tests/phpunit/includes/parser/ParserOutputTest.php

   1 <?php
   2
   3 use Wikimedia\TestingAccessWrapper;
   4
   5 /**
   6  * @group Database
   7  *        ^--- trigger DB shadowing because we are using Title magic
   8  */
   9 class ParserOutputTest extends MediaWikiTestCase {
  10
  11         public static function provideIsLinkInternal() {
  12                 return [
  13                         // Different domains
  14                         [ false, 'http://example.org', 'http://mediawiki.org' ],
  15                         // Same domains
  16                         [ true, 'http://example.org', 'http://example.org' ],
  17                         [ true, 'https://example.org', 'https://example.org' ],
  18                         [ true, '//example.org', '//example.org' ],
  19                         // Same domain different cases
  20                         [ true, 'http://example.org', 'http://EXAMPLE.ORG' ],
  21                         // Paths, queries, and fragments are not relevant
  22                         [ true, 'http://example.org', 'http://example.org/wiki/Main_Page' ],
  23                         [ true, 'http://example.org', 'http://example.org?my=query' ],
  24                         [ true, 'http://example.org', 'http://example.org#its-a-fragment' ],
  25                         // Different protocols
  26                         [ false, 'http://example.org', 'https://example.org' ],
  27                         [ false, 'https://example.org', 'http://example.org' ],
  28                         // Protocol relative servers always match http and https links
  29                         [ true, '//example.org', 'http://example.org' ],
  30                         [ true, '//example.org', 'https://example.org' ],
  31                         // But they don't match strange things like this
  32                         [ false, '//example.org', 'irc://example.org' ],
  33                 ];
  34         }
  35
  36         /**
  37          * Test to make sure ParserOutput::isLinkInternal behaves properly
  38          * @dataProvider provideIsLinkInternal
  39          * @covers ParserOutput::isLinkInternal
  40          */
  41         public function testIsLinkInternal( $shouldMatch, $server, $url ) {
  42                 $this->assertEquals( $shouldMatch, ParserOutput::isLinkInternal( $server, $url ) );
  43         }
  44
  45         /**
  46          * @covers ParserOutput::setExtensionData
  47          * @covers ParserOutput::getExtensionData
  48          */
  49         public function testExtensionData() {
  50                 $po = new ParserOutput();
  51
  52                 $po->setExtensionData( "one", "Foo" );
  53
  54                 $this->assertEquals( "Foo", $po->getExtensionData( "one" ) );
  55                 $this->assertNull( $po->getExtensionData( "spam" ) );
  56
  57                 $po->setExtensionData( "two", "Bar" );
  58                 $this->assertEquals( "Foo", $po->getExtensionData( "one" ) );
  59                 $this->assertEquals( "Bar", $po->getExtensionData( "two" ) );
  60
  61                 $po->setExtensionData( "one", null );
  62                 $this->assertNull( $po->getExtensionData( "one" ) );
  63                 $this->assertEquals( "Bar", $po->getExtensionData( "two" ) );
  64         }
  65
  66         /**
  67          * @covers ParserOutput::setProperty
  68          * @covers ParserOutput::getProperty
  69          * @covers ParserOutput::unsetProperty
  70          * @covers ParserOutput::getProperties
  71          */
  72         public function testProperties() {
  73                 $po = new ParserOutput();
  74
  75                 $po->setProperty( 'foo', 'val' );
  76
  77                 $properties = $po->getProperties();
  78                 $this->assertEquals( $po->getProperty( 'foo' ), 'val' );
  79                 $this->assertEquals( $properties['foo'], 'val' );
  80
  81                 $po->setProperty( 'foo', 'second val' );
  82
  83                 $properties = $po->getProperties();
  84                 $this->assertEquals( $po->getProperty( 'foo' ), 'second val' );
  85                 $this->assertEquals( $properties['foo'], 'second val' );
  86
  87                 $po->unsetProperty( 'foo' );
  88
  89                 $properties = $po->getProperties();
  90                 $this->assertEquals( $po->getProperty( 'foo' ), false );
  91                 $this->assertArrayNotHasKey( 'foo', $properties );
  92         }
  93
  94         /**
  95          * @covers ParserOutput::getText
  96          * @dataProvider provideGetText
  97          * @param array $options Options to getText()
  98          * @param array $poState ParserOptions state fields to set
  99          * @param string $text Parser text
 100          * @param string $expect Expected output
 101          */
 102         public function testGetText( $options, $poState, $text, $expect ) {
 103                 $this->setMwGlobals( [
 104                         'wgArticlePath' => '/wiki/$1',
 105                         'wgScriptPath' => '/w',
 106                         'wgScript' => '/w/index.php',
 107                 ] );
 108                 $this->hideDeprecated( 'ParserOutput stateful allowTOC' );
 109                 $this->hideDeprecated( 'ParserOutput stateful enableSectionEditLinks' );
 110
 111                 $po = new ParserOutput( $text );
 112
 113                 // Emulate Parser
 114                 $po->setEditSectionTokens( true );
 115
 116                 if ( $poState ) {
 117                         $wrap = TestingAccessWrapper::newFromObject( $po );
 118                         foreach ( $poState as $key => $value ) {
 119                                 $wrap->$key = $value;
 120                         }
 121                 }
 122
 123                 $actual = $po->getText( $options );
 124                 $this->assertSame( $expect, $actual );
 125         }
 126
 127         public static function provideGetText() {
 128                 // phpcs:disable Generic.Files.LineLength
 129                 $text = <<<EOF
 130 <div class="mw-parser-output"><p>Test document.
