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                                 // Using substr to remove the leading '/'
 141                                 $json[$key][$fqcn] = substr( $path, 1 );
 142                         }
 143                 }
 144                 foreach ( $this->overrides as $path => $fqcn ) {
 145                         // Using substr to remove the leading '/'
 146                         $json[$key][$fqcn] = substr( $path, 1 );
 147                 }
 148
 149                 // Sorting the list of autoload classes.
 150                 ksort( $json[$key] );
 151
 152                 // Return the whole JSON file
 153                 return FormatJson::encode( $json, "\t", FormatJson::ALL_OK ) . "\n";
 154         }
 155
 156         /**
 157          * Generates a PHP file setting up autoload information.
 158          *
 159          * @param string $commandName Command name to include in comment
 160          * @param string $filename of PHP file to put autoload information in.
 161          * @return string
 162          */
 163         protected function generatePHPAutoload( $commandName, $filename ) {
 164                 // No existing JSON file found; update/generate PHP file
 165                 $content = [];
 166
 167                 // We need to generate a line each rather than exporting the
 168                 // full array so __DIR__ can be prepended to all the paths
 169                 $format = "%s => __DIR__ . %s,";
 170                 foreach ( $this->classes as $path => $contained ) {
 171                         $exportedPath = var_export( $path, true );
 172                         foreach ( $contained as $fqcn ) {
 173                                 $content[$fqcn] = sprintf(
 174                                         $format,
 175                                         var_export( $fqcn, true ),
 176                                         $exportedPath
 177                                 );
 178                         }
 179                 }
 180
 181                 foreach ( $this->overrides as $fqcn => $path ) {
 182                         $content[$fqcn] = sprintf(
 183                                 $format,
 184                                 var_export( $fqcn, true ),
 185                                 var_export( $path, true )
 186                         );
 187                 }
 188
 189                 // sort for stable output
 190                 ksort( $content );
 191
 192                 // extensions using this generator are appending to the existing
 193                 // autoload.
 194                 if ( $this->variableName === 'wgAutoloadClasses' ) {
 195                         $op = '+=';
 196                 } else {
 197                         $op = '=';
 198                 }
 199
 200                 $output = implode( "\n\t", $content );
 201                 return
 202                         <<<EOD
 203 <?php
 204 // This file is generated by $commandName, do not adjust manually
 205 // @codingStandardsIgnoreFile
 206 global \${$this->variableName};
 207
 208 \${$this->variableName} {$op} [
 209         {$output}
 210 ];
 211
 212 EOD;
 213         }
 214
 215         /**
 216          * Returns all known classes as a string, which can be used to put into a target
 217          * file (e.g. extension.json, skin.json or autoload.php)
 218          *
 219          * @param string $commandName Value used in file comment to direct
 220          *  developers towards the appropriate way to update the autoload.
 221          * @return string
 222          */
 223         public function getAutoload( $commandName = 'AutoloadGenerator' ) {
 224                 // We need to check whether an extenson.json or skin.json exists or not, and
 225                 // incase it doesn't, update the autoload.php file.
 226
 227                 $fileinfo = $this->getTargetFileinfo();
 228
 229                 if ( $fileinfo['type'] === self::FILETYPE_JSON ) {
 230                         return $this->generateJsonAutoload( $fileinfo['filename'] );
 231                 } else {
 232                         return $this->generatePHPAutoload( $commandName, $fileinfo['filename'] );
 233                 }
 234         }
 235
 236         /**
 237          * Returns the filename of the extension.json of skin.json, if there's any, or
 238          * otherwise the path to the autoload.php file in an array as the "filename"
 239          * key and with the type (AutoloadGenerator::FILETYPE_JSON or AutoloadGenerator::FILETYPE_PHP)
 240          * of the file as the "type" key.
