Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 112 |
|
0.00% |
0 / 9 |
CRAP | |
0.00% |
0 / 1 |
| ConversionTraverser | |
0.00% |
0 / 112 |
|
0.00% |
0 / 9 |
1722 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 30 |
|
0.00% |
0 / 1 |
12 | |||
| noConvertHandler | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| anyHandler | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
12 | |||
| langContextHandler | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
2 | |||
| textHandler | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
2 | |||
| aHandler | |
0.00% |
0 / 28 |
|
0.00% |
0 / 1 |
132 | |||
| attrHandler | |
0.00% |
0 / 21 |
|
0.00% |
0 / 1 |
72 | |||
| lcHandler | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
90 | |||
| docFragToString | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
20 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace Wikimedia\Parsoid\Language; |
| 5 | |
| 6 | use Wikimedia\Assert\Assert; |
| 7 | use Wikimedia\Bcp47Code\Bcp47Code; |
| 8 | use Wikimedia\LangConv\ReplacementMachine; |
| 9 | use Wikimedia\Parsoid\Config\Env; |
| 10 | use Wikimedia\Parsoid\Core\DOMCompat; |
| 11 | use Wikimedia\Parsoid\DOM\DocumentFragment; |
| 12 | use Wikimedia\Parsoid\DOM\Element; |
| 13 | use Wikimedia\Parsoid\DOM\Node; |
| 14 | use Wikimedia\Parsoid\DOM\Text; |
| 15 | use Wikimedia\Parsoid\Utils\DOMDataUtils; |
| 16 | use Wikimedia\Parsoid\Utils\DOMTraverser; |
| 17 | use Wikimedia\Parsoid\Utils\DOMUtils; |
| 18 | use Wikimedia\Parsoid\Utils\Utils; |
| 19 | |
| 20 | class ConversionTraverser extends DOMTraverser { |
| 21 | |
| 22 | /** @var Bcp47Code a language code */ |
| 23 | private $toLang; |
| 24 | |
| 25 | /** @var Bcp47Code a language code */ |
| 26 | private $fromLang; |
| 27 | |
| 28 | /** @var LanguageGuesser */ |
| 29 | private $guesser; |
| 30 | |
| 31 | /** @var ReplacementMachine (uses MW-internal codes) */ |
| 32 | private $machine; |
| 33 | |
| 34 | /** |
| 35 | * @param Env $env |
| 36 | * @param Bcp47Code $toLang target language for conversion |
| 37 | * @param LanguageGuesser $guesser oracle to determine "original language" for round-tripping |
| 38 | * @param ReplacementMachine $machine machine to do actual conversion |
| 39 | */ |
| 40 | public function __construct( |
| 41 | Env $env, Bcp47Code $toLang, LanguageGuesser $guesser, ReplacementMachine $machine |
| 42 | ) { |
| 43 | parent::__construct(); |
| 44 | $this->toLang = $toLang; |
| 45 | $this->guesser = $guesser; |
| 46 | $this->machine = $machine; |
| 47 | |
| 48 | // No conversion inside <code>, <script>, <pre>, <cite>, or <style> |
| 49 | // (See adhoc regexps inside LanguageConverter.php::autoConvert) |
| 50 | // XXX: <cite> ought to probably be handled more generically |
| 51 | // as extension output, not special-cased as a HTML tag. |
| 52 | foreach ( [ 'code', 'script', 'pre', 'cite', 'style' ] as $el ) { |
| 53 | $this->addHandler( $el, function ( Element $el ) { |
| 54 | return $this->noConvertHandler( $el ); |
| 55 | } ); |
| 56 | } |
| 57 | // Setting/saving the language context |
| 58 | $this->addHandler( null, function ( Node $node ) { |
| 59 | return $this->anyHandler( $node ); |
| 60 | } ); |
| 61 | $this->addHandler( 'p', function ( Element $el ) { |
| 62 | return $this->langContextHandler( $el ); |
| 63 | } ); |
| 64 | $this->addHandler( 'body', function ( Element $el ) { |
| 65 | return $this->langContextHandler( $el ); |
| 66 | } ); |
| 67 | // Converting #text, <a> nodes, and title/alt attributes |
| 68 | $this->addHandler( '#text', function ( Node $node ) { |
| 69 | return $this->textHandler( $node ); |
| 70 | } ); |
| 71 | $this->addHandler( 'a', function ( Element $el ) use ( $env ){ |
| 72 | return $this->aHandler( $el, $env ); |
| 73 | } ); |
