nextcloud/lib/private/files/cache/scanner.php

331 lines
9.9 KiB
PHP
Raw Normal View History

<?php
/**
* Copyright (c) 2012 Robin Appelman <icewind@owncloud.com>
* This file is licensed under the Affero General Public License version 3 or
* later.
* See the COPYING-README file.
*/
namespace OC\Files\Cache;
use OC\Files\Filesystem;
2013-07-19 18:32:43 +04:00
use OC\Hooks\BasicEmitter;
use OCP\Config;
2013-07-19 18:32:43 +04:00
/**
* Class Scanner
*
* Hooks available in scope \OC\Files\Cache\Scanner:
* - scanFile(string $path, string $storageId)
* - scanFolder(string $path, string $storageId)
2013-11-07 19:22:29 +04:00
* - postScanFile(string $path, string $storageId)
* - postScanFolder(string $path, string $storageId)
2013-07-19 18:32:43 +04:00
*
* @package OC\Files\Cache
*/
class Scanner extends BasicEmitter {
/**
* @var \OC\Files\Storage\Storage $storage
*/
protected $storage;
/**
* @var string $storageId
*/
protected $storageId;
/**
* @var \OC\Files\Cache\Cache $cache
*/
protected $cache;
/**
* @var boolean $cacheActive If true, perform cache operations, if false, do not affect cache
*/
protected $cacheActive;
const SCAN_RECURSIVE = true;
const SCAN_SHALLOW = false;
const REUSE_ETAG = 1;
const REUSE_SIZE = 2;
public function __construct(\OC\Files\Storage\Storage $storage) {
$this->storage = $storage;
$this->storageId = $this->storage->getId();
$this->cache = $storage->getCache();
$this->cacheActive = !Config::getSystemValue('filesystem_cache_readonly', false);
}
/**
* get all the metadata of a file or folder
* *
*
* @param string $path
2014-05-11 21:13:51 +04:00
* @return array an array of metadata of the file
*/
public function getData($path) {
2013-10-24 16:24:56 +04:00
if (!$this->storage->isReadable($path)) {
//cant read, nothing we can do
2013-11-04 15:58:30 +04:00
\OCP\Util::writeLog('OC\Files\Cache\Scanner', "!!! Path '$path' is not readable !!!", \OCP\Util::DEBUG);
2013-10-24 16:24:56 +04:00
return null;
}
$data = array();
$data['mimetype'] = $this->storage->getMimeType($path);
$data['mtime'] = $this->storage->filemtime($path);
if ($data['mimetype'] == 'httpd/unix-directory') {
$data['size'] = -1; //unknown
} else {
$data['size'] = $this->storage->filesize($path);
}
2012-12-31 01:32:55 +04:00
$data['etag'] = $this->storage->getETag($path);
2013-02-10 15:27:35 +04:00
$data['storage_mtime'] = $data['mtime'];
$data['permissions'] = $this->storage->getPermissions($path);
return $data;
}
/**
* scan a single file and store it in the cache
*
* @param string $file
* @param int $reuseExisting
2014-05-11 21:13:51 +04:00
* @return array an array of metadata of the scanned file
*/
2014-06-10 17:42:37 +04:00
public function scanFile($file, $reuseExisting = 0) {
if (!self::isPartialFile($file)
and !Filesystem::isFileBlacklisted($file)
) {
2013-07-19 18:32:43 +04:00
$this->emit('\OC\Files\Cache\Scanner', 'scanFile', array($file, $this->storageId));
\OC_Hook::emit('\OC\Files\Cache\Scanner', 'scan_file', array('path' => $file, 'storage' => $this->storageId));
$data = $this->getData($file);
if ($data) {
$parent = dirname($file);
if ($parent === '.' or $parent === '/') {
$parent = '';
}
$parentId = $this->cache->getId($parent);
2014-06-10 17:37:43 +04:00
// scan the parent if it's not in the cache (id -1) and the current file is not the root folder
if ($file and $parentId === -1) {
$parentData = $this->scanFile($parent);
$parentId = $parentData['fileid'];
