$wt ] );
$pageConfig = new MockPageConfig( $opts, $content );
$html = $parsoid->wikitext2html( $pageConfig, [ "wrapSections" => false ] );
$doc = DOMUtils::parseHTML( $html );
$docBody = DOMCompat::getBody( $doc );
return( $docBody );
}
/**
* @param string $description
* @param string $wt
* @param array $search
* @param array $replace
* @param string $withoutSelser
* @param string $withSelser
*/
private function sharedTest(
string $description, string $wt, array $search, array $replace,
string $withoutSelser, string $withSelser
): void {
$siteConfig = new MockSiteConfig( [] );
$dataAccess = new MockDataAccess( [] );
$parsoid = new Parsoid( $siteConfig, $dataAccess );
$content = new MockPageContent( [ 'main' => $wt ] );
$pageConfig = new MockPageConfig( [], $content );
$html = $parsoid->wikitext2html( $pageConfig, [ "wrapSections" => false ] );
// This is mimicking a copy/paste in an editor
$editedHTML = str_replace( $search, $replace, $html );
// Without selser
$editedWT = $parsoid->html2wikitext( $pageConfig, $editedHTML, [], null );
$this->assertEquals( $withoutSelser, $editedWT, $description );
// With selser
$selserData = new SelserData( $wt, $html );
$editedWT = $parsoid->html2wikitext( $pageConfig, $editedHTML, [], $selserData );
$this->assertEquals( $withSelser, $editedWT, $description );
}
/**
* Wikilinks use ./ prefixed urls. For reasons of consistency,
* we should use a similar format for internal cite urls.
* This spec ensures that we don't inadvertently break that requirement.
* should use ./ prefixed urls for cite links
* @covers \Wikimedia\Parsoid\Wt2Html\ParserPipeline
*/
public function testWikilinkUseDotSlashPrefix(): void {
$description = "Regression Specs: should use ./ prefixed urls for cite links";
$wt = "a [[Foo]] [b]";
$docBody = $this->parseWT( $wt, [ 'title' => 'Main_Page' ] );
$attrib = DOMCompat::querySelectorAll( $docBody, ".mw-ref a" )[0]->
getAttribute( 'href' );
$this->assertEquals( './Main_Page#cite_note-1', $attrib, $description );
$attrib = DOMCompat::querySelectorAll( $docBody, "#cite_note-1 a" )[0]->
getAttribute( 'href' );
$this->assertEquals( './Main_Page#cite_ref-1', $attrib, $description );
}
/**
* should prevent regression of T153107
* @covers \Wikimedia\Parsoid\Wt2Html\ParserPipeline
*/
public function testPreventRegressionT153107(): void {
$description = "Regression Specs: should prevent regression of T153107";
$wt = "[[Foo|bar]]";
$searchItems = [ 'bar' ];
$replaceItems = [ 'Foo' ];
$withoutSelser = "[[Foo|Foo]]";
$withSelser = "[[Foo]]";
$this->sharedTest( $description, $wt, $searchItems, $replaceItems,
$withoutSelser, $withSelser );
}
/**
* should prevent regression of T268737
* @covers \Wikimedia\Parsoid\Html2Wt\Separators::recoverTrimmedWhitespace
*/
public function testPreventRegressionT268737(): void {
$description = "Regression Specs: should prevent regression of T262737";
$wt = "* [[Foo]]--[[Bar]]";
$searchItems = [ '--' ];
$replaceItems = [ '..' ];
$withoutSelser = "*[[Foo]]..[[Bar]]";
$withSelser = "* [[Foo]]..[[Bar]]";
$this->sharedTest( $description, $wt, $searchItems, $replaceItems,
$withoutSelser, $withSelser );
}
/**
* should ensure edited lists, headings, table cells preserve original
* whitespace in some scenarios
* @covers \Wikimedia\Parsoid\Wt2Html\ParserPipeline
*/
public function testPreserveWhitespace(): void {
$description = "should ensure edited lists, headings, table cells preserve original " .
