MediaWiki  1.28.1
updateCollation.php
Go to the documentation of this file.
1 <?php
27 require_once __DIR__ . '/Maintenance.php';
28 
36  const BATCH_SIZE = 100; // Number of rows to process in one batch
37  const SYNC_INTERVAL = 5; // Wait for replica DBs after this many batches
38 
39  public $sizeHistogram = [];
40 
41  public function __construct() {
42  parent::__construct();
43 
44  global $wgCategoryCollation;
45  $this->addDescription( <<<TEXT
46 This script will find all rows in the categorylinks table whose collation is
47 out-of-date (cl_collation != '$wgCategoryCollation') and repopulate cl_sortkey
48 using the page title and cl_sortkey_prefix. If all collations are
49 up-to-date, it will do nothing.
50 TEXT
51  );
52 
53  $this->addOption( 'force', 'Run on all rows, even if the collation is ' .
54  'supposed to be up-to-date.', false, false, 'f' );
55  $this->addOption( 'previous-collation', 'Set the previous value of ' .
56  '$wgCategoryCollation here to speed up this script, especially if your ' .
57  'categorylinks table is large. This will only update rows with that ' .
58  'collation, though, so it may miss out-of-date rows with a different, ' .
59  'even older collation.', false, true );
60  $this->addOption( 'target-collation', 'Set this to the new collation type to ' .
61  'use instead of $wgCategoryCollation. Usually you should not use this, ' .
62  'you should just update $wgCategoryCollation in LocalSettings.php.',
63  false, true );
64  $this->addOption( 'dry-run', 'Don\'t actually change the collations, just ' .
65  'compile statistics.' );
66  $this->addOption( 'verbose-stats', 'Show more statistics.' );
67  }
68 
69  public function execute() {
70  global $wgCategoryCollation;
71 
72  $dbw = $this->getDB( DB_MASTER );
73  $dbr = $this->getDB( DB_REPLICA );
74  $force = $this->getOption( 'force' );
75  $dryRun = $this->getOption( 'dry-run' );
76  $verboseStats = $this->getOption( 'verbose-stats' );
77  if ( $this->hasOption( 'target-collation' ) ) {
78  $collationName = $this->getOption( 'target-collation' );
79  $collation = Collation::factory( $collationName );
80  } else {
81  $collationName = $wgCategoryCollation;
82  $collation = Collation::singleton();
83  }
84 
85  // Collation sanity check: in some cases the constructor will work,
86  // but this will raise an exception, breaking all category pages
87  $collation->getFirstLetter( 'MediaWiki' );
88 
89  // Locally at least, (my local is a rather old version of mysql)
90  // mysql seems to filesort if there is both an equality
91  // (but not for an inequality) condition on cl_collation in the
92  // WHERE and it is also the first item in the ORDER BY.
93  if ( $this->hasOption( 'previous-collation' ) ) {
94  $orderBy = 'cl_to, cl_type, cl_from';
95  } else {
96  $orderBy = 'cl_collation, cl_to, cl_type, cl_from';
97  }
98  $options = [
99  'LIMIT' => self::BATCH_SIZE,
100  'ORDER BY' => $orderBy,
101  'STRAIGHT_JOIN' // per T58041
102  ];
103 
104  if ( $force ) {
105  $collationConds = [];
106  } else {
107  if ( $this->hasOption( 'previous-collation' ) ) {
108  $collationConds['cl_collation'] = $this->getOption( 'previous-collation' );
109  } else {
110  $collationConds = [ 0 =>
111  'cl_collation != ' . $dbw->addQuotes( $collationName )
112  ];
113  }
114 
115  $count = $dbr->estimateRowCount(
116  'categorylinks',
117  '*',
118  $collationConds,
119  __METHOD__
120  );
121  // Improve estimate if feasible
122  if ( $count < 1000000 ) {
123  $count = $dbr->selectField(
124  'categorylinks',
125  'COUNT(*)',
126  $collationConds,
127  __METHOD__
128  );
129  }
130  if ( $count == 0 ) {
131  $this->output( "Collations up-to-date.\n" );
132 
133  return;
134  }
135  if ( $dryRun ) {
136  $this->output( "$count rows would be updated.\n" );
137  } else {
138  $this->output( "Fixing collation for $count rows.\n" );
139  }
140  wfWaitForSlaves();
141  }
142  $count = 0;
143  $batchCount = 0;
144  $batchConds = [];
145  do {
146  $this->output( "Selecting next " . self::BATCH_SIZE . " rows..." );
147 
148  // cl_type must be selected as a number for proper paging because
149  // enums suck.
