MediaWiki master
compressOld.php
Go to the documentation of this file.
1<?php
47
48require_once __DIR__ . '/../Maintenance.php';
49
55class CompressOld extends Maintenance {
56 public function __construct() {
57 parent::__construct();
58 $this->addDescription( 'Compress the text of a wiki' );
59 $this->addOption( 'type', 'Set compression type to either: gzip|concat', false, true, 't' );
60 $this->addOption(
61 'chunksize',
62 'Maximum number of revisions in a concat chunk',
63 false,
64 true,
65 'c'
66 );
67 $this->addOption(
68 'begin-date',
69 'Earliest date to check for uncompressed revisions',
70 false,
71 true,
72 'b'
73 );
74 $this->addOption( 'end-date', 'Latest revision date to compress', false, true, 'e' );
75 $this->addOption(
76 'startid',
77 'The id to start from (gzip -> text table, concat -> page table)',
78 false,
79 true,
80 's'
81 );
82 $this->addOption(
83 'extdb',
84 'Store specified revisions in an external cluster (untested)',
85 false,
86 true
87 );
88 $this->addOption(
89 'endid',
90 'The page_id to stop at (only when using concat compression type)',
91 false,
92 true,
93 'n'
94 );
95 }
96
97 public function execute() {
98 global $wgDBname;
99 if ( !function_exists( "gzdeflate" ) ) {
100 $this->fatalError( "You must enable zlib support in PHP to compress old revisions!\n" .
101 "Please see https://www.php.net/manual/en/ref.zlib.php\n" );
102 }
103
104 $type = $this->getOption( 'type', 'concat' );
105 $chunkSize = $this->getOption( 'chunksize', 20 );
106 $startId = $this->getOption( 'startid', 0 );
107 $beginDate = $this->getOption( 'begin-date', '' );
108 $endDate = $this->getOption( 'end-date', '' );
109 $extDB = $this->getOption( 'extdb', '' );
110 $endId = $this->getOption( 'endid', false );
111
112 if ( $type != 'concat' && $type != 'gzip' ) {
113 $this->error( "Type \"{$type}\" not supported" );
114 }
115
116 if ( $extDB != '' ) {
117 $this->output( "Compressing database {$wgDBname} to external cluster {$extDB}\n"
118 . str_repeat( '-', 76 ) . "\n\n" );
119 } else {
120 $this->output( "Compressing database {$wgDBname}\n"
121 . str_repeat( '-', 76 ) . "\n\n" );
122 }
123
124 $success = true;
125 if ( $type == 'concat' ) {
126 $success = $this->compressWithConcat( $startId, $chunkSize, $beginDate,
127 $endDate, $extDB, $endId );
128 } else {
129 $this->compressOldPages( $startId, $extDB );
130 }
131
132 if ( $success ) {
133 $this->output( "Done.\n" );
134 }
135 }
136
143 private function compressOldPages( $start = 0, $extdb = '' ) {
144 $chunksize = 50;
145 $this->output( "Starting from old_id $start...\n" );
146 $dbw = $this->getPrimaryDB();
147 do {
148 $res = $dbw->newSelectQueryBuilder()
149 ->select( [ 'old_id', 'old_flags', 'old_text' ] )
150 ->forUpdate()
151 ->from( 'text' )
152 ->where( "old_id>=$start" )
153 ->orderBy( 'old_id' )
154 ->limit( $chunksize )
155 ->caller( __METHOD__ )->fetchResultSet();
156
157 if ( $res->numRows() == 0 ) {
158 break;
159 }
160
161 $last = $start;
162
163 foreach ( $res as $row ) {
