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
 109                 $po = new ParserOutput( $text );
 110
 111                 // Emulate Parser
 112                 $po->setEditSectionTokens( true );
 113
 114                 if ( $poState ) {
 115                         $wrap = TestingAccessWrapper::newFromObject( $po );
 116                         foreach ( $poState as $key => $value ) {
 117                                 $wrap->$key = $value;
 118                         }
 119                 }
 120
 121                 $actual = $po->getText( $options );
 122                 $this->assertSame( $expect, $actual );
 123         }
 124
 125         public static function provideGetText() {
 126                 // phpcs:disable Generic.Files.LineLength
 127                 $text = <<<EOF
 128 <div class="mw-parser-output"><p>Test document.
 129 </p>
 130 <mw:toc><div id="toc" class="toc"><div class="toctitle"><h2>Contents</h2></div>
 131 <ul>
 132 <li class="toclevel-1 tocsection-1"><a href="#Section_1"><span class="tocnumber">1</span> <span class="toctext">Section 1</span></a></li>
 133 <li class="toclevel-1 tocsection-2"><a href="#Section_2"><span class="tocnumber">2</span> <span class="toctext">Section 2</span></a>
 134 <ul>
 135 <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>
 136 </ul>
 137 </li>
 138 <li class="toclevel-1 tocsection-4"><a href="#Section_3"><span class="tocnumber">3</span> <span class="toctext">Section 3</span></a></li>
 139 </ul>
 140 </div>
 141 </mw:toc>
 142 <h2><span class="mw-headline" id="Section_1">Section 1</span><mw:editsection page="Test Page" section="1">Section 1</mw:editsection></h2>
 143 <p>One
 144 </p>
 145 <h2><span class="mw-headline" id="Section_2">Section 2</span><mw:editsection page="Test Page" section="2">Section 2</mw:editsection></h2>
 146 <p>Two
 147 </p>
 148 <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>
 149 <p>Two point one
 150 </p>
 151 <h2><span class="mw-headline" id="Section_3">Section 3</span><mw:editsection page="Test Page" section="4">Section 3</mw:editsection></h2>
 152 <p>Three
 153 </p></div>
 154 EOF;
 155
 156                 return [
 157                         'No stateless options, default state' => [
 158                                 [], [], $text, <<<EOF
 159 <div class="mw-parser-output"><p>Test document.
 160 </p>
 161 <div id="toc" class="toc"><div class="toctitle"><h2>Contents</h2></div>
 162 <ul>
 163 <li class="toclevel-1 tocsection-1"><a href="#Section_1"><span class="tocnumber">1</span> <span class="toctext">Section 1</span></a></li>
 164 <li class="toclevel-1 tocsection-2"><a href="#Section_2"><span class="tocnumber">2</span> <span class="toctext">Section 2</span></a>
 165 <ul>
 166 <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>
 167 </ul>
 168 </li>
 169 <li class="toclevel-1 tocsection-4"><a href="#Section_3"><span class="tocnumber">3</span> <span class="toctext">Section 3</span></a></li>
 170 </ul>
 171 </div>
 172
 173 <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>
 174 <p>One
 175 </p>
 176 <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>
 177 <p>Two
 178 </p>
 179 <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>
 180 <p>Two point one
 181 </p>
 182 <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>
 183 <p>Three
 184 </p></div>
 185 EOF
 186                         ],
 187                         'No stateless options, TOC statefully disabled' => [
 188                                 [], [ 'mTOCEnabled' => false ], $text, <<<EOF
 189 <div class="mw-parser-output"><p>Test document.
 190 </p>
 191
 192 <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>
 193 <p>One
 194 </p>
 195 <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>
 196 <p>Two
 197 </p>
 198 <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>
 199 <p>Two point one
 200 </p>
 201 <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>
 202 <p>Three
 203 </p></div>
 204 EOF
 205                         ],
 206                         'No stateless options, section edits statefully disabled' => [
 207                                 [], [ 'mEditSectionTokens' => false ], $text, <<<EOF
 208 <div class="mw-parser-output"><p>Test document.
 209 </p>
 210 <div id="toc" class="toc"><div class="toctitle"><h2>Contents</h2></div>
 211 <ul>
 212 <li class="toclevel-1 tocsection-1"><a href="#Section_1"><span class="tocnumber">1</span> <span class="toctext">Section 1</span></a></li>
 213 <li class="toclevel-1 tocsection-2"><a href="#Section_2"><span class="tocnumber">2</span> <span class="toctext">Section 2</span></a>
 214 <ul>
 215 <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>
 216 </ul>
 217 </li>
 218 <li class="toclevel-1 tocsection-4"><a href="#Section_3"><span class="tocnumber">3</span> <span class="toctext">Section 3</span></a></li>
 219 </ul>
 220 </div>
 221
 222 <h2><span class="mw-headline" id="Section_1">Section 1</span></h2>
 223 <p>One
 224 </p>
 225 <h2><span class="mw-headline" id="Section_2">Section 2</span></h2>
 226 <p>Two
 227 </p>
 228 <h3><span class="mw-headline" id="Section_2.1">Section 2.1</span></h3>
 229 <p>Two point one
 230 </p>
 231 <h2><span class="mw-headline" id="Section_3">Section 3</span></h2>
 232 <p>Three
 233 </p></div>
 234 EOF
 235                         ],
 236                         'Stateless options override stateful settings' => [
 237                                 [ 'allowTOC' => true, 'enableSectionEditLinks' => true ],
 238                                 [ 'mTOCEnabled' => false, 'mEditSectionTokens' => false ],
 239                                 $text, <<<EOF
 240 <div class="mw-parser-output"><p>Test document.
