MediaWiki  master
ForeignAPIRepo.php
Go to the documentation of this file.
1 <?php
27 
44 class ForeignAPIRepo extends FileRepo implements IForeignRepoWithMWApi {
45  /* This version string is used in the user agent for requests and will help
46  * server maintainers in identify ForeignAPI usage.
47  * Update the version every time you make breaking or significant changes. */
48  private const VERSION = "2.1";
49 
53  private const IMAGE_INFO_PROPS = [
54  'url',
55  'timestamp',
56  ];
57 
58  protected $fileFactory = [ ForeignAPIFile::class, 'newFromTitle' ];
60  protected $apiThumbCacheExpiry = 86400; // 1 day (24*3600)
61 
63  protected $fileCacheExpiry = 2592000; // 1 month (30*24*3600)
64 
76  protected $apiMetadataExpiry = 14400; // 4 hours
77 
79  protected $mFileExists = [];
80 
82  private $mApiBase;
83 
87  public function __construct( $info ) {
88  $localFileRepo = MediaWikiServices::getInstance()->getMainConfig()
89  ->get( MainConfigNames::LocalFileRepo );
90  parent::__construct( $info );
91 
92  // https://commons.wikimedia.org/w/api.php
93  $this->mApiBase = $info['apibase'] ?? null;
94 
95  if ( isset( $info['apiThumbCacheExpiry'] ) ) {
96  $this->apiThumbCacheExpiry = $info['apiThumbCacheExpiry'];
97  }
98  if ( isset( $info['fileCacheExpiry'] ) ) {
99  $this->fileCacheExpiry = $info['fileCacheExpiry'];
100  }
101  if ( isset( $info['apiMetadataExpiry'] ) ) {
102  $this->apiMetadataExpiry = $info['apiMetadataExpiry'];
103  }
104  if ( !$this->scriptDirUrl ) {
105  // hack for description fetches
106  $this->scriptDirUrl = dirname( $this->mApiBase );
107  }
108  // If we can cache thumbs we can guess sensible defaults for these
109  if ( $this->canCacheThumbs() && !$this->url ) {
110  $this->url = $localFileRepo['url'];
111  }
112  if ( $this->canCacheThumbs() && !$this->thumbUrl ) {
113  $this->thumbUrl = $this->url . '/thumb';
114  }
115  }
116 
120  private function getApiUrl() {
121  return $this->mApiBase;
122  }
123 
132  public function newFile( $title, $time = false ) {
133  if ( $time ) {
134  return false;
135  }
136 
137  return parent::newFile( $title, $time );
138  }
139 
144  public function fileExistsBatch( array $files ) {
145  $results = [];
146  foreach ( $files as $k => $f ) {
147  if ( isset( $this->mFileExists[$f] ) ) {
148  $results[$k] = $this->mFileExists[$f];
149  unset( $files[$k] );
150  } elseif ( self::isVirtualUrl( $f ) ) {
151  # @todo FIXME: We need to be able to handle virtual
152  # URLs better, at least when we know they refer to the
153  # same repo.
154  $results[$k] = false;
155  unset( $files[$k] );
156  } elseif ( FileBackend::isStoragePath( $f ) ) {
157  $results[$k] = false;
158  unset( $files[$k] );
159  wfWarn( "Got mwstore:// path '$f'." );
160  }
161  }
162 
163  $data = $this->fetchImageQuery( [
164  'titles' => implode( '|', $files ),
165  'prop' => 'imageinfo' ]
166  );
167 
168  if ( isset( $data['query']['pages'] ) ) {
169  # First, get results from the query. Note we only care whether the image exists,
170  # not whether it has a description page.
