MediaWiki  1.28.0
ApiStashEdit.php
Go to the documentation of this file.
1 <?php
25 
39 class ApiStashEdit extends ApiBase {
40  const ERROR_NONE = 'stashed';
41  const ERROR_PARSE = 'error_parse';
42  const ERROR_CACHE = 'error_cache';
43  const ERROR_UNCACHEABLE = 'uncacheable';
44  const ERROR_BUSY = 'busy';
45 
47  const MAX_CACHE_TTL = 300; // 5 minutes
48 
49  public function execute() {
50  $user = $this->getUser();
51  $params = $this->extractRequestParams();
52 
53  if ( $user->isBot() ) { // sanity
54  $this->dieUsage( 'This interface is not supported for bots', 'botsnotsupported' );
55  }
56 
58  $page = $this->getTitleOrPageId( $params );
59  $title = $page->getTitle();
60 
61  if ( !ContentHandler::getForModelID( $params['contentmodel'] )
62  ->isSupportedFormat( $params['contentformat'] )
63  ) {
64  $this->dieUsage( 'Unsupported content model/format', 'badmodelformat' );
65  }
66 
67  $text = null;
68  $textHash = null;
69  if ( strlen( $params['stashedtexthash'] ) ) {
70  // Load from cache since the client indicates the text is the same as last stash
71  $textHash = $params['stashedtexthash'];
72  $textKey = $cache->makeKey( 'stashedit', 'text', $textHash );
73  $text = $cache->get( $textKey );
74  if ( !is_string( $text ) ) {
75  $this->dieUsage( 'No stashed text found with the given hash', 'missingtext' );
76  }
77  } elseif ( $params['text'] !== null ) {
78  // Trim and fix newlines so the key SHA1's match (see WebRequest::getText())
79  $text = rtrim( str_replace( "\r\n", "\n", $params['text'] ) );
80  $textHash = sha1( $text );
81  } else {
82  $this->dieUsage(
83  'The text or stashedtexthash parameter must be given', 'missingtextparam' );
84  }
85 
86  $textContent = ContentHandler::makeContent(
87  $text, $title, $params['contentmodel'], $params['contentformat'] );
88 
90  if ( $page->exists() ) {
91  // Page exists: get the merged content with the proposed change
92  $baseRev = Revision::newFromPageId( $page->getId(), $params['baserevid'] );
93  if ( !$baseRev ) {
94  $this->dieUsage( "No revision ID {$params['baserevid']}", 'missingrev' );
95  }
96  $currentRev = $page->getRevision();
97  if ( !$currentRev ) {
98  $this->dieUsage( "No current revision of page ID {$page->getId()}", 'missingrev' );
99  }
100  // Merge in the new version of the section to get the proposed version
101  $editContent = $page->replaceSectionAtRev(
102  $params['section'],
103  $textContent,
104  $params['sectiontitle'],
105  $baseRev->getId()
106  );
107  if ( !$editContent ) {
108  $this->dieUsage( 'Could not merge updated section.', 'replacefailed' );
109  }
110  if ( $currentRev->getId() == $baseRev->getId() ) {
111  // Base revision was still the latest; nothing to merge
112  $content = $editContent;
113  } else {
114  // Merge the edit into the current version
115  $baseContent = $baseRev->getContent();
116  $currentContent = $currentRev->getContent();
117  if ( !$baseContent || !$currentContent ) {
118  $this->dieUsage( "Missing content for page ID {$page->getId()}", 'missingrev' );
119  }
120  $handler = ContentHandler::getForModelID( $baseContent->getModel() );
121  $content = $handler->merge3( $baseContent, $editContent, $currentContent );
122  }
123  } else {
124  // New pages: use the user-provided content model
125  $content = $textContent;
126  }
127 
128  if ( !$content ) { // merge3() failed
129  $this->getResult()->addValue( null,
130  $this->getModuleName(), [ 'status' => 'editconflict' ] );
131  return;
132  }
133 
134  // The user will abort the AJAX request by pressing "save", so ignore that
