Handle scp-like syntax in URL parsing, add more tests

pkg/url.go

@@ -1,28 +1,57 @@
 package pkg
 
 import (
-	"net/url"
+	urlpkg "net/url"
 	"path"
+	"regexp"
 	"strings"
 
 	"github.com/pkg/errors"
 )
 
-// TODO: maybe use https://github.com/whilp/git-urls?
+// scpSyntax matches the SCP-like addresses used by the ssh procotol (eg, [user@]host.xz:path/to/repo.git/).
+// See: https://golang.org/src/cmd/go/internal/get/vcs.go
+var scpSyntax = regexp.MustCompile(`^([a-zA-Z0-9_]+)@([a-zA-Z0-9._-]+):(.*)$`)
 
 func URLToPath(rawurl string) (string, error) {
-	parsed, err := url.Parse(rawurl)
+	url, err := parseURL(rawurl)
 	if err != nil {
-		return "", errors.Wrap(err, "Failed parsing URL")
+		return "", err
 	}
 
-	repoHost := strings.Split(parsed.Host, ":")[0]
+	// remove port numbers from host
+	repoHost := strings.Split(url.Host, ":")[0]
 
-	repoPath := parsed.Path
+	// remove trailing ".git" from repo name
-	//repoPath = strings.TrimSuffix(repoPath, ".git/")
+	localPath := path.Join(repoHost, url.Path)
-	//repoPath = strings.TrimSuffix(repoPath, ".git")
-
-	localPath := path.Join(repoHost, repoPath)
 	localPath = strings.TrimSuffix(localPath, ".git")
+
+	// remove tilde (~) char from username
+	localPath = strings.ReplaceAll(localPath, "~", "")
+
 	return localPath, nil
 }
+
+func parseURL(rawurl string) (url *urlpkg.URL, err error) {
+	// If rawurl matches SCP-like syntax, convert it into a URL.
+	// eg, "git@github.com:user/repo" becomes "ssh://git@github.com/user/repo".
+	if m := scpSyntax.FindStringSubmatch(rawurl); m != nil {
+		url = &urlpkg.URL{
+			Scheme: "ssh",
+			User:   urlpkg.User(m[1]),
+			Host:   m[2],
+			Path:   m[3],
+		}
+	} else {
+		url, err = urlpkg.Parse(rawurl)
+		if err != nil {
+			return nil, errors.Wrap(err, "Failed parsing URL")
+		}
+	}
+
+	if url.Host == "" && url.Path == "" {
+		return nil, errors.New("Parsed URL is empty")
+	}
+
+	return url, nil
+}

pkg/url_test.go

@@ -19,18 +19,15 @@ func TestURLParse(t *testing.T) {
 		in   string
 		want string
 	}{
-
 		{"ssh://github.com/grdl/git-get.git", "github.com/grdl/git-get"},
 		{"ssh://user@github.com/grdl/git-get.git", "github.com/grdl/git-get"},
-		{"ssh://user@github.com/~user/grdl/git-get.git", "github.com/~user/grdl/git-get"}, // TODO: is this what we want or should the result be github.com/grdl/git-get?
+		{"ssh://user@github.com:1234/grdl/git-get.git", "github.com/grdl/git-get"},
+		{"ssh://user@github.com/~user/grdl/git-get.git", "github.com/user/grdl/git-get"},
+		{"git+ssh://github.com/grdl/git-get.git", "github.com/grdl/git-get"},
 		{"git@github.com:grdl/git-get.git", "github.com/grdl/git-get"},
-		{"git@github.com:/~user/grdl/git-get.git", "github.com/grdl/git-get"},
+		{"git@github.com:/~user/grdl/git-get.git", "github.com/user/grdl/git-get"},
-
+		{"git://github.com/grdl/git-get.git", "github.com/grdl/git-get"},
-		{"https://github.com/grdl/git-get.git", "github.com/grdl/git-get"},
+		{"git://github.com/~user/grdl/git-get.git", "github.com/user/grdl/git-get"},
-		{"https://github.com/grdl/git-get.git", "github.com/grdl/git-get"},
-		{"https://github.com/grdl/git-get.git", "github.com/grdl/git-get"},
-		{"https://github.com/grdl/git-get.git", "github.com/grdl/git-get"},
-
 		{"https://github.com/grdl/git-get.git", "github.com/grdl/git-get"},
 		{"http://github.com/grdl/git-get.git", "github.com/grdl/git-get"},
 		{"https://github.com/grdl/git-get", "github.com/grdl/git-get"},
@@ -42,12 +39,11 @@ func TestURLParse(t *testing.T) {
 		{"https://github.com/grdl/git-get/", "github.com/grdl/git-get"},
 		{"https://github.com/grdl/git-get/////", "github.com/grdl/git-get"},
 		{"https://github.com/grdl/git-get.git/////", "github.com/grdl/git-get"},
-
+		{"ftp://github.com/grdl/git-get.git", "github.com/grdl/git-get"},
+		{"ftps://github.com/grdl/git-get.git", "github.com/grdl/git-get"},
+		{"rsync://github.com/grdl/git-get.git", "github.com/grdl/git-get"},
 		{"local/grdl/git-get/", "local/grdl/git-get"},
 		{"file://local/grdl/git-get", "local/grdl/git-get"},
-		{"https://github.com/grdl/git-get/////", "github.com/grdl/git-get"},
-		{"https://github.com/grdl/git-get/////", "github.com/grdl/git-get"},
-		{"https://github.com/grdl/git-get/////", "github.com/grdl/git-get"},
 	}
 
 	for _, test := range tests {
@@ -61,3 +57,20 @@ func TestURLParse(t *testing.T) {
 		}
 	}
 }
+
+func TestInvalidURLParse(t *testing.T) {
+	invalidURLs := []string{
+		"",
+		//TODO: This URL is technically a correct scp-like syntax. Not sure how to handle it
+		"github.com:grdl/git-git.get.git",
+		"git@github.com:1234:grdl/git-get.git",
+	}
+
+	for _, url := range invalidURLs {
+		got, err := URLToPath(url)
+		if err == nil {
+			t.Errorf("Wrong result of parsing invalid URL: %s, got: %s, want: error", url, got)
+		}
+	}
+
+}