MediaWiki master
ForeignAPIRepo.php
Go to the documentation of this file.
1<?php
27
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 = 24 * 3600; // 1 day
61
63 protected $fileCacheExpiry = 30 * 24 * 3600; // 1 month
64
76 protected $apiMetadataExpiry = 4 * 3600; // 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
125 public function newFile( $title, $time = false ) {
126 if ( $time ) {
127 return false;
128 }
129
130 return parent::newFile( $title, $time );
131 }
132
137 public function fileExistsBatch( array $files ) {
138 $results = [];
139 foreach ( $files as $k => $f ) {
140 if ( isset( $this->mFileExists[$f] ) ) {
141 $results[$k] = $this->mFileExists[$f];
142 unset( $files[$k] );
143 } elseif ( self::isVirtualUrl( $f ) ) {
144 # @todo FIXME: We need to be able to handle virtual
145 # URLs better, at least when we know they refer to the
146 # same repo.
147 $results[$k] = false;
148 unset( $files[$k] );
149 } elseif ( FileBackend::isStoragePath( $f ) ) {
150 $results[$k] = false;
151 unset( $files[$k] );
152 wfWarn( "Got mwstore:// path '$f'." );
153 }
154 }
155
156 $data = $this->fetchImageQuery( [
157 'titles' => implode( '|', $files ),
158 'prop' => 'imageinfo' ]
159 );
160
161 if ( isset( $data['query']['pages'] ) ) {
162 # First, get results from the query. Note we only care whether the image exists,
163 # not whether it has a description page.
164 foreach ( $data['query']['pages'] as $p ) {
165 $this->mFileExists[$p['title']] = ( $p['imagerepository'] !== '' );
166 }
167 # Second, copy the results to any redirects that were queried
168 if ( isset( $data['query']['redirects'] ) ) {
169 foreach ( $data['query']['redirects'] as $r ) {
170 $this->mFileExists[$r['from']] = $this->mFileExists[$r['to']];
171 }
172 }
173 # Third, copy the results to any non-normalized titles that were queried
174 if ( isset( $data['query']['normalized'] ) ) {
175 foreach ( $data['query']['normalized'] as $n ) {
176 $this->mFileExists[$n['from']] = $this->mFileExists[$n['to']];
177 }
178 }
179 # Finally, copy the results to the output
180 foreach ( $files as $key => $file ) {
181 $results[$key] = $this->mFileExists[$file];
182 }
183 }
184
185 return $results;
186 }
187
192 public function getFileProps( $virtualUrl ) {
193 return [];
194 }
195
202 public function fetchImageQuery( $query ) {
203 $languageCode = MediaWikiServices::getInstance()->getMainConfig()
204 ->get( MainConfigNames::LanguageCode );
205
206 $query = array_merge( $query,
207 [
208 'format' => 'json',
209 'action' => 'query',
210 'redirects' => 'true'
211 ] );
212
213 if ( !isset( $query['uselang'] ) ) { // uselang is unset or null
214 $query['uselang'] = $languageCode;
215 }
216
217 $data = $this->httpGetCached( 'Metadata', $query, $this->apiMetadataExpiry );
218
219 if ( $data ) {
220 return FormatJson::decode( $data, true );
221 } else {
222 return null;
223 }
224 }
225
230 public function getImageInfo( $data ) {
231 if ( $data && isset( $data['query']['pages'] ) ) {
232 foreach ( $data['query']['pages'] as $info ) {
233 if ( isset( $info['imageinfo'][0] ) ) {
234 $return = $info['imageinfo'][0];
235 if ( isset( $info['pageid'] ) ) {
236 $return['pageid'] = $info['pageid'];
237 }
238 return $return;
239 }
240 }
241 }
242
243 return false;
244 }
245
250 public function findBySha1( $hash ) {
251 $results = $this->fetchImageQuery( [
252 'aisha1base36' => $hash,
253 'aiprop' => ForeignAPIFile::getProps(),
254 'list' => 'allimages',
255 ] );
256 $ret = [];
257 if ( isset( $results['query']['allimages'] ) ) {
