<?php
-
/*
* Created on Sep 24, 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');
}
+/**
+ * This class contains a list of pages that the client has requested.
+ * Initially, when the client passes in titles=, pageids=, or revisions= parameter,
+ * an instance of the ApiPageSet class will normalize titles,
+ * determine if the pages/revisions exist, and prefetch any additional data page data requested.
+ *
+ * When generator is used, the result of the generator will become the input for the
+ * second instance of this class, and all subsequent actions will go use the second instance
+ * for all their work.
+ *
+ * @addtogroup API
+ */
class ApiPageSet extends ApiQueryBase {
private $mAllPages; // [ns][dbkey] => page_id or 0 when missing
- private $mGoodTitles, $mMissingTitles, $mRedirectTitles, $mNormalizedTitles;
+ private $mTitles, $mGoodTitles, $mMissingTitles, $mMissingPageIDs, $mRedirectTitles;
+ private $mNormalizedTitles, $mInterwikiTitles;
+ private $mResolveRedirects, $mPendingRedirectIDs;
+ private $mGoodRevIDs, $mMissingRevIDs;
+ private $mFakePageId;
- public function __construct($query) {
+ private $mRequestedPageFields;
+
+ public function __construct($query, $resolveRedirects = false) {
parent :: __construct($query, __CLASS__);
$this->mAllPages = array ();
+ $this->mTitles = array();
$this->mGoodTitles = array ();
$this->mMissingTitles = array ();
+ $this->mMissingPageIDs = array ();
$this->mRedirectTitles = array ();
$this->mNormalizedTitles = array ();
+ $this->mInterwikiTitles = array ();
+ $this->mGoodRevIDs = array();
+ $this->mMissingRevIDs = array();
+
+ $this->mRequestedPageFields = array ();
+ $this->mResolveRedirects = $resolveRedirects;
+ if($resolveRedirects)
+ $this->mPendingRedirectIDs = array();
+
+ $this->mFakePageId = -1;
+ }
+
+ public function isResolvingRedirects() {
+ return $this->mResolveRedirects;
+ }
+
+ public function requestField($fieldName) {
+ $this->mRequestedPageFields[$fieldName] = null;
+ }
+
+ public function getCustomField($fieldName) {
+ return $this->mRequestedPageFields[$fieldName];
+ }
+
+ /**
+ * Get fields that modules have requested from the page table
+ */
+ public function getPageTableFields() {
+ // Ensure we get minimum required fields
+ $pageFlds = array (
+ 'page_id' => null,
+ 'page_namespace' => null,
+ 'page_title' => null
+ );
+
+ // only store non-default fields
+ $this->mRequestedPageFields = array_diff_key($this->mRequestedPageFields, $pageFlds);
+
+ if ($this->mResolveRedirects)
+ $pageFlds['page_is_redirect'] = null;
+
+ $pageFlds = array_merge($pageFlds, $this->mRequestedPageFields);
+ return array_keys($pageFlds);
+ }
+
+ /**
+ * Returns an array [ns][dbkey] => page_id for all requested titles
+ * page_id is a unique negative number in case title was not found
+ */
+ public function getAllTitlesByNamespace() {
+ return $this->mAllPages;
+ }
+
+ /**
+ * All Title objects provided.
+ * @return array of Title objects
+ */
+ public function getTitles() {
+ return $this->mTitles;
+ }
+
+ /**
+ * Returns the number of unique pages (not revisions) in the set.
+ */
+ public function getTitleCount() {
+ return count($this->mTitles);
}
/**
return $this->mGoodTitles;
}
+ /**
+ * Returns the number of found unique pages (not revisions) in the set.
+ */
+ public function getGoodTitleCount() {
+ return count($this->mGoodTitles);
+ }
+
/**
* Title objects that were NOT found in the database.
+ * The array's index will be negative for each item
* @return array of Title objects
*/
public function getMissingTitles() {
return $this->mMissingTitles;
}
+ /**
+ * Page IDs that were not found in the database
+ * @return array of page IDs
+ */
+ public function getMissingPageIDs() {
+ return $this->mMissingPageIDs;
+ }
+
/**
* Get a list of redirects when doing redirect resolution
* @return array prefixed_title (string) => prefixed_title (string)
}
/**
- * Returns the number of unique pages (not revisions) in the set.
