thumbnails: extract pictures from audio files

go.mod

@@ -167,6 +167,7 @@ require (
 	github.com/desertbit/timer v0.0.0-20180107155436-c41aec40b27f // indirect
 	github.com/dgraph-io/ristretto v0.1.1 // indirect
 	github.com/dgryski/go-rendezvous v0.0.0-20200823014737-9f7001d12a5f // indirect
+	github.com/dhowden/tag v0.0.0-20230630033851-978a0926ee25 // indirect
 	github.com/dlclark/regexp2 v1.4.0 // indirect
 	github.com/docker/go-units v0.4.0 // indirect
 	github.com/dustin/go-humanize v1.0.1 // indirect

go.sum

@@ -1035,6 +1035,8 @@ github.com/dgryski/go-rendezvous v0.0.0-20200823014737-9f7001d12a5f/go.mod h1:cu
 github.com/dgryski/go-sip13 v0.0.0-20181026042036-e10d5fee7954/go.mod h1:vAd38F8PWV+bWy6jNmig1y/TA+kYO4g3RSRF0IAv0no=
 github.com/dgryski/trifles v0.0.0-20200323201526-dd97f9abfb48 h1:fRzb/w+pyskVMQ+UbP35JkH8yB7MYb4q/qhBarqZE6g=
 github.com/dgryski/trifles v0.0.0-20200323201526-dd97f9abfb48/go.mod h1:if7Fbed8SFyPtHLHbg49SI7NAdJiC5WIA09pe59rfAA=
+github.com/dhowden/tag v0.0.0-20230630033851-978a0926ee25 h1:simG0vMYFvNriGhaaat7QVVkaVkXzvqcohaBoLZl9Hg=
+github.com/dhowden/tag v0.0.0-20230630033851-978a0926ee25/go.mod h1:Z3Lomva4pyMWYezjMAU5QWRh0p1VvO4199OHlFnyKkM=
 github.com/dimchansky/utfbom v1.1.0/go.mod h1:rO41eb7gLfo8SF1jd9F8HplJm1Fewwi4mQvIirEdv+8=
 github.com/disintegration/imaging v1.6.2 h1:w1LecBlG2Lnp8B3jk5zSuNqd7b4DXhcjwek1ei82L+c=
 github.com/disintegration/imaging v1.6.2/go.mod h1:44/5580QXChDfwIclfc/PCwrr44amcmDAg8hxG0Ewe4=

services/thumbnails/pkg/preprocessor/lazyReadSeeker.go

@@ -0,0 +1,60 @@
+package preprocessor
+
+import (
+	"io"
+
+	"github.com/pkg/errors"
+)
+
+type LazyReadSeeker struct {
+	reader   io.Reader
+	buffer   []byte
+	position int64
+}
+
+func NewLazyReadSeeker(r io.Reader) *LazyReadSeeker {
+	return &LazyReadSeeker{
+		reader: r,
+		buffer: make([]byte, 0),
+	}
+}
+
+func (l *LazyReadSeeker) Read(p []byte) (int, error) {
+	// Fill buffer if necessary
+	if l.position >= int64(len(l.buffer)) {
+		temp := make([]byte, len(p))
+		n, err := l.reader.Read(temp)
+		if err != nil && err != io.EOF {
+			return 0, err
+		}
+		if n == 0 {
+			return 0, io.EOF
+		}
+		l.buffer = append(l.buffer, temp[:n]...)
+	}
+
+	// Read from buffer
+	n := copy(p, l.buffer[l.position:])
+	l.position += int64(n)
+
+	return n, nil
+}
+
+func (l *LazyReadSeeker) Seek(offset int64, whence int) (int64, error) {
+	var newPos int64
+	switch whence {
+	case io.SeekStart:
+		newPos = offset
+	case io.SeekCurrent:
+		newPos = l.position + offset
+	default:
+		return 0, errors.New("seekEnd is not supported")
+	}
+
+	if newPos < 0 {
+		return 0, errors.New("negative position is invalid")
+	}
+
+	l.position = newPos
+	return l.position, nil
+}

services/thumbnails/pkg/preprocessor/preprocessor.go

@@ -2,10 +2,12 @@ package preprocessor
 
 import (
 	"bufio"
+	"bytes"
 	"image"
 	"image/draw"
 	"image/gif"
 	"io"
+	"log"
 	"math"
 	"mime"
 	"strings"
@@ -15,6 +17,8 @@ import (
 	"golang.org/x/image/font"
 	"golang.org/x/image/font/opentype"
 	"golang.org/x/image/math/fixed"
+
+	"github.com/dhowden/tag"
 )
 
 type FileConverter interface {
@@ -41,6 +45,28 @@ func (i GifDecoder) Convert(r io.Reader) (interface{}, error) {
 	return img, nil
 }
 
+type AudioDecoder struct{}
+
+func (i AudioDecoder) Convert(r io.Reader) (interface{}, error) {
+	readSeeker := NewLazyReadSeeker(r)
+	m, err := tag.ReadFrom(readSeeker)
+	if err != nil {
+		log.Fatal(err)
+	}
+
+	picture := m.Picture()
+	if picture == nil {
+		return nil, errors.New(`could not extract image from audio file`)
+	}
+
+	converter := ForType(picture.MIMEType, nil)
+	if converter == nil {
+		return nil, errors.New(`could not find converter for image extraced from audio file`)
+	}
+
+	return converter.Convert(bytes.NewReader(picture.Data))
+}
+
 type TxtToImageConverter struct {
 	fontLoader *FontLoader
 }
@@ -197,6 +223,10 @@ func ForType(mimeType string, opts map[string]interface{}) FileConverter {
 		}
 	case "image/gif":
 		return GifDecoder{}
+	case "audio/flac":
+	case "audio/mpeg":
+	case "audio/ogg":
+		return AudioDecoder{}
 	default:
 		return ImageDecoder{}
 	}

services/thumbnails/pkg/thumbnail/thumbnail.go

@@ -21,6 +21,9 @@ var (
 		"image/x-ms-bmp": {},
 		"image/tiff":     {},
 		"text/plain":     {},
+		"audio/flac":     {},
+		"audio/mpeg":     {},
+		"audio/ogg":      {},
 	}
 )