135  ignore_user_abort( true );
136 
137  if ( $user->pingLimiter( 'stashedit' ) ) {
138  $status = 'ratelimited';
139  } else {
140  $status = self::parseAndStash( $page, $content, $user, $params['summary'] );
141  $textKey = $cache->makeKey( 'stashedit', 'text', $textHash );
142  $cache->set( $textKey, $text, self::MAX_CACHE_TTL );
143  }
144 
145  $stats = MediaWikiServices::getInstance()->getStatsdDataFactory();
146  $stats->increment( "editstash.cache_stores.$status" );
147 
148  $this->getResult()->addValue(
149  null,
150  $this->getModuleName(),
151  [
152  'status' => $status,
153  'texthash' => $textHash
154  ]
155  );
156  }
157 
168  $logger = LoggerFactory::getInstance( 'StashEdit' );
169 
170  $title = $page->getTitle();
171  $key = self::getStashKey( $title, self::getContentHash( $content ), $user );
172 
173  // Use the master DB for fast blocking locks
174  $dbw = wfGetDB( DB_MASTER );
175  if ( !$dbw->lock( $key, __METHOD__, 1 ) ) {
176  // De-duplicate requests on the same key
177  return self::ERROR_BUSY;
178  }
180  $unlocker = new ScopedCallback( function () use ( $dbw, $key ) {
181  $dbw->unlock( $key, __METHOD__ );
182  } );
183 
184  $cutoffTime = time() - self::PRESUME_FRESH_TTL_SEC;
185 
186  // Reuse any freshly build matching edit stash cache
187  $editInfo = $cache->get( $key );
188  if ( $editInfo && wfTimestamp( TS_UNIX, $editInfo->timestamp ) >= $cutoffTime ) {
189  $alreadyCached = true;
190  } else {
191  $format = $content->getDefaultFormat();
192  $editInfo = $page->prepareContentForEdit( $content, null, $user, $format, false );
193  $alreadyCached = false;
194  }
195 
196  if ( $editInfo && $editInfo->output ) {
197  // Let extensions add ParserOutput metadata or warm other caches
198  Hooks::run( 'ParserOutputStashForEdit',
199  [ $page, $content, $editInfo->output, $summary, $user ] );
200 
201  if ( $alreadyCached ) {
202  $logger->debug( "Already cached parser output for key '$key' ('$title')." );
203  return self::ERROR_NONE;
204  }
205 
206  list( $stashInfo, $ttl, $code ) = self::buildStashValue(
207  $editInfo->pstContent,
208  $editInfo->output,
209  $editInfo->timestamp,
210  $user
211  );
212 
213  if ( $stashInfo ) {
214  $ok = $cache->set( $key, $stashInfo, $ttl );
215  if ( $ok ) {
216  $logger->debug( "Cached parser output for key '$key' ('$title')." );
217  return self::ERROR_NONE;
218  } else {
219  $logger->error( "Failed to cache parser output for key '$key' ('$title')." );
220  return self::ERROR_CACHE;
221  }
222  } else {
223  $logger->info( "Uncacheable parser output for key '$key' ('$title') [$code]." );
224  return self::ERROR_UNCACHEABLE;
225  }
226  }
227 
228  return self::ERROR_PARSE;
229  }
230 
248  public static function checkCache( Title $title, Content $content, User $user ) {
249  if ( $user->isBot() ) {
250  return false; // bots never stash - don't pollute stats
251  }
252 
254  $logger = LoggerFactory::getInstance( 'StashEdit' );
255  $stats = MediaWikiServices::getInstance()->getStatsdDataFactory();
256 
257  $key = self::getStashKey( $title, self::getContentHash( $content ), $user );
258  $editInfo = $cache->get( $key );
259  if ( !is_object( $editInfo ) ) {
260  $start = microtime( true );
261  // We ignore user aborts and keep parsing. Block on any prior parsing
262  // so as to use its results and make use of the time spent parsing.
263  // Skip this logic if there no master connection in case this method
264  // is called on an HTTP GET request for some reason.
