diff options
Diffstat (limited to 'tests/phpunit/includes/parser/PreprocessorTest.php')
| -rw-r--r-- | tests/phpunit/includes/parser/PreprocessorTest.php | 195 |
1 files changed, 195 insertions, 0 deletions
diff --git a/tests/phpunit/includes/parser/PreprocessorTest.php b/tests/phpunit/includes/parser/PreprocessorTest.php new file mode 100644 index 00000000..7a5948d4 --- /dev/null +++ b/tests/phpunit/includes/parser/PreprocessorTest.php @@ -0,0 +1,195 @@ +<?php + +class PreprocessorTest extends MediaWikiTestCase { + var $mTitle = 'Page title'; + var $mPPNodeCount = 0; + var $mOptions; + + function setUp() { + global $wgParserConf; + $this->mOptions = new ParserOptions(); + $name = isset( $wgParserConf['preprocessorClass'] ) ? $wgParserConf['preprocessorClass'] : 'Preprocessor_DOM'; + + $this->mPreprocessor = new $name( $this ); + } + + function getStripList() { + return array( 'gallery', 'display map' /* Used by Maps, see r80025 CR */, '/foo' ); + } + + function provideCases() { + return array( + array( "Foo", "<root>Foo</root>" ), + array( "<!-- Foo -->", "<root><comment><!-- Foo --></comment></root>" ), + array( "<!-- Foo --><!-- Bar -->", "<root><comment><!-- Foo --></comment><comment><!-- Bar --></comment></root>" ), + array( "<!-- Foo --> <!-- Bar -->", "<root><comment><!-- Foo --></comment> <comment><!-- Bar --></comment></root>" ), + array( "<!-- Foo --> \n <!-- Bar -->", "<root><comment><!-- Foo --></comment> \n <comment><!-- Bar --></comment></root>" ), + array( "<!-- Foo --> \n <!-- Bar -->\n", "<root><comment><!-- Foo --></comment> \n<comment> <!-- Bar -->\n</comment></root>" ), + array( "<!-- Foo --> <!-- Bar -->\n", "<root><comment><!-- Foo --></comment> <comment><!-- Bar --></comment>\n</root>" ), + array( "<!-->Bar", "<root><comment><!-->Bar</comment></root>" ), + array( "<!-- Comment -- comment", "<root><comment><!-- Comment -- comment</comment></root>" ), + array( "== Foo ==\n <!-- Bar -->\n== Baz ==\n", "<root><h level=\"2\" i=\"1\">== Foo ==</h>\n<comment> <!-- Bar -->\n</comment><h level=\"2\" i=\"2\">== Baz ==</h>\n</root>" ), + array( "<gallery/>", "<root><ext><name>gallery</name><attr></attr></ext></root>" ), + array( "Foo <gallery/> Bar", "<root>Foo <ext><name>gallery</name><attr></attr></ext> Bar</root>" ), + array( "<gallery></gallery>", "<root><ext><name>gallery</name><attr></attr><inner></inner><close></gallery></close></ext></root>" ), + array( "<foo> <gallery></gallery>", "<root><foo> <ext><name>gallery</name><attr></attr><inner></inner><close></gallery></close></ext></root>" ), + array( "<foo> <gallery><gallery></gallery>", "<root><foo> <ext><name>gallery</name><attr></attr><inner><gallery></inner><close></gallery></close></ext></root>" ), + array( "<noinclude> Foo bar </noinclude>", "<root><ignore><noinclude></ignore> Foo bar <ignore></noinclude></ignore></root>" ), + array( "<noinclude>\n{{Foo}}\n</noinclude>", "<root><ignore><noinclude></ignore>\n<template lineStart=\"1\"><title>Foo</title></template>\n<ignore></noinclude></ignore></root>" ), + array( "<noinclude>\n{{Foo}}\n</noinclude>\n", "<root><ignore><noinclude></ignore>\n<template lineStart=\"1\"><title>Foo</title></template>\n<ignore></noinclude></ignore>\n</root>" ), + array( "<gallery>foo