44require_once __DIR__ .
'/../Maintenance.php';
63 parent::__construct();
65 $this->
addOption(
'type',
'Set compression type to either: gzip|concat',
false,
true,
't' );
68 'Maximum number of revisions in a concat chunk',
75 'Earliest date to check for uncompressed revisions',
80 $this->
addOption(
'end-date',
'Latest revision date to compress',
false,
true,
'e' );
83 'The id to start from (gzip -> text table, concat -> page table)',
90 'Store specified revisions in an external cluster (untested)',
96 'The page_id to stop at (only when using concat compression type)',
105 if ( !function_exists(
"gzdeflate" ) ) {
106 $this->
error(
"You must enable zlib support in PHP to compress old revisions!\n" .
107 "Please see http://www.php.net/manual/en/ref.zlib.php\n",
true );
111 $chunkSize = $this->
getOption(
'chunksize', 20 );
112 $startId = $this->
getOption(
'startid', 0 );
113 $beginDate = $this->
getOption(
'begin-date',
'' );
114 $endDate = $this->
getOption(
'end-date',
'' );
115 $extDB = $this->
getOption(
'extdb',
'' );
116 $endId = $this->
getOption(
'endid',
false );
119 $this->
error(
"Type \"{$type}\" not supported" );
122 if ( $extDB !=
'' ) {
123 $this->
output(
"Compressing database {$wgDBname} to external cluster {$extDB}\n"
124 . str_repeat(
'-', 76 ) .
"\n\n" );
126 $this->
output(
"Compressing database {$wgDBname}\n"
127 . str_repeat(
'-', 76 ) .
"\n\n" );
131 if (
$type ==
'concat' ) {
133 $endDate, $extDB, $endId );
139 $this->
output(
"Done.\n" );
151 $this->
output(
"Starting from old_id $start...\n" );
156 [
'old_id',
'old_flags',
'old_text' ],
159 [
'ORDER BY' =>
'old_id',
'LIMIT' => $chunksize,
'FOR UPDATE' ]
162 if (
$res->numRows() == 0 ) {
168 foreach (
$res as $row ) {
169 # print " {$row->old_id} - {$row->old_namespace}:{$row->old_title}\n";
171 $last = $row->old_id;
174 $start =
$last + 1; # Deletion may leave
long empty stretches
175 $this->
output(
"$start...\n" );
187 if (
false !== strpos( $row->old_flags,
'gzip' )
188 ||
false !== strpos( $row->old_flags,
'object' )
190 # print "Already compressed row {$row->old_id}\n";
194 $flags = $row->old_flags ?
"{$row->old_flags},gzip" :
"gzip";
195 $compress = gzdeflate( $row->old_text );
197 # Store in external storage if required
198 if ( $extdb !==
'' ) {
200 $compress = $storeObj->
store( $extdb, $compress );
201 if ( $compress ===
false ) {
202 $this->
error(
"Unable to store object" );
209 $dbw->update(
'text',
212 'old_text' => $compress
214 'old_id' => $row->old_id
234 $endDate, $extdb =
"", $maxPageId =
false
241 # Set up external storage
242 if ( $extdb !=
'' ) {
246 # Get all articles by page_id
248 $maxPageId =
$dbr->selectField(
'page',
'max(page_id)',
'', __METHOD__ );
250 $this->
output(
"Starting from $startId of $maxPageId\n" );
263 # For each article, get a list of revisions which fit the criteria
265 # No recompression, use a condition on old_flags
266 # Don't compress object type entities, because that might produce data loss when
267 # overwriting bulk storage concat rows. Don't compress external references, because
268 # the script doesn't yet delete rows from external storage.
270 'old_flags NOT ' .
$dbr->buildLike(
$dbr->anyString(),
'object',
$dbr->anyString() )
271 .
' AND old_flags NOT '
272 .
$dbr->buildLike(
$dbr->anyString(),
'external',
$dbr->anyString() )
276 if ( !preg_match(
'/^\d{14}$/', $beginDate ) ) {
277 $this->
error(
"Invalid begin date \"$beginDate\"\n" );
281 $conds[] =
"rev_timestamp>'" . $beginDate .
"'";
284 if ( !preg_match(
'/^\d{14}$/', $endDate ) ) {
285 $this->
error(
"Invalid end date \"$endDate\"\n" );
289 $conds[] =
"rev_timestamp<'" . $endDate .
"'";
291 if ( $loadStyle == self::LS_CHUNKED ) {
292 $tables = [
'revision',
'text' ];
293 $fields = [
'rev_id',
'rev_text_id',
'old_flags',
'old_text' ];
294 $conds[] =
'rev_text_id=old_id';
295 $revLoadOptions =
'FOR UPDATE';
298 $fields = [
'rev_id',
'rev_text_id' ];
299 $revLoadOptions = [];
302 # Don't work with current revisions
303 # Don't lock the page table for update either -- TS 2006-04-04
304 # $tables[] = 'page';
305 # $conds[] = 'page_id=rev_page AND rev_id != page_latest';
307 for ( $pageId = $startId; $pageId <= $maxPageId; $pageId++ ) {
314 $pageRes =
$dbr->select(
'page',
315 [
'page_id',
'page_namespace',
'page_title',
'page_latest' ],
316 $pageConds + [
'page_id' => $pageId ], __METHOD__ );
317 if ( $pageRes->numRows() == 0 ) {
320 $pageRow =
$dbr->fetchObject( $pageRes );
323 $titleObj = Title::makeTitle( $pageRow->page_namespace, $pageRow->page_title );
324 $this->
output(
"$pageId\t" . $titleObj->getPrefixedDBkey() .
