Warn if stateful ParserOutput transforms are used
[lhc/web/wiklou.git] / includes / parser / LinkHolderArray.php
1 <?php
2 /**
3 * Holder of replacement pairs for wiki links
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @file
21 * @ingroup Parser
22 */
23
24 /**
25 * @ingroup Parser
26 */
27 class LinkHolderArray {
28 public $internals = [];
29 public $interwikis = [];
30 public $size = 0;
31
32 /**
33 * @var Parser
34 */
35 public $parent;
36 protected $tempIdOffset;
37
38 /**
39 * @param Parser $parent
40 */
41 public function __construct( $parent ) {
42 $this->parent = $parent;
43 }
44
45 /**
46 * Reduce memory usage to reduce the impact of circular references
47 */
48 public function __destruct() {
49 foreach ( $this as $name => $value ) {
50 unset( $this->$name );
51 }
52 }
53
54 /**
55 * Don't serialize the parent object, it is big, and not needed when it is
56 * a parameter to mergeForeign(), which is the only application of
57 * serializing at present.
58 *
59 * Compact the titles, only serialize the text form.
60 * @return array
61 */
62 public function __sleep() {
63 foreach ( $this->internals as &$nsLinks ) {
64 foreach ( $nsLinks as &$entry ) {
65 unset( $entry['title'] );
66 }
67 }
68 unset( $nsLinks );
69 unset( $entry );
70
71 foreach ( $this->interwikis as &$entry ) {
72 unset( $entry['title'] );
73 }
74 unset( $entry );
75
76 return [ 'internals', 'interwikis', 'size' ];
77 }
78
79 /**
80 * Recreate the Title objects
81 */
82 public function __wakeup() {
83 foreach ( $this->internals as &$nsLinks ) {
84 foreach ( $nsLinks as &$entry ) {
85 $entry['title'] = Title::newFromText( $entry['pdbk'] );
86 }
87 }
88 unset( $nsLinks );
89 unset( $entry );
90
91 foreach ( $this->interwikis as &$entry ) {
92 $entry['title'] = Title::newFromText( $entry['pdbk'] );
93 }
94 unset( $entry );
95 }
96
97 /**
98 * Merge another LinkHolderArray into this one
99 * @param LinkHolderArray $other
100 */
101 public function merge( $other ) {
102 foreach ( $other->internals as $ns => $entries ) {
103 $this->size += count( $entries );
104 if ( !isset( $this->internals[$ns] ) ) {
105 $this->internals[$ns] = $entries;
106 } else {
107 $this->internals[$ns] += $entries;
108 }
109 }
110 $this->interwikis += $other->interwikis;
111 }
112
113 /**
114 * Merge a LinkHolderArray from another parser instance into this one. The
115 * keys will not be preserved. Any text which went with the old
116 * LinkHolderArray and needs to work with the new one should be passed in
117 * the $texts array. The strings in this array will have their link holders
118 * converted for use in the destination link holder. The resulting array of
119 * strings will be returned.
120 *
121 * @param LinkHolderArray $other
122 * @param array $texts Array of strings
123 * @return array
124 */
125 public function mergeForeign( $other, $texts ) {
126 $this->tempIdOffset = $idOffset = $this->parent->nextLinkID();
127 $maxId = 0;
128
129 # Renumber internal links
130 foreach ( $other->internals as $ns => $nsLinks ) {
131 foreach ( $nsLinks as $key => $entry ) {
132 $newKey = $idOffset + $key;
133 $this->internals[$ns][$newKey] = $entry;
134 $maxId = $newKey > $maxId ? $newKey : $maxId;
135 }
136 }
137 $texts = preg_replace_callback( '/(<!--LINK \d+:)(\d+)(-->)/',
138 [ $this, 'mergeForeignCallback' ], $texts );
139
140 # Renumber interwiki links
141 foreach ( $other->interwikis as $key => $entry ) {
142 $newKey = $idOffset + $key;
143 $this->interwikis[$newKey] = $entry;
144 $maxId = $newKey > $maxId ? $newKey : $maxId;
145 }
146 $texts = preg_replace_callback( '/(<!--IWLINK )(\d+)(-->)/',
147 [ $this, 'mergeForeignCallback' ], $texts );
148
149 # Set the parent link ID to be beyond the highest used ID
150 $this->parent->setLinkID( $maxId + 1 );
151 $this->tempIdOffset = null;
152 return $texts;
153 }
154
155 /**
156 * @param array $m
157 * @return string
158 */
159 protected function mergeForeignCallback( $m ) {
160 return $m[1] . ( $m[2] + $this->tempIdOffset ) . $m[3];
161 }
162
163 /**
164 * Get a subset of the current LinkHolderArray which is sufficient to
165 * interpret the given text.
