Merge "mw.Title: Use $wgIllegalFileChars for file title checking"
[lhc/web/wiklou.git] / maintenance / Maintenance.php
1 <?php
2 /**
3 * This program is free software; you can redistribute it and/or modify
4 * it under the terms of the GNU General Public License as published by
5 * the Free Software Foundation; either version 2 of the License, or
6 * (at your option) any later version.
7 *
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11 * GNU General Public License for more details.
12 *
13 * You should have received a copy of the GNU General Public License along
14 * with this program; if not, write to the Free Software Foundation, Inc.,
15 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
16 * http://www.gnu.org/copyleft/gpl.html
17 *
18 * @file
19 * @ingroup Maintenance
20 * @defgroup Maintenance Maintenance
21 */
22
23 // Bail on old versions of PHP, or if composer has not been run yet to install
24 // dependencies.
25 require_once __DIR__ . '/../includes/PHPVersionCheck.php';
26 wfEntryPointCheck( 'cli' );
27
28 /**
29 * @defgroup MaintenanceArchive Maintenance archives
30 * @ingroup Maintenance
31 */
32
33 // Define this so scripts can easily find doMaintenance.php
34 define( 'RUN_MAINTENANCE_IF_MAIN', __DIR__ . '/doMaintenance.php' );
35 define( 'DO_MAINTENANCE', RUN_MAINTENANCE_IF_MAIN ); // original name, harmless
36
37 $maintClass = false;
38
39 use MediaWiki\Logger\LoggerFactory;
40
41 /**
42 * Abstract maintenance class for quickly writing and churning out
43 * maintenance scripts with minimal effort. All that _must_ be defined
44 * is the execute() method. See docs/maintenance.txt for more info
45 * and a quick demo of how to use it.
46 *
47 * @author Chad Horohoe <chad@anyonecanedit.org>
48 * @since 1.16
49 * @ingroup Maintenance
50 */
51 abstract class Maintenance {
52 /**
53 * Constants for DB access type
54 * @see Maintenance::getDbType()
55 */
56 const DB_NONE = 0;
57 const DB_STD = 1;
58 const DB_ADMIN = 2;
59
60 // Const for getStdin()
61 const STDIN_ALL = 'all';
62
63 // This is the desired params
64 protected $mParams = [];
65
66 // Array of mapping short parameters to long ones
67 protected $mShortParamsMap = [];
68
69 // Array of desired args
70 protected $mArgList = [];
71
72 // This is the list of options that were actually passed
73 protected $mOptions = [];
74
75 // This is the list of arguments that were actually passed
76 protected $mArgs = [];
77
78 // Name of the script currently running
79 protected $mSelf;
80
81 // Special vars for params that are always used
82 protected $mQuiet = false;
83 protected $mDbUser, $mDbPass;
84
85 // A description of the script, children should change this via addDescription()
86 protected $mDescription = '';
87
88 // Have we already loaded our user input?
89 protected $mInputLoaded = false;
90
91 /**
92 * Batch size. If a script supports this, they should set
93 * a default with setBatchSize()
94 *
95 * @var int
96 */
97 protected $mBatchSize = null;
98
99 // Generic options added by addDefaultParams()
100 private $mGenericParameters = [];
101 // Generic options which might or not be supported by the script
102 private $mDependantParameters = [];
103
104 /**
105 * Used by getDB() / setDB()
106 * @var IDatabase
107 */
108 private $mDb = null;
109
110 /** @var float UNIX timestamp */
111 private $lastSlaveWait = 0.0;
112
113 /**
114 * Used when creating separate schema files.
115 * @var resource
116 */
117 public $fileHandle;
118
119 /**
120 * Accessible via getConfig()
121 *
122 * @var Config
123 */
124 private $config;
125
126 /**
127 * @see Maintenance::requireExtension
128 * @var array
129 */
130 private $requiredExtensions = [];
131
132 /**
133 * Used to read the options in the order they were passed.
134 * Useful for option chaining (Ex. dumpBackup.php). It will
135 * be an empty array if the options are passed in through
136 * loadParamsAndArgs( $self, $opts, $args ).
137 *
138 * This is an array of arrays where
139 * 0 => the option and 1 => parameter value.
140 *
141 * @var array
142 */
143 public $orderedOptions = [];
144
145 /**
146 * Default constructor. Children should call this *first* if implementing
147 * their own constructors
148 */
149 public function __construct() {
150 // Setup $IP, using MW_INSTALL_PATH if it exists
151 global $IP;
152 $IP = strval( getenv( 'MW_INSTALL_PATH' ) ) !== ''
153 ? getenv( 'MW_INSTALL_PATH' )
154 : realpath( __DIR__ . '/..' );
155
156 $this->addDefaultParams();
157 register_shutdown_function( [ $this, 'outputChanneled' ], false );
158 }
159
160 /**
161 * Should we execute the maintenance script, or just allow it to be included
162 * as a standalone class? It checks that the call stack only includes this
163 * function and "requires" (meaning was called from the file scope)
164 *
165 * @return bool
166 */
167 public static function shouldExecute() {
168 global $wgCommandLineMode;
169
170 if ( !function_exists( 'debug_backtrace' ) ) {
171 // If someone has a better idea...
172 return $wgCommandLineMode;
173 }
174
175 $bt = debug_backtrace();
176 $count = count( $bt );
177 if ( $count < 2 ) {
178 return false; // sanity
179 }
180 if ( $bt[0]['class'] !== 'Maintenance' || $bt[0]['function'] !== 'shouldExecute' ) {
181 return false; // last call should be to this function
182 }
183 $includeFuncs = [ 'require_once', 'require', 'include', 'include_once' ];
184 for ( $i = 1; $i < $count; $i++ ) {
185 if ( !in_array( $bt[$i]['function'], $includeFuncs ) ) {
186 return false; // previous calls should all be "requires"
187 }
188 }
189
190 return true;
191 }
192
193 /**
194 * Do the actual work. All child classes will need to implement this
195 */
196 abstract public function execute();
197
198 /**
199 * Add a parameter to the script. Will be displayed on --help
200 * with the associated description
201 *
202 * @param string $name The name of the param (help, version, etc)
203 * @param string $description The description of the param to show on --help
204 * @param bool $required Is the param required?
205 * @param bool $withArg Is an argument required with this option?
