4 * Copyright 2004-2010 Reini Urban
6 * This file is part of PhpWiki.
8 * PhpWiki is free software; you can redistribute it and/or modify
9 * it under the terms of the GNU General Public License as published by
10 * the Free Software Foundation; either version 2 of the License, or
11 * (at your option) any later version.
13 * PhpWiki is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 * GNU General Public License for more details.
18 * You should have received a copy of the GNU General Public License along
19 * with PhpWiki; if not, write to the Free Software Foundation, Inc.,
20 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
26 maintained by WikiPage
28 //:deleted (*) (Set if latest content is empty.)
36 %content (?should this be here?)
37 _supplanted : Time version ceased to be the current version
39 mtime (*) : Time of version edit.
42 author : nominal author
43 author_id : authenticated author
52 (types are scalars: strings, ints, bools)
56 * A WikiDB_backend handles the storage and retrieval of data for a WikiDB.
58 * It does not have to be this way, of course, but the standard WikiDB uses
59 * a WikiDB_backend. (Other WikiDB's could be written which use some other
60 * method to access their underlying data store.)
62 * The interface outlined here seems to work well with both RDBM based
63 * and flat DBM/hash based methods of data storage.
65 * Though it contains some default implementation of certain methods,
66 * this is an abstract base class. It is expected that most efficient
67 * backends will override nearly all the methods in this class.
75 * Get page meta-data from database.
77 * @param $pagename string Page name.
79 * Returns a hash containing the page meta-data.
80 * Returns an empty array if there is no meta-data for the requested page.
81 * Keys which might be present in the hash are:
83 * <dt> locked <dd> If the page is locked.
84 * <dt> hits <dd> The page hit count.
85 * <dt> created <dd> Unix time of page creation. (FIXME: Deprecated: I
86 * don't think we need this...)
89 function get_pagedata($pagename) {
90 trigger_error("virtual", E_USER_ERROR);
94 * Update the page meta-data.
98 * Only meta-data whose keys are preset in $newdata is affected.
102 * $backend->update_pagedata($pagename, array('locked' => 1));
104 * will set the value of 'locked' to 1 for the specified page, but it
105 * will not affect the value of 'hits' (or whatever other meta-data
106 * may have been stored for the page.)
108 * To delete a particular piece of meta-data, set its value to false.
110 * $backend->update_pagedata($pagename, array('locked' => false));
113 * @param $pagename string Page name.
114 * @param $newdata hash New meta-data.
116 function update_pagedata($pagename, $newdata) {
117 trigger_error("virtual", E_USER_ERROR);
122 * Get the current version number for a page.
124 * @param $pagename string Page name.
125 * @return int The latest version number for the page. Returns zero if
126 * no versions of a page exist.
128 function get_latest_version($pagename) {
129 trigger_error("virtual", E_USER_ERROR);
133 * Get preceding version number.
135 * @param $pagename string Page name.
136 * @param $version int Find version before this one.
137 * @return int The version number of the version in the database which
138 * immediately preceeds $version.
140 function get_previous_version($pagename, $version) {
141 trigger_error("virtual", E_USER_ERROR);
145 * Get revision meta-data and content.
147 * @param $pagename string Page name.
148 * @param $version integer Which version to get.
149 * @param $want_content boolean
150 * Indicates the caller really wants the page content. If this
151 * flag is not set, the backend is free to skip fetching of the
152 * page content (as that may be expensive). If the backend omits
153 * the content, the backend might still want to set the value of
154 * '%content' to the empty string if it knows there's no content.
156 * @return hash The version data, or false if specified version does not
159 * Some keys which might be present in the $versiondata hash are:
162 * <dd> This is a pseudo-meta-data element (since it's actually
163 * the page data, get it?) containing the page content.
164 * If the content was not fetched, this key may not be present.
166 * For description of other version meta-data see WikiDB_PageRevision::get().
167 * @see WikiDB_PageRevision::get
169 function get_versiondata($pagename, $version, $want_content = false) {
170 trigger_error("virtual", E_USER_ERROR);
174 * Delete page from the database with backup possibility.
175 * This should remove all links (from the named page) from
178 * @param $pagename string Page name.
179 * i.e save_page('') and DELETE nonempty id
180 * Can be undone and is seen in RecentChanges.
