Refactor submodule URL parsing (#7100)

Use combination of url.Parse and regex to parse refURL rather than by
hand with indexes & attempt to check if refURL is from same instance and
adjust output to match.

Also now return empty string instead of our original
guess at URL if we are unable to parse it.

Fixes #1526

modules/git/submodule.go

@@ -1,10 +1,19 @@
+// Copyright 2019 The Gitea Authors. All rights reserved.
 // Copyright 2015 The Gogs Authors. All rights reserved.
 // Use of this source code is governed by a MIT-style
 // license that can be found in the LICENSE file.
 
 package git
 
-import "strings"
+import (
+	"fmt"
+	"net"
+	"net/url"
+	"regexp"
+	"strings"
+)
+
+var scpSyntax = regexp.MustCompile(`^([a-zA-Z0-9_]+@)?([a-zA-Z0-9._-]+):(.*)$`)
 
 // SubModule submodule is a reference on git repository
 type SubModule struct {
@@ -34,46 +43,73 @@ func getRefURL(refURL, urlPrefix, parentPath string) string {
 		return ""
 	}
 
-	url := strings.TrimSuffix(refURL, ".git")
+	refURI := strings.TrimSuffix(refURL, ".git")
 
-	// git://xxx/user/repo
+	prefixURL, _ := url.Parse(urlPrefix)
-	if strings.HasPrefix(url, "git://") {
+	urlPrefixHostname, _, err := net.SplitHostPort(prefixURL.Host)
-		return "http://" + strings.TrimPrefix(url, "git://")
+	if err != nil {
-	}
+		urlPrefixHostname = prefixURL.Host
-
-	// http[s]://xxx/user/repo
-	if strings.HasPrefix(url, "http://") || strings.HasPrefix(url, "https://") {
-		return url
 	}
 
 	// Relative url prefix check (according to git submodule documentation)
-	if strings.HasPrefix(url, "./") || strings.HasPrefix(url, "../") {
+	if strings.HasPrefix(refURI, "./") || strings.HasPrefix(refURI, "../") {
 		// ...construct and return correct submodule url here...
 		idx := strings.Index(parentPath, "/src/")
 		if idx == -1 {
-			return url
+			return refURI
 		}
-		return strings.TrimSuffix(urlPrefix, "/") + parentPath[:idx] + "/" + url
+		return strings.TrimSuffix(urlPrefix, "/") + parentPath[:idx] + "/" + refURI
 	}
 
-	// sysuser@xxx:user/repo
+	if !strings.Contains(refURI, "://") {
-	i := strings.Index(url, "@")
+		// scp style syntax which contains *no* port number after the : (and is not parsed by net/url)
-	j := strings.LastIndex(url, ":")
+		// ex: git@try.gitea.io:go-gitea/gitea
+		match := scpSyntax.FindAllStringSubmatch(refURI, -1)
+		if len(match) > 0 {
 
-	// Only process when i < j because git+ssh://git@git.forwardbias.in/npploader.git
+			m := match[0]
-	if i > -1 && j > -1 && i < j {
+			refHostname := m[2]
-		// fix problem with reverse proxy works only with local server
+			path := m[3]
-		if strings.Contains(urlPrefix, url[i+1:j]) {
+
-			return urlPrefix + url[j+1:]
+			if !strings.HasPrefix(path, "/") {
+				path = "/" + path
+			}
+
+			if urlPrefixHostname == refHostname {
+				return prefixURL.Scheme + "://" + urlPrefixHostname + path
+			}
+			return "http://" + refHostname + path
 		}
-		if strings.HasPrefix(url, "ssh://") || strings.HasPrefix(url, "git+ssh://") {
-			k := strings.Index(url[j+1:], "/")
-			return "http://" + url[i+1:j] + "/" + url[j+1:][k+1:]
-		}
-		return "http://" + url[i+1:j] + "/" + url[j+1:]
 	}
 
-	return url
+	ref, err := url.Parse(refURI)
+	if err != nil {
+		return ""
+	}
+
+	refHostname, _, err := net.SplitHostPort(ref.Host)
+	if err != nil {
+		refHostname = ref.Host
+	}
+
+	supportedSchemes := []string{"http", "https", "git", "ssh", "git+ssh"}
+
+	for _, scheme := range supportedSchemes {
+		if ref.Scheme == scheme {
+			if urlPrefixHostname == refHostname {
+				return prefixURL.Scheme + "://" + prefixURL.Host + ref.Path
+			} else if ref.Scheme == "http" || ref.Scheme == "https" {
+				if len(ref.User.Username()) > 0 {
+					return ref.Scheme + "://" + fmt.Sprintf("%v", ref.User) + "@" + ref.Host + ref.Path
+				}
+				return ref.Scheme + "://" + ref.Host + ref.Path
+			} else {
+				return "http://" + refHostname + ref.Path
+			}
+		}
+	}
+
+	return ""
 }
 
 // RefURL guesses and returns reference URL.

modules/git/submodule_test.go

@@ -19,8 +19,19 @@ func TestGetRefURL(t *testing.T) {
 	}{
 		{"git://github.com/user1/repo1", "/", "/", "http://github.com/user1/repo1"},
 		{"https://localhost/user1/repo1.git", "/", "/", "https://localhost/user1/repo1"},
-		{"git@github.com/user1/repo1.git", "/", "/", "git@github.com/user1/repo1"},
+		{"http://localhost/user1/repo1.git", "/", "/", "http://localhost/user1/repo1"},
+		{"git@github.com:user1/repo1.git", "/", "/", "http://github.com/user1/repo1"},
 		{"ssh://git@git.zefie.net:2222/zefie/lge_g6_kernel_scripts.git", "/", "/", "http://git.zefie.net/zefie/lge_g6_kernel_scripts"},
+		{"git@git.zefie.net:2222/zefie/lge_g6_kernel_scripts.git", "/", "/", "http://git.zefie.net/2222/zefie/lge_g6_kernel_scripts"},
+		{"git@try.gitea.io:go-gitea/gitea", "https://try.gitea.io/go-gitea/gitea", "/", "https://try.gitea.io/go-gitea/gitea"},
+		{"ssh://git@try.gitea.io:9999/go-gitea/gitea", "https://try.gitea.io/go-gitea/gitea", "/", "https://try.gitea.io/go-gitea/gitea"},
+		{"git://git@try.gitea.io:9999/go-gitea/gitea", "https://try.gitea.io/go-gitea/log", "/", "https://try.gitea.io/go-gitea/gitea"},
+		{"ssh://git@127.0.0.1:9999/go-gitea/gitea", "https://127.0.0.1:3000/go-gitea/log", "/", "https://127.0.0.1:3000/go-gitea/gitea"},
+		{"https://gitea.com:3000/user1/repo1.git", "https://127.0.0.1:3000/go-gitea/gitea", "/", "https://gitea.com:3000/user1/repo1"},
+		{"https://username:password@github.com/username/repository.git", "/", "/", "https://username:password@github.com/username/repository"},
+		{"somethingbad", "https://127.0.0.1:3000/go-gitea/gitea", "/", ""},
+		{"git@localhost:user/repo", "https://localhost/user/repo2", "/", "https://localhost/user/repo"},
+		{"../path/to/repo.git/", "https://localhost/user/repo2/src/branch/master/test", "/", "../path/to/repo.git/"},
 	}
 
 	for _, kase := range kases {