Blame view
sources/lib/private/files/cache/scanner.php
9.94 KB
|
03e52840d
|
1 2 3 4 5 6 7 8 9 10 11 12 |
<?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; use OC\Hooks\BasicEmitter; |
|
6d9380f96
|
13 |
use OCP\Config; |
|
03e52840d
|
14 15 16 17 18 19 20 |
/** * Class Scanner * * Hooks available in scope \OC\Files\Cache\Scanner: * - scanFile(string $path, string $storageId) * - scanFolder(string $path, string $storageId) |
|
31b7f2792
|
21 22 |
* - postScanFile(string $path, string $storageId) * - postScanFolder(string $path, string $storageId) |
|
03e52840d
|
23 24 25 26 27 28 29 |
*
* @package OC\Files\Cache
*/
class Scanner extends BasicEmitter {
/**
* @var \OC\Files\Storage\Storage $storage
*/
|
|
6d9380f96
|
30 |
protected $storage; |
|
03e52840d
|
31 32 33 34 |
/** * @var string $storageId */ |
|
6d9380f96
|
35 |
protected $storageId; |
|
03e52840d
|
36 37 38 39 |
/** * @var \OC\Files\Cache\Cache $cache */ |
|
6d9380f96
|
40 |
protected $cache; |
|
03e52840d
|
41 |
|
|
31b7f2792
|
42 |
/** |
|
6d9380f96
|
43 |
* @var boolean $cacheActive If true, perform cache operations, if false, do not affect cache |
|
31b7f2792
|
44 |
*/ |
|
6d9380f96
|
45 |
protected $cacheActive; |
|
31b7f2792
|
46 |
|
|
03e52840d
|
47 48 49 50 51 52 53 54 55 56 |
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();
|
|
6d9380f96
|
57 |
$this->cacheActive = !Config::getSystemValue('filesystem_cache_readonly', false);
|
|
03e52840d
|
58 59 60 61 62 63 64 |
} /** * get all the metadata of a file or folder * * * * @param string $path |
|
6d9380f96
|
65 |
* @return array an array of metadata of the file |
|
03e52840d
|
66 67 |
*/
public function getData($path) {
|
|
31b7f2792
|
68 69 70 71 72 |
if (!$this->storage->isReadable($path)) {
//cant read, nothing we can do
\OCP\Util::writeLog('OC\Files\Cache\Scanner', "!!! Path '$path' is not readable !!!", \OCP\Util::DEBUG);
return null;
}
|
|
03e52840d
|
73 |
$data = array(); |
|
03e52840d
|
74 75 76 77 78 79 80 81 |
$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);
}
$data['etag'] = $this->storage->getETag($path);
|
|
31b7f2792
|
82 |
$data['storage_mtime'] = $data['mtime']; |
|
6d9380f96
|
83 |
$data['permissions'] = $this->storage->getPermissions($path); |
|
03e52840d
|
84 85 86 87 88 89 90 91 |
return $data; } /** * scan a single file and store it in the cache * * @param string $file * @param int $reuseExisting |
|
6d9380f96
|
92 |
* @return array an array of metadata of the scanned file |
|
03e52840d
|
93 |
*/ |
|
6d9380f96
|
94 |
public function scanFile($file, $reuseExisting = 0) {
|
|
03e52840d
|
95 96 97 98 99 100 101 |
if (!self::isPartialFile($file)
and !Filesystem::isFileBlacklisted($file)
) {
$this->emit('\OC\Files\Cache\Scanner', 'scanFile', array($file, $this->storageId));
\OC_Hook::emit('\OC\Files\Cache\Scanner', 'scan_file', array('path' => $file, 'storage' => $this->storageId));
$data = $this->getData($file);
if ($data) {
|
|
6d9380f96
|
102 103 104 105 106 107 108 109 110 111 112 113 114 |
$parent = dirname($file);
if ($parent === '.' or $parent === '/') {
$parent = '';
}
$parentId = $this->cache->getId($parent);
// 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;
|
|
03e52840d
|
115 |
} |
|
31b7f2792
|
116 |
$cacheData = $this->cache->get($file); |
|
6d9380f96
|
117 118 119 120 121 |
if ($cacheData and $reuseExisting) {
// prevent empty etag
if (empty($cacheData['etag'])) {
$etag = $data['etag'];
} else {
|
|
31b7f2792
|
122 |
$etag = $cacheData['etag']; |
|
6d9380f96
|
123 124 125 126 127 128 |
}
// only reuse data if the file hasn't explicitly changed
if (isset($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'] = $cacheData['size'];
|
|
03e52840d
|
129 |
} |
|
6d9380f96
|
130 131 |
if ($reuseExisting & self::REUSE_ETAG) {
$data['etag'] = $etag;
|
|
03e52840d
|
132 133 |
} } |
|
6d9380f96
|
134 135 136 137 138 139 140 141 142 143 144 145 146 |
// 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' !!!
