feat: allow disabling file detections by reading header (#1175)

cmd/root.go

@@ -66,6 +66,7 @@ func addServerFlags(flags *pflag.FlagSet) {
 	flags.Bool("disable-thumbnails", false, "disable image thumbnails")
 	flags.Bool("disable-preview-resize", false, "disable resize of image previews")
 	flags.Bool("disable-exec", false, "disables Command Runner feature")
+	flags.Bool("disable-type-detection-by-header", false, "disables type detection by reading file headers")
 }
 
 var rootCmd = &cobra.Command{
@@ -243,6 +244,9 @@ func getRunParams(flags *pflag.FlagSet, st *storage.Storage) *settings.Server {
 	_, disablePreviewResize := getParamB(flags, "disable-preview-resize")
 	server.ResizePreview = !disablePreviewResize
 
+	_, disableTypeDetectionByHeader := getParamB(flags, "disable-type-detection-by-header")
+	server.TypeDetectionByHeader = !disableTypeDetectionByHeader
+
 	_, disableExec := getParamB(flags, "disable-exec")
 	server.EnableExec = !disableExec
 

files/file.go

@@ -42,11 +42,12 @@ type FileInfo struct {
 
 // FileOptions are the options when getting a file info.
 type FileOptions struct {
-	Fs      afero.Fs
-	Path    string
-	Modify  bool
-	Expand  bool
-	Checker rules.Checker
+	Fs         afero.Fs
+	Path       string
+	Modify     bool
+	Expand     bool
+	ReadHeader bool
+	Checker    rules.Checker
 }
 
 // NewFileInfo creates a File object from a path and a given user. This File
@@ -75,13 +76,13 @@ func NewFileInfo(opts FileOptions) (*FileInfo, error) {
 
 	if opts.Expand {
 		if file.IsDir {
-			if err := file.readListing(opts.Checker); err != nil { //nolint:shadow
+			if err := file.readListing(opts.Checker, opts.ReadHeader); err != nil { //nolint:shadow
 				return nil, err
 			}
 			return file, nil
 		}
 
-		err = file.detectType(opts.Modify, true)
+		err = file.detectType(opts.Modify, true, true)
 		if err != nil {
 			return nil, err
 		}
@@ -134,7 +135,7 @@ func (i *FileInfo) Checksum(algo string) error {
 
 //nolint:goconst
 //TODO: use constants
-func (i *FileInfo) detectType(modify, saveContent bool) error {
+func (i *FileInfo) detectType(modify, saveContent, readHeader bool) error {
 	if IsNamedPipe(i.Mode) {
 		i.Type = "blob"
 		return nil
@@ -143,25 +144,13 @@ func (i *FileInfo) detectType(modify, saveContent bool) error {
 	// imagine the situation where a file in a dir with thousands
 	// of files couldn't be opened: we'd have immediately
 	// a 500 even though it doesn't matter. So we just log it.
-	reader, err := i.Fs.Open(i.Path)
-	if err != nil {
-		log.Print(err)
-		i.Type = "blob"
-		return nil
-	}
-	defer reader.Close()
 
-	buffer := make([]byte, 512)
-	n, err := reader.Read(buffer)
-	if err != nil && err != io.EOF {
-		log.Print(err)
-		i.Type = "blob"
-		return nil
-	}
+	var buffer []byte
 
 	mimetype := mime.TypeByExtension(i.Extension)
-	if mimetype == "" {
-		mimetype = http.DetectContentType(buffer[:n])
+	if mimetype == "" && readHeader {
+		buffer = i.readFirstBytes()
+		mimetype = http.DetectContentType(buffer)
 	}
 
 	switch {
@@ -175,10 +164,7 @@ func (i *FileInfo) detectType(modify, saveContent bool) error {
 	case strings.HasPrefix(mimetype, "image"):
 		i.Type = "image"
 		return nil
-	case isBinary(buffer[:n], n) || i.Size > 10*1024*1024: // 10 MB
-		i.Type = "blob"
-		return nil
-	default:
+	case (strings.HasPrefix(mimetype, "text") || (len(buffer) > 0 && !isBinary(buffer))) && i.Size <= 10*1024*1024: // 10 MB
 		i.Type = "text"
 
 		if !modify {
@@ -194,11 +180,34 @@ func (i *FileInfo) detectType(modify, saveContent bool) error {
 
 			i.Content = string(content)
 		}
+		return nil
+	default:
+		i.Type = "blob"
 	}
 
 	return nil
 }
 
+func (i *FileInfo) readFirstBytes() []byte {
+	reader, err := i.Fs.Open(i.Path)
+	if err != nil {
+		log.Print(err)
+		i.Type = "blob"
+		return nil
+	}
+	defer reader.Close()
+
+	buffer := make([]byte, 512)
+	n, err := reader.Read(buffer)
+	if err != nil && err != io.EOF {
+		log.Print(err)
+		i.Type = "blob"
+		return nil
+	}
+
+	return buffer[:n]
+}
+
 func (i *FileInfo) detectSubtitles() {
 	if i.Type != "video" {
 		return
@@ -215,7 +224,7 @@ func (i *FileInfo) detectSubtitles() {
 	}
 }
 
-func (i *FileInfo) readListing(checker rules.Checker) error {
+func (i *FileInfo) readListing(checker rules.Checker, readHeader bool) error {
 	afs := &afero.Afero{Fs: i.Fs}
 	dir, err := afs.ReadDir(i.Path)
 	if err != nil {
@@ -261,7 +270,7 @@ func (i *FileInfo) readListing(checker rules.Checker) error {
 		} else {
 			listing.NumFiles++
 
