Chris@13: <?php
Chris@13: 
Chris@13: /*
Chris@13:  * This file is part of Psy Shell.
Chris@13:  *
Chris@13:  * (c) 2012-2018 Justin Hileman
Chris@13:  *
Chris@13:  * For the full copyright and license information, please view the LICENSE
Chris@13:  * file that was distributed with this source code.
Chris@13:  */
Chris@13: 
Chris@13: namespace Psy;
Chris@13: 
Chris@13: use PhpParser\NodeTraverser;
Chris@13: use PhpParser\Parser;
Chris@13: use PhpParser\PrettyPrinter\Standard as Printer;
Chris@13: use Psy\CodeCleaner\AbstractClassPass;
Chris@13: use Psy\CodeCleaner\AssignThisVariablePass;
Chris@13: use Psy\CodeCleaner\CalledClassPass;
Chris@13: use Psy\CodeCleaner\CallTimePassByReferencePass;
Chris@13: use Psy\CodeCleaner\ExitPass;
Chris@13: use Psy\CodeCleaner\FinalClassPass;
Chris@13: use Psy\CodeCleaner\FunctionContextPass;
Chris@13: use Psy\CodeCleaner\FunctionReturnInWriteContextPass;
Chris@13: use Psy\CodeCleaner\ImplicitReturnPass;
Chris@13: use Psy\CodeCleaner\InstanceOfPass;
Chris@13: use Psy\CodeCleaner\LeavePsyshAlonePass;
Chris@13: use Psy\CodeCleaner\LegacyEmptyPass;
Chris@16: use Psy\CodeCleaner\ListPass;
Chris@13: use Psy\CodeCleaner\LoopContextPass;
Chris@13: use Psy\CodeCleaner\MagicConstantsPass;
Chris@13: use Psy\CodeCleaner\NamespacePass;
Chris@13: use Psy\CodeCleaner\PassableByReferencePass;
Chris@13: use Psy\CodeCleaner\RequirePass;
Chris@13: use Psy\CodeCleaner\StrictTypesPass;
Chris@13: use Psy\CodeCleaner\UseStatementPass;
Chris@13: use Psy\CodeCleaner\ValidClassNamePass;
Chris@13: use Psy\CodeCleaner\ValidConstantPass;
Chris@13: use Psy\CodeCleaner\ValidConstructorPass;
Chris@13: use Psy\CodeCleaner\ValidFunctionNamePass;
Chris@13: use Psy\Exception\ParseErrorException;
Chris@13: 
Chris@13: /**
Chris@13:  * A service to clean up user input, detect parse errors before they happen,
Chris@13:  * and generally work around issues with the PHP code evaluation experience.
Chris@13:  */
Chris@13: class CodeCleaner
Chris@13: {
Chris@13:     private $parser;
Chris@13:     private $printer;
Chris@13:     private $traverser;
Chris@13:     private $namespace;
Chris@13: 
Chris@13:     /**
Chris@13:      * CodeCleaner constructor.
Chris@13:      *
Chris@13:      * @param Parser        $parser    A PhpParser Parser instance. One will be created if not explicitly supplied
Chris@13:      * @param Printer       $printer   A PhpParser Printer instance. One will be created if not explicitly supplied
Chris@13:      * @param NodeTraverser $traverser A PhpParser NodeTraverser instance. One will be created if not explicitly supplied
Chris@13:      */
Chris@13:     public function __construct(Parser $parser = null, Printer $printer = null, NodeTraverser $traverser = null)
Chris@13:     {
Chris@13:         if ($parser === null) {
Chris@13:             $parserFactory = new ParserFactory();
Chris@13:             $parser        = $parserFactory->createParser();
Chris@13:         }
Chris@13: 
Chris@13:         $this->parser    = $parser;
Chris@13:         $this->printer   = $printer ?: new Printer();
Chris@13:         $this->traverser = $traverser ?: new NodeTraverser();
Chris@13: 
Chris@13:         foreach ($this->getDefaultPasses() as $pass) {
Chris@13:             $this->traverser->addVisitor($pass);
Chris@13:         }
Chris@13:     }
Chris@13: 
Chris@13:     /**
Chris@13:      * Get default CodeCleaner passes.