166 * @param string $text
167 * @return LinkHolderArray
168 */
169 public function getSubArray( $text ) {
170 $sub = new LinkHolderArray( $this->parent );
171
172 # Internal links
173 $pos = 0;
174 while ( $pos < strlen( $text ) ) {
175 if ( !preg_match( '/<!--LINK (\d+):(\d+)-->/',
176 $text, $m, PREG_OFFSET_CAPTURE, $pos )
177 ) {
178 break;
179 }
180 $ns = $m[1][0];
181 $key = $m[2][0];
182 $sub->internals[$ns][$key] = $this->internals[$ns][$key];
183 $pos = $m[0][1] + strlen( $m[0][0] );
184 }
185
186 # Interwiki links
187 $pos = 0;
188 while ( $pos < strlen( $text ) ) {
189 if ( !preg_match( '/<!--IWLINK (\d+)-->/', $text, $m, PREG_OFFSET_CAPTURE, $pos ) ) {
190 break;
191 }
192 $key = $m[1][0];
193 $sub->interwikis[$key] = $this->interwikis[$key];
194 $pos = $m[0][1] + strlen( $m[0][0] );
195 }
196 return $sub;
197 }
198
199 /**
200 * Returns true if the memory requirements of this object are getting large
201 * @return bool
202 */
203 public function isBig() {
204 global $wgLinkHolderBatchSize;
205 return $this->size > $wgLinkHolderBatchSize;
206 }
207
208 /**
209 * Clear all stored link holders.
210 * Make sure you don't have any text left using these link holders, before you call this
211 */
212 public function clear() {
213 $this->internals = [];
214 $this->interwikis = [];
215 $this->size = 0;
216 }
217
218 /**
219 * Make a link placeholder. The text returned can be later resolved to a real link with
220 * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
221 * parsing of interwiki links, and secondly to allow all existence checks and
222 * article length checks (for stub links) to be bundled into a single query.
223 *
224 * @param Title $nt
225 * @param string $text
226 * @param array $query [optional]
227 * @param string $trail [optional]
228 * @param string $prefix [optional]
229 * @return string
230 */
231 public function makeHolder( $nt, $text = '', $query = [], $trail = '', $prefix = '' ) {
232 if ( !is_object( $nt ) ) {
233 # Fail gracefully
234 $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
235 } else {
236 # Separate the link trail from the rest of the link
237 list( $inside, $trail ) = Linker::splitTrail( $trail );
238
239 $entry = [
240 'title' => $nt,
241 'text' => $prefix . $text . $inside,
242 'pdbk' => $nt->getPrefixedDBkey(),
243 ];
244 if ( $query !== [] ) {
245 $entry['query'] = $query;
246 }
247
248 if ( $nt->isExternal() ) {
249 // Use a globally unique ID to keep the objects mergable
250 $key = $this->parent->nextLinkID();
251 $this->interwikis[$key] = $entry;
252 $retVal = "<!--IWLINK $key-->{$trail}";
253 } else {
254 $key = $this->parent->nextLinkID();
255 $ns = $nt->getNamespace();
256 $this->internals[$ns][$key] = $entry;
257 $retVal = "<!--LINK $ns:$key-->{$trail}";
258 }
259 $this->size++;
260 }
261 return $retVal;
262 }
263
264 /**
265 * Replace <!--LINK--> link placeholders with actual links, in the buffer
266 *
267 * @param string &$text
268 */
269 public function replace( &$text ) {
270 $this->replaceInternal( $text );
271 $this->replaceInterwiki( $text );
272 }
273
274 /**
275 * Replace internal links
276 * @param string &$text
277 */
278 protected function replaceInternal( &$text ) {
279 if ( !$this->internals ) {
280 return;
281 }
282
283 global $wgContLang;
284
285 $colours = [];
286 $linkCache = LinkCache::singleton();
287 $output = $this->parent->getOutput();
288 $linkRenderer = $this->parent->getLinkRenderer();
289
290 $dbr = wfGetDB( DB_REPLICA );
291
292 # Sort by namespace
293 ksort( $this->internals );
294
295 $linkcolour_ids = [];
296
297 # Generate query
298 $lb = new LinkBatch();
299 $lb->setCaller( __METHOD__ );
300
301 foreach ( $this->internals as $ns => $entries ) {
302 foreach ( $entries as $entry ) {
303 /** @var Title $title */
304 $title = $entry['title'];
305 $pdbk = $entry['pdbk'];
306
307 # Skip invalid entries.
