includes/utils/AutoloadGenerator.php

   1 <?php
   2
   3 /**
   4  * Accepts a list of files and directories to search for
   5  * php files and generates $wgAutoloadLocalClasses or $wgAutoloadClasses
   6  * lines for all detected classes. These lines are written out
   7  * to an autoload.php file in the projects provided basedir.
   8  *
   9  * Usage:
  10  *
  11  *     $gen = new AutoloadGenerator( __DIR__ );
  12  *     $gen->readDir( __DIR__ . '/includes' );
  13  *     $gen->readFile( __DIR__ . '/foo.php' )
  14  *     $gen->getAutoload();
  15  */
  16 class AutoloadGenerator {
  17         const FILETYPE_JSON = 'json';
  18         const FILETYPE_PHP = 'php';
  19
  20         /**
  21          * @var string Root path of the project being scanned for classes
  22          */
  23         protected $basepath;
  24
  25         /**
  26          * @var ClassCollector Helper class extracts class names from php files
  27          */
  28         protected $collector;
  29
  30         /**
  31          * @var array Map of file shortpath to list of FQCN detected within file
  32          */
  33         protected $classes = [];
  34
  35         /**
  36          * @var string The global variable to write output to
  37          */
  38         protected $variableName = 'wgAutoloadClasses';
  39
  40         /**
  41          * @var array Map of FQCN to relative path(from self::$basepath)
  42          */
  43         protected $overrides = [];
  44
  45         /**
  46          * Directories that should be excluded
  47          *
  48          * @var string[]
  49          */
  50         protected $excludePaths = [];
  51
  52         /**
  53          * @param string $basepath Root path of the project being scanned for classes
  54          * @param array|string $flags
  55          *
  56          *  local - If this flag is set $wgAutoloadLocalClasses will be build instead
  57          *          of $wgAutoloadClasses
  58          */
  59         public function __construct( $basepath, $flags = [] ) {
  60                 if ( !is_array( $flags ) ) {
  61                         $flags = [ $flags ];
  62                 }
  63                 $this->basepath = self::normalizePathSeparator( realpath( $basepath ) );
  64                 $this->collector = new ClassCollector;
  65                 if ( in_array( 'local', $flags ) ) {
  66                         $this->variableName = 'wgAutoloadLocalClasses';
  67                 }
  68         }
  69
  70         /**
  71          * Directories that should be excluded
  72          *
  73          * @since 1.31
  74          * @param string[] $paths
  75          */
  76         public function setExcludePaths( array $paths ) {
  77                 foreach ( $paths as $path ) {
  78                         $this->excludePaths[] = self::normalizePathSeparator( $path );
  79                 }
  80         }
  81
  82         /**
  83          * Whether the file should be excluded
  84          *
  85          * @param string $path File path
  86          * @return bool
  87          */
  88         private function shouldExclude( $path ) {
  89                 foreach ( $this->excludePaths as $dir ) {
  90                         if ( strpos( $path, $dir ) === 0 ) {
  91                                 return true;
  92                         }
  93                 }
  94
  95                 return false;
  96         }
  97
  98         /**
  99          * Force a class to be autoloaded from a specific path, regardless of where
 100          * or if it was detected.
 101          *
 102          * @param string $fqcn FQCN to force the location of
 103          * @param string $inputPath Full path to the file containing the class
 104          * @throws Exception
 105          */
 106         public function forceClassPath( $fqcn, $inputPath ) {
 107                 $path = self::normalizePathSeparator( realpath( $inputPath ) );
 108                 if ( !$path ) {
 109                         throw new \Exception( "Invalid path: $inputPath" );
 110                 }
 111                 $len = strlen( $this->basepath );
 112                 if ( substr( $path, 0, $len ) !== $this->basepath ) {
 113                         throw new \Exception( "Path is not within basepath: $inputPath" );
 114                 }
 115                 $shortpath = substr( $path, $len );
 116                 $this->overrides[$fqcn] = $shortpath;
 117         }
 118
 119         /**
 120          * @param string $inputPath Path to a php file to find classes within
 121          * @throws Exception
 122          */
 123         public function readFile( $inputPath ) {
 124                 // NOTE: do NOT expand $inputPath using realpath(). It is perfectly
 125                 // reasonable for LocalSettings.php and similiar files to be symlinks
 126                 // to files that are outside of $this->basepath.
