// Copyright 2021 The Gitea Authors. All rights reserved. // SPDX-License-Identifier: MIT package typesniffer import ( "bytes" "encoding/base64" "encoding/hex" "strings" "testing" "github.com/stretchr/testify/assert" ) func TestDetectContentTypeLongerThanSniffLen(t *testing.T) { // Pre-condition: Shorter than sniffLen detects SVG. assert.Equal(t, "image/svg+xml", DetectContentType([]byte(``)).contentType) // Longer than sniffLen detects something else. assert.NotEqual(t, "image/svg+xml", DetectContentType([]byte(``)).contentType) } func TestIsTextFile(t *testing.T) { assert.True(t, DetectContentType([]byte{}).IsText()) assert.True(t, DetectContentType([]byte("lorem ipsum")).IsText()) } func TestIsSvgImage(t *testing.T) { assert.True(t, DetectContentType([]byte("")).IsSvgImage()) assert.True(t, DetectContentType([]byte(" ")).IsSvgImage()) assert.True(t, DetectContentType([]byte(``)).IsSvgImage()) assert.True(t, DetectContentType([]byte(``)).IsSvgImage()) assert.True(t, DetectContentType([]byte(` `)).IsSvgImage()) assert.True(t, DetectContentType([]byte(` `)).IsSvgImage()) assert.True(t, DetectContentType([]byte(` `)).IsSvgImage()) assert.True(t, DetectContentType([]byte(` `)).IsSvgImage()) assert.True(t, DetectContentType([]byte(` `)).IsSvgImage()) assert.True(t, DetectContentType([]byte(` `)).IsSvgImage()) assert.True(t, DetectContentType([]byte(` `)).IsSvgImage()) assert.True(t, DetectContentType([]byte(` `)).IsSvgImage()) // the DetectContentType should work for incomplete data, because only beginning bytes are used for detection assert.True(t, DetectContentType([]byte(`....`)).IsSvgImage()) assert.False(t, DetectContentType([]byte{}).IsSvgImage()) assert.False(t, DetectContentType([]byte("svg")).IsSvgImage()) assert.False(t, DetectContentType([]byte("")).IsSvgImage()) assert.False(t, DetectContentType([]byte("text")).IsSvgImage()) assert.False(t, DetectContentType([]byte("")).IsSvgImage()) assert.False(t, DetectContentType([]byte(``)).IsSvgImage()) assert.False(t, DetectContentType([]byte(` `)).IsSvgImage()) assert.False(t, DetectContentType([]byte(` `)).IsSvgImage()) assert.False(t, DetectContentType([]byte(`
`)).IsSvgImage()) assert.False(t, DetectContentType([]byte(`
`)).IsSvgImage()) assert.False(t, DetectContentType([]byte(``)).IsSvgImage()) assert.False(t, DetectContentType([]byte(``)).IsSvgImage()) } func TestIsPDF(t *testing.T) { pdf, _ := base64.StdEncoding.DecodeString("JVBERi0xLjYKJcOkw7zDtsOfCjIgMCBvYmoKPDwvTGVuZ3RoIDMgMCBSL0ZpbHRlci9GbGF0ZURlY29kZT4+CnN0cmVhbQp4nF3NPwsCMQwF8D2f4s2CNYk1baF0EHRwOwg4iJt/NsFb/PpevUE4Mjwe") assert.True(t, DetectContentType(pdf).IsPDF()) assert.False(t, DetectContentType([]byte("plain text")).IsPDF()) } func TestIsVideo(t *testing.T) { mp4, _ := base64.StdEncoding.DecodeString("AAAAGGZ0eXBtcDQyAAAAAGlzb21tcDQyAAEI721vb3YAAABsbXZoZAAAAADaBlwX2gZcFwAAA+gA") assert.True(t, DetectContentType(mp4).IsVideo()) assert.False(t, DetectContentType([]byte("plain text")).IsVideo()) } func TestIsAudio(t *testing.T) { mp3, _ := base64.StdEncoding.DecodeString("SUQzBAAAAAABAFRYWFgAAAASAAADbWFqb3JfYnJhbmQAbXA0MgBUWFhYAAAAEQAAA21pbm9yX3Zl") assert.True(t, DetectContentType(mp3).IsAudio()) assert.False(t, DetectContentType([]byte("plain text")).IsAudio()) assert.True(t, DetectContentType([]byte("ID3Toy\000")).IsAudio()) assert.True(t, DetectContentType([]byte("ID3Toy\n====\t* hi 🌞, ...")).IsText()) // test ID3 tag for plain text assert.True(t, DetectContentType([]byte("ID3Toy\n====\t* hi 🌞, ..."+"🌛"[0:2])).IsText()) // test ID3 tag with incomplete UTF8 char } func TestDetectContentTypeFromReader(t *testing.T) { mp3, _ := base64.StdEncoding.DecodeString("SUQzBAAAAAABAFRYWFgAAAASAAADbWFqb3JfYnJhbmQAbXA0MgBUWFhYAAAAEQAAA21pbm9yX3Zl") st, err := DetectContentTypeFromReader(bytes.NewReader(mp3)) assert.NoError(t, err) assert.True(t, st.IsAudio()) } func TestDetectContentTypeOgg(t *testing.T) { oggAudio, _ := hex.DecodeString("4f67675300020000000000000000352f0000000000007dc39163011e01766f72626973000000000244ac0000000000000071020000000000b8014f6767530000") st, err := DetectContentTypeFromReader(bytes.NewReader(oggAudio)) assert.NoError(t, err) assert.True(t, st.IsAudio()) oggVideo, _ := hex.DecodeString("4f676753000200000000000000007d9747ef000000009b59daf3012a807468656f7261030201001e00110001e000010e00020000001e00000001000001000001") st, err = DetectContentTypeFromReader(bytes.NewReader(oggVideo)) assert.NoError(t, err) assert.True(t, st.IsVideo()) } func TestDetectFileTypeBox(t *testing.T) { _, found := detectFileTypeBox([]byte("\x00\x00\xff\xffftypAAAA....")) assert.False(t, found) brands, found := detectFileTypeBox([]byte("\x00\x00\x00\x0cftypAAAA")) assert.True(t, found) assert.Equal(t, []string{"AAAA"}, brands) brands, found = detectFileTypeBox([]byte("\x00\x00\x00\x10ftypAAAA....BBBB")) assert.True(t, found) assert.Equal(t, []string{"AAAA"}, brands) brands, found = detectFileTypeBox([]byte("\x00\x00\x00\x14ftypAAAA....BBBB")) assert.True(t, found) assert.Equal(t, []string{"AAAA", "BBBB"}, brands) _, found = detectFileTypeBox([]byte("\x00\x00\x00\x14ftypAAAA....BBB")) assert.False(t, found) brands, found = detectFileTypeBox([]byte("\x00\x00\x00\x13ftypAAAA....BBB")) assert.True(t, found) assert.Equal(t, []string{"AAAA"}, brands) } func TestDetectContentTypeAvif(t *testing.T) { buf := []byte("\x00\x00\x00\x20ftypavif.......................") st := DetectContentType(buf) assert.Equal(t, MimeTypeImageAvif, st.contentType) }