182 function delete_page($pagename) {
184 $user =& $GLOBALS['request']->_user;
185 $vdata = array('author' => $user->getId(),
186 'author_id' => $user->getAuthenticatedId(),
189 $this->lock(); // critical section:
190 $version = $this->get_latest_version($pagename);
191 $this->set_versiondata($pagename, $version+1, $vdata);
192 $this->set_links($pagename, false); // links are purged.
193 // SQL needs to invalidate the non_empty id
194 if (! WIKIDB_NOCACHE_MARKUP) {
195 // need the hits, perms and LOCKED, otherwise you can reset the perm
196 // by action=remove and re-create it with default perms
197 $pagedata = $this->get_pagedata($pagename);
198 unset($pagedata['_cached_html']);
199 $this->update_pagedata($pagename, $pagedata);
205 * Delete page (and all its revisions) from the database.
208 function purge_page($pagename) {
209 trigger_error("virtual", E_USER_ERROR);
213 * Delete an old revision of a page.
215 * Note that one is never allowed to delete the most recent version,
216 * but that this requirement is enforced by WikiDB not by the backend.
218 * In fact, to be safe, backends should probably allow the deletion of
219 * the most recent version.
221 * @param $pagename string Page name.
222 * @param $version integer Version to delete.
224 function delete_versiondata($pagename, $version) {
225 trigger_error("virtual", E_USER_ERROR);
229 * Create a new page revision.
231 * If the given ($pagename,$version) is already in the database,
232 * this method completely overwrites any stored data for that version.
234 * @param $pagename string Page name.
235 * @param $version int New revisions content.
236 * @param $data hash New revision metadata.
238 * @see get_versiondata
240 function set_versiondata($pagename, $version, $data) {
241 trigger_error("virtual", E_USER_ERROR);
245 * Update page version meta-data.
247 * If the given ($pagename,$version) is already in the database,
248 * this method only changes those meta-data values whose keys are
249 * explicity listed in $newdata.
251 * @param $pagename string Page name.
252 * @param $version int New revisions content.
253 * @param $newdata hash New revision metadata.
254 * @see set_versiondata, get_versiondata
256 function update_versiondata($pagename, $version, $newdata) {
257 $data = $this->get_versiondata($pagename, $version, true);
262 foreach ($newdata as $key => $val) {
268 $this->set_versiondata($pagename, $version, $data);
272 * Set links for page.
274 * @param $pagename string Page name.
276 * @param $links array List of page(names) which page links to.
278 function set_links($pagename, $links) {
279 trigger_error("virtual", E_USER_ERROR);
283 * Find pages which link to or are linked from a page.
285 * @param $pagename string Page name.
286 * @param $reversed boolean True to get backlinks.
288 * FIXME: array or iterator?
289 * @return object A WikiDB_backend_iterator.
291 function get_links($pagename, $reversed, $include_empty=false,
292 $sortby='', $limit='', $exclude='') {
293 //FIXME: implement simple (but slow) link finder.
294 die("FIXME get_links");
298 * Get all revisions of a page.
300 * @param $pagename string The page name.
301 * @return object A WikiDB_backend_iterator.
303 function get_all_revisions($pagename) {
304 include_once('lib/WikiDB/backend/dumb/AllRevisionsIter.php');
305 return new WikiDB_backend_dumb_AllRevisionsIter($this, $pagename);
309 * Get all pages in the database.
311 * Pages should be returned in alphabetical order if that is
316 * @param $include_defaulted boolean
317 * If set, even pages with no content will be returned
318 * --- but still only if they have at least one revision (not
319 * counting the default revision 0) entered in the database.
321 * Normally pages whose current revision has empty content
322 * are not returned as these pages are considered to be
325 * @return object A WikiDB_backend_iterator.
327 function get_all_pages($include_defaulted, $orderby=false, $limit='', $exclude='') {
328 trigger_error("virtual", E_USER_ERROR);
332 * Title or full text search.
334 * Pages should be returned in alphabetical order if that is
339 * @param $search object A TextSearchQuery object describing the parsed query string,
340 * with efficient methods for SQL and PCRE match.
342 * @param $fullsearch boolean If true, a full text search is performed,
343 * otherwise a title search is performed.
345 * @return object A WikiDB_backend_iterator.
347 * @see WikiDB::titleSearch
349 function text_search($search, $fulltext=false, $sortby='',
350 $limit='', $exclude='')
352 // This method implements a simple linear search
353 // through all the pages in the database.
355 // It is expected that most backends will overload
356 // this method with something more efficient.
