assertCount( 1, $elts,
"Cannot match selector \"{$spec['selector']}\" in doc " .
ContentUtils::toXML( $doc ) );
$dp = DOMDataUtils::getDataParsoid( $elts[0] );
if ( $spec['dsrContent'] === null ) {
$this->assertTrue( empty( $dp->dsr ) );
return;
}
$dsr = $dp->dsr;
$this->assertTrue( !empty( $dsr ) );
$this->assertInstanceOf( DomSourceRange::class, $dsr );
$this->assertSubstring( $spec['dsrContent'][0], $wt, $dsr->start, $dsr->length(),
"all" );
if ( $spec['dsrContent'][1] === null ) {
$this->assertNull( $dsr->openWidth );
} else {
$this->assertSubstring( $spec['dsrContent'][1], $wt, $dsr->start, $dsr->openWidth,
"open" );
}
if ( $spec['dsrContent'][2] === null ) {
$this->assertNull( $dsr->closeWidth );
} else {
$this->assertSubstring( $spec['dsrContent'][2], $wt, $dsr->innerEnd(), $dsr->closeWidth,
"close" );
}
}
private function assertSubstring( $expected, $wt, $start, $length, $type ) {
$this->assertIsInt( $start );
$this->assertIsInt( $length );
$this->assertGreaterThanOrEqual( 0, $start,
"DSR $type start was less than zero" );
if ( $length > 0 ) {
$this->assertLessThan( strlen( $wt ), $start,
"DSR $type start was greater than or equal to the input length" );
}
$this->assertGreaterThanOrEqual( 0, $length,
"DSR $type length was less than zero" );
$this->assertLessThanOrEqual( strlen( $wt ), $length,
"DSR $type length was greater than the input length" );
$this->assertSame( $expected, PHPUtils::safeSubstr( $wt, $start, $length ),
"DSR $type did not match the expected string" );
}
/**
* For every test with a 'wt' property, provide the 'spec' property that is
* an array of DSR specs that need to be verified on the parsed output of 'wt'.
* Every spec should have:
* selector: a CSS selector for picking a DOM node in the parsed output.
* dsrContent: a 3-element array. The first element is the wikitext corresponding
* to the wikitext substring between dsr[0]..dsr[1]. The second and
* third elements are the opening/closing wikitext tag for that node.
* @dataProvider provideComputeDSR
* @param string $wt
* @param array $specs
*/
public function testComputeDSR( string $wt, array $specs ): 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 ] );
$doc = ContentUtils::createAndLoadDocument( $html );
foreach ( $specs as $spec ) {
$this->validateSpec( $wt, DOMCompat::getBody( $doc ), $spec );
}
}
/**
* @return array
*/
public function provideComputeDSR(): array {
return [
// Paragraph wrapping
'para 1' => [
'wt' => 'a', // disable first test so we can get to the juicy failure immediately.
'specs' => [
[ 'selector' => 'body > p', 'dsrContent' => [ 'a', '', '' ] ]
]
],
'para 2' => [
'wt' => "a\n\nb",
'specs' => [
[ 'selector' => 'body > p:nth-child(1)', 'dsrContent' => [ 'a', '', '' ] ],
[ 'selector' => 'body > p:nth-child(2)', 'dsrContent' => [ 'b', '', '' ] ]
]
],
// Lists
'list 1' => [
'wt' => "*a\n*b",
'specs' => [
[ 'selector' => 'ul', 'dsrContent' => [ "*a\n*b", '', '' ] ],
[ 'selector' => 'ul > li:nth-child(1)', 'dsrContent' => [ '*a', '*', '' ] ],
[ 'selector' => 'ul > li:nth-child(2)', 'dsrContent' => [ '*b', '*', '' ] ]
]
],
'list 2' => [
'wt' => "*a\n**b\n***c\n*d",
'specs' => [
[ 'selector' => 'body > ul', 'dsrContent' => [ "*a\n**b\n***c\n*d", '', '' ] ],
[ 'selector' => 'body > ul > li:first-child', 'dsrContent' => [ "*a\n**b\n***c", '*', '' ] ],
[ 'selector' => 'body > ul > li:first-child > ul', 'dsrContent' => [ "**b\n***c", '', '' ] ],
[
'selector' => 'body > ul > li:first-child > ul > li:first-child',
'dsrContent' => [ "**b\n***c", '**', '' ]
],
[
'selector' => 'body > ul > li:first-child > ul > li:first-child > ul > li',
'dsrContent' => [ '***c', '***', '' ]
],
[ 'selector' => 'body > ul > li:nth-child(2)', 'dsrContent' => [ '*d', '*', '' ] ]
]
],
'list 3' => [
// TODO: this test covers computeListEltWidth() but the return value does
// not find its way into the DOM, so we're only really testing that it
// doesn't throw.
