some improvements

pkg/scan/helper.go

@@ -3,24 +3,52 @@
 package scan
 
 import (
+	"bytes"
 	"net/http"
 	"os"
-	"path/filepath"
-	"strings"
 )
 
 const (
-	pdfContentType = "application/pdf"
-	maxSniffSize   = 512
+	pdfT         = "application/pdf"
+	jpegT        = "image/jpeg"
+	pngT         = "image/png"
+	bmpT         = "image/bmp"
+	gifT         = "image/gif"
+	webpT        = "image/webp"
+	pnmT         = "image/x-portable-anymap"
+	jp2T         = "image/jp2"
+	tiffT        = "image/tiff"
+	maxSniffSize = 512
 )
 
-func isPDF(path string) bool {
-	// TODO
-	// return checkContentType(path) || checkExtension(path)
-	return checkContentType(path)
+type extendedSig struct {
+	sig []byte
+	ct  string
 }
 
-func checkContentType(path string) bool {
+var extendedSigs = []extendedSig{
+	{[]byte("P1\x0A"), pnmT},
+	{[]byte("P2\x0A"), pnmT},
+	{[]byte("P3\x0A"), pnmT},
+	{[]byte("P4\x0A"), pnmT},
+	{[]byte("P5\x0A"), pnmT},
+	{[]byte("P6\x0A"), pnmT},
+	{[]byte("P7\x0A"), pnmT},
+	{[]byte("\x00\x00\x00\x0C\x6A\x50\x20\x20\x0D\x0A\x87\x0A\x00\x00\x00\x14\x66\x74\x79\x70\x6A\x70\x32"), jp2T},
+	{[]byte("\x4d\x4d\x00\x2a"), tiffT},
+	{[]byte("\x49\x49\x2a\x00"), tiffT},
+}
+
+func extendedContentType(buf []byte) string {
+	for _, sig := range extendedSigs {
+		if bytes.HasPrefix(buf, sig.sig) {
+			return sig.ct
+		}
+	}
+	return "application/octet-stream"
+}
+
+func isPDF(path string) bool {
 	f, err := os.Open(path)
 	if err != nil {
 		return false
@@ -31,9 +59,28 @@ func checkContentType(path string) bool {
 	if err != nil {
 		return false
 	}
-	return http.DetectContentType(buf) == pdfContentType
+	return http.DetectContentType(buf) == pdfT
 }
 
-func checkExtension(path string) bool {
-	return strings.ToLower(filepath.Ext(path)) == pdfExt
+func isImage(path string) bool {
+	f, err := os.Open(path)
+	if err != nil {
+		return false
+	}
+	defer f.Close()
+	buf := make([]byte, maxSniffSize)
+	_, err = f.Read(buf)
+	if err != nil {
+		return false
+	}
+	switch http.DetectContentType(buf) {
+	case jpegT, pngT, bmpT, gifT, webpT:
+		return true
+	default:
+		switch extendedContentType(buf) {
+		case pnmT, jp2T, tiffT:
+			return true
+		}
+	}
+	return false
 }

pkg/scan/scanner.go

@@ -106,6 +106,10 @@ func (s *Scanner) Scan(file string) (filename, text string, err error) {
 	var pdffile string
 	pdf := isPDF(scanfile)
 	if !pdf {
+		if !isImage(scanfile) {
+			err = errors.New("error: input file is not an image")
+			return
+		}
 		var (
 			pdf []byte
 			f   *os.File