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