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