357 include_once('lib/WikiDB/backend/dumb/TextSearchIter.php');
359 $pages = $this->get_all_pages(false, $sortby, false, $exclude);
360 return new WikiDB_backend_dumb_TextSearchIter($this, $pages, $search, $fulltext,
361 array('limit' => $limit,
362 'exclude' => $exclude));
369 * @param $pages object A TextSearchQuery object.
370 * @param $linkvalue object A TextSearchQuery object for the linkvalues
371 * (linkto, relation or backlinks or attribute values).
372 * @param $linktype string One of the 4 linktypes.
373 * @param $relation object A TextSearchQuery object or false.
374 * @param $options array Currently ignored. hash of sortby, limit, exclude.
375 * @return object A WikiDB_backend_iterator.
376 * @see WikiDB::linkSearch
378 function link_search( $pages, $linkvalue, $linktype, $relation=false, $options=array() ) {
379 include_once('lib/WikiDB/backend/dumb/LinkSearchIter.php');
380 $pageiter = $this->text_search($pages);
381 return new WikiDB_backend_dumb_LinkSearchIter($this, $pageiter, $linkvalue, $linktype, $relation, $options);
385 * Find pages with highest hit counts.
387 * Find the pages with the highest hit counts. The pages should
388 * be returned in reverse order by hit count.
391 * @param integer $limit No more than this many pages
392 * @return object A WikiDB_backend_iterator.
394 function most_popular($limit, $sortby='-hits') {
395 // This is method fetches all pages, then
396 // sorts them by hit count.
397 // (Not very efficient.)
399 // It is expected that most backends will overload
400 // method with something more efficient.
401 include_once('lib/WikiDB/backend/dumb/MostPopularIter.php');
402 $pages = $this->get_all_pages(false, $sortby, false);
403 return new WikiDB_backend_dumb_MostPopularIter($this, $pages, $limit);
407 * Find recent changes.
410 * @param $params hash See WikiDB::mostRecent for a description
411 * of parameters which can be included in this hash.
412 * @return object A WikiDB_backend_iterator.
413 * @see WikiDB::mostRecent
415 function most_recent($params) {
416 // This method is very inefficient and searches through
417 // all pages for the most recent changes.
419 // It is expected that most backends will overload
420 // method with something more efficient.
421 include_once('lib/WikiDB/backend/dumb/MostRecentIter.php');
422 $pages = $this->get_all_pages(true, '-mtime');
423 return new WikiDB_backend_dumb_MostRecentIter($this, $pages, $params);
426 function wanted_pages($exclude_from='', $exclude='', $sortby='', $limit='') {
427 include_once('lib/WikiDB/backend/dumb/WantedPagesIter.php');
428 $allpages = $this->get_all_pages(true,false,false,$exclude_from);
429 return new WikiDB_backend_dumb_WantedPagesIter($this, $allpages, $exclude, $sortby, $limit);
433 * Lock backend database.
435 * Calls may be nested.
437 * @param $write_lock boolean Unless this is set to false, a write lock
438 * is acquired, otherwise a read lock. If the backend doesn't support
439 * read locking, then it should make a write lock no matter which type
440 * of lock was requested.
442 * All backends <em>should</em> support write locking.
444 function lock($write_lock = true) {
448 * Unlock backend database.
450 * @param $force boolean Normally, the database is not unlocked until
451 * unlock() is called as many times as lock() has been. If $force is
452 * set to true, the the database is unconditionally unlocked.
454 function unlock($force = false) {
465 * Synchronize with filesystem.
467 * This should flush all unwritten data to the filesystem.
473 * Optimize the database.
475 function optimize() {
479 * Check database integrity.
481 * This should check the validity of the internal structure of the database.
482 * Errors should be reported via:
484 * trigger_error("Message goes here.", E_USER_WARNING);
487 * @return boolean True iff database is in a consistent state.
489 function check($args=false) {
493 * Put the database into a consistent state
494 * by reparsing and restoring all pages.
496 * This should put the database into a consistent state.
497 * (I.e. rebuild indexes, etc...)
499 * @return boolean True iff successful.
