API: Crusade against extract(). Left one extract() call alone in ApiQueryBacklinks...
[lhc/web/wiklou.git] / includes / api / ApiQueryRevisions.php
index d3129b0..af2afdc 100644 (file)
@@ -1,12 +1,11 @@
 <?php
 
-
 /*
  * Created on Sep 7, 2006
  *
  * API for MediaWiki 1.8+
  *
- * Copyright (C) 2006 Yuri Astrakhan <FirstnameLastname@gmail.com>
+ * Copyright (C) 2006 Yuri Astrakhan <Firstname><Lastname>@gmail.com
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
 
 if (!defined('MEDIAWIKI')) {
        // Eclipse helper - will be ignored in production
-       require_once ("ApiQueryBase.php");
+       require_once ('ApiQueryBase.php');
 }
 
+/**
+ * A query action to enumerate revisions of a given page, or show top revisions of multiple pages.
+ * Various pieces of information may be shown - flags, comments, and the actual wiki markup of the rev.
+ * In the enumeration mode, ranges of revisions may be requested and filtered.
+ *
+ * @ingroup API
+ */
 class ApiQueryRevisions extends ApiQueryBase {
 
-       public function __construct($query, $moduleName, $generator = false) {
-               parent :: __construct($query, $moduleName, $generator);
+       public function __construct($query, $moduleName) {
+               parent :: __construct($query, $moduleName, 'rv');
+       }
+
+       private $fld_ids = false, $fld_flags = false, $fld_timestamp = false, $fld_size = false,
+                       $fld_comment = false, $fld_user = false, $fld_content = false;
+
+       protected function getTokenFunctions() {
+               // tokenname => function
+               // function prototype is func($pageid, $title, $rev)
+               // should return token or false
+
+               // Don't call the hooks twice
+               if(isset($this->tokenFunctions))
+                       return $this->tokenFunctions;
+
+               // If we're in JSON callback mode, no tokens can be obtained
+               if(!is_null($this->getMain()->getRequest()->getVal('callback')))
+                       return array();
+
+               $this->tokenFunctions = array(
+                       'rollback' => array( 'ApiQueryRevisions', 'getRollbackToken' )
+               );
+               wfRunHooks('APIQueryRevisionsTokens', array(&$this->tokenFunctions));
+               return $this->tokenFunctions;
+       }
+
+       public static function getRollbackToken($pageid, $title, $rev)
+       {
+               global $wgUser;
+               if(!$wgUser->isAllowed('rollback'))
+                       return false;
+               return $wgUser->editToken(array($title->getPrefixedText(),
+                                               $rev->getUserText()));
        }
 
        public function execute() {
-               $rvlimit = $rvstartid = $rvendid = $rvstart = $rvend = $rvdir = $rvprop = null;
-               extract($this->extractRequestParams());
+               $params = $this->extractRequestParams(false);
 
-               //
-               // Parameter validation
-               //
+               // If any of those parameters are used, work in 'enumeration' mode.
+               // Enum mode can only be used when exactly one page is provided.
+               // Enumerating revisions on multiple pages make it extremely
+               // difficult to manage continuations and require additional SQL indexes
+               $enumRevMode = (!is_null($params['user']) || !is_null($params['excludeuser']) ||
+                               !is_null($params['limit']) || !is_null($params['startid']) ||
+                               !is_null($params['endid']) || $params['dir'] === 'newer' ||
+                               !is_null($params['start']) || !is_null($params['end']));
 
-               // true when ordered by timestamp from older to newer, false otherwise
-               $dirNewer = ($rvdir === 'newer');
 
-               // If any of those parameters are used, we work with single page only
-               $singePageMode = ($rvlimit !== 0 || $rvstartid !== 0 || $rvendid !== 0 || $dirNewer || isset ($rvstart) || isset ($rvend));
+               $pageSet = $this->getPageSet();
+               $pageCount = $pageSet->getGoodTitleCount();
+               $revCount = $pageSet->getRevisionCount();
 
-               if ($rvstartid !== 0 || $rvendid !== 0)
-                       $this->dieUsage('rvstartid/rvendid not implemented', 'notimplemented');
+               // Optimization -- nothing to do
+               if ($revCount === 0 && $pageCount === 0)
+                       return;
 
