From f7628adee5986e5d418c7c4ca82ca5535d4e3d30 Mon Sep 17 00:00:00 2001
From: Andrea Luzzardi <aluzzardi@gmail.com>
Date: Fri, 18 Mar 2022 12:29:44 -0700
Subject: [PATCH] telemetry: Normalize git URLs

Normalize `https://github.com/dagger/dagger` and
`git@github.com:dagger/dagger.git` to `github.com/dagger/dagger`

Signed-off-by: Andrea Luzzardi <aluzzardi@gmail.com>
---
 telemetry/git.go       | 65 ++++++++++++++++++++++++++++++++++++++++++
 telemetry/git_test.go  | 22 ++++++++++++++
 telemetry/telemetry.go | 19 ++++++++----
 3 files changed, 100 insertions(+), 6 deletions(-)
 create mode 100644 telemetry/git.go
 create mode 100644 telemetry/git_test.go

diff --git a/telemetry/git.go b/telemetry/git.go
new file mode 100644
index 00000000..1c0a8fc7
--- /dev/null
+++ b/telemetry/git.go
@@ -0,0 +1,65 @@
+package telemetry
+
+import (
+	"fmt"
+	"net/url"
+	"regexp"
+	"strings"
+)
+
+var (
+	urlSchemeRegExp  = regexp.MustCompile(`^[^:]+://`)
+	scpLikeURLRegExp = regexp.MustCompile(`^(?:(?P<user>[^@]+)@)?(?P<host>[^:\s]+):(?:(?P<port>[0-9]{1,5})(?:\/|:))?(?P<path>[^\\].*\/[^\\].*)$`)
+)
+
+func parseGitURL(endpoint string) (string, error) {
+	if e, ok := parseSCPLike(endpoint); ok {
+		return e, nil
+	}
+
+	return parseURL(endpoint)
+}
+
+func parseURL(endpoint string) (string, error) {
+	u, err := url.Parse(endpoint)
+	if err != nil {
+		return "", err
+	}
+
+	if !u.IsAbs() {
+		return "", fmt.Errorf(
+			"invalid endpoint: %s", endpoint,
+		)
+	}
+
+	return fmt.Sprintf("%s%s", u.Hostname(), u.Path), nil
+}
+
+func parseSCPLike(endpoint string) (string, bool) {
+	if matchesURLScheme(endpoint) || !matchesScpLike(endpoint) {
+		return "", false
+	}
+
+	_, host, _, path := findScpLikeComponents(endpoint)
+
+	return fmt.Sprintf("%s/%s", host, strings.TrimSuffix(path, ".git")), true
+}
+
+// matchesURLScheme returns true if the given string matches a URL-like
+// format scheme.
+func matchesURLScheme(url string) bool {
+	return urlSchemeRegExp.MatchString(url)
+}
+
+// matchesScpLike returns true if the given string matches an SCP-like
+// format scheme.
+func matchesScpLike(url string) bool {
+	return scpLikeURLRegExp.MatchString(url)
+}
+
+// findScpLikeComponents returns the user, host, port and path of the
+// given SCP-like URL.
+func findScpLikeComponents(url string) (user, host, port, path string) {
+	m := scpLikeURLRegExp.FindStringSubmatch(url)
+	return m[1], m[2], m[3], m[4]
+}
diff --git a/telemetry/git_test.go b/telemetry/git_test.go
new file mode 100644
index 00000000..e38e82c9
--- /dev/null
+++ b/telemetry/git_test.go
@@ -0,0 +1,22 @@
+package telemetry
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/require"
+)
+
+func TestParseGit(t *testing.T) {
+	var (
+		endpoint string
+		err      error
+	)
+
+	endpoint, err = parseGitURL("https://github.com/dagger/dagger")
+	require.NoError(t, err)
+	require.Equal(t, endpoint, "github.com/dagger/dagger")
+
+	endpoint, err = parseGitURL("git@github.com:dagger/dagger.git")
+	require.NoError(t, err)
+	require.Equal(t, endpoint, "github.com/dagger/dagger")
+}
diff --git a/telemetry/telemetry.go b/telemetry/telemetry.go
index 6177e4b1..547e212f 100644
--- a/telemetry/telemetry.go
+++ b/telemetry/telemetry.go
@@ -50,7 +50,7 @@ func Track(ctx context.Context, eventName string, properties ...*Property) {
 		return
 	}
 
-	repo := gitRepoURL(".")
+	repo := gitRepoURL(ctx, ".")
 
 	// Base properties
 	props := map[string]interface{}{
@@ -194,8 +194,8 @@ func hash(s string) string {
 	return fmt.Sprintf("%x", sha256.Sum256([]byte(s)))
 }
 
-// // gitRepoURL returns the git repository remote, if any.
-func gitRepoURL(path string) string {
+// gitRepoURL returns the git repository remote, if any.
+func gitRepoURL(ctx context.Context, path string) string {
 	repo, err := git.PlainOpenWithOptions(path, &git.PlainOpenOptions{
 		DetectDotGit: true,
 	})
@@ -208,9 +208,16 @@ func gitRepoURL(path string) string {
 		return ""
 	}
 
-	if urls := origin.Config().URLs; len(urls) > 0 {
-		return urls[0]
+	urls := origin.Config().URLs
+	if len(urls) == 0 {
+		return ""
 	}
 
-	return ""
+	endpoint, err := parseGitURL(urls[0])
+	if err != nil {
+		log.Ctx(ctx).Debug().Err(err).Str("url", urls[0]).Msg("failed to parse git URL")
+		return ""
+	}
+
+	return endpoint
 }