501 function rebuild($args=false) {
503 $dbh = $request->getDbh();
504 $iter = $dbh->getAllPages(false);
505 while ($page = $iter->next()) {
506 $current = $page->getCurrentRevision(true);
507 $pagename = $page->getName();
508 $meta = $current->_data;
509 $version = $current->getVersion();
510 $content =& $meta['%content'];
511 $formatted = new TransformedText($page, $content, $current->getMetaData());
512 $type = $formatted->getType();
513 $meta['pagetype'] = $type->getName();
514 $links = $formatted->getWikiPageLinks(); // linkto => relation
515 $this->lock(array('version','page','recent','link','nonempty'));
516 $this->set_versiondata($pagename, $version, $meta);
517 $this->set_links($pagename, $links);
518 $this->unlock(array('version','page','recent','link','nonempty'));
522 function _parse_searchwords($search) {
523 $search = strtolower(trim($search));
525 return array(array(),array());
527 $words = preg_split('/\s+/', $search);
529 foreach ($words as $key => $word) {
530 if ($word[0] == '-' && $word != '-') {
531 $word = substr($word, 1);
532 $exclude[] = preg_quote($word);
536 return array($words, $exclude);
540 * Split the given limit parameter into offset,limit. (offset is optional. default: 0)
541 * Duplicate the PageList function here to avoid loading the whole PageList.php
543 * list($offset,$count) = $this->limit($args['limit']);
545 function limit($limit) {
546 if (strstr($limit, ',')) {
547 list($from, $limit) = explode(',', $limit);
548 if ((!empty($from) && !is_numeric($from)) or (!empty($limit) && !is_numeric($limit))) {
549 return $this->error(_("Illegal 'limit' argument: must be numeric"));
551 return array($from, $limit);
554 if (!empty($limit) && !is_numeric($limit)) {
555 return $this->error(_("Illegal 'limit' argument: must be numeric"));
557 return array(0, $limit);
562 * Handle sortby requests for the DB iterator and table header links.
563 * Prefix the column with + or - like "+pagename","-mtime", ...
564 * supported actions: 'flip_order' "mtime" => "+mtime" => "-mtime" ...
565 * 'db' "-pagename" => "pagename DESC"
566 * In PageList all columns are sortable. (patch by DanFr)
567 * Here with the backend only some, the rest is delayed to PageList.
568 * (some kind of DumbIter)
569 * Duplicate the PageList function here to avoid loading the whole
570 * PageList.php, and it forces the backend specific sortable_columns()
572 function sortby ($column, $action, $sortable_columns=false) {
573 if (empty($column)) return '';
574 //support multiple comma-delimited sortby args: "+hits,+pagename"
575 if (strstr($column, ',')) {
577 foreach (explode(',', $column) as $col) {
579 $result[] = WikiDB_backend::sortby($col, $action);
581 $result[] = $this->sortby($col, $action);
583 return join(",",$result);
585 if (substr($column,0,1) == '+') {
586 $order = '+'; $column = substr($column,1);
587 } elseif (substr($column,0,1) == '-') {
588 $order = '-'; $column = substr($column,1);
590 // default order: +pagename, -mtime, -hits
592 if (in_array($column,array('mtime','hits')))
596 if ($action == 'flip_order') {
597 return ($order == '+' ? '-' : '+') . $column;
598 } elseif ($action == 'init') {
599 $this->_sortby[$column] = $order;
600 return $order . $column;
601 } elseif ($action == 'check') {
602 return (!empty($this->_sortby[$column]) or
603 ($GLOBALS['request']->getArg('sortby') and
604 strstr($GLOBALS['request']->getArg('sortby'),$column)));
605 } elseif ($action == 'db') {
606 // native sort possible?
607 if (!empty($this) and !$sortable_columns)
608 $sortable_columns = $this->sortable_columns();
609 if (in_array($column, $sortable_columns))
610 // asc or desc: +pagename, -pagename
611 return $column . ($order == '+' ? ' ASC' : ' DESC');
618 function sortable_columns() {
619 return array('pagename'/*,'mtime','author_id','author'*/);
622 // adds surrounding quotes
623 function quote ($s) { return "'".$s."'"; }
624 // no surrounding quotes because we know it's a string
625 function qstr ($s) { return $s; }
628 return in_array(DATABASE_TYPE, array('SQL','ADODB','PDO'));
631 function backendType() {
632 return DATABASE_TYPE;
635 function write_accesslog(&$entry) {
637 if (!$this->isSQL()) return;
639 $log_tbl = $entry->_accesslog->logtable;
640 // duration problem: sprintf "%f" might use comma e.g. "100,201" in european locales
641 $dbh->query("INSERT INTO $log_tbl"
642 . " (time_stamp,remote_host,remote_user,request_method,request_line,request_args,"
643 . "request_uri,request_time,status,bytes_sent,referer,agent,request_duration)"
644 . " VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?)",
646 // Problem: date formats are backend specific. Either use unixtime as %d (long),
647 // or the native timestamp format.