-			err := file.detectType(true, false)
+			err := file.detectType(true, false, readHeader)
 			if err != nil {
 				return err
 			}

files/utils.go

@@ -5,7 +5,7 @@ import (
 	"unicode/utf8"
 )
 
-func isBinary(content []byte, _ int) bool {
+func isBinary(content []byte) bool {
 	maybeStr := string(content)
 	runeCnt := utf8.RuneCount(content)
 	runeIndex := 0

http/preview.go

@@ -46,11 +46,12 @@ func previewHandler(imgSvc ImgService, fileCache FileCache, enableThumbnails, re
 		}
 
 		file, err := files.NewFileInfo(files.FileOptions{
-			Fs:      d.user.Fs,
-			Path:    "/" + vars["path"],
-			Modify:  d.user.Perm.Modify,
-			Expand:  true,
-			Checker: d,
+			Fs:         d.user.Fs,
+			Path:       "/" + vars["path"],
+			Modify:     d.user.Perm.Modify,
+			Expand:     true,
+			ReadHeader: d.server.TypeDetectionByHeader,
+			Checker:    d,
 		})
 		if err != nil {
 			return errToStatus(err), err

http/public.go

@@ -27,11 +27,12 @@ var withHashFile = func(fn handleFunc) handleFunc {
 		d.user = user
 
 		file, err := files.NewFileInfo(files.FileOptions{
-			Fs:      d.user.Fs,
-			Path:    link.Path,
-			Modify:  d.user.Perm.Modify,
-			Expand:  true,
-			Checker: d,
+			Fs:         d.user.Fs,
+			Path:       link.Path,
+			Modify:     d.user.Perm.Modify,
+			Expand:     true,
+			ReadHeader: d.server.TypeDetectionByHeader,
+			Checker:    d,
 		})
 		if err != nil {
 			return errToStatus(err), err

http/raw.go

@@ -84,11 +84,12 @@ var rawHandler = withUser(func(w http.ResponseWriter, r *http.Request, d *data)
 	}
 
 	file, err := files.NewFileInfo(files.FileOptions{
-		Fs:      d.user.Fs,
-		Path:    r.URL.Path,
-		Modify:  d.user.Perm.Modify,
-		Expand:  false,
-		Checker: d,
+		Fs:         d.user.Fs,
+		Path:       r.URL.Path,
+		Modify:     d.user.Perm.Modify,
+		Expand:     false,
+		ReadHeader: d.server.TypeDetectionByHeader,
+		Checker:    d,
 	})
 	if err != nil {
 		return errToStatus(err), err

http/resource.go

@@ -20,11 +20,12 @@ import (
 
 var resourceGetHandler = withUser(func(w http.ResponseWriter, r *http.Request, d *data) (int, error) {
 	file, err := files.NewFileInfo(files.FileOptions{
-		Fs:      d.user.Fs,
-		Path:    r.URL.Path,
-		Modify:  d.user.Perm.Modify,
-		Expand:  true,
-		Checker: d,
+		Fs:         d.user.Fs,
+		Path:       r.URL.Path,
+		Modify:     d.user.Perm.Modify,
+		Expand:     true,
+		ReadHeader: d.server.TypeDetectionByHeader,
+		Checker:    d,
 	})
 	if err != nil {
 		return errToStatus(err), err
@@ -58,11 +59,12 @@ func resourceDeleteHandler(fileCache FileCache) handleFunc {
 		}
 
 		file, err := files.NewFileInfo(files.FileOptions{
-			Fs:      d.user.Fs,
-			Path:    r.URL.Path,
-			Modify:  d.user.Perm.Modify,
-			Expand:  true,
-			Checker: d,
+			Fs:         d.user.Fs,
+			Path:       r.URL.Path,
+			Modify:     d.user.Perm.Modify,
+			Expand:     true,
+			ReadHeader: d.server.TypeDetectionByHeader,
+			Checker:    d,
 		})
 		if err != nil {
 			return errToStatus(err), err

settings/settings.go

@@ -30,17 +30,18 @@ func (s *Settings) GetRules() []rules.Rule {
 
 // Server specific settings.
 type Server struct {
-	Root             string `json:"root"`
-	BaseURL          string `json:"baseURL"`
-	Socket           string `json:"socket"`
-	TLSKey           string `json:"tlsKey"`
-	TLSCert          string `json:"tlsCert"`
-	Port             string `json:"port"`
-	Address          string `json:"address"`
-	Log              string `json:"log"`
-	EnableThumbnails bool   `json:"enableThumbnails"`
-	ResizePreview    bool   `json:"resizePreview"`
-	EnableExec       bool   `json:"enableExec"`
+	Root                  string `json:"root"`
+	BaseURL               string `json:"baseURL"`
+	Socket                string `json:"socket"`
+	TLSKey                string `json:"tlsKey"`
+	TLSCert               string `json:"tlsCert"`
+	Port                  string `json:"port"`
+	Address               string `json:"address"`
+	Log                   string `json:"log"`
+	EnableThumbnails      bool   `json:"enableThumbnails"`
+	ResizePreview         bool   `json:"resizePreview"`
+	EnableExec            bool   `json:"enableExec"`
+	TypeDetectionByHeader bool   `json:"typeDetectionByHeader"`
 }
 
 // Clean cleans any variables that might need cleaning.