tests/phpunit/includes/media/IPTCTest.php

   1 <?php
   2
   3 /**
   4  * @group Media
   5  */
   6 class IPTCTest extends MediaWikiTestCase {
   7
   8         /**
   9          * @covers IPTC::getCharset
  10          */
  11         public function testRecognizeUtf8() {
  12                 // utf-8 is the only one used in practise.
  13                 $res = IPTC::getCharset( "\x1b%G" );
  14                 $this->assertEquals( 'UTF-8', $res );
  15         }
  16
  17         /**
  18          * @covers IPTC::parse
  19          */
  20         public function testIPTCParseNoCharset88591() {
  21                 // basically IPTC for keyword with value of 0xBC which is 1/4 in iso-8859-1
  22                 // This data doesn't specify a charset. We're supposed to guess
  23                 // (which basically means utf-8 if valid, windows 1252 (iso 8859-1) if not)
  24                 $iptcData = "Photoshop 3.0\08BIM\4\4\0\0\0\0\0\x06\x1c\x02\x19\x00\x01\xBC";
  25                 $res = IPTC::parse( $iptcData );
  26                 $this->assertEquals( [ '¼' ], $res['Keywords'] );
  27         }
  28
  29         /**
  30          * @covers IPTC::parse
  31          */
  32         public function testIPTCParseNoCharset88591b() {
  33                 /* This one contains a sequence that's valid iso 8859-1 but not valid utf8 */
  34                 /* \xC3 = Ã, \xB8 = ¸  */
  35                 $iptcData = "Photoshop 3.0\08BIM\4\4\0\0\0\0\0\x09\x1c\x02\x19\x00\x04\xC3\xC3\xC3\xB8";
  36                 $res = IPTC::parse( $iptcData );
  37                 $this->assertEquals( [ 'ÃÃÃ¸' ], $res['Keywords'] );
  38         }
  39
  40         /**
  41          * Same as testIPTCParseNoCharset88591b, but forcing the charset to utf-8.
  42          * What should happen is the first "\xC3\xC3" should be dropped as invalid,
  43          * leaving \xC3\xB8, which is ø
  44          * @covers IPTC::parse
  45          */
  46         public function testIPTCParseForcedUTFButInvalid() {
  47                 $iptcData = "Photoshop 3.0\08BIM\4\4\0\0\0\0\0\x11\x1c\x02\x19\x00\x04\xC3\xC3\xC3\xB8"
  48                         . "\x1c\x01\x5A\x00\x03\x1B\x25\x47";
  49                 $res = IPTC::parse( $iptcData );
  50                 $this->assertEquals( [ 'ø' ], $res['Keywords'] );
  51         }
  52
  53         /**
  54          * @covers IPTC::parse
  55          */
  56         public function testIPTCParseNoCharsetUTF8() {
  57                 $iptcData = "Photoshop 3.0\08BIM\4\4\0\0\0\0\0\x07\x1c\x02\x19\x00\x02¼";
  58                 $res = IPTC::parse( $iptcData );
  59                 $this->assertEquals( [ '¼' ], $res['Keywords'] );
  60         }
  61
  62         /**
  63          * Testing something that has 2 values for keyword
  64          * @covers IPTC::parse
  65          */
  66         public function testIPTCParseMulti() {
  67                 $iptcData = /* identifier */ "Photoshop 3.0\08BIM\4\4"
  68                         /* length */ . "\0\0\0\0\0\x0D"
  69                         . "\x1c\x02\x19" . "\x00\x01" . "\xBC"
  70                         . "\x1c\x02\x19" . "\x00\x02" . "\xBC\xBD";
  71                 $res = IPTC::parse( $iptcData );
  72                 $this->assertEquals( [ '¼', '¼½' ], $res['Keywords'] );
  73         }
  74
  75         /**
  76          * @covers IPTC::parse
  77          */
  78         public function testIPTCParseUTF8() {
  79                 // This has the magic "\x1c\x01\x5A\x00\x03\x1B\x25\x47" which marks content as UTF8.
  80                 $iptcData =
  81                         "Photoshop 3.0\08BIM\4\4\0\0\0\0\0\x0F\x1c\x02\x19\x00\x02¼\x1c\x01\x5A\x00\x03\x1B\x25\x47";
  82                 $res = IPTC::parse( $iptcData );
  83                 $this->assertEquals( [ '¼' ], $res['Keywords'] );
  84         }
  85 }