tests/phpunit/includes/media/IPTCTest.php

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