pkg/vcs: extend ParseGitDiff

Return not just the modified files, but also their blob hashes.

3 files changed, 49 insertions, 12 deletions

diff --git a/pkg/manager/diff.go b/pkg/manager/diff.go
index 4ad759244..88710a59b 100644
--- a/pkg/manager/diff.go
+++ b/pkg/manager/diff.go
@@ -870,7 +870,9 @@ func affectedFiles(cfg *mgrconfig.Config, gitPatches [][]byte) (direct, transiti
 	transitiveMap := make(map[string]struct{})
 	var allFiles []string
 	for _, patch := range gitPatches {
-		allFiles = append(allFiles, vcs.ParseGitDiff(patch)...)
+		for _, diff := range vcs.ParseGitDiff(patch) {
+			allFiles = append(allFiles, diff.Name)
+		}
 	}
 	for _, file := range allFiles {
 		directMap[file] = struct{}{}
diff --git a/pkg/vcs/git.go b/pkg/vcs/git.go
index 72b4a32e6..a2975c034 100644
--- a/pkg/vcs/git.go
+++ b/pkg/vcs/git.go
@@ -578,15 +578,33 @@ func (git *gitRepo) PushCommit(repo, commit string) error {
 	return nil
 }
 
-var fileNameRe = regexp.MustCompile(`(?m)^diff.* b\/([^\s]+)`)
+var (
+	fileNameRe = regexp.MustCompile(`^diff.* b\/([^\s]+)$`)
+	indexRe    = regexp.MustCompile(`^index (\w+)\.\.(?:\w+)(?:\s\d+)?$`)
+)
+
+type ModifiedFile struct {
+	Name     string
+	LeftHash string
+}
 
 // ParseGitDiff extracts the files modified in the git patch.
-func ParseGitDiff(patch []byte) []string {
-	var files []string
-	for _, match := range fileNameRe.FindAllStringSubmatch(string(patch), -1) {
-		files = append(files, match[1])
+func ParseGitDiff(patch []byte) []ModifiedFile {
+	var ret []ModifiedFile
+	scanner := bufio.NewScanner(bytes.NewReader(patch))
+	for scanner.Scan() {
+		line := scanner.Text()
+		indexMatch := indexRe.FindStringSubmatch(line)
+		if indexMatch != nil && len(ret) > 0 {
+			ret[len(ret)-1].LeftHash = indexMatch[1]
+		} else {
+			fileNameMatch := fileNameRe.FindStringSubmatch(line)
+			if fileNameMatch != nil {
+				ret = append(ret, ModifiedFile{Name: fileNameMatch[1]})
+			}
+		}
 	}
-	return files
+	return ret
 }
 
 type Git struct {
diff --git a/pkg/vcs/git_test.go b/pkg/vcs/git_test.go
index 40207898c..f2d761713 100644
--- a/pkg/vcs/git_test.go
+++ b/pkg/vcs/git_test.go
@@ -455,7 +455,7 @@ func TestGitFetchShortHash(t *testing.T) {
 }
 
 func TestParseGitDiff(t *testing.T) {
-	files := ParseGitDiff([]byte(`diff --git a/a.txt b/a.txt
+	list := ParseGitDiff([]byte(`diff --git a/a.txt b/a.txt
 index 4c5fd91..8fe1e32 100644
 --- a/a.txt
 +++ b/a.txt
@@ -469,9 +469,26 @@ index 0000000..f8a9677
 +++ b/b.txt
 @@ -0,0 +1 @@
 +Second file.
-diff --git a/c/c.txt b/c/c.txt
-new file mode 100644
-index 0000000..e69de29
+diff --git a/c.txt b/c.txt
+deleted file mode 100644
+index f70f10e..0000000
+--- a/c.txt
++++ /dev/null
+@@ -1 +0,0 @@
+-A
 `))
-	assert.ElementsMatch(t, files, []string{"a.txt", "b.txt", "c/c.txt"})
+	assert.Equal(t, list, []ModifiedFile{
+		{
+			Name:     `a.txt`,
+			LeftHash: `4c5fd91`,
+		},
+		{
+			Name:     `b.txt`,
+			LeftHash: `0000000`,
+		},
+		{
+			Name:     `c.txt`,
+			LeftHash: `f70f10e`,
+		},
+	})
 }