651 $entry->request_method,
653 $entry->request_args,
655 $entry->_ncsa_time($entry->time),
665 * Iterator returned by backend methods which (possibly) return
668 * FIXME: This might be two seperate classes: page_iter and version_iter.
669 * For the versions we have WikiDB_backend_dumb_AllRevisionsIter.
671 class WikiDB_backend_iterator
674 * Get the next record in the iterator set.
676 * This returns a hash. The hash may contain the following keys:
678 * <dt> pagename <dt> (string) the page name or linked page name on link iterators
679 * <dt> version <dt> (int) the version number
680 * <dt> pagedata <dt> (hash) page meta-data (as returned from backend::get_pagedata().)
681 * <dt> versiondata <dt> (hash) page meta-data (as returned from backend::get_versiondata().)
682 * <dt> linkrelation <dt> (string) the page naming the relation (e.g. isa:=page <=> isa)
684 * If this is a page iterator, it must contain the 'pagename' entry --- the others
687 * If this is a version iterator, the 'pagename', 'version', <strong>and</strong> 'versiondata'
688 * entries are mandatory. ('pagedata' is optional.)
690 * If this is a link iterator, the 'pagename' is mandatory, 'linkrelation' is optional.
693 trigger_error("virtual", E_USER_ERROR);
697 if (!empty($this->_pages))
698 return count($this->_pages);
704 if (!empty($this->_pages)) {
705 reset($this->_pages);
706 return $this->_pages;
709 while ($page = $this->next())
716 * limit - if empty the pagelist iterator will do nothing.
717 * Some backends limit the result set itself (dba, file, flatfile),
718 * Some SQL based leave it to WikiDB/PageList - deferred filtering in the iterator.
721 return empty($this->_options['limit']) ? 0 : $this->_options['limit'];
725 * Release resources held by this iterator.
732 * search baseclass, pcre-specific
734 class WikiDB_backend_search
736 function WikiDB_backend_search($search, &$dbh) {
738 $this->_case_exact = $search->_case_exact;
739 $this->_stoplist =& $search->_stoplist;
740 $this->stoplisted = array();
742 function _quote($word) {
743 return preg_quote($word, "/");
745 //TODO: use word anchors
746 function EXACT($word) { return "^".$this->_quote($word)."$"; }
747 function STARTS_WITH($word) { return "^".$this->_quote($word); }
748 function ENDS_WITH($word) { return $this->_quote($word)."$"; }
749 function WORD($word) { return $this->_quote($word); }
750 function REGEX($word) { return $word; }
752 function _pagename_match_clause($node) {
754 $word = $this->$method($node->word);
755 return "preg_match(\"/\".$word.\"/\"".($this->_case_exact ? "i":"").")";
757 /* Eliminate stoplist words.
758 * Keep a list of Stoplisted words to inform the poor user.
760 function isStoplisted ($node) {
761 // check only on WORD or EXACT fulltext search
762 if ($node->op != 'WORD' and $node->op != 'EXACT')
764 if (preg_match("/^".$this->_stoplist."$/i", $node->word)) {
765 array_push($this->stoplisted, $node->word);
770 function getStoplisted($word) {
771 return $this->stoplisted;
776 * search baseclass, sql-specific
778 class WikiDB_backend_search_sql extends WikiDB_backend_search
780 function _pagename_match_clause($node) {
781 // word already quoted by TextSearchQuery_node_word::_sql_quote()
782 $word = $node->sql();
783 if ($word == '%') // ALL shortcut
786 return ($this->_case_exact
787 ? "pagename LIKE '$word'"
788 : "LOWER(pagename) LIKE '$word'");
790 function _fulltext_match_clause($node) {
791 // force word-style %word% for fulltext search
792 $word = '%' . $node->_sql_quote($node->word) . '%';
793 // eliminate stoplist words
794 if ($this->isStoplisted($node))
795 return "1=1"; // and (pagename or 1) => and 1
797 return $this->_pagename_match_clause($node)
798 // probably convert this MATCH AGAINST or SUBSTR/POSITION without wildcards
799 . ($this->_case_exact ? " OR content LIKE '$word'"
800 : " OR LOWER(content) LIKE '$word'");
808 // c-hanging-comment-ender-p: nil
809 // indent-tabs-mode: nil