feat(gitparse): track commit refs

Author: rgmz

hack/snifftest/main.go

@@ -197,15 +197,16 @@ func main() {
 					SkipBinaries: true,
 					SkipArchives: false,
 					Concurrency:  runtime.NumCPU(),
-					SourceMetadataFunc: func(file, email, commit, timestamp, repository string, line int64) *source_metadatapb.MetaData {
+					SourceMetadataFunc: func(repository, commit, ref, email, timestamp, file string, line int64) *source_metadatapb.MetaData {
 						return &source_metadatapb.MetaData{
 							Data: &source_metadatapb.MetaData_Git{
 								Git: &source_metadatapb.Git{
+									Repository: repository,
 									Commit:     commit,
-									File:       file,
+									CommitRef:  ref,
 									Email:      email,
-									Repository: repository,
 									Timestamp:  timestamp,
+									File:       file,
 								},
 							},
 						}

pkg/gitparse/gitparse.go

@@ -109,6 +109,7 @@ func (d *Diff) finalize() error { return d.contentWriter.CloseForWriting() }
 // Commit contains commit header info and diffs.
 type Commit struct {
 	Hash      string
+	SourceRef string
 	Author    string
 	Committer string
 	Date      time.Time
@@ -230,13 +231,16 @@ func (c *Parser) RepoPath(ctx context.Context, source string, head string, abbre
 		"--date=format:%a %b %d %H:%M:%S %Y %z",
 		"--pretty=fuller", // https://git-scm.com/docs/git-log#_pretty_formats
 		"--notes",         // https://git-scm.com/docs/git-log#Documentation/git-log.txt---notesltrefgt
+		"--source",        // https://git-scm.com/docs/git-log#Documentation/git-log.txt---source
 	}
 	if abbreviatedLog {
+		// https://git-scm.com/docs/git-log#Documentation/git-log.txt---diff-filterACDMRTUXB82308203
 		args = append(args, "--diff-filter=AM")
 	}
 	if head != "" {
 		args = append(args, head)
 	} else {
+		// https://git-scm.com/docs/git-log#Documentation/git-log.txt---all
 		args = append(args, "--all")
 	}
 	for _, glob := range excludedGlobs {
@@ -323,10 +327,9 @@ func (c *Parser) FromReader(ctx context.Context, stdOut io.Reader, diffChan chan
 	outReader := bufio.NewReader(stdOut)
 	var (
 		currentCommit *Commit
-
-		totalLogSize int
+		totalLogSize  int
+		latestState   = Initial
 	)
-	var latestState = Initial
 
 	diff := func(c *Commit, opts ...diffOption) *Diff {
 		opts = append(opts, withCustomContentWriter(bufferwriter.New()))
@@ -386,10 +389,18 @@ func (c *Parser) FromReader(ctx context.Context, stdOut io.Reader, diffChan chan
 			// Create a new currentDiff and currentCommit
 			currentCommit = &Commit{Message: strings.Builder{}}
 			currentDiff = diff(currentCommit)
-			// Check that the commit line contains a hash and set it.
-			if len(line) >= 47 {
-				currentCommit.Hash = string(line[7:47])
+
+			hash, ref := parseCommitLine(line)
+			if hash == nil || ref == nil {
+				ctx.Logger().Error(
+					fmt.Errorf(`expected line to match 'commit <hash> <ref>', got "%s"`, line),
+					"Failed to parse CommitLine")
+				latestState = ParseFailure
+				continue
 			}
+
+			currentCommit.Hash = string(hash)
+			currentCommit.SourceRef = string(ref)
 		case isMergeLine(isStaged, latestState, line):
 			latestState = MergeLine
 		case isAuthorLine(isStaged, latestState, line):
@@ -605,6 +616,22 @@ func isCommitLine(isStaged bool, latestState ParseState, line []byte) bool {
 	return false
 }
 
+func parseCommitLine(line []byte) (hash []byte, ref []byte) {
+	// Check that the commit line contains a 40-character hash and set it.
+	// `commit e5575cd6f2d21d3a1a604287c7bf4a7eab2266e0\n`
+	if len(line) >= 47 {
+		hash = line[7:47]
+	}
+
+	// Check if the commit line includes branch references.
+	// `commit 2dbbb28727c7c2954438666dafba57bb8c714d3b refs/heads/fix/github-enterprise-gist\n`
+	if len(line) > 48 {
+		ref = line[48 : len(line)-1]
+	}
+
+	return
+}
+
 // Author: Bill Rich <[email protected]>
 func isAuthorLine(isStaged bool, latestState ParseState, line []byte) bool {
 	if isStaged || !(latestState == CommitLine || latestState == MergeLine) {

pkg/gitparse/gitparse_test.go

@@ -703,6 +703,23 @@ func TestLineChecksNoStaged(t *testing.T) {
 	}
 }
 
+func Test_parseCommitLine(t *testing.T) {
+	cases := map[string][]string{
+		"commit 198c63cb8212a99cc4352bc72f25e5444a786291 refs/heads/main\n":                       {"198c63cb8212a99cc4352bc72f25e5444a786291", "refs/heads/main"},
+		"commit e76dfb98ab9001daa869191b6aebe8cf4cd3b22a refs/remotes/origin/debug/aws-logging\n": {"e76dfb98ab9001daa869191b6aebe8cf4cd3b22a", "refs/remotes/origin/debug/aws-logging"},
+	}
+
+	for line, expected := range cases {
+		hash, ref := parseCommitLine([]byte(line))
+		if string(hash) != expected[0] {
+			t.Errorf("Expected: %s, Got: %s", expected[0], hash)
+		}
+		if string(ref) != expected[1] {
+			t.Errorf("Expected: %s, Got: %s", expected[1], ref)
+		}
+	}
+}
+
 func TestBinaryPathParse(t *testing.T) {
 	cases := map[string]string{
 		"Binary files a/trufflehog_3.42.0_linux_arm64.tar.gz and /dev/null differ\n":                                                                                         "",

pkg/output/plain.go

@@ -87,6 +87,33 @@ func (p *PlainPrinter) Print(_ context.Context, r *detectors.ResultWithMetadata)
 
 	for _, data := range meta {
 		for k, v := range data {
+			// Only print Git commit refs when they're associated with a pull/merge request.
+			// Otherwise, this information is not useful.
+			if strings.EqualFold(k, "Commit_ref") {
+				ref, ok := v.(string)
+				if !ok {
+					continue
+				}
+
+				var prNum string
+				if strings.HasPrefix(ref, "refs/heads/trufflehog/pull/") {
+					prNum = ref[27:]
+				} else if strings.HasPrefix(ref, "refs/heads/trufflehog/merge-requests/") {
+					prNum = ref[37:]
+				}
+
+				if prNum == "" {
+					continue
+				}
+
+				k = "Pull Request"
+				if strings.HasSuffix(prNum, "/head") {
+					v = prNum[:len(prNum)-5]
+				} else if strings.HasSuffix(prNum, "/merge") {
+					v = prNum[:len(prNum)-6]
+				}
+			}
+
 			aggregateDataKeys = append(aggregateDataKeys, k)
 			aggregateData[k] = v
 		}