feat: expose patch operation API

Author: ThomasRooney

model.go

@@ -118,8 +118,13 @@ type binaryPatch struct {
 	Content string
 }
 
-// fileDiff Source of truth: https://github.com/git/git/blob/master/diffcore.h#L106
-// Implemented in https://github.com/git/git/blob/master/diff.c#L3496
+// fileDiff models the subset of Git patch metadata exposed by this parser.
+//
+// Git represents generated diff pairs as diff_filepair:
+// https://github.com/git/git/blob/aec3f587505a472db67e9462d0702e7d463a449d/diffcore.h#L107-L130
+//
+// Git emits unified patch file headers in builtin_diff:
+// https://github.com/git/git/blob/aec3f587505a472db67e9462d0702e7d463a449d/diff.c#L3838-L3930
 type fileDiff struct {
 	FromFile           string        `json:"from_file"`
 	ToFile             string        `json:"to_file"`
@@ -138,6 +143,16 @@ type fileDiff struct {
 	CopyTo             string        `json:"copy_to,omitempty"`
 	Hunks              []hunk        `json:"hunks"`
 	BinaryPatch        []binaryPatch `json:"binary_patch"`
+
+	// Parser-only paths from the "---" and "+++" file header lines. Git apply
+	// validates these against the diff --git/copy/rename names in apply.c:
+	// https://github.com/git/git/blob/aec3f587505a472db67e9462d0702e7d463a449d/apply.c#L929-L966
+	// https://github.com/git/git/blob/aec3f587505a472db67e9462d0702e7d463a449d/apply.c#L1330-L1453
+	//
+	// They are intentionally unexported and omitted from JSON because they are
+	// input syntax used for validation, not semantic diff metadata.
+	oldFileHeaderPath string
+	newFileHeaderPath string
 }
 
 func (fd *fileDiff) GoString() string {

parity_test.go

@@ -128,6 +128,47 @@ func TestApplyFile_ParityCorpus(t *testing.T) {
 	}
 }
 
+func TestApplyPatchOperations_ParityCorpus(t *testing.T) {
+	if testing.Short() {
+		t.Skip("parity corpus is an integration test stream")
+	}
+
+	requireGitBinary(t)
+
+	cases := loadParityCases(t)
+	require.NotEmpty(t, cases)
+
+	for _, tc := range cases {
+		tc := tc
+		t.Run(tc.name, func(t *testing.T) {
+			t.Parallel()
+
+			if !supportsPatchOperationParity(tc.fixture) {
+				t.Skip("fixture exercises git apply behavior outside the content-tree operation API")
+			}
+
+			oracles := runGitApplyOracles(t, tc)
+			require.NoError(t, oracles.exitErr)
+
+			operations, err := ParsePatchOperations(tc.patch)
+			require.NoError(t, err)
+
+			applied, err := ApplyPatchOperations(contentMap(tc.srcTree), operations)
+			require.NoError(t, err)
+			assertContentMap(t, oracles.tree, applied)
+		})
+	}
+}
+
+func supportsPatchOperationParity(fixture parityFixture) bool {
+	return !fixture.ExpectConflict &&
+		!fixture.ExpectGitError &&
+		!fixture.IgnoreWhitespace &&
+		len(fixture.GitArgs) == 0 &&
+		len(fixture.SrcModes) == 0 &&
+		len(fixture.OutModes) == 0
+}
+
 func runLibraryApply(t *testing.T, tc parityCase, rejectMode bool) (applyResult, error) {
 	t.Helper()
 
@@ -305,6 +346,10 @@ func loadParityTree(t *testing.T, legacyPath string, files map[string]string, mo
 		return tree
 	}
 
+	if info, err := os.Stat(legacyPath); err == nil && info.IsDir() {
+		return collectFixtureTree(t, legacyPath)
+	}
+
 	legacy := readParityFileMaybe(t, legacyPath)
 	if legacy == nil {
 		return nil
@@ -314,6 +359,30 @@ func loadParityTree(t *testing.T, legacyPath string, files map[string]string, mo
 	}
 }
 
