improve implementation of git FS (#1229)

git/api/fs.go

@@ -15,10 +15,12 @@
 package api
 
 import (
+	"bytes"
 	"context"
 	"fmt"
 	"io"
 	"io/fs"
+	"strings"
 	"time"
 
 	"github.com/harness/gitness/errors"
@@ -29,13 +31,23 @@ import (
 )
 
 // FS represents a git filesystem.
+// It implements fs.FS interface.
 type FS struct {
 	ctx context.Context
 	rev string
 	dir string
 }
 
-// NewFS creates a new git filesystem.
+var (
+	// Make sure all targeted interfaces are implemented.
+	_ fs.FS         = (*FS)(nil)
+	_ fs.ReadFileFS = (*FS)(nil)
+	_ fs.ReadDirFS  = (*FS)(nil)
+	_ fs.GlobFS     = (*FS)(nil)
+	_ fs.StatFS     = (*FS)(nil)
+)
+
+// NewFS creates a new git file system for the provided revision.
 func NewFS(ctx context.Context, rev, dir string) *FS {
 	return &FS{
 		ctx: ctx,
@@ -44,37 +56,57 @@ func NewFS(ctx context.Context, rev, dir string) *FS {
 	}
 }
 
+// Open opens a file.
+// It is part of the fs.FS interface.
 func (f *FS) Open(path string) (fs.File, error) {
+	if path != "" && !fs.ValidPath(path) {
+		return nil, &fs.PathError{
+			Op:   "open",
+			Path: path,
+			Err:  fs.ErrInvalid,
+		}
+	}
+
 	treeNode, err := GetTreeNode(f.ctx, f.dir, f.rev, path, true)
 	if err != nil {
-		return nil, err
+		if errors.IsNotFound(err) {
+			return nil, &fs.PathError{
+				Op:   "open",
+				Path: path,
+				Err:  fs.ErrNotExist,
+			}
+		}
+
+		return nil, &fs.PathError{
+			Op:   "open",
+			Path: path,
+			Err:  err,
+		}
 	}
 
-	if treeNode.IsDir() {
+	switch {
-		return nil, errors.InvalidArgument("can't open a directory")
+	case treeNode.IsDir():
-	}
+		return f.openTree(path, treeNode), nil
-	if treeNode.IsLink() {
+	case treeNode.IsSubmodule():
-		return nil, errors.InvalidArgument("can't open a link")
+		return f.openSubmodule(path, treeNode), nil
-	}
+	default:
-	if treeNode.IsSubmodule() {
+		return f.openBlob(path, treeNode), nil
-		return nil, errors.InvalidArgument("can't open a submodule")
 	}
+}
 
+func (f *FS) openBlob(path string, treeNode *TreeNode) *fsFile {
 	pipeRead, pipeWrite := io.Pipe()
 	ctx, cancelFn := context.WithCancel(f.ctx)
 	go func() {
-		var err error
-
-		defer func() {
-			// If running of the command below fails, make the pipe reader also fail with the same error.
-			_ = pipeWrite.CloseWithError(err)
-		}()
-
 		cmd := command.New("cat-file", command.WithFlag("-p"), command.WithArg(treeNode.SHA.String()))
-		err = cmd.Run(f.ctx, command.WithDir(f.dir), command.WithStdout(pipeWrite))
+		_ = pipeWrite.CloseWithError(cmd.Run(ctx, command.WithDir(f.dir), command.WithStdout(pipeWrite)))
+	}()
+	go func() {
+		<-ctx.Done()
+		_ = pipeWrite.CloseWithError(ctx.Err())
 	}()
 
-	pathFile := &file{
+	return &fsFile{
 		ctx:      ctx,
 		cancelFn: cancelFn,
 		path:     path,
@@ -83,8 +115,50 @@ func (f *FS) Open(path string) (fs.File, error) {
 		size:     treeNode.Size,
 		reader:   pipeRead,
 	}
+}
 
-	return pathFile, nil
+func (f *FS) openSubmodule(path string, treeNode *TreeNode) *fsFile {
+	content := treeNode.SHA.String() + "\n" // content of a submodule is the commit SHA plus end-of-line character.
+	return &fsFile{
+		ctx:      context.Background(),
+		cancelFn: func() {},
+		path:     path,
+		blobSHA:  treeNode.SHA,
+		mode:     treeNode.Mode,
+		size:     int64(len(content)),
+		reader:   strings.NewReader(content),
+	}
+}
+
+func (f *FS) openTree(path string, treeNode *TreeNode) *fsDir {
+	return &fsDir{
+		ctx:     context.Background(),
+		path:    path,
+		treeSHA: treeNode.SHA,
+		dir:     f.dir,
+		skip:    0,
+	}
+}
+
+// ReadFile reads the whole file.
+// It is part of the fs.ReadFileFS interface.
+func (f *FS) ReadFile(path string) ([]byte, error) {
+	file, err := f.Open(path)
+	if err != nil {
+		return nil, err
+	}
+
+	defer func() {
+		_ = file.Close()
+	}()
+
+	buffer := bytes.NewBuffer(nil)
+	_, err = io.Copy(buffer, file)
+	if err != nil {
+		return nil, err
+	}
+
+	return buffer.Bytes(), nil
 }
 
