<?php
/**
- * API for MediaWiki 1.8+
+ *
*
* Created on Sep 27, 2008
*
- * Copyright © 2008 Roan Kattow <Firstname>,<Lastname>@home.nl
+ * Copyright © 2008 Roan Kattouw "<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
* @file
*/
-if ( !defined( 'MEDIAWIKI' ) ) {
- // Eclipse helper - will be ignored in production
- require_once( "ApiQueryBase.php" );
-}
-
/**
* A query module to list duplicates of the given file(s)
*
*/
class ApiQueryDuplicateFiles extends ApiQueryGeneratorBase {
- public function __construct( $query, $moduleName ) {
+ public function __construct( ApiQuery $query, $moduleName ) {
parent::__construct( $query, $moduleName, 'df' );
}
$this->run( $resultPageSet );
}
+ /**
+ * @param ApiPageSet $resultPageSet
+ */
private function run( $resultPageSet = null ) {
$params = $this->extractRequestParams();
- $namespaces = $this->getPageSet()->getAllTitlesByNamespace();
+ $namespaces = $this->getPageSet()->getGoodAndMissingTitlesByNamespace();
if ( empty( $namespaces[NS_FILE] ) ) {
return;
}
$images = $namespaces[NS_FILE];
- $this->addTables( 'image', 'i1' );
- $this->addTables( 'image', 'i2' );
- $this->addFields( array(
- 'i1.img_name AS orig_name',
- 'i2.img_name AS dup_name',
- 'i2.img_user_text AS dup_user_text',
- 'i2.img_timestamp AS dup_timestamp'
- ) );
-
- $this->addWhere( array(
- 'i1.img_name' => array_keys( $images ),
- 'i1.img_sha1 = i2.img_sha1',
- 'i1.img_name != i2.img_name',
- ) );
+ if ( $params['dir'] == 'descending' ) {
+ $images = array_reverse( $images );
+ }
+ $skipUntilThisDup = false;
if ( isset( $params['continue'] ) ) {
$cont = explode( '|', $params['continue'] );
- if ( count( $cont ) != 2 ) {
- $this->dieUsage( 'Invalid continue param. You should pass the ' .
- 'original value returned by the previous query', '_badcontinue' );
+ $this->dieContinueUsageIf( count( $cont ) != 2 );
+ $fromImage = $cont[0];
+ $skipUntilThisDup = $cont[1];
+ // Filter out any images before $fromImage
+ foreach ( $images as $image => $pageId ) {
+ if ( $image < $fromImage ) {
+ unset( $images[$image] );
+ } else {
+ break;
+ }
}
- $orig = $this->getDB()->strencode( $this->titleTokey( $cont[0] ) );
- $dup = $this->getDB()->strencode( $this->titleToKey( $cont[1] ) );
- $this->addWhere(
- "i1.img_name > '$orig' OR " .
- "(i1.img_name = '$orig' AND " .
- "i2.img_name >= '$dup')"
- );
}
- $this->addOption( 'ORDER BY', 'i1.img_name' );
- $this->addOption( 'LIMIT', $params['limit'] + 1 );
+ $filesToFind = array_keys( $images );
+ if ( $params['localonly'] ) {
+ $files = RepoGroup::singleton()->getLocalRepo()->findFiles( $filesToFind );
+ } else {
+ $files = RepoGroup::singleton()->findFiles( $filesToFind );
+ }
- $res = $this->select( __METHOD__ );
+ $fit = true;
$count = 0;
- $titles = array();
- foreach ( $res as $row ) {
- if ( ++$count > $params['limit'] ) {
- // We've reached the one extra which shows that
- // there are additional pages to be had. Stop here...
- $this->setContinueEnumParameter( 'continue',
- $this->keyToTitle( $row->orig_name ) . '|' .
- $this->keyToTitle( $row->dup_name ) );
- break;
+ $titles = [];
+
+ $sha1s = [];
+ foreach ( $files as $file ) {
+ /** @var File $file */
+ $sha1s[$file->getName()] = $file->getSha1();
+ }
+
+ // find all files with the hashes, result format is:
+ // [ hash => [ dup1, dup2 ], hash1 => ... ]
+ $filesToFindBySha1s = array_unique( array_values( $sha1s ) );
+ if ( $params['localonly'] ) {
+ $filesBySha1s = RepoGroup::singleton()->getLocalRepo()->findBySha1s( $filesToFindBySha1s );
+ } else {
+ $filesBySha1s = RepoGroup::singleton()->findBySha1s( $filesToFindBySha1s );
+ }
+
+ // iterate over $images to handle continue param correct
+ foreach ( $images as $image => $pageId ) {
+ if ( !isset( $sha1s[$image] ) ) {
+ continue; // file does not exist
}
- if ( !is_null( $resultPageSet ) ) {
- $titles[] = Title::makeTitle( NS_FILE, $row->dup_name );
- } else {
- $r = array(
- 'name' => $row->dup_name,
- 'user' => $row->dup_user_text,
- 'timestamp' => wfTimestamp( TS_ISO_8601, $row->dup_timestamp )
- );
- $fit = $this->addPageSubItem( $images[$row->orig_name], $r );
- if ( !$fit ) {
- $this->setContinueEnumParameter( 'continue',
- $this->keyToTitle( $row->orig_name ) . '|' .
