MediaWiki  master
compressOld.php
Go to the documentation of this file.
1 <?php
45 
46 require_once __DIR__ . '/../Maintenance.php';
47 
53 class CompressOld extends Maintenance {
54  public function __construct() {
55  parent::__construct();
56  $this->addDescription( 'Compress the text of a wiki' );
57  $this->addOption( 'type', 'Set compression type to either: gzip|concat', false, true, 't' );
58  $this->addOption(
59  'chunksize',
60  'Maximum number of revisions in a concat chunk',
61  false,
62  true,
63  'c'
64  );
65  $this->addOption(
66  'begin-date',
67  'Earliest date to check for uncompressed revisions',
68  false,
69  true,
70  'b'
71  );
72  $this->addOption( 'end-date', 'Latest revision date to compress', false, true, 'e' );
73  $this->addOption(
74  'startid',
75  'The id to start from (gzip -> text table, concat -> page table)',
76  false,
77  true,
78  's'
79  );
80  $this->addOption(
81  'extdb',
82  'Store specified revisions in an external cluster (untested)',
83  false,
84  true
85  );
86  $this->addOption(
87  'endid',
88  'The page_id to stop at (only when using concat compression type)',
89  false,
90  true,
91  'n'
92  );
93  }
94 
95  public function execute() {
96  global $wgDBname;
97  if ( !function_exists( "gzdeflate" ) ) {
98  $this->fatalError( "You must enable zlib support in PHP to compress old revisions!\n" .
99  "Please see https://www.php.net/manual/en/ref.zlib.php\n" );
100  }
101 
102  $type = $this->getOption( 'type', 'concat' );
103  $chunkSize = $this->getOption( 'chunksize', 20 );
104  $startId = $this->getOption( 'startid', 0 );
105  $beginDate = $this->getOption( 'begin-date', '' );
106  $endDate = $this->getOption( 'end-date', '' );
107  $extDB = $this->getOption( 'extdb', '' );
108  $endId = $this->getOption( 'endid', false );
109 
110  if ( $type != 'concat' && $type != 'gzip' ) {
111  $this->error( "Type \"{$type}\" not supported" );
112  }
113 
114  if ( $extDB != '' ) {
115  $this->output( "Compressing database {$wgDBname} to external cluster {$extDB}\n"
116  . str_repeat( '-', 76 ) . "\n\n" );
117  } else {
118  $this->output( "Compressing database {$wgDBname}\n"
119  . str_repeat( '-', 76 ) . "\n\n" );
120  }
121 
122  $success = true;
123  if ( $type == 'concat' ) {
124  $success = $this->compressWithConcat( $startId, $chunkSize, $beginDate,
125  $endDate, $extDB, $endId );
126  } else {
127  $this->compressOldPages( $startId, $extDB );
128  }
129 
130  if ( $success ) {
131  $this->output( "Done.\n" );
132  }
133  }
134 
141  private function compressOldPages( $start = 0, $extdb = '' ) {
142  $chunksize = 50;
143  $this->output( "Starting from old_id $start...\n" );
144  $dbw = $this->getDB( DB_PRIMARY );
145  do {
146  $res = $dbw->newSelectQueryBuilder()
147  ->select( [ 'old_id', 'old_flags', 'old_text' ] )
148  ->forUpdate()
149  ->from( 'text' )
150  ->where( "old_id>=$start" )
151  ->orderBy( 'old_id' )
152  ->limit( $chunksize )
153  ->caller( __METHOD__ )->fetchResultSet();
154 
155  if ( $res->numRows() == 0 ) {
156  break;
157  }
158 
159  $last = $start;
160 
161  foreach ( $res as $row ) {
162  # print " {$row->old_id} - {$row->old_namespace}:{$row->old_title}\n";
163  $this->compressPage( $row, $extdb );
164  $last = $row->old_id;
165  }
166 
167  $start = $last + 1; # Deletion may leave long empty stretches
168  $this->output( "$start...\n" );
169  } while ( true );
170  }
171 
179  private function compressPage( $row, $extdb ) {
180  if ( strpos( $row->old_flags, 'gzip' ) !== false
181  || strpos( $row->old_flags, 'object' ) !== false
182  ) {
183  # print "Already compressed row {$row->old_id}\n";
184  return false;
185  }
186  $dbw = $this->getDB( DB_PRIMARY );
187  $flags = $row->old_flags ? "{$row->old_flags},gzip" : "gzip";
188  $compress = gzdeflate( $row->old_text );
189 
190  # Store in external storage if required
191  if ( $extdb !== '' ) {
192  $esFactory = $this->getServiceContainer()->getExternalStoreFactory();
194  $storeObj = $esFactory->getStore( 'DB' );
195  $compress = $storeObj->store( $extdb, $compress );
196  if ( $compress === false ) {
197  $this->error( "Unable to store object" );
198 
199  return false;
200  }
201  }
202 
203  # Update text row
204  $dbw->update( 'text',
205  [ /* SET */
206  'old_flags' => $flags,
207  'old_text' => $compress
208  ], [ /* WHERE */
209  'old_id' => $row->old_id
210  ], __METHOD__,
211  [ 'LIMIT' => 1 ]
212  );
213 
214  return true;
215  }
216 
228  private function compressWithConcat( $startId, $maxChunkSize, $beginDate,
229  $endDate, $extdb = "", $maxPageId = false
230  ) {
231  $dbr = $this->getDB( DB_REPLICA );
232  $dbw = $this->getDB( DB_PRIMARY );
233 