206 * @param string $shortName Character to use as short name
207 * @param bool $multiOccurrence Can this option be passed multiple times?
208 */
209 protected function addOption( $name, $description, $required = false,
210 $withArg = false, $shortName = false, $multiOccurrence = false
211 ) {
212 $this->mParams[$name] = [
213 'desc' => $description,
214 'require' => $required,
215 'withArg' => $withArg,
216 'shortName' => $shortName,
217 'multiOccurrence' => $multiOccurrence
218 ];
219
220 if ( $shortName !== false ) {
221 $this->mShortParamsMap[$shortName] = $name;
222 }
223 }
224
225 /**
226 * Checks to see if a particular param exists.
227 * @param string $name The name of the param
228 * @return bool
229 */
230 protected function hasOption( $name ) {
231 return isset( $this->mOptions[$name] );
232 }
233
234 /**
235 * Get an option, or return the default.
236 *
237 * If the option was added to support multiple occurrences,
238 * this will return an array.
239 *
240 * @param string $name The name of the param
241 * @param mixed $default Anything you want, default null
242 * @return mixed
243 */
244 protected function getOption( $name, $default = null ) {
245 if ( $this->hasOption( $name ) ) {
246 return $this->mOptions[$name];
247 } else {
248 // Set it so we don't have to provide the default again
249 $this->mOptions[$name] = $default;
250
251 return $this->mOptions[$name];
252 }
253 }
254
255 /**
256 * Add some args that are needed
257 * @param string $arg Name of the arg, like 'start'
258 * @param string $description Short description of the arg
259 * @param bool $required Is this required?
260 */
261 protected function addArg( $arg, $description, $required = true ) {
262 $this->mArgList[] = [
263 'name' => $arg,
264 'desc' => $description,
265 'require' => $required
266 ];
267 }
268
269 /**
270 * Remove an option. Useful for removing options that won't be used in your script.
271 * @param string $name The option to remove.
272 */
273 protected function deleteOption( $name ) {
274 unset( $this->mParams[$name] );
275 }
276
277 /**
278 * Set the description text.
279 * @param string $text The text of the description
280 */
281 protected function addDescription( $text ) {
282 $this->mDescription = $text;
283 }
284
285 /**
286 * Does a given argument exist?
287 * @param int $argId The integer value (from zero) for the arg
288 * @return bool
289 */
290 protected function hasArg( $argId = 0 ) {
291 return isset( $this->mArgs[$argId] );
292 }
293
294 /**
295 * Get an argument.
296 * @param int $argId The integer value (from zero) for the arg
297 * @param mixed $default The default if it doesn't exist
298 * @return mixed
299 */
300 protected function getArg( $argId = 0, $default = null ) {
301 return $this->hasArg( $argId ) ? $this->mArgs[$argId] : $default;
302 }
303
304 /**
305 * Set the batch size.
306 * @param int $s The number of operations to do in a batch
307 */
308 protected function setBatchSize( $s = 0 ) {
309 $this->mBatchSize = $s;
310
311 // If we support $mBatchSize, show the option.
312 // Used to be in addDefaultParams, but in order for that to
313 // work, subclasses would have to call this function in the constructor
314 // before they called parent::__construct which is just weird
315 // (and really wasn't done).
316 if ( $this->mBatchSize ) {
317 $this->addOption( 'batch-size', 'Run this many operations ' .
318 'per batch, default: ' . $this->mBatchSize, false, true );
319 if ( isset( $this->mParams['batch-size'] ) ) {
320 // This seems a little ugly...
321 $this->mDependantParameters['batch-size'] = $this->mParams['batch-size'];
322 }
323 }
324 }
325
326 /**
327 * Get the script's name
328 * @return string
329 */
330 public function getName() {
331 return $this->mSelf;
332 }
333
334 /**
335 * Return input from stdin.
336 * @param int $len The number of bytes to read. If null, just return the handle.
337 * Maintenance::STDIN_ALL returns the full length
338 * @return mixed
339 */
340 protected function getStdin( $len = null ) {
341 if ( $len == Maintenance::STDIN_ALL ) {
342 return file_get_contents( 'php://stdin' );
343 }
344 $f = fopen( 'php://stdin', 'rt' );
345 if ( !$len ) {
346 return $f;
347 }
348 $input = fgets( $f, $len );
349 fclose( $f );
350
351 return rtrim( $input );
352 }
353
354 /**
355 * @return bool
356 */
357 public function isQuiet() {
358 return $this->mQuiet;
359 }
360
361 /**
362 * Throw some output to the user. Scripts can call this with no fears,
363 * as we handle all --quiet stuff here
364 * @param string $out The text to show to the user
365 * @param mixed $channel Unique identifier for the channel. See function outputChanneled.
366 */
367 protected function output( $out, $channel = null ) {
368 if ( $this->mQuiet ) {
369 return;
370 }
371 if ( $channel === null ) {
372 $this->cleanupChanneled();
373 print $out;
374 } else {
375 $out = preg_replace( '/\n\z/', '', $out );
376 $this->outputChanneled( $out, $channel );
377 }
378 }
379
380 /**
381 * Throw an error to the user. Doesn't respect --quiet, so don't use
382 * this for non-error output
383 * @param string $err The error to display
384 * @param int $die If > 0, go ahead and die out using this int as the code
385 */
386 protected function error( $err, $die = 0 ) {
387 $this->outputChanneled( false );
388 if ( PHP_SAPI == 'cli' ) {
389 fwrite( STDERR, $err . "\n" );
390 } else {
391 print $err;
392 }
393 $die = intval( $die );
394 if ( $die > 0 ) {
395 die( $die );
396 }
397 }
398
399 private $atLineStart = true;
400 private $lastChannel = null;
401
402 /**
403 * Clean up channeled output. Output a newline if necessary.
404 */
405 public function cleanupChanneled() {
406 if ( !$this->atLineStart ) {
407 print "\n";
408 $this->atLineStart = true;
409 }
410 }
411
412 /**
413 * Message outputter with channeled message support. Messages on the
414 * same channel are concatenated, but any intervening messages in another
415 * channel start a new line.
416 * @param string $msg The message without trailing newline
417 * @param string $channel Channel identifier or null for no
418 * channel. Channel comparison uses ===.