}
if ($parent) {
$data['parent'] = $parentId;
}
$cacheData = $this->cache->get($file);
if ($cacheData and $reuseExisting) {
// prevent empty etag
if (empty($cacheData['etag'])) {
$etag = $data['etag'];
} else {
$etag = $cacheData['etag'];
}
// only reuse data if the file hasn't explicitly changed
if (isset($data['storage_mtime']) && isset($cacheData['storage_mtime']) && $data['storage_mtime'] === $cacheData['storage_mtime']) {
$data['mtime'] = $cacheData['mtime'];
if (($reuseExisting & self::REUSE_SIZE) && ($data['size'] === -1)) {
$data['size'] = $cacheData['size'];
}
if ($reuseExisting & self::REUSE_ETAG) {
$data['etag'] = $etag;
2013-10-24 16:24:56 +04:00
}
}
// Only update metadata that has changed
$newData = array_diff_assoc($data, $cacheData);
if (isset($newData['etag'])) {
$cacheDataString = print_r($cacheData, true);
$dataString = print_r($data, true);
\OCP\Util::writeLog('OC\Files\Cache\Scanner',
"!!! No reuse of etag for '$file' !!! \ncache: $cacheDataString \ndata: $dataString",
\OCP\Util::DEBUG);
}
} else {
$newData = $data;
2013-05-29 18:19:03 +04:00
}
if (!empty($newData)) {
$data['fileid'] = $this->addToCache($file, $newData);
2013-11-07 19:22:29 +04:00
$this->emit('\OC\Files\Cache\Scanner', 'postScanFile', array($file, $this->storageId));
\OC_Hook::emit('\OC\Files\Cache\Scanner', 'post_scan_file', array('path' => $file, 'storage' => $this->storageId));
}
2013-08-06 17:59:06 +04:00
} else {
$this->removeFromCache($file);
}
return $data;
}
return null;
}
protected function removeFromCache($path) {
\OC_Hook::emit('Scanner', 'removeFromCache', array('file' => $path));
$this->emit('\OC\Files\Cache\Scanner', 'removeFromCache', array($path));
if ($this->cacheActive) {
$this->cache->remove($path);
}
}
/**
* @param string $path
* @param array $data
* @return int the id of the added file
*/
protected function addToCache($path, $data) {
\OC_Hook::emit('Scanner', 'addToCache', array('file' => $path, 'data' => $data));
$this->emit('\OC\Files\Cache\Scanner', 'addToCache', array($path, $this->storageId, $data));
if ($this->cacheActive) {
return $this->cache->put($path, $data);
} else {
return -1;
}
}
/**
* @param string $path
* @param array $data
*/
protected function updateCache($path, $data) {
\OC_Hook::emit('Scanner', 'addToCache', array('file' => $path, 'data' => $data));
2014-06-02 17:17:00 +04:00
$this->emit('\OC\Files\Cache\Scanner', 'updateCache', array($path, $this->storageId, $data));
if ($this->cacheActive) {
$this->cache->put($path, $data);
}
}
/**
* scan a folder and all it's children
*
* @param string $path
2013-03-24 05:06:50 +04:00
* @param bool $recursive
* @param int $reuse
2014-05-11 21:13:51 +04:00
* @return array an array of the meta data of the scanned file or folder
*/
public function scan($path, $recursive = self::SCAN_RECURSIVE, $reuse = -1) {
if ($reuse === -1) {
$reuse = ($recursive === self::SCAN_SHALLOW) ? self::REUSE_ETAG | self::REUSE_SIZE : 0;
}
$data = $this->scanFile($path, $reuse);
$size = $this->scanChildren($path, $recursive, $reuse);
$data['size'] = $size;
return $data;
}
/**
* scan all the files and folders in a folder
*
* @param string $path
* @param bool $recursive
* @param int $reuse
* @return int the size of the scanned folder or -1 if the size is unknown at this stage
*/
public function scanChildren($path, $recursive = self::SCAN_RECURSIVE, $reuse = -1) {
if ($reuse === -1) {
