Chris@14: <?php
Chris@14: 
Chris@14: /*
Chris@14:  * This file is part of the Symfony package.
Chris@14:  *
Chris@14:  * (c) Fabien Potencier <fabien@symfony.com>
Chris@14:  *
Chris@14:  * For the full copyright and license information, please view the LICENSE
Chris@14:  * file that was distributed with this source code.
Chris@14:  */
Chris@14: 
Chris@14: namespace Symfony\Component\Translation\Extractor;
Chris@14: 
Chris@14: use Symfony\Component\Finder\Finder;
Chris@14: use Symfony\Component\Translation\MessageCatalogue;
Chris@14: 
Chris@14: /**
Chris@14:  * PhpExtractor extracts translation messages from a PHP template.
Chris@14:  *
Chris@14:  * @author Michel Salib <michelsalib@hotmail.com>
Chris@14:  */
Chris@14: class PhpExtractor extends AbstractFileExtractor implements ExtractorInterface
Chris@14: {
Chris@14:     const MESSAGE_TOKEN = 300;
Chris@14:     const METHOD_ARGUMENTS_TOKEN = 1000;
Chris@14:     const DOMAIN_TOKEN = 1001;
Chris@14: 
Chris@14:     /**
Chris@14:      * Prefix for new found message.
Chris@14:      *
Chris@14:      * @var string
Chris@14:      */
Chris@14:     private $prefix = '';
Chris@14: 
Chris@14:     /**
Chris@14:      * The sequence that captures translation messages.
Chris@14:      *
Chris@14:      * @var array
Chris@14:      */
Chris@17:     protected $sequences = [
Chris@17:         [
Chris@14:             '->',
Chris@14:             'trans',
Chris@14:             '(',
Chris@14:             self::MESSAGE_TOKEN,
Chris@14:             ',',
Chris@14:             self::METHOD_ARGUMENTS_TOKEN,
Chris@14:             ',',
Chris@14:             self::DOMAIN_TOKEN,
Chris@17:         ],
Chris@17:         [
Chris@14:             '->',
Chris@14:             'transChoice',
Chris@14:             '(',
Chris@14:             self::MESSAGE_TOKEN,
Chris@14:             ',',
Chris@14:             self::METHOD_ARGUMENTS_TOKEN,
Chris@14:             ',',
Chris@14:             self::METHOD_ARGUMENTS_TOKEN,
Chris@14:             ',',
Chris@14:             self::DOMAIN_TOKEN,
Chris@17:         ],
Chris@17:         [
Chris@14:             '->',
Chris@14:             'trans',
Chris@14:             '(',
Chris@14:             self::MESSAGE_TOKEN,
Chris@17:         ],
Chris@17:         [
Chris@14:             '->',
Chris@14:             'transChoice',
Chris@14:             '(',
Chris@14:             self::MESSAGE_TOKEN,
Chris@17:         ],
Chris@17:     ];
Chris@14: 
Chris@14:     /**
Chris@14:      * {@inheritdoc}
Chris@14:      */
Chris@14:     public function extract($resource, MessageCatalogue $catalog)
Chris@14:     {
Chris@14:         $files = $this->extractFiles($resource);
Chris@14:         foreach ($files as $file) {
Chris@14:             $this->parseTokens(token_get_all(file_get_contents($file)), $catalog);
Chris@14: 
Chris@14:             if (\PHP_VERSION_ID >= 70000) {
Chris@14:                 // PHP 7 memory manager will not release after token_get_all(), see https://bugs.php.net/70098
Chris@14:                 gc_mem_caches();
Chris@14:             }
Chris@14:         }
Chris@14:     }
Chris@14: 
Chris@14:     /**
Chris@14:      * {@inheritdoc}
Chris@14:      */
Chris@14:     public function setPrefix($prefix)
Chris@14:     {
Chris@14:         $this->prefix = $prefix;
Chris@14:     }
Chris@14: 
Chris@14:     /**
Chris@14:      * Normalizes a token.
Chris@14:      *
Chris@14:      * @param mixed $token
Chris@14:      *
Chris@14:      * @return string
Chris@14:      */
Chris@14:     protected function normalizeToken($token)
Chris@14:     {
Chris@14:         if (isset($token[1]) && 'b"' !== $token) {
Chris@14:             return $token[1];
Chris@14:         }
Chris@14: 
Chris@14:         return $token;
Chris@14:     }
Chris@14: 
Chris@14:     /**
Chris@14:      * Seeks to a non-whitespace token.
Chris@14:      */
Chris@14:     private function seekToNextRelevantToken(\Iterator $tokenIterator)
Chris@14:     {
Chris@14:         for (; $tokenIterator->valid(); $tokenIterator->next()) {
Chris@14:             $t = $tokenIterator->current();
Chris@14:             if (T_WHITESPACE !== $t[0]) {
Chris@14:                 break;
Chris@14:             }
Chris@14:         }
Chris@14:     }
Chris@14: 
Chris@14:     private function skipMethodArgument(\Iterator $tokenIterator)
Chris@14:     {
Chris@14:         $openBraces = 0;
Chris@14: 
Chris@14:         for (; $tokenIterator->valid(); $tokenIterator->next()) {
Chris@14:             $t = $tokenIterator->current();
Chris@14: 
Chris@14:             if ('[' === $t[0] || '(' === $t[0]) {
Chris@14:                 ++$openBraces;
Chris@14:             }
Chris@14: 
Chris@14:             if (']' === $t[0] || ')' === $t[0]) {
Chris@14:                 --$openBraces;
Chris@14:             }
Chris@14: 
Chris@14:             if ((0 === $openBraces && ',' === $t[0]) || (-1 === $openBraces && ')' === $t[0])) {
Chris@14:                 break;
Chris@14:             }
Chris@14:         }
Chris@14:     }
Chris@14: 
Chris@14:     /**
Chris@14:      * Extracts the message from the iterator while the tokens
Chris@14:      * match allowed message tokens.