164 # print " {$row->old_id} - {$row->old_namespace}:{$row->old_title}\n";
165 $this->compressPage( $row, $extdb );
166 $last = $row->old_id;
167 }
168
169 $start = $last + 1; # Deletion may leave long empty stretches
170 $this->output( "$start...\n" );
171 } while ( true );
172 }
173
181 private function compressPage( $row, $extdb ) {
182 if ( strpos( $row->old_flags, 'gzip' ) !== false
183 || strpos( $row->old_flags, 'object' ) !== false
184 ) {
185 # print "Already compressed row {$row->old_id}\n";
186 return false;
187 }
188 $dbw = $this->getPrimaryDB();
189 $flags = $row->old_flags ? "{$row->old_flags},gzip" : "gzip";
190 $compress = gzdeflate( $row->old_text );
191
192 # Store in external storage if required
193 if ( $extdb !== '' ) {
194 $esFactory = $this->getServiceContainer()->getExternalStoreFactory();
196 $storeObj = $esFactory->getStore( 'DB' );
197 $compress = $storeObj->store( $extdb, $compress );
198 if ( $compress === false ) {
199 $this->error( "Unable to store object" );
200
201 return false;
202 }
203 }
204
205 # Update text row
206 $dbw->update( 'text',
207 [ /* SET */
208 'old_flags' => $flags,
209 'old_text' => $compress
210 ], [ /* WHERE */
211 'old_id' => $row->old_id
212 ], __METHOD__,
213 [ 'LIMIT' => 1 ]
214 );
215
216 return true;
217 }
218
230 private function compressWithConcat( $startId, $maxChunkSize, $beginDate,
231 $endDate, $extdb = "", $maxPageId = false
232 ) {
233 $dbr = $this->getReplicaDB();
234 $dbw = $this->getPrimaryDB();
235
236 # Set up external storage
237 if ( $extdb != '' ) {
238 $esFactory = $this->getServiceContainer()->getExternalStoreFactory();
240 $storeObj = $esFactory->getStore( 'DB' );
241 }
242
243 $blobStore = $this->getServiceContainer()
244 ->getBlobStoreFactory()
245 ->newSqlBlobStore();
246
247 # Get all articles by page_id
248 if ( !$maxPageId ) {
249 $maxPageId = $dbr->newSelectQueryBuilder()
250 ->select( 'max(page_id)' )
251 ->from( 'page' )
252 ->caller( __METHOD__ )->fetchField();
253 }
254 $this->output( "Starting from $startId of $maxPageId\n" );
255 $pageConds = [];
256
257 /*
258 if ( $exclude_ns0 ) {
259 print "Excluding main namespace\n";
260 $pageConds[] = 'page_namespace<>0';
261 }
262 if ( $queryExtra ) {
263 $pageConds[] = $queryExtra;
264 }
265 */
266
267 # For each article, get a list of revisions which fit the criteria
268
269 # No recompression, use a condition on old_flags
270 # Don't compress object type entities, because that might produce data loss when
271 # overwriting bulk storage concat rows. Don't compress external references, because
272 # the script doesn't yet delete rows from external storage.
273 $slotRoleStore = $this->getServiceContainer()->getSlotRoleStore();
274 $queryBuilderTemplate = $dbw->newSelectQueryBuilder()
275 ->select( [ 'rev_id', 'old_id', 'old_flags', 'old_text' ] )
276 ->forUpdate()
277 ->from( 'revision' )
278 ->join( 'slots', null, 'rev_id=slot_revision_id' )
279 ->join( 'content', null, 'content_id=slot_content_id' )
280 ->join( 'text', null, 'SUBSTRING(content_address, 4)=old_id' )