 // ReadDir returns all entries for a directory.
@@ -97,7 +171,7 @@ func (f *FS) ReadDir(name string) ([]fs.DirEntry, error) {
 
 	result := make([]fs.DirEntry, len(treeNodes))
 	for i, treeNode := range treeNodes {
-		result[i] = entry{treeNode}
+		result[i] = fsEntry{treeNode}
 	}
 
 	return result, nil
@@ -117,10 +191,93 @@ func (f *FS) Stat(name string) (fs.FileInfo, error) {
 		return nil, fmt.Errorf("failed to read git directory: %w", err)
 	}
 
-	return entry{*treeInfo}, nil
+	return fsEntry{*treeInfo}, nil
 }
 
-type file struct {
+type fsDir struct {
+	ctx context.Context
+
+	path    string
+	treeSHA sha.SHA
+
+	dir  string
+	skip int
+}
+
+// Stat returns fs.FileInfo for the directory.
+// It is part of the fs.File interface.
+func (d *fsDir) Stat() (fs.FileInfo, error) { return d, nil }
+
+// Read always returns an error because it is not possible to read directory bytes.
+// It is part of the fs.File interface.
+func (d *fsDir) Read([]byte) (int, error) {
+	return 0, &fs.PathError{
+		Op:   "read",
+		Path: d.path,
+		Err:  fs.ErrInvalid,
+	}
+}
+
+// Close in a no-op for directories.
+// It is part of the fs.File interface.
+func (d *fsDir) Close() error { return nil }
+
+// ReadDir lists entries in the directory. The integer parameter can be used for pagination.
+// It is part of the fs.ReadDirFile interface.
+func (d *fsDir) ReadDir(n int) ([]fs.DirEntry, error) {
+	treeNodes, err := ListTreeNodes(d.ctx, d.dir, d.treeSHA.String(), "", true)
+	if err != nil {
+		return nil, fmt.Errorf("failed to read git directory: %w", err)
+	}
+
+	if d.skip >= len(treeNodes) {
+		treeNodes = treeNodes[len(treeNodes):]
+	} else {
+		treeNodes = treeNodes[d.skip:]
+	}
+
+	var result []fs.DirEntry
+	for _, treeNode := range treeNodes {
+		result = append(result, fsEntry{treeNode})
+		d.skip++
+
+		if n >= 1 && n == len(result) {
+			break
+		}
+	}
+
+	if len(result) == 0 {
+		return nil, io.EOF
+	}
+
+	return result, nil
+}
+
+// Name returns the path.
+// It is part of the fs.FileInfo interface.
+func (d *fsDir) Name() string { return d.path }
+
+// Size implementation for directories returns zero.
+// It is part of the fs.FileInfo interface.
+func (d *fsDir) Size() int64 { return 0 }
+
+// Mode always returns fs.ModeDir because a git tree is a directory.
+// It is part of the fs.FileInfo interface.
+func (d *fsDir) Mode() fs.FileMode { return fs.ModeDir }
+
+// ModTime is unimplemented.
+// It is part of the fs.FileInfo interface.
+func (d *fsDir) ModTime() time.Time { return time.Time{} }
+
+// IsDir implementation always returns true.
+// It is part of the fs.FileInfo interface.
+func (d *fsDir) IsDir() bool { return true }
+
+// Sys is unimplemented.
+// It is part of the fs.FileInfo interface.
+func (d *fsDir) Sys() any { return nil }
+
+type fsFile struct {
 	ctx      context.Context
 	cancelFn context.CancelFunc
 
@@ -132,109 +289,109 @@ type file struct {
 	reader io.Reader
 }
 
-func (f *file) Stat() (fs.FileInfo, error) {
+// Stat returns fs.FileInfo for the file.
-	return f, nil
+// It is part of the fs.File interface.
-}
+func (f *fsFile) Stat() (fs.FileInfo, error) { return f, nil }
 
 // Read bytes from the file.
-func (f *file) Read(bytes []byte) (int, error) {
+// It is part of the fs.File interface.
-	return f.reader.Read(bytes)
+func (f *fsFile) Read(bytes []byte) (int, error) { return f.reader.Read(bytes) }
-}
 
 // Close closes the file.
-func (f *file) Close() error {
+// It is part of the fs.File interface.
+func (f *fsFile) Close() error {
 	f.cancelFn()
 	return nil
 }
 
 // Name returns the name of the file.
 // It is part of the fs.FileInfo interface.
-func (f *file) Name() string {
+func (f *fsFile) Name() string { return f.path }
-	return f.path
-}
 
 // Size returns file size - the size of the git blob object.
 // It is part of the fs.FileInfo interface.
-func (f *file) Size() int64 {
+func (f *fsFile) Size() int64 { return f.size }
-	return f.size
-}
 