 241          *
 242          * @return array
 243          */
 244         public function getTargetFileinfo() {
 245                 $fileinfo = [
 246                         'filename' => $this->basepath . '/autoload.php',
 247                         'type' => self::FILETYPE_PHP
 248                 ];
 249                 if ( file_exists( $this->basepath . '/extension.json' ) ) {
 250                         $fileinfo = [
 251                                 'filename' => $this->basepath . '/extension.json',
 252                                 'type' => self::FILETYPE_JSON
 253                         ];
 254                 } elseif ( file_exists( $this->basepath . '/skin.json' ) ) {
 255                         $fileinfo = [
 256                                 'filename' => $this->basepath . '/skin.json',
 257                                 'type' => self::FILETYPE_JSON
 258                         ];
 259                 }
 260
 261                 return $fileinfo;
 262         }
 263
 264         /**
 265          * Ensure that Unix-style path separators ("/") are used in the path.
 266          *
 267          * @param string $path
 268          * @return string
 269          */
 270         protected static function normalizePathSeparator( $path ) {
 271                 return str_replace( '\\', '/', $path );
 272         }
 273
 274         /**
 275          * Initialize the source files and directories which are used for the MediaWiki default
 276          * autoloader in {mw-base-dir}/autoload.php including:
 277          *  * includes/
 278          *  * languages/
 279          *  * maintenance/
 280          *  * mw-config/
 281          *  * /*.php
 282          */
 283         public function initMediaWikiDefault() {
 284                 foreach ( [ 'includes', 'languages', 'maintenance', 'mw-config' ] as $dir ) {
 285                         $this->readDir( $this->basepath . '/' . $dir );
 286                 }
 287                 foreach ( glob( $this->basepath . '/*.php' ) as $file ) {
 288                         $this->readFile( $file );
 289                 }
 290         }
 291 }
 292
 293 /**
 294  * Reads PHP code and returns the FQCN of every class defined within it.
 295  */
 296 class ClassCollector {
 297
 298         /**
 299          * @var string Current namespace
 300          */
 301         protected $namespace = '';
 302
 303         /**
 304          * @var array List of FQCN detected in this pass
 305          */
 306         protected $classes;
 307
 308         /**
 309          * @var array Token from token_get_all() that started an expect sequence
 310          */
 311         protected $startToken;
 312
 313         /**
 314          * @var array List of tokens that are members of the current expect sequence
 315          */
 316         protected $tokens;
 317
 318         /**
 319          * @var array Class alias with target/name fields
 320          */
 321         protected $alias;
 322
 323         /**
 324          * @param string $code PHP code (including <?php) to detect class names from
 325          * @return array List of FQCN detected within the tokens
 326          */
 327         public function getClasses( $code ) {
 328                 $this->namespace = '';
 329                 $this->classes = [];
 330                 $this->startToken = null;
 331                 $this->alias = null;
 332                 $this->tokens = [];
 333
 334                 foreach ( token_get_all( $code ) as $token ) {
 335                         if ( $this->startToken === null ) {
 336                                 $this->tryBeginExpect( $token );
 337                         } else {
 338                                 $this->tryEndExpect( $token );
 339                         }
 340                 }
 341
 342                 return $this->classes;
 343         }
 344
 345         /**
 346          * Determine if $token begins the next expect sequence.
 347          *
 348          * @param array $token
 349          */
 350         protected function tryBeginExpect( $token ) {
 351                 if ( is_string( $token ) ) {
 352                         return;
 353                 }
 354                 // Note: When changing class name discovery logic,
 355                 // AutoLoaderTest.php may also need to be updated.
 356                 switch ( $token[0] ) {
 357                 case T_NAMESPACE:
 358                 case T_CLASS:
 359                 case T_INTERFACE:
 360                 case T_TRAIT:
 361                 case T_DOUBLE_COLON:
 362                         $this->startToken = $token;
 363                         break;
 364                 case T_STRING:
 365                         if ( $token[1] === 'class_alias' ) {
 366                                 $this->startToken = $token;
 367                                 $this->alias = [];
 368                         }
 369                 }
 370         }
 371
 372         /**
 373          * Accepts the next token in an expect sequence
 374          *
 375          * @param array $token
 376          */
 377         protected function tryEndExpect( $token ) {
 378                 switch ( $this->startToken[0] ) {
 379                 case T_DOUBLE_COLON:
 380                         // Skip over T_CLASS after T_DOUBLE_COLON because this is something like
 381                         // "self::static" which accesses the class name. It doens't define a new class.