281 ->where(
282 $dbr->expr(
283 'old_flags',
284 IExpression::NOT_LIKE,
285 new LikeValue( $dbr->anyString(), 'object', $dbr->anyString() )
286 )->and(
287 'old_flags',
288 IExpression::NOT_LIKE,
289 new LikeValue( $dbr->anyString(), 'external', $dbr->anyString() )
290 )
291 )
292 ->andWhere( [
293 'slot_role_id' => $slotRoleStore->getId( SlotRecord::MAIN ),
294 'SUBSTRING(content_address, 1, 3)=' . $dbr->addQuotes( 'tt:' ),
295 ] );
296
297 if ( $beginDate ) {
298 if ( !preg_match( '/^\d{14}$/', $beginDate ) ) {
299 $this->error( "Invalid begin date \"$beginDate\"\n" );
300
301 return false;
302 }
303 $queryBuilderTemplate->andWhere( "rev_timestamp>'" . $beginDate . "'" );
304 }
305 if ( $endDate ) {
306 if ( !preg_match( '/^\d{14}$/', $endDate ) ) {
307 $this->error( "Invalid end date \"$endDate\"\n" );
308
309 return false;
310 }
311 $queryBuilderTemplate->andWhere( "rev_timestamp<'" . $endDate . "'" );
312 }
313
314 for ( $pageId = $startId; $pageId <= $maxPageId; $pageId++ ) {
315 $this->waitForReplication();
316
317 # Wake up
318 $dbr->ping();
319
320 # Get the page row
321 $pageRow = $dbr->newSelectQueryBuilder()
322 ->select( [ 'page_id', 'page_namespace', 'page_title', 'rev_timestamp' ] )
323 ->from( 'page' )
324 ->straightJoin( 'revision', null, 'page_latest = rev_id' )
325 ->where( $pageConds )
326 ->andWhere( [ 'page_id' => $pageId ] )
327 ->caller( __METHOD__ )->fetchRow();
328 if ( $pageRow === false ) {
329 continue;
330 }
331
332 # Display progress
333 $titleObj = Title::makeTitle( $pageRow->page_namespace, $pageRow->page_title );
334 $this->output( "$pageId\t" . $titleObj->getPrefixedDBkey() . " " );
335
336 # Load revisions
337 $queryBuilder = clone $queryBuilderTemplate;
338 $revRes = $queryBuilder->where(
339 [
340 'rev_page' => $pageRow->page_id,
341 // Don't operate on the current revision
342 // Use < instead of <> in case the current revision has changed
343 // since the page select, which wasn't locking
344 'rev_timestamp < ' . (int)$pageRow->rev_timestamp
345 ] )
346 ->caller( __METHOD__ )->fetchResultSet();
347
348 $revs = [];
349 foreach ( $revRes as $revRow ) {
350 $revs[] = $revRow;
351 }
352
353 if ( count( $revs ) < 2 ) {
354 # No revisions matching, no further processing
355 $this->output( "\n" );
356 continue;
357 }
358
359 # For each chunk
360 $i = 0;
361 while ( $i < count( $revs ) ) {
362 if ( $i < count( $revs ) - $maxChunkSize ) {
363 $thisChunkSize = $maxChunkSize;
364 } else {
365 $thisChunkSize = count( $revs ) - $i;
366 }
367
368 $chunk = new ConcatenatedGzipHistoryBlob();
369 $stubs = [];
370 $this->beginTransaction( $dbw, __METHOD__ );
371 $usedChunk = false;
372 $primaryOldid = $revs[$i]->old_id;
373
374 # Get the text of each revision and add it to the object
375 for ( $j = 0; $j < $thisChunkSize && $chunk->isHappy(); $j++ ) {
376 $oldid = $revs[$i + $j]->old_id;
377
378 # Get text. We do not need the full `extractBlob` since the query is built
379 # to fetch non-externalstore blobs.