cache: $cacheDataString
data: $dataString",
\OCP\Util::DEBUG);
}
} else {
$newData = $data;
|
|
03e52840d
|
147 148 |
}
if (!empty($newData)) {
|
|
6d9380f96
|
149 |
$data['fileid'] = $this->addToCache($file, $newData); |
|
31b7f2792
|
150 151 |
$this->emit('\OC\Files\Cache\Scanner', 'postScanFile', array($file, $this->storageId));
\OC_Hook::emit('\OC\Files\Cache\Scanner', 'post_scan_file', array('path' => $file, 'storage' => $this->storageId));
|
|
03e52840d
|
152 153 |
}
} else {
|
|
6d9380f96
|
154 |
$this->removeFromCache($file); |
|
03e52840d
|
155 156 157 158 159 |
} return $data; } return null; } |
|
6d9380f96
|
160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 |
protected function removeFromCache($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));
$this->emit('\OC\Files\Cache\Scanner', 'updateCache', array($path, $this->storageId, $data));
if ($this->cacheActive) {
$this->cache->put($path, $data);
}
}
|
|
03e52840d
|
194 195 196 197 198 199 |
/** * scan a folder and all it's children * * @param string $path * @param bool $recursive * @param int $reuse |
|
6d9380f96
|
200 |
* @return array an array of the meta data of the scanned file or folder |
|
03e52840d
|
201 202 203 204 205 |
*/
public function scan($path, $recursive = self::SCAN_RECURSIVE, $reuse = -1) {
if ($reuse === -1) {
$reuse = ($recursive === self::SCAN_SHALLOW) ? self::REUSE_ETAG | self::REUSE_SIZE : 0;
}
|
|
6d9380f96
|
206 207 208 209 |
$data = $this->scanFile($path, $reuse); $size = $this->scanChildren($path, $recursive, $reuse); $data['size'] = $size; return $data; |
|
03e52840d
|
210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 |
}
/**
* 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;
}
$this->emit('\OC\Files\Cache\Scanner', 'scanFolder', array($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))) {
|
|
31b7f2792
|
236 |
$exceptionOccurred = false; |
|
03e52840d
|
237 |
\OC_DB::beginTransaction(); |
|
31b7f2792
|
238 |
if (is_resource($dh)) {
|
|
03e52840d
|
239 240 241 242 |
while (($file = readdir($dh)) !== false) {
$child = ($path) ? $path . '/' . $file : $file;
if (!Filesystem::isIgnoredDir($file)) {
$newChildren[] = $file;
|
|
31b7f2792
|
243 |
try {
|
|
6d9380f96
|
244 |
$data = $this->scanFile($child, $reuse); |
|
31b7f2792
|
245 |
if ($data) {
|
|
6d9380f96
|
246 247 248 249 |
if ($data['mimetype'] === 'httpd/unix-directory' and $recursive === self::SCAN_RECURSIVE) {
$childQueue[] = $child;
} else if ($data['size'] === -1) {
$size = -1;
|
|
31b7f2792
|
250 251 |
} else if ($size !== -1) {
$size += $data['size'];
|
|
03e52840d
|
252 |
} |
|
03e52840d
|
253 |
} |
|
6d9380f96
|
254 |
} catch (\Doctrine\DBAL\DBALException $ex) {
|
|
31b7f2792
|
255 256 257 258 259 260 |
// 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;
}
|
|
03e52840d
|
261 262 263 264 265 266 |
}
}
}
$removedChildren = \array_diff($existingChildren, $newChildren);
foreach ($removedChildren as $childName) {
$child = ($path) ? $path . '/' . $childName : $childName;
|
|
6d9380f96
|
267 |
$this->removeFromCache($child); |
|
03e52840d
|
268 269 |
} \OC_DB::commit(); |
|
6d9380f96
|
270 |
if ($exceptionOccurred) {
|
|
31b7f2792
|
271 272 273 274 275 276 |
// 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 $this->cache->loadMimetypes(); } |
|
03e52840d
|
277 278 279 280 |
foreach ($childQueue as $child) {
$childSize = $this->scanChildren($child, self::SCAN_RECURSIVE, $reuse);
if ($childSize === -1) {
$size = -1;
|
|
6d9380f96
|
281 |
} else if ($size !== -1) {
|
|
03e52840d
|
282 283 284 |
$size += $childSize; } } |
|
6d9380f96
|
285 |
$this->updateCache($path, array('size' => $size));
|
|
03e52840d
|
286 |
} |
|
31b7f2792
|
287 |
$this->emit('\OC\Files\Cache\Scanner', 'postScanFolder', array($path, $this->storageId));
|
|
03e52840d
|
288 289 290 291 |
return $size; } /** |
|
6d9380f96
|
292 |
* check if the file should be ignored when scanning |
|
03e52840d
|
293 294 |
* NOTE: files with a '.part' extension are ignored as well! * prevents unfinished put requests to be scanned |
|
6d9380f96
|
295 296 |
* * @param string $file |
|
03e52840d
|
297 298 299 300 301 302 303 304 305 306 307 308 309 |
* @return boolean
*/
public static function isPartialFile($file) {
if (pathinfo($file, PATHINFO_EXTENSION) === 'part') {
return true;
}
return false;
}
/**
* walk over any folders that are not fully scanned yet and scan them
*/
public function backgroundScan() {
|
|
31b7f2792
|
310 311 312 |
$lastPath = null;
while (($path = $this->cache->getIncomplete()) !== false && $path !== $lastPath) {
$this->scan($path, self::SCAN_RECURSIVE, self::REUSE_ETAG);
|
|
6d9380f96
|
313 314 315 316 |
\OC_Hook::emit('Scanner', 'correctFolderSize', array('path' => $path));
if ($this->cacheActive) {
$this->cache->correctFolderSize($path);
}
|
|
31b7f2792
|
317 |
$lastPath = $path; |
|
03e52840d
|
318 319 |
} } |
|
6d9380f96
|
320 321 322 323 324 325 326 327 328 |
/**
* 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;
}
|
|
03e52840d
|
329 |
} |