Complex classes like Doku_Indexer often do a lot of different things. To break such a class down, we need to identify a cohesive component within that class. A common approach to find such a component is to look for fields/methods that share the same prefixes, or suffixes. You can also have a look at the cohesion graph to spot any un-connected, or weakly-connected components.
Once you have determined the fields that belong together, you can apply the Extract Class refactoring. If the component makes sense as a sub-class, Extract Subclass is also a candidate, and is often faster.
While breaking up the class, it is a good idea to analyze how other classes use Doku_Indexer, and based on these observations, apply Extract Interface, too.
1 | <?php |
||
109 | class Doku_Indexer { |
||
110 | /** |
||
111 | * @var array $pidCache Cache for getPID() |
||
112 | */ |
||
113 | protected $pidCache = array(); |
||
114 | |||
115 | /** |
||
116 | * Adds the contents of a page to the fulltext index |
||
117 | * |
||
118 | * The added text replaces previous words for the same page. |
||
119 | * An empty value erases the page. |
||
120 | * |
||
121 | * @param string $page a page name |
||
122 | * @param string $text the body of the page |
||
123 | * @return string|boolean the function completed successfully |
||
124 | * |
||
125 | * @author Tom N Harris <[email protected]> |
||
126 | * @author Andreas Gohr <[email protected]> |
||
127 | */ |
||
128 | public function addPageWords($page, $text) { |
||
129 | if (!$this->lock()) |
||
130 | return "locked"; |
||
131 | |||
132 | // load known documents |
||
133 | $pid = $this->getPIDNoLock($page); |
||
134 | if ($pid === false) { |
||
135 | $this->unlock(); |
||
136 | return false; |
||
137 | } |
||
138 | |||
139 | $pagewords = array(); |
||
140 | // get word usage in page |
||
141 | $words = $this->getPageWords($text); |
||
142 | if ($words === false) { |
||
143 | $this->unlock(); |
||
144 | return false; |
||
145 | } |
||
146 | |||
147 | if (!empty($words)) { |
||
148 | foreach (array_keys($words) as $wlen) { |
||
149 | $index = $this->getIndex('i', $wlen); |
||
150 | foreach ($words[$wlen] as $wid => $freq) { |
||
151 | $idx = ($wid<count($index)) ? $index[$wid] : ''; |
||
152 | $index[$wid] = $this->updateTuple($idx, $pid, $freq); |
||
|
|||
153 | $pagewords[] = "$wlen*$wid"; |
||
154 | } |
||
155 | if (!$this->saveIndex('i', $wlen, $index)) { |
||
156 | $this->unlock(); |
||
157 | return false; |
||
158 | } |
||
159 | } |
||
160 | } |
||
161 | |||
162 | // Remove obsolete index entries |
||
163 | $pageword_idx = $this->getIndexKey('pageword', '', $pid); |
||
164 | if ($pageword_idx !== '') { |
||
165 | $oldwords = explode(':',$pageword_idx); |
||
166 | $delwords = array_diff($oldwords, $pagewords); |
||
167 | $upwords = array(); |
||
168 | foreach ($delwords as $word) { |
||
169 | if ($word != '') { |
||
170 | list($wlen,$wid) = explode('*', $word); |
||
171 | $wid = (int)$wid; |
||
172 | $upwords[$wlen][] = $wid; |
||
173 | } |
||
174 | } |
||
175 | foreach ($upwords as $wlen => $widx) { |
||
176 | $index = $this->getIndex('i', $wlen); |
||
177 | foreach ($widx as $wid) { |
||
178 | $index[$wid] = $this->updateTuple($index[$wid], $pid, 0); |
||
179 | } |
||
180 | $this->saveIndex('i', $wlen, $index); |
||
181 | } |
||
182 | } |
||
183 | // Save the reverse index |
||
184 | $pageword_idx = join(':', $pagewords); |
||
185 | if (!$this->saveIndexKey('pageword', '', $pid, $pageword_idx)) { |
||
186 | $this->unlock(); |
||
187 | return false; |
||
188 | } |
||
189 | |||
190 | $this->unlock(); |
||
191 | return true; |
||
192 | } |
||
193 | |||
194 | /** |
||
195 | * Split the words in a page and add them to the index. |
||
196 | * |
||
197 | * @param string $text content of the page |
||
198 | * @return array list of word IDs and number of times used |
||
199 | * |
||
200 | * @author Andreas Gohr <[email protected]> |
||
201 | * @author Christopher Smith <[email protected]> |
||
202 | * @author Tom N Harris <[email protected]> |
||
203 | */ |
||
204 | protected function getPageWords($text) { |
||
205 | |||
206 | $tokens = $this->tokenizer($text); |
||
207 | $tokens = array_count_values($tokens); // count the frequency of each token |
||
208 | |||
209 | $words = array(); |
||
210 | foreach ($tokens as $w=>$c) { |
||
211 | $l = wordlen($w); |
||
212 | if (isset($words[$l])){ |
||
213 | $words[$l][$w] = $c + (isset($words[$l][$w]) ? $words[$l][$w] : 0); |
||
214 | }else{ |
||
215 | $words[$l] = array($w => $c); |
||
216 | } |
||
217 | } |
||
218 | |||
219 | // arrive here with $words = array(wordlen => array(word => frequency)) |
||
220 | $word_idx_modified = false; |
||
221 | $index = array(); //resulting index |
||
222 | foreach (array_keys($words) as $wlen) { |
||
223 | $word_idx = $this->getIndex('w', $wlen); |
||
224 | foreach ($words[$wlen] as $word => $freq) { |
||
225 | $word = (string)$word; |
||
226 | $wid = array_search($word, $word_idx, true); |
||
227 | if ($wid === false) { |
||
228 | $wid = count($word_idx); |