 127                 $inputPath = self::normalizePathSeparator( $inputPath );
 128                 $len = strlen( $this->basepath );
 129                 if ( substr( $inputPath, 0, $len ) !== $this->basepath ) {
 130                         throw new \Exception( "Path is not within basepath: $inputPath" );
 131                 }
 132                 if ( $this->shouldExclude( $inputPath ) ) {
 133                         return;
 134                 }
 135                 $result = $this->collector->getClasses(
 136                         file_get_contents( $inputPath )
 137                 );
 138                 if ( $result ) {
 139                         $shortpath = substr( $inputPath, $len );
 140                         $this->classes[$shortpath] = $result;
 141                 }
 142         }
 143
 144         /**
 145          * @param string $dir Path to a directory to recursively search
 146          *  for php files with either .php or .inc extensions
 147          */
 148         public function readDir( $dir ) {
 149                 $it = new RecursiveDirectoryIterator(
 150                         self::normalizePathSeparator( realpath( $dir ) ) );
 151                 $it = new RecursiveIteratorIterator( $it );
 152
 153                 foreach ( $it as $path => $file ) {
 154                         $ext = pathinfo( $path, PATHINFO_EXTENSION );
 155                         // some older files in mw use .inc
 156                         if ( $ext === 'php' || $ext === 'inc' ) {
 157                                 $this->readFile( $path );
 158                         }
 159                 }
 160         }
 161
 162         /**
 163          * Updates the AutoloadClasses field at the given
 164          * filename.
 165          *
 166          * @param string $filename Filename of JSON
 167          *  extension/skin registration file
 168          * @return string Updated Json of the file given as the $filename parameter
 169          */
 170         protected function generateJsonAutoload( $filename ) {
 171                 $key = 'AutoloadClasses';
 172                 $json = FormatJson::decode( file_get_contents( $filename ), true );
 173                 unset( $json[$key] );
 174                 // Inverting the key-value pairs so that they become of the
 175                 // format class-name : path when they get converted into json.
 176                 foreach ( $this->classes as $path => $contained ) {
 177                         foreach ( $contained as $fqcn ) {
 178                                 // Using substr to remove the leading '/'
 179                                 $json[$key][$fqcn] = substr( $path, 1 );
 180                         }
 181                 }
 182                 foreach ( $this->overrides as $path => $fqcn ) {
 183                         // Using substr to remove the leading '/'
 184                         $json[$key][$fqcn] = substr( $path, 1 );
 185                 }
 186
 187                 // Sorting the list of autoload classes.
 188                 ksort( $json[$key] );
 189
 190                 // Return the whole JSON file
 191                 return FormatJson::encode( $json, "\t", FormatJson::ALL_OK ) . "\n";
 192         }
 193
 194         /**
 195          * Generates a PHP file setting up autoload information.
 196          *
 197          * @param string $commandName Command name to include in comment
 198          * @param string $filename of PHP file to put autoload information in.
 199          * @return string
 200          */
 201         protected function generatePHPAutoload( $commandName, $filename ) {
 202                 // No existing JSON file found; update/generate PHP file
 203                 $content = [];
 204
 205                 // We need to generate a line each rather than exporting the
 206                 // full array so __DIR__ can be prepended to all the paths
 207                 $format = "%s => __DIR__ . %s,";
 208                 foreach ( $this->classes as $path => $contained ) {
 209                         $exportedPath = var_export( $path, true );
 210                         foreach ( $contained as $fqcn ) {
 211                                 $content[$fqcn] = sprintf(
 212                                         $format,
 213                                         var_export( $fqcn, true ),
 214                                         $exportedPath
 215                                 );
 216                         }
 217                 }
 218
 219                 foreach ( $this->overrides as $fqcn => $path ) {
 220                         $content[$fqcn] = sprintf(
 221                                 $format,
 222                                 var_export( $fqcn, true ),
 223                                 var_export( $path, true )
 224                         );
 225                 }
 226
 227                 // sort for stable output
 228                 ksort( $content );
 229
 230                 // extensions using this generator are appending to the existing
 231                 // autoload.
 232                 if ( $this->variableName === 'wgAutoloadClasses' ) {
 233                         $op = '+=';
 234                 } else {
 235                         $op = '=';
 236                 }
 237
 238                 $output = implode( "\n\t", $content );
 239                 return <<<EOD
 240 <?php
 241 // This file is generated by $commandName, do not adjust manually
 242 // phpcs:ignoreFile Generic.Files.LineLength
 243 global \${$this->variableName};
 244
 245 \${$this->variableName} {$op} [
 246         {$output}
 247 ];
 248
 249 EOD;
 250         }
 251
 252         /**
 253          * Returns all known classes as a string, which can be used to put into a target
 254          * file (e.g. extension.json, skin.json or autoload.php)
 255          *
 256          * @param string $commandName Value used in file comment to direct
 257          *  developers towards the appropriate way to update the autoload.
