diff --git a/README.md b/README.md index ec09e22..c09851f 100644 --- a/README.md +++ b/README.md @@ -69,6 +69,7 @@ var examples = []string{ "Mon, 02 Jan 2006 15:04:05 MST", "Tue, 11 Jul 2017 16:28:13 +0200 (CEST)", "Mon, 02 Jan 2006 15:04:05 -0700", + "Thu, 4 Jan 2018 17:53:36 +0000", "Mon Aug 10 15:44:11 UTC+0100 2015", "Fri Jul 03 2015 18:04:07 GMT+0100 (GMT Daylight Time)", "12 Feb 2006, 19:17", @@ -80,6 +81,7 @@ var examples = []string{ "8/1/71", "4/8/2014 22:05", "04/08/2014 22:05", + "4/8/14 22:05", "04/2/2014 03:00:51", "8/8/1965 12:00:00 AM", "8/8/1965 01:00:01 PM", @@ -98,6 +100,8 @@ var examples = []string{ "2014/4/02 03:00:51", "2012/03/19 10:11:59", "2012/03/19 10:11:59.3186369", + // Chinese + "2014年04月08日", // yyyy-mm-ddThh "2006-01-02T15:04:05+0000", "2009-08-12T22:15:09-07:00", @@ -107,6 +111,7 @@ var examples = []string{ "2014-04-26 17:24:37.3186369", "2012-08-03 18:31:59.257000000", "2014-04-26 17:24:37.123", + "2013-04-01 22:43", "2013-04-01 22:43:22", "2014-12-16 06:20:00 UTC", "2014-12-16 06:20:00 GMT", @@ -176,6 +181,7 @@ func main() { | Mon, 02 Jan 2006 15:04:05 MST | 2006-01-02 15:04:05 +0000 MST | | Tue, 11 Jul 2017 16:28:13 +0200 (CEST) | 2017-07-11 16:28:13 +0200 +0200 | | Mon, 02 Jan 2006 15:04:05 -0700 | 2006-01-02 15:04:05 -0700 -0700 | +| Thu, 4 Jan 2018 17:53:36 +0000 | 2018-01-04 17:53:36 +0000 UTC | | Mon Aug 10 15:44:11 UTC+0100 2015 | 2015-08-10 15:44:11 +0000 UTC | | Fri Jul 03 2015 18:04:07 GMT+0100 (GMT Daylight Time) | 2015-07-03 18:04:07 +0100 GMT | | 12 Feb 2006, 19:17 | 2006-02-12 19:17:00 +0000 UTC | @@ -186,6 +192,7 @@ func main() { | 8/1/71 | 1971-08-01 00:00:00 +0000 UTC | | 4/8/2014 22:05 | 2014-04-08 22:05:00 +0000 UTC | | 04/08/2014 22:05 | 2014-04-08 22:05:00 +0000 UTC | +| 4/8/14 22:05 | 2014-04-08 22:05:00 +0000 UTC | | 04/2/2014 03:00:51 | 2014-04-02 03:00:51 +0000 UTC | | 8/8/1965 12:00:00 AM | 1965-08-08 00:00:00 +0000 UTC | | 8/8/1965 01:00:01 PM | 1965-08-08 13:00:01 +0000 UTC | @@ -203,6 +210,7 @@ func main() { | 2014/4/02 03:00:51 | 2014-04-02 03:00:51 +0000 UTC | | 2012/03/19 10:11:59 | 2012-03-19 10:11:59 +0000 UTC | | 2012/03/19 10:11:59.3186369 | 2012-03-19 10:11:59.3186369 +0000 UTC | +| 2014年04月08日 | 2014-04-08 00:00:00 +0000 UTC | | 2006-01-02T15:04:05+0000 | 2006-01-02 15:04:05 +0000 UTC | | 2009-08-12T22:15:09-07:00 | 2009-08-12 22:15:09 -0700 -0700 | | 2009-08-12T22:15:09 | 2009-08-12 22:15:09 +0000 UTC | @@ -210,6 +218,7 @@ func main() { | 2014-04-26 17:24:37.3186369 | 2014-04-26 17:24:37.3186369 +0000 UTC | | 2012-08-03 18:31:59.257000000 | 2012-08-03 18:31:59.257 +0000 UTC | | 2014-04-26 17:24:37.123 | 2014-04-26 17:24:37.123 +0000 UTC | +| 2013-04-01 22:43 | 2013-04-01 22:43:00 +0000 UTC | | 2013-04-01 22:43:22 | 2013-04-01 22:43:22 +0000 UTC | | 2014-12-16 06:20:00 UTC | 2014-12-16 06:20:00 +0000 UTC | | 2014-12-16 06:20:00 GMT | 2014-12-16 06:20:00 +0000 UTC | diff --git a/dateparse/main.go b/dateparse/main.go index 7e17d9f..cb543a8 100644 --- a/dateparse/main.go +++ b/dateparse/main.go @@ -56,13 +56,13 @@ func main() { for name, parser := range parsers { time.Local = nil - table.AddRow(name, "time.Local = nil", parser(datestr, nil), parser(datestr, nil).In(time.UTC)) + table.AddRow(name, "time.Local = nil", parser(datestr, nil, false), parser(datestr, nil, true)) if timezone != "" { time.Local = loc - table.AddRow(name, "time.Local = timezone arg", parser(datestr, loc), parser(datestr, loc).In(time.UTC)) + table.AddRow(name, "time.Local = timezone arg", parser(datestr, loc, false), parser(datestr, loc, true)) } time.Local = time.UTC - table.AddRow(name, "time.Local = time.UTC", parser(datestr, time.UTC), parser(datestr, time.UTC).In(time.UTC)) + table.AddRow(name, "time.Local = time.UTC", parser(datestr, time.UTC, false), parser(datestr, time.UTC, true)) } fmt.Println(table.Render()) @@ -72,20 +72,38 @@ func stuff() (string, string) { return "more", "stuff" } -type parser func(datestr string, loc *time.Location) time.Time +type parser func(datestr string, loc *time.Location, utc bool) string -func parseLocal(datestr string, loc *time.Location) time.Time { +func parseLocal(datestr string, loc *time.Location, utc bool) string { time.Local = loc - t, _ := dateparse.ParseLocal(datestr) - return t + t, err := dateparse.ParseLocal(datestr) + if err != nil { + return err.Error() + } + if utc { + return t.In(time.UTC).String() + } + return t.String() } -func parseIn(datestr string, loc *time.Location) time.Time { - t, _ := dateparse.ParseIn(datestr, loc) - return t +func parseIn(datestr string, loc *time.Location, utc bool) string { + t, err := dateparse.ParseIn(datestr, loc) + if err != nil { + return err.Error() + } + if utc { + return t.In(time.UTC).String() + } + return t.String() } -func parseAny(datestr string, loc *time.Location) time.Time { - t, _ := dateparse.ParseAny(datestr) - return t +func parseAny(datestr string, loc *time.Location, utc bool) string { + t, err := dateparse.ParseAny(datestr) + if err != nil { + return err.Error() + } + if utc { + return t.In(time.UTC).String() + } + return t.String() } diff --git a/example/main.go b/example/main.go index 8577394..72f99f0 100644 --- a/example/main.go +++ b/example/main.go @@ -18,6 +18,7 @@ var examples = []string{ "Mon, 02 Jan 2006 15:04:05 MST", "Tue, 11 Jul 2017 16:28:13 +0200 (CEST)", "Mon, 02 Jan 2006 15:04:05 -0700", + "Thu, 4 Jan 2018 17:53:36 +0000", "Mon Aug 10 15:44:11 UTC+0100 2015", "Fri Jul 03 2015 18:04:07 GMT+0100 (GMT Daylight Time)", "12 Feb 2006, 19:17", @@ -29,6 +30,7 @@ var examples = []string{ "8/1/71", "4/8/2014 22:05", "04/08/2014 22:05", + "4/8/14 22:05", "04/2/2014 03:00:51", "8/8/1965 12:00:00 AM", "8/8/1965 01:00:01 PM", @@ -47,6 +49,8 @@ var examples = []string{ "2014/4/02 03:00:51", "2012/03/19 10:11:59", "2012/03/19 10:11:59.3186369", + // Chinese + "2014年04月08日", // yyyy-mm-ddThh "2006-01-02T15:04:05+0000", "2009-08-12T22:15:09-07:00", @@ -56,6 +60,7 @@ var examples = []string{ "2014-04-26 17:24:37.3186369", "2012-08-03 18:31:59.257000000", "2014-04-26 17:24:37.123", + "2013-04-01 22:43", "2013-04-01 22:43:22", "2014-12-16 06:20:00 UTC", "2014-12-16 06:20:00 GMT", @@ -125,6 +130,7 @@ func