||
229 | $word_idx[] = $word; |
||
230 | $word_idx_modified = true; |
||
231 | } |
||
232 | if (!isset($index[$wlen])) |
||
233 | $index[$wlen] = array(); |
||
234 | $index[$wlen][$wid] = $freq; |
||
235 | } |
||
236 | // save back the word index |
||
237 | if ($word_idx_modified && !$this->saveIndex('w', $wlen, $word_idx)) |
||
238 | return false; |
||
239 | } |
||
240 | |||
241 | return $index; |
||
242 | } |
||
243 | |||
244 | /** |
||
245 | * Add/update keys to/of the metadata index. |
||
246 | * |
||
247 | * Adding new keys does not remove other keys for the page. |
||
248 | * An empty value will erase the key. |
||
249 | * The $key parameter can be an array to add multiple keys. $value will |
||
250 | * not be used if $key is an array. |
||
251 | * |
||
252 | * @param string $page a page name |
||
253 | * @param mixed $key a key string or array of key=>value pairs |
||
254 | * @param mixed $value the value or list of values |
||
255 | * @return boolean|string the function completed successfully |
||
256 | * |
||
257 | * @author Tom N Harris <[email protected]> |
||
258 | * @author Michael Hamann <[email protected]> |
||
259 | */ |
||
260 | public function addMetaKeys($page, $key, $value=null) { |
||
354 | |||
355 | /** |
||
356 | * Rename a page in the search index without changing the indexed content. This function doesn't check if the |
||
357 | * old or new name exists in the filesystem. It returns an error if the old page isn't in the page list of the |
||
358 | * indexer and it deletes all previously indexed content of the new page. |
||
359 | * |
||
360 | * @param string $oldpage The old page name |
||
361 | * @param string $newpage The new page name |
||
362 | * @return string|bool If the page was successfully renamed, can be a message in the case of an error |
||
363 | */ |
||
364 | public function renamePage($oldpage, $newpage) { |
||
365 | if (!$this->lock()) return 'locked'; |
||
366 | |||
367 | $pages = $this->getPages(); |
||
368 | |||
369 | $id = array_search($oldpage, $pages, true); |
||
370 | if ($id === false) { |
||
371 | $this->unlock(); |
||
372 | return 'page is not in index'; |
||
373 | } |
||
374 | |||
375 | $new_id = array_search($newpage, $pages, true); |
||
376 | if ($new_id !== false) { |
||
377 | // make sure the page is not in the index anymore |
||
378 | if ($this->deletePageNoLock($newpage) !== true) { |
||
379 | return false; |
||
380 | } |
||
381 | |||
382 | $pages[$new_id] = 'deleted:'.time().rand(0, 9999); |
||
383 | } |
||
384 | |||
385 | $pages[$id] = $newpage; |
||
386 | |||
387 | // update index |
||
388 | if (!$this->saveIndex('page', '', $pages)) { |
||
389 | $this->unlock(); |
||
390 | return false; |
||
391 | } |
||
392 | |||
393 | // reset the pid cache |
||
394 | $this->pidCache = array(); |
||
395 | |||
396 | $this->unlock(); |
||
397 | return true; |
||
398 | } |
||
399 | |||
400 | /** |
||
401 | * Renames a meta value in the index. This doesn't change the meta value in the pages, it assumes that all pages |
||
402 | * will be updated. |
||
403 | * |
||
404 | * @param string $key The metadata key of which a value shall be changed |
||
405 | * @param string $oldvalue The old value that shall be renamed |
||
406 | * @param string $newvalue The new value to which the old value shall be renamed, can exist (then values will be merged) |
||
407 | * @return bool|string If renaming the value has been successful, false or error message on error. |
||
408 | */ |
||
409 | public function renameMetaValue($key, $oldvalue, $newvalue) { |
||
410 | if (!$this->lock()) return 'locked'; |
||
411 | |||
412 | // change the relation references index |
||
413 | $metavalues = $this->getIndex($key, '_w'); |
||
414 | $oldid = array_search($oldvalue, $metavalues, true); |
||
415 | if ($oldid !== false) { |
||
416 | $newid = array_search($newvalue, $metavalues, true); |
||
417 | if ($newid !== false) { |
||
418 | // free memory |
||
419 | unset ($metavalues); |
||
420 | |||
421 | // okay, now we have two entries for the same value. we need to merge them. |
||
422 | $indexline = $this->getIndexKey($key.'_i', '', $oldid); |
||
423 | if ($indexline != '') { |
||
424 | $newindexline = $this->getIndexKey($key.'_i', '', $newid); |
||
425 | $pagekeys = $this->getIndex($key.'_p', ''); |
||
426 | $parts = explode(':', $indexline); |
||
427 | foreach ($parts as $part) { |
||
428 | list($id, $count) = explode('*', $part); |
||
429 | $newindexline = $this->updateTuple($newindexline, $id, $count); |
||
430 | |||
431 | $keyline = explode(':', $pagekeys[$id]); |
||
432 | // remove old meta value |
||
433 | $keyline = array_diff($keyline, array($oldid)); |
||
434 | // add new meta value when not already present |
||
435 | if (!in_array($newid, $keyline)) { |
||
436 | array_push($keyline, $newid); |
||
437 | } |
||
438 | $pagekeys[$id] = implode(':', $keyline); |
||
439 | } |
||
440 | $this->saveIndex($key.'_p', '', $pagekeys); |
||
441 | unset($pagekeys); |
||
442 | $this->saveIndexKey($key.'_i', '', $oldid, ''); |
||