Chris@13:      *
Chris@13:      * @return array
Chris@13:      */
Chris@13:     private function getDefaultPasses()
Chris@13:     {
Chris@13:         $useStatementPass = new UseStatementPass();
Chris@13:         $namespacePass    = new NamespacePass($this);
Chris@13: 
Chris@13:         // Try to add implicit `use` statements and an implicit namespace,
Chris@13:         // based on the file in which the `debug` call was made.
Chris@13:         $this->addImplicitDebugContext([$useStatementPass, $namespacePass]);
Chris@13: 
Chris@13:         return [
Chris@13:             // Validation passes
Chris@13:             new AbstractClassPass(),
Chris@13:             new AssignThisVariablePass(),
Chris@13:             new CalledClassPass(),
Chris@13:             new CallTimePassByReferencePass(),
Chris@13:             new FinalClassPass(),
Chris@13:             new FunctionContextPass(),
Chris@13:             new FunctionReturnInWriteContextPass(),
Chris@13:             new InstanceOfPass(),
Chris@13:             new LeavePsyshAlonePass(),
Chris@13:             new LegacyEmptyPass(),
Chris@16:             new ListPass(),
Chris@13:             new LoopContextPass(),
Chris@13:             new PassableByReferencePass(),
Chris@13:             new ValidConstructorPass(),
Chris@13: 
Chris@13:             // Rewriting shenanigans
Chris@13:             $useStatementPass,        // must run before the namespace pass
Chris@13:             new ExitPass(),
Chris@13:             new ImplicitReturnPass(),
Chris@13:             new MagicConstantsPass(),
Chris@13:             $namespacePass,           // must run after the implicit return pass
Chris@13:             new RequirePass(),
Chris@13:             new StrictTypesPass(),
Chris@13: 
Chris@13:             // Namespace-aware validation (which depends on aforementioned shenanigans)
Chris@13:             new ValidClassNamePass(),
Chris@13:             new ValidConstantPass(),
Chris@13:             new ValidFunctionNamePass(),
Chris@13:         ];
Chris@13:     }
Chris@13: 
Chris@13:     /**
Chris@13:      * "Warm up" code cleaner passes when we're coming from a debug call.
Chris@13:      *
Chris@13:      * This is useful, for example, for `UseStatementPass` and `NamespacePass`
Chris@13:      * which keep track of state between calls, to maintain the current
Chris@13:      * namespace and a map of use statements.
Chris@13:      *
Chris@13:      * @param array $passes
Chris@13:      */
Chris@13:     private function addImplicitDebugContext(array $passes)
Chris@13:     {
Chris@13:         $file = $this->getDebugFile();
Chris@13:         if ($file === null) {
Chris@13:             return;
Chris@13:         }
Chris@13: 
Chris@13:         try {
Chris@17:             $code = @\file_get_contents($file);
Chris@13:             if (!$code) {
Chris@13:                 return;
Chris@13:             }
Chris@13: 
Chris@13:             $stmts = $this->parse($code, true);
Chris@13:             if ($stmts === false) {
Chris@13:                 return;
Chris@13:             }
Chris@13: 
Chris@13:             // Set up a clean traverser for just these code cleaner passes
Chris@13:             $traverser = new NodeTraverser();
Chris@13:             foreach ($passes as $pass) {
Chris@13:                 $traverser->addVisitor($pass);
Chris@13:             }
Chris@13: 
Chris@13:             $traverser->traverse($stmts);
Chris@13:         } catch (\Throwable $e) {
Chris@13:             // Don't care.
Chris@13:         } catch (\Exception $e) {
Chris@13:             // Still don't care.
