From 18ec8c69f6c7f0a363d0cde8f541d4b3d71b5113 Mon Sep 17 00:00:00 2001
From: Klondike Dragon <klondikedragon@gmail.com>
Date: Fri, 15 Dec 2023 17:14:03 -0700
Subject: [PATCH] Expand Chinese date format support

Inspired by https://github.com/araddon/dateparse/pull/132 from https://github.com/xwjdsh -- made this more general to all time formats that could follow, and added format validation.

Also include the related README.md touchup from https://github.com/araddon/dateparse/pull/136
---
 README.md        |  2 +-
 parseany.go      | 29 ++++++++++++++++++++++++++---
 parseany_test.go | 10 +++++++++-
 3 files changed, 36 insertions(+), 5 deletions(-)

diff --git a/README.md b/README.md
index a38646d..0ca7b6a 100644
--- a/README.md
+++ b/README.md
@@ -285,7 +285,7 @@ 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           |
+| 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           |
diff --git a/parseany.go b/parseany.go
index 022afae..08b4450 100644
--- a/parseany.go
+++ b/parseany.go
@@ -454,6 +454,11 @@ iterRunes:
 			case '年':
 				// Chinese Year
 				p.stateDate = dateDigitChineseYear
+				p.yearlen = i - 2
+				p.moi = i + 1
+				if !p.setYear() {
+					return p, unknownErr(datestr)
+				}
 			case ',':
 				return p, unknownErr(datestr)
 			case 's', 'S', 'r', 'R', 't', 'T', 'n', 'N':
@@ -889,8 +894,26 @@ iterRunes:
 			//   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 == ' ' {
+			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)
+				}
 				p.stateDate = dateDigitChineseYearWs
+				p.stateTime = timeStart
 				break iterRunes
 			}
 		case dateDigitDot:
@@ -2305,11 +2328,11 @@ iterRunes:
 	case dateDigitChineseYear:
 		// dateDigitChineseYear
 		//   2014年04月08日
-		p.setEntireFormat([]byte("2006年01月02日"))
+		//   2014年4月12日
 		return p, nil
 
 	case dateDigitChineseYearWs:
-		p.setEntireFormat([]byte("2006年01月02日 15:04:05"))
+		//   2014年04月08日 00:00:00 ...
 		return p, nil
 
 	case dateAlphaSlashDigitSlash:
diff --git a/parseany_test.go b/parseany_test.go
index bf17542..c786cf4 100644
--- a/parseany_test.go
+++ b/parseany_test.go
@@ -216,9 +216,17 @@ var testInputs = []dateTest{
 	// 03 February 2013
 	{in: "03 February 2013", out: "2013-02-03 00:00:00 +0000 UTC"},
 	{in: "3 February 2013", out: "2013-02-03 00:00:00 +0000 UTC"},
-	// Chinese 2014年04月18日
+	// Chinese 2014年04月18日 - https://github.com/araddon/dateparse/pull/132
 	{in: "2014年04月08日", out: "2014-04-08 00:00:00 +0000 UTC"},
+	{in: "2014年4月8日", out: "2014-04-08 00:00:00 +0000 UTC"},
 	{in: "2014年04月08日 19:17:22", out: "2014-04-08 19:17:22 +0000 UTC"},
+	{in: "2014年04月08日 19:17:22 MDT", out: "2014-04-08 19:17:22 +0000 UTC", zname: "MDT"},
+	{in: "2014年04月08日 19:17:22 MDT-0700", out: "2014-04-09 02:17:22 +0000 UTC", zname: "MDT"},
+	{in: "2014年4月8日 19:17:22", out: "2014-04-08 19:17:22 +0000 UTC"},
+	{in: "2014年4月8日 19:17:22 MDT", out: "2014-04-08 19:17:22 +0000 UTC", zname: "MDT"},
+	{in: "2014年4月8日 19:17:22 MDT-0700", out: "2014-04-09 02:17:22 +0000 UTC", zname: "MDT"},
+	{in: "2014年4月8日 10:17pm", out: "2014-04-08 22:17:00 +0000 UTC"},
+	// TODO: support Chinese AM (上午) and PM (下午) indicators
 	//  mm/dd/yyyy
 	{in: "03/31/2014", out: "2014-03-31 00:00:00 +0000 UTC"},
 	{in: "3/31/2014", out: "2014-03-31 00:00:00 +0000 UTC"},