443 | $this->saveIndexKey($key.'_i', '', $newid, $newindexline); |
||
444 | } |
||
445 | } else { |
||
446 | $metavalues[$oldid] = $newvalue; |
||
447 | if (!$this->saveIndex($key.'_w', '', $metavalues)) { |
||
448 | $this->unlock(); |
||
449 | return false; |
||
450 | } |
||
451 | } |
||
452 | } |
||
453 | |||
454 | $this->unlock(); |
||
455 | return true; |
||
456 | } |
||
457 | |||
458 | /** |
||
459 | * Remove a page from the index |
||
460 | * |
||
461 | * Erases entries in all known indexes. |
||
462 | * |
||
463 | * @param string $page a page name |
||
464 | * @return string|boolean the function completed successfully |
||
465 | * |
||
466 | * @author Tom N Harris <[email protected]> |
||
467 | */ |
||
468 | public function deletePage($page) { |
||
469 | if (!$this->lock()) |
||
470 | return "locked"; |
||
471 | |||
472 | $result = $this->deletePageNoLock($page); |
||
473 | |||
474 | $this->unlock(); |
||
475 | |||
476 | return $result; |
||
477 | } |
||
478 | |||
479 | /** |
||
480 | * Remove a page from the index without locking the index, only use this function if the index is already locked |
||
481 | * |
||
482 | * Erases entries in all known indexes. |
||
483 | * |
||
484 | * @param string $page a page name |
||
485 | * @return boolean the function completed successfully |
||
486 | * |
||
487 | * @author Tom N Harris <[email protected]> |
||
488 | */ |
||
489 | protected function deletePageNoLock($page) { |
||
490 | // load known documents |
||
491 | $pid = $this->getPIDNoLock($page); |
||
492 | if ($pid === false) { |
||
493 | return false; |
||
494 | } |
||
495 | |||
496 | // Remove obsolete index entries |
||
497 | $pageword_idx = $this->getIndexKey('pageword', '', $pid); |
||
498 | if ($pageword_idx !== '') { |
||
499 | $delwords = explode(':',$pageword_idx); |
||
500 | $upwords = array(); |
||
501 | foreach ($delwords as $word) { |
||
502 | if ($word != '') { |
||
503 | list($wlen,$wid) = explode('*', $word); |
||
504 | $wid = (int)$wid; |
||
505 | $upwords[$wlen][] = $wid; |
||
506 | } |
||
507 | } |
||
508 | foreach ($upwords as $wlen => $widx) { |
||
509 | $index = $this->getIndex('i', $wlen); |
||
510 | foreach ($widx as $wid) { |
||
511 | $index[$wid] = $this->updateTuple($index[$wid], $pid, 0); |
||
512 | } |
||
513 | $this->saveIndex('i', $wlen, $index); |
||
514 | } |
||
515 | } |
||
516 | // Save the reverse index |
||
517 | if (!$this->saveIndexKey('pageword', '', $pid, "")) { |
||
518 | return false; |
||
519 | } |
||
520 | |||
521 | $this->saveIndexKey('title', '', $pid, ""); |
||
522 | $keyidx = $this->getIndex('metadata', ''); |
||
523 | foreach ($keyidx as $metaname) { |
||
524 | $val_idx = explode(':', $this->getIndexKey($metaname.'_p', '', $pid)); |
||
525 | $meta_idx = $this->getIndex($metaname.'_i', ''); |
||
526 | foreach ($val_idx as $id) { |
||
527 | if ($id === '') continue; |
||
528 | $meta_idx[$id] = $this->updateTuple($meta_idx[$id], $pid, 0); |
||
529 | } |
||
530 | $this->saveIndex($metaname.'_i', '', $meta_idx); |
||
531 | $this->saveIndexKey($metaname.'_p', '', $pid, ''); |
||
532 | } |
||
533 | |||
534 | return true; |
||
535 | } |
||
536 | |||
537 | /** |
||
538 | * Clear the whole index |
||
539 | * |
||
540 | * @return bool If the index has been cleared successfully |
||
541 | */ |
||
542 | public function clear() { |
||
543 | global $conf; |
||
544 | |||
545 | if (!$this->lock()) return false; |
||
546 | |||
547 | @unlink($conf['indexdir'].'/page.idx'); |
||
1 ignored issue
–
show
|
|||
548 | @unlink($conf['indexdir'].'/title.idx'); |
||
1 ignored issue
–
show
|
|||
549 | @unlink($conf['indexdir'].'/pageword.idx'); |
||
1 ignored issue
–
show
|
|||
550 | @unlink($conf['indexdir'].'/metadata.idx'); |
||
1 ignored issue
–
show
|
|||
551 | $dir = @opendir($conf['indexdir']); |
||
552 | if($dir!==false){ |
||
553 | while(($f = readdir($dir)) !== false){ |
||
554 | if(substr($f,-4)=='.idx' && |
||
555 | (substr($f,0,1)=='i' || substr($f,0,1)=='w' |
||
556 | || substr($f,-6)=='_w.idx' || substr($f,-6)=='_i.idx' || substr($f,-6)=='_p.idx')) |
||
557 | @unlink($conf['indexdir']."/$f"); |
||
1 ignored issue
–
show
|
|||
558 | } |
||
559 | } |
||
560 | @unlink($conf['indexdir'].'/lengths.idx'); |
||
1 ignored issue
–
show
|
|||
561 | |||
562 | // clear the pid cache |
||
563 | $this->pidCache = array(); |
||
564 | |||
565 | $this->unlock(); |
||
566 | return true; |
||
567 | } |
||
568 | |||
569 | /** |
||
570 | * Split the text into words for fulltext search |
||
571 | * |
||
572 | * TODO: does this also need &$stopwords ? |
||
573 | * |
||
574 | * @triggers INDEXER_TEXT_PREPARE |
||
575 | * This event allows plugins to modify the text before it gets tokenized. |
||
576 | * Plugins intercepting this event should also intercept INDEX_VERSION_GET |
||
577 | * |
||
578 | * @param string $text plain text |
||
579 | * @param boolean $wc are wildcards allowed? |
||
580 | * @return array list of words in the text |
||
581 | * |
||
582 | * @author Tom N Harris <[email protected]> |
||
583 | * @author Andreas Gohr <[email protected]> |