419 */
420 public function outputChanneled( $msg, $channel = null ) {
421 if ( $msg === false ) {
422 $this->cleanupChanneled();
423
424 return;
425 }
426
427 // End the current line if necessary
428 if ( !$this->atLineStart && $channel !== $this->lastChannel ) {
429 print "\n";
430 }
431
432 print $msg;
433
434 $this->atLineStart = false;
435 if ( $channel === null ) {
436 // For unchanneled messages, output trailing newline immediately
437 print "\n";
438 $this->atLineStart = true;
439 }
440 $this->lastChannel = $channel;
441 }
442
443 /**
444 * Does the script need different DB access? By default, we give Maintenance
445 * scripts normal rights to the DB. Sometimes, a script needs admin rights
446 * access for a reason and sometimes they want no access. Subclasses should
447 * override and return one of the following values, as needed:
448 * Maintenance::DB_NONE - For no DB access at all
449 * Maintenance::DB_STD - For normal DB access, default
450 * Maintenance::DB_ADMIN - For admin DB access
451 * @return int
452 */
453 public function getDbType() {
454 return Maintenance::DB_STD;
455 }
456
457 /**
458 * Add the default parameters to the scripts
459 */
460 protected function addDefaultParams() {
461
462 # Generic (non script dependant) options:
463
464 $this->addOption( 'help', 'Display this help message', false, false, 'h' );
465 $this->addOption( 'quiet', 'Whether to supress non-error output', false, false, 'q' );
466 $this->addOption( 'conf', 'Location of LocalSettings.php, if not default', false, true );
467 $this->addOption( 'wiki', 'For specifying the wiki ID', false, true );
468 $this->addOption( 'globals', 'Output globals at the end of processing for debugging' );
469 $this->addOption(
470 'memory-limit',
471 'Set a specific memory limit for the script, '
472 . '"max" for no limit or "default" to avoid changing it'
473 );
474 $this->addOption( 'server', "The protocol and server name to use in URLs, e.g. " .
475 "http://en.wikipedia.org. This is sometimes necessary because " .
476 "server name detection may fail in command line scripts.", false, true );
477 $this->addOption( 'profiler', 'Profiler output format (usually "text")', false, true );
478
479 # Save generic options to display them separately in help
480 $this->mGenericParameters = $this->mParams;
481
482 # Script dependant options:
483
484 // If we support a DB, show the options
485 if ( $this->getDbType() > 0 ) {
486 $this->addOption( 'dbuser', 'The DB user to use for this script', false, true );
487 $this->addOption( 'dbpass', 'The password to use for this script', false, true );
488 }
489
490 # Save additional script dependant options to display
491 #  them separately in help
492 $this->mDependantParameters = array_diff_key( $this->mParams, $this->mGenericParameters );
493 }
494
495 /**
496 * @since 1.24
497 * @return Config
498 */
499 public function getConfig() {
500 if ( $this->config === null ) {
501 $this->config = ConfigFactory::getDefaultInstance()->makeConfig( 'main' );
502 }
503
504 return $this->config;
505 }
506
507 /**
508 * @since 1.24
509 * @param Config $config
510 */
511 public function setConfig( Config $config ) {
512 $this->config = $config;
513 }
514
515 /**
516 * Indicate that the specified extension must be
517 * loaded before the script can run.
518 *
519 * This *must* be called in the constructor.
520 *
521 * @since 1.28
522 * @param string $name
523 */
524 protected function requireExtension( $name ) {
525 $this->requiredExtensions[] = $name;
526 }
527
528 /**
529 * Verify that the required extensions are installed
530 *
531 * @since 1.28
532 */
533 public function checkRequiredExtensions() {
534 $registry = ExtensionRegistry::getInstance();
535 $missing = [];
536 foreach ( $this->requiredExtensions as $name ) {
537 if ( !$registry->isLoaded( $name ) ) {
538 $missing[] = $name;
539 }
540 }
541
542 if ( $missing ) {
543 $joined = implode( ', ', $missing );
544 $msg = "The following extensions are required to be installed "
545 . "for this script to run: $joined. Please enable them and then try again.";
546 $this->error( $msg, 1 );
547 }
548
549 }
550
551 /**
552 * Run a child maintenance script. Pass all of the current arguments
553 * to it.
554 * @param string $maintClass A name of a child maintenance class
555 * @param string $classFile Full path of where the child is
556 * @return Maintenance
557 */
558 public function runChild( $maintClass, $classFile = null ) {
559 // Make sure the class is loaded first
560 if ( !class_exists( $maintClass ) ) {
561 if ( $classFile ) {
562 require_once $classFile;
563 }
564 if ( !class_exists( $maintClass ) ) {
565 $this->error( "Cannot spawn child: $maintClass" );
566 }
567 }
568
569 /**
570 * @var $child Maintenance
571 */
572 $child = new $maintClass();
573 $child->loadParamsAndArgs( $this->mSelf, $this->mOptions, $this->mArgs );
574 if ( !is_null( $this->mDb ) ) {
575 $child->setDB( $this->mDb );
576 }
577
578 return $child;
579 }
580
581 /**
582 * Do some sanity checking and basic setup
583 */
584 public function setup() {
585 global $IP, $wgCommandLineMode, $wgRequestTime;
586
587 # Abort if called from a web server
588 if ( isset( $_SERVER ) && isset( $_SERVER['REQUEST_METHOD'] ) ) {
589 $this->error( 'This script must be run from the command line', true );
590 }
591
592 if ( $IP === null ) {
593 $this->error( "\$IP not set, aborting!\n" .
594 '(Did you forget to call parent::__construct() in your maintenance script?)', 1 );
595 }
596
597 # Make sure we can handle script parameters
598 if ( !defined( 'HPHP_VERSION' ) && !ini_get( 'register_argc_argv' ) ) {
599 $this->error( 'Cannot get command line arguments, register_argc_argv is set to false', true );
600 }
601
602 // Send PHP warnings and errors to stderr instead of stdout.
603 // This aids in diagnosing problems, while keeping messages
604 // out of redirected output.