258 foreach ( $results['query']['allimages'] as $img ) {
259 // 1.14 was broken, doesn't return name attribute
260 if ( !isset( $img['name'] ) ) {
261 continue;
262 }
263 $ret[] = new ForeignAPIFile( Title::makeTitle( NS_FILE, $img['name'] ), $this, $img );
264 }
265 }
266
267 return $ret;
268 }
269
279 private function getThumbUrl(
280 $name, $width = -1, $height = -1, &$result = null, $otherParams = ''
281 ) {
282 $data = $this->fetchImageQuery( [
283 'titles' => 'File:' . $name,
284 'iiprop' => self::getIIProps(),
285 'iiurlwidth' => $width,
286 'iiurlheight' => $height,
287 'iiurlparam' => $otherParams,
288 'prop' => 'imageinfo' ] );
289 $info = $this->getImageInfo( $data );
290
291 if ( $data && $info && isset( $info['thumburl'] ) ) {
292 wfDebug( __METHOD__ . " got remote thumb " . $info['thumburl'] );
293 $result = $info;
294
295 return $info['thumburl'];
296 } else {
297 return false;
298 }
299 }
300
310 public function getThumbError(
311 $name, $width = -1, $height = -1, $otherParams = '', $lang = null
312 ) {
313 $data = $this->fetchImageQuery( [
314 'titles' => 'File:' . $name,
315 'iiprop' => self::getIIProps(),
316 'iiurlwidth' => $width,
317 'iiurlheight' => $height,
318 'iiurlparam' => $otherParams,
319 'prop' => 'imageinfo',
320 'uselang' => $lang,
321 ] );
322 $info = $this->getImageInfo( $data );
323
324 if ( $data && $info && isset( $info['thumberror'] ) ) {
325 wfDebug( __METHOD__ . " got remote thumb error " . $info['thumberror'] );
326
327 return new MediaTransformError(
328 'thumbnail_error_remote',
329 $width,
330 $height,
331 $this->getDisplayName(),
332 $info['thumberror'] // already parsed message from foreign repo
333 );
334 } else {
335 return false;
336 }
337 }
338
352 public function getThumbUrlFromCache( $name, $width, $height, $params = "" ) {
353 // We can't check the local cache using FileRepo functions because
354 // we override fileExistsBatch(). We have to use the FileBackend directly.
355 $backend = $this->getBackend(); // convenience
356
357 if ( !$this->canCacheThumbs() ) {
358 $result = null; // can't pass "null" by reference, but it's ok as default value
359
360 return $this->getThumbUrl( $name, $width, $height, $result, $params );
361 }
362
363 $key = $this->getLocalCacheKey( 'file-thumb-url', sha1( $name ) );
364 $sizekey = "$width:$height:$params";
365
366 /* Get the array of urls that we already know */
367 $knownThumbUrls = $this->wanCache->get( $key );
368 if ( !$knownThumbUrls ) {
369 /* No knownThumbUrls for this file */
370 $knownThumbUrls = [];
371 } elseif ( isset( $knownThumbUrls[$sizekey] ) ) {
372 wfDebug( __METHOD__ . ': Got thumburl from local cache: ' .
373 "{$knownThumbUrls[$sizekey]}" );
374
375 return $knownThumbUrls[$sizekey];
376 }
377
378 $metadata = null;
379 $foreignUrl = $this->getThumbUrl( $name, $width, $height, $metadata, $params );
380
381 if ( !$foreignUrl ) {
382 wfDebug( __METHOD__ . " Could not find thumburl" );
383
384 return false;
385 }
386
387 // We need the same filename as the remote one :)
388 $fileName = rawurldecode( pathinfo( $foreignUrl, PATHINFO_BASENAME ) );
389 if ( !$this->validateFilename( $fileName ) ) {
390 wfDebug( __METHOD__ . " The deduced filename $fileName is not safe" );
391
392 return false;
393 }
394 $localPath = $this->getZonePath( 'thumb' ) . "/" . $this->getHashPath( $name ) . $name;
395 $localFilename = $localPath . "/" . $fileName;
396 $localUrl = $this->getZoneUrl( 'thumb' ) . "/" . $this->getHashPath( $name ) .
