Custom decoder

7z.go

@@ -92,14 +92,17 @@ func (x *XFile) un7zip(zipFile *sevenzip.File) (int64, string, error) {
 	defer zFile.Close()
 
 	file := &file{
-		Path:     x.clean(zipFile.Name),
 		Data:     zFile,
 		FileMode: zipFile.Mode(),
 		DirMode:  x.DirMode,
 		Mtime:    zipFile.Modified,
 		Atime:    zipFile.Accessed,
 	}
 
+	if file.Path, err = x.clean(zipFile.Name); err != nil {
+		return 0, file.Path, err
+	}
+
 	if !strings.HasPrefix(file.Path, x.OutputDir) {
 		// The file being written is trying to write outside of our base path. Malicious archive?
 		err := fmt.Errorf("%s: %w: %s (from: %s)", zipFile.FileInfo().Name(), ErrInvalidPath, file.Path, zipFile.Name)

ar.go

@@ -37,13 +37,16 @@ func (x *XFile) unAr(reader io.Reader) (int64, []string, error) {
 		}
 
 		file := &file{
-			Path:     x.clean(header.Name),
 			Data:     arReader,
 			FileMode: os.FileMode(header.Mode), //nolint:gosec // what else ya gonna do with this?
 			DirMode:  x.DirMode,
 			Mtime:    header.ModTime,
 		}
 
+		if file.Path, err = x.clean(header.Name); err != nil {
+			return 0, files, err
+		}
+
 		if !strings.HasPrefix(file.Path, x.OutputDir) {
 			// The file being written is trying to write outside of our base path. Malicious archive?
 			return size, files, fmt.Errorf("%s: %w: %s (from: %s)", x.FilePath, ErrInvalidPath, file.Path, header.Name)

cpio.go

@@ -65,13 +65,17 @@ func (x *XFile) uncpio(reader io.Reader) (int64, []string, error) {
 
 func (x *XFile) uncpioFile(cpioFile *cpio.Header, cpioReader *cpio.Reader) (int64, error) {
 	file := &file{
-		Path:     x.clean(cpioFile.Name),
 		Data:     cpioReader,
 		FileMode: cpioFile.FileInfo().Mode(),
 		DirMode:  x.DirMode,
 		Mtime:    cpioFile.ModTime,
 	}
 
+	var err error
+	if file.Path, err = x.clean(cpioFile.Name); err != nil {
+		return 0, err
+	}
+
 	if !strings.HasPrefix(file.Path, x.OutputDir) {
 		// The file being written is trying to write outside of the base path. Malicious archive?
 		return 0, fmt.Errorf("%s: %w: %s (from: %s)", cpioFile.FileInfo().Name(), ErrInvalidPath, file.Path, cpioFile.Name)

decompress.go

@@ -30,14 +30,18 @@ func ExtractXZ(xFile *XFile) (size int64, filesList []string, err error) {
 		return 0, nil, fmt.Errorf("xz.NewReader: %w", err)
 	}
 
-	// Get the absolute path of the file being written.
 	file := &file{
-		Path:     xFile.clean(xFile.FilePath, ".xz"),
 		Data:     zipReader,
 		FileMode: xFile.FileMode,
 		DirMode:  xFile.DirMode,
 	}
 
+	// Get the absolute path of the file being written.
+	file.Path, err = xFile.clean(xFile.FilePath, ".xz")
+	if err != nil {
+		return 0, nil, err
+	}
+
 	size, err = xFile.write(file)
 	if err != nil {
 		return size, nil, err
@@ -60,14 +64,18 @@ func ExtractZlib(xFile *XFile) (size int64, filesList []string, err error) {
 	}
 	defer zipReader.Close()
 
-	// Get the absolute path of the file being written.
 	file := &file{
-		Path:     xFile.clean(xFile.FilePath, ".zz", ".zlib"),
 		Data:     zipReader,
 		FileMode: xFile.FileMode,
 		DirMode:  xFile.DirMode,
 	}
 
+	// Get the absolute path of the file being written.
+	file.Path, err = xFile.clean(xFile.FilePath, ".zz", ".zlib")
+	if err != nil {
+		return 0, nil, err
+	}
+
 	size, err = xFile.write(file)
 	if err != nil {
 		return size, nil, err
@@ -89,14 +97,18 @@ func ExtractLZMA(xFile *XFile) (size int64, filesList []string, err error) {
 		return 0, nil, fmt.Errorf("lzma.NewReader: %w", err)
 	}
 
-	// Get the absolute path of the file being written.
 	file := &file{
-		Path:     xFile.clean(xFile.FilePath, ".lzma", ".lz", ".lzip"),
 		Data:     zipReader,
 		FileMode: xFile.FileMode,
 		DirMode:  xFile.DirMode,
 	}
 