| 74 | $this->addHandler( null, function ( Node $node ) { |
| 75 | return $this->attrHandler( $node ); |
| 76 | } ); |
| 77 | // LanguageConverter markup |
| 78 | foreach ( [ 'meta', 'div', 'span' ] as $el ) { |
| 79 | $this->addHandler( $el, function ( Element $el ) { |
| 80 | return $this->lcHandler( $el ); |
| 81 | } ); |
| 82 | } |
| 83 | } |
| 84 | |
| 85 | /** |
| 86 | * @param Element $el |
| 87 | * @return ?Node|bool |
| 88 | */ |
| 89 | private function noConvertHandler( Element $el ) { |
| 90 | // Don't touch the inside of this node! |
| 91 | return $el->nextSibling; |
| 92 | } |
| 93 | |
| 94 | /** |
| 95 | * @param Node $node |
| 96 | * @return ?Node|bool |
| 97 | */ |
| 98 | private function anyHandler( Node $node ) { |
| 99 | /* Look for `lang` attributes */ |
| 100 | if ( $node instanceof Element ) { |
| 101 | if ( $node->hasAttribute( 'lang' ) ) { |
| 102 | $lang = DOMCompat::getAttribute( $node, 'lang' ); |
| 103 | // XXX validate lang! override fromLang? |
| 104 | // $this->>fromLang = $lang; |
| 105 | } |
| 106 | } |
| 107 | return true; // Continue with other handlers |
| 108 | } |
| 109 | |
| 110 | /** |
| 111 | * @param Element $el |
| 112 | * @return ?Node|bool |
| 113 | */ |
| 114 | private function langContextHandler( Element $el ) { |
| 115 | $this->fromLang = $this->guesser->guessLang( $el ); |
| 116 | // T320662: use internal MW language names for now :( |
| 117 | $fromLangMw = Utils::bcp47ToMwCode( $this->fromLang ); |
| 118 | $el->setAttribute( 'data-mw-variant-lang', $fromLangMw ); |
| 119 | return true; // Continue with other handlers |
| 120 | } |
| 121 | |
| 122 | /** |
| 123 | * @param Node $node |
| 124 | * @return ?Node|bool |
| 125 | */ |
| 126 | private function textHandler( Node $node ) { |
| 127 | Assert::invariant( $this->fromLang !== null, 'Text w/o a context' ); |
| 128 | $toLangMw = Utils::bcp47ToMwCode( $this->toLang ); |
| 129 | $fromLangMw = Utils::bcp47ToMwCode( $this->fromLang ); |
| 130 | // @phan-suppress-next-line PhanTypeMismatchArgument,PhanTypeMismatchReturn both declared as DOMNode |
| 131 | return $this->machine->replace( $node, $toLangMw, $fromLangMw ); |
| 132 | } |
| 133 | |
| 134 | /** |
| 135 | * @param Element $el |
| 136 | * @param Env $env |
| 137 | * @return ?Node|bool |
| 138 | */ |
| 139 | private function aHandler( Element $el, Env $env ) { |
| 140 | // Is this a wikilink? If so, extract title & convert it |
| 141 | if ( DOMUtils::hasRel( $el, 'mw:WikiLink' ) ) { |
| 142 | $href = preg_replace( '#^(\.\.?/)+#', '', DOMCompat::getAttribute( $el, 'href' ) ?? '', 1 ); |
| 143 | $fromPage = Utils::decodeURI( $href ); |
| 144 | $toPageFrag = $this->machine->convert( |
| 145 | $el->ownerDocument, $fromPage, |
| 146 | Utils::bcp47ToMwCode( $this->toLang ), |
| 147 | Utils::bcp47ToMwCode( $this->fromLang ) |
| 148 | ); |
| 149 | '@phan-var DocumentFragment $toPageFrag'; // @var DocumentFragment $toPageFrag |
| 150 | $toPage = $this->docFragToString( $toPageFrag ); |
| 151 | if ( $toPage === null ) { |
| 152 | // Non-reversible transform (sigh); mark this for rt. |
| 153 | $el->setAttribute( 'data-mw-variant-orig', $fromPage ); |
| 154 | $toPage = $this->docFragToString( $toPageFrag, true/* force */ ); |
| 155 | } |
| 156 | if ( $el->hasAttribute( 'title' ) ) { |
| 157 | $el->setAttribute( 'title', str_replace( '_', ' ', $toPage ) ); |
| 158 | } |
| 159 | $el->setAttribute( 'href', "./{$toPage}" ); |
| 160 | } elseif ( DOMUtils::hasRel( $el, 'mw:WikiLink/Interwiki' ) ) { |
| 161 | // Don't convert title or children of interwiki links |
| 162 | return $el->nextSibling; |
| 163 | } elseif ( DOMUtils::hasRel( $el, 'mw:ExtLink' ) ) { |
| 164 | // WTUtils.usesURLLinkSyntax uses data-parsoid, so don't use it, |
| 165 | // but syntactic free links should also have class="external free" |