-// Mode always returns 0 because a git blob is an ordinary file.
+// Mode returns file mode for the git blob.
 // It is part of the fs.FileInfo interface.
-func (f *file) Mode() fs.FileMode {
+func (f *fsFile) Mode() fs.FileMode {
-	return 0
+	switch f.mode { //nolint:exhaustive
-}
+	case TreeNodeModeSymlink:
-
+		return fs.ModeSymlink
-// ModTime is unimplemented.
+	case TreeNodeModeCommit:
-// It is part of the fs.FileInfo interface.
+		return fs.ModeIrregular
-func (f *file) ModTime() time.Time {
+	case TreeNodeModeExec:
-	// Git doesn't store file modification time directly.
+		return 0o555
-	// It's possible to find the last commit (and thus the commit time)
+	default:
-	// that modified touched the file, but this is out of scope for this implementation.
+		return 0o444
-	return time.Time{}
-}
-
-// IsDir implementation for the file struct always returns false.
-// It is part of the fs.FileInfo interface.
-func (f *file) IsDir() bool {
-	return false
-}
-
-// Sys is unimplemented.
-// It is part of the fs.FileInfo interface.
-func (f *file) Sys() any {
-	return nil
-}
-
-type entry struct {
-	TreeNode
-}
-
-// Name returns name of a git tree entry.
-// It is part of the fs.DirEntry interface.
-func (e entry) Name() string {
-	return e.TreeNode.Name
-}
-
-// IsDir returns if a git tree entry is a directory.
-// It is part of the fs.FileInfo and fs.DirEntry interfaces.
-func (e entry) IsDir() bool {
-	return e.TreeNode.IsDir()
-}
-
-// Type returns the type of git tree entry.
-// It is part of the fs.DirEntry interface.
-func (e entry) Type() fs.FileMode {
-	if e.TreeNode.IsDir() {
-		return fs.ModeDir
 	}
-	return 0
-}
-
-// Info returns FileInfo for a git tree entry.
-// It is part of the fs.DirEntry interface.
-func (e entry) Info() (fs.FileInfo, error) {
-	return e, nil
-}
-
-// Size returns file size - the size of the git blob object.
-// It is part of the fs.FileInfo interface.
-func (e entry) Size() int64 {
-	return e.TreeNode.Size
-}
-
-// Mode always returns 0 because a git blob is an ordinary file.
-// It is part of the fs.FileInfo interface.
-func (e entry) Mode() fs.FileMode {
-	return 0
 }
 
 // ModTime is unimplemented. Git doesn't store file modification time directly.
 // It's possible to find the last commit (and thus the commit time)
 // that modified touched the file, but this is out of scope for this implementation.
 // It is part of the fs.FileInfo interface.
-func (e entry) ModTime() time.Time { return time.Time{} }
+func (f *fsFile) ModTime() time.Time { return time.Time{} }
+
+// IsDir implementation always returns false.
+// It is part of the fs.FileInfo interface.
+func (f *fsFile) IsDir() bool { return false }
 
 // Sys is unimplemented.
 // It is part of the fs.FileInfo interface.
-func (e entry) Sys() any { return nil }
+func (f *fsFile) Sys() any { return nil }
+
+type fsEntry struct {
+	TreeNode
+}
+
+// Name returns name of a git tree entry.
+// It is part of the fs.DirEntry interface.
+func (e fsEntry) Name() string { return e.TreeNode.Name }
+
+// IsDir returns if a git tree entry is a directory.
+// It is part of the fs.FileInfo and fs.DirEntry interfaces.
+func (e fsEntry) IsDir() bool { return e.TreeNode.IsDir() }
+
+// Type returns the type of git tree entry.
+// It is part of the fs.DirEntry interface.
+func (e fsEntry) Type() fs.FileMode { return e.Mode() }
+
+// Info returns FileInfo for a git tree entry.
+// It is part of the fs.DirEntry interface.
+func (e fsEntry) Info() (fs.FileInfo, error) { return e, nil }
+
+// Size returns file size - the size of the git blob object.
+// It is part of the fs.FileInfo interface.
+func (e fsEntry) Size() int64 { return e.TreeNode.Size }
+
+// Mode always returns the filesystem entry mode.
+// It is part of the fs.FileInfo interface.
+func (e fsEntry) Mode() fs.FileMode {
+	var mode fs.FileMode
+	if e.TreeNode.Mode == TreeNodeModeExec {
+		mode = 0o555
+	} else {
+		mode = 0o444
+	}
+	if e.TreeNode.IsDir() {
+		mode |= fs.ModeDir
+	}
+	if e.TreeNode.IsLink() {
+		mode |= fs.ModeSymlink
+	}
+	if e.TreeNode.IsSubmodule() {
+		mode |= fs.ModeIrregular
+	}
+	return mode
+}
+
+// ModTime is unimplemented. Git doesn't store file modification time directly.
+// It's possible to find the last commit (and thus the commit time)
+// that modified touched the file, but this is out of scope for this implementation.
+// It is part of the fs.FileInfo interface.
+func (e fsEntry) ModTime() time.Time { return time.Time{} }
+
+// Sys is unimplemented.
+// It is part of the fs.FileInfo interface.
+func (e fsEntry) Sys() any { return nil }