Chris@0: <?php
Chris@0: 
Chris@0: /*
Chris@0:  * This file is part of the Symfony package.
Chris@0:  *
Chris@0:  * (c) Fabien Potencier <fabien@symfony.com>
Chris@0:  *
Chris@0:  * For the full copyright and license information, please view the LICENSE
Chris@0:  * file that was distributed with this source code.
Chris@0:  */
Chris@0: 
Chris@0: namespace Symfony\Component\Translation\Loader;
Chris@0: 
Chris@17: use Symfony\Component\Config\Resource\FileResource;
Chris@0: use Symfony\Component\Config\Util\XmlUtils;
Chris@17: use Symfony\Component\Translation\Exception\InvalidArgumentException;
Chris@0: use Symfony\Component\Translation\Exception\InvalidResourceException;
Chris@0: use Symfony\Component\Translation\Exception\NotFoundResourceException;
Chris@17: use Symfony\Component\Translation\MessageCatalogue;
Chris@0: 
Chris@0: /**
Chris@0:  * XliffFileLoader loads translations from XLIFF files.
Chris@0:  *
Chris@0:  * @author Fabien Potencier <fabien@symfony.com>
Chris@0:  */
Chris@0: class XliffFileLoader implements LoaderInterface
Chris@0: {
Chris@0:     /**
Chris@0:      * {@inheritdoc}
Chris@0:      */
Chris@0:     public function load($resource, $locale, $domain = 'messages')
Chris@0:     {
Chris@0:         if (!stream_is_local($resource)) {
Chris@0:             throw new InvalidResourceException(sprintf('This is not a local file "%s".', $resource));
Chris@0:         }
Chris@0: 
Chris@0:         if (!file_exists($resource)) {
Chris@0:             throw new NotFoundResourceException(sprintf('File "%s" not found.', $resource));
Chris@0:         }
Chris@0: 
Chris@0:         $catalogue = new MessageCatalogue($locale);
Chris@0:         $this->extract($resource, $catalogue, $domain);
Chris@0: 
Chris@0:         if (class_exists('Symfony\Component\Config\Resource\FileResource')) {
Chris@0:             $catalogue->addResource(new FileResource($resource));
Chris@0:         }
Chris@0: 
Chris@0:         return $catalogue;
Chris@0:     }
Chris@0: 
Chris@0:     private function extract($resource, MessageCatalogue $catalogue, $domain)
Chris@0:     {
Chris@0:         try {
Chris@0:             $dom = XmlUtils::loadFile($resource);
Chris@0:         } catch (\InvalidArgumentException $e) {
Chris@0:             throw new InvalidResourceException(sprintf('Unable to load "%s": %s', $resource, $e->getMessage()), $e->getCode(), $e);
Chris@0:         }
Chris@0: 
Chris@0:         $xliffVersion = $this->getVersionNumber($dom);
Chris@0:         $this->validateSchema($xliffVersion, $dom, $this->getSchema($xliffVersion));
Chris@0: 
Chris@0:         if ('1.2' === $xliffVersion) {
Chris@0:             $this->extractXliff1($dom, $catalogue, $domain);
Chris@0:         }
Chris@0: 
Chris@0:         if ('2.0' === $xliffVersion) {
Chris@0:             $this->extractXliff2($dom, $catalogue, $domain);
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Extract messages and metadata from DOMDocument into a MessageCatalogue.
Chris@0:      *
Chris@0:      * @param \DOMDocument     $dom       Source to extract messages and metadata
Chris@0:      * @param MessageCatalogue $catalogue Catalogue where we'll collect messages and metadata
Chris@0:      * @param string           $domain    The domain
Chris@0:      */
Chris@0:     private function extractXliff1(\DOMDocument $dom, MessageCatalogue $catalogue, $domain)
Chris@0:     {
Chris@0:         $xml = simplexml_import_dom($dom);
Chris@0:         $encoding = strtoupper($dom->encoding);
Chris@0: 
Chris@0:         $xml->registerXPathNamespace('xliff', 'urn:oasis:names:tc:xliff:document:1.2');
Chris@0:         foreach ($xml->xpath('//xliff:trans-unit') as $translation) {
Chris@0:             $attributes = $translation->attributes();
Chris@0: 
Chris@0:             if (!(isset($attributes['resname']) || isset($translation->source))) {
Chris@0:                 continue;
Chris@0:             }
Chris@0: 
Chris@0:             $source = isset($attributes['resname']) && $attributes['resname'] ? $attributes['resname'] : $translation->source;
Chris@0:             // If the xlf file has another encoding specified, try to convert it because
Chris@0:             // simple_xml will always return utf-8 encoded values
Chris@17:             $target = $this->utf8ToCharset((string) (isset($translation->target) ? $translation->target : $translation->source), $encoding);
Chris@0: 
Chris@0:             $catalogue->set((string) $source, $target, $domain);
Chris@0: 
Chris@17:             $metadata = [];
Chris@0:             if ($notes = $this->parseNotesMetadata($translation->note, $encoding)) {
Chris@0:                 $metadata['notes'] = $notes;
Chris@0:             }
Chris@0: 