+	// Get the absolute path of the file being written.
+	file.Path, err = xFile.clean(xFile.FilePath, ".lzma", ".lz", ".lzip")
+	if err != nil {
+		return 0, nil, err
+	}
+
 	size, err = xFile.write(file)
 	if err != nil {
 		return size, nil, err
@@ -118,14 +130,18 @@ func ExtractLZMA2(xFile *XFile) (size int64, filesList []string, err error) {
 		return 0, nil, fmt.Errorf("lzma.NewReader2: %w", err)
 	}
 
-	// Get the absolute path of the file being written.
 	file := &file{
-		Path:     xFile.clean(xFile.FilePath, ".lzma", ".lzma2"),
 		Data:     zipReader,
 		FileMode: xFile.FileMode,
 		DirMode:  xFile.DirMode,
 	}
 
+	// Get the absolute path of the file being written.
+	file.Path, err = xFile.clean(xFile.FilePath, ".lzma", ".lzma2")
+	if err != nil {
+		return 0, nil, err
+	}
+
 	size, err = xFile.write(file)
 	if err != nil {
 		return size, nil, err
@@ -148,14 +164,18 @@ func ExtractZstandard(xFile *XFile) (size int64, filesList []string, err error)
 	}
 	defer zipReader.Close()
 
-	// Get the absolute path of the file being written.
 	file := &file{
-		Path:     xFile.clean(xFile.FilePath, ".zstd", ".zst"),
 		Data:     zipReader,
 		FileMode: xFile.FileMode,
 		DirMode:  xFile.DirMode,
 	}
 
+	// Get the absolute path of the file being written.
+	file.Path, err = xFile.clean(xFile.FilePath, ".zstd", ".zst")
+	if err != nil {
+		return 0, nil, err
+	}
+
 	size, err = xFile.write(file)
 	if err != nil {
 		return size, nil, err
@@ -177,14 +197,18 @@ func ExtractLZW(xFile *XFile) (size int64, filesList []string, err error) {
 		return 0, nil, fmt.Errorf("lzw.NewReader: %w", err)
 	}
 
-	// Get the absolute path of the file being written.
 	file := &file{
-		Path:     xFile.clean(xFile.FilePath, ".Z"),
 		Data:     zipReader,
 		FileMode: xFile.FileMode,
 		DirMode:  xFile.DirMode,
 	}
 
+	// Get the absolute path of the file being written.
+	file.Path, err = xFile.clean(xFile.FilePath, ".Z")
+	if err != nil {
+		return 0, nil, err
+	}
+
 	size, err = xFile.write(file)
 	if err != nil {
 		return size, nil, err
@@ -201,14 +225,18 @@ func ExtractLZ4(xFile *XFile) (size int64, filesList []string, err error) {
 	}
 	defer compressedFile.Close()
 
-	// Get the absolute path of the file being written.
 	file := &file{
-		Path:     xFile.clean(xFile.FilePath, ".lz4"),
 		Data:     lz4.NewReader(compressedFile),
 		FileMode: xFile.FileMode,
 		DirMode:  xFile.DirMode,
 	}
 
+	// Get the absolute path of the file being written.
+	file.Path, err = xFile.clean(xFile.FilePath, ".lz4")
+	if err != nil {
+		return 0, nil, err
+	}
+
 	size, err = xFile.write(file)
 	if err != nil {
 		return size, nil, err
@@ -225,14 +253,18 @@ func ExtractSnappy(xFile *XFile) (size int64, filesList []string, err error) {
 	}
 	defer compressedFile.Close()
 
-	// Get the absolute path of the file being written.
 	file := &file{
-		Path:     xFile.clean(xFile.FilePath, ".snappy", ".sz"),
 		Data:     snappy.NewReader(compressedFile),
 		FileMode: xFile.FileMode,
 		DirMode:  xFile.DirMode,
 	}
 
+	// Get the absolute path of the file being written.
+	file.Path, err = xFile.clean(xFile.FilePath, ".snappy", ".sz")
+	if err != nil {
+		return 0, nil, err
+	}
+
 	size, err = xFile.write(file)
 	if err != nil {
 		return size, nil, err
@@ -249,14 +281,18 @@ func ExtractS2(xFile *XFile) (size int64, filesList []string, err error) {
 	}
 	defer compressedFile.Close()
 
-	// Get the absolute path of the file being written.
 	file := &file{
-		Path:     xFile.clean(xFile.FilePath, ".s2"),
 		Data:     s2.NewReader(compressedFile),
 		FileMode: xFile.FileMode,
 		DirMode:  xFile.DirMode,
 	}
 
+	// Get the absolute path of the file being written.
+	file.Path, err = xFile.clean(xFile.FilePath, ".s2")
+	if err != nil {
+		return 0, nil, err
+	}
+
 	size, err = xFile.write(file)
 	if err != nil {
 		return size, nil, err
@@ -273,14 +309,18 @@ func ExtractBrotli(xFile *XFile) (size int64, filesList []string, err error) {
 	}
 	defer compressedFile.Close()
 