+func collectFixtureTree(t *testing.T, root string) parityTree {
+	t.Helper()
+
+	tree := make(parityTree)
+	require.NoError(t, filepath.WalkDir(root, func(path string, d fs.DirEntry, err error) error {
+		require.NoError(t, err)
+		if path == root || d.IsDir() {
+			return nil
+		}
+		rel, err := filepath.Rel(root, path)
+		require.NoError(t, err)
+		content, err := os.ReadFile(path)
+		require.NoError(t, err)
+		info, err := d.Info()
+		require.NoError(t, err)
+		tree[filepath.ToSlash(rel)] = parityFile{
+			content: content,
+			mode:    info.Mode().Perm(),
+		}
+		return nil
+	}))
+	return tree
+}
+
 func parseParityMode(raw string) fs.FileMode {
 	if raw == "" {
 		return 0
@@ -406,6 +475,29 @@ func assertParityTree(t *testing.T, want, got parityTree) {
 	}
 }
 
+func contentMap(tree parityTree) map[string][]byte {
+	content := make(map[string][]byte, len(tree))
+	for path, file := range tree {
+		content[path] = append([]byte(nil), file.content...)
+	}
+	return content
+}
+
+func assertContentMap(t *testing.T, want parityTree, got map[string][]byte) {
+	t.Helper()
+
+	require.Len(t, got, len(want))
+	for path, expected := range want {
+		actual, ok := got[path]
+		require.True(t, ok, "missing file %s", path)
+		assert.Equal(t, expected.content, actual, "content mismatch for %s", path)
+	}
+	for path := range got {
+		_, ok := want[path]
+		assert.True(t, ok, "unexpected file %s", path)
+	}
+}
+
 func requireGitBinary(t *testing.T) {
 	t.Helper()
 

parser.go

@@ -206,8 +206,12 @@ func (p *parser) tryVisitHeader(diff string) bool {
 		return false
 	}
 
-	if strings.HasPrefix(diff, "+++ ") || strings.HasPrefix(diff, "--- ") {
-		// ignore -- we're still in the FileDiff and we've already captured the file names
+	if strings.HasPrefix(diff, "--- ") {
+		p.diff.FileDiff[fileHEAD].oldFileHeaderPath = parseFileHeaderPath(diff, "--- ")
+		return true
+	}
+	if strings.HasPrefix(diff, "+++ ") {
+		p.diff.FileDiff[fileHEAD].newFileHeaderPath = parseFileHeaderPath(diff, "+++ ")
 		return true
 	}
 	if strings.HasPrefix(diff, "index ") {
@@ -416,6 +420,40 @@ func (p *parser) parseDiffLine(line string) fileDiff {
 	}
 }
 
+func parseFileHeaderPath(line, prefix string) string {
+	path := firstFileHeaderToken(strings.TrimPrefix(line, prefix))
+	if path == "/dev/null" {
+		return ""
+	}
+	if strings.HasPrefix(path, "a/") || strings.HasPrefix(path, "b/") {
+		return path[2:]
+	}
+	return path
+}
+
+func firstFileHeaderToken(header string) string {
+	if !strings.HasPrefix(header, `"`) {
+		if fields := strings.Fields(header); len(fields) > 0 {
+			return fields[0]
+		}
+		return header
+	}
+
+	var escaped bool
+	for i := 1; i < len(header); i++ {
+		switch {
+		case escaped:
+			escaped = false
+		case header[i] == '\\':
+			escaped = true
+		case header[i] == '"':
+			return header[1:i]
+		}
+	}
+
+	return strings.Trim(header, `"`)
+}
+
 func parsePercentValue(raw string) int {
 	raw = strings.TrimSuffix(raw, "%")
 	value, err := strconv.Atoi(raw)