605 if ( ini_get( 'display_errors' ) ) {
606 ini_set( 'display_errors', 'stderr' );
607 }
608
609 $this->loadParamsAndArgs();
610 $this->maybeHelp();
611
612 # Set the memory limit
613 # Note we need to set it again later in cache LocalSettings changed it
614 $this->adjustMemoryLimit();
615
616 # Set max execution time to 0 (no limit). PHP.net says that
617 # "When running PHP from the command line the default setting is 0."
618 # But sometimes this doesn't seem to be the case.
619 ini_set( 'max_execution_time', 0 );
620
621 $wgRequestTime = microtime( true );
622
623 # Define us as being in MediaWiki
624 define( 'MEDIAWIKI', true );
625
626 $wgCommandLineMode = true;
627
628 # Turn off output buffering if it's on
629 while ( ob_get_level() > 0 ) {
630 ob_end_flush();
631 }
632
633 $this->validateParamsAndArgs();
634 }
635
636 /**
637 * Normally we disable the memory_limit when running admin scripts.
638 * Some scripts may wish to actually set a limit, however, to avoid
639 * blowing up unexpectedly. We also support a --memory-limit option,
640 * to allow sysadmins to explicitly set one if they'd prefer to override
641 * defaults (or for people using Suhosin which yells at you for trying
642 * to disable the limits)
643 * @return string
644 */
645 public function memoryLimit() {
646 $limit = $this->getOption( 'memory-limit', 'max' );
647 $limit = trim( $limit, "\" '" ); // trim quotes in case someone misunderstood
648 return $limit;
649 }
650
651 /**
652 * Adjusts PHP's memory limit to better suit our needs, if needed.
653 */
654 protected function adjustMemoryLimit() {
655 $limit = $this->memoryLimit();
656 if ( $limit == 'max' ) {
657 $limit = -1; // no memory limit
658 }
659 if ( $limit != 'default' ) {
660 ini_set( 'memory_limit', $limit );
661 }
662 }
663
664 /**
665 * Activate the profiler (assuming $wgProfiler is set)
666 */
667 protected function activateProfiler() {
668 global $wgProfiler, $wgProfileLimit, $wgTrxProfilerLimits;
669
670 $output = $this->getOption( 'profiler' );
671 if ( !$output ) {
672 return;
673 }
674
675 if ( is_array( $wgProfiler ) && isset( $wgProfiler['class'] ) ) {
676 $class = $wgProfiler['class'];
677 $profiler = new $class(
678 [ 'sampling' => 1, 'output' => [ $output ] ]
679 + $wgProfiler
680 + [ 'threshold' => $wgProfileLimit ]
681 );
682 $profiler->setTemplated( true );
683 Profiler::replaceStubInstance( $profiler );
684 }
685
686 $trxProfiler = Profiler::instance()->getTransactionProfiler();
687 $trxProfiler->setLogger( LoggerFactory::getInstance( 'DBPerformance' ) );
688 $trxProfiler->setExpectations( $wgTrxProfilerLimits['Maintenance'], __METHOD__ );
689 }
690
691 /**
692 * Clear all params and arguments.
693 */
694 public function clearParamsAndArgs() {
695 $this->mOptions = [];
696 $this->mArgs = [];
697 $this->mInputLoaded = false;
698 }
699
700 /**
701 * Load params and arguments from a given array
702 * of command-line arguments
703 *
704 * @since 1.27
705 * @param array $argv
706 */
707 public function loadWithArgv( $argv ) {
708 $options = [];
709 $args = [];
710 $this->orderedOptions = [];
711
712 # Parse arguments
713 for ( $arg = reset( $argv ); $arg !== false; $arg = next( $argv ) ) {
714 if ( $arg == '--' ) {
715 # End of options, remainder should be considered arguments
716 $arg = next( $argv );
717 while ( $arg !== false ) {
718 $args[] = $arg;
719 $arg = next( $argv );
720 }
721 break;
722 } elseif ( substr( $arg, 0, 2 ) == '--' ) {
723 # Long options
724 $option = substr( $arg, 2 );
725 if ( isset( $this->mParams[$option] ) && $this->mParams[$option]['withArg'] ) {
726 $param = next( $argv );
727 if ( $param === false ) {
728 $this->error( "\nERROR: $option parameter needs a value after it\n" );
729 $this->maybeHelp( true );
730 }
731
732 $this->setParam( $options, $option, $param );
733 } else {
734 $bits = explode( '=', $option, 2 );
735 if ( count( $bits ) > 1 ) {
736 $option = $bits[0];
737 $param = $bits[1];
738 } else {
739 $param = 1;
740 }
741
742 $this->setParam( $options, $option, $param );
743 }
744 } elseif ( $arg == '-' ) {
745 # Lonely "-", often used to indicate stdin or stdout.
746 $args[] = $arg;
747 } elseif ( substr( $arg, 0, 1 ) == '-' ) {
748 # Short options
749 $argLength = strlen( $arg );
750 for ( $p = 1; $p < $argLength; $p++ ) {
751 $option = $arg[$p];
752 if ( !isset( $this->mParams[$option] ) && isset( $this->mShortParamsMap[$option] ) ) {
753 $option = $this->mShortParamsMap[$option];
754 }
755
756 if ( isset( $this->mParams[$option]['withArg'] ) && $this->mParams[$option]['withArg'] ) {
757 $param = next( $argv );
758 if ( $param === false ) {
759 $this->error( "\nERROR: $option parameter needs a value after it\n" );
760 $this->maybeHelp( true );
761 }
762 $this->setParam( $options, $option, $param );
763 } else {
764 $this->setParam( $options, $option, 1 );
765 }
766 }
767 } else {
768 $args[] = $arg;
769 }
770 }
771
772 $this->mOptions = $options;
773 $this->mArgs = $args;
774 $this->loadSpecialVars();
775 $this->mInputLoaded = true;
776 }
777
778 /**
779 * Helper function used solely by loadParamsAndArgs
780 * to prevent code duplication
781 *
782 * This sets the param in the options array based on
783 * whether or not it can be specified multiple times.
