maintenance/Maintenance.php

   1 <?php
   2 // Define this so scripts can easily find doMaintenance.php
   3 define( 'DO_MAINTENANCE', dirname(__FILE__) . '/doMaintenance.php' );
   4
   5 /**
   6  * Abstract maintenance class for quickly writing and churning out
   7  * maintenance scripts with minimal effort. All that _must_ be defined
   8  * is the execute() method. See docs/maintenance.txt for more info
   9  * and a quick demo of how to use it.
  10  *
  11  * @author Chad Horohoe <chad@anyonecanedit.org>
  12  * @since 1.16
  13  * @ingroup Maintenance
  14  */
  15 abstract class Maintenance {
  16
  17         /**
  18          * Constants for DB access type
  19          * @see Maintenance::getDbType()
  20          */
  21         const NO_DB     = 0;
  22         const NORMAL_DB = 1;
  23         const ADMIN_DB  = 2;
  24
  25         // This is the desired params
  26         private $mParams = array();
  27
  28         // Array of desired args
  29         private $mArgList = array();
  30
  31         // This is the list of options that were actually passed
  32         private $mOptions = array();
  33
  34         // This is the list of arguments that were actually passed
  35         protected $mArgs = array();
  36
  37         // Name of the script currently running
  38         protected $mSelf;
  39
  40         // Special vars for params that are always used
  41         private $mQuiet = false;
  42         private $mDbUser, $mDbPass;
  43
  44         // A description of the script, children should change this
  45         protected $mDescription = '';
  46
  47         // Have we already loaded our user input?
  48         private $inputLoaded = false;
  49
  50         // Batch size
  51         protected $mBatchSize = 100;
  52
  53         /**
  54          * Default constructor. Children should call this if implementing
  55          * their own constructors
  56          */
  57         public function __construct() {
  58                 $this->addDefaultParams();
  59         }
  60
  61         /**
  62          * Do the actual work. All child classes will need to implement this
  63          */
  64         abstract public function execute();
  65
  66         /**
  67          * Add a parameter to the script. Will be displayed on --help
  68          * with the associated description
  69          *
  70          * @param $name String The name of the param (help, version, etc)
  71          * @param $description String The description of the param to show on --help
  72          * @param $required boolean Is the param required?
  73          * @param $withArg Boolean Is an argument required with this option?
  74          */
  75         protected function addParam( $name, $description, $required = false, $withArg = false ) {
  76                 $this->mParams[ $name ] = array( 'desc' => $description, 'require' => $required, 'withArg' => $withArg );
  77         }
  78
  79         /**
  80          * Checks to see if a particular param exists.
  81          * @param $name String The name of the param
  82          * @return boolean
  83          */
  84         protected function hasOption( $name ) {
  85                 return isset( $this->mOptions[ $name ] );
  86         }
  87
  88         /**
  89          * Get an option, or return the default
  90          * @param $name String The name of the param
  91          * @param $default mixed Anything you want, default null
  92          * @return mixed
  93          */
  94         protected function getOption( $name, $default = null ) {
  95                 if( $this->hasOption($name) ) {
  96                         return $this->mOptions[$name];
  97                 } else {
  98                         // Set it so we don't have to provide the default again
  99                         $this->mOptions[$name] = $default;
 100                         return $this->mOptions[$name];
 101                 }
 102         }
 103
 104         /**
 105          * Add some args that are needed. Used in formatting help
 106          */
 107         protected function addArgs( $args ) {
 108                 $this->mArgList = array_merge( $this->mArgList, $args );
 109         }
 110
 111         /**
 112          * Does a given argument exist?
 113          * @param $argId int The integer value (from zero) for the arg
 114          * @return boolean
 115          */
 116         protected function hasArg( $argId = 0 ) {
 117                 return isset( $this->mArgs[ $argId ] ) ;
 118         }
 119
 120         /**
 121          * Get an argument.
 122          * @param $argId int The integer value (from zero) for the arg
 123          * @param $default mixed The default if it doesn't exist
 124          * @return mixed
 125          */
 126         protected function getArg( $argId = 0, $default = null ) {
 127                 return $this->hasArg($name) ? $this->mArgs[$name] : $default;
 128         }
 129
 130         /**
 131          * Set the batch size.