$reuse = ($recursive === self::SCAN_SHALLOW) ? self::REUSE_ETAG | self::REUSE_SIZE : 0;
}
2013-07-19 18:32:43 +04:00
$this->emit('\OC\Files\Cache\Scanner', 'scanFolder', array($path, $this->storageId));
$size = 0;
$childQueue = array();
$existingChildren = array();
if ($this->cache->inCache($path)) {
$children = $this->cache->getFolderContents($path);
foreach ($children as $child) {
$existingChildren[] = $child['name'];
}
}
$newChildren = array();
if ($this->storage->is_dir($path) && ($dh = $this->storage->opendir($path))) {
$exceptionOccurred = false;
\OC_DB::beginTransaction();
if (is_resource($dh)) {
while (($file = readdir($dh)) !== false) {
$child = ($path) ? $path . '/' . $file : $file;
if (!Filesystem::isIgnoredDir($file)) {
$newChildren[] = $file;
try {
2014-06-10 17:42:37 +04:00
$data = $this->scanFile($child, $reuse);
if ($data) {
2014-06-02 23:04:36 +04:00
if ($data['mimetype'] === 'httpd/unix-directory' and $recursive === self::SCAN_RECURSIVE) {
$childQueue[] = $child;
} else if ($data['size'] === -1) {
$size = -1;
} else if ($size !== -1) {
$size += $data['size'];
}
}
} catch (\Doctrine\DBAL\DBALException $ex) {
// might happen if inserting duplicate while a scanning
// process is running in parallel
// log and ignore
\OC_Log::write('core', 'Exception while scanning file "' . $child . '": ' . $ex->getMessage(), \OC_Log::DEBUG);
$exceptionOccurred = true;
}
}
}
}
$removedChildren = \array_diff($existingChildren, $newChildren);
foreach ($removedChildren as $childName) {
$child = ($path) ? $path . '/' . $childName : $childName;
$this->removeFromCache($child);
}
\OC_DB::commit();
if ($exceptionOccurred) {
// It might happen that the parallel scan process has already
// inserted mimetypes but those weren't available yet inside the transaction
// To make sure to have the updated mime types in such cases,
// we reload them here
$this->cache->loadMimetypes();
}
foreach ($childQueue as $child) {
$childSize = $this->scanChildren($child, self::SCAN_RECURSIVE, $reuse);
if ($childSize === -1) {
$size = -1;
} else if ($size !== -1) {
$size += $childSize;
}
}
$this->updateCache($path, array('size' => $size));
}
2013-11-07 19:22:29 +04:00
$this->emit('\OC\Files\Cache\Scanner', 'postScanFolder', array($path, $this->storageId));
return $size;
}
2013-02-10 17:16:45 +04:00
/**
* check if the file should be ignored when scanning
2013-02-10 17:16:45 +04:00
* NOTE: files with a '.part' extension are ignored as well!
* prevents unfinished put requests to be scanned
*
* @param string $file
2013-02-10 17:16:45 +04:00
* @return boolean
*/
public static function isPartialFile($file) {
if (pathinfo($file, PATHINFO_EXTENSION) === 'part') {
2013-02-10 17:16:45 +04:00
return true;
}
return false;
}
2012-11-22 02:18:58 +04:00
/**
* walk over any folders that are not fully scanned yet and scan them
*/
public function backgroundScan() {
$lastPath = null;
while (($path = $this->cache->getIncomplete()) !== false && $path !== $lastPath) {
$this->scan($path, self::SCAN_RECURSIVE, self::REUSE_ETAG);
\OC_Hook::emit('Scanner', 'correctFolderSize', array('path' => $path));
if ($this->cacheActive) {
$this->cache->correctFolderSize($path);
}
$lastPath = $path;
2012-11-22 02:18:58 +04:00
}
}
/**
* Set whether the cache is affected by scan operations
*
* @param boolean $active The active state of the cache
*/
public function setCacheActive($active) {
$this->cacheActive = $active;
}
}