⎣ Added text decoding support ! ⎤

go.mod

@@ -1,3 +1,5 @@
 module github.com/Unbewohnte/id3ed
 
 go 1.16
+
+require golang.org/x/text v0.3.6

go.sum

@@ -0,0 +1,3 @@
+golang.org/x/text v0.3.6 h1:aRYxNxv6iGQlyVaZmk6ZgYEDa+Jg18DxebPSrd6bg1M=
+golang.org/x/text v0.3.6/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
+golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=

util/conversion.go

@@ -4,6 +4,8 @@ import (
 	"fmt"
 	"strconv"
 	"strings"
+
+	euni "golang.org/x/text/encoding/unicode"
 )
 
 // Decodes given byte into integer
@@ -39,13 +41,55 @@ func BytesToIntIgnoreFirstBit(gBytes []byte) (int64, error) {
 // Converts given bytes into string, ignoring the first 31 non-printable ASCII characters.
 // (LOSSY, if given bytes contain some nasty ones)
 func ToStringLossy(gBytes []byte) string {
-	var filteredBytes []byte
+	var runes []rune
 	for _, b := range gBytes {
 		if b <= 31 {
 			continue
 		}
-		filteredBytes = append(filteredBytes, b)
+		runes = append(runes, rune(b))
+	}
+
+	return strings.ToValidUTF8(string(runes), "")
+}
+
+// Decodes the given frame`s contents
+func DecodeText(fContents []byte) string {
+	textEncoding := fContents[0] // the first byte is the encoding
+
+	switch textEncoding {
+	case 0:
+		// ISO-8859-1
+		return ToStringLossy(fContents[1:])
+	case 1:
+		// UTF-16 with BOM
+		encoding := euni.UTF16(euni.BigEndian, euni.ExpectBOM)
+		decoder := encoding.NewDecoder()
+
+		decodedBytes := make([]byte, len(fContents)*2)
+		_, _, err := decoder.Transform(decodedBytes, fContents[1:], true)
+		if err != nil {
+			return ""
+		}
+
+		return string(decodedBytes)
+
+	case 2:
+		// UTF-16
+		encoding := euni.UTF16(euni.BigEndian, euni.IgnoreBOM)
+		decoder := encoding.NewDecoder()
+
+		decodedBytes := make([]byte, len(fContents)*2)
+		_, _, err := decoder.Transform(decodedBytes, fContents[1:], true)
+		if err != nil {
+			return ""
+		}
+
+		return string(decodedBytes)
+
+	case 3:
+		// UTF-8
+		return ToStringLossy(fContents[1:])
 	}
 
-	return strings.ToValidUTF8(string(filteredBytes), "")
+	return ""
 }

util/conversion_test.go

@@ -2,7 +2,7 @@ package util
 
 import "testing"
 
-func TestToString(t *testing.T) {
+func TestToStringLossy(t *testing.T) {
 	someVeryNastyBytes := []byte{0, 1, 2, 3, 4, 5, 6, 50, 7, 8, 9, 10, 11, 50, 50}
 
 	gString := ToStringLossy(someVeryNastyBytes)
@@ -11,3 +11,14 @@ func TestToString(t *testing.T) {
 		t.Errorf("ToString failed: expected output: %s; got %s", "222", gString)
 	}
 }
+
+func TestDecodeText(t *testing.T) {
+	// 3 - UTF-8 encoding
+	someFrameContents := []byte{3, 50, 50, 50, 50, 0, 0, 0, 0, 50}
+
+	decodedUtf8text := DecodeText(someFrameContents)
+
+	if decodedUtf8text != "22222" {
+		t.Errorf("DecodeText failed: expected text %s, got %s", "22222", decodedUtf8text)
+	}
+}

v2/frame.go

@@ -153,3 +153,11 @@ func ReadNextFrame(r io.Reader, h Header) (Frame, uint64, error) {
 
