Allow Gemfury storage backend for jobs using a Gemfury registry

internal/handlers/egress_dynamic_hosts.go

@@ -18,22 +18,34 @@ import (
 // here — it has no capture group, and its "*" spans dots.
 var ecrHostPattern = regexp.MustCompile(`^[0-9]{12}\.dkr\.ecr\.([a-z0-9-]+)\.amazonaws\.com$`)
 
+// gemfuryStorageHost is the single Gemfury-owned S3 bucket that every Gemfury
+// registry (pypi.fury.io, npm.fury.io, ...) 302-redirects package downloads to
+// via short-lived pre-signed URLs.
+const gemfuryStorageHost = "gemfury.s3-accelerate.dualstack.amazonaws.com"
+
 // registryRedirectHosts returns storage backends that a configured registry
 // redirects to on download but that appear in no credential field.
 //
 // Private ECR 307-redirects layer downloads to a per-region, AWS-owned S3
 // bucket. It is derived per job rather than globbed into the static defaults
 // because "prod-<anything>-starport-layer-bucket" is a claimable S3 name, so a
 // glob would hand every job an attacker-registrable destination.
+//
+// Gemfury redirects to one fixed bucket shared by all Gemfury accounts, with
+// the account in the URL path. It is derived per job rather than added to the
+// static defaults because allowing a shared multi-tenant host there would open
+// every tenant's content to every job. The derived host is a constant, so a
+// crafted credential cannot widen it.
 func registryRedirectHosts(credHosts []string) []string {
 	var hosts []string
 	for _, h := range credHosts {
-		m := ecrHostPattern.FindStringSubmatch(h)
-		if m == nil {
-			continue
+		if m := ecrHostPattern.FindStringSubmatch(h); m != nil {
+			region := m[1]
+			hosts = append(hosts, fmt.Sprintf("prod-%s-starport-layer-bucket.s3.%s.amazonaws.com", region, region))
+		}
+		if strings.HasSuffix(h, ".fury.io") {
+			hosts = append(hosts, gemfuryStorageHost)
 		}
-		region := m[1]
-		hosts = append(hosts, fmt.Sprintf("prod-%s-starport-layer-bucket.s3.%s.amazonaws.com", region, region))
 	}
 	return hosts
 }

internal/handlers/egress_dynamic_hosts_test.go

@@ -147,6 +147,67 @@ func TestRegistryRedirectHosts_OnlyCanonicalECRHosts(t *testing.T) {
 		registryRedirectHosts([]string{"123456789012.dkr.ecr.us-east-2.amazonaws.com"}))
 }
 
+func TestRegistryRedirectHosts_GemfuryStorageDerived(t *testing.T) {
+	// Gemfury 302-redirects package downloads from every registry endpoint to a
+	// single Gemfury-owned S3 bucket via pre-signed URLs, so the bucket is
+	// derived from the job's own Gemfury credential.
+	creds := config.Credentials{
+		{"type": "python_index", "index-url": "https://pypi.fury.io/acme/"},
+	}
+	h := newEgressHandlerWithCreds(creds)
+
+	assert.Nil(t, egressResult(t, h, "https://pypi.fury.io/acme/-/ver_x/pkg-1.0.0-py3-none-any.whl"),
+		"the Gemfury registry itself must be allowed")
+	assert.Nil(t, egressResult(t, h, "https://gemfury.s3-accelerate.dualstack.amazonaws.com/gems/x/pkg_whl?X-Amz-Signature=x"),
+		"the Gemfury storage bucket must be allowed")
+
+	for _, blocked := range []string{
+		// Dynamic hosts are matched exactly, so no child or lookalike widens it.
+		"https://evil.gemfury.s3-accelerate.dualstack.amazonaws.com/loot",
+		"https://gemfuryx.s3-accelerate.dualstack.amazonaws.com/loot",
+		"https://gemfury.s3.amazonaws.com/loot",
+		// The shared parent namespace stays closed.
+		"https://attacker.s3-accelerate.dualstack.amazonaws.com/loot",
+	} {
+		assert.NotNil(t, egressResult(t, h, blocked), "must remain blocked: "+blocked)
+	}
+}
+
+func TestRegistryRedirectHosts_OnlyGemfuryHosts(t *testing.T) {
+	for _, h := range []string{
+		"pypi.fury.io",
+		"npm.fury.io",
+		"npm-proxy.fury.io",
+		"gem.fury.io",
+	} {
+		assert.Equal(t, []string{gemfuryStorageHost}, registryRedirectHosts([]string{h}),
+			"must derive the Gemfury bucket from %q", h)
+	}
+
+	for _, h := range []string{
+		"fury.io",               // apex is not a registry endpoint
+		"pypi.fury.io.evil.com", // suffix must be fury.io
+		"pypifury.io",
+		"evil.com",
+	} {
+		assert.Empty(t, registryRedirectHosts([]string{h}), "must derive nothing from %q", h)
+	}
+
+	// Several Gemfury credentials still yield a single entry.
+	assert.Equal(t, []string{"pypi.fury.io", "npm.fury.io", gemfuryStorageHost}, dynamicHosts(config.Credentials{
+		{"type": "python_index", "index-url": "https://pypi.fury.io/acme/"},
+		{"type": "npm_registry", "registry": "https://npm.fury.io/acme/"},
+	}))
+}
+
+func TestRegistryRedirectHosts_NotAddedWithoutGemfuryCredential(t *testing.T) {
+	h := newEgressHandlerWithCreds(config.Credentials{
+		{"type": "python_index", "index-url": "https://pypi.internal.example.com/simple"},
+	})
+	assert.NotNil(t, egressResult(t, h, "https://gemfury.s3-accelerate.dualstack.amazonaws.com/gems/x/loot"),
+		"Gemfury bucket must not be allowed for a job with no Gemfury credential")
+}
+
 func TestRegistryRedirectHosts_NotAddedWithoutECRCredential(t *testing.T) {
 	h := newEgressHandlerWithCreds(config.Credentials{
 		{"type": "docker_registry", "registry": "https://registry.internal.example.com"},