150  if ( $dbw->getType() === 'mysql' ) {
151  $clType = 'cl_type+0 AS "cl_type_numeric"';
152  } else {
153  $clType = 'cl_type';
154  }
155  $res = $dbw->select(
156  [ 'categorylinks', 'page' ],
157  [ 'cl_from', 'cl_to', 'cl_sortkey_prefix', 'cl_collation',
158  'cl_sortkey', $clType,
159  'page_namespace', 'page_title'
160  ],
161  array_merge( $collationConds, $batchConds, [ 'cl_from = page_id' ] ),
162  __METHOD__,
163  $options
164  );
165  $this->output( " processing..." );
166 
167  if ( !$dryRun ) {
168  $this->beginTransaction( $dbw, __METHOD__ );
169  }
170  foreach ( $res as $row ) {
171  $title = Title::newFromRow( $row );
172  if ( !$row->cl_collation ) {
173  # This is an old-style row, so the sortkey needs to be
174  # converted.
175  if ( $row->cl_sortkey == $title->getText()
176  || $row->cl_sortkey == $title->getPrefixedText()
177  ) {
178  $prefix = '';
179  } else {
180  # Custom sortkey, use it as a prefix
181  $prefix = $row->cl_sortkey;
182  }
183  } else {
184  $prefix = $row->cl_sortkey_prefix;
185  }
186  # cl_type will be wrong for lots of pages if cl_collation is 0,
187  # so let's update it while we're here.
188  if ( $title->getNamespace() == NS_CATEGORY ) {
189  $type = 'subcat';
190  } elseif ( $title->getNamespace() == NS_FILE ) {
191  $type = 'file';
192  } else {
193  $type = 'page';
194  }
195  $newSortKey = $collation->getSortKey(
196  $title->getCategorySortkey( $prefix ) );
197  if ( $verboseStats ) {
198  $this->updateSortKeySizeHistogram( $newSortKey );
199  }
200 
201  if ( !$dryRun ) {
202  $dbw->update(
203  'categorylinks',
204  [
205  'cl_sortkey' => $newSortKey,
206  'cl_sortkey_prefix' => $prefix,
207  'cl_collation' => $collationName,
208  'cl_type' => $type,
209  'cl_timestamp = cl_timestamp',
210  ],
211  [ 'cl_from' => $row->cl_from, 'cl_to' => $row->cl_to ],
212  __METHOD__
213  );
214  }
215  if ( $row ) {
216  $batchConds = [ $this->getBatchCondition( $row, $dbw ) ];
217  }
218  }
219  if ( !$dryRun ) {
220  $this->commitTransaction( $dbw, __METHOD__ );
221  }
222 
223  $count += $res->numRows();
224  $this->output( "$count done.\n" );
225 
226  if ( !$dryRun && ++$batchCount % self::SYNC_INTERVAL == 0 ) {
227  $this->output( "Waiting for replica DBs ... " );
228  wfWaitForSlaves();
229  $this->output( "done\n" );
230  }
231  } while ( $res->numRows() == self::BATCH_SIZE );
232 
233  $this->output( "$count rows processed\n" );
234 
235  if ( $verboseStats ) {
236  $this->output( "\n" );
237  $this->showSortKeySizeHistogram();
238  }
239  }
240 
248  function getBatchCondition( $row, $dbw ) {
249  if ( $this->hasOption( 'previous-collation' ) ) {
250  $fields = [ 'cl_to', 'cl_type', 'cl_from' ];
251  } else {
252  $fields = [ 'cl_collation', 'cl_to', 'cl_type', 'cl_from' ];
253  }
254  $first = true;
255  $cond = false;
256  $prefix = false;
257  foreach ( $fields as $field ) {
258  if ( $dbw->getType() === 'mysql' && $field === 'cl_type' ) {
259  // Range conditions with enums are weird in mysql
260  // This must be a numeric literal, or it won't work.