380 $text = $blobStore->decompressData(
381 $revs[$i + $j]->old_text,
382 explode( ',', $revs[$i + $j]->old_flags )
383 );
384
385 if ( $text === false ) {
386 $this->error( "\nError, unable to get text in old_id $oldid" );
387 # $dbw->delete( 'old', [ 'old_id' => $oldid ] );
388 }
389
390 if ( $extdb == "" && $j == 0 ) {
391 $chunk->setText( $text );
392 $this->output( '.' );
393 } else {
394 # Don't make a stub if it's going to be longer than the article
395 # Stubs are typically about 100 bytes
396 if ( strlen( $text ) < 120 ) {
397 $stub = false;
398 $this->output( 'x' );
399 } else {
400 $stub = new HistoryBlobStub( $chunk->addItem( $text ) );
401 $stub->setLocation( $primaryOldid );
402 $stub->setReferrer( $oldid );
403 $this->output( '.' );
404 $usedChunk = true;
405 }
406 $stubs[$j] = $stub;
407 }
408 }
409 $thisChunkSize = $j;
410
411 # If we couldn't actually use any stubs because the pages were too small, do nothing
412 if ( $usedChunk ) {
413 if ( $extdb != "" ) {
414 # Move blob objects to External Storage
415 // @phan-suppress-next-line PhanPossiblyUndeclaredVariable storeObj is set when used
416 $stored = $storeObj->store( $extdb, serialize( $chunk ) );
417 if ( $stored === false ) {
418 $this->error( "Unable to store object" );
419
420 return false;
421 }
422 # Store External Storage URLs instead of Stub placeholders
423 foreach ( $stubs as $stub ) {
424 if ( $stub === false ) {
425 continue;
426 }
427 # $stored should provide base path to a BLOB
428 $url = $stored . "/" . $stub->getHash();
429 $dbw->update( 'text',
430 [ /* SET */
431 'old_text' => $url,
432 'old_flags' => 'external,utf-8',
433 ], [ /* WHERE */
434 'old_id' => $stub->getReferrer(),
435 ],
436 __METHOD__
437 );
438 }
439 } else {
440 # Store the main object locally
441 $dbw->update( 'text',
442 [ /* SET */
443 'old_text' => serialize( $chunk ),
444 'old_flags' => 'object,utf-8',
445 ], [ /* WHERE */
446 'old_id' => $primaryOldid
447 ],
448 __METHOD__
449 );
450
451 # Store the stub objects
452 for ( $j = 1; $j < $thisChunkSize; $j++ ) {
453 # Skip if not compressing and don't overwrite the first revision
454 if ( $stubs[$j] !== false && $revs[$i + $j]->old_id != $primaryOldid ) {
455 $dbw->update( 'text',
456 [ /* SET */
457 'old_text' => serialize( $stubs[$j] ),
458 'old_flags' => 'object,utf-8',
459 ], [ /* WHERE */
460 'old_id' => $revs[$i + $j]->old_id
461 ],
462 __METHOD__
463 );
464 }
465 }
466 }
467 }
468 # Done, next
469 $this->output( "/" );
470 $this->commitTransaction( $dbw, __METHOD__ );
471 $i += $thisChunkSize;
472 }
473 $this->output( "\n" );
474 }
475
476 return true;
477 }
478}
479
480$maintClass = CompressOld::class;
481require_once RUN_MAINTENANCE_IF_MAIN;
Maintenance script that compress the text of a wiki.
execute()
Do the actual work.
__construct()
Default constructor.
Concatenated gzip (CGZ) storage Improves compression ratio by concatenating like objects before gzipp...
Pointer object for an item within a CGZ blob stored in the text table.
Abstract maintenance class for quickly writing and churning out maintenance scripts with minimal effo...
error( $err, $die=0)
Throw an error to the user.
beginTransaction(IDatabase $dbw, $fname)
Begin a transaction on a DB.
commitTransaction(IDatabase $dbw, $fname)
Commit the transaction on a DB handle and wait for replica DBs to catch up.
output( $out, $channel=null)
Throw some output to the user.
waitForReplication()
Wait for replica DBs to catch up.
getServiceContainer()
Returns the main service container.
addDescription( $text)
Set the description text.
addOption( $name, $description, $required=false, $withArg=false, $shortName=false, $multiOccurrence=false)
Add a parameter to the script.
getOption( $name, $default=null)
Get an option, or return the default.
fatalError( $msg, $exitCode=1)
Output a message and terminate the current script.
Value object representing a content slot associated with a page revision.
Represents a title within MediaWiki.
Definition Title.php:78
Content of like value.
Definition LikeValue.php:14
$maintClass
$wgDBname
Config variable stub for the DBname setting, for use by phpdoc and IDEs.