397 rawurlencode( $name ) . "/" . rawurlencode( $fileName );
398
399 if ( $backend->fileExists( [ 'src' => $localFilename ] )
400 && isset( $metadata['timestamp'] )
401 ) {
402 wfDebug( __METHOD__ . " Thumbnail was already downloaded before" );
403 $modified = (int)wfTimestamp( TS_UNIX, $backend->getFileTimestamp( [ 'src' => $localFilename ] ) );
404 $remoteModified = (int)wfTimestamp( TS_UNIX, $metadata['timestamp'] );
405 $current = (int)wfTimestamp( TS_UNIX );
406 $diff = abs( $modified - $current );
407 if ( $remoteModified < $modified && $diff < $this->fileCacheExpiry ) {
408 /* Use our current and already downloaded thumbnail */
409 $knownThumbUrls[$sizekey] = $localUrl;
410 $this->wanCache->set( $key, $knownThumbUrls, $this->apiThumbCacheExpiry );
411
412 return $localUrl;
413 }
414 /* There is a new Commons file, or existing thumbnail older than a month */
415 }
416
417 $thumb = self::httpGet( $foreignUrl, 'default', [], $mtime );
418 if ( !$thumb ) {
419 wfDebug( __METHOD__ . " Could not download thumb" );
420
421 return false;
422 }
423
424 # @todo FIXME: Delete old thumbs that aren't being used. Maintenance script?
425 $backend->prepare( [ 'dir' => dirname( $localFilename ) ] );
426 $params = [ 'dst' => $localFilename, 'content' => $thumb ];
427 if ( !$backend->quickCreate( $params )->isOK() ) {
428 wfDebug( __METHOD__ . " could not write to thumb path '$localFilename'" );
429
430 return $foreignUrl;
431 }
432 $knownThumbUrls[$sizekey] = $localUrl;
433
434 $ttl = $mtime
435 ? $this->wanCache->adaptiveTTL( $mtime, $this->apiThumbCacheExpiry )
437 $this->wanCache->set( $key, $knownThumbUrls, $ttl );
438 wfDebug( __METHOD__ . " got local thumb $localUrl, saving to cache" );
439
440 return $localUrl;
441 }
442
449 public function getZoneUrl( $zone, $ext = null ) {
450 switch ( $zone ) {
451 case 'public':
452 return $this->url;
453 case 'thumb':
454 return $this->thumbUrl;
455 default:
456 return parent::getZoneUrl( $zone, $ext );
457 }
458 }
459
465 public function getZonePath( $zone ) {
466 $supported = [ 'public', 'thumb' ];
467 if ( in_array( $zone, $supported ) ) {
468 return parent::getZonePath( $zone );
469 }
470
471 return false;
472 }
473
478 public function canCacheThumbs() {
479 return ( $this->apiThumbCacheExpiry > 0 );
480 }
481
486 public static function getUserAgent() {
487 return MediaWikiServices::getInstance()->getHttpRequestFactory()->getUserAgent() .