784 *
785 * @since 1.27
786 * @param array $options
787 * @param string $option
788 * @param mixed $value
789 */
790 private function setParam( &$options, $option, $value ) {
791 $this->orderedOptions[] = [ $option, $value ];
792
793 if ( isset( $this->mParams[$option] ) ) {
794 $multi = $this->mParams[$option]['multiOccurrence'];
795 } else {
796 $multi = false;
797 }
798 $exists = array_key_exists( $option, $options );
799 if ( $multi && $exists ) {
800 $options[$option][] = $value;
801 } elseif ( $multi ) {
802 $options[$option] = [ $value ];
803 } elseif ( !$exists ) {
804 $options[$option] = $value;
805 } else {
806 $this->error( "\nERROR: $option parameter given twice\n" );
807 $this->maybeHelp( true );
808 }
809 }
810
811 /**
812 * Process command line arguments
813 * $mOptions becomes an array with keys set to the option names
814 * $mArgs becomes a zero-based array containing the non-option arguments
815 *
816 * @param string $self The name of the script, if any
817 * @param array $opts An array of options, in form of key=>value
818 * @param array $args An array of command line arguments
819 */
820 public function loadParamsAndArgs( $self = null, $opts = null, $args = null ) {
821 # If we were given opts or args, set those and return early
822 if ( $self ) {
823 $this->mSelf = $self;
824 $this->mInputLoaded = true;
825 }
826 if ( $opts ) {
827 $this->mOptions = $opts;
828 $this->mInputLoaded = true;
829 }
830 if ( $args ) {
831 $this->mArgs = $args;
832 $this->mInputLoaded = true;
833 }
834
835 # If we've already loaded input (either by user values or from $argv)
836 # skip on loading it again. The array_shift() will corrupt values if
837 # it's run again and again
838 if ( $this->mInputLoaded ) {
839 $this->loadSpecialVars();
840
841 return;
842 }
843
844 global $argv;
845 $this->mSelf = $argv[0];
846 $this->loadWithArgv( array_slice( $argv, 1 ) );
847 }
848
849 /**
850 * Run some validation checks on the params, etc
851 */
852 protected function validateParamsAndArgs() {
853 $die = false;
854 # Check to make sure we've got all the required options
855 foreach ( $this->mParams as $opt => $info ) {
856 if ( $info['require'] && !$this->hasOption( $opt ) ) {
857 $this->error( "Param $opt required!" );
858 $die = true;
859 }
860 }
861 # Check arg list too
862 foreach ( $this->mArgList as $k => $info ) {
863 if ( $info['require'] && !$this->hasArg( $k ) ) {
864 $this->error( 'Argument <' . $info['name'] . '> required!' );
865 $die = true;
866 }
867 }
868
869 if ( $die ) {
870 $this->maybeHelp( true );
871 }
872 }
873
874 /**
875 * Handle the special variables that are global to all scripts
876 */
877 protected function loadSpecialVars() {
878 if ( $this->hasOption( 'dbuser' ) ) {
879 $this->mDbUser = $this->getOption( 'dbuser' );
880 }
881 if ( $this->hasOption( 'dbpass' ) ) {
882 $this->mDbPass = $this->getOption( 'dbpass' );
883 }
884 if ( $this->hasOption( 'quiet' ) ) {
885 $this->mQuiet = true;
886 }
887 if ( $this->hasOption( 'batch-size' ) ) {
888 $this->mBatchSize = intval( $this->getOption( 'batch-size' ) );
889 }
890 }
891
892 /**
893 * Maybe show the help.
894 * @param bool $force Whether to force the help to show, default false
895 */
896 protected function maybeHelp( $force = false ) {
897 if ( !$force && !$this->hasOption( 'help' ) ) {
898 return;
899 }
900
901 $screenWidth = 80; // TODO: Calculate this!
902 $tab = " ";
903 $descWidth = $screenWidth - ( 2 * strlen( $tab ) );
904
905 ksort( $this->mParams );
906 $this->mQuiet = false;
907
908 // Description ...
909 if ( $this->mDescription ) {
910 $this->output( "\n" . $this->mDescription . "\n" );
911 }
912 $output = "\nUsage: php " . basename( $this->mSelf );
913
914 // ... append parameters ...
915 if ( $this->mParams ) {
916 $output .= " [--" . implode( array_keys( $this->mParams ), "|--" ) . "]";
917 }
918
919 // ... and append arguments.
920 if ( $this->mArgList ) {
921 $output .= ' ';
922 foreach ( $this->mArgList as $k => $arg ) {
923 if ( $arg['require'] ) {
924 $output .= '<' . $arg['name'] . '>';
925 } else {
926 $output .= '[' . $arg['name'] . ']';
927 }
928 if ( $k < count( $this->mArgList ) - 1 ) {
929 $output .= ' ';
930 }
931 }
932 }
933 $this->output( "$output\n\n" );
934
935 # TODO abstract some repetitive code below
936
937 // Generic parameters
938 $this->output( "Generic maintenance parameters:\n" );
939 foreach ( $this->mGenericParameters as $par => $info ) {
940 if ( $info['shortName'] !== false ) {
941 $par .= " (-{$info['shortName']})";
942 }
943 $this->output(
944 wordwrap( "$tab--$par: " . $info['desc'], $descWidth,
945 "\n$tab$tab" ) . "\n"
946 );
947 }
948 $this->output( "\n" );
949
950 $scriptDependantParams = $this->mDependantParameters;
951 if ( count( $scriptDependantParams ) > 0 ) {
952 $this->output( "Script dependant parameters:\n" );
953 // Parameters description
954 foreach ( $scriptDependantParams as $par => $info ) {
955 if ( $info['shortName'] !== false ) {
956 $par .= " (-{$info['shortName']})";
957 }
958 $this->output(
959 wordwrap( "$tab--$par: " . $info['desc'], $descWidth,
960 "\n$tab$tab" ) . "\n"
961 );
962 }
963 $this->output( "\n" );
964 }
965
966 // Script specific parameters not defined on construction by
967 // Maintenance::addDefaultParams()
968 $scriptSpecificParams = array_diff_key(
969 # all script parameters:
970 $this->mParams,
971 # remove the Maintenance default parameters:
972 $this->mGenericParameters,
973 $this->mDependantParameters
974 );
975 if ( count( $scriptSpecificParams ) > 0 ) {
976 $this->output( "Script specific parameters:\n" );
977 // Parameters description
978 foreach ( $scriptSpecificParams as $par => $info ) {
979 if ( $info['shortName'] !== false ) {
980 $par .= " (-{$info['shortName']})";
981 }
982 $this->output(
983 wordwrap( "$tab--$par: " . $info['desc'], $descWidth,
984 "\n$tab$tab" ) . "\n"
985 );
986 }
987 $this->output( "\n" );
988 }
989
990 // Print arguments
991 if ( count( $this->mArgList ) > 0 ) {
992 $this->output( "Arguments:\n" );
993 // Arguments description
994 foreach ( $this->mArgList as $info ) {
995 $openChar = $info['require'] ? '<' : '[';
996 $closeChar = $info['require'] ? '>' : ']';
997 $this->output(
998 wordwrap( "$tab$openChar" . $info['name'] . "$closeChar: " .