bar", "<root><ext><name>gallery</name><attr></attr><inner>foo bar</inner></ext></root>" ), + array( "<{{foo}}>", "<root><<template><title>foo</title></template>></root>" ), + array( "<{{{foo}}}>", "<root><<tplarg><title>foo</title></tplarg>></root>" ), + array( "<gallery></gallery</gallery>", "<root><ext><name>gallery</name><attr></attr><inner></gallery</inner><close></gallery></close></ext></root>" ), + array( "=== Foo === ", "<root><h level=\"3\" i=\"1\">=== Foo === </h></root>" ), + array( "==<!-- -->= Foo === ", "<root><h level=\"2\" i=\"1\">==<comment><!-- --></comment>= Foo === </h></root>" ), + array( "=== Foo ==<!-- -->= ", "<root><h level=\"1\" i=\"1\">=== Foo ==<comment><!-- --></comment>= </h></root>" ), + array( "=== Foo ===<!-- -->\n", "<root><h level=\"3\" i=\"1\">=== Foo ===<comment><!-- --></comment></h>\n</root>" ), + array( "=== Foo ===<!-- --> <!-- -->\n", "<root><h level=\"3\" i=\"1\">=== Foo ===<comment><!-- --></comment> <comment><!-- --></comment></h>\n</root>" ), + array( "== Foo ==\n== Bar == \n", "<root><h level=\"2\" i=\"1\">== Foo ==</h>\n<h level=\"2\" i=\"2\">== Bar == </h>\n</root>" ), + array( "===========", "<root><h level=\"5\" i=\"1\">===========</h></root>" ), + array( "Foo\n=\n==\n=\n", "<root>Foo\n=\n==\n=\n</root>" ), + array( "{{Foo}}", "<root><template lineStart=\"1\"><title>Foo</title></template></root>" ), + array( "\n{{Foo}}", "<root>\n<template lineStart=\"1\"><title>Foo</title></template></root>" ), + array( "{{Foo|bar}}", "<root><template lineStart=\"1\"><title>Foo</title><part><name index=\"1\" /><value>bar</value></part></template></root>" ), + array( "{{Foo|bar}}a", "<root><template lineStart=\"1\"><title>Foo</title><part><name index=\"1\" /><value>bar</value></part></template>a</root>" ), + array( "{{Foo|bar|baz}}", "<root><template lineStart=\"1\"><title>Foo</title><part><name index=\"1\" /><value>bar</value></part><part><name index=\"2\" /><value>baz</value></part></template></root>" ), + array( "{{Foo|1=bar}}", "<root><template lineStart=\"1\"><title>Foo</title><part><name>1</name>=<value>bar</value></part></template></root>" ), + array( "{{Foo|=bar}}", "<root><template lineStart=\"1\"><title>Foo</title><part><name></name>=<value>bar</value></part></template></root>" ), + array( "{{Foo|bar=baz}}", "<root><template lineStart=\"1\"><title>Foo</title><part><name>bar</name>=<value>baz</value></part></template></root>" ), + array( "{{Foo|1=bar|baz}}", "<root><template lineStart=\"1\"><title>Foo</title><part><name>1</name>=<value>bar</value></part><part><name index=\"1\" /><value>baz</value></part></template></root>" ), + array( "{{Foo|1=bar|2=baz}}", "<root><template lineStart=\"1\"><title>Foo</title><part><name>1</name>=<value>bar</value></part><part><name>2</name>=<value>baz</value></part></template></root>" ), + array( "{{Foo|bar|foo=baz}}", "<root><template lineStart=\"1\"><title>Foo</title><part><name index=\"1\" /><value>bar</value></part><part><name>foo</name>=<value>baz</value></part></template></root>" ), + array( "{{{1}}}", "<root><tplarg lineStart=\"1\"><title>1</title></tplarg></root>" ), + array( "{{{1|}}}", "<root><tplarg lineStart=\"1\"><title>1</title><part><name index=\"1\" /><value></value></part></tplarg></root>" ), + array( "{{{Foo}}}", "<root><tplarg lineStart=\"1\"><title>Foo</title></tplarg></root>" ), + array( "{{{Foo|}}}", "<root><tplarg lineStart=\"1\"><title>Foo</title><part><name index=\"1\" /><value></value></part></tplarg></root>" ), + array( "{{{Foo|bar|baz}}}", "<root><tplarg lineStart=\"1\"><title>Foo</title><part><name index=\"1\" /><value>bar</value></part><part><name index=\"2\" /><value>baz</value></part></tplarg></root>" ), + array( "{<!