Add base64 detectors. (#414)

pkg/decoders/base64.go

@@ -0,0 +1,78 @@
+package decoders
+
+import (
+	"bytes"
+	"encoding/base64"
+	"strings"
+
+	"github.com/trufflesecurity/trufflehog/v3/pkg/sources"
+)
+
+type Base64 struct{}
+
+var (
+	b64Charset  = []byte("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/=")
+	b64EndChars = "+/="
+)
+
+func getSubstringsOfCharacterSet(data []byte, charset []byte, threshold int) []string {
+	count := 0
+	substrings := []string{}
+	letters := strings.Builder{}
+	if len(data) == 0 {
+		return nil
+	}
+	for _, char := range string(data) {
+		if bytes.ContainsRune(charset, char) {
+			letters.WriteRune(char)
+			count++
+		} else {
+			if count > threshold {
+				substrings = appendB64Substring(letters, substrings)
+			}
+			letters.Reset()
+			count = 0
+		}
+	}
+
+	if count > threshold && len(letters.String()) > 0 {
+		substrings = appendB64Substring(letters, substrings)
+	}
+
+	return substrings
+}
+
+func appendB64Substring(letters strings.Builder, substrings []string) []string {
+
+	substring := strings.TrimLeft(letters.String(), b64EndChars)
+	// handle key=value
+	if strings.Contains(strings.TrimRight(substring, b64EndChars), "=") {
+		split := strings.SplitN(substring, "=", 2)
+		substrings = append(substrings, split[len(split)-1])
+	} else {
+		substrings = append(substrings, substring)
+	}
+	return substrings
+}
+
+func (d *Base64) FromChunk(chunk *sources.Chunk) *sources.Chunk {
+
+	encodedSubstrings := getSubstringsOfCharacterSet(chunk.Data, b64Charset, 20)
+	decodedSubstrings := map[string][]byte{}
+
+	for _, str := range encodedSubstrings {
+		dec, err := base64.StdEncoding.DecodeString(str)
+		if err == nil && len(dec) > 0 {
+			decodedSubstrings[str] = dec
+		}
+	}
+
+	if len(decodedSubstrings) > 0 {
+		for substring, dec := range decodedSubstrings {
+			chunk.Data = bytes.Replace(chunk.Data, []byte(substring), dec, 1)
+		}
+		return chunk
+	}
+
+	return nil
+}

pkg/decoders/base64_test.go

@@ -0,0 +1,102 @@
+package decoders
+
+import (
+	"testing"
+
+	"github.com/kylelemons/godebug/pretty"
+
+	"github.com/trufflesecurity/trufflehog/v3/pkg/detectors"
+	"github.com/trufflesecurity/trufflehog/v3/pkg/sources"
+)
+
+func TestBase64_FromChunk(t *testing.T) {
+	tests := []struct {
+		name  string
+		chunk *sources.Chunk
+		want  *sources.Chunk
+	}{
+		{
+			name: "only b64 chunk",
+			chunk: &sources.Chunk{
+				Data: []byte(`bG9uZ2VyLWVuY29kZWQtc2VjcmV0LXRlc3Q=`),
+			},
+			want: &sources.Chunk{
+				Data: []byte(`longer-encoded-secret-test`),
+			},
+		},
+		{
+			name: "mixed content",
+			chunk: &sources.Chunk{
+				Data: []byte(`token: bG9uZ2VyLWVuY29kZWQtc2VjcmV0LXRlc3Q=`),
+			},
+			want: &sources.Chunk{
+				Data: []byte(`token: longer-encoded-secret-test`),
+			},
+		},
+		{
+			name: "no chunk",
+			chunk: &sources.Chunk{
+				Data: []byte(``),
+			},
+			want: nil,
+		},
+		{
+			name: "env var (looks like all b64 decodable but has `=` in the middle)",
+			chunk: &sources.Chunk{
+				Data: []byte(`some-encoded-secret=dGVzdHNlY3JldA==`),
+			},
+			want: &sources.Chunk{
+				Data: []byte(`some-encoded-secret=testsecret`),
+			},
+		},
+		{
+			name: "has longer b64 inside",
+			chunk: &sources.Chunk{
+				Data: []byte(`some-encoded-secret="bG9uZ2VyLWVuY29kZWQtc2VjcmV0LXRlc3Q="`),
+			},
+			want: &sources.Chunk{
+				Data: []byte(`some-encoded-secret="longer-encoded-secret-test"`),
+			},
+		},
+		{
+			name: "many possible substrings",
+			chunk: &sources.Chunk{
+				Data: []byte(`Many substrings in this slack message could be base64 decoded
+				but only dGhpcyBlbmNhcHN1bGF0ZWQgc2VjcmV0 should be decoded.`),
+			},
+			want: &sources.Chunk{
+				Data: []byte(`Many substrings in this slack message could be base64 decoded
+				but only this encapsulated secret should be decoded.`),
+			},
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			d := &Base64{}
+			got := d.FromChunk(tt.chunk)
+			if tt.want != nil {
+				if got == nil {
+					t.Fatal("got nil, did not want nil")
+				}
+				if diff := pretty.Compare(string(got.Data), string(tt.want.Data)); diff != "" {
+					t.Errorf("Base64FromChunk() %s diff: (-got +want)\n%s", tt.name, diff)
+				}
+			} else {
+				if got != nil {
+					t.Error("Expected nil chunk")
+				}
+			}
+		})
+	}
+}
+
+func BenchmarkFromChunk(benchmark *testing.B) {
+	d := Base64{}
+	for name, data := range detectors.MustGetBenchmarkData() {
+		benchmark.Run(name, func(b *testing.B) {
+			for n := 0; n < b.N; n++ {
+				d.FromChunk(&sources.Chunk{Data: data})
+			}
+		})
+	}
+}

pkg/decoders/decoders.go

@@ -7,6 +7,7 @@ import (
 func DefaultDecoders() []Decoder {
 	return []Decoder{
 		&Plain{},
+		&Base64{},
 	}
 }