+ * Get a list of interwiki titles - maps the title given
+ * with to the interwiki prefix.
+ * @return array raw_prefixed_title (string) => interwiki_prefix (string)
*/
- public function getGoodTitleCount() {
- return count($this->getGoodTitles());
+ public function getInterwikiTitles() {
+ return $this->mInterwikiTitles;
}
-
+
/**
* Get the list of revision IDs (requested with revids= parameter)
+ * @return array revID (int) => pageID (int)
*/
public function getRevisionIDs() {
- $this->dieUsage(__FUNCTION__ . " is not implemented", 'notimplemented');
+ return $this->mGoodRevIDs;
}
-
+
+ /**
+ * Revision IDs that were not found in the database
+ * @return array of revision IDs
+ */
+ public function getMissingRevisionIDs() {
+ return $this->mMissingRevIDs;
+ }
+
/**
* Returns the number of revisions (requested with revids= parameter)
*/
public function getRevisionCount() {
- return 0; // TODO: implement
+ return count($this->getRevisionIDs());
}
+ /**
+ * Populate from the request parameters
+ */
+ public function execute() {
+ $this->profileIn();
+ $titles = $pageids = $revids = null;
+ extract($this->extractRequestParams());
+
+ // Only one of the titles/pageids/revids is allowed at the same time
+ $dataSource = null;
+ if (isset ($titles))
+ $dataSource = 'titles';
+ if (isset ($pageids)) {
+ if (isset ($dataSource))
+ $this->dieUsage("Cannot use 'pageids' at the same time as '$dataSource'", 'multisource');
+ $dataSource = 'pageids';
+ }
+ if (isset ($revids)) {
+ if (isset ($dataSource))
+ $this->dieUsage("Cannot use 'revids' at the same time as '$dataSource'", 'multisource');
+ $dataSource = 'revids';
+ }
+
+ switch ($dataSource) {
+ case 'titles' :
+ $this->initFromTitles($titles);
+ break;
+ case 'pageids' :
+ $this->initFromPageIds($pageids);
+ break;
+ case 'revids' :
+ if($this->mResolveRedirects)
+ $this->dieUsage('revids may not be used with redirect resolution', 'params');
+ $this->initFromRevIDs($revids);
+ break;
+ default :
+ // Do nothing - some queries do not need any of the data sources.
+ break;
+ }
+ $this->profileOut();
+ }
+
+ /**
+ * Initialize PageSet from a list of Titles
+ */
+ public function populateFromTitles($titles) {
+ $this->profileIn();
+ $this->initFromTitles($titles);
+ $this->profileOut();
+ }
+
+ /**
+ * Initialize PageSet from a list of Page IDs
+ */
+ public function populateFromPageIDs($pageIDs) {
+ $this->profileIn();
+ $this->initFromPageIds($pageIDs);
+ $this->profileOut();
+ }
+
+ /**
+ * Initialize PageSet from a rowset returned from the database
+ */
+ public function populateFromQueryResult($db, $queryResult) {
+ $this->profileIn();
+ $this->initFromQueryResult($db, $queryResult);
+ $this->profileOut();
+ }
+
+ /**
+ * Initialize PageSet from a list of Revision IDs
+ */
+ public function populateFromRevisionIDs($revIDs) {
+ $this->profileIn();
+ $revIDs = array_map('intval', $revIDs); // paranoia
+ $this->initFromRevIDs($revIDs);
+ $this->profileOut();
+ }
+
+ /**
+ * Extract all requested fields from the row received from the database
+ */
+ public function processDbRow($row) {
+
+ // Store Title object in various data structures
+ $title = Title :: makeTitle($row->page_namespace, $row->page_title);
+
+ $pageId = intval($row->page_id);
+ $this->mAllPages[$row->page_namespace][$row->page_title] = $pageId;
+ $this->mTitles[] = $title;
+
+ if ($this->mResolveRedirects && $row->page_is_redirect == '1') {
+ $this->mPendingRedirectIDs[$pageId] = $title;
+ } else {
+ $this->mGoodTitles[$pageId] = $title;
+ }
+
+ foreach ($this->mRequestedPageFields as $fieldName => & $fieldValues)
+ $fieldValues[$pageId] = $row-> $fieldName;
+ }
+
+ public function finishPageSetGeneration() {
+ $this->profileIn();
+ $this->resolvePendingRedirects();
+ $this->profileOut();
+ }
+
/**
* This method populates internal variables with page information
* based on the given array of title strings.