 241 </p>
 242 <div id="toc" class="toc"><div class="toctitle"><h2>Contents</h2></div>
 243 <ul>
 244 <li class="toclevel-1 tocsection-1"><a href="#Section_1"><span class="tocnumber">1</span> <span class="toctext">Section 1</span></a></li>
 245 <li class="toclevel-1 tocsection-2"><a href="#Section_2"><span class="tocnumber">2</span> <span class="toctext">Section 2</span></a>
 246 <ul>
 247 <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>
 248 </ul>
 249 </li>
 250 <li class="toclevel-1 tocsection-4"><a href="#Section_3"><span class="tocnumber">3</span> <span class="toctext">Section 3</span></a></li>
 251 </ul>
 252 </div>
 253
 254 <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>
 255 <p>One
 256 </p>
 257 <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>
 258 <p>Two
 259 </p>
 260 <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>
 261 <p>Two point one
 262 </p>
 263 <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>
 264 <p>Three
 265 </p></div>
 266 EOF
 267                         ],
 268                         'Statelessly disable section edit links' => [
 269                                 [ 'enableSectionEditLinks' => false ], [], $text, <<<EOF
 270 <div class="mw-parser-output"><p>Test document.
 271 </p>
 272 <div id="toc" class="toc"><div class="toctitle"><h2>Contents</h2></div>
 273 <ul>
 274 <li class="toclevel-1 tocsection-1"><a href="#Section_1"><span class="tocnumber">1</span> <span class="toctext">Section 1</span></a></li>
 275 <li class="toclevel-1 tocsection-2"><a href="#Section_2"><span class="tocnumber">2</span> <span class="toctext">Section 2</span></a>
 276 <ul>
 277 <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>
 278 </ul>
 279 </li>
 280 <li class="toclevel-1 tocsection-4"><a href="#Section_3"><span class="tocnumber">3</span> <span class="toctext">Section 3</span></a></li>
 281 </ul>
 282 </div>
 283
 284 <h2><span class="mw-headline" id="Section_1">Section 1</span></h2>
 285 <p>One
 286 </p>
 287 <h2><span class="mw-headline" id="Section_2">Section 2</span></h2>
 288 <p>Two
 289 </p>
 290 <h3><span class="mw-headline" id="Section_2.1">Section 2.1</span></h3>
 291 <p>Two point one
 292 </p>
 293 <h2><span class="mw-headline" id="Section_3">Section 3</span></h2>
 294 <p>Three
 295 </p></div>
 296 EOF
 297                         ],
 298                         'Statelessly disable TOC' => [
 299                                 [ 'allowTOC' => false ], [], $text, <<<EOF
 300 <div class="mw-parser-output"><p>Test document.
 301 </p>
 302
 303 <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>
 304 <p>One
 305 </p>
 306 <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>
 307 <p>Two
 308 </p>
 309 <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>
 310 <p>Two point one
 311 </p>
 312 <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>
 313 <p>Three
 314 </p></div>
 315 EOF
 316                         ],
 317                         'Statelessly unwrap text' => [
 318                                 [ 'unwrap' => true ], [], $text, <<<EOF
 319 <p>Test document.
 320 </p>
 321 <div id="toc" class="toc"><div class="toctitle"><h2>Contents</h2></div>
 322 <ul>
 323 <li class="toclevel-1 tocsection-1"><a href="#Section_1"><span class="tocnumber">1</span> <span class="toctext">Section 1</span></a></li>
 324 <li class="toclevel-1 tocsection-2"><a href="#Section_2"><span class="tocnumber">2</span> <span class="toctext">Section 2</span></a>
 325 <ul>
 326 <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>
 327 </ul>
 328 </li>
 329 <li class="toclevel-1 tocsection-4"><a href="#Section_3"><span class="tocnumber">3</span> <span class="toctext">Section 3</span></a></li>
 330 </ul>
 331 </div>
 332
 333 <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>
 334 <p>One
 335 </p>
 336 <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>
 337 <p>Two
 338 </p>
 339 <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>
 340 <p>Two point one
 341 </p>
 342 <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>
 343 <p>Three
 344 </p>
 345 EOF
 346                         ],
 347                         'Unwrap without a mw-parser-output wrapper' => [
 348                                 [ 'unwrap' => true ], [], '<div class="foobar">Content</div>', '<div class="foobar">Content</div>'
 349                         ],
 350                 ];
 351                 // phpcs:enable
 352         }
 353
 354 }