999 $info['desc'], $descWidth, "\n$tab$tab" ) . "\n"
1000 );
1001 }
1002 $this->output( "\n" );
1003 }
1004
1005 die( 1 );
1006 }
1007
1008 /**
1009 * Handle some last-minute setup here.
1010 */
1011 public function finalSetup() {
1012 global $wgCommandLineMode, $wgShowSQLErrors, $wgServer;
1013 global $wgDBadminuser, $wgDBadminpassword;
1014 global $wgDBuser, $wgDBpassword, $wgDBservers, $wgLBFactoryConf;
1015
1016 # Turn off output buffering again, it might have been turned on in the settings files
1017 if ( ob_get_level() ) {
1018 ob_end_flush();
1019 }
1020 # Same with these
1021 $wgCommandLineMode = true;
1022
1023 # Override $wgServer
1024 if ( $this->hasOption( 'server' ) ) {
1025 $wgServer = $this->getOption( 'server', $wgServer );
1026 }
1027
1028 # If these were passed, use them
1029 if ( $this->mDbUser ) {
1030 $wgDBadminuser = $this->mDbUser;
1031 }
1032 if ( $this->mDbPass ) {
1033 $wgDBadminpassword = $this->mDbPass;
1034 }
1035
1036 if ( $this->getDbType() == self::DB_ADMIN && isset( $wgDBadminuser ) ) {
1037 $wgDBuser = $wgDBadminuser;
1038 $wgDBpassword = $wgDBadminpassword;
1039
1040 if ( $wgDBservers ) {
1041 /**
1042 * @var $wgDBservers array
1043 */
1044 foreach ( $wgDBservers as $i => $server ) {
1045 $wgDBservers[$i]['user'] = $wgDBuser;
1046 $wgDBservers[$i]['password'] = $wgDBpassword;
1047 }
1048 }
1049 if ( isset( $wgLBFactoryConf['serverTemplate'] ) ) {
1050 $wgLBFactoryConf['serverTemplate']['user'] = $wgDBuser;
1051 $wgLBFactoryConf['serverTemplate']['password'] = $wgDBpassword;
1052 }
1053 LBFactory::destroyInstance();
1054 }
1055
1056 // Per-script profiling; useful for debugging
1057 $this->activateProfiler();
1058
1059 $this->afterFinalSetup();
1060
1061 $wgShowSQLErrors = true;
1062
1063 MediaWiki\suppressWarnings();
1064 set_time_limit( 0 );
1065 MediaWiki\restoreWarnings();
1066
1067 $this->adjustMemoryLimit();
1068 }
1069
1070 /**
1071 * Execute a callback function at the end of initialisation
1072 */
1073 protected function afterFinalSetup() {
1074 if ( defined( 'MW_CMDLINE_CALLBACK' ) ) {
1075 call_user_func( MW_CMDLINE_CALLBACK );
1076 }
1077 }
1078
1079 /**
1080 * Potentially debug globals. Originally a feature only
1081 * for refreshLinks
1082 */
1083 public function globals() {
1084 if ( $this->hasOption( 'globals' ) ) {
1085 print_r( $GLOBALS );
1086 }
1087 }
1088
1089 /**
1090 * Generic setup for most installs. Returns the location of LocalSettings
1091 * @return string
1092 */
1093 public function loadSettings() {
1094 global $wgCommandLineMode, $IP;
1095
1096 if ( isset( $this->mOptions['conf'] ) ) {
1097 $settingsFile = $this->mOptions['conf'];
1098 } elseif ( defined( "MW_CONFIG_FILE" ) ) {
1099 $settingsFile = MW_CONFIG_FILE;
1100 } else {
1101 $settingsFile = "$IP/LocalSettings.php";
1102 }
1103 if ( isset( $this->mOptions['wiki'] ) ) {
1104 $bits = explode( '-', $this->mOptions['wiki'] );
1105 if ( count( $bits ) == 1 ) {
1106 $bits[] = '';
1107 }
1108 define( 'MW_DB', $bits[0] );
1109 define( 'MW_PREFIX', $bits[1] );
1110 }
1111
1112 if ( !is_readable( $settingsFile ) ) {
1113 $this->error( "A copy of your installation's LocalSettings.php\n" .
1114 "must exist and be readable in the source directory.\n" .
1115 "Use --conf to specify it.", true );
1116 }
1117 $wgCommandLineMode = true;
1118
1119 return $settingsFile;
1120 }
1121
1122 /**
1123 * Support function for cleaning up redundant text records
1124 * @param bool $delete Whether or not to actually delete the records
1125 * @author Rob Church <robchur@gmail.com>
1126 */
1127 public function purgeRedundantText( $delete = true ) {
1128 # Data should come off the master, wrapped in a transaction
1129 $dbw = $this->getDB( DB_MASTER );
1130 $this->beginTransaction( $dbw, __METHOD__ );
1131
1132 # Get "active" text records from the revisions table
1133 $this->output( 'Searching for active text records in revisions table...' );
1134 $res = $dbw->select( 'revision', 'rev_text_id', [], __METHOD__, [ 'DISTINCT' ] );
1135 foreach ( $res as $row ) {
1136 $cur[] = $row->rev_text_id;
1137 }
1138 $this->output( "done.\n" );
1139
1140 # Get "active" text records from the archive table
1141 $this->output( 'Searching for active text records in archive table...' );
1142 $res = $dbw->select( 'archive', 'ar_text_id', [], __METHOD__, [ 'DISTINCT' ] );
1143 foreach ( $res as $row ) {
1144 # old pre-MW 1.5 records can have null ar_text_id's.
