Add PEF (Pentax Electronic Format) support

gen/main.go

@@ -25,7 +25,7 @@ func main() {
 		if err != nil {
 			return err
 		}
-		if info.IsDir() || strings.HasPrefix(info.Name(), ".") {
+		if info.IsDir() || strings.HasPrefix(info.Name(), ".") || strings.HasSuffix(info.Name(), ".txt") {
 			return nil
 		}
 

gen/testdata_exiftool/images/bep/jølstravatnet.pef.config.json

@@ -0,0 +1 @@
+{"width": 6040, "height": 4032}

gen/testdata_exiftool/images/sample.arw.json

@@ -1,12 +1,12 @@
 [{
-  "SourceFile": "testdata/images/sample.arw",
+  "SourceFile": "../testdata/images/sample.arw",
   "ExifTool": {
     "ExifToolVersion": 13.50
   },
   "File": {
     "FileName": "sample.arw",
-    "Directory": "testdata/images",
-    "FileSize": 9601024,
+    "Directory": "../testdata/images",
+    "FileSize": 102400,
     "FilePermissions": 100644,
     "FileType": "ARW",
     "FileTypeExtension": "ARW",

gen/testdata_exiftool/images/sample.cr2.json

@@ -1,12 +1,12 @@
 [{
-  "SourceFile": "testdata/images/sample.cr2",
+  "SourceFile": "../testdata/images/sample.cr2",
   "ExifTool": {
     "ExifToolVersion": 13.50
   },
   "File": {
     "FileName": "sample.cr2",
-    "Directory": "testdata/images",
-    "FileSize": 13287537,
+    "Directory": "../testdata/images",
+    "FileSize": 51200,
     "FilePermissions": 100644,
     "FileType": "CR2",
     "FileTypeExtension": "CR2",

gen/testdata_exiftool/images/sample.dng.json

@@ -1,12 +1,12 @@
 [{
-  "SourceFile": "testdata/images/sample.dng",
+  "SourceFile": "../testdata/images/sample.dng",
   "ExifTool": {
     "ExifToolVersion": 13.50
   },
   "File": {
     "FileName": "sample.dng",
-    "Directory": "testdata/images",
-    "FileSize": 6611520,
+    "Directory": "../testdata/images",
+    "FileSize": 51200,
     "FilePermissions": 100644,
     "FileType": "DNG",
     "FileTypeExtension": "DNG",

gen/testdata_exiftool/images/sample.nef.json

@@ -1,12 +1,12 @@
 [{
-  "SourceFile": "testdata/images/sample.nef",
+  "SourceFile": "../testdata/images/sample.nef",
   "ExifTool": {
     "ExifToolVersion": 13.50
   },
   "File": {
     "FileName": "sample.nef",
-    "Directory": "testdata/images",
-    "FileSize": 4055474,
+    "Directory": "../testdata/images",
+    "FileSize": 76800,
     "FilePermissions": 100644,
     "FileType": "NEF",
     "FileTypeExtension": "NEF",

imagemeta.go

@@ -25,7 +25,7 @@ const (
 	XMP
 
 	// CONFIG source, which currently the image dimensions encoded in the image.
-	// Note that this must not be confused with the dimensions stored in EXIF tags.
+	// Note that this may not be the same as the dimensions stored in EXIF tags.
 	CONFIG
 )
 
@@ -60,6 +60,8 @@ const (
 	NEF
 	// ARW is the Sony ARW RAW image format.
 	ARW
+	// PEF is the Pentax PEF RAW image format.
+	PEF
 )
 
 // ImageConfig contains basic image configuration.
@@ -75,7 +77,8 @@ type DecodeResult struct {
 	ImageConfig ImageConfig
 }
 
-// Decode reads EXIF and IPTC metadata from r and returns a Meta struct.
+// Decode reads the metadata and passes it to the provided handlers in the options.
+// It returns a DecodeResult containing the image configuration if the CONFIG source was requested.
 func Decode(opts Options) (result DecodeResult, err error) {
 	var base *baseStreamingDecoder
 