-               $data = $this->getData();
-               $pageCount = $data->getPageCount();
-               if ($singePageMode && $pageCount > 1)
-                       $this->dieUsage('You have supplied multiple pages, but the specified revisions parameters may only be used with one page.', 'rv_multpages');
+               if ($revCount > 0 && $enumRevMode)
+                       $this->dieUsage('The revids= parameter may not be used with the list options (limit, startid, endid, dirNewer, start, end).', 'revids');
 
-               $tables = array (
-                       'revision'
-               );
-               $fields = array (
-                       'rev_id',
-                       'rev_page',
-                       'rev_text_id',
-                       'rev_minor_edit'
-               );
-               $conds = array (
-                       'rev_deleted' => 0
-               );
-               $options = array ();
-
-               $showTimestamp = $showUser = $showComment = $showContent = false;
-               if (isset ($rvprop)) {
-                       foreach ($rvprop as $prop) {
-                               switch ($prop) {
-                                       case 'timestamp' :
-                                               $fields[] = 'rev_timestamp';
-                                               $showTimestamp = true;
-                                               break;
-                                       case 'user' :
-                                               $fields[] = 'rev_user';
-                                               $fields[] = 'rev_user_text';
-                                               $showUser = true;
-                                               break;
-                                       case 'comment' :
-                                               $fields[] = 'rev_comment';
-                                               $showComment = true;
-                                               break;
-                                       case 'content' :
-                                               // todo: check the page count/limit when requesting content
-                                               //$this->validateLimit( 'content: (rvlimit*pages)+revids',
-                                               //$rvlimit * count($this->existingPageIds) + count($this->revIdsArray), 50, 200 );
-                                               $tables[] = 'text';
-                                               $conds[] = 'rev_text_id=old_id';
-                                               $fields[] = 'old_id';
-                                               $fields[] = 'old_text';
-                                               $fields[] = 'old_flags';
-                                               break;
-                                       default :
-                                               $this->dieDebug("unknown rvprop $prop");
-                               }
+               if ($pageCount > 1 && $enumRevMode)
+                       $this->dieUsage('titles, pageids or a generator was used to supply multiple pages, but the limit, startid, endid, dirNewer, user, excludeuser, start and end parameters may only be used on a single page.', 'multpages');
+
+               $this->addTables('revision');
+               $this->addFields( Revision::selectFields() );
+               $this->addTables( 'page' );
+               $this->addWhere('page_id = rev_page');
+
+               $prop = array_flip($params['prop']);
+
+               // Optional fields
+               $this->fld_ids = isset ($prop['ids']);
+               // $this->addFieldsIf('rev_text_id', $this->fld_ids); // should this be exposed?
+               $this->fld_flags = isset ($prop['flags']);
+               $this->fld_timestamp = isset ($prop['timestamp']);
+               $this->fld_comment = isset ($prop['comment']);
+               $this->fld_size = isset ($prop['size']);
+               $this->fld_user = isset ($prop['user']);
+               $this->token = $params['token'];
+
+               if ( !is_null($this->token) || $pageCount > 0) {
+                       $this->addFields( Revision::selectPageFields() );
+               }
+
+               if (isset ($prop['content'])) {
+
+                       // For each page we will request, the user must have read rights for that page
+                       foreach ($pageSet->getGoodTitles() as $title) {
+                               if( !$title->userCanRead() )
+                                       $this->dieUsage(
+                                               'The current user is not allowed to read ' . $title->getPrefixedText(),
+                                               'accessdenied');
                        }
+
+                       $this->addTables('text');
+                       $this->addWhere('rev_text_id=old_id');
+                       $this->addFields('old_id');
+                       $this->addFields( Revision::selectTextFields() );
+
+                       $this->fld_content = true;
+
+                       $this->expandTemplates = $params['expandtemplates'];
+                       $this->generateXML = $params['generatexml'];
+                       if(isset($params['section']))
+                               $this->section = $params['section'];
+                       else
+                               $this->section = false;
+               }
+
+               $userMax = ( $this->fld_content ? ApiBase::LIMIT_SML1 : ApiBase::LIMIT_BIG1 );
+               $botMax  = ( $this->fld_content ? ApiBase::LIMIT_SML2 : ApiBase::LIMIT_BIG2 );
+               $limit = $params['limit'];
+               if( $limit == 'max' ) {
+                       $limit = $this->getMain()->canApiHighLimits() ? $botMax : $userMax;
+                       $this->getResult()->addValue( 'limits', $this->getModuleName(), $limit );
                }
 