261  $encValue = intval( $row->cl_type_numeric );
262  } else {
263  $encValue = $dbw->addQuotes( $row->$field );
264  }
265  $inequality = "$field > $encValue";
266  $equality = "$field = $encValue";
267  if ( $first ) {
268  $cond = $inequality;
269  $prefix = $equality;
270  $first = false;
271  } else {
272  $cond .= " OR ($prefix AND $inequality)";
273  $prefix .= " AND $equality";
274  }
275  }
276 
277  return $cond;
278  }
279 
280  function updateSortKeySizeHistogram( $key ) {
281  $length = strlen( $key );
282  if ( !isset( $this->sizeHistogram[$length] ) ) {
283  $this->sizeHistogram[$length] = 0;
284  }
285  $this->sizeHistogram[$length]++;
286  }
287 
289  $maxLength = max( array_keys( $this->sizeHistogram ) );
290  if ( $maxLength == 0 ) {
291  return;
292  }
293  $numBins = 20;
294  $coarseHistogram = array_fill( 0, $numBins, 0 );
295  $coarseBoundaries = [];
296  $boundary = 0;
297  for ( $i = 0; $i < $numBins - 1; $i++ ) {
298  $boundary += $maxLength / $numBins;
299  $coarseBoundaries[$i] = round( $boundary );
300  }
301  $coarseBoundaries[$numBins - 1] = $maxLength + 1;
302  $raw = '';
303  for ( $i = 0; $i <= $maxLength; $i++ ) {
304  if ( $raw !== '' ) {
305  $raw .= ', ';
306  }
307  if ( !isset( $this->sizeHistogram[$i] ) ) {
308  $val = 0;
309  } else {
310  $val = $this->sizeHistogram[$i];
311  }
312  for ( $coarseIndex = 0; $coarseIndex < $numBins - 1; $coarseIndex++ ) {
313  if ( $coarseBoundaries[$coarseIndex] > $i ) {
314  $coarseHistogram[$coarseIndex] += $val;
315  break;
316  }
317  }
318  if ( $coarseIndex == $numBins - 1 ) {
319  $coarseHistogram[$coarseIndex] += $val;
320  }
321  $raw .= $val;
322  }
323 
324  $this->output( "Sort key size histogram\nRaw data: $raw\n\n" );
325 
326  $maxBinVal = max( $coarseHistogram );
327  $scale = 60 / $maxBinVal;
328  $prevBoundary = 0;
329  for ( $coarseIndex = 0; $coarseIndex < $numBins; $coarseIndex++ ) {
330  if ( !isset( $coarseHistogram[$coarseIndex] ) ) {
331  $val = 0;
332  } else {
333  $val = $coarseHistogram[$coarseIndex];
334  }
335  $boundary = $coarseBoundaries[$coarseIndex];
336  $this->output( sprintf( "%-10s %-10d |%s\n",
337  $prevBoundary . '-' . ( $boundary - 1 ) . ': ',
338  $val,
339  str_repeat( '*', $scale * $val ) ) );
340  $prevBoundary = $boundary;
341  }
342  }
343 }
344 
345 $maintClass = "UpdateCollation";
346 require_once RUN_MAINTENANCE_IF_MAIN;
static newFromRow($row)
Make a Title object from a DB row.