171  foreach ( $data['query']['pages'] as $p ) {
172  $this->mFileExists[$p['title']] = ( $p['imagerepository'] !== '' );
173  }
174  # Second, copy the results to any redirects that were queried
175  if ( isset( $data['query']['redirects'] ) ) {
176  foreach ( $data['query']['redirects'] as $r ) {
177  $this->mFileExists[$r['from']] = $this->mFileExists[$r['to']];
178  }
179  }
180  # Third, copy the results to any non-normalized titles that were queried
181  if ( isset( $data['query']['normalized'] ) ) {
182  foreach ( $data['query']['normalized'] as $n ) {
183  $this->mFileExists[$n['from']] = $this->mFileExists[$n['to']];
184  }
185  }
186  # Finally, copy the results to the output
187  foreach ( $files as $key => $file ) {
188  $results[$key] = $this->mFileExists[$file];
189  }
190  }
191 
192  return $results;
193  }
194 
199  public function getFileProps( $virtualUrl ) {
200  return [];
201  }
202 
209  public function fetchImageQuery( $query ) {
210  $languageCode = MediaWikiServices::getInstance()->getMainConfig()
211  ->get( MainConfigNames::LanguageCode );
212 
213  $query = array_merge( $query,
214  [
215  'format' => 'json',
216  'action' => 'query',
217  'redirects' => 'true'
218  ] );
219 
220  if ( !isset( $query['uselang'] ) ) { // uselang is unset or null
221  $query['uselang'] = $languageCode;
222  }
223 
224  $data = $this->httpGetCached( 'Metadata', $query, $this->apiMetadataExpiry );
225 
226  if ( $data ) {
227  return FormatJson::decode( $data, true );
228  } else {
229  return null;
230  }
231  }
232 
237  public function getImageInfo( $data ) {
238  if ( $data && isset( $data['query']['pages'] ) ) {
239  foreach ( $data['query']['pages'] as $info ) {
240  if ( isset( $info['imageinfo'][0] ) ) {
241  $return = $info['imageinfo'][0];
242  if ( isset( $info['pageid'] ) ) {
243  $return['pageid'] = $info['pageid'];
244  }
245  return $return;
246  }
247  }
248  }
249 
250  return false;
251  }
252 
257  public function findBySha1( $hash ) {
258  $results = $this->fetchImageQuery( [
259  'aisha1base36' => $hash,
260  'aiprop' => ForeignAPIFile::getProps(),
261  'list' => 'allimages',
262  ] );
263  $ret = [];
264  if ( isset( $results['query']['allimages'] ) ) {
265  foreach ( $results['query']['allimages'] as $img ) {
266  // 1.14 was broken, doesn't return name attribute
267  if ( !isset( $img['name'] ) ) {
268  continue;
269  }
270  $ret[] = new ForeignAPIFile( Title::makeTitle( NS_FILE, $img['name'] ), $this, $img );
271  }
272  }
273 
274  return $ret;
275  }
276 
286  private function getThumbUrl(
287  $name, $width = -1, $height = -1, &$result = null, $otherParams = ''
288  ) {
289  $data = $this->fetchImageQuery( [
290  'titles' => 'File:' . $name,
291  'iiprop' => self::getIIProps(),
292  'iiurlwidth' => $width,
293  'iiurlheight' => $height,
294  'iiurlparam' => $otherParams,
295  'prop' => 'imageinfo' ] );
296  $info = $this->getImageInfo( $data );
297 
298  if ( $data && $info && isset( $info['thumburl'] ) ) {
299  wfDebug( __METHOD__ . " got remote thumb " . $info['thumburl'] );
300  $result = $info;
301 
302  return $info['thumburl'];
303  } else {
304  return false;
305  }
306  }
307 
317  public function getThumbError(
318  $name, $width = -1, $height = -1, $otherParams = '', $lang = null
319  ) {
320  $data = $this->fetchImageQuery( [
321  'titles' => 'File:' . $name,
322  'iiprop' => self::getIIProps(),
323  'iiurlwidth' => $width,
324  'iiurlheight' => $height,
325  'iiurlparam' => $otherParams,
326  'prop' => 'imageinfo',
327  'uselang' => $lang,
328  ] );
329  $info = $this->getImageInfo( $data );
330 
331  if ( $data && $info && isset( $info['thumberror'] ) ) {
332  wfDebug( __METHOD__ . " got remote thumb error " . $info['thumberror'] );
333 
334  return new MediaTransformError(
335  'thumbnail_error_remote',
336  $width,
337  $height,
338  $this->getDisplayName(),
339  $info['thumberror'] // already parsed message from foreign repo
340  );
341  } else {
342  return false;
343  }
344  }
345 
359  public function getThumbUrlFromCache( $name, $width, $height, $params = "" ) {
360  // We can't check the local cache using FileRepo functions because
361  // we override fileExistsBatch(). We have to use the FileBackend directly.
