resolved: SAAS-5256 new sanitise formula for scm id. (#92)

aquasecurity · May 9, 2022 · 6da91d4 · 6da91d4
1 parent 6950435
commit 6da91d4
Show file tree

Hide file tree

Showing 3 changed files with 19 additions and 22 deletions.
diff --git a/Makefile b/Makefile
@@ -33,4 +33,4 @@ build:
   -w /build \
   -v `pwd`:/build \
   golang:1.18 \
-  go build -o /build/bin/aqua cmd/aqua/main.go|| exit 1
+  go build -o build/bin/aqua cmd/aqua/main.go|| exit 1
diff --git a/pkg/metadata/collector.go b/pkg/metadata/collector.go
@@ -85,12 +85,11 @@ func GetImageDetails(imageName string) (prefix, repo, tag string) {
 func GetScmID(scanPath string) (string, error) {
 	gitConfigFile := filepath.Join(scanPath, ".git", "config")
 	gitConfig, err := ioutil.ReadFile(gitConfigFile)
-	system := GetBuildSystem()
 	if err == nil {
 		re := regexp.MustCompile(`(?m)^\s*url\s?=\s*(.*)\s*$`)
 		if re.Match(gitConfig) {
 			scmID := re.FindStringSubmatch(string(gitConfig))[1]
-			scmID = sanitiseScmId(system, scmID)
+			scmID = sanitiseScmId(scmID)
 
 			return scmID, nil
 		}
@@ -99,10 +98,13 @@ func GetScmID(scanPath string) (string, error) {
 	return filepath.Base(scanPath), nil
 }
 
-func sanitiseScmId(system string, scmID string) string {
-	if system == "gitlab" {
-		scmID = strings.Split(scmID, "@")[1]
-	}
+// This function is the formula on the DB, on Atlas side and on Argon side,
+// do not change without coordinating the change
+func sanitiseScmId(scmID string) string {
+	scmID = regexp.MustCompile("^.*@").ReplaceAllLiteralString(scmID, "")
+	scmID = regexp.MustCompile("^https?://").ReplaceAllLiteralString(scmID, "")
+	scmID = strings.TrimSuffix(scmID, ".git")
+	scmID = strings.ReplaceAll(scmID, ":", "/")
 	return scmID
 }
 

diff --git a/pkg/metadata/collector_test.go b/pkg/metadata/collector_test.go
@@ -6,8 +6,7 @@ import (
 
 func Test_convertScmId(t *testing.T) {
 	type args struct {
-		system string
-		scmID  string
+		scmID string
 	}
 	tests := []struct {
 		name string
@@ -17,42 +16,38 @@ func Test_convertScmId(t *testing.T) {
 		{
 			name: "happy path - bitbucket",
 			args: args{
-				scmID:  "git@bitbucket.org:repo.git",
-				system: "bitbucket",
+				scmID: "git@bitbucket.org:repo.git",
 			},
-			want: "git@bitbucket.org:repo.git",
+			want: "bitbucket.org/repo",
 		},
 		{
 			name: "happy path - github",
 			args: args{
-				scmID:  "https://github.com/repo.git",
-				system: "github",
+				scmID: "https://github.com/repo.git",
 			},
-			want: "https://github.com/repo.git",
+			want: "github.com/repo",
 		},
 		{
 			name: "happy path - other",
 			args: args{
-				scmID:  "git@othertest.com/repo.git",
-				system: "other",
+				scmID: "git@othertest.com/repo.git",
 			},
-			want: "git@othertest.com/repo.git",
+			want: "othertest.com/repo",
 		},
 
 		{
 			name: "happy path - gitlab",
 			args: args{
-				scmID:  "https://gitlab-ci-token:123456-abcdef@gitlab.com/aqua/repo.git",
-				system: "gitlab",
+				scmID: "https://gitlab-ci-token:123456-abcdef@gitlab.com/aqua/repo.git",
 			},
-			want: "gitlab.com/aqua/repo.git",
+			want: "gitlab.com/aqua/repo",
 		},
 	}
 	for _, tt := range tests {
 		tt := tt
 		t.Run(tt.name, func(t *testing.T) {
 			t.Parallel()
-			if got := sanitiseScmId(tt.args.system, tt.args.scmID); got != tt.want {
+			if got := sanitiseScmId(tt.args.scmID); got != tt.want {
 				t.Errorf("sanitiseScmId() = %v, want %v", got, tt.want)
 			}
 		})