265  $lb = MediaWikiServices::getInstance()->getDBLoadBalancer();
266  $dbw = $lb->getAnyOpenConnection( $lb->getWriterIndex() );
267  if ( $dbw && $dbw->lock( $key, __METHOD__, 30 ) ) {
268  $editInfo = $cache->get( $key );
269  $dbw->unlock( $key, __METHOD__ );
270  }
271 
272  $timeMs = 1000 * max( 0, microtime( true ) - $start );
273  $stats->timing( 'editstash.lock_wait_time', $timeMs );
274  }
275 
276  if ( !is_object( $editInfo ) || !$editInfo->output ) {
277  $stats->increment( 'editstash.cache_misses.no_stash' );
278  $logger->debug( "Empty cache for key '$key' ('$title'); user '{$user->getName()}'." );
279  return false;
280  }
281 
282  $age = time() - wfTimestamp( TS_UNIX, $editInfo->output->getCacheTime() );
283  if ( $age <= self::PRESUME_FRESH_TTL_SEC ) {
284  // Assume nothing changed in this time
285  $stats->increment( 'editstash.cache_hits.presumed_fresh' );
286  $logger->debug( "Timestamp-based cache hit for key '$key' (age: $age sec)." );
287  } elseif ( isset( $editInfo->edits ) && $editInfo->edits === $user->getEditCount() ) {
288  // Logged-in user made no local upload/template edits in the meantime
289  $stats->increment( 'editstash.cache_hits.presumed_fresh' );
290  $logger->debug( "Edit count based cache hit for key '$key' (age: $age sec)." );
291  } elseif ( $user->isAnon()
292  && self::lastEditTime( $user ) < $editInfo->output->getCacheTime()
293  ) {
294  // Logged-out user made no local upload/template edits in the meantime
295  $stats->increment( 'editstash.cache_hits.presumed_fresh' );
296  $logger->debug( "Edit check based cache hit for key '$key' (age: $age sec)." );
297  } else {
298  // User may have changed included content
299  $editInfo = false;
300  }
301 
302  if ( !$editInfo ) {
303  $stats->increment( 'editstash.cache_misses.proven_stale' );
304  $logger->info( "Stale cache for key '$key'; old key with outside edits. (age: $age sec)" );
305  } elseif ( $editInfo->output->getFlag( 'vary-revision' ) ) {
306  // This can be used for the initial parse, e.g. for filters or doEditContent(),
307  // but a second parse will be triggered in doEditUpdates(). This is not optimal.
308  $logger->info( "Cache for key '$key' ('$title') has vary_revision." );
309  } elseif ( $editInfo->output->getFlag( 'vary-revision-id' ) ) {
310  // Similar to the above if we didn't guess the ID correctly.
311  $logger->info( "Cache for key '$key' ('$title') has vary_revision_id." );
312  }
313 
314  return $editInfo;
315  }
316 
321  private static function lastEditTime( User $user ) {
322  $time = wfGetDB( DB_REPLICA )->selectField(
323  'recentchanges',
324  'MAX(rc_timestamp)',
325  [ 'rc_user_text' => $user->getName() ],
326  __METHOD__
327  );
328 
329  return wfTimestampOrNull( TS_MW, $time );
330  }
331 
338  private static function getContentHash( Content $content ) {
339  return sha1( implode( "\n", [
340  $content->getModel(),
341  $content->getDefaultFormat(),
342  $content->serialize( $content->getDefaultFormat() )
343  ] ) );
344  }
345 
358  private static function getStashKey( Title $title, $contentHash, User $user ) {
359  return ObjectCache::getLocalClusterInstance()->makeKey(
360  'prepared-edit',
361  md5( $title->getPrefixedDBkey() ),
362  // Account for the edit model/text
363  $contentHash,
364  // Account for user name related variables like signatures
365  md5( $user->getId() . "\n" . $user->getName() )
366  );
367  }
368 
380  private static function buildStashValue(
382  ) {
383  // If an item is renewed, mind the cache TTL determined by config and parser functions.
384  // Put an upper limit on the TTL for sanity to avoid extreme template/file staleness.