1145 if ( $row->ar_text_id !== null ) {
1146 $cur[] = $row->ar_text_id;
1147 }
1148 }
1149 $this->output( "done.\n" );
1150
1151 # Get the IDs of all text records not in these sets
1152 $this->output( 'Searching for inactive text records...' );
1153 $cond = 'old_id NOT IN ( ' . $dbw->makeList( $cur ) . ' )';
1154 $res = $dbw->select( 'text', 'old_id', [ $cond ], __METHOD__, [ 'DISTINCT' ] );
1155 $old = [];
1156 foreach ( $res as $row ) {
1157 $old[] = $row->old_id;
1158 }
1159 $this->output( "done.\n" );
1160
1161 # Inform the user of what we're going to do
1162 $count = count( $old );
1163 $this->output( "$count inactive items found.\n" );
1164
1165 # Delete as appropriate
1166 if ( $delete && $count ) {
1167 $this->output( 'Deleting...' );
1168 $dbw->delete( 'text', [ 'old_id' => $old ], __METHOD__ );
1169 $this->output( "done.\n" );
1170 }
1171
1172 # Done
1173 $this->commitTransaction( $dbw, __METHOD__ );
1174 }
1175
1176 /**
1177 * Get the maintenance directory.
1178 * @return string
1179 */
1180 protected function getDir() {
1181 return __DIR__;
1182 }
1183
1184 /**
1185 * Returns a database to be used by current maintenance script. It can be set by setDB().
1186 * If not set, wfGetDB() will be used.
1187 * This function has the same parameters as wfGetDB()
1188 *
1189 * @param integer $db DB index (DB_SLAVE/DB_MASTER)
1190 * @param array $groups; default: empty array
1191 * @param string|bool $wiki; default: current wiki
1192 * @return IDatabase
1193 */
1194 protected function getDB( $db, $groups = [], $wiki = false ) {
1195 if ( is_null( $this->mDb ) ) {
1196 return wfGetDB( $db, $groups, $wiki );
1197 } else {
1198 return $this->mDb;
1199 }
1200 }
1201
1202 /**
1203 * Sets database object to be returned by getDB().
1204 *
1205 * @param IDatabase $db Database object to be used
1206 */
1207 public function setDB( IDatabase $db ) {
1208 $this->mDb = $db;
1209 }
1210
1211 /**
1212 * Begin a transcation on a DB
1213 *
1214 * This method makes it clear that begin() is called from a maintenance script,
1215 * which has outermost scope. This is safe, unlike $dbw->begin() called in other places.
1216 *
1217 * @param IDatabase $dbw
1218 * @param string $fname Caller name
1219 * @since 1.27
1220 */
1221 protected function beginTransaction( IDatabase $dbw, $fname ) {
1222 $dbw->begin( $fname );
1223 }
1224
1225 /**
1226 * Commit the transcation on a DB handle and wait for slaves to catch up
1227 *
1228 * This method makes it clear that commit() is called from a maintenance script,
1229 * which has outermost scope. This is safe, unlike $dbw->commit() called in other places.
1230 *
1231 * @param IDatabase $dbw
1232 * @param string $fname Caller name
1233 * @return bool Whether the slave wait succeeded
1234 * @since 1.27
1235 */
1236 protected function commitTransaction( IDatabase $dbw, $fname ) {
1237 $dbw->commit( $fname );
1238
1239 $ok = wfWaitForSlaves( $this->lastSlaveWait, false, '*', 30 );
1240 $this->lastSlaveWait = microtime( true );
1241
1242 return $ok;
1243 }
1244
1245 /**
1246 * Rollback the transcation on a DB handle
1247 *
1248 * This method makes it clear that rollback() is called from a maintenance script,
1249 * which has outermost scope. This is safe, unlike $dbw->rollback() called in other places.
1250 *
1251 * @param IDatabase $dbw
1252 * @param string $fname Caller name
1253 * @since 1.27
1254 */
1255 protected function rollbackTransaction( IDatabase $dbw, $fname ) {
1256 $dbw->rollback( $fname );
1257 }
1258
1259 /**
1260 * Lock the search index
1261 * @param DatabaseBase &$db
1262 */
1263 private function lockSearchindex( $db ) {
1264 $write = [ 'searchindex' ];
1265 $read = [
1266 'page',
1267 'revision',
1268 'text',
1269 'interwiki',
1270 'l10n_cache',
1271 'user',
1272 'page_restrictions'
1273 ];
1274 $db->lockTables( $read, $write, __CLASS__ . '::' . __METHOD__ );
1275 }
1276
1277 /**
1278 * Unlock the tables
1279 * @param DatabaseBase &$db
1280 */
1281 private function unlockSearchindex( $db ) {
1282 $db->unlockTables( __CLASS__ . '::' . __METHOD__ );
1283 }
1284
1285 /**
1286 * Unlock and lock again
1287 * Since the lock is low-priority, queued reads will be able to complete
1288 * @param DatabaseBase &$db
1289 */
1290 private function relockSearchindex( $db ) {
1291 $this->unlockSearchindex( $db );
1292 $this->lockSearchindex( $db );
1293 }
1294
1295 /**
1296 * Perform a search index update with locking
1297 * @param int $maxLockTime The maximum time to keep the search index locked.
1298 * @param string $callback The function that will update the function.
1299 * @param DatabaseBase $dbw
1300 * @param array $results
1301 */
1302 public function updateSearchIndex( $maxLockTime, $callback, $dbw, $results ) {
1303 $lockTime = time();
1304
1305 # Lock searchindex
1306 if ( $maxLockTime ) {
1307 $this->output( " --- Waiting for lock ---" );
1308 $this->lockSearchindex( $dbw );
1309 $lockTime = time();
1310 $this->output( "\n" );
1311 }
1312
1313 # Loop through the results and do a search update
1314 foreach ( $results as $row ) {
1315 # Allow reads to be processed
1316 if ( $maxLockTime && time() > $lockTime + $maxLockTime ) {
1317 $this->output( " --- Relocking ---" );
1318 $this->relockSearchindex( $dbw );
1319 $lockTime = time();
1320 $this->output( "\n" );
1321 }
1322 call_user_func( $callback, $dbw, $row );
1323 }
1324
1325 # Unlock searchindex
1326 if ( $maxLockTime ) {
1327 $this->output( " --- Unlocking --" );
1328 $this->unlockSearchindex( $dbw );
1329 $this->output( "\n" );
1330 }
1331 }
1332
1333 /**
1334 * Update the searchindex table for a given pageid
1335 * @param DatabaseBase $dbw A database write handle
1336 * @param int $pageId The page ID to update.