||
584 | */ |
||
585 | public function tokenizer($text, $wc=false) { |
||
586 | $wc = ($wc) ? '' : '\*'; |
||
587 | $stopwords =& idx_get_stopwords(); |
||
588 | |||
589 | // prepare the text to be tokenized |
||
590 | $evt = new Doku_Event('INDEXER_TEXT_PREPARE', $text); |
||
591 | if ($evt->advise_before(true)) { |
||
592 | if (preg_match('/[^0-9A-Za-z ]/u', $text)) { |
||
593 | // handle asian chars as single words (may fail on older PHP version) |
||
594 | $asia = @preg_replace('/('.IDX_ASIAN.')/u', ' \1 ', $text); |
||
595 | if (!is_null($asia)) $text = $asia; // recover from regexp falure |
||
596 | } |
||
597 | } |
||
598 | $evt->advise_after(); |
||
599 | unset($evt); |
||
600 | |||
601 | $text = strtr($text, |
||
602 | array( |
||
603 | "\r" => ' ', |
||
604 | "\n" => ' ', |
||
605 | "\t" => ' ', |
||
606 | "\xC2\xAD" => '', //soft-hyphen |
||
607 | ) |
||
608 | ); |
||
609 | if (preg_match('/[^0-9A-Za-z ]/u', $text)) |
||
610 | $text = utf8_stripspecials($text, ' ', '\._\-:'.$wc); |
||
611 | |||
612 | $wordlist = explode(' ', $text); |
||
613 | foreach ($wordlist as $i => $word) { |
||
614 | $wordlist[$i] = (preg_match('/[^0-9A-Za-z]/u', $word)) ? |
||
615 | utf8_strtolower($word) : strtolower($word); |
||
616 | } |
||
617 | |||
618 | foreach ($wordlist as $i => $word) { |
||
619 | if ((!is_numeric($word) && strlen($word) < IDX_MINWORDLENGTH) |
||
620 | || array_search($word, $stopwords, true) !== false) |
||
621 | unset($wordlist[$i]); |
||
622 | } |
||
623 | return array_values($wordlist); |
||
624 | } |
||
625 | |||
626 | /** |
||
627 | * Get the numeric PID of a page |
||
628 | * |
||
629 | * @param string $page The page to get the PID for |
||
630 | * @return bool|int The page id on success, false on error |
||
631 | */ |
||
632 | public function getPID($page) { |
||
649 | |||
650 | /** |
||
651 | * Get the numeric PID of a page without locking the index. |
||
652 | * Only use this function when the index is already locked. |
||
653 | * |
||
654 | * @param string $page The page to get the PID for |
||
655 | * @return bool|int The page id on success, false on error |
||
656 | */ |
||
657 | protected function getPIDNoLock($page) { |
||
658 | // avoid expensive addIndexKey operation for the most recently requested pages by using a cache |
||
659 | if (isset($this->pidCache[$page])) return $this->pidCache[$page]; |
||
660 | $pid = $this->addIndexKey('page', '', $page); |
||
661 | // limit cache to 10 entries by discarding the oldest element as in DokuWiki usually only the most recently |
||
662 | // added item will be requested again |
||
663 | if (count($this->pidCache) > 10) array_shift($this->pidCache); |
||
664 | $this->pidCache[$page] = $pid; |
||
665 | return $pid; |
||
666 | } |
||
667 | |||
668 | /** |
||
669 | * Get the page id of a numeric PID |
||
670 | * |
||
671 | * @param int $pid The PID to get the page id for |
||
672 | * @return string The page id |
||
673 | */ |
||
674 | public function getPageFromPID($pid) { |
||
677 | |||
678 | /** |
||
679 | * Find pages in the fulltext index containing the words, |
||
680 | * |
||
681 | * The search words must be pre-tokenized, meaning only letters and |
||
682 | * numbers with an optional wildcard |
||
683 | * |
||
684 | * The returned array will have the original tokens as key. The values |
||
685 | * in the returned list is an array with the page names as keys and the |
||
686 | * number of times that token appears on the page as value. |
||
687 | * |
||
688 | * @param array $tokens list of words to search for |
||
689 | * @return array list of page names with usage counts |
||
690 | * |
||
691 | * @author Tom N Harris <[email protected]> |
||
692 | * @author Andreas Gohr <[email protected]> |
||
693 | */ |
||
694 | public function lookup(&$tokens) { |
||
695 | $result = array(); |
||
696 | $wids = $this->getIndexWords($tokens, $result); |
||
697 | if (empty($wids)) return array(); |
||
698 | // load known words and documents |
||
699 | $page_idx = $this->getIndex('page', ''); |
||
700 | $docs = array(); |
||
701 | foreach (array_keys($wids) as $wlen) { |
||
702 | $wids[$wlen] = array_unique($wids[$wlen]); |
||
703 | $index = $this->getIndex('i', $wlen); |
||
704 | foreach($wids[$wlen] as $ixid) { |
||
705 | if ($ixid < count($index)) |
||
706 | $docs["$wlen*$ixid"] = $this->parseTuples($page_idx, $index[$ixid]); |
||
707 | } |
||
708 | } |
||
709 | // merge found pages into final result array |
||
710 | $final = array(); |
||
711 | foreach ($result as $word => $res) { |
||
712 | $final[$word] = array(); |
||
713 | foreach ($res as $wid) { |
||
714 | // handle the case when ($ixid < count($index)) has been false |
||
715 | // and thus $docs[$wid] hasn't been set. |
||
716 | if (!isset($docs[$wid])) continue; |
||
717 | $hits = &$docs[$wid]; |
||
718 | foreach ($hits as $hitkey => $hitcnt) { |
||
719 | // make sure the document still exists |
||
720 | if (!page_exists($hitkey, '', false)) continue; |
||
721 | if (!isset($final[$word][$hitkey])) |
||