-- -->{Foo}}", "<root>{<comment><!-- --></comment>{Foo}}</root>" ), + array( "{{{{Foobar}}}}", "<root>{<tplarg><title>Foobar</title></tplarg>}</root>" ), + array( "{{{ {{Foo}} }}}", "<root><tplarg lineStart=\"1\"><title> <template><title>Foo</title></template> </title></tplarg></root>" ), + array( "{{ {{{Foo}}} }}", "<root><template lineStart=\"1\"><title> <tplarg><title>Foo</title></tplarg> </title></template></root>" ), + array( "{{{{{Foo}}}}}", "<root><template lineStart=\"1\"><title><tplarg><title>Foo</title></tplarg></title></template></root>" ), + array( "{{{{{Foo}} }}}", "<root><tplarg lineStart=\"1\"><title><template><title>Foo</title></template> </title></tplarg></root>" ), + array( "{{{{{{Foo}}}}}}", "<root><tplarg lineStart=\"1\"><title><tplarg><title>Foo</title></tplarg></title></tplarg></root>" ), + array( "{{{{{{Foo}}}}}", "<root>{<template><title><tplarg><title>Foo</title></tplarg></title></template></root>" ), + array( "[[[Foo]]", "<root>[[[Foo]]</root>" ), + array( "{{Foo|[[[[bar]]|baz]]}}", "<root><template lineStart=\"1\"><title>Foo</title><part><name index=\"1\" /><value>[[[[bar]]|baz]]</value></part></template></root>" ), // This test is important, since it means the difference between having the [[ rule stacked or not + array( "{{Foo|[[[[bar]|baz]]}}", "<root>{{Foo|[[[[bar]|baz]]}}</root>" ), + array( "{{Foo|Foo [[[[bar]|baz]]}}", "<root>{{Foo|Foo [[[[bar]|baz]]}}</root>" ), + array( "Foo <display map>Bar</display map >Baz", "<root>Foo <ext><name>display map</name><attr></attr><inner>Bar</inner><close></display map ></close></ext>Baz</root>" ), + array( "Foo <display map foo>Bar</display map >Baz", "<root>Foo <ext><name>display map</name><attr> foo</attr><inner>Bar</inner><close></display map ></close></ext>Baz</root>" ), + array( "Foo <gallery bar=\"baz\" />", "<root>Foo <ext><name>gallery</name><attr> bar="baz" </attr></ext></root>" ), + array( "</foo>Foo<//foo>", "<root><ext><name>/foo</name><attr></attr><inner>Foo</inner><close><//foo></close></ext></root>" ), # Worth blacklisting IMHO + array( "{{#ifexpr: ({{{1|1}}} = 2) | Foo | Bar }}", "<root><template lineStart=\"1\"><title>#ifexpr: (<tplarg><title>1</title><part><name index=\"1\" /><value>1</value></part></tplarg> = 2) </title><part><name index=\"1\" /><value> Foo </value></part><part><name index=\"2\" /><value> Bar </value></part></template></root>"), + array( "{{#if: {{{1|}}} | Foo | {{Bar}} }}", "<root><template lineStart=\"1\"><title>#if: <tplarg><title>1</title><part><name index=\"1\" /><value></value></part></tplarg> </title><part><name index=\"1\" /><value> Foo </value></part><part><name index=\"2\" /><value> <template><title>Bar</title></template> </value></part></template></root>"), + array( "{{#if: {{{1|}}} | Foo | [[Bar]] }}", "<root><template lineStart=\"1\"><title>#if: <tplarg><title>1</title><part><name index=\"1\" /><value></value></part></tplarg> </title><part><name index=\"1\" /><value> Foo </value></part><part><name index=\"2\" /><value> [[Bar]] </value></part></template></root>"), + array( "{{#if: {{{1|}}} | [[Foo]] | Bar }}", "<root><template lineStart=\"1\"><title>#if: <tplarg><title>1</title><part><name index=\"1\" /><value></value></part></tplarg> </title><part><name index=\"1\" /><value> [[Foo]] </value></part><part><name index=\"2\" /><value> Bar </value></part></template></root>"), + array( "{{#if: {{{1|}}} | 1 | {{#if: {{{1|}}} | 2 | 3 }} }}", "<root><template lineStart=\"1\"><title>#if: <tplarg><title>1</title><part><name index=\"1\" /><value></value></part></tplarg> </title><part><name index=\"1\" /><value> 1 </value></part><part><name index=\"2\" /><value> <template><title>#if: <tplarg><title>1</title><part><name index=\"1\" /><value></value></part></tplarg> </title><part><name index=\"1\" /><value> 2 </value></part><part><name index=\"2\" /><value> 3 </value></part></template> </value></part></template></root>"), + array( "{{ {{Foo}}", "<root>{{ <template><title>Foo</title></template></root>"), + array( "{{Foobar {{Foo}} {{Bar}} {{Baz}} ", "<root>{{Foobar <template><title>Foo</title></template> <template><title>Bar</title></template> <template><title>Baz</title></template> </root>"), + array( "[[Foo]] |", "<root>[[Foo]] |</root>"), + array( "{{Foo|Bar|", "<root>{{Foo|Bar|</root>"), + array( "[[Foo]", "<root>[[Foo]</root>"), + array( "[[Foo|Bar]", "<root>[[Foo|Bar]</root>"), + array( "{{Foo| [[Bar] }}", "<root>{{Foo| [[Bar] }}</root>"), + array( "{{Foo| [[Bar|Baz] }}", "<root>{{Foo| [[Bar|Baz] }}</root>"), + array( "{{Foo|bar=[[baz]}}", "<root>{{Foo|bar=[[baz]}}</root>"), + array( "{{foo|", "<root>{{foo|</root>"), + array( "{{foo|}", "<root>{{foo|}</root>"), + array( "{{foo|} }}", "<root><template lineStart=\"1\"><title>foo</title><part><name index=\"1\" /><value>} </value></part></template></root>"), + array( "{{foo|bar=|}", "<root>{{foo|bar=|}</root>"), + array( "{{Foo|} Bar=", "<root>{{Foo|} Bar=</root>"), + array( "{{Foo|} Bar=}}", "<root><template lineStart=\"1\"><title>Foo</title><part><name>} Bar</name>=<value></value></part></template></root>"), + /* array( file_get_contents( dirname( __FILE__ ) . '/QuoteQuran.txt' ), file_get_contents( dirname( __FILE__ ) . '/QuoteQuranExpanded.txt' ) ), */ + ); + } + + /** + * @dataProvider provideCases + */ + function testPreprocessorOutput( $wikiText, $expectedXml ) { + $this->assertEquals( $expectedXml, $this->mPreprocessor->preprocessToXml( $wikiText ) ); + } + + /** + * These are more complex test cases taken out of wiki articles. + */ + function provideFiles() { + return array( + array( "QuoteQuran" ), # http://en.wikipedia.org/w/index.php?title=Template:QuoteQuran/sandbox&oldid=237348988 GFDL + CC-BY-SA by Striver + array( "Factorial" ), # http://en.wikipedia.org/w/index.php?title=Template:Factorial&oldid=98548758 GFDL + CC-BY-SA by Polonium + array( "All_system_messages" ), # http://tl.wiktionary.org/w/index.php?title=Suleras:All_system_messages&oldid=2765 GPL text generated by MediaWiki + array( "Fundraising" ), # http://tl.wiktionary.org/w/index.php?title=MediaWiki:Sitenotice&oldid=5716 