| 166 | if ( DOMUtils::hasClass( $el, 'free' ) ) { |
| 167 | // Don't convert children of syntactic "free links" |
| 168 | return $el->nextSibling; |
| 169 | } |
| 170 | // Other external link text is protected from conversion iff |
| 171 | // (a) it doesn't starts/end with -{ ... }- |
| 172 | if ( $el->firstChild && |
| 173 | DOMUtils::hasTypeOf( $el->firstChild, 'mw:LanguageVariant' ) ) { |
| 174 | return true; |
| 175 | } |
| 176 | // (b) it looks like a URL (protocol-relative links excluded) |
| 177 | $linkText = $el->textContent; // XXX: this could be expensive |
| 178 | if ( Utils::isProtocolValid( $linkText, $env ) |
| 179 | && !str_starts_with( $linkText, '//' ) |
| 180 | ) { |
| 181 | return $el->nextSibling; |
| 182 | } |
| 183 | } |
| 184 | return true; |
| 185 | } |
| 186 | |
| 187 | /** |
| 188 | * @param Node $node |
| 189 | * @return ?Node|bool |
| 190 | */ |
| 191 | private function attrHandler( Node $node ) { |
| 192 | // Convert `alt` and `title` attributes on elements |
| 193 | // (Called before aHandler, so the `title` might get overwritten there) |
| 194 | if ( !( $node instanceof Element ) ) { |
| 195 | return true; |
| 196 | } |
| 197 | foreach ( [ 'title', 'alt' ] as $attr ) { |
| 198 | $orig = DOMCompat::getAttribute( $node, $attr ); |
| 199 | if ( $orig === null ) { |
| 200 | continue; |
| 201 | } |
| 202 | if ( $attr === 'title' && DOMUtils::hasRel( $node, 'mw:WikiLink' ) ) { |
| 203 | // We've already converted the title in aHandler above. |
| 204 | continue; |
| 205 | } |
| 206 | if ( str_contains( $orig, '://' ) ) { |
| 207 | continue; /* Don't convert URLs */ |
| 208 | } |
| 209 | $toFrag = $this->machine->convert( |
| 210 | $node->ownerDocument, $orig, |
| 211 | Utils::bcp47ToMwCode( $this->toLang ), |
| 212 | Utils::bcp47ToMwCode( $this->fromLang ) |
| 213 | ); |
| 214 | '@phan-var DocumentFragment $toFrag'; // @var DocumentFragment $toFrag |
| 215 | $to = $this->docFragToString( $toFrag ); |
| 216 | if ( $to === null ) { |
| 217 | // Non-reversible transform (sigh); mark for rt. |
| 218 | $node->setAttribute( "data-mw-variant-{$attr}", $orig ); |
| 219 | $to = $this->docFragToString( $toFrag, true/* force */ ); |
| 220 | } |
| 221 | $node->setAttribute( $attr, $to ); |
| 222 | } |
| 223 | return true; |
| 224 | } |
| 225 | |
| 226 | /** |
| 227 | * Handler for LanguageConverter markup |
| 228 | * |
| 229 | * @param Element $el |
| 230 | * @return ?Node|bool |
| 231 | */ |
| 232 | private function lcHandler( Element $el ) { |
| 233 | if ( !DOMUtils::hasTypeOf( $el, 'mw:LanguageVariant' ) ) { |
| 234 | return true; /* not language converter markup */ |
| 235 | } |
| 236 | $dmv = DOMDataUtils::getDataMwVariant( $el ); |
| 237 | if ( $dmv->disabled ?? false ) { |
| 238 | $df = DOMDataUtils::cloneDocumentFragment( $dmv->disabled ); |
| 239 | while ( $el->firstChild !== null ) { |
| 240 | $el->removeChild( $el->firstChild ); |
| 241 | } |
| 242 | DOMCompat::appendChild( $el, $df ); |
| 243 | return $el->nextSibling; |
| 244 | } elseif ( $dmv->twoway ?? false ) { |
| 245 | // FIXME |
| 246 | } elseif ( $dmv->oneway ?? false ) { |
| 247 | // FIXME |
| 248 | } elseif ( $dmv->name ?? false ) { |
| 249 | // FIXME |
| 250 | } elseif ( $dmv->filter ?? false ) { |
| 251 | // FIXME |
| 252 | } elseif ( $dmv->describe ?? false ) { |
| 253 | // FIXME |
| 254 | } |
| 255 | return true; |
| 256 | } |
| 257 | |
| 258 | private function docFragToString( |
| 259 | DocumentFragment $docFrag, bool $force = false |
| 260 | ): ?string { |
| 261 | if ( !$force ) { |
| 262 | for ( $child = $docFrag->firstChild; $child; $child = $child->nextSibling ) { |
| 263 | if ( !( $child instanceof Text ) ) { |
| 264 | return null; /* unsafe */ |
| 265 | } |
| 266 | } |
| 267 | } |
| 268 | return $docFrag->textContent; |
| 269 | } |
| 270 | } |