Add test for zero length input and refactor some funcs

internal/matchers/archive.go

@@ -33,7 +33,7 @@ func Gzip(in []byte) bool {
 
 // Crx matches a Chrome extension file: a zip archive prepended by "Cr24".
 func Crx(in []byte) bool {
-	return len(in) > 4 && bytes.Equal(in[:4], []byte("Cr24"))
+	return bytes.HasPrefix(in, []byte("Cr24"))
 }
 
 // Tar matches a (t)ape (ar)chive file.

internal/matchers/audio.go

@@ -26,7 +26,7 @@ func Ape(in []byte) bool {
 
 // MusePack matches a Musepack file.
 func MusePack(in []byte) bool {
-	return len(in) > 4 && bytes.Equal(in[:4], []byte("MPCK"))
+	return bytes.HasPrefix(in, []byte("MPCK"))
 }
 
 // Wav matches a Waveform Audio File Format file.

internal/matchers/binary.go

@@ -6,20 +6,19 @@ import (
 
 // Class matches an java class file.
 func Class(in []byte) bool {
-	return len(in) > 4 && bytes.Equal(in[:4], []byte{0xCA, 0xFE, 0xBA, 0xBE})
+	return bytes.HasPrefix(in, []byte{0xCA, 0xFE, 0xBA, 0xBE})
 }
 
 // Swf matches an Adobe Flash swf file.
 func Swf(in []byte) bool {
-	return len(in) > 3 &&
-		bytes.Equal(in[:3], []byte("CWS")) ||
-		bytes.Equal(in[:3], []byte("FWS")) ||
-		bytes.Equal(in[:3], []byte("ZWS"))
+	return bytes.HasPrefix(in, []byte("CWS")) ||
+		bytes.HasPrefix(in, []byte("FWS")) ||
+		bytes.HasPrefix(in, []byte("ZWS"))
 }
 
 // Wasm matches a web assembly File Format file.
 func Wasm(in []byte) bool {
-	return len(in) > 4 && bytes.Equal(in[:4], []byte{0x00, 0x61, 0x73, 0x6D})
+	return bytes.HasPrefix(in, []byte{0x00, 0x61, 0x73, 0x6D})
 }
 
 // Dbf matches a dBase file.

internal/matchers/document.go

@@ -4,7 +4,7 @@ import "bytes"
 
 // Pdf matches a Portable Document Format file.
 func Pdf(in []byte) bool {
-	return len(in) > 4 && bytes.Equal(in[:4], []byte{0x25, 0x50, 0x44, 0x46})
+	return bytes.HasPrefix(in, []byte{0x25, 0x50, 0x44, 0x46})
 }
 
 // DjVu matches a DjVu file
@@ -15,7 +15,10 @@ func DjVu(in []byte) bool {
 	if len(in) < 15 {
 		return false
 	}
-	return bytes.HasPrefix(in[12:], []byte("DJVM")) || bytes.HasPrefix(in[12:], []byte("DJVU")) || bytes.HasPrefix(in[12:], []byte("DJVI")) || bytes.HasPrefix(in[12:], []byte("THUM"))
+	return bytes.HasPrefix(in[12:], []byte("DJVM")) ||
+		bytes.HasPrefix(in[12:], []byte("DJVU")) ||
+		bytes.HasPrefix(in[12:], []byte("DJVI")) ||
+		bytes.HasPrefix(in[12:], []byte("THUM"))
 }
 
 // Mobi matches a Mobi file

internal/matchers/font.go

@@ -4,7 +4,7 @@ import "bytes"
 
 // Woff matches a Web Open Font Format file.
 func Woff(in []byte) bool {
-	return len(in) > 4 && bytes.Equal(in[:4], []byte("wOFF"))
+	return bytes.HasPrefix(in, []byte("wOFF"))
 }
 
 // Woff2 matches a Web Open Font Format version 2 file.