'wt' => '{{1x|** foo}}',
'specs' => [
[
'selector' => 'body > ul > li',
'dsrContent' => null
],
]
],
// Headings
'heading 1' => [
'wt' => "=A=\n==B==\n===C===\n====D====",
'specs' => [
[ 'selector' => 'body > h1', 'dsrContent' => [ '=A=', '=', '=' ] ],
[ 'selector' => 'body > h2', 'dsrContent' => [ '==B==', '==', '==' ] ],
[ 'selector' => 'body > h3', 'dsrContent' => [ '===C===', '===', '===' ] ],
[ 'selector' => 'body > h4', 'dsrContent' => [ '====D====', '====', '====' ] ]
]
],
'heading 2' => [
'wt' => "=A New Use for the = Sign=\n==The == Operator==",
'specs' => [
[
'selector' => 'body > h1',
'dsrContent' => [ '=A New Use for the = Sign=', '=', '=' ]
],
[ 'selector' => 'body > h2', 'dsrContent' => [ '==The == Operator==', '==', '==' ] ]
]
],
// Quotes
'quotes 1' => [
'wt' => "''a''\n'''b'''",
'specs' => [
[ 'selector' => 'p > i', 'dsrContent' => [ "''a''", "''", "''" ] ],
[ 'selector' => 'p > b', 'dsrContent' => [ "'''b'''", "'''", "'''" ] ]
]
],
// Tables
'tables 1' => [
'wt' => "{|\n|-\n|A\n|}",
'specs' => [
[ 'selector' => 'body > table', 'dsrContent' => [ "{|\n|-\n|A\n|}", '{|', '|}' ] ],
[ 'selector' => 'body > table > tbody > tr', 'dsrContent' => [ "|-\n|A", '|-', '' ] ],
[ 'selector' => 'body > table > tbody > tr > td', 'dsrContent' => [ '|A', '|', '' ] ]
]
],
'tables 2' => [
'wt' => "{|\nA\n|}",
'specs' => [
[
'selector' => 'body > table',
'dsrContent' => [ "{|\nA\n|}", '{|', '|}' ]
],
// FIXME: throws due to negative length
/*
[
'selector' => 'body > p > span',
'dsrContent' => [ "A", '', '' ]
],
*/
]
],
'tables 3 mixed stx' => [
'wt' => "{|\n|foo\n",
'specs' => [
[
'selector' => 'body > table',
'dsrContent' => [ "{|\n|foo\n", '{|', '' ]
],
[
'selector' => 'body > table > tbody',
'dsrContent' => [ "|foo\n", '', '' ]
],
[
'selector' => 'tr > td',
'dsrContent' => [ "|foo", '|', '' ]
]
]
],
// Pre
'pre 1' => [
'wt' => " Preformatted text ",
'specs' => [
[ 'selector' => 'body > pre', 'dsrContent' => [ " Preformatted text ", ' ', '' ] ]
]
],
// Regression test for robust DSR computation when PreHandler has a bug.
// PreHandler swallows the category link into which creates an 1-char
// difference in DSR offset by the time the tag is encountered.
// The endTSR value on the tag should override the buggy computed value.