 132          * @param $s int The number of operations to do in a batch
 133          */
 134         protected function setBatchSize( $s = 0 ) {
 135                 $this->mBatchSize = $s;
 136         }
 137
 138         /**
 139          * Return input from stdin.
 140          * @param $length int The number of bytes to read. If null,
 141          *        just return the handle
 142          * @return mixed
 143          */
 144         protected function getStdin( $len = null ) {
 145                 $f = fopen( 'php://stdin', 'rt' );
 146                 if( !$len ) {
 147                         return $f;
 148                 }
 149                 $input = fgets( $f, $len );
 150                 fclose ( $f );
 151                 return rtrim( $input );
 152         }
 153
 154         /**
 155          * Throw some output to the user. Scripts can call this with no fears,
 156          * as we handle all --quiet stuff here
 157          * @param $out String The text to show to the user
 158          */
 159         protected function output( $out ) {
 160                 if( $this->mQuiet ) {
 161                         return;
 162                 }
 163                 $f = fopen( 'php://stdout', 'w' );
 164                 fwrite( $f, $out );
 165                 fclose( $f );
 166         }
 167
 168         /**
 169          * Throw an error to the user. Doesn't respect --quiet, so don't use
 170          * this for non-error output
 171          * @param $err String The error to display
 172          * @param $die boolean If true, go ahead and die out.
 173          */
 174         protected function error( $err, $die = false ) {
 175                 $f = fopen( 'php://stderr', 'w' );
 176                 fwrite( $f, $err );
 177                 fclose( $f );
 178                 if( $die ) die();
 179         }
 180
 181         /**
 182          * Does the script need normal DB access? By default, we give Maintenance
 183          * scripts admin rights to the DB (when available). Sometimes, a script needs
 184          * normal access for a reason and sometimes they want no access. Subclasses
 185          * should override and return one of the following values, as needed:
 186          *    Maintenance::NO_DB      -  For no DB access at all
 187          *    Maintenance::NORMAL_DB  -  For normal DB access
 188          *    Maintenance::ADMIN_DB   -  For admin DB access, default
 189          * @return int
 190          */
 191         protected function getDbType() {
 192                 return Maintenance :: ADMIN_DB;
 193         }
 194
 195         /**
 196          * Add the default parameters to the scripts
 197          */
 198         private function addDefaultParams() {
 199                 $this->addParam( 'help', "Display this help message" );
 200                 $this->addParam( 'quiet', "Whether to supress non-error output" );
 201                 $this->addParam( 'conf', "Location of LocalSettings.php, if not default", false, true );
 202                 $this->addParam( 'wiki', "For specifying the wiki ID", false, true );
 203                 if( $this->getDbType() > 0 ) {
 204                         $this->addParam( 'dbuser', "The DB user to use for this script", false, true );
 205                         $this->addParam( 'dbpass', "The password to use for this script", false, true );
 206                 }
 207         }
 208
 209         /**
 210          * Spawn a child maintenance script. Pass all of the current arguments
 211          * to it.
 212          * @param $maintClass String A name of a child maintenance class
 213          * @param $classFile String Full path of where the child is
 214          * @return Maintenance child
 215          */
 216         protected function spawnChild( $maintClass, $classFile = null ) {
 217                 // If we haven't already specified, kill setup procedures
 218                 // for child scripts, we've already got a sane environment
 219                 if( !defined( 'MW_NO_SETUP' ) ) {
 220                         define( 'MW_NO_SETUP', true );
 221                 }
 222
 223                 // Make sure the class is loaded first
 224                 if( !class_exists( $maintClass ) ) {
 225                         if( $classFile ) {
 226                                 require_once( $classFile );
 227                         }
 228                         if( !class_exists( $maintClass ) ) {
 229                                 $this->error( "Cannot spawn child: $maintClass\n" );
 230                         }
 231                 }
 232
 233                 $child = new $maintClass();
 234                 $child->loadParamsAndArgs( $this->mSelf, $this->mOptions, $this->mArgs );
 235                 return $child;
 236         }
 237
 238         /**
 239          * Do some sanity checking and basic setup
 240          */
 241         public function setup() {
 242                 global $IP, $wgCommandLineMode, $wgUseNormalUser, $wgRequestTime;
 243
 244                 # Abort if called from a web server
 245                 if ( isset( $_SERVER ) && array_key_exists( 'REQUEST_METHOD', $_SERVER ) ) {
 246                         $this->error( "This script must be run from the command line\n", true );
 247                 }
 248
 249                 # Make sure we can handle script parameters
 250                 if( !ini_get( 'register_argc_argv' ) ) {
 251                         $this->error( "Cannot get command line arguments, register_argc_argv is set to false", true );
 252                 }
 253
 254                 # Make sure we're on PHP5 or better
 255                 if( version_compare( PHP_VERSION, '5.0.0' ) < 0 ) {
 256                         $this->error( "Sorry! This version of MediaWiki requires PHP 5; you are running " .
