From 8c077cefae3bfe0ad3725ffcd5488c5cb4612784 Mon Sep 17 00:00:00 2001
From: Lucas Pinheiro <lpinheiro@renderedtext.com>
Date: Tue, 17 Dec 2024 08:17:32 -0300
Subject: [PATCH] fix: handle custom domain URL (#62)

---
 pkg/api/signed_url.go      | 20 ++++++++++++++++++++
 pkg/api/signed_url_test.go | 18 ++++++++++++++++--
 2 files changed, 36 insertions(+), 2 deletions(-)

diff --git a/pkg/api/signed_url.go b/pkg/api/signed_url.go
index bd716f5..523cb4f 100644
--- a/pkg/api/signed_url.go
+++ b/pkg/api/signed_url.go
@@ -14,6 +14,10 @@ import (
 	log "github.com/sirupsen/logrus"
 )
 
+var (
+	customDomainRegex = regexp.MustCompile(`https:\/\/[a-z0-9\-\.]+\/[a-z0-9\-]+\/[a-z0-9\-]+\/([^?]+)\?`)
+)
+
 type SignedURL struct {
 	URL    string `json:"url,omitempty"`
 	Method string `json:"method,omitempty"`
@@ -221,6 +225,10 @@ func (u *SignedURL) GetObject() (string, error) {
 		log.Debugf("Parsing localhost URL: %s\n", u.URL)
 		return parseLocalhostURL(URL)
 
+	case customDomainRegex.Match([]byte(URL.String())):
+		log.Debugf("Parsing custom domain URL: %s\n", u.URL)
+		return parseCustomDomainURL(URL)
+
 	default:
 		log.Warnf("Failed to parse URL '%s' - unrecognized host '%s'\n", u.URL, host)
 		return "", fmt.Errorf("unrecognized host %s", host)
@@ -257,6 +265,18 @@ func parseS3URL(URL *url.URL) (string, error) {
 	return parsed[3], nil
 }
 
+// Custom domain URLs are used when minio or some other s3-compatible storage is being used.
+// The URL will be of the form: https://<domain>/<bucket>/<semaphore-project-id>/<path>
+// We are only interested in the <path> part here.
+func parseCustomDomainURL(URL *url.URL) (string, error) {
+	parsed := customDomainRegex.FindStringSubmatch(URL.String())
+	if len(parsed) < 2 {
+		return "", fmt.Errorf("Failed to parse custom domain URL '%s'\n", URL)
+	}
+
+	return parsed[1], nil
+}
+
 // Localhost URLs are used during tests
 func parseLocalhostURL(URL *url.URL) (string, error) {
 	// we don't want the leading slash
diff --git a/pkg/api/signed_url_test.go b/pkg/api/signed_url_test.go
index 4fe5604..32b24c4 100644
--- a/pkg/api/signed_url_test.go
+++ b/pkg/api/signed_url_test.go
@@ -63,8 +63,22 @@ func Test__GetObject(t *testing.T) {
 		assert.Equal(t, "artifacts/project/projectid/mydir/myfile.txt", obj)
 	})
 
-	t.Run("bad host", func(t *testing.T) {
-		signedURL := SignedURL{URL: "https://somehost.com/projectid/artifacts/project/projectid/myfile.txt"}
+	t.Run("custom domain - file", func(t *testing.T) {
+		signedURL := SignedURL{URL: "https://artifacts.somedomain.com/my-bucket1/projectid/artifacts/project/projectid/myfile.txt?X-Amz-Algorithm"}
+		obj, err := signedURL.GetObject()
+		assert.Nil(t, err)
+		assert.Equal(t, "artifacts/project/projectid/myfile.txt", obj)
+	})
+
+	t.Run("custom domain - file inside directory", func(t *testing.T) {
+		signedURL := SignedURL{URL: "https://artifacts.somedomain.com/my-bucket1/projectid/artifacts/project/projectid/mydir/myfile.txt?Expires=231256754712"}
+		obj, err := signedURL.GetObject()
+		assert.Nil(t, err)
+		assert.Equal(t, "artifacts/project/projectid/mydir/myfile.txt", obj)
+	})
+
+	t.Run("bad URL", func(t *testing.T) {
+		signedURL := SignedURL{URL: "http://somehost.com/projectid/artifacts/project/projectid/myfile.txt"}
 		_, err := signedURL.GetObject()
 		assert.NotNil(t, err)
 	})