feat: [PIPE-22535]: Include previous commit in blame (#2812)

git/api/blame.go

@@ -27,6 +27,8 @@ import (
 	"github.com/harness/gitness/errors"
 	"github.com/harness/gitness/git/command"
 	"github.com/harness/gitness/git/sha"
+
+	"github.com/gotidy/ptr"
 )
 
 var (
@@ -37,8 +39,14 @@ var (
 )
 
 type BlamePart struct {
-	Commit *Commit  `json:"commit"`
+	Commit   *Commit            `json:"commit"`
-	Lines  []string `json:"lines"`
+	Lines    []string           `json:"lines"`
+	Previous *BlamePartPrevious `json:"previous,omitempty"`
+}
+
+type BlamePartPrevious struct {
+	CommitSHA sha.SHA `json:"commit_sha"`
+	FileName  string  `json:"file_name"`
 }
 
 type BlameNextReader interface {
@@ -92,17 +100,22 @@ func (g *Git) Blame(
 	}()
 
 	return &BlameReader{
-		scanner:     bufio.NewScanner(pipeRead),
+		scanner:   bufio.NewScanner(pipeRead),
-		commitCache: make(map[string]*Commit),
+		cache:     make(map[string]blameReaderCacheItem),
-		errReader:   stderr, // Any stderr output will cause the BlameReader to fail.
+		errReader: stderr, // Any stderr output will cause the BlameReader to fail.
 	}
 }
 
+type blameReaderCacheItem struct {
+	commit   *Commit
+	previous *BlamePartPrevious
+}
+
 type BlameReader struct {
-	scanner     *bufio.Scanner
+	scanner   *bufio.Scanner
-	lastLine    string
+	lastLine  string
-	commitCache map[string]*Commit
+	cache     map[string]blameReaderCacheItem
-	errReader   io.Reader
+	errReader io.Reader
 }
 
 func (r *BlameReader) nextLine() (string, error) {
@@ -132,6 +145,7 @@ func (r *BlameReader) unreadLine(line string) {
 //nolint:complexity,gocognit,nestif // it's ok
 func (r *BlameReader) NextPart() (*BlamePart, error) {
 	var commit *Commit
+	var previous *BlamePartPrevious
 	var lines []string
 	var err error
 
@@ -146,8 +160,10 @@ func (r *BlameReader) NextPart() (*BlamePart, error) {
 			commitSHA := sha.Must(matches[1])
 
 			if commit == nil {
-				commit = r.commitCache[commitSHA.String()]
+				if cacheItem, ok := r.cache[commitSHA.String()]; ok {
-				if commit == nil {
+					commit = cacheItem.commit
+					previous = cacheItem.previous
+				} else {
 					commit = &Commit{SHA: commitSHA}
 				}
 
@@ -164,11 +180,15 @@ func (r *BlameReader) NextPart() (*BlamePart, error) {
 
 			if !commit.SHA.Equal(commitSHA) {
 				r.unreadLine(line)
-				r.commitCache[commit.SHA.String()] = commit
+				r.cache[commit.SHA.String()] = blameReaderCacheItem{
+					commit:   commit,
+					previous: previous,
+				}
 
 				return &BlamePart{
-					Commit: commit,
+					Commit:   commit,
-					Lines:  lines,
+					Lines:    lines,
+					Previous: previous,
 				}, nil
 			}
 
@@ -187,7 +207,7 @@ func (r *BlameReader) NextPart() (*BlamePart, error) {
 			continue
 		}
 
-		parseBlameHeaders(line, commit)
+		parseBlameHeaders(line, commit, &previous)
 	}
 