 257                                         PHP_VERSION . ".\n\n" .
 258                                         "If you are sure you already have PHP 5 installed, it may be installed\n" .
 259                                         "in a different path from PHP 4. Check with your system administrator.\n", true );
 260                 }
 261
 262                 if( version_compare( phpversion(), '5.2.4' ) >= 0 ) {
 263                         // Send PHP warnings and errors to stderr instead of stdout.
 264                         // This aids in diagnosing problems, while keeping messages
 265                         // out of redirected output.
 266                         if( ini_get( 'display_errors' ) ) {
 267                                 ini_set( 'display_errors', 'stderr' );
 268                         }
 269
 270                         // Don't touch the setting on earlier versions of PHP,
 271                         // as setting it would disable output if you'd wanted it.
 272
 273                         // Note that exceptions are also sent to stderr when
 274                         // command-line mode is on, regardless of PHP version.
 275                 }
 276
 277                 # Set the memory limit
 278                 ini_set( 'memory_limit', -1 );
 279
 280                 $wgRequestTime = microtime(true);
 281
 282                 # Define us as being in Mediawiki
 283                 define( 'MEDIAWIKI', true );
 284
 285                 # Setup $IP, using MW_INSTALL_PATH if it exists
 286                 $IP = strval( getenv('MW_INSTALL_PATH') ) !== ''
 287                         ? getenv('MW_INSTALL_PATH')
 288                         : realpath( dirname( __FILE__ ) . '/..' );
 289
 290                 $wgCommandLineMode = true;
 291                 # Turn off output buffering if it's on
 292                 @ob_end_flush();
 293
 294                 if (!isset( $wgUseNormalUser ) ) {
 295                         $wgUseNormalUser = false;
 296                 }
 297
 298                 $this->loadParamsAndArgs();
 299                 $this->maybeHelp();
 300         }
 301
 302         /**
 303          * Clear all params and arguments.
 304          */
 305         public function clearParamsAndArgs() {
 306                 $this->mOptions = array();
 307                 $this->mArgs = array();
 308                 $this->inputLoaded = false;
 309         }
 310
 311         /**
 312          * Process command line arguments
 313          * $mOptions becomes an array with keys set to the option names
 314          * $mArgs becomes a zero-based array containing the non-option arguments
 315          *
 316          * @param $self String The name of the script, if any
 317          * @param $opts Array An array of options, in form of key=>value
 318          * @param $args Array An array of command line arguments
 319          */
 320         public function loadParamsAndArgs( $self = null, $opts = null, $args = null ) {
 321                 # If we were given opts or args, set those and return early
 322                 if( $self ) {
 323                         $this->mSelf = $self;
 324                         $this->inputLoaded = true;
 325                 }
 326                 if( $opts ) {
 327                         $this->mOptions = $opts;
 328                         $this->inputLoaded = true;
 329                 }
 330                 if( $args ) {
 331                         $this->mArgs = $args;
 332                         $this->inputLoaded = true;
 333                 }
 334
 335                 # If we've already loaded input (either by user values or from $argv)
 336                 # skip on loading it again. The array_shift() will corrupt values if
 337                 # it's run again and again
 338                 if( $this->inputLoaded ) {
 339                         $this->loadSpecialVars();
 340                         return;
 341                 }
 342
 343                 global $argv;
 344                 $this->mSelf = array_shift( $argv );
 345
 346                 $options = array();
 347                 $args = array();
 348
 349                 # Parse arguments
 350                 for( $arg = reset( $argv ); $arg !== false; $arg = next( $argv ) ) {
 351                         if ( $arg == '--' ) {
 352                                 # End of options, remainder should be considered arguments
 353                                 $arg = next( $argv );
 354                                 while( $arg !== false ) {
 355                                         $args[] = $arg;
 356                                         $arg = next( $argv );
 357                                 }
 358                                 break;
 359                         } elseif ( substr( $arg, 0, 2 ) == '--' ) {
 360                                 # Long options
 361                                 $option = substr( $arg, 2 );