Chris@13:         }
Chris@13:     }
Chris@13: 
Chris@13:     /**
Chris@13:      * Search the stack trace for a file in which the user called Psy\debug.
Chris@13:      *
Chris@13:      * @return string|null
Chris@13:      */
Chris@13:     private static function getDebugFile()
Chris@13:     {
Chris@17:         $trace = \debug_backtrace(DEBUG_BACKTRACE_IGNORE_ARGS);
Chris@13: 
Chris@17:         foreach (\array_reverse($trace) as $stackFrame) {
Chris@13:             if (!self::isDebugCall($stackFrame)) {
Chris@13:                 continue;
Chris@13:             }
Chris@13: 
Chris@17:             if (\preg_match('/eval\(/', $stackFrame['file'])) {
Chris@17:                 \preg_match_all('/([^\(]+)\((\d+)/', $stackFrame['file'], $matches);
Chris@13: 
Chris@13:                 return $matches[1][0];
Chris@13:             }
Chris@13: 
Chris@13:             return $stackFrame['file'];
Chris@13:         }
Chris@13:     }
Chris@13: 
Chris@13:     /**
Chris@13:      * Check whether a given backtrace frame is a call to Psy\debug.
Chris@13:      *
Chris@13:      * @param array $stackFrame
Chris@13:      *
Chris@13:      * @return bool
Chris@13:      */
Chris@13:     private static function isDebugCall(array $stackFrame)
Chris@13:     {
Chris@13:         $class    = isset($stackFrame['class']) ? $stackFrame['class'] : null;
Chris@13:         $function = isset($stackFrame['function']) ? $stackFrame['function'] : null;
Chris@13: 
Chris@13:         return ($class === null && $function === 'Psy\debug') ||
Chris@13:             ($class === 'Psy\Shell' && $function === 'debug');
Chris@13:     }
Chris@13: 
Chris@13:     /**
Chris@13:      * Clean the given array of code.
Chris@13:      *
Chris@13:      * @throws ParseErrorException if the code is invalid PHP, and cannot be coerced into valid PHP
Chris@13:      *
Chris@13:      * @param array $codeLines
Chris@13:      * @param bool  $requireSemicolons
Chris@13:      *
Chris@13:      * @return string|false Cleaned PHP code, False if the input is incomplete
Chris@13:      */
Chris@13:     public function clean(array $codeLines, $requireSemicolons = false)
Chris@13:     {
Chris@17:         $stmts = $this->parse('<?php ' . \implode(PHP_EOL, $codeLines) . PHP_EOL, $requireSemicolons);
Chris@13:         if ($stmts === false) {
Chris@13:             return false;
Chris@13:         }
Chris@13: 
Chris@13:         // Catch fatal errors before they happen
Chris@13:         $stmts = $this->traverser->traverse($stmts);
Chris@13: 
Chris@13:         // Work around https://github.com/nikic/PHP-Parser/issues/399
Chris@17:         $oldLocale = \setlocale(LC_NUMERIC, 0);
Chris@17:         \setlocale(LC_NUMERIC, 'C');
Chris@13: 
Chris@13:         $code = $this->printer->prettyPrint($stmts);
Chris@13: 
Chris@13:         // Now put the locale back
Chris@17:         \setlocale(LC_NUMERIC, $oldLocale);
Chris@13: 
Chris@13:         return $code;
Chris@13:     }
Chris@13: 
Chris@13:     /**
Chris@13:      * Set the current local namespace.
Chris@13:      *
Chris@13:      * @param null|array $namespace (default: null)
Chris@13:      *
Chris@13:      * @return null|array
Chris@13:      */
Chris@13:     public function setNamespace(array $namespace = null)
Chris@13:     {
Chris@13:         $this->namespace = $namespace;
Chris@13:     }
Chris@13: 
Chris@13:     /**
Chris@13:      * Get the current local namespace.
Chris@13:      *
Chris@13:      * @return null|array
Chris@13:      */
Chris@13:     public function getNamespace()
Chris@13:     {
Chris@13:         return $this->namespace;
Chris@13:     }
Chris@13: 
Chris@13:     /**
Chris@13:      * Lex and parse a block of code.