Definition: Title.php:450
#define the
table suitable for use with IDatabase::select()
commitTransaction(IDatabase $dbw, $fname)
Commit the transcation on a DB handle and wait for replica DBs to catch up.
wfWaitForSlaves($ifWritesSince=null, $wiki=false, $cluster=false, $timeout=null)
Waits for the replica DBs to catch up to the master position.
$maintClass
static factory($collationName)
Definition: Collation.php:48
We use the convention $dbr for read and $dbw for write to help you keep track of whether the database object is a the world will explode Or to be a subsequent write query which succeeded on the master may fail when replicated to the slave due to a unique key collision Replication on the slave will stop and it may take hours to repair the database and get it back online Setting read_only in my cnf on the slave will avoid this but given the dire we prefer to have as many checks as possible We provide a but the wrapper functions like please read the documentation for except in special pages derived from QueryPage It s a common pitfall for new developers to submit code containing SQL queries which examine huge numbers of rows Remember that COUNT * is(N), counting rows in atable is like counting beans in a bucket.------------------------------------------------------------------------Replication------------------------------------------------------------------------The largest installation of MediaWiki, Wikimedia, uses a large set ofslave MySQL servers replicating writes made to a master MySQL server.Itis important to understand the issues associated with this setup if youwant to write code destined for Wikipedia.It's often the case that the best algorithm to use for a given taskdepends on whether or not replication is in use.Due to our unabashedWikipedia-centrism, we often just use the replication-friendly version, but if you like, you can use wfGetLB() ->getServerCount() > 1 tocheck to see if replication is in use.===Lag===Lag primarily occurs when large write queries are sent to the master.Writes on the master are executed in parallel, but they are executed inserial when they are replicated to the slaves.The master writes thequery to the binlog when the transaction is committed.The slaves pollthe binlog and start executing the query as soon as it appears.They canservice reads while they are performing a write query, but will not readanything more from the binlog and thus will perform no more writes.Thismeans that if the write query runs for a long time, the slaves will lagbehind the master for the time it takes for the write query to complete.Lag can be exacerbated by high read load.MediaWiki's load balancer willstop sending reads to a slave when it is lagged by more than 30 seconds.If the load ratios are set incorrectly, or if there is too much loadgenerally, this may lead to a slave permanently hovering around 30seconds lag.If all slaves are lagged by more than 30 seconds, MediaWiki will stopwriting to the database.All edits and other write operations will berefused, with an error returned to the user.This gives the slaves achance to catch up.Before we had this mechanism, the slaves wouldregularly lag by several minutes, making review of recent editsdifficult.In addition to this, MediaWiki attempts to ensure that the user seesevents occurring on the wiki in chronological order.A few seconds of lagcan be tolerated, as long as the user sees a consistent picture fromsubsequent requests.This is done by saving the master binlog positionin the session, and then at the start of each request, waiting for theslave to catch up to that position before doing any reads from it.Ifthis wait times out, reads are allowed anyway, but the request isconsidered to be in"lagged slave mode".Lagged slave mode can bechecked by calling wfGetLB() ->getLaggedSlaveMode().The onlypractical consequence at present is a warning displayed in the pagefooter.===Lag avoidance===To avoid excessive lag, queries which write large numbers of rows shouldbe split up, generally to write one row at a time.Multi-row INSERT...SELECT queries are the worst offenders should be avoided altogether.Instead do the select first and then the insert.===Working with lag===Despite our best efforts, it's not practical to guarantee a low-lagenvironment.Lag will usually be less than one second, but mayoccasionally be up to 30 seconds.For scalability, it's very importantto keep load on the master low, so simply sending all your queries tothe master is not the answer.So when you have a genuine need forup-to-date data, the following approach is advised:1) Do a quick query to the master for a sequence number or timestamp 2) Run the full query on the slave and check if it matches the data you gotfrom the master 3) If it doesn't, run the full query on the masterTo avoid swamping the master every time the slaves lag, use of thisapproach should be kept to a minimum.In most cases you should just readfrom the slave and let the user deal with the delay.