 362                                 if ( isset( $this->mParams[$option] ) && $this->mParams[$option]['withArg'] ) {
 363                                         $param = next( $argv );
 364                                         if ( $param === false ) {
 365                                                 $this->error( "$arg needs a value after it\n", true );
 366                                         }
 367                                         $options[$option] = $param;
 368                                 } else {
 369                                         $bits = explode( '=', $option, 2 );
 370                                         if( count( $bits ) > 1 ) {
 371                                                 $option = $bits[0];
 372                                                 $param = $bits[1];
 373                                         } else {
 374                                                 $param = 1;
 375                                         }
 376                                         $options[$option] = $param;
 377                                 }
 378                         } elseif ( substr( $arg, 0, 1 ) == '-' ) {
 379                                 # Short options
 380                                 for ( $p=1; $p<strlen( $arg ); $p++ ) {
 381                                         $option = $arg{$p};
 382                                         if ( isset( $this->mParams[$option]['withArg'] ) ) {
 383                                                 $param = next( $argv );
 384                                                 if ( $param === false ) {
 385                                                         $this->error( "$arg needs a value after it\n", true );
 386                                                 }
 387                                                 $options[$option] = $param;
 388                                         } else {
 389                                                 $options[$option] = 1;
 390                                         }
 391                                 }
 392                         } else {
 393                                 $args[] = $arg;
 394                         }
 395                 }
 396
 397                 # Check to make sure we've got all the required ones
 398                 foreach( $this->mParams as $opt => $info ) {
 399                         if( $info['require'] && !$this->hasOption($opt) ) {
 400                                 $this->error( "Param $opt required.\n", true );
 401                         }
 402                 }
 403
 404                 # Also make sure we've got enough arguments
 405                 if ( count( $args ) < count( $this->mArgList ) ) {
 406                         $this->error( "Not enough arguments passed", true );
 407                 }
 408
 409                 $this->mOptions = $options;
 410                 $this->mArgs = $args;
 411                 $this->loadSpecialVars();
 412                 $this->inputLoaded = true;
 413         }
 414
 415         /**
 416          * Handle the special variables that are global to all scripts
 417          */
 418         private function loadSpecialVars() {
 419                 if( $this->hasOption( 'dbuser' ) )
 420                         $this->mDbUser = $this->getOption( 'dbuser' );
 421                 if( $this->hasOption( 'dbpass' ) )
 422                         $this->mDbPass = $this->getOption( 'dbpass' );
 423                 if( $this->hasOption( 'quiet' ) )
 424                         $this->mQuiet = true;
 425         }
 426
 427         /**
 428          * Maybe show the help.
 429          * @param $force boolean Whether to force the help to show, default false
 430          */
 431         private function maybeHelp( $force = false ) {
 432                 if( $this->hasOption('help') || in_array( 'help', $this->mArgs ) || $force ) {
 433                         $this->mQuiet = false;
 434                         if( $this->mDescription ) {
 435                                 $this->output( $this->mDescription . "\n" );
 436                         }
 437                         $this->output( "\nUsage: php " . $this->mSelf );
 438                         if( $this->mParams ) {
 439                                 $this->output( " [--" . implode( array_keys( $this->mParams ), "|--" ) . "]" );
 440                         }
 441                         if( $this->mArgList ) {
 442                                 $this->output( " <" . implode( $this->mArgList, "> <" ) . ">" );
 443                         }
 444                         $this->output( "\n" );
 445                         foreach( $this->mParams as $par => $info ) {
 446                                 $this->output( "\t$par : " . $info['desc'] . "\n" );
 447                         }
 448                         die( 1 );
 449                 }
 450         }
 451
 452         /**
 453          * Handle some last-minute setup here.