Chris@14:      */
Chris@14:     private function getValue(\Iterator $tokenIterator)
Chris@14:     {
Chris@14:         $message = '';
Chris@14:         $docToken = '';
Chris@17:         $docPart = '';
Chris@14: 
Chris@14:         for (; $tokenIterator->valid(); $tokenIterator->next()) {
Chris@14:             $t = $tokenIterator->current();
Chris@17:             if ('.' === $t) {
Chris@17:                 // Concatenate with next token
Chris@17:                 continue;
Chris@17:             }
Chris@14:             if (!isset($t[1])) {
Chris@14:                 break;
Chris@14:             }
Chris@14: 
Chris@14:             switch ($t[0]) {
Chris@14:                 case T_START_HEREDOC:
Chris@14:                     $docToken = $t[1];
Chris@14:                     break;
Chris@14:                 case T_ENCAPSED_AND_WHITESPACE:
Chris@14:                 case T_CONSTANT_ENCAPSED_STRING:
Chris@17:                     if ('' === $docToken) {
Chris@17:                         $message .= PhpStringTokenParser::parse($t[1]);
Chris@17:                     } else {
Chris@17:                         $docPart = $t[1];
Chris@17:                     }
Chris@14:                     break;
Chris@14:                 case T_END_HEREDOC:
Chris@17:                     $message .= PhpStringTokenParser::parseDocString($docToken, $docPart);
Chris@17:                     $docToken = '';
Chris@17:                     $docPart = '';
Chris@17:                     break;
Chris@17:                 case T_WHITESPACE:
Chris@17:                     break;
Chris@14:                 default:
Chris@14:                     break 2;
Chris@14:             }
Chris@14:         }
Chris@14: 
Chris@14:         return $message;
Chris@14:     }
Chris@14: 
Chris@14:     /**
Chris@14:      * Extracts trans message from PHP tokens.
Chris@14:      *
Chris@14:      * @param array            $tokens
Chris@14:      * @param MessageCatalogue $catalog
Chris@14:      */
Chris@14:     protected function parseTokens($tokens, MessageCatalogue $catalog)
Chris@14:     {
Chris@14:         $tokenIterator = new \ArrayIterator($tokens);
Chris@14: 
Chris@14:         for ($key = 0; $key < $tokenIterator->count(); ++$key) {
Chris@14:             foreach ($this->sequences as $sequence) {
Chris@14:                 $message = '';
Chris@14:                 $domain = 'messages';
Chris@14:                 $tokenIterator->seek($key);
Chris@14: 
Chris@14:                 foreach ($sequence as $sequenceKey => $item) {
Chris@14:                     $this->seekToNextRelevantToken($tokenIterator);
Chris@14: 
Chris@14:                     if ($this->normalizeToken($tokenIterator->current()) === $item) {
Chris@14:                         $tokenIterator->next();
Chris@14:                         continue;
Chris@14:                     } elseif (self::MESSAGE_TOKEN === $item) {
Chris@14:                         $message = $this->getValue($tokenIterator);
Chris@14: 
Chris@17:                         if (\count($sequence) === ($sequenceKey + 1)) {
Chris@14:                             break;
Chris@14:                         }
Chris@14:                     } elseif (self::METHOD_ARGUMENTS_TOKEN === $item) {
Chris@14:                         $this->skipMethodArgument($tokenIterator);
Chris@14:                     } elseif (self::DOMAIN_TOKEN === $item) {
Chris@18:                         $domainToken = $this->getValue($tokenIterator);
Chris@18:                         if ('' !== $domainToken) {
Chris@18:                             $domain = $domainToken;
Chris@18:                         }
Chris@14: 
Chris@14:                         break;
Chris@14:                     } else {
Chris@14:                         break;
Chris@14:                     }
Chris@14:                 }
Chris@14: 
Chris@14:                 if ($message) {
Chris@14:                     $catalog->set($message, $this->prefix.$message, $domain);
Chris@14:                     break;
Chris@14:                 }
Chris@14:             }
Chris@14:         }
Chris@14:     }
Chris@14: 
Chris@14:     /**
Chris@14:      * @param string $file
Chris@14:      *
Chris@14:      * @return bool
Chris@14:      *
Chris@14:      * @throws \InvalidArgumentException
Chris@14:      */
Chris@14:     protected function canBeExtracted($file)
Chris@14:     {
Chris@14:         return $this->isFile($file) && 'php' === pathinfo($file, PATHINFO_EXTENSION);
Chris@14:     }
Chris@14: 
Chris@14:     /**
Chris@14:      * @param string|array $directory
Chris@14:      *
Chris@14:      * @return array
Chris@14:      */
Chris@14:     protected function extractFromDirectory($directory)
Chris@14:     {
Chris@14:         $finder = new Finder();
Chris@14: 
Chris@14:         return $finder->files()->name('*.php')->in($directory);
Chris@14:     }
Chris@14: }