-               if (isset ($rvstart))
-                       $conds[] = 'rev_timestamp >= ' . $this->prepareTimestamp($rvstart);
-               if (isset ($rvend))
-                       $conds[] = 'rev_timestamp <= ' . $this->prepareTimestamp($rvend);
+               if ($enumRevMode) {
 
-               if ($singePageMode) {
-                       if (!isset ($rvlimit))
-                               $rvlimit = 10;
+                       // This is mostly to prevent parameter errors (and optimize SQL?)
+                       if (!is_null($params['startid']) && !is_null($params['start']))
+                               $this->dieUsage('start and startid cannot be used together', 'badparams');
 
-                       $options['LIMIT'] = $rvlimit + 1;
-                       $options['ORDER BY'] = 'rev_timestamp' . ($dirNewer ? '' : ' DESC');
-                       
-                       // get the first (and only) pageid => title pair
-                       foreach($data->getGoodTitles() as $pageId => $titleObj) {
-                               $conds['rev_page'] = $pageId;
-                               break;
+                       if (!is_null($params['endid']) && !is_null($params['end']))
+                               $this->dieUsage('end and endid cannot be used together', 'badparams');
+
+                       if(!is_null($params['user']) && !is_null($params['excludeuser']))
+                               $this->dieUsage('user and excludeuser cannot be used together', 'badparams');
+
+                       // This code makes an assumption that sorting by rev_id and rev_timestamp produces
+                       // the same result. This way users may request revisions starting at a given time,
+                       // but to page through results use the rev_id returned after each page.
+                       // Switching to rev_id removes the potential problem of having more than
+                       // one row with the same timestamp for the same page.
+                       // The order needs to be the same as start parameter to avoid SQL filesort.
+
+                       if (is_null($params['startid']) && is_null($params['endid']))
+                               $this->addWhereRange('rev_timestamp', $params['dir'],
+                                       $params['start'], $params['end']);
+                       else
+                               $this->addWhereRange('rev_id', $params['dir'],
+                                       $params['startid'], $params['endid']);
+
+                       // must manually initialize unset limit
+                       if (is_null($limit))
+                               $limit = 10;
+                       $this->validateLimit('limit', $limit, 1, $userMax, $botMax);
+
+                       // There is only one ID, use it
+                       $this->addWhereFld('rev_page', current(array_keys($pageSet->getGoodTitles())));
+
+                       if(!is_null($params['user'])) {
+                               $this->addWhereFld('rev_user_text', $params['user']);
+                       } elseif (!is_null( $params['excludeuser'])) {
+                               $this->addWhere('rev_user_text != ' .
+                                       $this->getDB()->addQuotes($params['excludeuser']));
                        }
                }
+               elseif ($revCount > 0) {
+                       $max = $this->getMain()->canApiHighLimits() ? $botMax : $userMax;
+                       $revs = $pageSet->getRevisionIDs();
+                       if(self::truncateArray($revs, $max))
+                               $this->setWarning("Too many values supplied for parameter 'revids': the limit is $max"); 
 