 	return frame, read, err
 }
+
+// Returns decoded string from f.Contents.
+// Note that it can and probably will return
+// corrupted data if you use it on non-text frames such as APIC
+// for such cases please deal with raw []byte
+func (f *Frame) Text() string {
+	return util.DecodeText(f.Contents)
+}

v2/read_test.go

@@ -4,8 +4,6 @@ import (
 	"os"
 	"path/filepath"
 	"testing"
-
-	"github.com/Unbewohnte/id3ed/util"
 )
 
 func TestReadV2Tag(t *testing.T) {
@@ -21,14 +19,14 @@ func TestReadV2Tag(t *testing.T) {
 
 	titleFrame := tag.GetFrame("TIT2")
 
-	if util.ToStringLossy(titleFrame.Contents) != "title" {
+	if titleFrame.Text() != "title" {
 		t.Errorf("ReadV2Tag failed: expected contents of the title frame to be %s; got %s",
-			"title", util.ToStringLossy(titleFrame.Contents))
+			"title", titleFrame.Text())
 	}
 
 	album := tag.Album()
 	if album != "album" {
-		t.Errorf("ReadV2Tag failed: expected contents of the album to be %s; got %s",
+		t.Errorf("ReadV2Tag failed: expected contents of the album frame to be %s; got %s",
 			"album", album)
 	}
 }

v2/v2tag.go

@@ -1,7 +1,5 @@
 package v2
 
-import "github.com/Unbewohnte/id3ed/util"
-
 type ID3v2Tag struct {
 	Header Header
 	Frames []Frame
@@ -22,9 +20,9 @@ func (tag *ID3v2Tag) GetFrame(id string) *Frame {
 func (tag *ID3v2Tag) Title() string {
 	switch tag.Header.Version {
 	case V2_2:
-		return util.ToStringLossy(tag.GetFrame("TT2").Contents)
+		return tag.GetFrame("TT2").Text()
 	default:
-		return util.ToStringLossy(tag.GetFrame("TIT2").Contents)
+		return tag.GetFrame("TIT2").Text()
 	}
 }
 
@@ -32,9 +30,9 @@ func (tag *ID3v2Tag) Title() string {
 func (tag *ID3v2Tag) Album() string {
 	switch tag.Header.Version {
 	case V2_2:
-		return util.ToStringLossy(tag.GetFrame("TAL").Contents)
+		return tag.GetFrame("TAL").Text()
 	default:
-		return util.ToStringLossy(tag.GetFrame("TALB").Contents)
+		return tag.GetFrame("TALB").Text()
 	}
 }
 
@@ -42,9 +40,9 @@ func (tag *ID3v2Tag) Album() string {
 func (tag *ID3v2Tag) Artist() string {
 	switch tag.Header.Version {
 	case V2_2:
-		return util.ToStringLossy(tag.GetFrame("TP1").Contents)
+		return tag.GetFrame("TP1").Text()
 	default:
-		return util.ToStringLossy(tag.GetFrame("TPE1").Contents)
+		return tag.GetFrame("TPE1").Text()
 	}
 }
 
@@ -52,9 +50,9 @@ func (tag *ID3v2Tag) Artist() string {
 func (tag *ID3v2Tag) Year() string {
 	switch tag.Header.Version {
 	case V2_2:
-		return util.ToStringLossy(tag.GetFrame("TYE").Contents)
+		return tag.GetFrame("TYE").Text()
 	default:
-		return util.ToStringLossy(tag.GetFrame("TYER").Contents)
+		return tag.GetFrame("TYER").Text()
 	}
 }
 
@@ -62,8 +60,8 @@ func (tag *ID3v2Tag) Year() string {
 func (tag *ID3v2Tag) Comment() string {
 	switch tag.Header.Version {
 	case V2_2:
-		return util.ToStringLossy(tag.GetFrame("COM").Contents)
+		return tag.GetFrame("COM").Text()
 	default:
-		return util.ToStringLossy(tag.GetFrame("COMM").Contents)
+		return tag.GetFrame("COMM").Text()
 	}
 }