 454          */
 455         private function finalSetup() {
 456                 global $wgCommandLineMode, $wgUseNormalUser, $wgShowSQLErrors;
 457                 global $wgTitle, $wgProfiling, $IP, $wgDBadminuser, $wgDBadminpassword;
 458                 global $wgDBuser, $wgDBpassword, $wgDBservers, $wgLBFactoryConf;
 459
 460                 # Turn off output buffering again, it might have been turned on in the settings files
 461                 if( ob_get_level() ) {
 462                         ob_end_flush();
 463                 }
 464                 # Same with these
 465                 $wgCommandLineMode = true;
 466
 467                 # If these were passed, use them
 468                 if( $this->mDbUser )
 469                         $wgDBadminuser = $this->mDbUser;
 470                 if( $this->mDbPass )
 471                         $wgDBadminpass = $this->mDbPass;
 472
 473                 if ( empty( $wgUseNormalUser ) && isset( $wgDBadminuser ) ) {
 474                         $wgDBuser = $wgDBadminuser;
 475                         $wgDBpassword = $wgDBadminpassword;
 476
 477                         if( $wgDBservers ) {
 478                                 foreach ( $wgDBservers as $i => $server ) {
 479                                         $wgDBservers[$i]['user'] = $wgDBuser;
 480                                         $wgDBservers[$i]['password'] = $wgDBpassword;
 481                                 }
 482                         }
 483                         if( isset( $wgLBFactoryConf['serverTemplate'] ) ) {
 484                                 $wgLBFactoryConf['serverTemplate']['user'] = $wgDBuser;
 485                                 $wgLBFactoryConf['serverTemplate']['password'] = $wgDBpassword;
 486                         }
 487                 }
 488
 489                 if ( defined( 'MW_CMDLINE_CALLBACK' ) ) {
 490                         $fn = MW_CMDLINE_CALLBACK;
 491                         $fn();
 492                 }
 493
 494                 $wgShowSQLErrors = true;
 495                 @set_time_limit( 0 );
 496
 497                 $wgProfiling = false; // only for Profiler.php mode; avoids OOM errors
 498         }
 499
 500         /**
 501          * Do setup specific to WMF
 502          */
 503         public function loadWikimediaSettings() {
 504                 global $IP, $wgNoDBParam, $wgUseNormalUser, $wgConf;
 505
 506                 if ( empty( $wgNoDBParam ) ) {
 507                         # Check if we were passed a db name
 508                         if ( isset( $this->mOptions['wiki'] ) ) {
 509                                 $db = $this->mOptions['wiki'];
 510                         } else {
 511                                 $db = array_shift( $this->mArgs );
 512                         }
 513                         list( $site, $lang ) = $wgConf->siteFromDB( $db );
 514
 515                         # If not, work out the language and site the old way
 516                         if ( is_null( $site ) || is_null( $lang ) ) {
 517                                 if ( !$db ) {
 518                                         $lang = 'aa';
 519                                 } else {
 520                                         $lang = $db;
 521                                 }
 522                                 if ( isset( $this->mArgs[0] ) ) {
 523                                         $site = array_shift( $this->mArgs );
 524                                 } else {
 525                                         $site = 'wikipedia';
 526                                 }
 527                         }
 528                 } else {
 529                         $lang = 'aa';
 530                         $site = 'wikipedia';
 531                 }
 532
 533                 # This is for the IRC scripts, which now run as the apache user
 534                 # The apache user doesn't have access to the wikiadmin_pass command
 535                 if ( $_ENV['USER'] == 'apache' ) {
 536                 #if ( posix_geteuid() == 48 ) {
 537                         $wgUseNormalUser = true;
 538                 }
 539
 540                 putenv( 'wikilang=' . $lang );
 541
 542                 $DP = $IP;
 543                 ini_set( 'include_path', ".:$IP:$IP/includes:$IP/languages:$IP/maintenance" );
 544
 545                 if ( $lang == 'test' && $site == 'wikipedia' ) {