 382                         $this->startToken = null;
 383                         break;
 384                 case T_NAMESPACE:
 385                         if ( $token === ';' || $token === '{' ) {
 386                                 $this->namespace = $this->implodeTokens() . '\\';
 387                         } else {
 388                                 $this->tokens[] = $token;
 389                         }
 390                         break;
 391
 392                 case T_STRING:
 393                         if ( $this->alias !== null ) {
 394                                 // Flow 1 - Two string literals:
 395                                 // - T_STRING  class_alias
 396                                 // - '('
 397                                 // - T_CONSTANT_ENCAPSED_STRING 'TargetClass'
 398                                 // - ','
 399                                 // - T_WHITESPACE
 400                                 // - T_CONSTANT_ENCAPSED_STRING 'AliasName'
 401                                 // - ')'
 402                                 // Flow 2 - Use of ::class syntax for first parameter
 403                                 // - T_STRING  class_alias
 404                                 // - '('
 405                                 // - T_STRING TargetClass
 406                                 // - T_DOUBLE_COLON ::
 407                                 // - T_CLASS class
 408                                 // - ','
 409                                 // - T_WHITESPACE
 410                                 // - T_CONSTANT_ENCAPSED_STRING 'AliasName'
 411                                 // - ')'
 412                                 if ( $token === '(' ) {
 413                                         // Start of a function call to class_alias()
 414                                         $this->alias = [ 'target' => false, 'name' => false ];
 415                                 } elseif ( $token === ',' ) {
 416                                         // Record that we're past the first parameter
 417                                         if ( $this->alias['target'] === false ) {
 418                                                 $this->alias['target'] = true;
 419                                         }
 420                                 } elseif ( is_array( $token ) && $token[0] === T_CONSTANT_ENCAPSED_STRING ) {
 421                                         if ( $this->alias['target'] === true ) {
 422                                                 // We already saw a first argument, this must be the second.
 423                                                 // Strip quotes from the string literal.
 424                                                 $this->alias['name'] = substr( $token[1], 1, -1 );
 425                                         }
 426                                 } elseif ( $token === ')' ) {
 427                                         // End of function call
 428                                         $this->classes[] = $this->alias['name'];
 429                                         $this->alias = null;
 430                                         $this->startToken = null;
 431                                 } elseif ( !is_array( $token ) || (
 432                                         $token[0] !== T_STRING &&
 433                                         $token[0] !== T_DOUBLE_COLON &&
 434                                         $token[0] !== T_CLASS &&
 435                                         $token[0] !== T_WHITESPACE
 436                                 ) ) {
 437                                         // Ignore this call to class_alias() - compat/Timestamp.php
 438                                         $this->alias = null;
 439                                         $this->startToken = null;
 440                                 }
 441                         }
 442                         break;
 443
 444                 case T_CLASS:
 445                 case T_INTERFACE:
 446                 case T_TRAIT:
 447                         $this->tokens[] = $token;
 448                         if ( is_array( $token ) && $token[0] === T_STRING ) {
 449                                 $this->classes[] = $this->namespace . $this->implodeTokens();
 450                         }
 451                 }
 452         }
 453
 454         /**
 455          * Returns the string representation of the tokens within the
 456          * current expect sequence and resets the sequence.
 457          *
 458          * @return string
 459          */
 460         protected function implodeTokens() {
 461                 $content = [];
 462                 foreach ( $this->tokens as $token ) {
 463                         $content[] = is_string( $token ) ? $token : $token[1];
 464                 }
 465
 466                 $this->tokens = [];
 467                 $this->startToken = null;
 468
 469                 return trim( implode( '', $content ), " \n\t" );
 470         }
 471 }