308 # Result will be ugly, but prevents crash.
309 if ( is_null( $title ) ) {
310 continue;
311 }
312
313 # Check if it's a static known link, e.g. interwiki
314 if ( $title->isAlwaysKnown() ) {
315 $colours[$pdbk] = '';
316 } elseif ( $ns == NS_SPECIAL ) {
317 $colours[$pdbk] = 'new';
318 } else {
319 $id = $linkCache->getGoodLinkID( $pdbk );
320 if ( $id != 0 ) {
321 $colours[$pdbk] = $linkRenderer->getLinkClasses( $title );
322 $output->addLink( $title, $id );
323 $linkcolour_ids[$id] = $pdbk;
324 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
325 $colours[$pdbk] = 'new';
326 } else {
327 # Not in the link cache, add it to the query
328 $lb->addObj( $title );
329 }
330 }
331 }
332 }
333 if ( !$lb->isEmpty() ) {
334 $fields = array_merge(
335 LinkCache::getSelectFields(),
336 [ 'page_namespace', 'page_title' ]
337 );
338
339 $res = $dbr->select(
340 'page',
341 $fields,
342 $lb->constructSet( 'page', $dbr ),
343 __METHOD__
344 );
345
346 # Fetch data and form into an associative array
347 # non-existent = broken
348 foreach ( $res as $s ) {
349 $title = Title::makeTitle( $s->page_namespace, $s->page_title );
350 $pdbk = $title->getPrefixedDBkey();
351 $linkCache->addGoodLinkObjFromRow( $title, $s );
352 $output->addLink( $title, $s->page_id );
353 $colours[$pdbk] = $linkRenderer->getLinkClasses( $title );
354 // add id to the extension todolist
355 $linkcolour_ids[$s->page_id] = $pdbk;
356 }
357 unset( $res );
358 }
359 if ( count( $linkcolour_ids ) ) {
360 // pass an array of page_ids to an extension
361 Hooks::run( 'GetLinkColours', [ $linkcolour_ids, &$colours ] );
362 }
363
364 # Do a second query for different language variants of links and categories
365 if ( $wgContLang->hasVariants() ) {
366 $this->doVariants( $colours );
367 }
368
369 # Construct search and replace arrays
370 $replacePairs = [];
371 foreach ( $this->internals as $ns => $entries ) {
372 foreach ( $entries as $index => $entry ) {
373 $pdbk = $entry['pdbk'];
374 $title = $entry['title'];
375 $query = isset( $entry['query'] ) ? $entry['query'] : [];
376 $key = "$ns:$index";
377 $searchkey = "<!--LINK $key-->";
378 $displayText = $entry['text'];
379 if ( isset( $entry['selflink'] ) ) {
380 $replacePairs[$searchkey] = Linker::makeSelfLinkObj( $title, $displayText, $query );
381 continue;
382 }
383 if ( $displayText === '' ) {
384 $displayText = null;
385 } else {
386 $displayText = new HtmlArmor( $displayText );
387 }
388 if ( !isset( $colours[$pdbk] ) ) {
389 $colours[$pdbk] = 'new';
390 }
391 $attribs = [];
392 if ( $colours[$pdbk] == 'new' ) {
393 $linkCache->addBadLinkObj( $title );
394 $output->addLink( $title, 0 );
395 $link = $linkRenderer->makeBrokenLink(
396 $title, $displayText, $attribs, $query
397 );
398 } else {
399 $link = $linkRenderer->makePreloadedLink(
400 $title, $displayText, $colours[$pdbk], $attribs, $query
401 );
402 }
403
404 $replacePairs[$searchkey] = $link;
405 }
406 }
407 $replacer = new HashtableReplacer( $replacePairs, 1 );
408
409 # Do the thing
410 $text = preg_replace_callback(
