Let notebook detection code use underlying metadata if available (#1574)

## Changes

If we're using a `vfs.Path` backed by a workspace filesystem filer, we
have access to the `workspace.ObjectInfo` value for every file. By
providing access to this value we can use it directly and avoid reading
the first line of the underlying file.

A follow-up change will implement the interface defined in this change
for the workspace filesystem filer.

## Tests

Unit tests.

libs/notebook/detect.go

@@ -12,27 +12,69 @@ import (
 	"github.com/databricks/databricks-sdk-go/service/workspace"
 )
 
+// FileInfoWithWorkspaceObjectInfo is an interface implemented by [fs.FileInfo] values that
+// contain a file's underlying [workspace.ObjectInfo].
+//
+// This may be the case when working with a [filer.Filer] backed by the workspace API.
+// For these files we do not need to read a file's header to know if it is a notebook;
+// we can use the [workspace.ObjectInfo] value directly.
+type FileInfoWithWorkspaceObjectInfo interface {
+	WorkspaceObjectInfo() workspace.ObjectInfo
+}
+
 // Maximum length in bytes of the notebook header.
 const headerLength = 32
 
-// readHeader reads the first N bytes from a file.
+// file wraps an fs.File and implements a few helper methods such that
-func readHeader(fsys fs.FS, name string) ([]byte, error) {
+// they don't need to be inlined in the [DetectWithFS] function below.
+type file struct {
+	f fs.File
+}
+
+func openFile(fsys fs.FS, name string) (*file, error) {
 	f, err := fsys.Open(name)
 	if err != nil {
 		return nil, err
 	}
 
-	defer f.Close()
+	return &file{f: f}, nil
+}
 
+func (f file) close() error {
+	return f.f.Close()
+}
+
+func (f file) readHeader() (string, error) {
 	// Scan header line with some padding.
 	var buf = make([]byte, headerLength)
-	n, err := f.Read([]byte(buf))
+	n, err := f.f.Read([]byte(buf))
 	if err != nil && err != io.EOF {
-		return nil, err
+		return "", err
 	}
 
 	// Trim buffer to actual read bytes.
-	return buf[:n], nil
+	buf = buf[:n]
+
+	// Read the first line from the buffer.
+	scanner := bufio.NewScanner(bytes.NewReader(buf))
+	scanner.Scan()
+	return scanner.Text(), nil
+}
+
+// getObjectInfo returns the [workspace.ObjectInfo] for the file if it is
+// part of the [fs.FileInfo] value returned by the [fs.Stat] call.
+func (f file) getObjectInfo() (oi workspace.ObjectInfo, ok bool, err error) {
+	stat, err := f.f.Stat()
+	if err != nil {
+		return workspace.ObjectInfo{}, false, err
+	}
+
+	// Use object info if available.
+	if i, ok := stat.(FileInfoWithWorkspaceObjectInfo); ok {
+		return i.WorkspaceObjectInfo(), true, nil
+	}
+
+	return workspace.ObjectInfo{}, false, nil
 }
 
 // Detect returns whether the file at path is a Databricks notebook.
@@ -40,13 +82,27 @@ func readHeader(fsys fs.FS, name string) ([]byte, error) {
 func DetectWithFS(fsys fs.FS, name string) (notebook bool, language workspace.Language, err error) {
 	header := ""
 
-	buf, err := readHeader(fsys, name)
+	f, err := openFile(fsys, name)
+	if err != nil {
+		return false, "", err
+	}
+
+	defer f.close()
+
+	// Use object info if available.
+	oi, ok, err := f.getObjectInfo()
+	if err != nil {
+		return false, "", err
+	}
+	if ok {
+		return oi.ObjectType == workspace.ObjectTypeNotebook, oi.Language, nil
+	}
+
+	// Read the first line of the file.
+	fileHeader, err := f.readHeader()
 	if err != nil {
 		return false, "", err
 	}
-	scanner := bufio.NewScanner(bytes.NewReader(buf))
-	scanner.Scan()
-	fileHeader := scanner.Text()
 
 	// Determine which header to expect based on filename extension.
 	ext := strings.ToLower(filepath.Ext(name))

libs/notebook/detect_test.go

@@ -99,3 +99,21 @@ func TestDetectFileWithLongHeader(t *testing.T) {
 	require.NoError(t, err)
 	assert.False(t, nb)
 }
+
+func TestDetectWithObjectInfo(t *testing.T) {
+	fakeFS := &fakeFS{
+		fakeFile{
+			fakeFileInfo{
+				workspace.ObjectInfo{
+					ObjectType: workspace.ObjectTypeNotebook,
+					Language:   workspace.LanguagePython,
+				},
+			},
+		},
+	}
+
+	nb, lang, err := DetectWithFS(fakeFS, "doesntmatter")
+	require.NoError(t, err)
+	assert.True(t, nb)
+	assert.Equal(t, workspace.LanguagePython, lang)
+}

libs/notebook/fakefs_test.go

@@ -0,0 +1,77 @@
+package notebook
+
+import (
+	"fmt"
+	"io/fs"
+	"time"
+
+	"github.com/databricks/databricks-sdk-go/service/workspace"
+)
+
+type fakeFS struct {
+	fakeFile
+}
+
+type fakeFile struct {
+	fakeFileInfo
+}
+
+func (f fakeFile) Close() error {
+	return nil
+}
+
+func (f fakeFile) Read(p []byte) (n int, err error) {
+	return 0, fmt.Errorf("not implemented")
+}
+
+func (f fakeFile) Stat() (fs.FileInfo, error) {
+	return f.fakeFileInfo, nil
+}
+
+type fakeFileInfo struct {
+	oi workspace.ObjectInfo
+}
+
+func (f fakeFileInfo) WorkspaceObjectInfo() workspace.ObjectInfo {
+	return f.oi
+}
+
+func (f fakeFileInfo) Name() string {
+	return ""
+}
+
+func (f fakeFileInfo) Size() int64 {
+	return 0
+}
+
+func (f fakeFileInfo) Mode() fs.FileMode {
+	return 0
+}
+
+func (f fakeFileInfo) ModTime() time.Time {
+	return time.Time{}
+}
+
+func (f fakeFileInfo) IsDir() bool {
+	return false
+}
+
+func (f fakeFileInfo) Sys() any {
+	return nil
+}
+
+func (f fakeFS) Open(name string) (fs.File, error) {
+	return f.fakeFile, nil
+}
+
+func (f fakeFS) Stat(name string) (fs.FileInfo, error) {
+	panic("not implemented")
+}
+
+func (f fakeFS) ReadDir(name string) ([]fs.DirEntry, error) {
+	panic("not implemented")
+}
+
+func (f fakeFS) ReadFile(name string) ([]byte, error) {
+	panic("not implemented")
+}