GFDL + CC-BY-SA, copied there by Sky Harbor. + ); + } + + /** + * @dataProvider provideFiles + */ + function testPreprocessorOutputFiles( $filename ) { + $folder = dirname( __FILE__ ) . "/../../../parser/preprocess"; + $wikiText = file_get_contents( "$folder/$filename.txt" ); + $output = $this->mPreprocessor->preprocessToXml( $wikiText ); + + $expectedFilename = "$folder/$filename.expected"; + if ( file_exists( $expectedFilename ) ) { + $this->assertStringEqualsFile( $expectedFilename, $output ); + } else { + $tempFilename = tempnam( $folder, "$filename." ); + file_put_contents( $tempFilename, $output ); + $this->markTestIncomplete( "File $expectedFilename missing. Output stored as $tempFilename" ); + } + } + + /** + * Tests from Bug 28642 ยท https://bugzilla.wikimedia.org/28642 + */ + function provideHeadings() { + return array( /* These should become headings: */ + array( "== h ==<!--c1-->", "<root><h level=\"2\" i=\"1\">== h ==<comment><!--c1--></comment></h></root>" ), + array( "== h == <!--c1-->", "<root><h level=\"2\" i=\"1\">== h == <comment><!--c1--></comment></h></root>" ), + array( "== h ==<!--c1--> ", "<root><h level=\"2\" i=\"1\">== h ==<comment><!--c1--></comment> </h></root>" ), + array( "== h == <!--c1--> ", "<root><h level=\"2\" i=\"1\">== h == <comment><!--c1--></comment> </h></root>" ), + array( "== h ==<!--c1--><!--c2-->", "<root><h level=\"2\" i=\"1\">== h ==<comment><!--c1--></comment><comment><!--c2--></comment></h></root>" ), + array( "== h == <!--c1--><!--c2-->", "<root><h level=\"2\" i=\"1\">== h == <comment><!--c1--></comment><comment><!--c2--></comment></h></root>" ), + array( "== h ==<!--c1--><!--c2--> ", "<root><h level=\"2\" i=\"1\">== h ==<comment><!--c1--></comment><comment><!--c2--></comment> </h></root>" ), + array( "== h == <!--c1--><!--c2--> ", "<root><h level=\"2\" i=\"1\">== h == <comment><!--c1--></comment><comment><!--c2--></comment> </h></root>" ), + array( "== h == <!--c1--> <!--c2-->", "<root><h level=\"2\" i=\"1\">== h == <comment><!--c1--></comment> <comment><!--c2--></comment></h></root>" ), + array( "== h ==<!--c1--> <!--c2--> ", "<root><h level=\"2\" i=\"1\">== h ==<comment><!--c1--></comment> <comment><!--c2--></comment> </h></root>" ), + array( "== h == <!--c1--> <!--c2--> ", "<root><h level=\"2\" i=\"1\">== h == <comment><!--c1--></comment> <comment><!--c2--></comment> </h></root>" ), + array( "== h ==<!--c1--><!--c2--><!--c3-->", "<root><h level=\"2\" i=\"1\">== h ==<comment><!--c1--></comment><comment><!--c2--></comment><comment><!--c3--></comment></h></root>" ), + array( "== h ==<!--c1--> <!--c2--><!--c3-->", "<root><h level=\"2\" i=\"1\">== h ==<comment><!--c1--></comment> <comment><!--c2--></comment><comment><!--c3--></comment></h></root>" ), + array( "== h ==<!--c1--><!--c2--> <!--c3-->", "<root><h level=\"2\" i=\"1\">== h ==<comment><!--c1--></comment><comment><!--c2--></comment> <comment><!--c3--></comment></h></root>" ), + array( "== h ==<!--c1--> <!--c2--> <!--c3-->", "<root><h level=\"2\" i=\"1\">== h ==<comment><!--c1--></comment> <comment><!--c2--></comment> <comment><!--c3--></comment></h></root>" ), + array( "== h == <!--c1--><!--c2--><!--c3-->", "<root><h level=\"2\" i=\"1\">== h == <comment><!--c1--></comment><comment><!