Support RabbitMQ log format (dd-mon-yyyy::hh:mm:ss)

Adapt https://github.com/araddon/dateparse/pull/122 by https://github.com/bizy01 to add support for RMQ log format. Refactor to avoid redundant code. Add format validations.

As a side note, will also support the format dd-mm-yyyy:hh:mm:ss.

parseany.go

@@ -598,44 +598,76 @@ iterRunes:
 			//   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 ' ':
+			case ' ', ':':
-				// we need to find if this was 4 digits, aka year
+				doubleColonTimeConnector := false
-				// or 2 digits which makes it ambiguous year/day
+				if r == ':' {
-				length := i - (p.moi + p.molen + 1)
+					p.link++
-				if length == 4 {
+					if p.link == 2 {
-					p.yearlen = 4
+						if i+1 < len(p.datestr) {
-					p.set(p.yeari, "2006")
+							// only legitimate content to follow "::" is the start of the time
-					// We now also know that part1 was the day
+							nextChar, _ := utf8.DecodeRuneInString(p.datestr[i+1:])
-					p.dayi = 0
+							if unicode.IsDigit(nextChar) {
-					p.daylen = p.part1Len
+								doubleColonTimeConnector = true
-					if !p.setDay() {
+							}
-						return p, unknownErr(datestr)
+						}
+						if !doubleColonTimeConnector {
+							return p, unknownErr(datestr)
+						}
 					}
-				} else if length == 2 {
+				} else if p.link > 0 {
-					// We have no idea if this is
+					return p, unknownErr(datestr)
-					// yy-mon-dd   OR  dd-mon-yy
+				}
-					// (or for dateDigitDashDigitDash, yy-mm-dd  OR  dd-mm-yy)
+				if r == ' ' || doubleColonTimeConnector {
-					//
+					// we need to find if this was 4 digits, aka year
-					// We are going to ASSUME (bad, bad) that it is dd-mon-yy (dd-mm-yy),
+					// or 2 digits which makes it ambiguous year/day
-					// which is a horrible assumption, but seems to be the convention for
+					var sepLen int
-					// dates that are formatted in this way.
+					if doubleColonTimeConnector {
-					p.ambiguousMD = true
+						sepLen = 2
-					p.yearlen = 2
+					} else {
-					p.set(p.yeari, "06")
+						sepLen = 1
-					// We now also know that part1 was the day
+					}
-					p.dayi = 0
+					length := i - (p.moi + p.molen + sepLen)
-					p.daylen = p.part1Len
+					if length == 4 {
-					if !p.setDay() {
+						p.yearlen = 4
-						return p, unknownErr(datestr)
+						p.set(p.yeari, "2006")
-					}
+						// We now also know that part1 was the day
-				} else {
+						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)
+					}
+					p.stateTime = timeStart
+					break iterRunes
+				}
+			default:
+				if !unicode.IsDigit(r) && !unicode.IsLetter(r) && p.link > 0 {
 					return p, unknownErr(datestr)
 				}
-				p.stateTime = timeStart
-				break iterRunes
 			}
 
 		case dateDigitYearSlash:
@@ -2397,6 +2429,7 @@ type parser struct {
 	fullMonth                  string
 	parsedAMPM                 bool
 	skip                       int
+	link                       int
 	extra                      int
 	part1Len                   int
 	yeari                      int

parseany_test.go

@@ -446,6 +446,9 @@ var testInputs = []dateTest{
 	// Git log default date format - https://github.com/araddon/dateparse/pull/92
 	{in: "Thu Apr 7 15:13:13 2005 -0700", out: "2005-04-07 22:13:13 +0000 UTC"},
 	{in: "Tue Dec 12 23:07:11 2023 -0700", out: "2023-12-13 06:07:11 +0000 UTC"},
+	// RabbitMQ log format - https://github.com/araddon/dateparse/pull/122
+	{in: "8-Mar-2018::14:09:27", out: "2018-03-08 14:09:27 +0000 UTC"},
+	{in: "08-03-2018::02:09:29 PM", out: "2018-03-08 14:09:29 +0000 UTC"},
 	//   yyyy-mm-dd hh:mm:ss,000
 	{in: "2014-05-11 08:20:13,787", out: "2014-05-11 08:20:13.787 +0000 UTC"},
 	//   yyyy-mm-dd hh:mm:ss +0000
@@ -827,6 +830,10 @@ var testParseErrors = []dateTest{
 	{in: "2014-02-13 00:00:00 utc", err: true}, // lowercase timezones are not valid
 	{in: "2014-02-13t00:00:00.0z", err: true},  // lowercase 't' separator is not supported
 	{in: "2014-02-13T00:00:00.0z", err: true},  // lowercase 'z' zulu timezone indicator not a valid format
+	// Invalid variants of RabbitMQ log format
+	{in: "8-Mar-2018:14:09:27", err: true},
+	{in: "8-Mar-2018: 14:09:27", err: true},
+	{in: "8-Mar-2018:::14:09:27", err: true},
 }
 
 func TestParseErrors(t *testing.T) {
@@ -1093,5 +1100,5 @@ func TestRetryAmbiguousDateWithSwap(t *testing.T) {
 
 // Convenience function for debugging a particular broken test case
 func TestDebug(t *testing.T) {
-	MustParse("Tue Dec 12 23:07:11 2023 -0700")
+	MustParse("8-Mar-2018::14:09:27")
 }