234  # Set up external storage
235  if ( $extdb != '' ) {
236  $esFactory = $this->getServiceContainer()->getExternalStoreFactory();
238  $storeObj = $esFactory->getStore( 'DB' );
239  }
240 
241  $blobStore = $this->getServiceContainer()
242  ->getBlobStoreFactory()
243  ->newSqlBlobStore();
244 
245  # Get all articles by page_id
246  if ( !$maxPageId ) {
247  $maxPageId = $dbr->newSelectQueryBuilder()
248  ->select( 'max(page_id)' )
249  ->from( 'page' )
250  ->caller( __METHOD__ )->fetchField();
251  }
252  $this->output( "Starting from $startId of $maxPageId\n" );
253  $pageConds = [];
254 
255  /*
256  if ( $exclude_ns0 ) {
257  print "Excluding main namespace\n";
258  $pageConds[] = 'page_namespace<>0';
259  }
260  if ( $queryExtra ) {
261  $pageConds[] = $queryExtra;
262  }
263  */
264 
265  # For each article, get a list of revisions which fit the criteria
266 
267  # No recompression, use a condition on old_flags
268  # Don't compress object type entities, because that might produce data loss when
269  # overwriting bulk storage concat rows. Don't compress external references, because
270  # the script doesn't yet delete rows from external storage.
271  $conds = [
272  'old_flags NOT ' . $dbr->buildLike( $dbr->anyString(), 'object', $dbr->anyString() )
273  . ' AND old_flags NOT '
274  . $dbr->buildLike( $dbr->anyString(), 'external', $dbr->anyString() )
275  ];
276 
277  if ( $beginDate ) {
278  if ( !preg_match( '/^\d{14}$/', $beginDate ) ) {
279  $this->error( "Invalid begin date \"$beginDate\"\n" );
280 
281  return false;
282  }
283  $conds[] = "rev_timestamp>'" . $beginDate . "'";
284  }
285  if ( $endDate ) {
286  if ( !preg_match( '/^\d{14}$/', $endDate ) ) {
287  $this->error( "Invalid end date \"$endDate\"\n" );
288 
289  return false;
290  }
291  $conds[] = "rev_timestamp<'" . $endDate . "'";
292  }
293 
294  $slotRoleStore = $this->getServiceContainer()->getSlotRoleStore();
295  $tables = [ 'revision', 'slots', 'content', 'text' ];
296  $conds = array_merge( [
297  'rev_id=slot_revision_id',
298  'slot_role_id=' . $slotRoleStore->getId( SlotRecord::MAIN ),
299  'content_id=slot_content_id',
300  'SUBSTRING(content_address, 1, 3)=' . $dbr->addQuotes( 'tt:' ),
301  'SUBSTRING(content_address, 4)=old_id',
302  ], $conds );
303 
304  $fields = [ 'rev_id', 'old_id', 'old_flags', 'old_text' ];
305  $revLoadOptions = 'FOR UPDATE';
306 
307  # Don't work with current revisions
308  # Don't lock the page table for update either -- TS 2006-04-04
309  # $tables[] = 'page';
310  # $conds[] = 'page_id=rev_page AND rev_id != page_latest';
311 
312  for ( $pageId = $startId; $pageId <= $maxPageId; $pageId++ ) {
313  $this->waitForReplication();
314 
315  # Wake up
316  $dbr->ping();
317 
318  # Get the page row
319  $pageRow = $dbr->newSelectQueryBuilder()
320  ->select( [ 'page_id', 'page_namespace', 'page_title', 'rev_timestamp' ] )
321  ->from( 'page' )
322  ->straightJoin( 'revision', null, 'page_latest = rev_id' )
323  ->where( $pageConds )
324  ->andWhere( [ 'page_id' => $pageId ] )
325  ->caller( __METHOD__ )->fetchRow();
326  if ( $pageRow === false ) {
327  continue;
328  }
329 
330  # Display progress
331  $titleObj = Title::makeTitle( $pageRow->page_namespace, $pageRow->page_title );
332  $this->output( "$pageId\t" . $titleObj->getPrefixedDBkey() . " " );
333 
334  # Load revisions
335  $revRes = $dbw->select( $tables, $fields,
336  array_merge( [
337  'rev_page' => $pageRow->page_id,
338  # Don't operate on the current revision
339  # Use < instead of <> in case the current revision has changed
340  # since the page select, which wasn't locking
341  'rev_timestamp < ' . (int)$pageRow->rev_timestamp
342  ], $conds ),
343  __METHOD__,
344  $revLoadOptions
345  );
346  $revs = [];
347  foreach ( $revRes as $revRow ) {
348  $revs[] = $revRow;
349  }
350 
351  if ( count( $revs ) < 2 ) {
352  # No revisions matching, no further processing
353  $this->output( "\n" );
354  continue;
355  }
356 
357  # For each chunk
358  $i = 0;
359  while ( $i < count( $revs ) ) {
360  if ( $i < count( $revs ) - $maxChunkSize ) {
361  $thisChunkSize = $maxChunkSize;
362  } else {
363  $thisChunkSize = count( $revs ) - $i;
364  }
365 
366  $chunk = new ConcatenatedGzipHistoryBlob();
367  $stubs = [];
368  $this->beginTransaction( $dbw, __METHOD__ );
369  $usedChunk = false;
370  $primaryOldid = $revs[$i]->old_id;
371 
372  # Get the text of each revision and add it to the object
373  for ( $j = 0; $j < $thisChunkSize && $chunk->isHappy(); $j++ ) {
374  $oldid = $revs[$i + $j]->old_id;
375 
376  # Get text. We do not need the full `extractBlob` since the query is built
377  # to fetch non-externalstore blobs.