-               $db = $this->getDB();
-               $this->profileDBIn();
-               $res = $db->select($tables, $fields, $conds, __CLASS__ . '::' . __FUNCTION__, $options);
-               $this->profileDBOut();
+                       // Get all revision IDs
+                       $this->addWhereFld('rev_id', array_keys($revs));
+
+                       // assumption testing -- we should never get more then $revCount rows.
+                       $limit = $revCount;
+               }
+               elseif ($pageCount > 0) {
+                       $max = $this->getMain()->canApiHighLimits() ? $botMax : $userMax;
+                       $titles = $pageSet->getGoodTitles();
+                       if(self::truncateArray($titles, $max))
+                               $this->setWarning("Too many values supplied for parameter 'titles': the limit is $max");
+                       
+                       // When working in multi-page non-enumeration mode,
+                       // limit to the latest revision only
+                       $this->addWhere('page_id=rev_page');
+                       $this->addWhere('page_latest=rev_id');
+                       
+                       // Get all page IDs
+                       $this->addWhereFld('page_id', array_keys($titles));
+
+                       // assumption testing -- we should never get more then $pageCount rows.
+                       $limit = $pageCount;
+               } else
+                       ApiBase :: dieDebug(__METHOD__, 'param validation?');
+
+               $this->addOption('LIMIT', $limit +1);
 
                $data = array ();
                $count = 0;
+               $res = $this->select(__METHOD__);
+
+               $db = $this->getDB();
                while ($row = $db->fetchObject($res)) {
 
-                       if (++ $count > $rvlimit) {
+                       if (++ $count > $limit) {
                                // We've reached the one extra which shows that there are additional pages to be had. Stop here...
-                               $startStr = 'rvstartid=' . $row->rev_id;
-                               $msg = array ('continue' => $startStr );
-                               $this->getResult()->addMessage('query-status', 'revisions', $msg);
+                               if (!$enumRevMode)
+                                       ApiBase :: dieDebug(__METHOD__, 'Got more rows then expected'); // bug report
+                               $this->setContinueEnumParameter('startid', intval($row->rev_id));
                                break;
                        }
 
+                       $revision = new Revision( $row );
+                       $this->getResult()->addValue(
+                               array (
+                                       'query',
+                                       'pages',
+                                       $revision->getPage(),
+                                       'revisions'),
+                               null,
+                               $this->extractRowInfo( $revision ));
+               }
+               $db->freeResult($res);
 
-                       $revid = intval($row->rev_id);
-                       $pageid = intval($row->rev_page);
+               // Ensure that all revisions are shown as '<rev>' elements
+               $result = $this->getResult();
+               if ($result->getIsRawMode()) {
+                       $data =& $result->getData();
+                       foreach ($data['query']['pages'] as & $page) {
+                               if (is_array($page) && array_key_exists('revisions', $page)) {
+                                       $result->setIndexedTagName($page['revisions'], 'rev');
+                               }
+                       }
+               }
+       }
 
-                       $vals = array (
-                               'revid' => $revid,
-                               'oldid' => intval($row->rev_text_id
-                       ));
+       private function extractRowInfo( $revision ) {
 
-                       if( $row->rev_minor_edit ) {
-                               $vals['minor'] = '';
-                       }
-       
-                       if ($showTimestamp)
-                               $vals['timestamp'] = wfTimestamp(TS_ISO_8601, $row->rev_timestamp);
-       
-                       if ($showUser) {
-                               $vals['user'] = $row->rev_user_text;
-                               if( !$row->rev_user )
-                                       $vals['anon'] = '';
-                       }
+               $vals = array ();
 
-                       if ($showComment)
-                               $vals['comment'] = $row->rev_comment;
+               if ($this->fld_ids) {
+                       $vals['revid'] = $revision->getId();
+                       // $vals['oldid'] = intval($row->rev_text_id);  // todo: should this be exposed?
+               }
 
-                       if ($showContent) {
-                               $vals['xml:space'] = 'preserve';
-                               $vals['*'] = Revision::getRevisionText( $row );
-                       } else {
-                               $vals['*'] = '';        // Force all elements to be attributes
-                       }
+               if ($this->fld_flags && $revision->isMinor())
+                       $vals['minor'] = '';
 
-                       $data[$pageid]['revisions']['_element'] = 'rv';
-                       $data[$pageid]['revisions'][$revid] = $vals;
+               if ($this->fld_user) {
+                       $vals['user'] = $revision->getUserText();
+                       if (!$revision->getUser())
+                               $vals['anon'] = '';
                }
-               $db->freeResult($res);
 
