From 04c9bb535e5d548deec0f30c9a16e3b7695b2a59 Mon Sep 17 00:00:00 2001
From: ahrav <ahravdutta02@gmail.com>
Date: Wed, 12 Oct 2022 16:28:24 -0700
Subject: [PATCH] [THOG-768] - Add ability to skip scanning Github repos (#846)

* Add ability to skip scanning Github repos.

* remove old change.

* rename method.
---
 pkg/sources/github/github.go      | 49 +++++++++++++++++++++++++------
 pkg/sources/github/github_test.go | 19 ++++++++----
 2 files changed, 54 insertions(+), 14 deletions(-)

diff --git a/pkg/sources/github/github.go b/pkg/sources/github/github.go
index f60f108fb..6a0ff0698 100644
--- a/pkg/sources/github/github.go
+++ b/pkg/sources/github/github.go
@@ -42,22 +42,23 @@ const (
 )
 
 type Source struct {
-	name            string
-	sourceID        int64
-	jobID           int64
-	verify          bool
-	repos           []string
-	orgs            []string
-	members         []string
+	name     string
+	token    string
+	sourceID int64
+	jobID    int64
+	verify   bool
+	repos,
+	orgs,
+	members,
+	ignoreRepos []string
 	git             *git.Git
 	httpClient      *http.Client
 	aCtx            context.Context
 	log             *log.Entry
-	token           string
 	conn            *sourcespb.GitHub
 	jobPool         *errgroup.Group
-	resumeInfoSlice []string
 	resumeInfoMutex sync.Mutex
+	resumeInfoSlice []string
 	apiClient       *github.Client
 	publicMap       map[string]source_metadatapb.Visibility
 	sources.Progress
@@ -131,6 +132,7 @@ func (s *Source) Init(aCtx context.Context, name string, jobID, sourceID int64,
 
 	s.repos = s.conn.Repositories
 	s.orgs = s.conn.Organizations
+	s.ignoreRepos = s.conn.IgnoreRepos
 
 	// Head or base should only be used with incoming webhooks
 	if (len(s.conn.Head) > 0 || len(s.conn.Base) > 0) && len(s.repos) != 1 {
@@ -600,6 +602,7 @@ func (s *Source) getReposByOrg(ctx context.Context, org string) ([]string, error
 			PerPage: defaultPagination,
 		},
 	}
+
 	var numRepos, numForks int
 	for {
 		someRepos, res, err := s.apiClient.Repositories.ListByOrg(ctx, org, opts)
@@ -615,8 +618,13 @@ func (s *Source) getReposByOrg(ctx context.Context, org string) ([]string, error
 		if len(someRepos) == 0 || res == nil {
 			break
 		}
+
 		s.log.Debugf("Listed repos for org %s page %d/%d", org, opts.Page, res.LastPage)
 		for _, r := range someRepos {
+			if s.ignoreRepo(r.GetName()) {
+				continue
+			}
+
 			numRepos++
 			if r.GetFork() {
 				numForks++
@@ -654,6 +662,7 @@ func (s *Source) getReposByUser(ctx context.Context, user string) ([]string, err
 			PerPage: 50,
 		},
 	}
+
 	for {
 		someRepos, res, err := s.apiClient.Repositories.List(ctx, user, opts)
 		if err == nil {
@@ -668,8 +677,13 @@ func (s *Source) getReposByUser(ctx context.Context, user string) ([]string, err
 		if res == nil {
 			break
 		}
+
 		s.log.Debugf("Listed repos for user %s page %d/%d", user, opts.Page, res.LastPage)
 		for _, r := range someRepos {
+			if s.ignoreRepo(r.GetName()) {
+				continue
+			}
+
 			if r.GetFork() && !s.conn.IncludeForks {
 				continue
 			}
@@ -683,6 +697,23 @@ func (s *Source) getReposByUser(ctx context.Context, user string) ([]string, err
 	return repos, nil
 }
 
+func (s *Source) ignoreRepo(r string) bool {
+	if stringInSlice(r, s.ignoreRepos) {
+		s.log.Debugf("ignoring repo %s", r)
+		return true
+	}
+	return false
+}
+
+func stringInSlice(s string, l []string) bool {
+	for _, b := range l {
+		if b == s {
+			return true
+		}
+	}
+	return false
+}
+
 func (s *Source) getGistsByUser(ctx context.Context, user string) ([]string, error) {
 	var gistURLs []string
 	gistOpts := &github.GistListOptions{}
diff --git a/pkg/sources/github/github_test.go b/pkg/sources/github/github_test.go
index 41bd5f23b..a3dc968da 100644
--- a/pkg/sources/github/github_test.go
+++ b/pkg/sources/github/github_test.go
@@ -17,12 +17,13 @@ import (
 	"github.com/google/go-github/v42/github"
 	"github.com/sirupsen/logrus"
 	"github.com/stretchr/testify/assert"
-	"github.com/trufflesecurity/trufflehog/v3/pkg/context"
-	"github.com/trufflesecurity/trufflehog/v3/pkg/pb/credentialspb"
-	"github.com/trufflesecurity/trufflehog/v3/pkg/pb/sourcespb"
 	"golang.org/x/sync/errgroup"
 	"google.golang.org/protobuf/types/known/anypb"
 	"gopkg.in/h2non/gock.v1"
+
+	"github.com/trufflesecurity/trufflehog/v3/pkg/context"
+	"github.com/trufflesecurity/trufflehog/v3/pkg/pb/credentialspb"
+	"github.com/trufflesecurity/trufflehog/v3/pkg/pb/sourcespb"
 )
 
 func createTestSource(src *sourcespb.GitHub) (*Source, *anypb.Any) {
@@ -64,9 +65,13 @@ func TestAddReposByOrg(t *testing.T) {
 	gock.New("https://api.github.com").
 		Get("/orgs/super-secret-org/repos").
 		Reply(200).
-		JSON([]map[string]string{{"clone_url": "super-secret-repo"}})
+		JSON([]map[string]string{
+			{"clone_url": "super-secret-repo", "name": "super-secret-repo"},
+			{"clone_url": "super-secret-repo2", "name": "super-secret-repo2"},
+		})
 
 	s := initTestSource(nil)
+	s.ignoreRepos = []string{"super-secret-repo2"}
 	// gock works here because github.NewClient is using the default HTTP Transport
 	err := s.addRepos(context.TODO(), "super-secret-org", s.getReposByOrg)
 	assert.Nil(t, err)
@@ -81,9 +86,13 @@ func TestAddReposByUser(t *testing.T) {
 	gock.New("https://api.github.com").
 		Get("/users/super-secret-user/repos").
 		Reply(200).
-		JSON([]map[string]string{{"clone_url": "super-secret-repo"}})
+		JSON([]map[string]string{
+			{"clone_url": "super-secret-repo", "name": "super-secret-repo"},
+			{"clone_url": "super-secret-repo2", "name": "super-secret-repo2"},
+		})
 
 	s := initTestSource(nil)
+	s.ignoreRepos = []string{"super-secret-repo2"}
 	err := s.addRepos(context.TODO(), "super-secret-user", s.getReposByUser)
 	assert.Nil(t, err)
 	assert.Equal(t, 1, len(s.repos))