 258          * @return string
 259          */
 260         public function getAutoload( $commandName = 'AutoloadGenerator' ) {
 261                 // We need to check whether an extenson.json or skin.json exists or not, and
 262                 // incase it doesn't, update the autoload.php file.
 263
 264                 $fileinfo = $this->getTargetFileinfo();
 265
 266                 if ( $fileinfo['type'] === self::FILETYPE_JSON ) {
 267                         return $this->generateJsonAutoload( $fileinfo['filename'] );
 268                 } else {
 269                         return $this->generatePHPAutoload( $commandName, $fileinfo['filename'] );
 270                 }
 271         }
 272
 273         /**
 274          * Returns the filename of the extension.json of skin.json, if there's any, or
 275          * otherwise the path to the autoload.php file in an array as the "filename"
 276          * key and with the type (AutoloadGenerator::FILETYPE_JSON or AutoloadGenerator::FILETYPE_PHP)
 277          * of the file as the "type" key.
 278          *
 279          * @return array
 280          */
 281         public function getTargetFileinfo() {
 282                 $fileinfo = [
 283                         'filename' => $this->basepath . '/autoload.php',
 284                         'type' => self::FILETYPE_PHP
 285                 ];
 286                 if ( file_exists( $this->basepath . '/extension.json' ) ) {
 287                         $fileinfo = [
 288                                 'filename' => $this->basepath . '/extension.json',
 289                                 'type' => self::FILETYPE_JSON
 290                         ];
 291                 } elseif ( file_exists( $this->basepath . '/skin.json' ) ) {
 292                         $fileinfo = [
 293                                 'filename' => $this->basepath . '/skin.json',
 294                                 'type' => self::FILETYPE_JSON
 295                         ];
 296                 }
 297
 298                 return $fileinfo;
 299         }
 300
 301         /**
 302          * Ensure that Unix-style path separators ("/") are used in the path.
 303          *
 304          * @param string $path
 305          * @return string
 306          */
 307         protected static function normalizePathSeparator( $path ) {
 308                 return str_replace( '\\', '/', $path );
 309         }
 310
 311         /**
 312          * Initialize the source files and directories which are used for the MediaWiki default
 313          * autoloader in {mw-base-dir}/autoload.php including:
 314          *  * includes/
 315          *  * languages/
 316          *  * maintenance/
 317          *  * mw-config/
 318          *  * /*.php
 319          */
 320         public function initMediaWikiDefault() {
 321                 foreach ( [ 'includes', 'languages', 'maintenance', 'mw-config' ] as $dir ) {
 322                         $this->readDir( $this->basepath . '/' . $dir );
 323                 }
 324                 foreach ( glob( $this->basepath . '/*.php' ) as $file ) {
 325                         $this->readFile( $file );
 326                 }
 327         }
 328 }
 329
 330 /**
 331  * Reads PHP code and returns the FQCN of every class defined within it.
 332  */
 333 class ClassCollector {
 334
 335         /**
 336          * @var string Current namespace
 337          */
 338         protected $namespace = '';
 339
 340         /**
 341          * @var array List of FQCN detected in this pass
 342          */
 343         protected $classes;
 344
 345         /**
 346          * @var array Token from token_get_all() that started an expect sequence
 347          */
 348         protected $startToken;
 349
 350         /**
 351          * @var array List of tokens that are members of the current expect sequence
 352          */
 353         protected $tokens;
 354
 355         /**
 356          * @var array Class alias with target/name fields
 357          */
 358         protected $alias;
 359
 360         /**
 361          * @param string $code PHP code (including <?php) to detect class names from
 362          * @return array List of FQCN detected within the tokens
 363          */
 364         public function getClasses( $code ) {
 365                 $this->namespace = '';
 366                 $this->classes = [];
 367                 $this->startToken = null;
 368                 $this->alias = null;
 369                 $this->tokens = [];
 370
 371                 foreach ( token_get_all( $code ) as $token ) {
 372                         if ( $this->startToken === null ) {
 373                                 $this->tryBeginExpect( $token );
 374                         } else {
 375                                 $this->tryEndExpect( $token );
 376                         }
 377                 }
 378
 379                 return $this->classes;
 380         }
 381
 382         /**
 383          * Determine if $token begins the next expect sequence.
 384          *
 385          * @param array $token
 386          */
 387         protected function tryBeginExpect( $token ) {
 388                 if ( is_string( $token ) ) {
 389                         return;
 390                 }
 391                 // Note: When changing class name discovery logic,
 392                 // AutoLoaderTest.php may also need to be updated.