"whitespace in some scenarios";
$wt = implode( "\n", [
"* item",
"* item",
"* item
",
"* [[Link|item]]",
"* wrap",
"== heading ==",
"== heading ==",
"== heading
==",
"== [[Link|heading]] ==",
"{|",
"| cell",
"| cell",
"| cell
",
"| [[Link|cell]]",
"| unedited c1 || cell || unedited c3 || cell",
"| unedited c1 || cell || unedited c3 || unedited c4",
"|}"
] );
$searchItems = [ 'item', 'heading', 'cell', 'wrap' ];
$replaceItems = [ 'edited item', 'edited heading', 'edited cell', 'wrap' ];
// Without selser, we should see normalized wikitext
$withoutSelser = implode( "\n", [
"*edited item",
"*edited item",
"*edited item
",
"*[[Link|edited item]]",
"*wrap",
"==edited heading==",
"==edited heading==",
"==edited heading
==",
"==[[Link|edited heading]]==",
"{|",
"|edited cell",
"|edited cell",
"|edited cell
",
"|[[Link|edited cell]]",
"|unedited c1||edited cell||unedited c3||edited cell",
"|unedited c1||edited cell||unedited c3||unedited c4",
"|}"
] );
// With selser, we should have whitespace heuristics applied
$withSelser = implode( "\n", [
"* edited item",
"* edited item",
"* edited item
",
"* [[Link|edited item]]",
"* wrap",
"== edited heading ==",
"== edited heading ==",
"== edited heading
==",
"== [[Link|edited heading]] ==",
"{|",
"| edited cell",
"| edited cell",
"| edited cell
",
"| [[Link|edited cell]]",
"| unedited c1 || edited cell || unedited c3 || edited cell",
"| unedited c1 || edited cell || unedited c3 || unedited c4",
"|}"
] );
$this->sharedTest( $description, $wt, $searchItems, $replaceItems,
$withoutSelser, $withSelser );
}
/**
* should not apply whitespace heuristics for HTML versions older than 1.7.0
* @covers \Wikimedia\Parsoid\Wt2Html\ParserPipeline
*/
public function testNoWhitespaceHeuristics(): void {
$description = "should not apply whitespace heuristics for HTML versions older than 1.7.0";
$wt = implode( "\n", [
"* item",
"* item",
"* item
",
"* [[Link|item]]",
"* wrap",
"== heading ==",
"== heading ==",
"== heading
==",
"== [[Link|heading]] ==",
"{|",
"| cell",
"| cell",
"| cell
",
"| [[Link|cell]]",
"|}"
] );
$siteConfig = new MockSiteConfig( [] );
$dataAccess = new MockDataAccess( [] );
$parsoid = new Parsoid( $siteConfig, $dataAccess );
$content = new MockPageContent( [ 'main' => $wt ] );
$pageConfig = new MockPageConfig( [], $content );
$html = $parsoid->wikitext2html( $pageConfig, [ "wrapSections" => false ] );
$search = [ 'item', 'heading', 'cell', 'wrap' ];
$replace = [ 'edited item', 'edited heading', 'edited cell', 'wrap' ];
$editedBody = str_replace( $search, $replace, $html );
// Whitespace heuristics are enabled
$newVersion = implode( "\n", [
"* edited item",
"* edited item",
"* edited item
",
"* [[Link|edited item]]",
"* wrap",
"== edited heading ==",
"== edited heading ==",
"== edited heading
==",
"== [[Link|edited heading]] ==",
"{|",
"| edited cell",
"| edited cell",
"| edited cell
",
"| [[Link|edited cell]]",
"|}"
] );
$selserData = new SelserData( $wt, $html );
$editedWT = $parsoid->html2wikitext( $pageConfig, $editedBody, [], $selserData );
$this->assertEquals( $newVersion, $editedWT, $description );
// Whitespace heuristics are disabled, but selser's buildSep heuristics
// will do the magic for old non-text and non-comment nodes.
