MediaWiki  master
BackupDumper.php
Go to the documentation of this file.
1 <?php
28 require_once __DIR__ . '/../Maintenance.php';
29 require_once __DIR__ . '/../../includes/export/WikiExporter.php';
30 
34 
39 abstract class BackupDumper extends Maintenance {
40  public $reporting = true;
41  public $pages = null; // all pages
42  public $skipHeader = false; // don't output <mediawiki> and <siteinfo>
43  public $skipFooter = false; // don't output </mediawiki>
44  public $startId = 0;
45  public $endId = 0;
46  public $revStartId = 0;
47  public $revEndId = 0;
48  public $dumpUploads = false;
49  public $dumpUploadFileContents = false;
50  public $orderRevs = false;
51  public $limitNamespaces = [];
53  public $stderr;
54 
55  protected $reportingInterval = 100;
56  protected $pageCount = 0;
57  protected $revCount = 0;
58  protected $schemaVersion = null; // use default
59  protected $server = null; // use default
60  protected $sink = null; // Output filters
61  protected $lastTime = 0;
62  protected $pageCountLast = 0;
63  protected $revCountLast = 0;
64 
65  protected $outputTypes = [];
66  protected $filterTypes = [];
67 
68  protected $ID = 0;
69 
71  protected $startTime;
73  protected $pageCountPart;
75  protected $revCountPart;
77  protected $maxCount;
79  protected $timeOfCheckpoint;
81  protected $egress;
83  protected $buffer;
85  protected $openElement;
87  protected $atStart;
89  protected $thisRevModel;
91  protected $thisRevFormat;
93  protected $lastName;
95  protected $state;
96 
104  protected $forcedDb = null;
105 
107  protected $lb;
108 
112  public function __construct( $args = null ) {
113  parent::__construct();
114  $this->stderr = fopen( "php://stderr", "wt" );
115 
116  // Built-in output and filter plugins
117  $this->registerOutput( 'file', DumpFileOutput::class );
118  $this->registerOutput( 'gzip', DumpGZipOutput::class );
119  $this->registerOutput( 'bzip2', DumpBZip2Output::class );
120  $this->registerOutput( 'dbzip2', DumpDBZip2Output::class );
121  $this->registerOutput( 'lbzip2', DumpLBZip2Output::class );
122  $this->registerOutput( '7zip', Dump7ZipOutput::class );
123 
124  $this->registerFilter( 'latest', DumpLatestFilter::class );
125  $this->registerFilter( 'notalk', DumpNotalkFilter::class );
126  $this->registerFilter( 'namespace', DumpNamespaceFilter::class );
127 
128  // These three can be specified multiple times
129  $this->addOption( 'plugin', 'Load a dump plugin class. Specify as <class>[:<file>].',
130  false, true, false, true );
131  $this->addOption( 'output', 'Begin a filtered output stream; Specify as <type>:<file>. ' .
132  '<type>s: file, gzip, bzip2, 7zip, dbzip2, lbzip2', false, true, false, true );
133  $this->addOption( 'filter', 'Add a filter on an output branch. Specify as ' .
134  '<type>[:<options>]. <types>s: latest, notalk, namespace', false, true, false, true );
135  $this->addOption( 'report', 'Report position and speed after every n pages processed. ' .
136  'Default: 100.', false, true );
137  $this->addOption( 'schema-version', 'Schema version to use for output. ' .
138  'Default: ' . WikiExporter::schemaVersion(), false, true );
139  $this->addOption( 'server', 'Force reading from MySQL server', false, true );
140  $this->addOption( '7ziplevel', '7zip compression level for all 7zip outputs. Used for ' .