1337 * @return null|string
1338 */
1339 public function updateSearchIndexForPage( $dbw, $pageId ) {
1340 // Get current revision
1341 $rev = Revision::loadFromPageId( $dbw, $pageId );
1342 $title = null;
1343 if ( $rev ) {
1344 $titleObj = $rev->getTitle();
1345 $title = $titleObj->getPrefixedDBkey();
1346 $this->output( "$title..." );
1347 # Update searchindex
1348 $u = new SearchUpdate( $pageId, $titleObj->getText(), $rev->getContent() );
1349 $u->doUpdate();
1350 $this->output( "\n" );
1351 }
1352
1353 return $title;
1354 }
1355
1356 /**
1357 * Wrapper for posix_isatty()
1358 * We default as considering stdin a tty (for nice readline methods)
1359 * but treating stout as not a tty to avoid color codes
1360 *
1361 * @param mixed $fd File descriptor
1362 * @return bool
1363 */
1364 public static function posix_isatty( $fd ) {
1365 if ( !function_exists( 'posix_isatty' ) ) {
1366 return !$fd;
1367 } else {
1368 return posix_isatty( $fd );
1369 }
1370 }
1371
1372 /**
1373 * Prompt the console for input
1374 * @param string $prompt What to begin the line with, like '> '
1375 * @return string Response
1376 */
1377 public static function readconsole( $prompt = '> ' ) {
1378 static $isatty = null;
1379 if ( is_null( $isatty ) ) {
1380 $isatty = self::posix_isatty( 0 /*STDIN*/ );
1381 }
1382
1383 if ( $isatty && function_exists( 'readline' ) ) {
1384 $resp = readline( $prompt );
1385 if ( $resp === null ) {
1386 // Workaround for https://github.com/facebook/hhvm/issues/4776
1387 return false;
1388 } else {
1389 return $resp;
1390 }
1391 } else {
1392 if ( $isatty ) {
1393 $st = self::readlineEmulation( $prompt );
1394 } else {
1395 if ( feof( STDIN ) ) {
1396 $st = false;
1397 } else {
1398 $st = fgets( STDIN, 1024 );
1399 }
1400 }
1401 if ( $st === false ) {
1402 return false;
1403 }
1404 $resp = trim( $st );
1405
1406 return $resp;
1407 }
1408 }
1409
1410 /**
1411 * Emulate readline()
1412 * @param string $prompt What to begin the line with, like '> '
1413 * @return string
1414 */
1415 private static function readlineEmulation( $prompt ) {
1416 $bash = Installer::locateExecutableInDefaultPaths( [ 'bash' ] );
1417 if ( !wfIsWindows() && $bash ) {
1418 $retval = false;
1419 $encPrompt = wfEscapeShellArg( $prompt );
1420 $command = "read -er -p $encPrompt && echo \"\$REPLY\"";
1421 $encCommand = wfEscapeShellArg( $command );
1422 $line = wfShellExec( "$bash -c $encCommand", $retval, [], [ 'walltime' => 0 ] );
1423
1424 if ( $retval == 0 ) {
1425 return $line;
1426 } elseif ( $retval == 127 ) {
1427 // Couldn't execute bash even though we thought we saw it.
1428 // Shell probably spit out an error message, sorry :(
1429 // Fall through to fgets()...
1430 } else {
1431 // EOF/ctrl+D
1432 return false;
1433 }
1434 }
1435
1436 // Fallback... we'll have no editing controls, EWWW
1437 if ( feof( STDIN ) ) {
1438 return false;
1439 }
1440 print $prompt;
1441
1442 return fgets( STDIN, 1024 );
1443 }
1444 }
1445
1446 /**
1447 * Fake maintenance wrapper, mostly used for the web installer/updater
1448 */
1449 class FakeMaintenance extends Maintenance {
1450 protected $mSelf = "FakeMaintenanceScript";
1451
1452 public function execute() {
1453 return;
1454 }
1455 }
1456
1457 /**
1458 * Class for scripts that perform database maintenance and want to log the
1459 * update in `updatelog` so we can later skip it
1460 */
1461 abstract class LoggedUpdateMaintenance extends Maintenance {
1462 public function __construct() {
1463 parent::__construct();
1464 $this->addOption( 'force', 'Run the update even if it was completed already' );
1465 $this->setBatchSize( 200 );
1466 }
1467
1468 public function execute() {
1469 $db = $this->getDB( DB_MASTER );
1470 $key = $this->getUpdateKey();
1471
1472 if ( !$this->hasOption( 'force' )
1473 && $db->selectRow( 'updatelog', '1', [ 'ul_key' => $key ], __METHOD__ )
1474 ) {
1475 $this->output( "..." . $this->updateSkippedMessage() . "\n" );
1476
1477 return true;
1478 }
1479
1480 if ( !$this->doDBUpdates() ) {
1481 return false;
1482 }
1483
1484 if ( $db->insert( 'updatelog', [ 'ul_key' => $key ], __METHOD__, 'IGNORE' ) ) {
1485 return true;
1486 } else {
1487 $this->output( $this->updatelogFailedMessage() . "\n" );
1488
1489 return false;
1490 }
1491 }
1492
1493 /**
1494 * Message to show that the update was done already and was just skipped
1495 * @return string
1496 */
1497 protected function updateSkippedMessage() {
1498 $key = $this->getUpdateKey();
1499
1500 return "Update '{$key}' already logged as completed.";
1501 }
1502
1503 /**
1504 * Message to show that the update log was unable to log the completion of this update
1505 * @return string
1506 */
1507 protected function updatelogFailedMessage() {
1508 $key = $this->getUpdateKey();
1509
1510 return "Unable to log update '{$key}' as completed.";
1511 }
1512
1513 /**
1514 * Do the actual work. All child classes will need to implement this.
1515 * Return true to log the update as done or false (usually on failure).
1516 * @return bool
1517 */
1518 abstract protected function doDBUpdates();
1519
1520 /**
1521 * Get the update key name to go in the update log table
1522 * @return string
1523 */
1524 abstract protected function getUpdateKey();
1525 }