patch_file_operations.go

@@ -0,0 +1,133 @@
+package git_diff_parser
+
+import "fmt"
+
+type PatchOperationType string
+
+const (
+	PatchOperationTypeModify     PatchOperationType = "modify"
+	PatchOperationTypeCreate     PatchOperationType = "create"
+	PatchOperationTypeDelete     PatchOperationType = "delete"
+	PatchOperationTypeRename     PatchOperationType = "rename"
+	PatchOperationTypeCopy       PatchOperationType = "copy"
+	PatchOperationTypeModeChange PatchOperationType = "mode_change"
+	PatchOperationTypeBinary     PatchOperationType = "binary"
+)
+
+// PatchOperation describes one file-level operation in a git patchset.
+//
+// Values returned by ParsePatchOperations can be passed to ApplyPatchOperations.
+type PatchOperation struct {
+	Type       PatchOperationType
+	SourcePath string
+	TargetPath string
+	OldMode    string
+	NewMode    string
+	IndexMode  string
+	IsBinary   bool
+	Patch      []byte
+
+	file *patchsetFile
+}
+
+// MutatesFileSet reports whether this operation adds, removes, or moves a file.
+func (op *PatchOperation) MutatesFileSet() bool {
+	switch op.Type {
+	case PatchOperationTypeCreate, PatchOperationTypeDelete, PatchOperationTypeRename, PatchOperationTypeCopy:
+		return true
+	default:
+		return false
+	}
+}
+
+// ParsePatchOperations parses patchData into ordered file-level operations.
+func ParsePatchOperations(patchData []byte) ([]PatchOperation, error) {
+	patchset, errs := parsePatchset(patchData)
+	if len(errs) > 0 {
+		return nil, fmt.Errorf("unsupported patch syntax: %w", errs[0])
+	}
+
+	operations := make([]PatchOperation, 0, len(patchset.Files))
+	for i := range patchset.Files {
+		operation, err := patchOperationFromFile(&patchset.Files[i])
+		if err != nil {
+			return nil, err
+		}
+		operations = append(operations, operation)
+	}
+
+	return operations, nil
+}
+
+// ApplyPatchOperations applies ordered patch operations to a copy of tree.
+func ApplyPatchOperations(tree map[string][]byte, operations []PatchOperation) (map[string][]byte, error) {
+	files, err := patchsetFilesFromOperations(operations)
+	if err != nil {
+		return nil, err
+	}
+	return applyPatchsetFiles(tree, files)
+}
+
+func patchsetFilesFromOperations(operations []PatchOperation) ([]*patchsetFile, error) {
+	files := make([]*patchsetFile, 0, len(operations))
+	for i := range operations {
+		file, err := operations[i].patchsetFile()
+		if err != nil {
+			return nil, err
+		}
+		files = append(files, file)
+	}
+
+	return files, nil
+}
+
+func patchOperationFromFile(file *patchsetFile) (PatchOperation, error) {
+	return PatchOperation{
+		Type:       publicPatchOperationType(file.Operation),
+		SourcePath: file.SourcePath,
+		TargetPath: file.TargetPath,
+		OldMode:    file.Diff.OldMode,
+		NewMode:    file.Diff.NewMode,
+		IndexMode:  file.Diff.IndexMode,
+		IsBinary:   file.Diff.IsBinary,
+		Patch:      append([]byte(nil), file.Patch...),
+		file:       file,
+	}, nil
+}
+
+func publicPatchOperationType(op patchsetOperation) PatchOperationType {
+	switch op {
+	case patchsetOperationCreate:
+		return PatchOperationTypeCreate
+	case patchsetOperationDelete:
+		return PatchOperationTypeDelete
+	case patchsetOperationRename:
+		return PatchOperationTypeRename
+	case patchsetOperationCopy:
+		return PatchOperationTypeCopy
+	case patchsetOperationModeChange:
+		return PatchOperationTypeModeChange
+	case patchsetOperationBinary:
+		return PatchOperationTypeBinary
+	default:
+		return PatchOperationTypeModify
+	}
+}
+
+func (op *PatchOperation) patchsetFile() (*patchsetFile, error) {
+	if op.file != nil {
+		return op.file, nil
+	}
+	if len(op.Patch) > 0 {
+		patchset, errs := parsePatchset(op.Patch)
+		if len(errs) > 0 {
+			return nil, fmt.Errorf("unsupported patch syntax: %w", errs[0])
+		}
+		if len(patchset.Files) != 1 {
+			return nil, fmt.Errorf("patch operation contains %d file diffs, expected 1", len(patchset.Files))
+		}
+		return &patchset.Files[0], nil
+	}
+
+	return nil, fmt.Errorf("patch operation has no patch data")
+}