362  $backend = $this->getBackend(); // convenience
363 
364  if ( !$this->canCacheThumbs() ) {
365  $result = null; // can't pass "null" by reference, but it's ok as default value
366 
367  return $this->getThumbUrl( $name, $width, $height, $result, $params );
368  }
369 
370  $key = $this->getLocalCacheKey( 'file-thumb-url', sha1( $name ) );
371  $sizekey = "$width:$height:$params";
372 
373  /* Get the array of urls that we already know */
374  $knownThumbUrls = $this->wanCache->get( $key );
375  if ( !$knownThumbUrls ) {
376  /* No knownThumbUrls for this file */
377  $knownThumbUrls = [];
378  } elseif ( isset( $knownThumbUrls[$sizekey] ) ) {
379  wfDebug( __METHOD__ . ': Got thumburl from local cache: ' .
380  "{$knownThumbUrls[$sizekey]}" );
381 
382  return $knownThumbUrls[$sizekey];
383  }
384 
385  $metadata = null;
386  $foreignUrl = $this->getThumbUrl( $name, $width, $height, $metadata, $params );
387 
388  if ( !$foreignUrl ) {
389  wfDebug( __METHOD__ . " Could not find thumburl" );
390 
391  return false;
392  }
393 
394  // We need the same filename as the remote one :)
395  $fileName = rawurldecode( pathinfo( $foreignUrl, PATHINFO_BASENAME ) );
396  if ( !$this->validateFilename( $fileName ) ) {
397  wfDebug( __METHOD__ . " The deduced filename $fileName is not safe" );
398 
399  return false;
400  }
401  $localPath = $this->getZonePath( 'thumb' ) . "/" . $this->getHashPath( $name ) . $name;
402  $localFilename = $localPath . "/" . $fileName;
403  $localUrl = $this->getZoneUrl( 'thumb' ) . "/" . $this->getHashPath( $name ) .
404  rawurlencode( $name ) . "/" . rawurlencode( $fileName );
405 
406  if ( $backend->fileExists( [ 'src' => $localFilename ] )
407  && isset( $metadata['timestamp'] )
408  ) {
409  wfDebug( __METHOD__ . " Thumbnail was already downloaded before" );
410  $modified = (int)wfTimestamp( TS_UNIX, $backend->getFileTimestamp( [ 'src' => $localFilename ] ) );
411  $remoteModified = (int)wfTimestamp( TS_UNIX, $metadata['timestamp'] );
412  $current = (int)wfTimestamp( TS_UNIX );
413  $diff = abs( $modified - $current );
414  if ( $remoteModified < $modified && $diff < $this->fileCacheExpiry ) {
415  /* Use our current and already downloaded thumbnail */
416  $knownThumbUrls[$sizekey] = $localUrl;
417  $this->wanCache->set( $key, $knownThumbUrls, $this->apiThumbCacheExpiry );
418 
419  return $localUrl;
420  }
421  /* There is a new Commons file, or existing thumbnail older than a month */
422  }
423 
424  $thumb = self::httpGet( $foreignUrl, 'default', [], $mtime );
425  if ( !$thumb ) {
426  wfDebug( __METHOD__ . " Could not download thumb" );
427 
428  return false;
429  }
430 
431  # @todo FIXME: Delete old thumbs that aren't being used. Maintenance script?
432  $backend->prepare( [ 'dir' => dirname( $localFilename ) ] );
433  $params = [ 'dst' => $localFilename, 'content' => $thumb ];
434  if ( !$backend->quickCreate( $params )->isOK() ) {
435  wfDebug( __METHOD__ . " could not write to thumb path '$localFilename'" );
436 
437  return $foreignUrl;
438  }
439  $knownThumbUrls[$sizekey] = $localUrl;
440 
441  $ttl = $mtime
442  ? $this->wanCache->adaptiveTTL( $mtime, $this->apiThumbCacheExpiry )
444  $this->wanCache->set( $key, $knownThumbUrls, $ttl );
445  wfDebug( __METHOD__ . " got local thumb $localUrl, saving to cache" );
446 
447  return $localUrl;
448  }
449 
456  public function getZoneUrl( $zone, $ext = null ) {
457  switch ( $zone ) {
458  case 'public':
459  return $this->url;
460  case 'thumb':
461  return $this->thumbUrl;
462  default:
463  return parent::getZoneUrl( $zone, $ext );
464  }
465  }
466 
472  public function getZonePath( $zone ) {
473  $supported = [ 'public', 'thumb' ];
474  if ( in_array( $zone, $supported ) ) {
475  return parent::getZonePath( $zone );
476  }
477 
478  return false;
479  }
480 
485  public function canCacheThumbs() {
486  return ( $this->apiThumbCacheExpiry > 0 );
487  }
488 
493  public static function getUserAgent() {
494  return MediaWikiServices::getInstance()->getHttpRequestFactory()->getUserAgent() .