@@ -201,7 +204,7 @@ func Decode(opts Options) (result DecodeResult, err error) {
 		sourceSet = EXIF | XMP | IPTC | CONFIG
 	case HEIF, AVIF:
 		sourceSet = EXIF | XMP | CONFIG
-	case DNG, CR2, NEF, ARW:
+	case DNG, CR2, NEF, ARW, PEF:
 		sourceSet = EXIF | XMP | IPTC | CONFIG
 	default:
 		return result, fmt.Errorf("unsupported image format")
@@ -243,7 +246,7 @@ func Decode(opts Options) (result DecodeResult, err error) {
 		dec = &imageDecoderPNG{baseStreamingDecoder: base}
 	case HEIF, AVIF:
 		dec = &imageDecoderHEIF{baseStreamingDecoder: base}
-	case DNG, CR2, NEF, ARW:
+	case DNG, CR2, NEF, ARW, PEF:
 		dec = &imageDecoderRAW{baseStreamingDecoder: base}
 	}
 

imagemeta_fuzz_test.go

@@ -129,6 +129,16 @@ func FuzzDecodeARW(f *testing.F) {
 	})
 }
 
+func FuzzDecodePEF(f *testing.F) {
+	filenames := []string{"bep/jølstravatnet.pef"}
+	for _, filename := range filenames {
+		f.Add(readTestDataFileAll(f, filename))
+	}
+	f.Fuzz(func(t *testing.T, imageBytes []byte) {
+		fuzzDecodeBytes(t, imageBytes, imagemeta.PEF)
+	})
+}
+
 func fuzzDecodeBytes(t *testing.T, imageBytes []byte, f imagemeta.ImageFormat) error {
 	r := bytes.NewReader(imageBytes)
 	_, err := imagemeta.Decode(imagemeta.Options{R: r, ImageFormat: f, Sources: imagemeta.EXIF | imagemeta.IPTC | imagemeta.XMP | imagemeta.CONFIG, Timeout: 600 * time.Millisecond})

imagemeta_test.go

@@ -138,6 +138,12 @@ func TestDecodeRAW(t *testing.T) {
 	c.Assert(err, qt.IsNil)
 	c.Assert(tags.EXIF()["Make"].Value, qt.Equals, "SONY")
 	c.Assert(tags.EXIF()["Model"].Value, qt.Equals, "DSLR-A330")
+
+	// PEF (Pentax K-3 II)
+	_, tags, err = extractTags(t, "bep/jølstravatnet.pef", imagemeta.EXIF)
+	c.Assert(err, qt.IsNil)
+	c.Assert(tags.EXIF()["Make"].Value, qt.Equals, "RICOH IMAGING COMPANY, LTD.")
+	c.Assert(tags.EXIF()["Model"].Value, qt.Equals, "PENTAX K-3 II")
 }
 
 func TestDecodeRAWConfig(t *testing.T) {
@@ -149,10 +155,11 @@ func TestDecodeRAWConfig(t *testing.T) {
 		width    int
 		height   int
 	}{
-		{"sample.dng", imagemeta.DNG, 2592, 1944}, // DefaultCropSize
-		{"sample.cr2", imagemeta.CR2, 3648, 2736}, // ExifIFD dimensions
-		{"sample.nef", imagemeta.NEF, 2012, 1324}, // SubIFD dimensions
-		{"sample.arw", imagemeta.ARW, 3880, 2600}, // ExifIFD dimensions
+		{"sample.dng", imagemeta.DNG, 2592, 1944},            // DefaultCropSize
+		{"sample.cr2", imagemeta.CR2, 3648, 2736},            // ExifIFD dimensions
+		{"sample.nef", imagemeta.NEF, 2012, 1324},            // SubIFD dimensions
+		{"sample.arw", imagemeta.ARW, 3880, 2600},            // ExifIFD dimensions
+		{"bep/jølstravatnet.pef", imagemeta.PEF, 6080, 4032}, // IFD0 dimensions
 	}
 
 	for _, tt := range tests {
@@ -1061,6 +1068,8 @@ func extToFormat(ext string) imagemeta.ImageFormat {
 		return imagemeta.NEF
 	case ".arw":
 		return imagemeta.ARW
+	case ".pef":
+		return imagemeta.PEF
 	case ".txt":
 		return -1
 	default:
@@ -1192,7 +1201,7 @@ func withTestDataFile(t testing.TB, fn func(path string, info os.FileInfo, err e
 			return err
 		}
 