------------------------------------------------------------------------Lock contention------------------------------------------------------------------------Due to the high write rate on Wikipedia(and some other wikis), MediaWiki developers need to be very careful to structure their writesto avoid long-lasting locks.By default, MediaWiki opens a transactionat the first query, and commits it before the output is sent.Locks willbe held from the time when the query is done until the commit.So youcan reduce lock time by doing as much processing as possible before youdo your write queries.Often this approach is not good enough, and it becomes necessary toenclose small groups of queries in their own transaction.Use thefollowing syntax:$dbw=wfGetDB(DB_MASTER
globals txt Globals are evil The original MediaWiki code relied on globals for processing context far too often MediaWiki development since then has been a story of slowly moving context out of global variables and into objects Storing processing context in object member variables allows those objects to be reused in a much more flexible way Consider the elegance of
database rows
Definition: globals.txt:10
static singleton()
Definition: Collation.php:34
Abstract maintenance class for quickly writing and churning out maintenance scripts with minimal effo...
Definition: maintenance.txt:39
null for the local wiki Added in
Definition: hooks.txt:1555
getDB($db, $groups=[], $wiki=false)
Returns a database to be used by current maintenance script.
hasOption($name)
Checks to see if a particular param exists.
require_once RUN_MAINTENANCE_IF_MAIN
Definition: maintenance.txt:50
updateSortKeySizeHistogram($key)
when a variable name is used in a it is silently declared as a new local masking the global
Definition: design.txt:93
title
const DB_MASTER
Definition: defines.php:23
The ContentHandler facility adds support for arbitrary content types on wiki instead of relying on wikitext for everything It was introduced in MediaWiki Each kind of and so on Built in content types are
addOption($name, $description, $required=false, $withArg=false, $shortName=false, $multiOccurrence=false)
Add a parameter to the script.
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 as context as context $options
Definition: hooks.txt:1046
$res
Definition: database.txt:21
const NS_CATEGORY
Definition: Defines.php:70
and(b) You must cause any modified files to carry prominent notices stating that You changed the files
namespace and then decline to actually register it file or subcat img or subcat $title
Definition: hooks.txt:953
const NS_FILE
Definition: Defines.php:62
addDescription($text)
Set the description text.
static configuration should be added through ResourceLoaderGetConfigVars instead can be used to get the real title after the basic globals have been set up
Definition: hooks.txt:2159
This document is intended to provide useful advice for parties seeking to redistribute MediaWiki to end users It s targeted particularly at maintainers for Linux since it s been observed that distribution packages of MediaWiki often break We ve consistently had to recommend that users seeking support use official tarballs instead of their distribution s and this often solves whatever problem the user is having It would be nice if this could such as
Definition: distributors.txt:9
getOption($name, $default=null)
Get an option, or return the default.
output($out, $channel=null)
Throw some output to the user.
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 $wgServiceWiringFiles lists includes ServiceWiring php
Definition: injection.txt:35
deferred txt A few of the database updates required by various functions here can be deferred until after the result page is displayed to the user For updating the view updating the linked to tables after a etc PHP does not yet have any way to tell the server to actually return and disconnect while still running these but it might have such a feature in the future We handle these by creating a deferred update object and putting those objects on a global then executing the whole list after the page is displayed We don t do anything smart like collating updates to the same table or such because the list is almost always going to have just one item on if so it s not worth the trouble Since there is a job queue in the jobs table
Definition: deferred.txt:11
getBatchCondition($row, $dbw)
Return an SQL expression selecting rows which sort above the given row, assuming an ordering of cl_co...