* #5 Substitute the original LinkBatch object with the new list
* #6 Repeat from step #1
*/
- private function populateTitles($titles, $redirects) {
-
- $pageFlds = array (
- 'page_id',
- 'page_namespace',
- 'page_title'
- );
- if ($redirects) {
- $pageFlds[] = 'page_is_redirect';
- }
+ private function initFromTitles($titles) {
// Get validated and normalized title objects
- $linkBatch = $this->processTitlesStrArray($titles);
-
+ $linkBatch = $this->processTitlesArray($titles);
+ if($linkBatch->isEmpty())
+ return;
+
$db = $this->getDB();
+ $set = $linkBatch->constructSet('page', $db);
- //
- // Repeat until all redirects have been resolved
- //
- while (false !== ($set = $linkBatch->constructSet('page', $db))) {
+ // Get pageIDs data from the `page` table
+ $this->profileDBIn();
+ $res = $db->select('page', $this->getPageTableFields(), $set, __METHOD__);
+ $this->profileDBOut();
- // Hack: get the ns:titles stored in array(ns => array(titles)) format
- $remaining = $linkBatch->data;
+ // Hack: get the ns:titles stored in array(ns => array(titles)) format
+ $this->initFromQueryResult($db, $res, $linkBatch->data, true); // process Titles
- $redirectIds = array ();
+ // Resolve any found redirects
+ $this->resolvePendingRedirects();
+ }
- //
- // Get data about $linkBatch from `page` table
- //
- $this->profileDBIn();
- $res = $db->select('page', $pageFlds, $set, __CLASS__ . '::' . __FUNCTION__);
- $this->profileDBOut();
- while ($row = $db->fetchObject($res)) {
+ private function initFromPageIds($pageids) {
+ if(empty($pageids))
+ return;
- unset ($remaining[$row->page_namespace][$row->page_title]);
- $title = Title :: makeTitle($row->page_namespace, $row->page_title);
- $this->mAllPages[$row->page_namespace][$row->page_title] = $row->page_id;
+ $pageids = array_map('intval', $pageids); // paranoia
+ $set = array (
+ 'page_id' => $pageids
+ );
- if ($redirects && $row->page_is_redirect == '1') {
- $redirectIds[$row->page_id] = $title;
- } else {
- $this->mGoodTitles[$row->page_id] = $title;
- }
+ $db = $this->getDB();
+
+ // Get pageIDs data from the `page` table
+ $this->profileDBIn();
+ $res = $db->select('page', $this->getPageTableFields(), $set, __METHOD__);
+ $this->profileDBOut();
+
+ $this->initFromQueryResult($db, $res, array_flip($pageids), false); // process PageIDs
+
+ // Resolve any found redirects
+ $this->resolvePendingRedirects();
+ }
+
+ /**
+ * Iterate through the result of the query on 'page' table,
+ * and for each row create and store title object and save any extra fields requested.
+ * @param $db Database
+ * @param $res DB Query result
+ * @param $remaining Array of either pageID or ns/title elements (optional).
+ * If given, any missing items will go to $mMissingPageIDs and $mMissingTitles
+ * @param $processTitles bool Must be provided together with $remaining.