 393                 switch ( $token[0] ) {
 394                         case T_NAMESPACE:
 395                         case T_CLASS:
 396                         case T_INTERFACE:
 397                         case T_TRAIT:
 398                         case T_DOUBLE_COLON:
 399                                 $this->startToken = $token;
 400                                 break;
 401                         case T_STRING:
 402                                 if ( $token[1] === 'class_alias' ) {
 403                                         $this->startToken = $token;
 404                                         $this->alias = [];
 405                                 }
 406                 }
 407         }
 408
 409         /**
 410          * Accepts the next token in an expect sequence
 411          *
 412          * @param array $token
 413          */
 414         protected function tryEndExpect( $token ) {
 415                 switch ( $this->startToken[0] ) {
 416                         case T_DOUBLE_COLON:
 417                                 // Skip over T_CLASS after T_DOUBLE_COLON because this is something like
 418                                 // "self::static" which accesses the class name. It doens't define a new class.
 419                                 $this->startToken = null;
 420                                 break;
 421                         case T_NAMESPACE:
 422                                 if ( $token === ';' || $token === '{' ) {
 423                                         $this->namespace = $this->implodeTokens() . '\\';
 424                                 } else {
 425                                         $this->tokens[] = $token;
 426                                 }
 427                                 break;
 428
 429                         case T_STRING:
 430                                 if ( $this->alias !== null ) {
 431                                         // Flow 1 - Two string literals:
 432                                         // - T_STRING  class_alias
 433                                         // - '('
 434                                         // - T_CONSTANT_ENCAPSED_STRING 'TargetClass'
 435                                         // - ','
 436                                         // - T_WHITESPACE
 437                                         // - T_CONSTANT_ENCAPSED_STRING 'AliasName'
 438                                         // - ')'
 439                                         // Flow 2 - Use of ::class syntax for first parameter
 440                                         // - T_STRING  class_alias
 441                                         // - '('
 442                                         // - T_STRING TargetClass
 443                                         // - T_DOUBLE_COLON ::
 444                                         // - T_CLASS class
 445                                         // - ','
 446                                         // - T_WHITESPACE
 447                                         // - T_CONSTANT_ENCAPSED_STRING 'AliasName'
 448                                         // - ')'
 449                                         if ( $token === '(' ) {
 450                                                 // Start of a function call to class_alias()
 451                                                 $this->alias = [ 'target' => false, 'name' => false ];
 452                                         } elseif ( $token === ',' ) {
 453                                                 // Record that we're past the first parameter
 454                                                 if ( $this->alias['target'] === false ) {
 455                                                         $this->alias['target'] = true;
 456                                                 }
 457                                         } elseif ( is_array( $token ) && $token[0] === T_CONSTANT_ENCAPSED_STRING ) {
 458                                                 if ( $this->alias['target'] === true ) {
 459                                                         // We already saw a first argument, this must be the second.
 460                                                         // Strip quotes from the string literal.
 461                                                         $this->alias['name'] = substr( $token[1], 1, -1 );
 462                                                 }
 463                                         } elseif ( $token === ')' ) {
 464                                                 // End of function call
 465                                                 $this->classes[] = $this->alias['name'];
 466                                                 $this->alias = null;
 467                                                 $this->startToken = null;
 468                                         } elseif ( !is_array( $token ) || (
 469                                                 $token[0] !== T_STRING &&
 470                                                 $token[0] !== T_DOUBLE_COLON &&
 471                                                 $token[0] !== T_CLASS &&
 472                                                 $token[0] !== T_WHITESPACE
 473                                         ) ) {
 474                                                 // Ignore this call to class_alias() - compat/Timestamp.php
 475                                                 $this->alias = null;
 476                                                 $this->startToken = null;
 477                                         }
 478                                 }
 479                                 break;
 480
 481                         case T_CLASS:
 482                         case T_INTERFACE:
 483                         case T_TRAIT:
 484                                 $this->tokens[] = $token;
 485                                 if ( is_array( $token ) && $token[0] === T_STRING ) {
 486                                         $this->classes[] = $this->namespace . $this->implodeTokens();
 487                                 }
 488                 }
 489         }
 490
 491         /**
 492          * Returns the string representation of the tokens within the
 493          * current expect sequence and resets the sequence.
 494          *
 495          * @return string
 496          */
 497         protected function implodeTokens() {
 498                 $content = [];
 499                 foreach ( $this->tokens as $token ) {
 500                         $content[] = is_string( $token ) ? $token : $token[1];
 501                 }
 502
 503                 $this->tokens = [];
 504                 $this->startToken = null;
 505
 506                 return trim( implode( '', $content ), " \n\t" );
 507         }
 508 }