$oldVersion = implode( "\n", [
"*edited item",
"*edited item",
"* edited item
",
"* [[Link|edited item]]",
"*wrap",
"==edited heading==",
"==edited heading==",
"== edited heading
==",
"== [[Link|edited heading]] ==",
"{|",
"|edited cell",
"|edited cell",
"| edited cell
",
"| [[Link|edited cell]]",
"|}"
] );
// Pretend we are in 1.6.1 version to disable whitespace heuristics
$htmlVersion = '1.6.1';
$selserData = new SelserData( $wt, $html );
$editedWT = $parsoid->html2wikitext( $pageConfig, $editedBody,
[ 'inputContentVersion' => $htmlVersion ], $selserData );
$this->assertEquals( $oldVersion, $editedWT, $description );
}
/**
* should not wrap templatestyles style tags in p-wrappers
* @covers \Wikimedia\Parsoid\Wt2Html\TT\TemplateHandler
*/
public function testNoWrapTemplateStyles(): void {
$description = "Regression Specs: should not wrap templatestyles style tags in p-wrappers";
$wt = "foo
";
$docBody = $this->parseWT( $wt );
$node = DOMCompat::nodeName( $docBody->firstChild );
$this->assertEquals( "style", $node, $description );
}
/**
* For https://phabricator.wikimedia.org/T208901
* should not split p-wrappers around templatestyles
* @covers \Wikimedia\Parsoid\Wt2Html\TT\TemplateHandler
*/
public function testNoSplitPWrapper(): void {
$description = "Regression Specs: should not split p-wrappers around templatestyles";
$wt = 'abc {{1x| def}} ghi ' .
'[[File:Thumb.png|thumb|250px]]';
$docBody = $this->parseWT( $wt );
$node = DOMCompat::nodeName( $docBody->firstChild );
$this->assertEquals( "p", $node, $description );
$node = DOMCompat::nodeName( DOMUtils::nextNonSepSibling( $docBody->firstChild ) );
$this->assertEquals( "figure", $node, $description );
}
/**
* Regression Specs: should deduplicate templatestyles style tags
* @covers \Wikimedia\Parsoid\Wt2Html\TT\TemplateHandler
*/
public function testNoDupTemplateStyleTags(): void {
$description = "Regression Specs: should deduplicate templatestyles style tags";
$wt = [
'a',
'b'
];
$wt = implode( "\n", $wt );
$docBody = $this->parseWT( $wt );
$firstStyle = $docBody->firstChild->firstChild;
$this->assertEquals( "style", DOMCompat::nodeName( $firstStyle ), $description );
$secondStyle = $firstStyle->nextSibling->nextSibling->nextSibling;
$this->assertEquals( "link", DOMCompat::nodeName( $secondStyle ), $description );
$this->assertEquals( "mw-deduplicated-inline-style",
$secondStyle->getAttribute( 'rel' ), $description );
$this->assertEquals( 'mw-data:' .
$firstStyle->getAttribute( 'data-mw-deduplicate' ),
$secondStyle->getAttribute( 'href' ), $description );
$keys = [ 'about','typeof','data-mw','data-parsoid' ];
foreach ( $keys as $key ) {
$this->assertTrue( $secondStyle->hasAttribute( $key ), $description );
}
}
/**
* I1f572f996a7c2b3b852752f5348ebb60d8e21c47 introduces a backwards
* incompatibility. This test asserts that selser will restore content
* for invalid follows that would otherwise be dropped since it wasn't
* span wrapped.
* @covers \Wikimedia\Parsoid\Ext\Cite\Ref::domToWikitext
*/
public function testSelserFollowsWrap(): void {
$wt = 'Hi ho [hi ho]';
$html = <<Hi ho [1]
EOT;
$editedHtml = <<Ha ha [1]
EOT;
$siteConfig = new MockSiteConfig( [] );
$dataAccess = new MockDataAccess( [] );
$parsoid = new Parsoid( $siteConfig, $dataAccess );
$content = new MockPageContent( [ 'main' => $wt ] );
$pageConfig = new MockPageConfig( [], $content );
// Without selser
$editedWt = $parsoid->html2wikitext( $pageConfig, $editedHtml, [], null );
$this->assertEquals( "Ha ha [hi ho]\n\n", $editedWt );
}
}