MediaWiki
REL1_19
|
00001 <?php 00002 class IPTCTest extends MediaWikiTestCase { 00003 public function testRecognizeUtf8() { 00004 // utf-8 is the only one used in practise. 00005 $res = IPTC::getCharset( "\x1b%G" ); 00006 $this->assertEquals( 'UTF-8', $res ); 00007 } 00008 00009 public function testIPTCParseNoCharset88591() { 00010 // basically IPTC for keyword with value of 0xBC which is 1/4 in iso-8859-1 00011 // This data doesn't specify a charset. We're supposed to guess 00012 // (which basically means utf-8 if valid, windows 1252 (iso 8859-1) if not) 00013 $iptcData = "Photoshop 3.0\08BIM\4\4\0\0\0\0\0\x06\x1c\x02\x19\x00\x01\xBC"; 00014 $res = IPTC::Parse( $iptcData ); 00015 $this->assertEquals( array( '¼' ), $res['Keywords'] ); 00016 } 00017 /* This one contains a sequence that's valid iso 8859-1 but not valid utf8 */ 00018 /* \xC3 = Ã, \xB8 = ¸ */ 00019 public function testIPTCParseNoCharset88591b() { 00020 $iptcData = "Photoshop 3.0\08BIM\4\4\0\0\0\0\0\x09\x1c\x02\x19\x00\x04\xC3\xC3\xC3\xB8"; 00021 $res = IPTC::Parse( $iptcData ); 00022 $this->assertEquals( array( 'ÃÃø' ), $res['Keywords'] ); 00023 } 00024 /* Same as testIPTCParseNoCharset88591b, but forcing the charset to utf-8. 00025 * What should happen is the first "\xC3\xC3" should be dropped as invalid, 00026 * leaving \xC3\xB8, which is ø 00027 */ 00028 public function testIPTCParseForcedUTFButInvalid() { 00029 $iptcData = "Photoshop 3.0\08BIM\4\4\0\0\0\0\0\x11\x1c\x02\x19\x00\x04\xC3\xC3\xC3\xB8" 00030 . "\x1c\x01\x5A\x00\x03\x1B\x25\x47"; 00031 $res = IPTC::Parse( $iptcData ); 00032 $this->assertEquals( array( 'ø' ), $res['Keywords'] ); 00033 } 00034 public function testIPTCParseNoCharsetUTF8() { 00035 $iptcData = "Photoshop 3.0\08BIM\4\4\0\0\0\0\0\x07\x1c\x02\x19\x00\x02¼"; 00036 $res = IPTC::Parse( $iptcData ); 00037 $this->assertEquals( array( '¼' ), $res['Keywords'] ); 00038 } 00039 // Testing something that has 2 values for keyword 00040 public function testIPTCParseMulti() { 00041 $iptcData = /* identifier */ "Photoshop 3.0\08BIM\4\4" 00042 /* length */ . "\0\0\0\0\0\x0D" 00043 . "\x1c\x02\x19" . "\x00\x01" . "\xBC" 00044 . "\x1c\x02\x19" . "\x00\x02" . "\xBC\xBD"; 00045 $res = IPTC::Parse( $iptcData ); 00046 $this->assertEquals( array( '¼', '¼½' ), $res['Keywords'] ); 00047 } 00048 public function testIPTCParseUTF8() { 00049 // This has the magic "\x1c\x01\x5A\x00\x03\x1B\x25\x47" which marks content as UTF8. 00050 $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"; 00051 $res = IPTC::Parse( $iptcData ); 00052 $this->assertEquals( array( '¼' ), $res['Keywords'] ); 00053 } 00054 00055 }