tests/phpunit/includes/media/JpegMetadataExtractorTest.php

   1 <?php
   2 /**
   3  * @todo Could use a test of extended XMP segments. Hard to find programs that
   4  * create example files, and creating my own in vim probably wouldn't
   5  * serve as a very good "test". (Adobe photoshop probably creates such files
   6  * but it costs money). The implementation of it currently in MediaWiki is based
   7  * solely on reading the standard, without any real world test files.
   8  *
   9  * @group Media
  10  * @covers \JpegMetadataExtractor
  11  */
  12 class JpegMetadataExtractorTest extends MediaWikiIntegrationTestCase {
  13         private const FILE_PATH = __DIR__ . '/../../data/media/';
  14
  15         /**
  16          * We also use this test to test padding bytes don't
  17          * screw stuff up
  18          *
  19          * @param string $file Filename
  20          *
  21          * @dataProvider provideUtf8Comment
  22          */
  23         public function testUtf8Comment( $file ) {
  24                 $res = JpegMetadataExtractor::segmentSplitter( self::FILE_PATH . $file );
  25                 $this->assertEquals( [ 'UTF-8 JPEG Comment — ¼' ], $res['COM'] );
  26         }
  27
  28         public static function provideUtf8Comment() {
  29                 return [
  30                         [ 'jpeg-comment-utf.jpg' ],
  31                         [ 'jpeg-padding-even.jpg' ],
  32                         [ 'jpeg-padding-odd.jpg' ],
  33                 ];
  34         }
  35
  36         /** The file is iso-8859-1, but it should get auto converted */
  37         public function testIso88591Comment() {
  38                 $res = JpegMetadataExtractor::segmentSplitter( self::FILE_PATH . 'jpeg-comment-iso8859-1.jpg' );
  39                 $this->assertEquals( [ 'ISO-8859-1 JPEG Comment - ¼' ], $res['COM'] );
  40         }
  41
  42         /** Comment values that are non-textual (random binary junk) should not be shown.
  43          * The example test file has a comment with a 0x5 byte in it which is a control character
  44          * and considered binary junk for our purposes.
  45          */
  46         public function testBinaryCommentStripped() {
  47                 $res = JpegMetadataExtractor::segmentSplitter( self::FILE_PATH . 'jpeg-comment-binary.jpg' );
  48                 $this->assertSame( [], $res['COM'] );
  49         }
  50
  51         /* Very rarely a file can have multiple comments.
  52          *   Order of comments is based on order inside the file.
  53          */
  54         public function testMultipleComment() {
  55                 $res = JpegMetadataExtractor::segmentSplitter( self::FILE_PATH . 'jpeg-comment-multiple.jpg' );
  56                 $this->assertEquals( [ 'foo', 'bar' ], $res['COM'] );
  57         }
  58
  59         public function testXMPExtraction() {
  60                 $res = JpegMetadataExtractor::segmentSplitter( self::FILE_PATH . 'jpeg-xmp-psir.jpg' );
  61                 $expected = file_get_contents( self::FILE_PATH . 'jpeg-xmp-psir.xmp' );
  62                 $this->assertEquals( $expected, $res['XMP'] );
  63         }
  64
  65         public function testPSIRExtraction() {
  66                 $res = JpegMetadataExtractor::segmentSplitter( self::FILE_PATH . 'jpeg-xmp-psir.jpg' );
  67                 $expected = '50686f746f73686f7020332e30003842494d04040000000'
  68                         . '000181c02190004746573741c02190003666f6f1c020000020004';
  69                 $this->assertEquals( $expected, bin2hex( $res['PSIR'][0] ) );
  70         }
  71
  72         public function testXMPExtractionNullChar() {
  73                 $res = JpegMetadataExtractor::segmentSplitter( self::FILE_PATH . 'jpeg-xmp-nullchar.jpg' );
  74                 $expected = file_get_contents( self::FILE_PATH . 'jpeg-xmp-psir.xmp' );
  75                 $this->assertEquals( $expected, $res['XMP'] );
  76         }
  77
  78         public function testXMPExtractionAltAppId() {
  79                 $res = JpegMetadataExtractor::segmentSplitter( self::FILE_PATH . 'jpeg-xmp-alt.jpg' );
  80                 $expected = file_get_contents( self::FILE_PATH . 'jpeg-xmp-psir.xmp' );
  81                 $this->assertEquals( $expected, $res['XMP'] );
  82         }
  83
  84         public function testIPTCHashComparisionNoHash() {
  85                 $segments = JpegMetadataExtractor::segmentSplitter( self::FILE_PATH . 'jpeg-xmp-psir.jpg' );
  86                 $res = JpegMetadataExtractor::doPSIR( $segments['PSIR'][0] );
  87
  88                 $this->assertEquals( 'iptc-no-hash', $res );
  89         }
  90
  91         public function testIPTCHashComparisionBadHash() {
  92                 $segments = JpegMetadataExtractor::segmentSplitter( self::FILE_PATH . 'jpeg-iptc-bad-hash.jpg' );
  93                 $res = JpegMetadataExtractor::doPSIR( $segments['PSIR'][0] );
  94
  95                 $this->assertEquals( 'iptc-bad-hash', $res );
  96         }
  97
  98         public function testIPTCHashComparisionGoodHash() {
  99                 $segments = JpegMetadataExtractor::segmentSplitter( self::FILE_PATH . 'jpeg-iptc-good-hash.jpg' );
 100                 $res = JpegMetadataExtractor::doPSIR( $segments['PSIR'][0] );
 101
 102                 $this->assertEquals( 'iptc-good-hash', $res );
 103         }
 104
 105         public function testExifByteOrder() {
 106                 $res = JpegMetadataExtractor::segmentSplitter( self::FILE_PATH . 'exif-user-comment.jpg' );
 107                 $expected = 'BE';
 108                 $this->assertEquals( $expected, $res['byteOrder'] );
 109         }
 110
 111         public function testInfiniteRead() {
 112                 // test file truncated right after a segment, which previously
 113                 // caused an infinite loop looking for the next segment byte.
 114                 // Should get past infinite loop and throw in StringUtils::unpack()
 115                 $this->expectException( InvalidJpegException::class );
 116                 $res = JpegMetadataExtractor::segmentSplitter( self::FILE_PATH . 'jpeg-segment-loop1.jpg' );
 117         }
 118
 119         public function testInfiniteRead2() {
 120                 // test file truncated after a segment's marker and size, which
 121                 // would cause a seek past end of file. Seek past end of file
 122                 // doesn't actually fail, but prevents further reading and was
 123                 // devolving into the previous case (testInfiniteRead).
 124                 $this->expectException( InvalidJpegException::class );
 125                 $res = JpegMetadataExtractor::segmentSplitter( self::FILE_PATH . 'jpeg-segment-loop2.jpg' );
 126         }
 127 }