-               $this->getResult()->addMessage('query', 'allpages', $data);
+               if ($this->fld_timestamp) {
+                       $vals['timestamp'] = wfTimestamp(TS_ISO_8601, $revision->getTimestamp());
+               }
+
+               if ($this->fld_size && !is_null($revision->getSize())) {
+                       $vals['size'] = $revision->getSize();
+               }
+
+               if ($this->fld_comment) {
+                       $comment = $revision->getComment();
+                       if (strval($comment) !== '')
+                               $vals['comment'] = $comment;
+               }
+
+               if(!is_null($this->token) || ($this->fld_content && $this->expandTemplates))
+                       $title = $revision->getTitle();
+
+               if(!is_null($this->token))
+               {
+                       $tokenFunctions = $this->getTokenFunctions();
+                       foreach($this->token as $t)
+                       {
+                               $val = call_user_func($tokenFunctions[$t], $title->getArticleID(), $title, $revision);
+                               if($val === false)
+                                       $this->setWarning("Action '$t' is not allowed for the current user");
+                               else
+                                       $vals[$t . 'token'] = $val;
+                       }
+               }
+
+               if ($this->fld_content) {
+                       global $wgParser;
+                       $text = $revision->getText();
+                       # Expand templates after getting section content because
+                       # template-added sections don't count and Parser::preprocess()
+                       # will have less input
+                       if ($this->section !== false) {
+                               $text = $wgParser->getSection( $text, $this->section, false);
+                               if($text === false)
+                                       $this->dieUsage("There is no section {$this->section} in r".$revision->getId(), 'nosuchsection');
+                       }
+                       if ($this->generateXML) {
+                               $wgParser->startExternalParse( $title, new ParserOptions(), OT_PREPROCESS );
+                               $dom = $wgParser->preprocessToDom( $text );
+                               if ( is_callable( array( $dom, 'saveXML' ) ) ) {
+                                       $xml = $dom->saveXML();
+                               } else {
+                                       $xml = $dom->__toString();
+                               }
+                               $vals['parsetree'] = $xml;
+                               
+                       }
+                       if ($this->expandTemplates) {
+                               $text = $wgParser->preprocess( $text, $title, new ParserOptions() );
+                       }
+                       ApiResult :: setContent($vals, $text);
+               }
+               return $vals;
        }
 
-       protected function getAllowedParams() {
+       public function getAllowedParams() {
                return array (
-                       'rvlimit' => array (
-                               GN_ENUM_DFLT => 0,
-                               GN_ENUM_TYPE => 'limit',
-                               GN_ENUM_MIN => 0,
-                               GN_ENUM_MAX1 => 50,
-                               GN_ENUM_MAX2 => 500
+                       'prop' => array (
+                               ApiBase :: PARAM_ISMULTI => true,
+                               ApiBase :: PARAM_DFLT => 'ids|timestamp|flags|comment|user',
+                               ApiBase :: PARAM_TYPE => array (
+                                       'ids',
+                                       'flags',
+                                       'timestamp',
+                                       'user',
+                                       'size',
+                                       'comment',
+                                       'content',
+                               )
+                       ),
+                       'limit' => array (
+                               ApiBase :: PARAM_TYPE => 'limit',
+                               ApiBase :: PARAM_MIN => 1,
+                               ApiBase :: PARAM_MAX => ApiBase :: LIMIT_BIG1,
+                               ApiBase :: PARAM_MAX2 => ApiBase :: LIMIT_BIG2
                        ),
-                       'rvstartid' => 0,
-                       'rvendid' => 0,
-                       'rvstart' => array (
-                               GN_ENUM_TYPE => 'timestamp'
+                       'startid' => array (
+                               ApiBase :: PARAM_TYPE => 'integer'
                        ),
-                       'rvend' => array (
-                               GN_ENUM_TYPE => 'timestamp'
+                       'endid' => array (
+                               ApiBase :: PARAM_TYPE => 'integer'
                        ),
-                       'rvdir' => array (
-                               GN_ENUM_DFLT => 'older',
-                               GN_ENUM_TYPE => array (
+                       'start' => array (
+                               ApiBase :: PARAM_TYPE => 'timestamp'
+                       ),
+                       'end' => array (
+                               ApiBase :: PARAM_TYPE => 'timestamp'
+                       ),
+                       'dir' => array (
+                               ApiBase :: PARAM_DFLT => 'older',
+                               ApiBase :: PARAM_TYPE => array (
                                        'newer',
                                        'older'
                                )
                        ),
-                       'rvprop' => array (
-                               GN_ENUM_ISMULTI => true,
-                               GN_ENUM_TYPE => array (
-                                       'timestamp',
-                                       'user',
-                                       'comment',
-                                       'content'
-                               )
-                       )
+                       'user' => array(
+                               ApiBase :: PARAM_TYPE => 'user'
+                       ),
+                       'excludeuser' => array(
+                               ApiBase :: PARAM_TYPE => 'user'
+                       ),
+                       'expandtemplates' => false,
+                       'generatexml' => false,
+                       'section' => null,
+                       'token' => array(
+                               ApiBase :: PARAM_TYPE => array_keys($this->getTokenFunctions()),
+                               ApiBase :: PARAM_ISMULTI => true
+                       ),
                );
        }
 