141  '-mx option to 7za command.', false, true );
142 
143  if ( $args ) {
144  // Args should be loaded and processed so that dump() can be called directly
145  // instead of execute()
146  $this->loadWithArgv( $args );
147  $this->processOptions();
148  }
149  }
150 
155  function registerOutput( $name, $class ) {
156  $this->outputTypes[$name] = $class;
157  }
158 
163  function registerFilter( $name, $class ) {
164  $this->filterTypes[$name] = $class;
165  }
166 
174  function loadPlugin( $class, $file ) {
175  if ( $file != '' ) {
176  require_once $file;
177  }
178  $register = [ $class, 'register' ];
179  $register( $this );
180  }
181 
182  public function execute() {
183  throw new MWException( 'execute() must be overridden in subclasses' );
184  }
185 
189  function processOptions() {
190  $sink = null;
191  $sinks = [];
192 
193  $this->schemaVersion = WikiExporter::schemaVersion();
194 
195  $options = $this->orderedOptions;
196  foreach ( $options as $arg ) {
197  list( $opt, $param ) = $arg;
198 
199  switch ( $opt ) {
200  case 'plugin':
201  $val = explode( ':', $param, 2 );
202 
203  if ( count( $val ) === 1 ) {
204  $this->loadPlugin( $val[0], '' );
205  } elseif ( count( $val ) === 2 ) {
206  $this->loadPlugin( $val[0], $val[1] );
207  }
208 
209  break;
210  case 'output':
211  $split = explode( ':', $param, 2 );
212  if ( count( $split ) !== 2 ) {
213  $this->fatalError( 'Invalid output parameter' );
214  }
215  list( $type, $file ) = $split;
216  if ( !is_null( $sink ) ) {
217  $sinks[] = $sink;
218  }
219  if ( !isset( $this->outputTypes[$type] ) ) {
220  $this->fatalError( "Unrecognized output sink type '$type'" );
221  }
222  $class = $this->outputTypes[$type];
223  if ( $type === "7zip" ) {
224  $sink = new $class( $file, intval( $this->getOption( '7ziplevel' ) ) );
225  } else {
226  $sink = new $class( $file );
227  }
228 
229  break;
230  case 'filter':
231  if ( is_null( $sink ) ) {
232  $sink = new DumpOutput();
233  }
234 
235  $split = explode( ':', $param, 2 );
236  $key = $split[0];
237 
238  if ( !isset( $this->filterTypes[$key] ) ) {
239  $this->fatalError( "Unrecognized filter type '$key'" );
240  }
241 
242  $type = $this->filterTypes[$key];
243 
244  if ( count( $split ) === 1 ) {
245  $filter = new $type( $sink );
246  } elseif ( count( $split ) === 2 ) {
247  $filter = new $type( $sink, $split[1] );
248  }
249 
250  // references are lame in php...
251  unset( $sink );
252  $sink = $filter;
253 
254  break;
255  case 'schema-version':
256  if ( !in_array( $param, XmlDumpWriter::$supportedSchemas ) ) {
257  $this->fatalError(
258  "Unsupported schema version $param. Supported versions: " .
259  implode( ', ', XmlDumpWriter::$supportedSchemas )
260  );
261  }
262  $this->schemaVersion = $param;
263  break;
264  }
265  }
266 
267  if ( $this->hasOption( 'report' ) ) {
268  $this->reportingInterval = intval( $this->getOption( 'report' ) );
269  }
270 
271  if ( $this->hasOption( 'server' ) ) {
272  $this->server = $this->getOption( 'server' );
273  }
274 
275  if ( is_null( $sink ) ) {
276  $sink = new DumpOutput();
277  }
278  $sinks[] = $sink;
279 
280  if ( count( $sinks ) > 1 ) {
281  $this->sink = new DumpMultiWriter( $sinks );
282  } else {
283  $this->sink = $sink;
284  }
285  }
286 
287  function dump( $history, $text = WikiExporter::TEXT ) {
288  # Notice messages will foul up your XML output even if they're
289  # relatively harmless.