" " );
327 $revRes = $dbw->select(
$tables, $fields,
329 'rev_page' => $pageRow->page_id,
330 # Don
't operate on the current revision
331 # Use < instead of <> in case the current revision has changed
332 # since the page select, which wasn't locking
333 'rev_id < ' . $pageRow->page_latest
339 foreach ( $revRes as $revRow ) {
343 if ( count( $revs ) < 2 ) {
344 # No revisions matching, no further processing
351 while ( $i < count( $revs ) ) {
352 if ( $i < count( $revs ) - $maxChunkSize ) {
353 $thisChunkSize = $maxChunkSize;
355 $thisChunkSize = count( $revs ) - $i;
362 $primaryOldid = $revs[$i]->rev_text_id;
365 # Get the text of each revision and add it to the object
366 for ( $j = 0; $j < $thisChunkSize && $chunk->isHappy(); $j++ ) {
368 $oldid = $revs[$i + $j]->rev_text_id;
371 if ( $loadStyle == self::LS_INDIVIDUAL ) {
372 $textRow = $dbw->selectRow(
'text',
373 [
'old_flags',
'old_text' ],
374 [
'old_id' => $oldid ],
383 if ( $text ===
false ) {
384 $this->
error(
"\nError, unable to get text in old_id $oldid" );
385 # $dbw->delete( 'old', [ 'old_id' => $oldid ] );
388 if ( $extdb ==
"" && $j == 0 ) {
389 $chunk->setText( $text );
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 ) {
399 $stub->setLocation( $primaryOldid );
400 $stub->setReferrer( $oldid );
409 # If we couldn't actually use any stubs because the pages were too small, do nothing
411 if ( $extdb !=
"" ) {
412 # Move blob objects to External Storage
413 $stored = $storeObj->store( $extdb,
serialize( $chunk ) );
414 if ( $stored ===
false ) {
415 $this->
error(
"Unable to store object" );
419 # Store External Storage URLs instead of Stub placeholders
420 foreach ( $stubs as $stub ) {
421 if ( $stub ===
false ) {
424 # $stored should provide base path to a BLOB
425 $url = $stored .
"/" . $stub->getHash();
426 $dbw->update(
'text',
429 'old_flags' =>
'external,utf-8',
431 'old_id' => $stub->getReferrer(),
436 # Store the main object locally
437 $dbw->update(
'text',
440 'old_flags' =>
'object,utf-8',
442 'old_id' => $primaryOldid
446 # Store the stub objects
447 for ( $j = 1; $j < $thisChunkSize; $j++ ) {
448 # Skip if not compressing and don't overwrite the first revision
449 if ( $stubs[$j] !==
false && $revs[$i + $j]->rev_text_id != $primaryOldid ) {
450 $dbw->update(
'text',
453 'old_flags' =>
'object,utf-8',
455 'old_id' => $revs[$i + $j]->rev_text_id
465 $i += $thisChunkSize;
wfWaitForSlaves( $ifWritesSince=null, $wiki=false, $cluster=false, $timeout=null)
Waits for the replica DBs to catch up to the master position.
Maintenance script that compress the text of a wiki.
execute()
Do the actual work.
compressPage( $row, $extdb)
Compress the text in gzip format.
__construct()
Default constructor.
const LS_CHUNKED
Option to load revisions in chunks.
compressWithConcat( $startId, $maxChunkSize, $beginDate, $endDate, $extdb="", $maxPageId=false)
Compress the text in chunks after concatenating the revisions.
compressOldPages( $start=0, $extdb='')
Fetch the text row-by-row to 'compressPage' function for compression.
const LS_INDIVIDUAL
Option to load each revision individually.
Concatenated gzip (CGZ) storage Improves compression ratio by concatenating like objects before gzipp...
DB accessable external objects.
store( $location, $data)
Insert a data item into a given location.
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...
beginTransaction(IDatabase $dbw, $fname)
Begin a transcation on a DB.
commitTransaction(IDatabase $dbw, $fname)
Commit the transcation on a DB handle and wait for replica DBs to catch up.
getDB( $db, $groups=[], $wiki=false)
Returns a database to be used by current maintenance script.
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.
static getRevisionText( $row, $prefix='old_', $wiki=false)
Get revision text associated with an old or archive row.
design txt This is a brief overview of the new design More thorough and up to date information is available on the documentation wiki at etc Handles the details of getting and saving to the user table of the and dealing with sessions and cookies OutputPage Encapsulates the entire HTML page that will be sent in response to any server request It is used by calling its functions to add in any and then calling output() to send it all. It could be easily changed to send incrementally if that becomes useful
this hook is for auditing only RecentChangesLinked and Watchlist RecentChangesLinked and Watchlist Do not use this to implement individual filters if they are compatible with the ChangesListFilter and ChangesListFilterGroup structure use sub classes of those in conjunction with the ChangesListSpecialPageStructuredFilters hook This hook can be used to implement filters that do not implement that or custom behavior that is not an individual filter e g Watchlist & $tables
do that in ParserLimitReportFormat instead use this to modify the parameters of the image all existing parser cache entries will be invalid To avoid you ll need to handle that somehow(e.g. with the RejectParserCacheValue hook) because MediaWiki won 't do it for you. & $defaults error
it s the revision text itself In either if gzip is the revision text is gzipped $flags
require_once RUN_MAINTENANCE_IF_MAIN
controlled by $wgMainCacheType controlled by $wgParserCacheType controlled by $wgMessageCacheType If you set CACHE_NONE to one of the three control default value for MediaWiki still create a but requests to it are no ops and we always fall through to the database If the cache daemon can t be it should also disable itself fairly smoothly By $wgMemc is used but when it is $parserMemc or $messageMemc this is mentioned $wgDBname