tests/phpunit/includes/media/IPTCTest.php

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