722 | $final[$word][$hitkey] = $hitcnt; |
||
723 | else |
||
724 | $final[$word][$hitkey] += $hitcnt; |
||
725 | } |
||
726 | } |
||
727 | } |
||
728 | return $final; |
||
729 | } |
||
730 | |||
731 | /** |
||
732 | * Find pages containing a metadata key. |
||
733 | * |
||
734 | * The metadata values are compared as case-sensitive strings. Pass a |
||
735 | * callback function that returns true or false to use a different |
||
736 | * comparison function. The function will be called with the $value being |
||
737 | * searched for as the first argument, and the word in the index as the |
||
738 | * second argument. The function preg_match can be used directly if the |
||
739 | * values are regexes. |
||
740 | * |
||
741 | * @param string $key name of the metadata key to look for |
||
742 | * @param string $value search term to look for, must be a string or array of strings |
||
743 | * @param callback $func comparison function |
||
744 | * @return array lists with page names, keys are query values if $value is array |
||
745 | * |
||
746 | * @author Tom N Harris <[email protected]> |
||
747 | * @author Michael Hamann <[email protected]> |
||
748 | */ |
||
749 | public function lookupKey($key, &$value, $func=null) { |
||
750 | if (!is_array($value)) |
||
751 | $value_array = array($value); |
||
752 | else |
||
753 | $value_array =& $value; |
||
754 | |||
755 | // the matching ids for the provided value(s) |
||
756 | $value_ids = array(); |
||
757 | |||
758 | $metaname = idx_cleanName($key); |
||
759 | |||
760 | // get all words in order to search the matching ids |
||
761 | if ($key == 'title') { |
||
762 | $words = $this->getIndex('title', ''); |
||
763 | } else { |
||
764 | $words = $this->getIndex($metaname.'_w', ''); |
||
765 | } |
||
766 | |||
767 | if (!is_null($func)) { |
||
768 | foreach ($value_array as $val) { |
||
769 | foreach ($words as $i => $word) { |
||
770 | if (call_user_func_array($func, array($val, $word))) |
||
771 | $value_ids[$i][] = $val; |
||
772 | } |
||
773 | } |
||
774 | } else { |
||
775 | foreach ($value_array as $val) { |
||
776 | $xval = $val; |
||
777 | $caret = '^'; |
||
778 | $dollar = '$'; |
||
779 | // check for wildcards |
||
780 | if (substr($xval, 0, 1) == '*') { |
||
781 | $xval = substr($xval, 1); |
||
782 | $caret = ''; |
||
783 | } |
||
784 | if (substr($xval, -1, 1) == '*') { |
||
785 | $xval = substr($xval, 0, -1); |
||
786 | $dollar = ''; |
||
787 | } |
||
788 | if (!$caret || !$dollar) { |
||
789 | $re = $caret.preg_quote($xval, '/').$dollar; |
||
790 | foreach(array_keys(preg_grep('/'.$re.'/', $words)) as $i) |
||
791 | $value_ids[$i][] = $val; |
||
792 | } else { |
||
793 | if (($i = array_search($val, $words, true)) !== false) |
||
794 | $value_ids[$i][] = $val; |
||
795 | } |
||
796 | } |
||
797 | } |
||
798 | |||
799 | unset($words); // free the used memory |
||
800 | |||
801 | // initialize the result so it won't be null |
||
802 | $result = array(); |
||
803 | foreach ($value_array as $val) { |
||
804 | $result[$val] = array(); |
||
805 | } |
||
806 | |||
807 | $page_idx = $this->getIndex('page', ''); |
||
808 | |||
809 | // Special handling for titles |
||
810 | if ($key == 'title') { |
||
811 | foreach ($value_ids as $pid => $val_list) { |
||
812 | $page = $page_idx[$pid]; |
||
813 | foreach ($val_list as $val) { |
||
814 | $result[$val][] = $page; |
||
815 | } |
||
816 | } |
||
817 | } else { |
||
818 | // load all lines and pages so the used lines can be taken and matched with the pages |
||
819 | $lines = $this->getIndex($metaname.'_i', ''); |
||
820 | |||
821 | foreach ($value_ids as $value_id => $val_list) { |
||
822 | // parse the tuples of the form page_id*1:page2_id*1 and so on, return value |
||
823 | // is an array with page_id => 1, page2_id => 1 etc. so take the keys only |
||
824 | $pages = array_keys($this->parseTuples($page_idx, $lines[$value_id])); |
||
825 | foreach ($val_list as $val) { |
||
826 | $result[$val] = array_merge($result[$val], $pages); |
||
827 | } |
||
828 | } |
||
829 | } |
||
830 | if (!is_array($value)) $result = $result[$value]; |
||
831 | return $result; |
||
832 | } |
||
833 | |||
834 | /** |
||
835 | * Find the index ID of each search term. |
||
836 | * |
||
837 | * The query terms should only contain valid characters, with a '*' at |
||
838 | * either the beginning or end of the word (or both). |
||
839 | * The $result parameter can be used to merge the index locations with |
||
840 | * the appropriate query term. |
||
841 | * |
||
842 | * @param array $words The query terms. |
||
843 | * @param array $result Set to word => array("length*id" ...) |
||
844 | * @return array Set to length => array(id ...) |
||
845 | * |
||
846 | * @author Tom N Harris <[email protected]> |
||
847 | */ |
||
848 | protected function getIndexWords(&$words, &$result) { |
||
849 | $tokens = array(); |
||
850 | $tokenlength = array(); |
||
851 | $tokenwild = array(); |
||
852 | foreach ($words as $word) { |
||
853 | $result[$word] = array(); |
||
854 | $caret = '^'; |
||