385  $since = time() - wfTimestamp( TS_UNIX, $parserOutput->getTimestamp() );
386  $ttl = min( $parserOutput->getCacheExpiry() - $since, self::MAX_CACHE_TTL );
387  if ( $ttl <= 0 ) {
388  return [ null, 0, 'no_ttl' ];
389  }
390 
391  // Only store what is actually needed
392  $stashInfo = (object)[
393  'pstContent' => $pstContent,
394  'output' => $parserOutput,
395  'timestamp' => $timestamp,
396  'edits' => $user->getEditCount()
397  ];
398 
399  return [ $stashInfo, $ttl, 'ok' ];
400  }
401 
402  public function getAllowedParams() {
403  return [
404  'title' => [
405  ApiBase::PARAM_TYPE => 'string',
407  ],
408  'section' => [
409  ApiBase::PARAM_TYPE => 'string',
410  ],
411  'sectiontitle' => [
412  ApiBase::PARAM_TYPE => 'string'
413  ],
414  'text' => [
415  ApiBase::PARAM_TYPE => 'text',
416  ApiBase::PARAM_DFLT => null
417  ],
418  'stashedtexthash' => [
419  ApiBase::PARAM_TYPE => 'string',
420  ApiBase::PARAM_DFLT => null
421  ],
422  'summary' => [
423  ApiBase::PARAM_TYPE => 'string',
424  ],
425  'contentmodel' => [
428  ],
429  'contentformat' => [
432  ],
433  'baserevid' => [
434  ApiBase::PARAM_TYPE => 'integer',
436  ]
437  ];
438  }
439 
440  public function needsToken() {
441  return 'csrf';
442  }
443 
444  public function mustBePosted() {
445  return true;
446  }
447 
448  public function isWriteMode() {
449  return true;
450  }
451 
452  public function isInternal() {
453  return true;
454  }
455 }
static factory(Title $title)
Create a WikiPage object of the appropriate class for the given title.
Definition: WikiPage.php:115
const PARAM_TYPE
(string|string[]) Either an array of allowed value strings, or a string type as described below...
Definition: ApiBase.php:88
static getStashKey(Title $title, $contentHash, User $user)
Get the temporary prepared edit stash key for a user.
deferred txt A few of the database updates required by various functions here can be deferred until after the result page is displayed to the user For updating the view updating the linked to tables after a etc PHP does not yet have any way to tell the server to actually return and disconnect while still running these but it might have such a feature in the future We handle these by creating a deferred update object and putting those objects on a global list
Definition: deferred.txt:11
wfGetDB($db, $groups=[], $wiki=false)
Get a Database object.
getResult()
Get the result object.
Definition: ApiBase.php:577
serialize($format=null)
Convenience method for serializing this Content object.
Prepare an edit in shared cache so that it can be reused on edit.
static checkCache(Title $title, Content $content, User $user)
Check that a prepared edit is in cache and still up-to-date.
static parseAndStash(WikiPage $page, Content $content, User $user, $summary)
static getAllContentFormats()
Apache License January AND DISTRIBUTION Definitions License shall mean the terms and conditions for use
static getForModelID($modelId)
Returns the ContentHandler singleton for the given model ID.
const PARAM_DFLT
(null|boolean|integer|string) Default value of the parameter.
Definition: ApiBase.php:50
static getContentModels()
const PARAM_REQUIRED
(boolean) Is the parameter required?
Definition: ApiBase.php:112
extractRequestParams($parseLimit=true)
Using getAllowedParams(), this function makes an array of the values provided by the user...
Definition: ApiBase.php:678
static getLocalClusterInstance()
Get the main cluster-local cache object.
injection txt This is an overview of how MediaWiki makes use of dependency injection The design described here grew from the discussion of RFC T384 The term dependency this means that anything an object needs to operate should be injected from the the object itself should only know narrow no concrete implementation of the logic it relies on The requirement to inject everything typically results in an architecture that based on two main types of and essentially stateless service objects that use other service objects to operate on the value objects As of the beginning MediaWiki is only starting to use the DI approach Much of the code still relies on global state or direct resulting in a highly cyclical dependency MediaWikiServices
Definition: injection.txt:23
static newFromPageId($pageId, $revId=0, $flags=0)
Load either the current, or a specified, revision that's attached to a given page ID...
Definition: Revision.php:159
const MAX_CACHE_TTL
const DB_MASTER
Definition: defines.php:23
globals will be eliminated from MediaWiki replaced by an application object which would be passed to constructors Whether that would be an convenient solution remains to be but certainly PHP makes such object oriented programming models easier than they were in previous versions For the time being MediaWiki programmers will have to work in an environment with some global context At the time of globals were initialised on startup by MediaWiki of these were configuration which are documented in DefaultSettings php There is no comprehensive documentation for the remaining however some of the most important ones are listed below They are typically initialised either in index php or in Setup php For a description of the see design txt $wgTitle Title object created from the request URL $wgOut OutputPage object for HTTP response $wgUser User object for the user associated with the current request $wgLang Language object selected by user preferences $wgContLang Language object associated with the wiki being viewed $wgParser Parser object Parser extensions register their hooks here $wgRequest WebRequest object
Definition: globals.txt:25
getName()
Get the user name, or the IP of an anonymous user.