- $this->keyToTitle( $row->dup_name ) );
+ $sha1 = $sha1s[$image];
+ $dupFiles = $filesBySha1s[$sha1];
+ if ( $params['dir'] == 'descending' ) {
+ $dupFiles = array_reverse( $dupFiles );
+ }
+ /** @var File $dupFile */
+ foreach ( $dupFiles as $dupFile ) {
+ $dupName = $dupFile->getName();
+ if ( $image == $dupName && $dupFile->isLocal() ) {
+ continue; // ignore the local file itself
+ }
+ if ( $skipUntilThisDup !== false && $dupName < $skipUntilThisDup ) {
+ continue; // skip to pos after the image from continue param
+ }
+ $skipUntilThisDup = false;
+ if ( ++$count > $params['limit'] ) {
+ $fit = false; // break outer loop
+ // We're one over limit which shows that
+ // there are additional images to be had. Stop here...
+ $this->setContinueEnumParameter( 'continue', $image . '|' . $dupName );
break;
}
+ if ( !is_null( $resultPageSet ) ) {
+ $titles[] = $dupFile->getTitle();
+ } else {
+ $r = [
+ 'name' => $dupName,
+ 'user' => $dupFile->getUser( 'text' ),
+ 'timestamp' => wfTimestamp( TS_ISO_8601, $dupFile->getTimestamp() ),
+ 'shared' => !$dupFile->isLocal(),
+ ];
+ $fit = $this->addPageSubItem( $pageId, $r );
+ if ( !$fit ) {
+ $this->setContinueEnumParameter( 'continue', $image . '|' . $dupName );
+ break;
+ }
+ }
+ }
+ if ( !$fit ) {
+ break;
}
}
if ( !is_null( $resultPageSet ) ) {
}
public function getAllowedParams() {
- return array(
- 'limit' => array(
+ return [
+ 'limit' => [
ApiBase::PARAM_DFLT => 10,
ApiBase::PARAM_TYPE => 'limit',
ApiBase::PARAM_MIN => 1,
ApiBase::PARAM_MAX => ApiBase::LIMIT_BIG1,
ApiBase::PARAM_MAX2 => ApiBase::LIMIT_BIG2
- ),
- 'continue' => null,
- );
- }
-
- public function getParamDescription() {
- return array(
- 'limit' => 'How many files to return',
- 'continue' => 'When more results are available, use this to continue',
- );
- }
-
- public function getDescription() {
- return 'List all files that are duplicates of the given file(s)';
- }
-
- public function getPossibleErrors() {
- return array_merge( parent::getPossibleErrors(), array(
- array( 'code' => '_badcontinue', 'info' => 'Invalid continue param. You should pass the original value returned by the previous query' ),
- ) );
+ ],
+ 'continue' => [
+ ApiBase::PARAM_HELP_MSG => 'api-help-param-continue',
+ ],
+ 'dir' => [
+ ApiBase::PARAM_DFLT => 'ascending',
+ ApiBase::PARAM_TYPE => [
+ 'ascending',
+ 'descending'
+ ]
+ ],
+ 'localonly' => false,
+ ];
}
- protected function getExamples() {
- return array(
- 'api.php?action=query&titles=File:Albert_Einstein_Head.jpg&prop=duplicatefiles',
- 'api.php?action=query&generator=allimages&prop=duplicatefiles',
- );
+ protected function getExamplesMessages() {
+ return [
+ 'action=query&titles=File:Albert_Einstein_Head.jpg&prop=duplicatefiles'
+ => 'apihelp-query+duplicatefiles-example-simple',
+ 'action=query&generator=allimages&prop=duplicatefiles'
+ => 'apihelp-query+duplicatefiles-example-generated',
+ ];
}
- public function getVersion() {
- return __CLASS__ . ': $Id$';
+ public function getHelpUrls() {
+ return 'https://www.mediawiki.org/wiki/Special:MyLanguage/API:Duplicatefiles';
}
}