855 | $dollar = '$'; |
||
856 | $xword = $word; |
||
857 | $wlen = wordlen($word); |
||
858 | |||
859 | // check for wildcards |
||
860 | if (substr($xword, 0, 1) == '*') { |
||
861 | $xword = substr($xword, 1); |
||
862 | $caret = ''; |
||
863 | $wlen -= 1; |
||
864 | } |
||
865 | if (substr($xword, -1, 1) == '*') { |
||
866 | $xword = substr($xword, 0, -1); |
||
867 | $dollar = ''; |
||
868 | $wlen -= 1; |
||
869 | } |
||
870 | if ($wlen < IDX_MINWORDLENGTH && $caret && $dollar && !is_numeric($xword)) |
||
871 | continue; |
||
872 | if (!isset($tokens[$xword])) |
||
873 | $tokenlength[$wlen][] = $xword; |
||
874 | if (!$caret || !$dollar) { |
||
875 | $re = $caret.preg_quote($xword, '/').$dollar; |
||
876 | $tokens[$xword][] = array($word, '/'.$re.'/'); |
||
877 | if (!isset($tokenwild[$xword])) |
||
878 | $tokenwild[$xword] = $wlen; |
||
879 | } else { |
||
880 | $tokens[$xword][] = array($word, null); |
||
881 | } |
||
882 | } |
||
883 | asort($tokenwild); |
||
884 | // $tokens = array( base word => array( [ query term , regexp ] ... ) ... ) |
||
885 | // $tokenlength = array( base word length => base word ... ) |
||
886 | // $tokenwild = array( base word => base word length ... ) |
||
887 | $length_filter = empty($tokenwild) ? $tokenlength : min(array_keys($tokenlength)); |
||
888 | $indexes_known = $this->indexLengths($length_filter); |
||
889 | if (!empty($tokenwild)) sort($indexes_known); |
||
890 | // get word IDs |
||
891 | $wids = array(); |
||
892 | foreach ($indexes_known as $ixlen) { |
||
893 | $word_idx = $this->getIndex('w', $ixlen); |
||
894 | // handle exact search |
||
895 | if (isset($tokenlength[$ixlen])) { |
||
896 | foreach ($tokenlength[$ixlen] as $xword) { |
||
897 | $wid = array_search($xword, $word_idx, true); |
||
898 | if ($wid !== false) { |
||
899 | $wids[$ixlen][] = $wid; |
||
900 | foreach ($tokens[$xword] as $w) |
||
901 | $result[$w[0]][] = "$ixlen*$wid"; |
||
902 | } |
||
903 | } |
||
904 | } |
||
905 | // handle wildcard search |
||
906 | foreach ($tokenwild as $xword => $wlen) { |
||
907 | if ($wlen >= $ixlen) break; |
||
908 | foreach ($tokens[$xword] as $w) { |
||
909 | if (is_null($w[1])) continue; |
||
910 | foreach(array_keys(preg_grep($w[1], $word_idx)) as $wid) { |
||
911 | $wids[$ixlen][] = $wid; |
||
912 | $result[$w[0]][] = "$ixlen*$wid"; |
||
913 | } |
||
914 | } |
||
915 | } |
||
916 | } |
||
917 | return $wids; |
||
918 | } |
||
919 | |||
920 | /** |
||
921 | * Return a list of all pages |
||
922 | * Warning: pages may not exist! |
||
923 | * |
||
924 | * @param string $key list only pages containing the metadata key (optional) |
||
925 | * @return array list of page names |
||
926 | * |
||
927 | * @author Tom N Harris <[email protected]> |
||
928 | */ |
||
929 | public function getPages($key=null) { |
||
930 | $page_idx = $this->getIndex('page', ''); |
||
931 | if (is_null($key)) return $page_idx; |
||
932 | |||
933 | $metaname = idx_cleanName($key); |
||
934 | |||
935 | // Special handling for titles |
||
936 | if ($key == 'title') { |
||
937 | $title_idx = $this->getIndex('title', ''); |
||
938 | array_splice($page_idx, count($title_idx)); |
||
939 | foreach ($title_idx as $i => $title) |
||
940 | if ($title === "") unset($page_idx[$i]); |
||
941 | return array_values($page_idx); |
||
942 | } |
||
943 | |||
944 | $pages = array(); |
||
945 | $lines = $this->getIndex($metaname.'_i', ''); |
||
946 | foreach ($lines as $line) { |
||
947 | $pages = array_merge($pages, $this->parseTuples($page_idx, $line)); |
||
948 | } |
||
949 | return array_keys($pages); |
||
950 | } |
||
951 | |||
952 | /** |
||
953 | * Return a list of words sorted by number of times used |
||
954 | * |
||
955 | * @param int $min bottom frequency threshold |
||
956 | * @param int $max upper frequency limit. No limit if $max<$min |
||
957 | * @param int $minlen minimum length of words to count |
||
958 | * @param string $key metadata key to list. Uses the fulltext index if not given |
||
959 | * @return array list of words as the keys and frequency as values |
||
960 | * |
||
961 | * @author Tom N Harris <[email protected]> |
||
962 | */ |
||
963 | public function histogram($min=1, $max=0, $minlen=3, $key=null) { |
||
964 | if ($min < 1) |
||
965 | $min = 1; |
||
966 | if ($max < $min) |
||
967 | $max = 0; |
||
968 | |||
969 | $result = array(); |
||
970 | |||
971 | if ($key == 'title') { |
||
972 | $index = $this->getIndex('title', ''); |
||
973 | $index = array_count_values($index); |
||
974 | foreach ($index as $val => $cnt) { |
||
975 | if ($cnt >= $min && (!$max || $cnt <= $max) && strlen($val) >= $minlen) |
||
976 | $result[$val] = $cnt; |
||
977 | } |
||
978 | } |
||
979 | elseif (!is_null($key)) { |
||
980 | $metaname = idx_cleanName($key); |
||
981 | $index = $this->getIndex($metaname.'_i', ''); |
||
982 | $val_idx = array(); |
||
983 | foreach ($index as $wid => $line) { |
||
984 | $freq = $this->countTuples($line); |
||
985 | if ($freq >= $min && (!$max || $freq <= $max)) |
||
986 | $val_idx[$wid] = $freq; |
||
987 | } |