 131 </p>
 132 <mw:toc><div id="toc" class="toc"><div class="toctitle"><h2>Contents</h2></div>
 133 <ul>
 134 <li class="toclevel-1 tocsection-1"><a href="#Section_1"><span class="tocnumber">1</span> <span class="toctext">Section 1</span></a></li>
 135 <li class="toclevel-1 tocsection-2"><a href="#Section_2"><span class="tocnumber">2</span> <span class="toctext">Section 2</span></a>
 136 <ul>
 137 <li class="toclevel-2 tocsection-3"><a href="#Section_2.1"><span class="tocnumber">2.1</span> <span class="toctext">Section 2.1</span></a></li>
 138 </ul>
 139 </li>
 140 <li class="toclevel-1 tocsection-4"><a href="#Section_3"><span class="tocnumber">3</span> <span class="toctext">Section 3</span></a></li>
 141 </ul>
 142 </div>
 143 </mw:toc>
 144 <h2><span class="mw-headline" id="Section_1">Section 1</span><mw:editsection page="Test Page" section="1">Section 1</mw:editsection></h2>
 145 <p>One
 146 </p>
 147 <h2><span class="mw-headline" id="Section_2">Section 2</span><mw:editsection page="Test Page" section="2">Section 2</mw:editsection></h2>
 148 <p>Two
 149 </p>
 150 <h3><span class="mw-headline" id="Section_2.1">Section 2.1</span><mw:editsection page="Test Page" section="3">Section 2.1</mw:editsection></h3>
 151 <p>Two point one
 152 </p>
 153 <h2><span class="mw-headline" id="Section_3">Section 3</span><mw:editsection page="Test Page" section="4">Section 3</mw:editsection></h2>
 154 <p>Three
 155 </p></div>
 156 EOF;
 157
 158                 $dedupText = <<<EOF
 159 <p>This is a test document.</p>
 160 <style data-mw-deduplicate="duplicate1">.Duplicate1 {}</style>
 161 <style data-mw-deduplicate="duplicate1">.Duplicate1 {}</style>
 162 <style data-mw-deduplicate="duplicate2">.Duplicate2 {}</style>
 163 <style data-mw-deduplicate="duplicate1">.Duplicate1 {}</style>
 164 <style data-mw-deduplicate="duplicate2">.Duplicate2 {}</style>
 165 <style data-mw-not-deduplicate="duplicate1">.Duplicate1 {}</style>
 166 <style data-mw-deduplicate="duplicate1">.Same-attribute-different-content {}</style>
 167 <style data-mw-deduplicate="duplicate3">.Duplicate1 {}</style>
 168 <style>.Duplicate1 {}</style>
 169 EOF;
 170
 171                 return [
 172                         'No stateless options, default state' => [
 173                                 [], [], $text, <<<EOF
 174 <div class="mw-parser-output"><p>Test document.