main() { | Mon, 02 Jan 2006 15:04:05 MST | 2006-01-02 15:04:05 +0000 MST | | Tue, 11 Jul 2017 16:28:13 +0200 (CEST) | 2017-07-11 16:28:13 +0200 +0200 | | Mon, 02 Jan 2006 15:04:05 -0700 | 2006-01-02 15:04:05 -0700 -0700 | +| Thu, 4 Jan 2018 17:53:36 +0000 | 2018-01-04 17:53:36 +0000 UTC | | Mon Aug 10 15:44:11 UTC+0100 2015 | 2015-08-10 15:44:11 +0000 UTC | | Fri Jul 03 2015 18:04:07 GMT+0100 (GMT Daylight Time) | 2015-07-03 18:04:07 +0100 GMT | | 12 Feb 2006, 19:17 | 2006-02-12 19:17:00 +0000 UTC | @@ -135,6 +141,7 @@ func main() { | 8/1/71 | 1971-08-01 00:00:00 +0000 UTC | | 4/8/2014 22:05 | 2014-04-08 22:05:00 +0000 UTC | | 04/08/2014 22:05 | 2014-04-08 22:05:00 +0000 UTC | +| 4/8/14 22:05 | 2014-04-08 22:05:00 +0000 UTC | | 04/2/2014 03:00:51 | 2014-04-02 03:00:51 +0000 UTC | | 8/8/1965 12:00:00 AM | 1965-08-08 00:00:00 +0000 UTC | | 8/8/1965 01:00:01 PM | 1965-08-08 13:00:01 +0000 UTC | @@ -152,6 +159,7 @@ func main() { | 2014/4/02 03:00:51 | 2014-04-02 03:00:51 +0000 UTC | | 2012/03/19 10:11:59 | 2012-03-19 10:11:59 +0000 UTC | | 2012/03/19 10:11:59.3186369 | 2012-03-19 10:11:59.3186369 +0000 UTC | +| 2014年04月08日 | 2014-04-08 00:00:00 +0000 UTC | | 2006-01-02T15:04:05+0000 | 2006-01-02 15:04:05 +0000 UTC | | 2009-08-12T22:15:09-07:00 | 2009-08-12 22:15:09 -0700 -0700 | | 2009-08-12T22:15:09 | 2009-08-12 22:15:09 +0000 UTC | @@ -159,6 +167,7 @@ func main() { | 2014-04-26 17:24:37.3186369 | 2014-04-26 17:24:37.3186369 +0000 UTC | | 2012-08-03 18:31:59.257000000 | 2012-08-03 18:31:59.257 +0000 UTC | | 2014-04-26 17:24:37.123 | 2014-04-26 17:24:37.123 +0000 UTC | +| 2013-04-01 22:43 | 2013-04-01 22:43:00 +0000 UTC | | 2013-04-01 22:43:22 | 2013-04-01 22:43:22 +0000 UTC | | 2014-12-16 06:20:00 UTC | 2014-12-16 06:20:00 +0000 UTC | | 2014-12-16 06:20:00 GMT | 2014-12-16 06:20:00 +0000 UTC | diff --git a/parseany.go b/parseany.go index 6badec9..6a30dc7 100644 --- a/parseany.go +++ b/parseany.go @@ -8,15 +8,9 @@ import ( "strconv" "time" "unicode" - - u "github.com/araddon/gou" + "unicode/utf8" ) -func init() { - u.SetupLogging("debug") - u.SetColorOutput() -} - type dateState int const ( @@ -50,6 +44,8 @@ const ( stateDigitSlashWSColonAMPM stateDigitSlashWSColonColon stateDigitSlashWSColonColonAMPM + stateDigitChineseYear + stateDigitChineseYearWs stateDigitAlpha stateAlpha stateAlphaWS @@ -134,11 +130,11 @@ func parseTime(datestr string, loc *time.Location) (time.Time, error) { // we figure it out and then attempt a parse iterRunes: for i := 0; i < len(datestr); i++ { - r := rune(datestr[i]) - // r, bytesConsumed := utf8.DecodeRuneInString(datestr[ri:]) - // if bytesConsumed > 1 { - // ri += (bytesConsumed - 1) - // } + //r := rune(datestr[i]) + r, bytesConsumed := utf8.DecodeRuneInString(datestr[i:]) + if bytesConsumed > 1 { + i += (bytesConsumed - 1) + } switch state { case stateStart: @@ -151,6 +147,11 @@ iterRunes: if unicode.IsDigit(r) { continue } else if unicode.IsLetter(r) { + if r == '年' { + // Chinese Year + state = stateDigitChineseYear + continue + } state = stateDigitAlpha continue } @@ -387,6 +388,7 @@ iterRunes: // 04/2/2014 03:00:37 // 3/1/2012 10:11:59 // 4/8/2014 22:05 + // 4/8/14 22:05 switch r { case ':': state = stateDigitSlashWSColon @@ -397,6 +399,7 @@ iterRunes: // 04/2/2014 03:00:37 // 3/1/2012 10:11:59 // 4/8/2014 22:05 + // 4/8/14 22:05 // 3/1/2012 10:11:59 AM switch r { case ':': @@ -410,11 +413,21 @@ iterRunes: // 04/2/2014 03:00:37 // 3/1/2012 10:11:59 // 4/8/2014 22:05 + // 4/8/14 22:05 // 3/1/2012 10:11:59 AM switch r { case 'A', 'P': state = stateDigitSlashWSColonColonAMPM } + case stateDigitChineseYear: + // stateDigitChineseYear + // 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 + if r == ' ' { + state = stateDigitChineseYearWs + break + } case stateDigitAlpha: // 12 Feb 2006, 19:17 // 12 Feb 2006, 19:17:22 @@ -488,6 +501,7 @@ iterRunes: // stateWeekdayAbbrevCommaOffset // Mon, 02 Jan 2006 15:04:05 -0700 // Thu, 13 Jul 2017 08:58:40 +0100 + // Thu, 4 Jan 2018 17:53:36 +0000 // stateWeekdayAbbrevCommaOffsetZone // Tue, 11 Jul 2017 16:28:13 +0200 (CEST) switch { @@ -504,6 +518,7 @@ iterRunes: // stateWeekdayAbbrevCommaOffset // Mon, 02 Jan 2006 15:04:05 -0700 // Thu, 13 Jul 2017 08:58:40 +0100 + // Thu, 4 Jan 2018 17:53:36 +0000 // stateWeekdayAbbrevCommaOffsetZone // Tue, 11 Jul 2017 16:28:13 +0200 (CEST) if r == '(' { @@ -669,7 +684,13 @@ iterRunes: } case stateDigitDashWs: // starts digit then dash 02- then whitespace 1 << 2 << 5 + 3 // 2013-04-01 22:43:22 - return parse("2006-01-02 15:04:05", datestr, loc) + // 2013-04-01 22:43 + switch len(datestr) { + case len("2013-04-01 22:43"): + return parse("2006-01-02 15:04", datestr, loc) + default: + return parse("2006-01-02 15:04:05", datestr, loc) + } case stateDigitDashWsWsOffset: // 2006-01-02 15:04:05 -0700 @@ -814,7 +835,7 @@ iterRunes: } } } else { - for _, layout := range []string{"01/02/2006 15:04", "01/2/2006 15:04", "1/02/2006 15:04", "1/2/2006 15:04"} { + for _, layout := range []string{"01/02/2006 15:04", "01/2/2006 15:04", "1/02/2006 15:04", "1/2/2006 15:04", "1/2/06 15:04", "01/02/06 15:04"} { if t, err := parse(layout, datestr, loc); err == nil { return t, nil } @@ -851,6 +872,7 @@ iterRunes: // 3/1/2012 10:11:59 // 03/1/2012 10:11:59 // 3/01/2012 10:11:59 + // 4/8/14 22:05 if part1Len == 4 { for _, layout := range []string{"2006/01/02 15:04:05", "2006/1/02 15:04:05", "2006/01/2 15:04:05", "2006/1/2 15:04:05"} { if t, err := parse(layout, datestr, loc); err == nil { @@ -886,7 +908,12 @@ iterRunes: } } } - + case stateDigitChineseYear: + // stateDigitChineseYear + // 2014年04月08日 + return parse("2006年01月02日", datestr, loc) + case stateDigitChineseYearWs: + return parse("2006年01月02日 15:04:05", datestr, loc) case stateWeekdayCommaOffset: // Monday, 02 Jan 2006 15:04:05 -0700 // Monday, 02 Jan 2006 15:04:05 +0100 @@ -899,6 +926,12 @@ iterRunes: // Mon, 02 Jan 2006 15:04:05 -0700 // Thu, 13 Jul 2017 08:58:40 +0100 // RFC1123Z = "Mon, 02 Jan 2006 15:04:05 -0700" // RFC1123 with numeric zone + // + // Thu, 4 Jan 2018 17:53:36 +0000 + if len(datestr) > 10 && datestr[6] == ' ' { + // this really appears to be an invalid RFC1123 with non zero filled day + return parse("Mon, 2 Jan 2006 15:04:05 -0700", datestr, loc) + } return parse("Mon, 02 Jan 2006 15:04:05 -0700", datestr, loc) case stateWeekdayAbbrevCommaOffsetZone: // Tue, 11 Jul 2017 16:28:13 +0200 (CEST) diff --git a/parseany_test.go b/parseany_test.go index 96b1f3c..1a59dba 100644 --- a/parseany_test.go +++ b/parseany_test.go @@ -14,7 +14,7 @@ import ( // !!!!! The time-zone of local machine effects the results! // https://play.golang.org/p/IDHRalIyXh // https://github.com/golang/go/issues/18012 -func TestParseInLocation(t *testing.T) { +func TestInLocation(t *testing.T) { denverLoc, err := time.LoadLocation("America/Denver") assert.Equal(t, nil, err) @@ -170,6 +170,9 @@ func TestParse(t *testing.T) { ts = MustParse("Mon, 02 Jan 2006 15:04:05 -0700") assert.Equal(t, "2006-01-02 22:04:05 +0000 UTC", fmt.Sprintf("%v", ts.In(time.UTC))) + ts = MustParse("Thu, 4 Jan 2018 17:53:36 +0000") + assert.Equal(t, "2018-01-04 17:53:36 +0000 UTC", fmt.Sprintf("%v", ts.In(time.UTC))) + // not sure if this is anything close to a standard, never seen it before ts = MustParse("12 Feb 2006, 19:17") assert.Equal(t, "2006-02-12 19:17:00 +0000 UTC", fmt.Sprintf("%v", ts.In(time.UTC))) @@ -180,6 +183,15 @@ func TestParse(t *testing.T) { ts = MustParse("2013-Feb-03") assert.Equal(t, "2013-02-03 00:00:00 +0000 UTC", fmt.Sprintf("%v", ts.In(time.UTC))) + //--------------------------------------------- + // Chinese 2014年04月18日 + + ts = MustParse("2014年04月08日") + assert.Equal(t, "2014-04-08 00:00:00 +0000 UTC", fmt.Sprintf("%v", ts.In(time.UTC))) + + ts = MustParse("2014年04月08日 19:17:22") + assert.Equal(t, "2014-04-08 19:17:22 +0000 UTC", fmt.Sprintf("%v", ts.In(time.UTC))) + //--------------------------------------------- // mm/dd/yyyy ? @@ -199,10 +211,18 @@ func TestParse(t *testing.T) { ts = MustParse("4/8/2014 22:05") assert.Equal(t, "2014-04-08 22:05:00 +0000 UTC", fmt.Sprintf("%v", ts.In(time.UTC))) - + ts = MustParse("4/18/2014 22:05") + assert.Equal(t, "2014-04-18 22:05:00 +0000 UTC", fmt.Sprintf("%v", ts.In(time.UTC))) ts = MustParse("04/08/2014 22:05") assert.Equal(t, "2014-04-08 22:05:00 +0000 UTC", fmt.Sprintf("%v", ts.In(time.UTC))) + ts = MustParse("4/8/14 22:05") + assert.Equal(t, "2014-04-08 22:05:00 +0000 UTC", fmt.Sprintf("%v", ts.In(time.UTC))) + ts = MustParse("4/18/14 22:05") + assert.Equal(t, "2014-04-18 22:05:00 +0000 UTC", fmt.Sprintf("%v", ts.In(time.UTC))) + ts = MustParse("10/18/14 22:05") + assert.Equal(t, "2014-10-18 22:05:00 +0000 UTC", fmt.Sprintf("%v", ts.In(time.UTC))) + ts = MustParse("04/2/2014 4:00:51") assert.Equal(t, "2014-04-02 04:00:51 +0000 UTC", fmt.Sprintf("%v", ts.In(time.UTC))) @@ -348,6 +368,9 @@ func TestParse(t *testing.T) { ts = MustParse("2012-08-03 18:31:59.257000000") assert.Equal(t, "2012-08-03 18:31:59.257 +0000 UTC", fmt.Sprintf("%v", ts.In(time.UTC))) + ts = MustParse("2013-04-01 22:43") + assert.Equal(t, "2013-04-01 22:43:00 +0000 UTC", fmt.Sprintf("%v", ts.In(time.UTC))) + ts = MustParse("2013-04-01 22:43:22") assert.Equal(t, "2013-04-01 22:43:22 +0000 UTC", fmt.Sprintf("%v", ts.In(time.UTC)))