Support PDF files with magic byte in the header

vips/foreign.go

@@ -6,13 +6,12 @@ import (
 	"bytes"
 	"encoding/xml"
 	"fmt"
+	"golang.org/x/image/bmp"
+	"golang.org/x/net/html/charset"
 	"image/png"
 	"math"
 	"runtime"
 	"unsafe"
-
-	"golang.org/x/image/bmp"
-	"golang.org/x/net/html/charset"
 )
 
 // SubsampleMode correlates to a libvips subsample mode
@@ -154,14 +153,14 @@ func DetermineImageType(buf []byte) ImageType {
 		return ImageTypeHEIF
 	} else if isSVG(buf) {
 		return ImageTypeSVG
-	} else if isPDF(buf) {
-		return ImageTypePDF
 	} else if isBMP(buf) {
 		return ImageTypeBMP
 	} else if isJP2K(buf) {
 		return ImageTypeJP2K
 	} else if isJXL(buf) {
 		return ImageTypeJXL
+	} else if isPDF(buf) {
+		return ImageTypePDF
 	} else {
 		return ImageTypeUnknown
 	}
@@ -240,7 +239,10 @@ func isSVG(buf []byte) bool {
 var pdf = []byte("\x25\x50\x44\x46")
 
 func isPDF(buf []byte) bool {
-	return bytes.HasPrefix(buf, pdf)
+	if len(buf) <= 1024 {
+		return bytes.Contains(buf, pdf)
+	}
+	return bytes.Contains(buf[:1024], pdf)
 }
 
 var bmpHeader = []byte("BM")

vips/foreign_test.go

@@ -106,6 +106,17 @@ func Test_DetermineImageType__PDF(t *testing.T) {
 	assert.Equal(t, ImageTypePDF, imageType)
 }
 
+func Test_DetermineImageType__PDF_1(t *testing.T) {
+	Startup(&Config{})
+
+	buf, err := os.ReadFile(resources + "PDF-2.0-with-offset-start.pdf")
+	assert.NoError(t, err)
+	assert.NotNil(t, buf)
+
+	imageType := DetermineImageType(buf)
+	assert.Equal(t, ImageTypePDF, imageType)
+}
+
 func Test_DetermineImageType__BMP(t *testing.T) {
 	Startup(&Config{})
 

vips/image_golden_test.go

@@ -954,6 +954,14 @@ func TestImage_Pixelate(t *testing.T) {
 		nil, nil)
 }
 
+func TestPDF_WithOffsetStart(t *testing.T) {
+	goldenTest(t, resources+"PDF-2.0-with-offset-start.pdf",
+		nil, func(img *ImageRef) {
+			assert.Equal(t, 612, img.Width())
+			assert.Equal(t, 396, img.Height())
+		}, nil)
+}
+
 func testWebpOptimizeIccProfile(t *testing.T, exportParams *WebpExportParams) []byte {
 	return goldenTest(t, resources+"has-icc-profile.png",
 		func(img *ImageRef) error {

vips/image_test.go

@@ -1190,6 +1190,17 @@ func TestImageRef_SetGamma(t *testing.T) {
 	require.NoError(t, err)
 }
 
+func Test_NewImageFromFile(t *testing.T) {
+	Startup(nil)
+
+	image, err := NewImageFromFile(resources + "PDF-2.0-with-offset-start.pdf")
+	require.NoError(t, err)
+
+	assert.Equal(t, ImageTypePDF, image.originalFormat)
+	assert.Equal(t, ImageTypePDF, image.format)
+	assert.Equal(t, 1, image.Pages())
+}
+
 // TODO unit tests to cover:
 // NewImageFromReader failing test
 // NewImageFromFile failing test