From d285c2b223ad3f9e5e1a66d1309b79bf4a3f3368 Mon Sep 17 00:00:00 2001
From: Balazs Nagy <julsevern@gmail.com>
Date: Thu, 19 Aug 2021 08:26:27 +0200
Subject: [PATCH] handle s3.amazonaws.com.cn URL for s3 getter

---
 get_s3.go      | 28 +++++++++++++++++++++++++++-
 get_s3_test.go | 16 ++++++++++++++++
 2 files changed, 43 insertions(+), 1 deletion(-)

diff --git a/get_s3.go b/get_s3.go
index ec864428e..c7328146d 100644
--- a/get_s3.go
+++ b/get_s3.go
@@ -220,7 +220,7 @@ func (g *S3Getter) parseUrl(u *url.URL) (region, bucket, path, version string, c
 	// This just check whether we are dealing with S3 or
 	// any other S3 compliant service. S3 has a predictable
 	// url as others do not
-	if strings.Contains(u.Host, "amazonaws.com") {
+	if strings.HasSuffix(u.Host, "amazonaws.com") {
 		// Amazon S3 supports both virtual-hosted–style and path-style URLs to access a bucket, although path-style is deprecated
 		// In both cases few older regions supports dash-style region indication (s3-Region) even if AWS discourages their use.
 		// The same bucket could be reached with:
@@ -266,6 +266,32 @@ func (g *S3Getter) parseUrl(u *url.URL) (region, bucket, path, version string, c
 		}
 		version = u.Query().Get("version")
 
+	} else if strings.HasSuffix(u.Host, "amazonaws.com.cn") {
+		// S3 buckets in China:
+		// s3.region.amazonaws.com.cn/bucket/path
+		// bucket.s3.region.amazonaws.com.cn/path
+		hostParts := strings.Split(u.Host, ".")
+		switch len(hostParts) {
+		case 5:
+			if hostParts[0] != "s3" {
+				err = fmt.Errorf("URL is not a valid S3 URL")
+				return
+			}
+			region = hostParts[1]
+			pathParts := strings.SplitN(u.Path, "/", 3)
+			bucket = pathParts[1]
+			path = pathParts[2]
+		case 6:
+			region = hostParts[2]
+			pathParts := strings.SplitN(u.Path, "/", 2)
+			bucket = hostParts[0]
+			path = pathParts[1]
+		default:
+			err = fmt.Errorf("URL is not a valid S3 URL")
+			return
+		}
+		version = u.Query().Get("version")
+
 	} else {
 		pathParts := strings.SplitN(u.Path, "/", 3)
 		if len(pathParts) != 3 {
diff --git a/get_s3_test.go b/get_s3_test.go
index 25ce48367..57ac5849a 100644
--- a/get_s3_test.go
+++ b/get_s3_test.go
@@ -189,6 +189,14 @@ func TestS3Getter_Url(t *testing.T) {
 			path:    "foo/bar.baz",
 			version: "1234",
 		},
+		{
+			name:    "AWSv1234-CN",
+			url:     "s3::https://s3.cn-north-1.amazonaws.com.cn/bucket/foo/bar.baz?version=1234",
+			region:  "cn-north-1",
+			bucket:  "bucket",
+			path:    "foo/bar.baz",
+			version: "1234",
+		},
 		{
 			name:    "AWSVhostDot",
 			url:     "s3::https://bucket.s3.eu-west-1.amazonaws.com/foo/bar.baz?version=1234",
@@ -197,6 +205,14 @@ func TestS3Getter_Url(t *testing.T) {
 			path:    "foo/bar.baz",
 			version: "1234",
 		},
+		{
+			name:    "AWSVhostDot-CN",
+			url:     "s3::https://bucket.s3.cn-north-1.amazonaws.com.cn/foo/bar.baz?version=1234",
+			region:  "cn-north-1",
+			bucket:  "bucket",
+			path:    "foo/bar.baz",
+			version: "1234",
+		},
 		{
 			name:    "AWSVhostDash",
 			url:     "s3::https://bucket.s3-eu-west-1.amazonaws.com/foo/bar.baz?version=1234",