 	// Check if there's something in the error buffer... If yes, that's the error!
@@ -223,15 +243,16 @@ func (r *BlameReader) NextPart() (*BlamePart, error) {
 
 	if commit != nil && len(lines) > 0 {
 		part = &BlamePart{
-			Commit: commit,
+			Commit:   commit,
-			Lines:  lines,
+			Lines:    lines,
+			Previous: previous,
 		}
 	}
 
 	return part, err
 }
 
-func parseBlameHeaders(line string, commit *Commit) {
+func parseBlameHeaders(line string, commit *Commit, previous **BlamePartPrevious) {
 	// This is the list of git blame headers that we process. Other headers we ignore.
 	const (
 		headerSummary       = "summary "
@@ -241,6 +262,7 @@ func parseBlameHeaders(line string, commit *Commit) {
 		headerCommitterName = "committer "
 		headerCommitterMail = "committer-mail "
 		headerCommitterTime = "committer-time "
+		headerPrevious      = "previous "
 	)
 
 	switch {
@@ -258,6 +280,8 @@ func parseBlameHeaders(line string, commit *Commit) {
 		commit.Committer.Identity.Email = extractEmail(line[len(headerCommitterMail):])
 	case strings.HasPrefix(line, headerCommitterTime):
 		commit.Committer.When = extractTime(line[len(headerCommitterTime):])
+	case strings.HasPrefix(line, headerPrevious):
+		*previous = ptr.Of(extractPrevious(line[len(headerPrevious):]))
 	}
 }
 
@@ -265,6 +289,19 @@ func extractName(s string) string {
 	return s
 }
 
+// extractPrevious extracts the sha and filename of the previous commit.
+// example: previous 999d2ed306a916423d18e022abe258e92419ab9a README.md
+func extractPrevious(s string) BlamePartPrevious {
+	rawSHA, fileName, _ := strings.Cut(s, " ")
+	if len(fileName) > 0 && fileName[0] == '"' {
+		fileName, _ = strconv.Unquote(fileName)
+	}
+	return BlamePartPrevious{
+		CommitSHA: sha.Must(rawSHA),
+		FileName:  fileName,
+	}
+}
+
 // extractEmail extracts email from git blame output.
 // The email address is wrapped between "<" and ">" characters.
 // If "<" or ">" are not in place it returns the string as it.

git/api/blame_test.go

@@ -41,6 +41,7 @@ committer-mail <noreply@harness.io>
 committer-time 1669812989
 committer-tz +0100
 summary Pull request 1
+previous ec84ae5018520efdead481c81a31950b82196ec6 "\"\\n\\123\342\210\206'ex' \\\\t.\\ttxt\""
 filename file_name_before_rename.go
 	Line 10
 16f267ad4f731af1b2e36f42e170ed8921377398 12 11 1
@@ -55,7 +56,7 @@ committer-mail <noreply@harness.io>
 committer-time 1673952128
 committer-tz +0100
 summary Pull request 2
-previous 6561a7b86e1a5e74ea0e4e73ccdfc18b486a2826 file_name.go
+previous 6561a7b86e1a5e74ea0e4e73ccdfc18b486a2826  file_name.go 
 filename file_name.go
 	Line 12
 16f267ad4f731af1b2e36f42e170ed8921377398 13 13 2
@@ -86,6 +87,10 @@ filename file_name.go
 			When:     time.Unix(1669812989, 0),
 		},
 	}
+	previous1 := &BlamePartPrevious{
+		CommitSHA: sha.Must("ec84ae5018520efdead481c81a31950b82196ec6"),
+		FileName:  `"\n\123∆'ex' \\t.\ttxt"`,
+	}
 
 	commit2 := &Commit{
 		SHA:     sha.Must("dcb4b6b63e86f06ed4e4c52fbc825545dc0b6200"),
@@ -100,26 +105,33 @@ filename file_name.go
 			When:     time.Unix(1673952128, 0),
 		},
 	}
+	previous2 := &BlamePartPrevious{
+		CommitSHA: sha.Must("6561a7b86e1a5e74ea0e4e73ccdfc18b486a2826"),
+		FileName:  " file_name.go ",
+	}
 