495  " ForeignAPIRepo/" . self::VERSION;
496  }
497 
504  public function getInfo() {
505  $info = parent::getInfo();
506  $info['apiurl'] = $this->getApiUrl();
507 
508  $query = [
509  'format' => 'json',
510  'action' => 'query',
511  'meta' => 'siteinfo',
512  'siprop' => 'general',
513  ];
514 
515  $data = $this->httpGetCached( 'SiteInfo', $query, 7200 );
516 
517  if ( $data ) {
518  $siteInfo = FormatJson::decode( $data, true );
519  $general = $siteInfo['query']['general'];
520 
521  $info['articlepath'] = $general['articlepath'];
522  $info['server'] = $general['server'];
523  if ( !isset( $info['favicon'] ) && isset( $general['favicon'] ) ) {
524  $info['favicon'] = $general['favicon'];
525  }
526  }
527 
528  return $info;
529  }
530 
538  public static function httpGet(
539  $url, $timeout = 'default', $options = [], &$mtime = false
540  ) {
541  $options['timeout'] = $timeout;
542  $url = MediaWikiServices::getInstance()->getUrlUtils()
543  ->expand( $url, PROTO_HTTP );
544  wfDebug( "ForeignAPIRepo: HTTP GET: $url" );
545  if ( !$url ) {
546  return false;
547  }
548  $options['method'] = "GET";
549 
550  if ( !isset( $options['timeout'] ) ) {
551  $options['timeout'] = 'default';
552  }
553 
554  $options['userAgent'] = self::getUserAgent();
555 
556  $req = MediaWikiServices::getInstance()->getHttpRequestFactory()
557  ->create( $url, $options, __METHOD__ );
558  $status = $req->execute();
559 
560  if ( $status->isOK() ) {
561  $lmod = $req->getResponseHeader( 'Last-Modified' );
562  $mtime = $lmod ? (int)wfTimestamp( TS_UNIX, $lmod ) : false;
563 
564  return $req->getContent();
565  } else {
566  $logger = LoggerFactory::getInstance( 'http' );
567  $logger->warning(
568  $status->getWikiText( false, false, 'en' ),
569  [ 'caller' => 'ForeignAPIRepo::httpGet' ]
570  );
571 
572  return false;
573  }
574  }
575 
580  protected static function getIIProps() {
581  return implode( '|', self::IMAGE_INFO_PROPS );
582  }
583 
591  public function httpGetCached( $attribute, $query, $cacheTTL = 3600 ) {
592  if ( $this->mApiBase ) {
593  $url = wfAppendQuery( $this->mApiBase, $query );
594  } else {
595  $url = $this->makeUrl( $query, 'api' );
596  }
597 
598  return $this->wanCache->getWithSetCallback(
599  // Allow reusing the same cached data across wikis (T285271).
600  // This does not use getSharedCacheKey() because caching here
601  // is transparent to client wikis (which are not expected to issue purges).
602  $this->wanCache->makeGlobalKey( "filerepo-$attribute", sha1( $url ) ),
603  $cacheTTL,
604  function ( $curValue, &$ttl ) use ( $url ) {
605  $html = self::httpGet( $url, 'default', [], $mtime );
606  // FIXME: This should use the mtime from the api response body
607  // not the mtime from the last-modified header which usually is not set.
608  if ( $html !== false ) {
609  $ttl = $mtime ? $this->wanCache->adaptiveTTL( $mtime, $ttl ) : $ttl;
610  } else {
611  $ttl = $this->wanCache->adaptiveTTL( $mtime, $ttl );
612  $html = null; // caches negatives
613  }
614 
615  return $html;
616  },
617  [ 'pcGroup' => 'http-get:3', 'pcTTL' => WANObjectCache::TTL_PROC_LONG ]
618  );
619  }
620 
625  public function enumFiles( $callback ) {
626  // @phan-suppress-previous-line PhanPluginNeverReturnMethod
627  throw new MWException( 'enumFiles is not supported by ' . static::class );
628  }
629 
630  protected function assertWritableRepo() {
631  // @phan-suppress-previous-line PhanPluginNeverReturnMethod
632  throw new MWException( static::class . ': write operations are not supported.' );
633  }
634 }
const NS_FILE
Definition: Defines.php:70
const PROTO_HTTP
Definition: Defines.php:193
wfDebug( $text, $dest='all', array $context=[])
Sends a line to the debug log if enabled or, optionally, to a comment in output.