-		if info.IsDir() || strings.HasPrefix(info.Name(), ".") {
+		if info.IsDir() || strings.HasPrefix(info.Name(), ".") || strings.HasSuffix(info.Name(), ".txt") {
 			return nil
 		}
 
@@ -1210,6 +1219,7 @@ var goldenSkip = map[string]bool{
 	"invalid-encoding-usercomment.jpg": true, // The file has an EXIF error that produces a warning in imagemeta. It's tested separately.
 	"sample.cr2":                       true, // CR2 chains 4 IFDs; exiftool reports IFD3 StripByteCounts which we don't reach.
 	"sample.arw":                       true, // IFD0 0x0201/0x0202 are preview offsets; exiftool renames to PreviewImageStart/Length so values differ.
+	"bep/jølstravatnet.pef":            true, // GPSProcessingMethod includes encoding prefix; CONFIG dimensions differ (identify uses libraw active area).
 }
 
 var isSpaceDelimitedFloatRe = regexp.MustCompile(`^(\d+\.\d+)( \d+\.?\d*)+$`)
@@ -1431,6 +1441,7 @@ func BenchmarkDecode(b *testing.B) {
 		{"cr2", imagemeta.CR2, "sample.cr2"},
 		{"nef", imagemeta.NEF, "sample.nef"},
 		{"arw", imagemeta.ARW, "sample.arw"},
+		{"pef", imagemeta.PEF, "bep/jølstravatnet.pef"},
 	} {
 		imageFormat = tt.format
 		runBenchmarkWithFile(b, "exif/"+tt.name, tt.format, tt.filename, func(r io.ReadSeeker) error {

metadecoder_exif.go

@@ -145,7 +145,10 @@ var (
 		"BlackLevelRepeatDim":     exifConverters.convertNumbersToSpaceLimited,
 		"MaskedAreas":             exifConverters.convertNumbersToSpaceLimited,
 		"CFAPattern": func(ctx valueConverterContext, v any) any {
-			b := v.([]byte)
+			b, ok := typeAssert[[]byte](ctx, v)
+			if !ok {
+				return v
+			}
 			horizontalRepeat := ctx.s.byteOrder.Uint16(b[:2])
 			verticalRepeat := ctx.s.byteOrder.Uint16(b[2:])
 			repeatLen := int(horizontalRepeat) * int(verticalRepeat)
@@ -508,7 +511,9 @@ func (e *metaDecoderEXIF) decodeTag(namespace string) error {
 
 	if tagName == tagNameThumbnailOffset {
 		// When set, thumbnailOffset is set to the offset of the EXIF data in the original file.
-		val = val.(uint32) + uint32(e.readerOffset+e.thumbnailOffset)
+		if v, ok := typeAssert[uint32](e.valueConverterCtx, val); ok {
+			val = v + uint32(e.readerOffset+e.thumbnailOffset)
+		}
 	}
 
 	tagInfo.Value = val

testdata/fuzz/FuzzDecodeDNG/85bf2db7be71f22e

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("II*\x00\b\x00\x00\x000\x00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000i\x87\x04\x00\x01\x00\x00\x00\x84\x02\x00\x000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000 \x00000000000000000000000000000000000000000000000000000000000000000000000000000000000000\x02\xa3\x05\x00\x01\x00\x00\x00\x01\x04\x00\x00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000")

testdata/fuzz/FuzzDecodePEF/284853a7f4021859

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("II*\x00\b\x00\x00\x000\x00\x01\x02\f\x00\x00\x00\x00\x0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000")