488 " ForeignAPIRepo/" . self::VERSION;
489 }
490
497 public function getInfo() {
498 $info = parent::getInfo();
499 $info['apiurl'] = $this->mApiBase;
500
501 $query = [
502 'format' => 'json',
503 'action' => 'query',
504 'meta' => 'siteinfo',
505 'siprop' => 'general',
506 ];
507
508 $data = $this->httpGetCached( 'SiteInfo', $query, 7200 );
509
510 if ( $data ) {
511 $siteInfo = FormatJson::decode( $data, true );
512 $general = $siteInfo['query']['general'];
513
514 $info['articlepath'] = $general['articlepath'];
515 $info['server'] = $general['server'];
516 if ( !isset( $info['favicon'] ) && isset( $general['favicon'] ) ) {
517 $info['favicon'] = $general['favicon'];
518 }
519 }
520
521 return $info;
522 }
523
531 public static function httpGet(
532 $url, $timeout = 'default', $options = [], &$mtime = false
533 ) {
534 $options['timeout'] = $timeout;
535 $url = MediaWikiServices::getInstance()->getUrlUtils()
536 ->expand( $url, PROTO_HTTP );
537 wfDebug( "ForeignAPIRepo: HTTP GET: $url" );
538 if ( !$url ) {
539 return false;
540 }
541 $options['method'] = "GET";
542
543 if ( !isset( $options['timeout'] ) ) {
544 $options['timeout'] = 'default';
545 }
546
547 $options['userAgent'] = self::getUserAgent();
548
549 $req = MediaWikiServices::getInstance()->getHttpRequestFactory()
550 ->create( $url, $options, __METHOD__ );
551 $status = $req->execute();
552
553 if ( $status->isOK() ) {
554 $lmod = $req->getResponseHeader( 'Last-Modified' );
555 $mtime = $lmod ? (int)wfTimestamp( TS_UNIX, $lmod ) : false;
556
557 return $req->getContent();
558 } else {
559 $logger = LoggerFactory::getInstance( 'http' );
560 $logger->warning(
561 $status->getWikiText( false, false, 'en' ),
562 [ 'caller' => 'ForeignAPIRepo::httpGet' ]
563 );
564
565 return false;
566 }
567 }
568
573 protected static function getIIProps() {
574 return implode( '|', self::IMAGE_INFO_PROPS );
575 }
576
584 public function httpGetCached( $attribute, $query, $cacheTTL = 3600 ) {
585 if ( $this->mApiBase ) {
586 $url = wfAppendQuery( $this->mApiBase, $query );
587 } else {
588 $url = $this->makeUrl( $query, 'api' );
589 }
590
591 return $this->wanCache->getWithSetCallback(
592 // Allow reusing the same cached data across wikis (T285271).
593 // This does not use getSharedCacheKey() because caching here
594 // is transparent to client wikis (which are not expected to issue purges).
595 $this->wanCache->makeGlobalKey( "filerepo-$attribute", sha1( $url ) ),
596 $cacheTTL,
597 function ( $curValue, &$ttl ) use ( $url ) {
598 $html = self::httpGet( $url, 'default', [], $mtime );
599 // FIXME: This should use the mtime from the api response body
600 // not the mtime from the last-modified header which usually is not set.
601 if ( $html !== false ) {
602 $ttl = $mtime ? $this->wanCache->adaptiveTTL( $mtime, $ttl ) : $ttl;
603 } else {
604 $ttl = $this->wanCache->adaptiveTTL( $mtime, $ttl );
605 $html = null; // caches negatives
606 }
607
608 return $html;
609 },
610 [ 'pcGroup' => 'http-get:3', 'pcTTL' => WANObjectCache::TTL_PROC_LONG ]
611 );
612 }
613
618 public function enumFiles( $callback ) {
619 throw new RuntimeException( 'enumFiles is not supported by ' . static::class );
620 }
621
625 protected function assertWritableRepo() {
626 throw new LogicException( static::class . ': write operations are not supported.' );
627 }
628}
const NS_FILE
Definition Defines.php:70
const PROTO_HTTP
Definition Defines.php:200
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:51
getDisplayName()
Get the human-readable name of the repo.
getLocalCacheKey( $kClassSuffix,... $components)
Get a site-local, repository-qualified, WAN cache key.
makeUrl( $query='', $entry='index')
Make an url to this repo.
Definition FileRepo.php:808
FileBackend $backend
Definition FileRepo.php:74
string false $url
Public zone URL.
Definition FileRepo.php:115
validateFilename( $filename)
Determine if a relative path is valid, i.e.
string $name
Definition FileRepo.php:165
string false $thumbUrl
The base thumbnail URL.
Definition FileRepo.php:118
getHashPath( $name)
Get a relative path including trailing slash, e.g.
Definition FileRepo.php:747
getBackend()
Get the file backend instance.
Definition FileRepo.php:253
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)
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)
Basic media transform error class.
Create PSR-3 logger objects.
A class containing constants representing the names of configuration variables.
Service locator for MediaWiki core services.
Represents a title within MediaWiki.
Definition Title.php:78
A foreign repo that implement support for API queries.
Represents the target of a wiki link.
Interface for objects (potentially) representing an editable wiki page.