411 '/(<!--LINK .*?-->)/',
412 $replacer->cb(),
413 $text
414 );
415 }
416
417 /**
418 * Replace interwiki links
419 * @param string &$text
420 */
421 protected function replaceInterwiki( &$text ) {
422 if ( empty( $this->interwikis ) ) {
423 return;
424 }
425
426 # Make interwiki link HTML
427 $output = $this->parent->getOutput();
428 $replacePairs = [];
429 $linkRenderer = $this->parent->getLinkRenderer();
430 foreach ( $this->interwikis as $key => $link ) {
431 $replacePairs[$key] = $linkRenderer->makeLink(
432 $link['title'],
433 new HtmlArmor( $link['text'] )
434 );
435 $output->addInterwikiLink( $link['title'] );
436 }
437 $replacer = new HashtableReplacer( $replacePairs, 1 );
438
439 $text = preg_replace_callback(
440 '/<!--IWLINK (.*?)-->/',
441 $replacer->cb(),
442 $text );
443 }
444
445 /**
446 * Modify $this->internals and $colours according to language variant linking rules
447 * @param array &$colours
448 */
449 protected function doVariants( &$colours ) {
450 global $wgContLang;
451 $linkBatch = new LinkBatch();
452 $variantMap = []; // maps $pdbkey_Variant => $keys (of link holders)
453 $output = $this->parent->getOutput();
454 $linkCache = LinkCache::singleton();
455 $titlesToBeConverted = '';
456 $titlesAttrs = [];
457
458 // Concatenate titles to a single string, thus we only need auto convert the
459 // single string to all variants. This would improve parser's performance
460 // significantly.
461 foreach ( $this->internals as $ns => $entries ) {
462 if ( $ns == NS_SPECIAL ) {
463 continue;
464 }
465 foreach ( $entries as $index => $entry ) {
466 $pdbk = $entry['pdbk'];
467 // we only deal with new links (in its first query)
468 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] === 'new' ) {
469 $titlesAttrs[] = [ $index, $entry['title'] ];
470 // separate titles with \0 because it would never appears
471 // in a valid title
472 $titlesToBeConverted .= $entry['title']->getText() . "\0";
473 }
474 }
475 }
476
477 // Now do the conversion and explode string to text of titles
478 $titlesAllVariants = $wgContLang->autoConvertToAllVariants( rtrim( $titlesToBeConverted, "\0" ) );
479 $allVariantsName = array_keys( $titlesAllVariants );
480 foreach ( $titlesAllVariants as &$titlesVariant ) {
481 $titlesVariant = explode( "\0", $titlesVariant );
482 }
483
484 // Then add variants of links to link batch
485 $parentTitle = $this->parent->getTitle();
486 foreach ( $titlesAttrs as $i => $attrs ) {
487 /** @var Title $title */
488 list( $index, $title ) = $attrs;
489 $ns = $title->getNamespace();
490 $text = $title->getText();
491
492 foreach ( $allVariantsName as $variantName ) {
493 $textVariant = $titlesAllVariants[$variantName][$i];
494 if ( $textVariant === $text ) {
495 continue;
496 }
497
498 $variantTitle = Title::makeTitle( $ns, $textVariant );
499
500 // Self-link checking for mixed/different variant titles. At this point, we
501 // already know the exact title does not exist, so the link cannot be to a
502 // variant of the current title that exists as a separate page.