378  $text = $blobStore->decompressData(
379  $revs[$i + $j]->old_text,
380  explode( ',', $revs[$i + $j]->old_flags )
381  );
382 
383  if ( $text === false ) {
384  $this->error( "\nError, unable to get text in old_id $oldid" );
385  # $dbw->delete( 'old', [ 'old_id' => $oldid ] );
386  }
387 
388  if ( $extdb == "" && $j == 0 ) {
389  $chunk->setText( $text );
390  $this->output( '.' );
391  } else {
392  # Don't make a stub if it's going to be longer than the article
393  # Stubs are typically about 100 bytes
394  if ( strlen( $text ) < 120 ) {
395  $stub = false;
396  $this->output( 'x' );
397  } else {
398  $stub = new HistoryBlobStub( $chunk->addItem( $text ) );
399  $stub->setLocation( $primaryOldid );
400  $stub->setReferrer( $oldid );
401  $this->output( '.' );
402  $usedChunk = true;
403  }
404  $stubs[$j] = $stub;
405  }
406  }
407  $thisChunkSize = $j;
408 
409  # If we couldn't actually use any stubs because the pages were too small, do nothing
410  if ( $usedChunk ) {
411  if ( $extdb != "" ) {
412  # Move blob objects to External Storage
413  // @phan-suppress-next-line PhanPossiblyUndeclaredVariable storeObj is set when used
414  $stored = $storeObj->store( $extdb, serialize( $chunk ) );
415  if ( $stored === false ) {
416  $this->error( "Unable to store object" );
417 
418  return false;
419  }
420  # Store External Storage URLs instead of Stub placeholders
421  foreach ( $stubs as $stub ) {
422  if ( $stub === false ) {
423  continue;
424  }
425  # $stored should provide base path to a BLOB
426  $url = $stored . "/" . $stub->getHash();
427  $dbw->update( 'text',
428  [ /* SET */
429  'old_text' => $url,
430  'old_flags' => 'external,utf-8',
431  ], [ /* WHERE */
432  'old_id' => $stub->getReferrer(),
433  ],
434  __METHOD__
435  );
436  }
437  } else {
438  # Store the main object locally
439  $dbw->update( 'text',
440  [ /* SET */
441  'old_text' => serialize( $chunk ),
442  'old_flags' => 'object,utf-8',
443  ], [ /* WHERE */
444  'old_id' => $primaryOldid
445  ],
446  __METHOD__
447  );
448 
449  # Store the stub objects
450  for ( $j = 1; $j < $thisChunkSize; $j++ ) {
451  # Skip if not compressing and don't overwrite the first revision
452  if ( $stubs[$j] !== false && $revs[$i + $j]->old_id != $primaryOldid ) {
453  $dbw->update( 'text',
454  [ /* SET */
455  'old_text' => serialize( $stubs[$j] ),
456  'old_flags' => 'object,utf-8',
457  ], [ /* WHERE */
458  'old_id' => $revs[$i + $j]->old_id
459  ],
460  __METHOD__
461  );
462  }
463  }
464  }
465  }
466  # Done, next
467  $this->output( "/" );
468  $this->commitTransaction( $dbw, __METHOD__ );
469  $i += $thisChunkSize;
470  }
471  $this->output( "\n" );
472  }
473 
474  return true;
475  }
476 }
477 
478 $maintClass = CompressOld::class;
479 require_once RUN_MAINTENANCE_IF_MAIN;
$success
Maintenance script that compress the text of a wiki.
Definition: compressOld.php:53
execute()
Do the actual work.
Definition: compressOld.php:95
__construct()
Default constructor.
Definition: compressOld.php:54
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...
Definition: Maintenance.php:66
getDB( $db, $groups=[], $dbDomain=false)
Returns a database to be used by current maintenance script.
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.
Definition: SlotRecord.php:40
Represents a title within MediaWiki.
Definition: Title.php:76
$maintClass
$wgDBname
Config variable stub for the DBname setting, for use by phpdoc and IDEs.
const DB_REPLICA
Definition: defines.php:26
const DB_PRIMARY
Definition: defines.php:28