+ * If true, treat $remaining as an array of [ns][title]
+ * If false, treat it as an array of [pageIDs]
+ * @return Array of redirect IDs (only when resolving redirects)
+ */
+ private function initFromQueryResult($db, $res, &$remaining = null, $processTitles = null) {
+ if (!is_null($remaining) && is_null($processTitles))
+ ApiBase :: dieDebug(__METHOD__, 'Missing $processTitles parameter when $remaining is provided');
+
+ while ($row = $db->fetchObject($res)) {
+
+ $pageId = intval($row->page_id);
+
+ // Remove found page from the list of remaining items
+ if (isset($remaining)) {
+ if ($processTitles)
+ unset ($remaining[$row->page_namespace][$row->page_title]);
+ else
+ unset ($remaining[$pageId]);
}
- $db->freeResult($res);
-
- //
- // The remaining titles in $remaining are non-existant pages
- //
- foreach ($remaining as $ns => $dbkeys) {
- foreach ($dbkeys as $dbkey => $nothing) {
- $this->mMissingTitles[] = Title :: makeTitle($ns, $dbkey);
- $this->mAllPages[$ns][$dbkey] = 0;
+
+ // Store any extra fields requested by modules
+ $this->processDbRow($row);
+ }
+ $db->freeResult($res);
+
+ if(isset($remaining)) {
+ // Any items left in the $remaining list are added as missing
+ if($processTitles) {
+ // The remaining titles in $remaining are non-existant pages
+ foreach ($remaining as $ns => $dbkeys) {
+ foreach ( $dbkeys as $dbkey => $unused ) {
+ $title = Title :: makeTitle($ns, $dbkey);
+ $this->mAllPages[$ns][$dbkey] = $this->mFakePageId;
+ $this->mMissingTitles[$this->mFakePageId] = $title;
+ $this->mFakePageId--;
+ $this->mTitles[] = $title;
+ }
}
}
+ else
+ {
+ // The remaining pageids do not exist
+ if(empty($this->mMissingPageIDs))
+ $this->mMissingPageIDs = array_keys($remaining);
+ else
+ $this->mMissingPageIDs = array_merge($this->mMissingPageIDs, array_keys($remaining));
+ }
+ }
+ }
- if (!$redirects || empty ($redirectIds))
- break;
+ private function initFromRevIDs($revids) {
- //
- // Resolve redirects by querying the pagelinks table, and repeat the process
- //
-
- // Create a new linkBatch object for the next pass
- $linkBatch = new LinkBatch();
-
- // find redirect targets for all redirect pages
- $this->profileDBIn();
- $res = $db->select('pagelinks', array (
- 'pl_from',
- 'pl_namespace',
- 'pl_title'
- ), array (
- 'pl_from' => array_keys($redirectIds
- )), __CLASS__ . '::' . __FUNCTION__);
- $this->profileDBOut();
-
- while ($row = $db->fetchObject($res)) {
-
- // Bug 7304 workaround
- // ( http://bugzilla.wikipedia.org/show_bug.cgi?id=7304 )
- // A redirect page may have more than one link.
- // This code will only use the first link returned.
- if (isset ($redirectIds[$row->pl_from])) { // remove line when 7304 is fixed
-
- $titleStrFrom = $redirectIds[$row->pl_from]->getPrefixedText();
- $titleStrTo = Title :: makeTitle($row->pl_namespace, $row->pl_title)->getPrefixedText();
- $this->mRedirectTitles[$titleStrFrom] = $titleStrTo;
-
- unset ($redirectIds[$row->pl_from]); // remove line when 7304 is fixed
-
- // Avoid an infinite loop by checking if we have already processed this target
- if (!isset ($this->mAllPages[$row->pl_namespace][$row->pl_title])) {
- $linkBatch->add($row->pl_namespace, $row->pl_title);
- }
+ if(empty($revids))
+ return;
+
+ $db = $this->getDB();
+ $pageids = array();
+ $remaining = array_flip($revids);
+
+ $tables = array('revision');
+ $fields = array('rev_id','rev_page');
+ $where = array('rev_deleted' => 0, 'rev_id' => $revids);
+
+ // Get pageIDs data from the `page` table
+ $this->profileDBIn();
+ $res = $db->select( $tables, $fields, $where, __METHOD__ );
+ while ( $row = $db->fetchObject( $res ) ) {
+ $revid = intval($row->rev_id);
+ $pageid = intval($row->rev_page);
+ $this->mGoodRevIDs[$revid] = $pageid;
+ $pageids[$pageid] = '';