||
988 | if (!empty($val_idx)) { |
||
989 | $words = $this->getIndex($metaname.'_w', ''); |
||
990 | foreach ($val_idx as $wid => $freq) { |
||
991 | if (strlen($words[$wid]) >= $minlen) |
||
992 | $result[$words[$wid]] = $freq; |
||
993 | } |
||
994 | } |
||
995 | } |
||
996 | else { |
||
997 | $lengths = idx_listIndexLengths(); |
||
998 | foreach ($lengths as $length) { |
||
999 | if ($length < $minlen) continue; |
||
1000 | $index = $this->getIndex('i', $length); |
||
1001 | $words = null; |
||
1002 | foreach ($index as $wid => $line) { |
||
1003 | $freq = $this->countTuples($line); |
||
1004 | if ($freq >= $min && (!$max || $freq <= $max)) { |
||
1005 | if ($words === null) |
||
1006 | $words = $this->getIndex('w', $length); |
||
1007 | $result[$words[$wid]] = $freq; |
||
1008 | } |
||
1009 | } |
||
1010 | } |
||
1011 | } |
||
1012 | |||
1013 | arsort($result); |
||
1014 | return $result; |
||
1015 | } |
||
1016 | |||
1017 | /** |
||
1018 | * Lock the indexer. |
||
1019 | * |
||
1020 | * @author Tom N Harris <[email protected]> |
||
1021 | * |
||
1022 | * @return bool|string |
||
1023 | */ |
||
1024 | protected function lock() { |
||
1025 | global $conf; |
||
1026 | $status = true; |
||
1027 | $run = 0; |
||
1028 | $lock = $conf['lockdir'].'/_indexer.lock'; |
||
1029 | while (!@mkdir($lock, $conf['dmode'])) { |
||
1030 | usleep(50); |
||
1031 | if(is_dir($lock) && time()-@filemtime($lock) > 60*5){ |
||
1032 | // looks like a stale lock - remove it |
||
1033 | if (!@rmdir($lock)) { |
||
1034 | $status = "removing the stale lock failed"; |
||
1035 | return false; |
||
1036 | } else { |
||
1037 | $status = "stale lock removed"; |
||
1038 | } |
||
1039 | }elseif($run++ == 1000){ |
||
1040 | // we waited 5 seconds for that lock |
||
1041 | return false; |
||
1042 | } |
||
1043 | } |
||
1044 | if (!empty($conf['dperm'])) { |
||
1045 | chmod($lock, $conf['dperm']); |
||
1046 | } |
||
1047 | return $status; |
||
1048 | } |
||
1049 | |||
1050 | /** |
||
1051 | * Release the indexer lock. |
||
1052 | * |
||
1053 | * @author Tom N Harris <[email protected]> |
||
1054 | * |
||
1055 | * @return bool |
||
1056 | */ |
||
1057 | protected function unlock() { |
||
1058 | global $conf; |
||
1059 | @rmdir($conf['lockdir'].'/_indexer.lock'); |
||
1 ignored issue
–
show
|
|||
1060 | return true; |
||
1061 | } |
||
1062 | |||
1063 | /** |
||
1064 | * Retrieve the entire index. |
||
1065 | * |
||
1066 | * The $suffix argument is for an index that is split into |
||
1067 | * multiple parts. Different index files should use different |
||
1068 | * base names. |
||
1069 | * |
||
1070 | * @param string $idx name of the index |
||
1071 | * @param string $suffix subpart identifier |
||
1072 | * @return array list of lines without CR or LF |
||
1073 | * |
||
1074 | * @author Tom N Harris <[email protected]> |
||
1075 | */ |
||
1076 | protected function getIndex($idx, $suffix) { |
||
1077 | global $conf; |
||
1078 | $fn = $conf['indexdir'].'/'.$idx.$suffix.'.idx'; |
||
1079 | if (!file_exists($fn)) return array(); |
||
1080 | return file($fn, FILE_IGNORE_NEW_LINES); |
||
1081 | } |
||
1082 | |||
1083 | /** |
||
1084 | * Replace the contents of the index with an array. |
||
1085 | * |
||
1086 | * @param string $idx name of the index |
||
1087 | * @param string $suffix subpart identifier |
||
1088 | * @param array $lines list of lines without LF |
||
1089 | * @return bool If saving succeeded |
||
1090 | * |
||
1091 | * @author Tom N Harris <[email protected]> |
||
1092 | */ |
||
1093 | protected function saveIndex($idx, $suffix, &$lines) { |
||
1107 | |||
1108 | /** |
||
1109 | * Retrieve a line from the index. |
||
1110 | * |
||
1111 | * @param string $idx name of the index |
||
1112 | * @param string $suffix subpart identifier |
||
1113 | * @param int $id the line number |
||
1114 | * @return string a line with trailing whitespace removed |
||
1115 | * |
||
1116 | * @author Tom N Harris <[email protected]> |
||
1117 | */ |
||
1118 | protected function getIndexKey($idx, $suffix, $id) { |
||
1119 | global $conf; |
||
1120 | $fn = $conf['indexdir'].'/'.$idx.$suffix.'.idx'; |
||
1121 | if (!file_exists($fn)) return ''; |
||
1122 | $fh = @fopen($fn, 'r'); |
||
1123 | if (!$fh) return ''; |
||
1124 | $ln = -1; |
||
1125 | while (($line = fgets($fh)) !== false) { |
||
1126 | if (++$ln == $id) break; |
||
1127 | } |
||
1128 | fclose($fh); |
||
1129 | return rtrim((string)$line); |
||
1130 | } |
||
1131 | |||
1132 | /** |
||
1133 | * Write a line into the index. |
||
1134 | * |
||
1135 | * @param string $idx name of the index |
||
1136 | * @param string $suffix subpart identifier |
||
1137 | * @param int $id the line number |
||
1138 | * @param string $line line to write |
||
1139 | * @return bool If saving succeeded |
||
1140 | * |
||
1141 | * @author Tom N Harris <[email protected]> |
||
1142 | */ |
||
1143 | protected function saveIndexKey($idx, $suffix, $id, $line) { |
||
1144 | global $conf; |
||
1145 | if (substr($line, -1) != "\n") |
||
1146 | $line .= "\n"; |
||
