Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
Total | |
0.00% |
0 / 110 |
|
0.00% |
0 / 9 |
CRAP | |
0.00% |
0 / 1 |
ConversionTraverser | |
0.00% |
0 / 110 |
|
0.00% |
0 / 9 |
1640 | |
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 / 22 |
|
0.00% |
0 / 1 |
72 | |||
lcHandler | |
0.00% |
0 / 12 |
|
0.00% |
0 / 1 |
72 | |||
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\DOM\DocumentFragment; |
11 | use Wikimedia\Parsoid\DOM\Element; |
12 | use Wikimedia\Parsoid\DOM\Node; |
13 | use Wikimedia\Parsoid\DOM\Text; |
14 | use Wikimedia\Parsoid\Utils\DOMCompat; |
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> |
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' ] 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 | && substr( $linkText, 0, 2 ) !== '//' |
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 | DOMUtils::assertElt( $node ); |
198 | foreach ( [ 'title', 'alt' ] as $attr ) { |
199 | $orig = DOMCompat::getAttribute( $node, $attr ); |
200 | if ( $orig === null ) { |
201 | continue; |
202 | } |
203 | if ( $attr === 'title' && DOMUtils::hasRel( $node, 'mw:WikiLink' ) ) { |
204 | // We've already converted the title in aHandler above. |
205 | continue; |
206 | } |
207 | if ( str_contains( $orig, '://' ) ) { |
208 | continue; /* Don't convert URLs */ |
209 | } |
210 | $toFrag = $this->machine->convert( |
211 | $node->ownerDocument, $orig, |
212 | Utils::bcp47ToMwCode( $this->toLang ), |
213 | Utils::bcp47ToMwCode( $this->fromLang ) |
214 | ); |
215 | '@phan-var DocumentFragment $toFrag'; // @var DocumentFragment $toFrag |
216 | $to = $this->docFragToString( $toFrag ); |
217 | if ( $to === null ) { |
218 | // Non-reversible transform (sigh); mark for rt. |
219 | $node->setAttribute( "data-mw-variant-{$attr}", $orig ); |
220 | $to = $this->docFragToString( $toFrag, true/* force */ ); |
221 | } |
222 | $node->setAttribute( $attr, $to ); |
223 | } |
224 | return true; |
225 | } |
226 | |
227 | /** |
228 | * Handler for LanguageConverter markup |
229 | * |
230 | * @param Element $el |
231 | * @return ?Node|bool |
232 | */ |
233 | private function lcHandler( Element $el ) { |
234 | if ( !DOMUtils::hasTypeOf( $el, 'mw:LanguageVariant' ) ) { |
235 | return true; /* not language converter markup */ |
236 | } |
237 | $dmv = DOMDataUtils::getJSONAttribute( $el, 'data-mw-variant', [] ); |
238 | if ( isset( $dmv->disabled ) ) { |
239 | DOMCompat::setInnerHTML( $el, $dmv->disabled->t ); |
240 | // XXX check handling of embedded data-parsoid |
241 | // XXX check handling of nested constructs |
242 | return $el->nextSibling; |
243 | } elseif ( isset( $dmv->twoway ) ) { |
244 | // FIXME |
245 | } elseif ( isset( $dmv->oneway ) ) { |
246 | // FIXME |
247 | } elseif ( isset( $dmv->name ) ) { |
248 | // FIXME |
249 | } elseif ( isset( $dmv->filter ) ) { |
250 | // FIXME |
251 | } elseif ( isset( $dmv->describe ) ) { |
252 | // FIXME |
253 | } |
254 | return true; |
255 | } |
256 | |
257 | private function docFragToString( |
258 | DocumentFragment $docFrag, bool $force = false |
259 | ): ?string { |
260 | if ( !$force ) { |
261 | for ( $child = $docFrag->firstChild; $child; $child = $child->nextSibling ) { |
262 | if ( !( $child instanceof Text ) ) { |
263 | return null; /* unsafe */ |
264 | } |
265 | } |
266 | } |
267 | return $docFrag->textContent; |
268 | } |
269 | } |