+ unset($remaining[$revid]);
+ }
+ $db->freeResult( $res );
+ $this->profileDBOut();
+
+ $this->mMissingRevIDs = array_keys($remaining);
+
+ // Populate all the page information
+ if($this->mResolveRedirects)
+ ApiBase :: dieDebug(__METHOD__, 'revids may not be used with redirect resolution');
+ $this->initFromPageIds(array_keys($pageids));
+ }
+
+ private function resolvePendingRedirects() {
+
+ if($this->mResolveRedirects) {
+ $db = $this->getDB();
+ $pageFlds = $this->getPageTableFields();
+
+ // Repeat until all redirects have been resolved
+ // The infinite loop is prevented by keeping all known pages in $this->mAllPages
+ while (!empty ($this->mPendingRedirectIDs)) {
+
+ // Resolve redirects by querying the pagelinks table, and repeat the process
+ // Create a new linkBatch object for the next pass
+ $linkBatch = $this->getRedirectTargets();
+
+ if ($linkBatch->isEmpty())
+ break;
+
+ $set = $linkBatch->constructSet('page', $db);
+ if(false === $set)
+ break;
+
+ // Get pageIDs data from the `page` table
+ $this->profileDBIn();
+ $res = $db->select('page', $pageFlds, $set, __METHOD__);
+ $this->profileDBOut();
+
+ // Hack: get the ns:titles stored in array(ns => array(titles)) format
+ $this->initFromQueryResult($db, $res, $linkBatch->data, true);
+ }
+ }
+ }
+
+ private function getRedirectTargets() {
+
+ $linkBatch = new LinkBatch();
+ $db = $this->getDB();
+
+ // find redirect targets for all redirect pages
+ $this->profileDBIn();
+ $res = $db->select('pagelinks', array (
+ 'pl_from',
+ 'pl_namespace',
+ 'pl_title'
+ ), array (
+ 'pl_from' => array_keys($this->mPendingRedirectIDs
+ )), __METHOD__);
+ $this->profileDBOut();
+
+ while ($row = $db->fetchObject($res)) {
+
+ $plfrom = intval($row->pl_from);
+
+ // Bug 7304 workaround
+ // ( http://bugzilla.wikipedia.org/show_bug.cgi?id=7304 )
+ // A redirect page may have more than one link.
+ // This code will only use the first link returned.
+ if (isset ($this->mPendingRedirectIDs[$plfrom])) { // remove line when bug 7304 is fixed
+
+ $titleStrFrom = $this->mPendingRedirectIDs[$plfrom]->getPrefixedText();
+ $titleStrTo = Title :: makeTitle($row->pl_namespace, $row->pl_title)->getPrefixedText();
+ unset ($this->mPendingRedirectIDs[$plfrom]); // remove line when bug 7304 is fixed
+
+ // Avoid an infinite loop by checking if we have already processed this target
+ if (!isset ($this->mAllPages[$row->pl_namespace][$row->pl_title])) {
+ $linkBatch->add($row->pl_namespace, $row->pl_title);
+ }
+ } else {
+ // This redirect page has more than one link.
+ // This is very slow, but safer until bug 7304 is resolved
+ $title = Title :: newFromID($plfrom);
+ $titleStrFrom = $title->getPrefixedText();
+
+ $article = new Article($title);
+ $text = $article->getContent();
+ $titleTo = Title :: newFromRedirect($text);
+ $titleStrTo = $titleTo->getPrefixedText();
+
+ if (is_null($titleStrTo))
+ ApiBase :: dieDebug(__METHOD__, 'Bug7304 workaround: redir target from {$title->getPrefixedText()} not found');
+
+ // Avoid an infinite loop by checking if we have already processed this target
+ if (!isset ($this->mAllPages[$titleTo->getNamespace()][$titleTo->getDBkey()])) {
+ $linkBatch->addObj($titleTo);
}
}
- $db->freeResult($res);
+
+ $this->mRedirectTitles[$titleStrFrom] = $titleStrTo;
}
+ $db->freeResult($res);
+
+ // All IDs must exist in the page table
+ if (!empty($this->mPendingRedirectIDs[$plfrom]))
+ ApiBase :: dieDebug(__METHOD__, 'Invalid redirect IDs were found');
+
+ return $linkBatch;
}
/**
* Given an array of title strings, convert them into Title objects.
+ * Alternativelly, an array of Title objects may be given.
* This method validates access rights for the title,
* and appends normalization values to the output.
*
* @return LinkBatch of title objects.
