParser.format)
p.datestr = newParser.datestr
}
}
}
}()
}
// 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(bytesToString(p.format), p.formatSetLen, len(p.format), false, false, true)
if validFormatTo < len(p.format) {
return time.Time{}, p.unexpectedTail(p.formatSetLen)
}
}
if p.skip > 0 && len(p.format) > p.skip {
// copy and then re-slice to shorten to avoid losing the header of the pooled format string
copy(p.format, p.format[p.skip:])
p.format = p.format[:len(p.format)-p.skip]
p.formatSetLen -= p.skip
if p.formatSetLen < 0 {
p.formatSetLen = 0
}
p.datestr = p.datestr[p.skip:]
}
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)
return time.Parse(bytesToString(p.format), p.datestr)
} else {
//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)
return time.ParseInLocation(bytesToString(p.format), p.datestr, p.loc)
}
}
func isDay(alpha string) bool {
_, ok := knownDays[alpha]
return ok
}
func isMonthFull(alpha string) bool {
_, ok := knownMonths[alpha]
return ok
}
Reference in New Issue View Git Blame Copy Permalink