From 66c5647c687e722dbbfc52e295ddf6613c283654 Mon Sep 17 00:00:00 2001
From: Sam <sfowler@redhat.com>
Date: Thu, 19 Dec 2024 14:59:38 +1000
Subject: [PATCH] Custom directory ignore

---
 cmd/deplist/deplist.go  |  9 ++++++++-
 deplist.go              | 25 ++++++++++++++++++++-----
 deplist_test.go         |  2 +-
 internal/utils/utils.go | 16 ----------------
 4 files changed, 29 insertions(+), 23 deletions(-)

diff --git a/cmd/deplist/deplist.go b/cmd/deplist/deplist.go
index c57d3f3..68b30f3 100644
--- a/cmd/deplist/deplist.go
+++ b/cmd/deplist/deplist.go
@@ -3,6 +3,7 @@ package main
 import (
 	"flag"
 	"fmt"
+	"strings"
 
 	"github.com/RedHatProductSecurity/deplist"
 	purl "github.com/mcoops/packageurl-go"
@@ -12,6 +13,7 @@ import (
 func main() {
 	deptypePtr := flag.Int("deptype", -1, "golang, nodejs, python etc")
 	debugPtr := flag.Bool("debug", false, "debug logging (default false)")
+	ignorePtr := flag.String("ignore", "", "comma separated list of directory names to ignore (default '')")
 
 	flag.Parse()
 
@@ -19,6 +21,11 @@ func main() {
 		log.SetLevel(log.DebugLevel)
 	}
 
+	var ignoreDirs []string
+	if ignorePtr != nil {
+		ignoreDirs = strings.Split(*ignorePtr, ",")
+	}
+
 	if flag.Args() == nil || len(flag.Args()) == 0 {
 		fmt.Println("No path to scan was specified, i.e. deplist /tmp/files/")
 		return
@@ -26,7 +33,7 @@ func main() {
 
 	path := flag.Args()[0]
 
-	deps, _, err := deplist.GetDeps(path)
+	deps, _, err := deplist.GetDeps(path, ignoreDirs)
 	if err != nil {
 		fmt.Println(err.Error())
 	}
diff --git a/deplist.go b/deplist.go
index 5d09061..a39c0fc 100644
--- a/deplist.go
+++ b/deplist.go
@@ -5,6 +5,7 @@ import (
 	"os"
 	"os/exec"
 	"path/filepath"
+	"slices"
 	"strings"
 
 	"github.com/RedHatProductSecurity/deplist/internal/scan"
@@ -61,7 +62,17 @@ func addPackagesToDeps(discovered Discovered, pkgs map[string]string, lang Bitma
 	return discovered
 }
 
-func getDeps(fullPath string) ([]Dependency, Bitmask, error) {
+var defaultIgnore []string = []string{
+	"node_modules",
+	"vendor",
+	"scripts",
+	"docs",
+	"test",
+	"tests",
+	".git",
+}
+
+func getDeps(fullPath string, ignoreDirs []string) ([]Dependency, Bitmask, error) {
 	var discovered Discovered
 	// special var so we don't double handle both repos with both
 	// a Gemfile and Gemfile.lock
@@ -78,6 +89,9 @@ func getDeps(fullPath string) ([]Dependency, Bitmask, error) {
 	rubyPath := filepath.Join(fullPath, "Gemfile") // Later we translate Gemfile.lock -> Gemfile to handle both cases
 	pythonPath := filepath.Join(fullPath, "requirements.txt")
 
+	ignoreDirs = append(ignoreDirs, defaultIgnore...)
+	log.Debugf("directories ignored: %s", ignoreDirs)
+
 	// point at the parent repo, but can't assume where the indicators will be
 	err := filepath.Walk(fullPath, func(path string, info os.FileInfo, err error) error {
 		if err != nil {
@@ -87,7 +101,8 @@ func getDeps(fullPath string) ([]Dependency, Bitmask, error) {
 
 		if info.IsDir() {
 			// prevent walking down the vendors, docs, etc
-			if utils.BelongsToIgnoreList(info.Name()) {
+			if slices.Contains(ignoreDirs, info.Name()) {
+				log.Debugf("Skipping '%s', directory  name '%s' in ignore list", path, info.Name())
 				return filepath.SkipDir
 			}
 		} else {
@@ -293,13 +308,13 @@ func findBaseDir(fullPath string) (string, error) {
 }
 
 // GetDeps scans a given repository and returns all dependencies found in a DependencyList struct.
-func GetDeps(fullPath string) ([]Dependency, Bitmask, error) {
+func GetDeps(fullPath string, ignoreDirs []string) ([]Dependency, Bitmask, error) {
 	fullPath, err := findBaseDir(fullPath)
 	if err != nil {
 		return nil, 0, err
 	}
 
-	deps, foundTypes, err := getDeps(fullPath)
+	deps, foundTypes, err := getDeps(fullPath, ignoreDirs)
 	if err != nil {
 		return deps, foundTypes, err
 	}
@@ -309,7 +324,7 @@ func GetDeps(fullPath string) ([]Dependency, Bitmask, error) {
 		fullPath = filepath.Join(fullPath, "src")
 		if _, err := os.Stat(fullPath); err != nil {
 			log.Debugf("No deps found, trying %s", fullPath)
-			deps, foundTypes, _ = getDeps(fullPath)
+			deps, foundTypes, _ = getDeps(fullPath, ignoreDirs)
 		}
 	}
 
diff --git a/deplist_test.go b/deplist_test.go
index 1d38523..2f7717c 100644
--- a/deplist_test.go
+++ b/deplist_test.go
@@ -352,7 +352,7 @@ func depToKey(pkg Dependency) string {
 func TestGetDeps(t *testing.T) {
 	want := BuildWant()
 
-	got, gotBitmask, err := GetDeps("test/testRepo")
+	got, gotBitmask, err := GetDeps("test/testRepo", []string{})
 	if err != nil {
 		t.Errorf("GetDeps failed: %s", err)
 		return
diff --git a/internal/utils/utils.go b/internal/utils/utils.go
index ae8b1d7..728f2c1 100644
--- a/internal/utils/utils.go
+++ b/internal/utils/utils.go
@@ -5,22 +5,6 @@ import (
 	"path/filepath"
 )
 
-// BelongsToIgnoreList is fastest way we can do a string compare on a list
-func BelongsToIgnoreList(needle string) bool {
-	switch needle {
-	case
-		"node_modules",
-		"vendor",
-		"scripts",
-		"docs",
-		"test",
-		"tests",
-		".git":
-		return true
-	}
-	return false
-}
-
 func CharIsDigit(c string) bool {
 	if len(c) == 0 {
 		return false