290  if ( ini_get( 'display_errors' ) ) {
291  ini_set( 'display_errors', 'stderr' );
292  }
293 
294  $this->initProgress( $history );
295 
296  $db = $this->backupDb();
297  $exporter = new WikiExporter( $db, $history, $text, $this->limitNamespaces );
298  $exporter->setSchemaVersion( $this->schemaVersion );
299  $exporter->dumpUploads = $this->dumpUploads;
300  $exporter->dumpUploadFileContents = $this->dumpUploadFileContents;
301 
302  $wrapper = new ExportProgressFilter( $this->sink, $this );
303  $exporter->setOutputSink( $wrapper );
304 
305  if ( !$this->skipHeader ) {
306  $exporter->openStream();
307  }
308  # Log item dumps: all or by range
309  if ( $history & WikiExporter::LOGS ) {
310  if ( $this->startId || $this->endId ) {
311  $exporter->logsByRange( $this->startId, $this->endId );
312  } else {
313  $exporter->allLogs();
314  }
315  } elseif ( is_null( $this->pages ) ) {
316  # Page dumps: all or by page ID range
317  if ( $this->startId || $this->endId ) {
318  $exporter->pagesByRange( $this->startId, $this->endId, $this->orderRevs );
319  } elseif ( $this->revStartId || $this->revEndId ) {
320  $exporter->revsByRange( $this->revStartId, $this->revEndId );
321  } else {
322  $exporter->allPages();
323  }
324  } else {
325  # Dump of specific pages
326  $exporter->pagesByName( $this->pages );
327  }
328 
329  if ( !$this->skipFooter ) {
330  $exporter->closeStream();
331  }
332 
333  $this->report( true );
334  }
335 
342  function initProgress( $history = WikiExporter::FULL ) {
343  $table = ( $history == WikiExporter::CURRENT ) ? 'page' : 'revision';
344  $field = ( $history == WikiExporter::CURRENT ) ? 'page_id' : 'rev_id';
345 
347  if ( $this->forcedDb === null ) {
348  $dbr = $this->getDB( DB_REPLICA );
349  }
350  $this->maxCount = $dbr->selectField( $table, "MAX($field)", '', __METHOD__ );
351  $this->startTime = microtime( true );
352  $this->lastTime = $this->startTime;
353  $this->ID = getmypid();
354  }
355 
362  function backupDb() {
363  if ( $this->forcedDb !== null ) {
364  return $this->forcedDb;
365  }
366 
367  $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
368  $this->lb = $lbFactory->newMainLB();
369  $db = $this->lb->getMaintenanceConnectionRef( DB_REPLICA, 'dump' );
370 
371  // Discourage the server from disconnecting us if it takes a long time
372  // to read out the big ol' batch query.
373  $db->setSessionOptions( [ 'connTimeout' => 3600 * 24 ] );
374 
375  return $db;
376  }
377 
384  public function setDB( IMaintainableDatabase $db ) {
385  parent::setDB( $db );
386  $this->forcedDb = $db;
387  }
388 
389  function __destruct() {
390  if ( isset( $this->lb ) ) {
391  $this->lb->closeAll();
392  }
393  }
394 
395  function backupServer() {
396  global $wgDBserver;
397 
398  return $this->server ?: $wgDBserver;
399  }
400 
401  function reportPage() {
402  $this->pageCount++;
403  }
404 
405  function revCount() {
406  $this->revCount++;
407  $this->report();
408  }
409 
410  function report( $final = false ) {
411  if ( $final xor ( $this->revCount % $this->reportingInterval == 0 ) ) {
412  $this->showReport();
413  }
414  }
415 
416  function showReport() {
417  if ( $this->reporting ) {
418  $now = wfTimestamp( TS_DB );
419  $nowts = microtime( true );
420  $deltaAll = $nowts - $this->startTime;
421  $deltaPart = $nowts - $this->lastTime;
422  $this->pageCountPart = $this->pageCount - $this->pageCountLast;
423  $this->revCountPart = $this->revCount - $this->revCountLast;
424 
425  if ( $deltaAll ) {
