1 files changed, 94 insertions, 49 deletions

diff --git a/includes/api/ApiQueryDuplicateFiles.php b/includes/api/ApiQueryDuplicateFiles.php
index beca5879..8f0fd3be 100644
--- a/includes/api/ApiQueryDuplicateFiles.php
+++ b/includes/api/ApiQueryDuplicateFiles.php
@@ -4,7 +4,7 @@
  *
  * Created on Sep 27, 2008
  *
- * Copyright © 2008 Roan Kattouw <Firstname>.<Lastname>@gmail.com
+ * 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
@@ -59,67 +59,99 @@ class ApiQueryDuplicateFiles extends ApiQueryGeneratorBase {
 		}
 		$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' );
 			}
-			$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')"
-			);
+			$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;
+				}
+			}
 		}
 
-		$dir = ( $params['dir'] == 'descending' ? ' DESC' : '' );
-		$this->addOption( 'ORDER BY', 'i1.img_name' . $dir );
-		$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;
+
+		$sha1s = array();
+		foreach ( $files as $file ) {
+			$sha1s[$file->getName()] = $file->getSha1();
+		}
+
+		// find all files with the hashes, result format is: array( hash => array( 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
+			}
+			$sha1 = $sha1s[$image];
+			$dupFiles = $filesBySha1s[$sha1];
+			if( $params['dir'] == 'descending' ) {
+				$dupFiles = array_reverse( $dupFiles );
 			}
-			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 ) );
+			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[] = $file->getTitle();
+				} else {
+					$r = array(
+						'name' => $dupName,
+						'user' => $dupFile->getUser( 'text' ),
+						'timestamp' => wfTimestamp( TS_ISO_8601, $dupFile->getTimestamp() )
+					);
+					if( !$dupFile->isLocal() ) {
+						$r['shared'] = '';
+					}
+					$fit = $this->addPageSubItem( $pageId, $r );
+					if ( !$fit ) {
+						$this->setContinueEnumParameter( 'continue', $image . '|' . $dupName );
+						break;
+					}
+				}
+			}
+			if( !$fit ) {
+				break;
 			}
 		}
 		if ( !is_null( $resultPageSet ) ) {
@@ -144,19 +176,32 @@ class ApiQueryDuplicateFiles extends ApiQueryGeneratorBase {
 					'descending'
 				)
 			),
+			'localonly' => false,
 		);
 	}
 
 	public function getParamDescription() {
 		return array(
-			'limit' => 'How many files to return',
+			'limit' => 'How many duplicate files to return',
 			'continue' => 'When more results are available, use this to continue',
 			'dir' => 'The direction in which to list',
+			'localonly' => 'Look only for files in the local repository',
+		);
+	}
+
+	public function getResultProperties() {
+		return array(
+			'' => array(
+				'name' => 'string',
+				'user' => 'string',
+				'timestamp' => 'timestamp',
+				'shared' => 'boolean',
+			)
 		);
 	}
 
 	public function getDescription() {
-		return 'List all files that are duplicates of the given file(s)';
+		return 'List all files that are duplicates of the given file(s) based on hash values';
 	}
 
 	public function getPossibleErrors() {