-	// Get the absolute path of the file being written.
 	file := &file{
-		Path:     xFile.clean(xFile.FilePath, ".brotli", ".br"),
 		Data:     brotli.NewReader(compressedFile),
 		FileMode: xFile.FileMode,
 		DirMode:  xFile.DirMode,
 	}
 
+	// Get the absolute path of the file being written.
+	file.Path, err = xFile.clean(xFile.FilePath, ".brotli", ".br")
+	if err != nil {
+		return 0, nil, err
+	}
+
 	size, err = xFile.write(file)
 	if err != nil {
 		return size, nil, err
@@ -297,14 +337,18 @@ func ExtractBzip(xFile *XFile) (size int64, filesList []string, err error) {
 	}
 	defer compressedFile.Close()
 
-	// Get the absolute path of the file being written.
 	file := &file{
-		Path:     xFile.clean(xFile.FilePath, ".bz", ".bz2"),
 		Data:     bzip2.NewReader(compressedFile),
 		FileMode: xFile.FileMode,
 		DirMode:  xFile.DirMode,
 	}
 
+	// Get the absolute path of the file being written.
+	file.Path, err = xFile.clean(xFile.FilePath, ".bz", ".bz2")
+	if err != nil {
+		return 0, nil, err
+	}
+
 	size, err = xFile.write(file)
 	if err != nil {
 		return size, nil, err
@@ -327,15 +371,19 @@ func ExtractGzip(xFile *XFile) (size int64, filesList []string, err error) {
 	}
 	defer zipReader.Close()
 
-	// Get the absolute path of the file being written.
 	file := &file{
-		Path:     xFile.clean(xFile.FilePath, ".gz"),
 		Data:     zipReader,
 		FileMode: xFile.FileMode,
 		DirMode:  xFile.DirMode,
 		Mtime:    zipReader.ModTime,
 	}
 
+	// Get the absolute path of the file being written.
+	file.Path, err = xFile.clean(xFile.FilePath, ".gz")
+	if err != nil {
+		return 0, nil, err
+	}
+
 	size, err = xFile.write(file)
 	if err != nil {
 		return size, nil, err

encoding.go

@@ -0,0 +1,30 @@
+package xtractr
+
+import "fmt"
+
+/* This file will surely grow when someone writes a proper character encoding detector. */
+
+// EncoderInput is used as input for a custom encoder procedure.
+type EncoderInput struct {
+	FileName string
+	XFile    *XFile
+}
+
+// decode a string using the provided decoder.
+func (x *XFile) decode(input string) (string, error) {
+	if x.Encoder == nil {
+		return input, nil
+	}
+
+	encoding := x.Encoder(&EncoderInput{FileName: input, XFile: x})
+	if encoding == nil {
+		return input, nil
+	}
+
+	output, err := encoding.String(input)
+	if err != nil {
+		return "", fmt.Errorf("decoding file name: %w", err)
+	}
+
+	return output, nil
+}

files.go

@@ -10,6 +10,8 @@ import (
 	"regexp"
 	"strings"
 	"time"
+
+	"golang.org/x/text/encoding"
 )
 
 // ArchiveList is the value returned when searching for compressed files.
@@ -112,6 +114,9 @@ type XFile struct {
 	Password string
 	// (RAR/7z) Archive passwords (to try multiple).
 	Passwords []string
+	// If file names are not UTF8 encoded, pass your own encoder here.
+	// Provide a function that takes in a file name and returns an encoder for it.
+	Encoder func(*EncoderInput) *encoding.Decoder
 	// If the archive only has one directory in the root, then setting
 	// this true will cause the extracted content to be moved into the
 	// output folder, and the root folder in the archive to be removed.
@@ -473,16 +478,22 @@ func (x *Xtractr) Rename(oldpath, newpath string) error {
 }
 
 // clean returns an absolute path for a file inside the OutputDir.
+// clean also decodes the file name using a provided decoder.
 // If trim length is > 0, then the suffixes are trimmed, and filepath removed.
-func (x *XFile) clean(filePath string, trim ...string) string {
+func (x *XFile) clean(filePath string, trim ...string) (string, error) {
+	filePath, err := x.decode(filePath)
+	if err != nil {
+		return "", err
+	}
+
 	if len(trim) != 0 {
 		filePath = filepath.Base(filePath)
 		for _, suffix := range trim {
 			filePath = strings.TrimSuffix(filePath, suffix)
 		}
 	}
 
-	return filepath.Clean(filepath.Join(x.OutputDir, filePath))
+	return filepath.Clean(filepath.Join(x.OutputDir, filePath)), nil
 }
 
 // AllExcept can be used as an input to ExcludeSuffix in a Filter.