Definition: User.php:2108
const TS_UNIX
Unix time - the number of seconds since 1970-01-01 00:00:00 UTC.
Definition: defines.php:6
wfTimestamp($outputtype=TS_UNIX, $ts=0)
Get a timestamp string in one of various formats.
null means default in associative array with keys and values unescaped Should be merged with default with a value of false meaning to suppress the attribute in associative array with keys and values unescaped noclasses just before the function returns a value If you return true
Definition: hooks.txt:1936
getTitleOrPageId($params, $load=false)
Get a WikiPage object from a title or pageid param, if possible.
Definition: ApiBase.php:833
getCacheExpiry()
Returns the number of seconds after which this object should expire.
Definition: CacheTime.php:110
const ERROR_BUSY
const ERROR_NONE
isAnon()
Get whether the user is anonymous.
Definition: User.php:3388
if($limit) $timestamp
this hook is for auditing only RecentChangesLinked and Watchlist RecentChangesLinked and Watchlist e g Watchlist removed from all revisions and log entries to which it was applied This gives extensions a chance to take it off their books as the deletion has already been partly carried out by this point or something similar the user will be unable to create the tag set and then return false from the hook function Ensure you consume the ChangeTagAfterDelete hook to carry out custom deletion actions as context $parserOutput
Definition: hooks.txt:1046
$summary
Base interface for content objects.
Definition: Content.php:34
$cache
Definition: mcc.php:33
$params
getTitle()
Get the title object of the article.
Definition: WikiPage.php:232
const TS_MW
MediaWiki concatenated string timestamp (YYYYMMDDHHMMSS)
Definition: defines.php:11
namespace and then decline to actually register it file or subcat img or subcat $title
Definition: hooks.txt:953
const ERROR_PARSE
getModuleName()
Get the name of the module being executed by this instance.
Definition: ApiBase.php:457
const ERROR_CACHE
static run($event, array $args=[], $deprecatedVersion=null)
Call hook functions defined in Hooks::register and $wgHooks.
Definition: Hooks.php:131
isBot()
Definition: User.php:3396
static makeContent($text, Title $title=null, $modelId=null, $format=null)
Convenience function for creating a Content object from a given textual representation.
static lastEditTime(User $user)
this hook is for auditing only or null if authentication failed before getting that far or null if we can t even determine that probably a stub it is not rendered in wiki pages or galleries in category pages allow injecting custom HTML after the section Any uses of the hook need to handle escaping see BaseTemplate::getToolbox and BaseTemplate::makeListItem for details on the format of individual items inside of this array or by returning and letting standard HTTP rendering take place modifiable or by returning false and taking over the output modifiable & $code
Definition: hooks.txt:802
please add to it if you re going to add events to the MediaWiki code where normally authentication against an external auth plugin would be creating a local account $user
Definition: hooks.txt:242
Class representing a MediaWiki article and history.
Definition: WikiPage.php:32
prepareContentForEdit(Content $content, $revision=null, User $user=null, $serialFormat=null, $useCache=true)
Prepare content which is about to be saved.
Definition: WikiPage.php:2103
injection txt This is an overview of how MediaWiki makes use of dependency injection The design described here grew from the discussion of RFC T384 The term dependency this means that anything an object needs to operate should be injected from the the object itself should only know narrow no concrete implementation of the logic it relies on The requirement to inject everything typically results in an architecture that based on two main types of and essentially stateless service objects that use other service objects to operate on the value objects As of the beginning MediaWiki is only starting to use the DI approach Much of the code still relies on global state or direct resulting in a highly cyclical dependency which acts as the top level factory for services in MediaWiki which can be used to gain access to default instances of various services MediaWikiServices however also allows new services to be defined and default services to be redefined Services are defined or redefined by providing a callback the instantiator that will return a new instance of the service When it will create an instance of MediaWikiServices and populate it with the services defined in the files listed by thereby bootstrapping the DI framework Per $wgServiceWiringFiles lists includes ServiceWiring php
Definition: injection.txt:35
getId()
Get the user's ID.