 175 </p>
 176 <div id="toc" class="toc"><div class="toctitle"><h2>Contents</h2></div>
 177 <ul>
 178 <li class="toclevel-1 tocsection-1"><a href="#Section_1"><span class="tocnumber">1</span> <span class="toctext">Section 1</span></a></li>
 179 <li class="toclevel-1 tocsection-2"><a href="#Section_2"><span class="tocnumber">2</span> <span class="toctext">Section 2</span></a>
 180 <ul>
 181 <li class="toclevel-2 tocsection-3"><a href="#Section_2.1"><span class="tocnumber">2.1</span> <span class="toctext">Section 2.1</span></a></li>
 182 </ul>
 183 </li>
 184 <li class="toclevel-1 tocsection-4"><a href="#Section_3"><span class="tocnumber">3</span> <span class="toctext">Section 3</span></a></li>
 185 </ul>
 186 </div>
 187
 188 <h2><span class="mw-headline" id="Section_1">Section 1</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=1" title="Edit section: Section 1">edit</a><span class="mw-editsection-bracket">]</span></span></h2>
 189 <p>One
 190 </p>
 191 <h2><span class="mw-headline" id="Section_2">Section 2</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=2" title="Edit section: Section 2">edit</a><span class="mw-editsection-bracket">]</span></span></h2>
 192 <p>Two
 193 </p>
 194 <h3><span class="mw-headline" id="Section_2.1">Section 2.1</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=3" title="Edit section: Section 2.1">edit</a><span class="mw-editsection-bracket">]</span></span></h3>
 195 <p>Two point one
 196 </p>
 197 <h2><span class="mw-headline" id="Section_3">Section 3</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=4" title="Edit section: Section 3">edit</a><span class="mw-editsection-bracket">]</span></span></h2>
 198 <p>Three
 199 </p></div>
 200 EOF
 201                         ],
 202                         'No stateless options, TOC statefully disabled' => [
 203                                 [], [ 'mTOCEnabled' => false ], $text, <<<EOF
 204 <div class="mw-parser-output"><p>Test document.
 205 </p>
 206
 207 <h2><span class="mw-headline" id="Section_1">Section 1</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=1" title="Edit section: Section 1">edit</a><span class="mw-editsection-bracket">]</span></span></h2>
 208 <p>One
 209 </p>
 210 <h2><span class="mw-headline" id="Section_2">Section 2</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=2" title="Edit section: Section 2">edit</a><span class="mw-editsection-bracket">]</span></span></h2>
 211 <p>Two
 212 </p>
 213 <h3><span class="mw-headline" id="Section_2.1">Section 2.1</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=3" title="Edit section: Section 2.1">edit</a><span class="mw-editsection-bracket">]</span></span></h3>
 214 <p>Two point one
 215 </p>
 216 <h2><span class="mw-headline" id="Section_3">Section 3</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=4" title="Edit section: Section 3">edit</a><span class="mw-editsection-bracket">]</span></span></h2>
 217 <p>Three
 218 </p></div>
 219 EOF
 220                         ],
 221                         'No stateless options, section edits statefully disabled' => [
 222                                 [], [ 'mEditSectionTokens' => false ], $text, <<<EOF
 223 <div class="mw-parser-output"><p>Test document.
 224 </p>
 225 <div id="toc" class="toc"><div class="toctitle"><h2>Contents</h2></div>
 226 <ul>
 227 <li class="toclevel-1 tocsection-1"><a href="#Section_1"><span class="tocnumber">1</span> <span class="toctext">Section 1</span></a></li>
 228 <li class="toclevel-1 tocsection-2"><a href="#Section_2"><span class="tocnumber">2</span> <span class="toctext">Section 2</span></a>
 229 <ul>
 230 <li class="toclevel-2 tocsection-3"><a href="#Section_2.1"><span class="tocnumber">2.1</span> <span class="toctext">Section 2.1</span></a></li>
 231 </ul>
 232 </li>
 233 <li class="toclevel-1 tocsection-4"><a href="#Section_3"><span class="tocnumber">3</span> <span class="toctext">Section 3</span></a></li>
 234 </ul>
 235 </div>
 236
 237 <h2><span class="mw-headline" id="Section_1">Section 1</span></h2>
 238 <p>One
 239 </p>
 240 <h2><span class="mw-headline" id="Section_2">Section 2</span></h2>
 241 <p>Two
 242 </p>
 243 <h3><span class="mw-headline" id="Section_2.1">Section 2.1</span></h3>
 244 <p>Two point one
 245 </p>
 246 <h2><span class="mw-headline" id="Section_3">Section 3</span></h2>
 247 <p>Three
 248 </p></div>
 249 EOF
 250                         ],
 251                         'Stateless options override stateful settings' => [
 252                                 [ 'allowTOC' => true, 'enableSectionEditLinks' => true ],
 253                                 [ 'mTOCEnabled' => false, 'mEditSectionTokens' => false ],
 254                                 $text, <<<EOF
 255 <div class="mw-parser-output"><p>Test document.
