summaryrefslogtreecommitdiff
path: root/tests/phpunit/includes/CollationTest.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/CollationTest.php
parent08aa4418c30cfc18ccc69a0f0f9cb9e17be6c196 (diff)
Update to MediaWiki 1.21.2
Diffstat (limited to 'tests/phpunit/includes/CollationTest.php')
-rw-r--r--tests/phpunit/includes/CollationTest.php109
1 files changed, 109 insertions, 0 deletions
diff --git a/tests/phpunit/includes/CollationTest.php b/tests/phpunit/includes/CollationTest.php
new file mode 100644
index 00000000..c746208b
--- /dev/null
+++ b/tests/phpunit/includes/CollationTest.php
@@ -0,0 +1,109 @@
+<?php
+class CollationTest extends MediaWikiLangTestCase {
+ protected function setUp() {
+ parent::setUp();
+ if ( !extension_loaded( 'intl' ) ) {
+ $this->markTestSkipped( 'These tests require intl extension' );
+ }
+ }
+
+ /**
+ * Test to make sure, that if you
+ * have "X" and "XY", the binary
+ * sortkey also has "X" being a
+ * prefix of "XY". Our collation
+ * code makes this assumption.
+ *
+ * @param $lang String Language code for collator
+ * @param $base String Base string
+ * @param $extended String String containing base as a prefix.
+ *
+ * @dataProvider prefixDataProvider
+ */
+ function testIsPrefix( $lang, $base, $extended ) {
+ $cp = Collator::create( $lang );
+ $cp->setStrength( Collator::PRIMARY );
+ $baseBin = $cp->getSortKey( $base );
+ // Remove sortkey terminator
+ $baseBin = rtrim( $baseBin, "\0" );
+ $extendedBin = $cp->getSortKey( $extended );
+ $this->assertStringStartsWith( $baseBin, $extendedBin, "$base is not a prefix of $extended" );
+ }
+
+ function prefixDataProvider() {
+ return array(
+ array( 'en', 'A', 'AA' ),
+ array( 'en', 'A', 'AAA' ),
+ array( 'en', 'Д', 'ДЂ' ),
+ array( 'en', 'Д', 'ДA' ),
+ // 'Ʒ' should expand to 'Z ' (note space).
+ array( 'fi', 'Z', 'Ʒ' ),
+ // 'Þ' should expand to 'th'
+ array( 'sv', 't', 'Þ' ),
+ // Javanese is a limited use alphabet, so should have 3 bytes
+ // per character, so do some tests with it.
+ array( 'en', 'ꦲ', 'ꦲꦤ' ),
+ array( 'en', 'ꦲ', 'ꦲД' ),
+ array( 'en', 'A', 'Aꦲ' ),
+ );
+ }
+ /**
+ * Opposite of testIsPrefix
+ *
+ * @dataProvider notPrefixDataProvider
+ */
+ function testNotIsPrefix( $lang, $base, $extended ) {
+ $cp = Collator::create( $lang );
+ $cp->setStrength( Collator::PRIMARY );
+ $baseBin = $cp->getSortKey( $base );
+ // Remove sortkey terminator
+ $baseBin = rtrim( $baseBin, "\0" );
+ $extendedBin = $cp->getSortKey( $extended );
+ $this->assertStringStartsNotWith( $baseBin, $extendedBin, "$base is a prefix of $extended" );
+ }
+
+ function notPrefixDataProvider() {
+ return array(
+ array( 'en', 'A', 'B' ),
+ array( 'en', 'AC', 'ABC' ),
+ array( 'en', 'Z', 'Ʒ' ),
+ array( 'en', 'A', 'ꦲ' ),
+ );
+ }
+
+ /**
+ * Test correct first letter is fetched.
+ *
+ * @param $collation String Collation name (aka uca-en)
+ * @param $string String String to get first letter of
+ * @param $firstLetter String Expected first letter.
+ *
+ * @dataProvider firstLetterProvider
+ */
+ function testGetFirstLetter( $collation, $string, $firstLetter ) {
+ $col = Collation::factory( $collation );
+ $this->assertEquals( $firstLetter, $col->getFirstLetter( $string ) );
+ }
+ function firstLetterProvider() {
+ return array(
+ array( 'uppercase', 'Abc', 'A' ),
+ array( 'uppercase', 'abc', 'A' ),
+ array( 'identity', 'abc', 'a' ),
+ array( 'uca-en', 'abc', 'A' ),
+ array( 'uca-en', ' ', ' ' ),
+ array( 'uca-en', 'Êveryone', 'E' ),
+ array( 'uca-vi', 'Êveryone', 'Ê' ),
+ // Make sure thorn is not a first letter.
+ array( 'uca-sv', 'The', 'T' ),
+ array( 'uca-sv', 'Å', 'Å' ),
+ array( 'uca-hu', 'dzsdo', 'Dzs' ),
+ array( 'uca-hu', 'dzdso', 'Dz' ),
+ array( 'uca-hu', 'CSD', 'Cs' ),
+ array( 'uca-root', 'CSD', 'C' ),
+ array( 'uca-fi', 'Ǥ', 'G' ),
+ array( 'uca-fi', 'Ŧ', 'T' ),
+ array( 'uca-fi', 'Ʒ', 'Z' ),
+ array( 'uca-fi', 'Ŋ', 'N' ),
+ );
+ }
+}