Chris@13:      *
Chris@13:      * @see Parser::parse
Chris@13:      *
Chris@13:      * @throws ParseErrorException for parse errors that can't be resolved by
Chris@13:      *                             waiting a line to see what comes next
Chris@13:      *
Chris@13:      * @param string $code
Chris@13:      * @param bool   $requireSemicolons
Chris@13:      *
Chris@13:      * @return array|false A set of statements, or false if incomplete
Chris@13:      */
Chris@13:     protected function parse($code, $requireSemicolons = false)
Chris@13:     {
Chris@13:         try {
Chris@13:             return $this->parser->parse($code);
Chris@13:         } catch (\PhpParser\Error $e) {
Chris@13:             if ($this->parseErrorIsUnclosedString($e, $code)) {
Chris@13:                 return false;
Chris@13:             }
Chris@13: 
Chris@13:             if ($this->parseErrorIsUnterminatedComment($e, $code)) {
Chris@13:                 return false;
Chris@13:             }
Chris@13: 
Chris@13:             if ($this->parseErrorIsTrailingComma($e, $code)) {
Chris@13:                 return false;
Chris@13:             }
Chris@13: 
Chris@13:             if (!$this->parseErrorIsEOF($e)) {
Chris@13:                 throw ParseErrorException::fromParseError($e);
Chris@13:             }
Chris@13: 
Chris@13:             if ($requireSemicolons) {
Chris@13:                 return false;
Chris@13:             }
Chris@13: 
Chris@13:             try {
Chris@13:                 // Unexpected EOF, try again with an implicit semicolon
Chris@13:                 return $this->parser->parse($code . ';');
Chris@13:             } catch (\PhpParser\Error $e) {
Chris@13:                 return false;
Chris@13:             }
Chris@13:         }
Chris@13:     }
Chris@13: 
Chris@13:     private function parseErrorIsEOF(\PhpParser\Error $e)
Chris@13:     {
Chris@13:         $msg = $e->getRawMessage();
Chris@13: 
Chris@17:         return ($msg === 'Unexpected token EOF') || (\strpos($msg, 'Syntax error, unexpected EOF') !== false);
Chris@13:     }
Chris@13: 
Chris@13:     /**
Chris@13:      * A special test for unclosed single-quoted strings.
Chris@13:      *
Chris@13:      * Unlike (all?) other unclosed statements, single quoted strings have
Chris@13:      * their own special beautiful snowflake syntax error just for
Chris@13:      * themselves.
Chris@13:      *
Chris@13:      * @param \PhpParser\Error $e
Chris@13:      * @param string           $code
Chris@13:      *
Chris@13:      * @return bool
Chris@13:      */
Chris@13:     private function parseErrorIsUnclosedString(\PhpParser\Error $e, $code)
Chris@13:     {
Chris@13:         if ($e->getRawMessage() !== 'Syntax error, unexpected T_ENCAPSED_AND_WHITESPACE') {
Chris@13:             return false;
Chris@13:         }
Chris@13: 
Chris@13:         try {
Chris@13:             $this->parser->parse($code . "';");
Chris@13:         } catch (\Exception $e) {
Chris@13:             return false;
Chris@13:         }
Chris@13: 
Chris@13:         return true;
Chris@13:     }
Chris@13: 
Chris@13:     private function parseErrorIsUnterminatedComment(\PhpParser\Error $e, $code)
Chris@13:     {
Chris@13:         return $e->getRawMessage() === 'Unterminated comment';
Chris@13:     }
Chris@13: 
Chris@13:     private function parseErrorIsTrailingComma(\PhpParser\Error $e, $code)
Chris@13:     {
Chris@17:         return ($e->getRawMessage() === 'A trailing comma is not allowed here') && (\substr(\rtrim($code), -1) === ',');
Chris@13:     }
Chris@13: }