 256 </p>
 257 <div id="toc" class="toc"><div class="toctitle"><h2>Contents</h2></div>
 258 <ul>
 259 <li class="toclevel-1 tocsection-1"><a href="#Section_1"><span class="tocnumber">1</span> <span class="toctext">Section 1</span></a></li>
 260 <li class="toclevel-1 tocsection-2"><a href="#Section_2"><span class="tocnumber">2</span> <span class="toctext">Section 2</span></a>
 261 <ul>
 262 <li class="toclevel-2 tocsection-3"><a href="#Section_2.1"><span class="tocnumber">2.1</span> <span class="toctext">Section 2.1</span></a></li>
 263 </ul>
 264 </li>
 265 <li class="toclevel-1 tocsection-4"><a href="#Section_3"><span class="tocnumber">3</span> <span class="toctext">Section 3</span></a></li>
 266 </ul>
 267 </div>
 268
 269 <h2><span class="mw-headline" id="Section_1">Section 1</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=1" title="Edit section: Section 1">edit</a><span class="mw-editsection-bracket">]</span></span></h2>
 270 <p>One
 271 </p>
 272 <h2><span class="mw-headline" id="Section_2">Section 2</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=2" title="Edit section: Section 2">edit</a><span class="mw-editsection-bracket">]</span></span></h2>
 273 <p>Two
 274 </p>
 275 <h3><span class="mw-headline" id="Section_2.1">Section 2.1</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=3" title="Edit section: Section 2.1">edit</a><span class="mw-editsection-bracket">]</span></span></h3>
 276 <p>Two point one
 277 </p>
 278 <h2><span class="mw-headline" id="Section_3">Section 3</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=4" title="Edit section: Section 3">edit</a><span class="mw-editsection-bracket">]</span></span></h2>
 279 <p>Three
 280 </p></div>
 281 EOF
 282                         ],
 283                         'Statelessly disable section edit links' => [
 284                                 [ 'enableSectionEditLinks' => false ], [], $text, <<<EOF
 285 <div class="mw-parser-output"><p>Test document.
 286 </p>
 287 <div id="toc" class="toc"><div class="toctitle"><h2>Contents</h2></div>
 288 <ul>
 289 <li class="toclevel-1 tocsection-1"><a href="#Section_1"><span class="tocnumber">1</span> <span class="toctext">Section 1</span></a></li>
 290 <li class="toclevel-1 tocsection-2"><a href="#Section_2"><span class="tocnumber">2</span> <span class="toctext">Section 2</span></a>
 291 <ul>
 292 <li class="toclevel-2 tocsection-3"><a href="#Section_2.1"><span class="tocnumber">2.1</span> <span class="toctext">Section 2.1</span></a></li>
 293 </ul>
 294 </li>
 295 <li class="toclevel-1 tocsection-4"><a href="#Section_3"><span class="tocnumber">3</span> <span class="toctext">Section 3</span></a></li>
 296 </ul>
 297 </div>
 298
 299 <h2><span class="mw-headline" id="Section_1">Section 1</span></h2>
 300 <p>One
 301 </p>
 302 <h2><span class="mw-headline" id="Section_2">Section 2</span></h2>
 303 <p>Two
 304 </p>
 305 <h3><span class="mw-headline" id="Section_2.1">Section 2.1</span></h3>
 306 <p>Two point one
 307 </p>
 308 <h2><span class="mw-headline" id="Section_3">Section 3</span></h2>
 309 <p>Three
 310 </p></div>
 311 EOF
 312                         ],
 313                         'Statelessly disable TOC' => [
 314                                 [ 'allowTOC' => false ], [], $text, <<<EOF
 315 <div class="mw-parser-output"><p>Test document.