'pre 2' => [
'wt' => " ''a'' b\n[[Category:Foo]]",
'specs' => [
[ 'selector' => 'body > pre > i', 'dsrContent' => [ "''a''", "''", "''" ] ]
]
],
// Elements
'elt 1' => [
'wt' => "'''bold'''",
'specs' => [
[
'selector' => 'body > p',
'dsrContent' => [ "'''bold'''", '', '' ]
],
[
'selector' => 'body > p > small',
'dsrContent' => [ "'''bold'''", '', '' ]
]
]
],
'elt 2 misnested' => [
'wt' => 'A',
'specs' => [
[
'selector' => 'body > p > b',
'dsrContent' => [ 'A', '', '' ]
],
[
'selector' => 'body > p > b > i',
'dsrContent' => [ 'A', '', '' ]
],
]
],
'elt 3 p-wrap node splitting' => [
'wt' => 'ab
c',
'specs' => [
[
'selector' => 'body > p:nth-of-type(1) > b',
'dsrContent' => [ 'a', '', '' ]
],
[
'selector' => 'body > b',
'dsrContent' => [ 'b
', '', '' ]
],
[
'selector' => 'body > p:nth-of-type(2) > b',
'dsrContent' => [ 'c', '', '' ]
]
]
],
'elt 4 AFE reconstruction' => [
'wt' => 'a
b',
'specs' => [
[
'selector' => 'body > div > p:nth-of-type(1) > b',
'dsrContent' => [ 'a', '', '' ]
],
[
'selector' => 'body > div > p:nth-of-type(2) > b',
'dsrContent' => [ 'b', '', '' ]
],
]
],
'elt 5 AAA' => [
'wt' => 'abc',
'specs' => [
[
'selector' => 'body > div > b',
'dsrContent' => [ '
a', '', '' ]
],
[
'selector' => 'body > div > div',
'dsrContent' => [ 'bc', '
', '' ]
],
]
],
'elt 6 misnested' => [
'wt' => "
''''AB",
'specs' => [
[
'selector' => 'body > p > span > b',
'dsrContent' => [ "'''A", "'''", '' ]
],
[
'selector' => 'body > p > b',
'dsrContent' => [ '
B', '', '' ]
],
]
],
// Links
'link 1' => [
'wt' => 'Foo https://en.wikipedia.org/ bar',
'specs' => [
[
'selector' => 'a',
'dsrContent' => [ 'https://en.wikipedia.org/', '', '' ]
],
]
],
'link 2' => [
'wt' => 'http://example.com/?foo=bar',
'specs' => [
[
'selector' => 'a',
'dsrContent' => [ 'http://example.com/?foo=bar', '', '' ]
]
]
],
'link 3' => [
'wt' => 'RFC 2616',
'specs' => [
[
'selector' => 'a',
'dsrContent' => [ 'RFC 2616', '', '' ]
]
]
],
'link 4' => [
'wt' => '[[Foo]]',
'specs' => [
[
'selector' => 'a',
'dsrContent' => [ '[[Foo]]', '[[', ']]' ]
]
]
],
'link 5' => [
'wt' => '[[Foo|bar]]',
'specs' => [
[
'selector' => 'a',
'dsrContent' => [ '[[Foo|bar]]', '[[Foo|', ']]' ]
]
]
],
'link 6' => [
'wt' => '[https://en.wikipedia.org/ Wikipedia]',
'specs' => [
[
'selector' => 'a',
'dsrContent' => [ '[https://en.wikipedia.org/ Wikipedia]', '[https://en.wikipedia.org/ ', ']' ],
]
]
],
// xmlish extension tags
'xmlish 1' => [
'wt' => '
Like rain gone wooden',
'specs' => [
[
'selector' => 'div.poem',
'dsrContent' => [ '
Like rain gone wooden', '
', '' ]
]
]
],
// Comments
'comment 1' => [
'wt' => '* b',
'specs' => [
[
'selector' => 'li',
'dsrContent' => [ '* b', '*', '' ]
]
]
],
'comment 2' => [
// Misnesting is a way to get the comment length to appear in the output
'wt' => "
AAA
",
'specs' => [
[
'selector' => 'body > div',
'dsrContent' => [ '
AAA
', '
', '
' ]
]
]
],
// Entities
'entity 1' => [
'wt' => '—',
'specs' => [
[
'selector' => 'body > p > span',
'dsrContent' => [ '—', null, null ]
],
]
]
];
}
}