summaryrefslogtreecommitdiff
path: root/tests/phpunit/includes/media/IPTCTest.php
diff options
context:
space:
mode:
authorPierre Schmitz <pierre@archlinux.de>2013-09-04 05:51:59 +0200
committerPierre Schmitz <pierre@archlinux.de>2013-09-04 05:51:59 +0200
commit91e194556c52d2f354344f930419eef2dd6267f0 (patch)
tree0cd12490d3cd3499274017c9b799d0f738d3719e /tests/phpunit/includes/media/IPTCTest.php
parent08aa4418c30cfc18ccc69a0f0f9cb9e17be6c196 (diff)
Update to MediaWiki 1.21.2
Diffstat (limited to 'tests/phpunit/includes/media/IPTCTest.php')
-rw-r--r--tests/phpunit/includes/media/IPTCTest.php60
1 files changed, 60 insertions, 0 deletions
diff --git a/tests/phpunit/includes/media/IPTCTest.php b/tests/phpunit/includes/media/IPTCTest.php
new file mode 100644
index 00000000..c9648a79
--- /dev/null
+++ b/tests/phpunit/includes/media/IPTCTest.php
@@ -0,0 +1,60 @@
+<?php
+class IPTCTest extends MediaWikiTestCase {
+ public function testRecognizeUtf8() {
+ // utf-8 is the only one used in practise.
+ $res = IPTC::getCharset( "\x1b%G" );
+ $this->assertEquals( 'UTF-8', $res );
+ }
+
+ public function testIPTCParseNoCharset88591() {
+ // basically IPTC for keyword with value of 0xBC which is 1/4 in iso-8859-1
+ // This data doesn't specify a charset. We're supposed to guess
+ // (which basically means utf-8 if valid, windows 1252 (iso 8859-1) if not)
+ $iptcData = "Photoshop 3.0\08BIM\4\4\0\0\0\0\0\x06\x1c\x02\x19\x00\x01\xBC";
+ $res = IPTC::Parse( $iptcData );
+ $this->assertEquals( array( '¼' ), $res['Keywords'] );
+ }
+
+ /* This one contains a sequence that's valid iso 8859-1 but not valid utf8 */
+ /* \xC3 = Ã, \xB8 = ¸ */
+ public function testIPTCParseNoCharset88591b() {
+ $iptcData = "Photoshop 3.0\08BIM\4\4\0\0\0\0\0\x09\x1c\x02\x19\x00\x04\xC3\xC3\xC3\xB8";
+ $res = IPTC::Parse( $iptcData );
+ $this->assertEquals( array( 'ÃÃø' ), $res['Keywords'] );
+ }
+
+ /* Same as testIPTCParseNoCharset88591b, but forcing the charset to utf-8.
+ * What should happen is the first "\xC3\xC3" should be dropped as invalid,
+ * leaving \xC3\xB8, which is ø
+ */
+ public function testIPTCParseForcedUTFButInvalid() {
+ $iptcData = "Photoshop 3.0\08BIM\4\4\0\0\0\0\0\x11\x1c\x02\x19\x00\x04\xC3\xC3\xC3\xB8"
+ . "\x1c\x01\x5A\x00\x03\x1B\x25\x47";
+ $res = IPTC::Parse( $iptcData );
+ $this->assertEquals( array( 'ø' ), $res['Keywords'] );
+ }
+
+ public function testIPTCParseNoCharsetUTF8() {
+ $iptcData = "Photoshop 3.0\08BIM\4\4\0\0\0\0\0\x07\x1c\x02\x19\x00\x02¼";
+ $res = IPTC::Parse( $iptcData );
+ $this->assertEquals( array( '¼' ), $res['Keywords'] );
+ }
+
+ // Testing something that has 2 values for keyword
+ public function testIPTCParseMulti() {
+ $iptcData = /* identifier */ "Photoshop 3.0\08BIM\4\4"
+ /* length */ . "\0\0\0\0\0\x0D"
+ . "\x1c\x02\x19" . "\x00\x01" . "\xBC"
+ . "\x1c\x02\x19" . "\x00\x02" . "\xBC\xBD";
+ $res = IPTC::Parse( $iptcData );
+ $this->assertEquals( array( '¼', '¼½' ), $res['Keywords'] );
+ }
+
+ public function testIPTCParseUTF8() {
+ // This has the magic "\x1c\x01\x5A\x00\x03\x1B\x25\x47" which marks content as UTF8.
+ $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";
+ $res = IPTC::Parse( $iptcData );
+ $this->assertEquals( array( '¼' ), $res['Keywords'] );
+ }
+
+}