503 if ( $variantTitle->equals( $parentTitle ) && !$title->hasFragment() ) {
504 $this->internals[$ns][$index]['selflink'] = true;
505 continue 2;
506 }
507
508 $linkBatch->addObj( $variantTitle );
509 $variantMap[$variantTitle->getPrefixedDBkey()][] = "$ns:$index";
510 }
511 }
512
513 // process categories, check if a category exists in some variant
514 $categoryMap = []; // maps $category_variant => $category (dbkeys)
515 $varCategories = []; // category replacements oldDBkey => newDBkey
516 foreach ( $output->getCategoryLinks() as $category ) {
517 $categoryTitle = Title::makeTitleSafe( NS_CATEGORY, $category );
518 $linkBatch->addObj( $categoryTitle );
519 $variants = $wgContLang->autoConvertToAllVariants( $category );
520 foreach ( $variants as $variant ) {
521 if ( $variant !== $category ) {
522 $variantTitle = Title::makeTitleSafe( NS_CATEGORY, $variant );
523 if ( is_null( $variantTitle ) ) {
524 continue;
525 }
526 $linkBatch->addObj( $variantTitle );
527 $categoryMap[$variant] = [ $category, $categoryTitle ];
528 }
529 }
530 }
531
532 if ( !$linkBatch->isEmpty() ) {
533 // construct query
534 $dbr = wfGetDB( DB_REPLICA );
535 $fields = array_merge(
536 LinkCache::getSelectFields(),
537 [ 'page_namespace', 'page_title' ]
538 );
539
540 $varRes = $dbr->select( 'page',
541 $fields,
542 $linkBatch->constructSet( 'page', $dbr ),
543 __METHOD__
544 );
545
546 $linkcolour_ids = [];
547 $linkRenderer = $this->parent->getLinkRenderer();
548
549 // for each found variants, figure out link holders and replace
550 foreach ( $varRes as $s ) {
551 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
552 $varPdbk = $variantTitle->getPrefixedDBkey();
553 $vardbk = $variantTitle->getDBkey();
554
555 $holderKeys = [];
556 if ( isset( $variantMap[$varPdbk] ) ) {
557 $holderKeys = $variantMap[$varPdbk];
558 $linkCache->addGoodLinkObjFromRow( $variantTitle, $s );
559 $output->addLink( $variantTitle, $s->page_id );
560 }
561
562 // loop over link holders
563 foreach ( $holderKeys as $key ) {
564 list( $ns, $index ) = explode( ':', $key, 2 );
565 $entry =& $this->internals[$ns][$index];
566 $pdbk = $entry['pdbk'];
567
568 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] === 'new' ) {
569 // found link in some of the variants, replace the link holder data
570 $entry['title'] = $variantTitle;
571 $entry['pdbk'] = $varPdbk;
572
573 // set pdbk and colour
574 $colours[$varPdbk] = $linkRenderer->getLinkClasses( $variantTitle );
575 $linkcolour_ids[$s->page_id] = $pdbk;
576 }
577 }
578
579 // check if the object is a variant of a category
580 if ( isset( $categoryMap[$vardbk] ) ) {
581 list( $oldkey, $oldtitle ) = $categoryMap[$vardbk];
582 if ( !isset( $varCategories[$oldkey] ) && !$oldtitle->exists() ) {
583 $varCategories[$oldkey] = $vardbk;
584 }
585 }
586 }
587 Hooks::run( 'GetLinkColours', [ $linkcolour_ids, &$colours ] );
588
589 // rebuild the categories in original order (if there are replacements)
590 if ( count( $varCategories ) > 0 ) {
591 $newCats = [];
592 $originalCats = $output->getCategories();
593 foreach ( $originalCats as $cat => $sortkey ) {
594 // make the replacement
595 if ( array_key_exists( $cat, $varCategories ) ) {
596 $newCats[$varCategories[$cat]] = $sortkey;
597 } else {
598 $newCats[$cat] = $sortkey;
599 }
600 }
601 $output->setCategoryLinks( $newCats );
602 }
603 }
604 }
605
606 /**
607 * Replace <!--LINK--> link placeholders with plain text of links
608 * (not HTML-formatted).
609 *
610 * @param string $text
611 * @return string
612 */
613 public function replaceText( $text ) {
614 $text = preg_replace_callback(
615 '/<!--(LINK|IWLINK) (.*?)-->/',
616 [ $this, 'replaceTextCallback' ],
617 $text );
618
619 return $text;
620 }
621
622 /**
623 * Callback for replaceText()
624 *
625 * @param array $matches
626 * @return string
627 * @private
628 */
629 public function replaceTextCallback( $matches ) {
630 $type = $matches[1];
631 $key = $matches[2];
632 if ( $type == 'LINK' ) {
633 list( $ns, $index ) = explode( ':', $key, 2 );
634 if ( isset( $this->internals[$ns][$index]['text'] ) ) {
635 return $this->internals[$ns][$index]['text'];
636 }
637 } elseif ( $type == 'IWLINK' ) {
638 if ( isset( $this->interwikis[$key]['text'] ) ) {
639 return $this->interwikis[$key]['text'];
640 }
641 }
642 return $matches[0];
643 }
644 }