1147 | $fn = $conf['indexdir'].'/'.$idx.$suffix; |
||
1148 | $fh = @fopen($fn.'.tmp', 'w'); |
||
1149 | if (!$fh) return false; |
||
1150 | $ih = @fopen($fn.'.idx', 'r'); |
||
1151 | if ($ih) { |
||
1152 | $ln = -1; |
||
1153 | while (($curline = fgets($ih)) !== false) { |
||
1154 | fwrite($fh, (++$ln == $id) ? $line : $curline); |
||
1155 | } |
||
1156 | if ($id > $ln) { |
||
1157 | while ($id > ++$ln) |
||
1158 | fwrite($fh, "\n"); |
||
1159 | fwrite($fh, $line); |
||
1160 | } |
||
1161 | fclose($ih); |
||
1162 | } else { |
||
1163 | $ln = -1; |
||
1164 | while ($id > ++$ln) |
||
1165 | fwrite($fh, "\n"); |
||
1166 | fwrite($fh, $line); |
||
1167 | } |
||
1168 | fclose($fh); |
||
1169 | if (isset($conf['fperm'])) |
||
1170 | chmod($fn.'.tmp', $conf['fperm']); |
||
1171 | io_rename($fn.'.tmp', $fn.'.idx'); |
||
1172 | return true; |
||
1173 | } |
||
1174 | |||
1175 | /** |
||
1176 | * Retrieve or insert a value in the index. |
||
1177 | * |
||
1178 | * @param string $idx name of the index |
||
1179 | * @param string $suffix subpart identifier |
||
1180 | * @param string $value line to find in the index |
||
1181 | * @return int|bool line number of the value in the index or false if writing the index failed |
||
1182 | * |
||
1183 | * @author Tom N Harris <[email protected]> |
||
1184 | */ |
||
1185 | protected function addIndexKey($idx, $suffix, $value) { |
||
1186 | $index = $this->getIndex($idx, $suffix); |
||
1187 | $id = array_search($value, $index, true); |
||
1188 | if ($id === false) { |
||
1189 | $id = count($index); |
||
1190 | $index[$id] = $value; |
||
1191 | if (!$this->saveIndex($idx, $suffix, $index)) { |
||
1192 | trigger_error("Failed to write $idx index", E_USER_ERROR); |
||
1193 | return false; |
||
1194 | } |
||
1195 | } |
||
1196 | return $id; |
||
1197 | } |
||
1198 | |||
1199 | /** |
||
1200 | * Get the list of lengths indexed in the wiki. |
||
1201 | * |
||
1202 | * Read the index directory or a cache file and returns |
||
1203 | * a sorted array of lengths of the words used in the wiki. |
||
1204 | * |
||
1205 | * @author YoBoY <[email protected]> |
||
1206 | * |
||
1207 | * @return array |
||
1208 | */ |
||
1209 | protected function listIndexLengths() { |
||
1212 | |||
1213 | /** |
||
1214 | * Get the word lengths that have been indexed. |
||
1215 | * |
||
1216 | * Reads the index directory and returns an array of lengths |
||
1217 | * that there are indices for. |
||
1218 | * |
||
1219 | * @author YoBoY <[email protected]> |
||
1220 | * |
||
1221 | * @param array|int $filter |
||
1222 | * @return array |
||
1223 | */ |
||
1224 | protected function indexLengths($filter) { |
||
1225 | global $conf; |
||
1226 | $idx = array(); |
||
1227 | if (is_array($filter)) { |
||
1228 | // testing if index files exist only |
||
1229 | $path = $conf['indexdir']."/i"; |
||
1230 | foreach ($filter as $key => $value) { |
||
1231 | if (file_exists($path.$key.'.idx')) |
||
1232 | $idx[] = $key; |
||
1233 | } |
||
1234 | } else { |
||
1235 | $lengths = idx_listIndexLengths(); |
||
1236 | foreach ($lengths as $key => $length) { |
||
1237 | // keep all the values equal or superior |
||
1238 | if ((int)$length >= (int)$filter) |
||
1239 | $idx[] = $length; |
||
1240 | } |
||
1241 | } |
||
1242 | return $idx; |
||
1243 | } |
||
1244 | |||
1245 | /** |
||
1246 | * Insert or replace a tuple in a line. |
||
1247 | * |
||
1248 | * @author Tom N Harris <[email protected]> |
||
1249 | * |
||
1250 | * @param string $line |
||
1251 | * @param string|int $id |
||
1252 | * @param int $count |
||
1253 | * @return string |
||
1254 | */ |
||
1255 | protected function updateTuple($line, $id, $count) { |
||
1269 | |||
1270 | /** |
||
1271 | * Split a line into an array of tuples. |
||
1272 | * |
||
1273 | * @author Tom N Harris <[email protected]> |
||
1274 | * @author Andreas Gohr <[email protected]> |
||
1275 | * |
||
1276 | * @param array $keys |
||
1277 | * @param string $line |
||
1278 | * @return array |
||
1279 | */ |
||
1280 | protected function parseTuples(&$keys, $line) { |
||
1281 | $result = array(); |
||
1282 | if ($line == '') return $result; |
||
1283 | $parts = explode(':', $line); |
||
1284 | foreach ($parts as $tuple) { |
||
1285 | if ($tuple === '') continue; |
||
1286 | list($key, $cnt) = explode('*', $tuple); |
||
1287 | if (!$cnt) continue; |
||
1288 | $key = $keys[$key]; |
||
1289 | if (!$key) continue; |
||
1290 | $result[$key] = $cnt; |
||
1294 | |||
1295 | /** |
||
1296 | * Sum the counts in a list of tuples. |
||
1297 | * |
||
1298 | * @author Tom N Harris <[email protected]> |
||
1299 | * |
||
1300 | * @param string $line |
||
1301 | * @return int |
||
1302 | */ |
||
1303 | protected function countTuples($line) { |
||
1313 | } |
||
1314 | |||
1608 |
If a method or function can return multiple different values and unless you are sure that you only can receive a single value in this context, we recommend to add an additional type check:
If this a common case that PHP Analyzer should handle natively, please let us know by opening an issue.