internal/matchers/geo.go

@@ -40,5 +40,5 @@ func Shp(in []byte) bool {
 // Shx matches a shape index format file.
 // https://www.esri.com/library/whitepapers/pdfs/shapefile.pdf
 func Shx(in []byte) bool {
-	return len(in) > 4 && bytes.Equal(in[:4], []byte{0x00, 0x00, 0x27, 0x0A})
+	return bytes.HasPrefix(in, []byte{0x00, 0x00, 0x27, 0x0A})
 }

internal/matchers/image.go

@@ -4,13 +4,12 @@ import "bytes"
 
 // Png matches a Portable Network Graphics file.
 func Png(in []byte) bool {
-	return len(in) > 8 &&
-		bytes.Equal(in[:8], []byte{0x89, 0x50, 0x4E, 0x47, 0x0D, 0x0A, 0x1A, 0x0A})
+	return bytes.HasPrefix(in, []byte{0x89, 0x50, 0x4E, 0x47, 0x0D, 0x0A, 0x1A, 0x0A})
 }
 
 // Jpg matches a Joint Photographic Experts Group file.
 func Jpg(in []byte) bool {
-	return len(in) > 3 && bytes.Equal(in[:3], []byte{0xFF, 0xD8, 0xFF})
+	return bytes.HasPrefix(in, []byte{0xFF, 0xD8, 0xFF})
 }
 
 // Gif matches a Graphics Interchange Format file.
@@ -50,9 +49,8 @@ func Ico(in []byte) bool {
 
 // Tiff matches a Tagged Image File Format file.
 func Tiff(in []byte) bool {
-	return len(in) > 4 &&
-		(bytes.Equal(in[:4], []byte{0x49, 0x49, 0x2A, 0x00}) ||
-			bytes.Equal(in[:4], []byte{0x4D, 0x4D, 0x00, 0x2A}))
+	return bytes.HasPrefix(in, []byte{0x49, 0x49, 0x2A, 0x00}) ||
+		bytes.HasPrefix(in, []byte{0x4D, 0x4D, 0x00, 0x2A})
 }
 
 // Bpg matches a Better Portable Graphics file.
@@ -62,7 +60,7 @@ func Bpg(in []byte) bool {
 
 // Dwg matches a CAD drawing file.
 func Dwg(in []byte) bool {
-	if in[0] != 0x41 || in[1] != 0x43 || len(in) < 6 {
+	if len(in) < 6 || in[0] != 0x41 || in[1] != 0x43 {
 		return false
 	}
 	dwgVersions := [][]byte{

internal/matchers/text.go

@@ -158,6 +158,9 @@ func Json(in []byte) bool {
 //
 // BUG(gabriel-vasile): The "type" key should be searched for in the root object.
 func GeoJson(in []byte) bool {
+	if len(in) == 0 {
+		return false
+	}
 	in = trimLWS(in)
 	// geojson is always an object
 	if in[0] != '{' {
@@ -227,7 +230,7 @@ func Tcl(in []byte) bool {
 
 // Rtf matches a Rich Text Format file.
 func Rtf(in []byte) bool {
-	return len(in) > 6 && bytes.Equal(in[:6], []byte("{\\rtf1"))
+	return bytes.HasPrefix(in, []byte("{\\rtf1"))
 }
 
 // Svg matches a SVG file.

internal/matchers/video.go

@@ -52,7 +52,7 @@ func Flv(in []byte) bool {
 
 // Mpeg matches a Moving Picture Experts Group file.
 func Mpeg(in []byte) bool {
-	return len(in) > 3 && bytes.Equal(in[:3], []byte("\x00\x00\x01")) &&
+	return bytes.HasPrefix(in, []byte{0x00, 0x00, 0x01}) &&
 		in[3] >= 0xB0 && in[3] <= 0xBF
 }
 

mime_test.go

@@ -233,6 +233,12 @@ Extension | MIME type
 	}
 }
 
+func TestIndexOutOfRange(t *testing.T) {
+	for _, n := range root.flatten() {
+		_ = n.matchFunc(nil)
+	}
+}
+
 func BenchmarkMatchDetect(b *testing.B) {
 	files := []string{"png.png", "jpg.jpg", "pdf.pdf", "zip.zip", "docx.docx", "doc.doc"}
 	data, fLen := [][matchers.ReadLimit]byte{}, len(files)