*/
- private function processTitlesStrArray($titles) {
+ private function processTitlesArray($titles) {
$linkBatch = new LinkBatch();
- foreach ($titles as $titleString) {
- $titleObj = Title :: newFromText($titleString);
-
- // Validation
+ foreach ($titles as $title) {
+
+ $titleObj = is_string($title) ? Title :: newFromText($title) : $title;
if (!$titleObj)
- $this->dieUsage("bad title $titleString", 'invalidtitle');
- if ($titleObj->getNamespace() < 0)
- $this->dieUsage("No support for special page $titleString has been implemented", 'unsupportednamespace');
- if (!$titleObj->userCanRead())
- $this->dieUsage("No read permission for $titleString", 'titleaccessdenied');
+ $this->dieUsage("bad title", 'invalidtitle');
+
+ $iw = $titleObj->getInterwiki();
+ if (!empty($iw)) {
+ // This title is an interwiki link.
+ $this->mInterwikiTitles[$titleObj->getPrefixedText()] = $iw;
+ } else {
- $linkBatch->addObj($titleObj);
+ // Validation
+ if ($titleObj->getNamespace() < 0)
+ $this->dieUsage("No support for special pages has been implemented", 'unsupportednamespace');
+ $linkBatch->addObj($titleObj);
+ }
+
// Make sure we remember the original title that was given to us
// This way the caller can correlate new titles with the originally requested,
// i.e. namespace is localized or capitalization is different
- if ($titleString !== $titleObj->getPrefixedText()) {
- $this->mNormalizedTitles[$titleString] = $titleObj->getPrefixedText();
+ if (is_string($title) && $title !== $titleObj->getPrefixedText()) {
+ $this->mNormalizedTitles[$title] = $titleObj->getPrefixedText();
}
}
return $linkBatch;
}
-
- private function populatePageIDs($pageids) {
- $this->dieUsage(__FUNCTION__ . " is not implemented", 'notimplemented');
- }
-
- public function execute() {
- $titles = $pageids = $revids = $redirects = null;
- extract($this->extractRequestParams());
-
- // Only one of the titles/pageids/revids is allowed at the same time
- $dataSource = null;
- if (isset ($titles))
- $dataSource = 'titles';
- if (isset ($pageids)) {
- if (isset ($dataSource))
- $this->dieUsage("Cannot use 'pageids' at the same time as '$dataSource'", 'multisource');
- $dataSource = 'pageids';
- }
- if (isset ($revids)) {
- if (isset ($dataSource))
- $this->dieUsage("Cannot use 'revids' at the same time as '$dataSource'", 'multisource');
- $dataSource = 'revids';
- }
-
- switch ($dataSource) {
- case 'titles' :
- $this->populateTitles($titles, $redirects);
- break;
- case 'pageids' :
- $this->populatePageIDs($pageids, $redirects);
- break;
- case 'revids' :
- $this->populateRevIDs($revids);
- break;
- default :
- // Do nothing - some queries do not need any of the data sources.
- break;
- }
- }
-
+
protected function getAllowedParams() {
return array (
'titles' => array (
- GN_ENUM_ISMULTI => true
+ ApiBase :: PARAM_ISMULTI => true
),
'pageids' => array (
- GN_ENUM_TYPE => 'integer',
- GN_ENUM_ISMULTI => true
+ ApiBase :: PARAM_TYPE => 'integer',
+ ApiBase :: PARAM_ISMULTI => true
),
'revids' => array (
- GN_ENUM_TYPE => 'integer',
- GN_ENUM_ISMULTI => true
- ),
- 'redirects' => false
+ ApiBase :: PARAM_TYPE => 'integer',
+ ApiBase :: PARAM_ISMULTI => true
+ )
);
}
-
+
protected function getParamDescription() {
return array (
'titles' => 'A list of titles to work on',
'pageids' => 'A list of page IDs to work on',
- 'revids' => 'A list of revision IDs to work on',
- 'redirects' => 'Automatically resolve redirects'
+ 'revids' => 'A list of revision IDs to work on'
);
}
+
+ public function getVersion() {
+ return __CLASS__ . ': $Id$';
+ }
}
-?>
\ No newline at end of file
+
dépôts / lhc / web / wiklou.git / blobdiff