$count
Maintenance script that will find all rows in the categorylinks table whose collation is out-of-date...
const DB_REPLICA
Definition: defines.php:22
</td >< td > &</td >< td > t want your writing to be edited mercilessly and redistributed at will
=Architecture==Two class hierarchies are used to provide the functionality associated with the different content models:*Content interface(and AbstractContent base class) define functionality that acts on the concrete content of a page, and *ContentHandler base class provides functionality specific to a content model, but not acting on concrete content.The most important function of ContentHandler is to act as a factory for the appropriate implementation of Content.These Content objects are to be used by MediaWiki everywhere, instead of passing page content around as text.All manipulation and analysis of page content must be done via the appropriate methods of the Content object.For each content model, a subclass of ContentHandler has to be registered with $wgContentHandlers.The ContentHandler object for a given content model can be obtained using ContentHandler::getForModelID($id).Also Title, WikiPage and Revision now have getContentHandler() methods for convenience.ContentHandler objects are singletons that provide functionality specific to the content type, but not directly acting on the content of some page.ContentHandler::makeEmptyContent() and ContentHandler::unserializeContent() can be used to create a Content object of the appropriate type.However, it is recommended to instead use WikiPage::getContent() resp.Revision::getContent() to get a page's content as a Content object.These two methods should be the ONLY way in which page content is accessed.Another important function of ContentHandler objects is to define custom action handlers for a content model, see ContentHandler::getActionOverrides().This is similar to what WikiPage::getActionOverrides() was already doing.==Serialization==With the ContentHandler facility, page content no longer has to be text based.Objects implementing the Content interface are used to represent and handle the content internally.For storage and data exchange, each content model supports at least one serialization format via ContentHandler::serializeContent($content).The list of supported formats for a given content model can be accessed using ContentHandler::getSupportedFormats().Content serialization formats are identified using MIME type like strings.The following formats are built in:*text/x-wiki-wikitext *text/javascript-for js pages *text/css-for css pages *text/plain-for future use, e.g.with plain text messages.*text/html-for future use, e.g.with plain html messages.*application/vnd.php.serialized-for future use with the api and for extensions *application/json-for future use with the api, and for use by extensions *application/xml-for future use with the api, and for use by extensions In PHP, use the corresponding CONTENT_FORMAT_XXX constant.Note that when using the API to access page content, especially action=edit, action=parse and action=query &prop=revisions, the model and format of the content should always be handled explicitly.Without that information, interpretation of the provided content is not reliable.The same applies to XML dumps generated via maintenance/dumpBackup.php or Special:Export.Also note that the API will provide encapsulated, serialized content-so if the API was called with format=json, and contentformat is also json(or rather, application/json), the page content is represented as a string containing an escaped json structure.Extensions that use JSON to serialize some types of page content may provide specialized API modules that allow access to that content in a more natural form.==Compatibility==The ContentHandler facility is introduced in a way that should allow all existing code to keep functioning at least for pages that contain wikitext or other text based content.However, a number of functions and hooks have been deprecated in favor of new versions that are aware of the page's content model, and will now generate warnings when used.Most importantly, the following functions have been deprecated:*Revisions::getText() is deprecated in favor Revisions::getContent()*WikiPage::getText() is deprecated in favor WikiPage::getContent() Also, the old Article::getContent()(which returns text) is superceded by Article::getContentObject().However, both methods should be avoided since they do not provide clean access to the page's actual content.For instance, they may return a system message for non-existing pages.Use WikiPage::getContent() instead.Code that relies on a textual representation of the page content should eventually be rewritten.However, ContentHandler::getContentText() provides a stop-gap that can be used to get text for a page.Its behavior is controlled by $wgContentHandlerTextFallback it
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 one of or reset my talk page
Definition: hooks.txt:2491
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 one of or reset my talk my contributions etc etc otherwise the built in rate limiting checks are if enabled allows for interception of redirect as a string mapping parameter names to values & $type
Definition: hooks.txt:2491
beginTransaction(IDatabase $dbw, $fname)
Begin a transcation on a DB.