426  $portion = $this->revCount / $this->maxCount;
427  $eta = $this->startTime + $deltaAll / $portion;
428  $etats = wfTimestamp( TS_DB, intval( $eta ) );
429  $pageRate = $this->pageCount / $deltaAll;
430  $revRate = $this->revCount / $deltaAll;
431  } else {
432  $pageRate = '-';
433  $revRate = '-';
434  $etats = '-';
435  }
436  if ( $deltaPart ) {
437  $pageRatePart = $this->pageCountPart / $deltaPart;
438  $revRatePart = $this->revCountPart / $deltaPart;
439  } else {
440  $pageRatePart = '-';
441  $revRatePart = '-';
442  }
443 
444  $dbDomain = WikiMap::getCurrentWikiDbDomain()->getId();
445  $this->progress( sprintf(
446  "%s: %s (ID %d) %d pages (%0.1f|%0.1f/sec all|curr), "
447  . "%d revs (%0.1f|%0.1f/sec all|curr), ETA %s [max %d]",
448  $now, $dbDomain, $this->ID, $this->pageCount, $pageRate,
449  $pageRatePart, $this->revCount, $revRate, $revRatePart, $etats,
450  $this->maxCount
451  ) );
452  $this->lastTime = $nowts;
453  $this->revCountLast = $this->revCount;
454  }
455  }
456 
457  function progress( $string ) {
458  if ( $this->reporting ) {
459  fwrite( $this->stderr, $string . "\n" );
460  }
461  }
462 }
if(PHP_SAPI !='cli-server') if(!isset( $_SERVER['SCRIPT_FILENAME'])) $file
Item class for a filearchive table row.
Definition: router.php:42
processOptions()
Processes arguments and sets $this->$sink accordingly.
int $timeOfCheckpoint
dump( $history, $text=WikiExporter::TEXT)
LoadBalancer $lb
setSessionOptions(array $options)
Override database&#39;s default behavior.
getOption( $name, $default=null)
Get an option, or return the default.
progress( $string)
array $orderedOptions
Used to read the options in the order they were passed.
registerFilter( $name, $class)
bool resource $stderr
Abstract maintenance class for quickly writing and churning out maintenance scripts with minimal effo...
Definition: Maintenance.php:86
__construct( $args=null)
$wgDBserver
Database host name or IP address.
string null $thisRevFormat
hasOption( $name)
Checks to see if a particular option exists.
loadWithArgv( $argv)
Load params and arguments from a given array of command-line arguments.
if( $line===false) $args
Definition: cdb.php:64
loadPlugin( $class, $file)
Load a plugin and register it.
initProgress( $history=WikiExporter::FULL)
Initialise starting time and maximum revision count.
wfTimestamp( $outputtype=TS_UNIX, $ts=0)
Get a timestamp string in one of various formats.
IMaintainableDatabase null $forcedDb
The dependency-injected database to use.
ExportProgressFilter $egress
string $lastName
string $buffer
$filter
static getCurrentWikiDbDomain()
Definition: WikiMap.php:293
setDB(IMaintainableDatabase $db)
Force the dump to use the provided database connection for database operations, wherever possible...
Advanced database interface for IDatabase handles that include maintenance methods.
string null $thisRevModel
report( $final=false)
static schemaVersion()
Returns the default export schema version, as defined by $wgXmlDumpSchemaVersion. ...
addOption( $name, $description, $required=false, $withArg=false, $shortName=false, $multiOccurrence=false)
Add a parameter to the script.
static string [] $supportedSchemas
the schema versions supported for output
const DB_REPLICA
Definition: defines.php:25
registerOutput( $name, $class)
fatalError( $msg, $exitCode=1)
Output a message and terminate the current script.
getDB( $db, $groups=[], $dbDomain=false)
Returns a database to be used by current maintenance script.
array false $openElement