 316 </p>
 317
 318 <h2><span class="mw-headline" id="Section_1">Section 1</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=1" title="Edit section: Section 1">edit</a><span class="mw-editsection-bracket">]</span></span></h2>
 319 <p>One
 320 </p>
 321 <h2><span class="mw-headline" id="Section_2">Section 2</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=2" title="Edit section: Section 2">edit</a><span class="mw-editsection-bracket">]</span></span></h2>
 322 <p>Two
 323 </p>
 324 <h3><span class="mw-headline" id="Section_2.1">Section 2.1</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=3" title="Edit section: Section 2.1">edit</a><span class="mw-editsection-bracket">]</span></span></h3>
 325 <p>Two point one
 326 </p>
 327 <h2><span class="mw-headline" id="Section_3">Section 3</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=4" title="Edit section: Section 3">edit</a><span class="mw-editsection-bracket">]</span></span></h2>
 328 <p>Three
 329 </p></div>
 330 EOF
 331                         ],
 332                         'Statelessly unwrap text' => [
 333                                 [ 'unwrap' => true ], [], $text, <<<EOF
 334 <p>Test document.
 335 </p>
 336 <div id="toc" class="toc"><div class="toctitle"><h2>Contents</h2></div>
 337 <ul>
 338 <li class="toclevel-1 tocsection-1"><a href="#Section_1"><span class="tocnumber">1</span> <span class="toctext">Section 1</span></a></li>
 339 <li class="toclevel-1 tocsection-2"><a href="#Section_2"><span class="tocnumber">2</span> <span class="toctext">Section 2</span></a>
 340 <ul>
 341 <li class="toclevel-2 tocsection-3"><a href="#Section_2.1"><span class="tocnumber">2.1</span> <span class="toctext">Section 2.1</span></a></li>
 342 </ul>
 343 </li>
 344 <li class="toclevel-1 tocsection-4"><a href="#Section_3"><span class="tocnumber">3</span> <span class="toctext">Section 3</span></a></li>
 345 </ul>
 346 </div>
 347
 348 <h2><span class="mw-headline" id="Section_1">Section 1</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=1" title="Edit section: Section 1">edit</a><span class="mw-editsection-bracket">]</span></span></h2>
 349 <p>One
 350 </p>
 351 <h2><span class="mw-headline" id="Section_2">Section 2</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=2" title="Edit section: Section 2">edit</a><span class="mw-editsection-bracket">]</span></span></h2>
 352 <p>Two
 353 </p>
 354 <h3><span class="mw-headline" id="Section_2.1">Section 2.1</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=3" title="Edit section: Section 2.1">edit</a><span class="mw-editsection-bracket">]</span></span></h3>
 355 <p>Two point one
 356 </p>
 357 <h2><span class="mw-headline" id="Section_3">Section 3</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Test_Page&amp;action=edit&amp;section=4" title="Edit section: Section 3">edit</a><span class="mw-editsection-bracket">]</span></span></h2>
 358 <p>Three
 359 </p>
 360 EOF
 361                         ],
 362                         'Unwrap without a mw-parser-output wrapper' => [
 363                                 [ 'unwrap' => true ], [], '<div class="foobar">Content</div>', '<div class="foobar">Content</div>'
 364                         ],
 365                         'Unwrap with extra comment at end' => [
 366                                 [ 'unwrap' => true ], [], '<div class="mw-parser-output"><p>Test document.</p></div>
 367 <!-- Saved in parser cache... -->', '<p>Test document.</p>
 368 <!-- Saved in parser cache... -->'
 369                         ],
 370                         'Style deduplication' => [
 371                                 [], [], $dedupText, <<<EOF
 372 <p>This is a test document.</p>
 373 <style data-mw-deduplicate="duplicate1">.Duplicate1 {}</style>
 374 <link rel="mw-deduplicated-inline-style" href="mw-data:duplicate1"/>
 375 <style data-mw-deduplicate="duplicate2">.Duplicate2 {}</style>
 376 <link rel="mw-deduplicated-inline-style" href="mw-data:duplicate1"/>
 377 <link rel="mw-deduplicated-inline-style" href="mw-data:duplicate2"/>
 378 <style data-mw-not-deduplicate="duplicate1">.Duplicate1 {}</style>
 379 <link rel="mw-deduplicated-inline-style" href="mw-data:duplicate1"/>
 380 <style data-mw-deduplicate="duplicate3">.Duplicate1 {}</style>
 381 <style>.Duplicate1 {}</style>
 382 EOF
 383                         ],
 384                         'Style deduplication disabled' => [
 385                                 [ 'deduplicateStyles' => false ], [], $dedupText, $dedupText
 386                         ],
 387                 ];
 388                 // phpcs:enable
 389         }
 390
 391 }