trufflehog/pkg/detectors/dropbox/dropbox_test.go

package dropbox

import (
	"context"
	"strings"
	"testing"

	"github.com/google/go-cmp/cmp"

	"github.com/trufflesecurity/trufflehog/v3/pkg/detectors"
	"github.com/trufflesecurity/trufflehog/v3/pkg/engine/ahocorasick"
)

var (
	validPattern = `
		# Configuration File: config.yaml
		database:
			host: $DB_HOST
			port: $DB_PORT
			username: $DB_USERNAME
			password: $DB_PASS  # IMPORTANT: Do not share this password publicly

		api:
			auth_type: "Bearer"
			in: "Header"
			api_version: v1
			dropbox_secret: "sl.4ihqlizKRm9J8tJvdBUecLPfYunjh3Nx73cUBGcRKpTFxRny3cYKdaQdzVF_rBIEO9emJaHyRWeM_tm5pYJFTc1TwYjM2fHlhSdhKkzHJjf5dx86fUlaO_eKY9r4ijZ8eD"
			base_url: "https://api.example.com/$api_version/example"
			response_code: 200

		# Notes:
		# - Remember to rotate the secret every 90 days.
		# - The above credentials should only be used in a secure environment.
	`
	secret = "sl.4ihqlizKRm9J8tJvdBUecLPfYunjh3Nx73cUBGcRKpTFxRny3cYKdaQdzVF_rBIEO9emJaHyRWeM_tm5pYJFTc1TwYjM2fHlhSdhKkzHJjf5dx86fUlaO_eKY9r4ijZ8eD"
)

func TestDropBox_Pattern(t *testing.T) {
	d := Scanner{}
	ahoCorasickCore := ahocorasick.NewAhoCorasickCore([]detectors.Detector{d})

	tests := []struct {
		name  string
		input string
		want  []string
	}{
		{
			name:  "valid pattern",
			input: validPattern,
			want:  []string{secret},
		},
	}

	for _, test := range tests {
		t.Run(test.name, func(t *testing.T) {
			matchedDetectors := ahoCorasickCore.FindDetectorMatches([]byte(test.input))
			if len(matchedDetectors) == 0 {
				t.Errorf("keywords '%v' not matched by: %s", d.Keywords(), test.input)
				return
			}

			results, err := d.FromData(context.Background(), false, []byte(test.input))
			if err != nil {
				t.Errorf("error = %v", err)
				return
			}

			if len(results) != len(test.want) {
				if len(results) == 0 {
					t.Errorf("did not receive result")
				} else {
					t.Errorf("expected %d results, only received %d", len(test.want), len(results))
				}
				return
			}

			actual := make(map[string]struct{}, len(results))
			for _, r := range results {
				if len(r.RawV2) > 0 {
					actual[string(r.RawV2)] = struct{}{}
				} else {
					actual[string(r.Raw)] = struct{}{}
				}
			}
			expected := make(map[string]struct{}, len(test.want))
			for _, v := range test.want {
				expected[v] = struct{}{}
			}

			if diff := cmp.Diff(expected, actual); diff != "" {
				t.Errorf("%s diff: (-want +got)\n%s", test.name, diff)
			}
		})
	}
}

// TestDropBox_LongTokenThroughEngineWindow guards the MaxSecretSize override. The scanning
// engine only passes a keyword-centered window of the chunk to FromData (512 bytes by
// default). Newer scoped Dropbox tokens (sl.u.…) can be ~1.5KB, so without MaxSecretSize the
// token is truncated before the regex sees it and verification fails. This exercises the full
// Aho-Corasick windowing path rather than calling FromData on the whole input directly.
func TestDropBox_LongTokenThroughEngineWindow(t *testing.T) {
	token := "sl.u." + strings.Repeat("aB3-_", 300) // 5 + 1500 = 1505 chars, single base64url run
	chunk := []byte("DROPBOX_TOKEN=" + token + "\n")

	d := Scanner{}
	core := ahocorasick.NewAhoCorasickCore([]detectors.Detector{d})
	matches := core.FindDetectorMatches(chunk)
	if len(matches) == 0 {
		t.Fatal("no detector matches for long token")
	}

	var found bool
	for _, m := range matches {
		for _, data := range m.Matches() {
			results, err := d.FromData(context.Background(), false, data)
			if err != nil {
				t.Fatal(err)
			}
			for _, r := range results {
				if string(r.Raw) == token {
					found = true
				}
			}
		}
	}
	if !found {
		t.Errorf("full %d-char token was not captured through the engine window (MaxSecretSize regression?)", len(token))
	}
}