 546                         define( 'TESTWIKI', 1 );
 547                 }
 548         }
 549
 550         /**
 551          * Generic setup for most installs. Returns the location of LocalSettings
 552          * @return String
 553          */
 554         public function loadSettings() {
 555                 global $wgWikiFarm, $wgCommandLineMode, $IP, $DP;
 556
 557                 $wgWikiFarm = false;
 558                 if ( isset( $this->mOptions['conf'] ) ) {
 559                         $settingsFile = $this->mOptions['conf'];
 560                 } else {
 561                         $settingsFile = "$IP/LocalSettings.php";
 562                 }
 563                 if ( isset( $this->mOptions['wiki'] ) ) {
 564                         $bits = explode( '-', $this->mOptions['wiki'] );
 565                         if ( count( $bits ) == 1 ) {
 566                                 $bits[] = '';
 567                         }
 568                         define( 'MW_DB', $bits[0] );
 569                         define( 'MW_PREFIX', $bits[1] );
 570                 }
 571
 572                 if ( ! is_readable( $settingsFile ) ) {
 573                         $this->error( "A copy of your installation's LocalSettings.php\n" .
 574                                                 "must exist and be readable in the source directory.\n", true );
 575                 }
 576                 $wgCommandLineMode = true;
 577                 $DP = $IP;
 578                 $this->finalSetup();
 579                 return $settingsFile;
 580         }
 581
 582         /**
 583          * Support function for cleaning up redundant text records
 584          * @param $delete boolean Whether or not to actually delete the records
 585          * @author Rob Church <robchur@gmail.com>
 586          */
 587         protected function purgeRedundantText( $delete = true ) {
 588                 # Data should come off the master, wrapped in a transaction
 589                 $dbw = wfGetDB( DB_MASTER );
 590                 $dbw->begin();
 591
 592                 $tbl_arc = $dbw->tableName( 'archive' );
 593                 $tbl_rev = $dbw->tableName( 'revision' );
 594                 $tbl_txt = $dbw->tableName( 'text' );
 595
 596                 # Get "active" text records from the revisions table
 597                 $this->output( "Searching for active text records in revisions table..." );
 598                 $res = $dbw->query( "SELECT DISTINCT rev_text_id FROM $tbl_rev" );
 599                 while( $row = $dbw->fetchObject( $res ) ) {
 600                         $cur[] = $row->rev_text_id;
 601                 }
 602                 $this->output( "done.\n" );
 603
 604                 # Get "active" text records from the archive table
 605                 $this->output( "Searching for active text records in archive table..." );
 606                 $res = $dbw->query( "SELECT DISTINCT ar_text_id FROM $tbl_arc" );
 607                 while( $row = $dbw->fetchObject( $res ) ) {
 608                         $cur[] = $row->ar_text_id;
 609                 }
 610                 $this->output( "done.\n" );
 611
 612                 # Get the IDs of all text records not in these sets
 613                 $this->output( "Searching for inactive text records..." );
 614                 $set = implode( ', ', $cur );
 615                 $res = $dbw->query( "SELECT old_id FROM $tbl_txt WHERE old_id NOT IN ( $set )" );
 616                 $old = array();
 617                 while( $row = $dbw->fetchObject( $res ) ) {
 618                         $old[] = $row->old_id;
 619                 }
 620                 $this->output( "done.\n" );
 621
 622                 # Inform the user of what we're going to do
 623                 $count = count( $old );
 624                 $this->output( "$count inactive items found.\n" );
 625
 626                 # Delete as appropriate
 627                 if( $delete && $count ) {
 628                         $this->output( "Deleting..." );
 629                         $set = implode( ', ', $old );
 630                         $dbw->query( "DELETE FROM $tbl_txt WHERE old_id IN ( $set )" );
 631                         $this->output( "done.\n" );
 632                 }
 633
 634                 # Done
 635                 $dbw->commit();
 636
 637         }
 638 }
 639