From 26bdedb5a25460e986bd791306998dbcf4ba0b8a Mon Sep 17 00:00:00 2001
From: Seednode <git@seedno.de>
Date: Mon, 11 Sep 2023 19:38:38 -0500
Subject: [PATCH] Added UTF-8 validation for files when using --text

---
 cmd/root.go       |  4 ++--
 formats/audio.go  |  3 +++
 formats/images.go |  3 +++
 formats/text.go   | 20 ++++++++++++++++----
 formats/types.go  |  4 +++-
 formats/video.go  |  3 +++
 6 files changed, 30 insertions(+), 7 deletions(-)

diff --git a/cmd/root.go b/cmd/root.go
index f807860..c2d70db 100644
--- a/cmd/root.go
+++ b/cmd/root.go
@@ -12,7 +12,7 @@ import (
 )
 
 const (
-	Version string = "0.68.0"
+	Version string = "0.69.0"
 )
 
 var (
@@ -94,7 +94,7 @@ func init() {
 	rootCmd.Flags().BoolVarP(&sorting, "sort", "s", false, "enable sorting")
 	rootCmd.Flags().BoolVar(&statistics, "stats", false, "expose stats endpoint")
 	rootCmd.Flags().StringVar(&statisticsFile, "stats-file", "", "path to optional persistent stats file")
-	rootCmd.Flags().BoolVar(&text, "text", true, "enable support for text files")
+	rootCmd.Flags().BoolVar(&text, "text", false, "enable support for text files")
 	rootCmd.Flags().BoolVarP(&verbose, "verbose", "v", false, "log accessed files to stdout")
 	rootCmd.Flags().BoolVarP(&version, "version", "V", false, "display version and exit")
 	rootCmd.Flags().BoolVar(&videos, "video", false, "enable support for video files")
diff --git a/formats/audio.go b/formats/audio.go
index 5ed2bdd..7b90d17 100644
--- a/formats/audio.go
+++ b/formats/audio.go
@@ -34,5 +34,8 @@ func RegisterAudioFormats() *SupportedFormat {
 			`audio/ogg`,
 			`audio/wav`,
 		},
+		Validate: func(filePath string) bool {
+			return true
+		},
 	}
 }
diff --git a/formats/images.go b/formats/images.go
index bed3713..0ef1931 100644
--- a/formats/images.go
+++ b/formats/images.go
@@ -67,6 +67,9 @@ func RegisterImageFormats() *SupportedFormat {
 			`image/png`,
 			`image/webp`,
 		},
+		Validate: func(filePath string) bool {
+			return true
+		},
 	}
 }
 
diff --git a/formats/text.go b/formats/text.go
index 6d9c474..64ee81e 100644
--- a/formats/text.go
+++ b/formats/text.go
@@ -5,6 +5,7 @@ Copyright © 2023 Seednode <seednode@seedno.de>
 package formats
 
 import (
+	"errors"
 	"fmt"
 	"os"
 	"unicode/utf8"
@@ -24,10 +25,6 @@ func RegisterTextFormats() *SupportedFormat {
 				body = []byte{}
 			}
 
-			if !utf8.Valid(body) {
-				body = []byte(`Unable to parse binary file as text.`)
-			}
-
 			return fmt.Sprintf(`<a href="/%s"><pre>%s</pre></a>`,
 				queryParams,
 				body)
@@ -52,5 +49,20 @@ func RegisterTextFormats() *SupportedFormat {
 			`text/plain`,
 			`text/plain; charset=utf-8`,
 		},
+		Validate: func(path string) bool {
+			file, err := os.Open(path)
+			switch {
+			case errors.Is(err, os.ErrNotExist):
+				return false
+			case err != nil:
+				return false
+			}
+			defer file.Close()
+
+			head := make([]byte, 512)
+			file.Read(head)
+
+			return utf8.Valid(head)
+		},
 	}
 }
diff --git a/formats/types.go b/formats/types.go
index b09b659..579dcbf 100644
--- a/formats/types.go
+++ b/formats/types.go
@@ -11,6 +11,7 @@ import (
 )
 
 type FormatFunction func(queryParams, fileUri, filePath, fileName, mime string) string
+type ValidateFunction func(filePath string) bool
 
 type SupportedFormat struct {
 	Css        FormatFunction
@@ -18,6 +19,7 @@ type SupportedFormat struct {
 	Body       FormatFunction
 	Extensions []string
 	MimeTypes  []string
+	Validate   ValidateFunction
 }
 
 type SupportedFormats struct {
@@ -79,7 +81,7 @@ func FileType(path string, types *SupportedFormats) (bool, *SupportedFormat, str
 		if mimeType == v {
 			fileType := types.Type(mimeType)
 
-			return true, fileType, mimeType, nil
+			return fileType.Validate(path), fileType, mimeType, nil
 		}
 	}
 
diff --git a/formats/video.go b/formats/video.go
index 04b3207..d7565f4 100644
--- a/formats/video.go
+++ b/formats/video.go
@@ -33,5 +33,8 @@ func RegisterVideoFormats() *SupportedFormat {
 			`video/ogg`,
 			`video/webm`,
 		},
+		Validate: func(filePath string) bool {
+			return true
+		},
 	}
 }