 	want := []*BlamePart{
 		{
-			Commit: commit1,
+			Commit:   commit1,
-			Lines:  []string{"Line 10", "Line 11"},
+			Lines:    []string{"Line 10", "Line 11"},
+			Previous: previous1,
 		},
 		{
-			Commit: commit2,
+			Commit:   commit2,
-			Lines:  []string{"Line 12"},
+			Lines:    []string{"Line 12"},
+			Previous: previous2,
 		},
 		{
-			Commit: commit1,
+			Commit:   commit1,
-			Lines:  []string{"Line 13", "Line 14"},
+			Lines:    []string{"Line 13", "Line 14"},
+			Previous: previous1,
 		},
 	}
 
 	reader := BlameReader{
-		scanner:     bufio.NewScanner(strings.NewReader(blameOut)),
+		scanner:   bufio.NewScanner(strings.NewReader(blameOut)),
-		commitCache: make(map[string]*Commit),
+		cache:     make(map[string]blameReaderCacheItem),
-		errReader:   strings.NewReader(""),
+		errReader: strings.NewReader(""),
 	}
 
 	var got []*BlamePart
@@ -144,9 +156,9 @@ filename file_name.go
 
 func TestBlameReader_NextPart_UserError(t *testing.T) {
 	reader := BlameReader{
-		scanner:     bufio.NewScanner(strings.NewReader("")),
+		scanner:   bufio.NewScanner(strings.NewReader("")),
-		commitCache: make(map[string]*Commit),
+		cache:     make(map[string]blameReaderCacheItem),
-		errReader:   strings.NewReader("fatal: no such path\n"),
+		errReader: strings.NewReader("fatal: no such path\n"),
 	}
 
 	_, err := reader.NextPart()
@@ -157,9 +169,9 @@ func TestBlameReader_NextPart_UserError(t *testing.T) {
 
 func TestBlameReader_NextPart_CmdError(t *testing.T) {
 	reader := BlameReader{
-		scanner:     bufio.NewScanner(iotest.ErrReader(errors.New("dummy error"))),
+		scanner:   bufio.NewScanner(iotest.ErrReader(errors.New("dummy error"))),
-		commitCache: make(map[string]*Commit),
+		cache:     make(map[string]blameReaderCacheItem),
-		errReader:   strings.NewReader(""),
+		errReader: strings.NewReader(""),
 	}
 
 	_, err := reader.NextPart()

git/blame.go

@@ -20,6 +20,7 @@ import (
 	"io"
 
 	"github.com/harness/gitness/errors"
+	"github.com/harness/gitness/git/sha"
 )
 
 type BlameParams struct {
@@ -65,8 +66,14 @@ func (params *BlameParams) Validate() error {
 }
 
 type BlamePart struct {
-	Commit *Commit  `json:"commit"`
+	Commit   *Commit            `json:"commit"`
-	Lines  []string `json:"lines"`
+	Lines    []string           `json:"lines"`
+	Previous *BlamePartPrevious `json:"previous,omitempty"`
+}
+
+type BlamePartPrevious struct {
+	CommitSHA sha.SHA `json:"commit_sha"`
+	FileName  string  `json:"file_name"`
 }
 
 // Blame processes and streams the git blame output data.
@@ -94,7 +101,6 @@ func (s *Service) Blame(ctx context.Context, params *BlameParams) (<-chan *Blame
 
 		for {
 			part, errRead := reader.NextPart()
-
 			if part == nil {
 				return
 			}
@@ -108,7 +114,18 @@ func (s *Service) Blame(ctx context.Context, params *BlameParams) (<-chan *Blame
 			lines := make([]string, len(part.Lines))
 			copy(lines, part.Lines)
 
-			ch <- &BlamePart{Commit: commit, Lines: lines}
+			next := &BlamePart{
+				Commit: commit,
+				Lines:  lines,
+			}
+			if part.Previous != nil {
+				next.Previous = &BlamePartPrevious{
+					CommitSHA: part.Previous.CommitSHA,
+					FileName:  part.Previous.FileName,
+				}
+			}
+
+			ch <- next
 
 			if errRead != nil && errors.Is(errRead, io.EOF) {
 				return