From acc79ebd436c5b4390f861bc6f50d8e9f75f80fe Mon Sep 17 00:00:00 2001
From: Richard Gomez <rmgomez368@gmail.com>
Date: Sat, 13 Jan 2024 17:22:38 -0500
Subject: [PATCH] update ConditionalDetector logic

---
 pkg/detectors/detectors.go       | 16 ++++++++-----
 pkg/detectors/parseur/parseur.go |  7 +++---
 pkg/engine/engine.go             |  2 +-
 pkg/sources/sources.go           | 40 +++++++++++++++++++++++++++-----
 4 files changed, 48 insertions(+), 17 deletions(-)

diff --git a/pkg/detectors/detectors.go b/pkg/detectors/detectors.go
index 0618435e91fc..4f9e40b24718 100644
--- a/pkg/detectors/detectors.go
+++ b/pkg/detectors/detectors.go
@@ -31,18 +31,22 @@ type Detector interface {
 // ConditionalDetector is an optional interface that a detector can implement to
 // skip chunks based on specific criteria.
 type ConditionalDetector interface {
-	// ScanChunk determines whether the detector should run.
-	ScanChunk(chunk sources.Chunk) bool
+	// MatchesChunk determines whether the detector should run.
+	MatchesChunk(chunk sources.Chunk) bool
 }
 
-// FilenameConditions is a set of common conditions to be used by ConditionalDetector.
+var lockFilePat = regexp.MustCompile(`(^|/)(package(-lock)?\.json|yarn\.lock)$`)
+
+// Conditions is a set of common conditions to be used by ConditionalDetector.
 // (Using anonymous structs is weird, but Go has no concept of static members... https://stackoverflow.com/a/55390104)
-var FilenameConditions = struct {
+var Conditions = struct {
 	// LockFiles are a common source of false-positives.
 	// https://github.com/trufflesecurity/trufflehog/issues/1460
-	LockFiles *regexp.Regexp
+	IsLockFile func(path string) bool
 }{
-	LockFiles: regexp.MustCompile(`(^|/)(package(-lock)?\.json|yarn\.lock)$`),
+	IsLockFile: func(path string) bool {
+		return lockFilePat.MatchString(path)
+	},
 }
 
 // Versioner is an optional interface that a detector can implement to
diff --git a/pkg/detectors/parseur/parseur.go b/pkg/detectors/parseur/parseur.go
index adfa2023fa10..c7453630efd5 100644
--- a/pkg/detectors/parseur/parseur.go
+++ b/pkg/detectors/parseur/parseur.go
@@ -33,10 +33,9 @@ func (s Scanner) Keywords() []string {
 	return []string{"parseur"}
 }
 
-func (s Scanner) ScanChunk(chunk sources.Chunk) bool {
-	// TODO: Can |chunk.SourceMetadata| be nil?
-	if m, ok := chunk.SourceMetadata.GetData().(sources.GitSourceMetadata); ok {
-		return !detectors.FilenameConditions.LockFiles.MatchString(m.GetFile())
+func (s Scanner) MatchesChunk(chunk sources.Chunk) bool {
+	if m, ok := sources.NewGitSourceMetadata(chunk.SourceType, chunk.SourceMetadata); ok {
+		return !detectors.Conditions.IsLockFile(m.File)
 	}
 	return true
 }
diff --git a/pkg/engine/engine.go b/pkg/engine/engine.go
index e11461b3e238..ec8dadab4d71 100644
--- a/pkg/engine/engine.go
+++ b/pkg/engine/engine.go
@@ -478,7 +478,7 @@ func (e *Engine) detectorWorker(ctx context.Context) {
 
 				for k, detector := range chunkSpecificDetectors {
 					d, ok := detector.(detectors.ConditionalDetector)
-					if ok && !d.ScanChunk(*chunk) {
+					if ok && !d.MatchesChunk(*chunk) {
 						ctx.Logger().V(4).Info("skipping detector for chunk", "detector", detector.Type().String(), "chunk", chunk)
 						delete(chunkSpecificDetectors, k)
 						continue
diff --git a/pkg/sources/sources.go b/pkg/sources/sources.go
index a929261dd741..149ae3c8093e 100644
--- a/pkg/sources/sources.go
+++ b/pkg/sources/sources.go
@@ -38,12 +38,40 @@ type Chunk struct {
 	Verify bool
 }
 
-// GitSourceMetadata defines a common interface for Git-based source metadata.
-// For example, this should match Git, Azure, Bitbucket, GitHub, and Gitlab.
-type GitSourceMetadata interface {
-	GetRepository() string
-	GetCommit() string
-	GetFile() string
+// GitSourceMetadata defines a common struct for Git-based source metadata.
+type GitSourceMetadata struct {
+	Repository string
+	Commit     string
+	File       string
+}
+
+func NewGitSourceMetadata(source sourcespb.SourceType, data *source_metadatapb.MetaData) (*GitSourceMetadata, bool) {
+	if data == nil {
+		return nil, false
+	}
+
+	switch source {
+	case sourcespb.SourceType_SOURCE_TYPE_GIT:
+		md := data.GetGit()
+		return &GitSourceMetadata{md.GetRepository(), md.GetCommit(), md.GetFile()}, true
+	case sourcespb.SourceType_SOURCE_TYPE_AZURE_REPOS:
+		md := data.GetAzureRepos()
+		return &GitSourceMetadata{md.GetRepository(), md.GetCommit(), md.GetFile()}, true
+	case sourcespb.SourceType_SOURCE_TYPE_BITBUCKET:
+		md := data.GetBitbucket()
+		return &GitSourceMetadata{md.GetRepository(), md.GetCommit(), md.GetFile()}, true
+	case sourcespb.SourceType_SOURCE_TYPE_GERRIT:
+		md := data.GetGerrit()
+		return &GitSourceMetadata{md.GetProject(), md.GetCommit(), md.GetFile()}, true
+	case sourcespb.SourceType_SOURCE_TYPE_GITHUB:
+		md := data.GetGithub()
+		return &GitSourceMetadata{md.GetRepository(), md.GetCommit(), md.GetFile()}, true
+	case sourcespb.SourceType_SOURCE_TYPE_GITLAB:
+		md := data.GetGitlab()
+		return &GitSourceMetadata{md.GetRepository(), md.GetCommit(), md.GetFile()}, true
+	default:
+		return nil, false
+	}
 }
 
 // ChunkingTarget specifies criteria for a targeted chunking process.