wfWarn( $msg, $callerOffset=1, $level=E_USER_NOTICE)
Send a warning either to the debug log or in a PHP error depending on $wgDevelopmentWarnings.
wfAppendQuery( $url, $query)
Append a query string to an existing URL, which may or may not already have query string parameters a...
wfTimestamp( $outputtype=TS_UNIX, $ts=0)
Get a timestamp string in one of various formats.
static isStoragePath( $path)
Check if a given path is a "mwstore://" path.
quickCreate(array $params, array $opts=[])
Performs a single quick create operation.
fileExists(array $params)
Check if a file exists at a storage path in the backend.
getFileTimestamp(array $params)
Get the last-modified timestamp of the file at a storage path.
prepare(array $params)
Prepare a storage directory for usage.
Base class for file repositories.
Definition: FileRepo.php:48
getDisplayName()
Get the human-readable name of the repo.
Definition: FileRepo.php:1841
getLocalCacheKey( $kClassSuffix,... $components)
Get a site-local, repository-qualified, WAN cache key.
Definition: FileRepo.php:1903
makeUrl( $query='', $entry='index')
Make an url to this repo.
Definition: FileRepo.php:807
FileBackend $backend
Definition: FileRepo.php:71
string false $url
Public zone URL.
Definition: FileRepo.php:112
validateFilename( $filename)
Determine if a relative path is valid, i.e.
Definition: FileRepo.php:1742
string $name
Definition: FileRepo.php:162
string false $thumbUrl
The base thumbnail URL.
Definition: FileRepo.php:115
getHashPath( $name)
Get a relative path including trailing slash, e.g.
Definition: FileRepo.php:746
getBackend()
Get the file backend instance.
Definition: FileRepo.php:251
Foreign file accessible through api.php requests.
static getProps()
Get the property string for iiprop and aiprop.
A foreign repository for a remote MediaWiki accessible through api.php requests.
int $fileCacheExpiry
Redownload thumbnail files after this expiry.
newFile( $title, $time=false)
Per docs in FileRepo, this needs to return false if we don't support versioned files.
int $apiMetadataExpiry
API metadata cache time.
static httpGet( $url, $timeout='default', $options=[], &$mtime=false)
enumFiles( $callback)
getInfo()
Get information about the repo - overrides/extends the parent class's information.
static getUserAgent()
The user agent the ForeignAPIRepo will use.
fetchImageQuery( $query)
Make an API query in the foreign repo, caching results.
getThumbError( $name, $width=-1, $height=-1, $otherParams='', $lang=null)
assertWritableRepo()
Throw an exception if this repo is read-only by design.
canCacheThumbs()
Are we locally caching the thumbnails?
int $apiThumbCacheExpiry
Check back with Commons after this expiry.
getFileProps( $virtualUrl)
getThumbUrlFromCache( $name, $width, $height, $params="")
Return the imageurl from cache if possible.
fileExistsBatch(array $files)
getZoneUrl( $zone, $ext=null)
getZonePath( $zone)
Get the local directory corresponding to one of the basic zones.
httpGetCached( $attribute, $query, $cacheTTL=3600)
HTTP GET request to a mediawiki API (with caching)
static decode( $value, $assoc=false)
Decodes a JSON string.
Definition: FormatJson.php:146
MediaWiki exception.
Definition: MWException.php:32
Basic media transform error class.
PSR-3 logger instance factory.
A class containing constants representing the names of configuration variables.
Service locator for MediaWiki core services.
Represents a title within MediaWiki.
Definition: Title.php:82
A foreign repo that implement support for API queries.
Interface for objects (potentially) representing an editable wiki page.
if(PHP_SAPI !='cli-server') if(!isset( $_SERVER['SCRIPT_FILENAME'])) $file
Item class for a filearchive table row.
Definition: router.php:42
if(!is_readable( $file)) $ext
Definition: router.php:48
if(!isset( $args[0])) $lang