--c2--></comment><comment><!--c3--></comment></h></root>" ), + array( "== h == <!--c1--> <!--c2--><!--c3-->", "<root><h level=\"2\" i=\"1\">== h == <comment><!--c1--></comment> <comment><!--c2--></comment><comment><!--c3--></comment></h></root>" ), + array( "== h == <!--c1--><!--c2--> <!--c3-->", "<root><h level=\"2\" i=\"1\">== h == <comment><!--c1--></comment><comment><!--c2--></comment> <comment><!--c3--></comment></h></root>" ), + array( "== h == <!--c1--> <!--c2--> <!--c3-->", "<root><h level=\"2\" i=\"1\">== h == <comment><!--c1--></comment> <comment><!--c2--></comment> <comment><!--c3--></comment></h></root>" ), + array( "== h ==<!--c1--><!--c2--><!--c3--> ", "<root><h level=\"2\" i=\"1\">== h ==<comment><!--c1--></comment><comment><!--c2--></comment><comment><!--c3--></comment> </h></root>" ), + array( "== h ==<!--c1--> <!--c2--><!--c3--> ", "<root><h level=\"2\" i=\"1\">== h ==<comment><!--c1--></comment> <comment><!--c2--></comment><comment><!--c3--></comment> </h></root>" ), + array( "== h ==<!--c1--><!--c2--> <!--c3--> ", "<root><h level=\"2\" i=\"1\">== h ==<comment><!--c1--></comment><comment><!--c2--></comment> <comment><!--c3--></comment> </h></root>" ), + array( "== h ==<!--c1--> <!--c2--> <!--c3--> ", "<root><h level=\"2\" i=\"1\">== h ==<comment><!--c1--></comment> <comment><!--c2--></comment> <comment><!--c3--></comment> </h></root>" ), + array( "== h == <!--c1--><!--c2--><!--c3--> ", "<root><h level=\"2\" i=\"1\">== h == <comment><!--c1--></comment><comment><!--c2--></comment><comment><!--c3--></comment> </h></root>" ), + array( "== h == <!--c1--> <!--c2--><!--c3--> ", "<root><h level=\"2\" i=\"1\">== h == <comment><!--c1--></comment> <comment><!--c2--></comment><comment><!--c3--></comment> </h></root>" ), + array( "== h == <!--c1--><!--c2--> <!--c3--> ", "<root><h level=\"2\" i=\"1\">== h == <comment><!--c1--></comment><comment><!--c2--></comment> <comment><!--c3--></comment> </h></root>" ), + array( "== h == <!--c1--> <!--c2--> <!--c3--> ", "<root><h level=\"2\" i=\"1\">== h == <comment><!--c1--></comment> <comment><!--c2--></comment> <comment><!--c3--></comment> </h></root>" ), + + /* These are not working: */ + array( "== h ==<!--c1--> <!--c2-->", "<root>== h ==<comment><!--c1--></comment> <comment><!--c2--></comment></root>" ), + array( "== h == <!--c1--> <!--c2-->", "<root>== h == <comment><!--c1--></comment> <comment><!--c2--></comment></root>" ), + array( "== h ==<!--c1--> <!--c2--> ", "<root>== h ==<comment><!--c1--></comment> <comment><!--c2--></comment> </root>" ), + array( "== h == x <!--c1--><!--c2--><!--c3--> ", "<root>== h == x <comment><!--c1--></comment><comment><!--c2--></comment><comment><!--c3--></comment> </root>" ), + array( "== h ==<!--c1--> x <!--c2--><!--c3--> ", "<root>== h ==<comment><!--c1--></comment> x <comment><!--c2--></comment><comment><!--c3--></comment> </root>" ), + array( "== h ==<!--c1--><!--c2--><!--c3--> x ", "<root>== h ==<comment><!--c1--></comment><comment><!--c2--></comment><comment><!--c3--></comment> x </root>" ), + ); + } + + /** + * @dataProvider provideHeadings + */ + function testHeadings( $wikiText, $expectedXml ) { + $this->assertEquals( $expectedXml, $this->mPreprocessor->preprocessToXml( $wikiText ) ); + } +} + |