Definition: User.php:2083
this hook is for auditing only RecentChangesLinked and Watchlist RecentChangesLinked and Watchlist e g Watchlist removed from all revisions and log entries to which it was applied This gives extensions a chance to take it off their books as the deletion has already been partly carried out by this point or something similar the user will be unable to create the tag set and then return false from the hook function Ensure you consume the ChangeTagAfterDelete hook to carry out custom deletion actions as context called by AbstractContent::getParserOutput May be used to override the normal model specific rendering of page content $content
Definition: hooks.txt:1046
const ERROR_UNCACHEABLE
getEditCount()
Get the user's edit count.
Definition: User.php:3270
static buildStashValue(Content $pstContent, ParserOutput $parserOutput, $timestamp, User $user)
Build a value to store in memcached based on the PST content and parser output.
dieUsage($description, $errorCode, $httpRespCode=0, $extradata=null)
Throw a UsageException, which will (if uncaught) call the main module's error handler and die with an...
Definition: ApiBase.php:1574
this hook is for auditing only RecentChangesLinked and Watchlist RecentChangesLinked and Watchlist e g Watchlist removed from all revisions and log entries to which it was applied This gives extensions a chance to take it off their books as the deletion has already been partly carried out by this point or something similar the user will be unable to create the tag set $status
Definition: hooks.txt:1046
MediaWiki Logger LoggerFactory implements a PSR[0] compatible message logging system Named Psr Log LoggerInterface instances can be obtained from the MediaWiki Logger LoggerFactory::getInstance() static method.MediaWiki\Logger\LoggerFactory expects a class implementing the MediaWiki\Logger\Spi interface to act as a factory for new Psr\Log\LoggerInterface instances.The"Spi"in MediaWiki\Logger\Spi stands for"service provider interface".An SPI is an API intended to be implemented or extended by a third party.This software design pattern is intended to enable framework extension and replaceable components.It is specifically used in the MediaWiki\Logger\LoggerFactory service to allow alternate PSR-3 logging implementations to be easily integrated with MediaWiki.The service provider interface allows the backend logging library to be implemented in multiple ways.The $wgMWLoggerDefaultSpi global provides the classname of the default MediaWiki\Logger\Spi implementation to be loaded at runtime.This can either be the name of a class implementing the MediaWiki\Logger\Spi with a zero argument const ructor or a callable that will return an MediaWiki\Logger\Spi instance.Alternately the MediaWiki\Logger\LoggerFactory MediaWiki Logger LoggerFactory
Definition: logger.txt:5
This abstract class implements many basic API functions, and is the base of all API classes...
Definition: ApiBase.php:39
const DB_REPLICA
Definition: defines.php:22
this hook is for auditing only or null if authentication failed before getting that far or null if we can t even determine that probably a stub it is not rendered in wiki pages or galleries in category pages allow injecting custom HTML after the section Any uses of the hook need to handle escaping see BaseTemplate::getToolbox and BaseTemplate::makeListItem for details on the format of individual items inside of this array or by returning and letting standard HTTP rendering take place modifiable or by returning false and taking over the output modifiable modifiable after all normalizations have been except for the $wgMaxImageArea check set to true or false to override the $wgMaxImageArea check result gives extension the possibility to transform it themselves $handler
Definition: hooks.txt:802
getModel()
Returns the ID of the content model used by this Content object.
wfTimestampOrNull($outputtype=TS_UNIX, $ts=null)
Return a formatted timestamp, or null if input is null.
getUser()
Get the User object.
see documentation in includes Linker php for Linker::makeImageLink & $time
Definition: hooks.txt:1749
getDefaultFormat()
Convenience method that returns the default serialization format for the content model that this Cont...
const PRESUME_FRESH_TTL_SEC
do that in ParserLimitReportFormat instead use this to modify the parameters of the image and a DIV can begin in one section and end in another Make sure your code can handle that case gracefully See the EditSectionClearerLink extension for an example zero but section is usually empty its values are the globals values before the output is cached $page
Definition: hooks.txt:2491
getPrefixedDBkey()
Get the prefixed database key form.
Definition: Title.php:1443
static getContentHash(Content $content)
Get hash of the content, factoring in model/format.