-       protected function getDescription() {
-               return 'module a';
+       public function getParamDescription() {
+               return array (
+                       'prop' => 'Which properties to get for each revision.',
+                       'limit' => 'limit how many revisions will be returned (enum)',
+                       'startid' => 'from which revision id to start enumeration (enum)',
+                       'endid' => 'stop revision enumeration on this revid (enum)',
+                       'start' => 'from which revision timestamp to start enumeration (enum)',
+                       'end' => 'enumerate up to this timestamp (enum)',
+                       'dir' => 'direction of enumeration - towards "newer" or "older" revisions (enum)',
+                       'user' => 'only include revisions made by user',
+                       'excludeuser' => 'exclude revisions made by user',
+                       'expandtemplates' => 'expand templates in revision content',
+                       'generatexml' => 'generate XML parse tree for revision content',
+                       'section' => 'only retrieve the content of this section',
+                       'token' => 'Which tokens to obtain for each revision',
+               );
+       }
+
+       public function getDescription() {
+               return array (
+                       'Get revision information.',
+                       'This module may be used in several ways:',
+                       ' 1) Get data about a set of pages (last revision), by setting titles or pageids parameter.',
+                       ' 2) Get revisions for one given page, by using titles/pageids with start/end/limit params.',
+                       ' 3) Get data about a set of revisions by setting their IDs with revids parameter.',
+                       'All parameters marked as (enum) may only be used with a single page (#2).'
+               );
        }
 
        protected function getExamples() {
                return array (
-                       'api.php?action=query&prop=revisions&titles=ArticleA&rvprop=timestamp|user|comment|content'
+                       'Get data with content for the last revision of titles "API" and "Main Page":',
+                       '  api.php?action=query&prop=revisions&titles=API|Main%20Page&rvprop=timestamp|user|comment|content',
+                       'Get last 5 revisions of the "Main Page":',
+                       '  api.php?action=query&prop=revisions&titles=Main%20Page&rvlimit=5&rvprop=timestamp|user|comment',
+                       'Get first 5 revisions of the "Main Page":',
+                       '  api.php?action=query&prop=revisions&titles=Main%20Page&rvlimit=5&rvprop=timestamp|user|comment&rvdir=newer',
+                       'Get first 5 revisions of the "Main Page" made after 2006-05-01:',
+                       '  api.php?action=query&prop=revisions&titles=Main%20Page&rvlimit=5&rvprop=timestamp|user|comment&rvdir=newer&rvstart=20060501000000',
+                       'Get first 5 revisions of the "Main Page" that were not made made by anonymous user "127.0.0.1"',
+                       '  api.php?action=query&prop=revisions&titles=Main%20Page&rvlimit=5&rvprop=timestamp|user|comment&rvexcludeuser=127.0.0.1',
+                       'Get first 5 revisions of the "Main Page" that were made by the user "MediaWiki default"',
+                       '  api.php?action=query&prop=revisions&titles=Main%20Page&rvlimit=5&rvprop=timestamp|user|comment&rvuser=MediaWiki%20default',
                );
        }
+
+       public function getVersion() {
+               return __CLASS__ . ': $Id$';
+       }
 }
-?>
\ No newline at end of file