Chris@0:             if (isset($translation->target) && $translation->target->attributes()) {
Chris@17:                 $metadata['target-attributes'] = [];
Chris@0:                 foreach ($translation->target->attributes() as $key => $value) {
Chris@0:                     $metadata['target-attributes'][$key] = (string) $value;
Chris@0:                 }
Chris@0:             }
Chris@0: 
Chris@0:             if (isset($attributes['id'])) {
Chris@0:                 $metadata['id'] = (string) $attributes['id'];
Chris@0:             }
Chris@0: 
Chris@0:             $catalogue->setMetadata((string) $source, $metadata, $domain);
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * @param \DOMDocument     $dom
Chris@0:      * @param MessageCatalogue $catalogue
Chris@0:      * @param string           $domain
Chris@0:      */
Chris@0:     private function extractXliff2(\DOMDocument $dom, MessageCatalogue $catalogue, $domain)
Chris@0:     {
Chris@0:         $xml = simplexml_import_dom($dom);
Chris@0:         $encoding = strtoupper($dom->encoding);
Chris@0: 
Chris@0:         $xml->registerXPathNamespace('xliff', 'urn:oasis:names:tc:xliff:document:2.0');
Chris@0: 
Chris@14:         foreach ($xml->xpath('//xliff:unit') as $unit) {
Chris@14:             foreach ($unit->segment as $segment) {
Chris@14:                 $source = $segment->source;
Chris@0: 
Chris@14:                 // If the xlf file has another encoding specified, try to convert it because
Chris@14:                 // simple_xml will always return utf-8 encoded values
Chris@14:                 $target = $this->utf8ToCharset((string) (isset($segment->target) ? $segment->target : $source), $encoding);
Chris@0: 
Chris@14:                 $catalogue->set((string) $source, $target, $domain);
Chris@0: 
Chris@17:                 $metadata = [];
Chris@14:                 if (isset($segment->target) && $segment->target->attributes()) {
Chris@17:                     $metadata['target-attributes'] = [];
Chris@14:                     foreach ($segment->target->attributes() as $key => $value) {
Chris@14:                         $metadata['target-attributes'][$key] = (string) $value;
Chris@14:                     }
Chris@0:                 }
Chris@14: 
Chris@14:                 if (isset($unit->notes)) {
Chris@17:                     $metadata['notes'] = [];
Chris@14:                     foreach ($unit->notes->note as $noteNode) {
Chris@17:                         $note = [];
Chris@14:                         foreach ($noteNode->attributes() as $key => $value) {
Chris@14:                             $note[$key] = (string) $value;
Chris@14:                         }
Chris@14:                         $note['content'] = (string) $noteNode;
Chris@14:                         $metadata['notes'][] = $note;
Chris@14:                     }
Chris@14:                 }
Chris@14: 
Chris@14:                 $catalogue->setMetadata((string) $source, $metadata, $domain);
Chris@0:             }
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Convert a UTF8 string to the specified encoding.
Chris@0:      *
Chris@0:      * @param string $content  String to decode
Chris@0:      * @param string $encoding Target encoding
Chris@0:      *
Chris@0:      * @return string
Chris@0:      */
Chris@0:     private function utf8ToCharset($content, $encoding = null)
Chris@0:     {
Chris@0:         if ('UTF-8' !== $encoding && !empty($encoding)) {
Chris@0:             return mb_convert_encoding($content, $encoding, 'UTF-8');
Chris@0:         }
Chris@0: 
Chris@0:         return $content;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Validates and parses the given file into a DOMDocument.
Chris@0:      *
Chris@0:      * @param string       $file
Chris@0:      * @param \DOMDocument $dom
Chris@0:      * @param string       $schema source of the schema
Chris@0:      *
Chris@0:      * @throws InvalidResourceException
Chris@0:      */
Chris@0:     private function validateSchema($file, \DOMDocument $dom, $schema)
Chris@0:     {
Chris@0:         $internalErrors = libxml_use_internal_errors(true);
Chris@0: 
Chris@0:         $disableEntities = libxml_disable_entity_loader(false);
Chris@0: 
Chris@0:         if (!@$dom->schemaValidateSource($schema)) {
Chris@0:             libxml_disable_entity_loader($disableEntities);
Chris@0: 
Chris@0:             throw new InvalidResourceException(sprintf('Invalid resource provided: "%s"; Errors: %s', $file, implode("\n", $this->getXmlErrors($internalErrors))));
Chris@0:         }
Chris@0: 
Chris@0:         libxml_disable_entity_loader($disableEntities);
Chris@0: 
Chris@0:         $dom->normalizeDocument();
Chris@0: 
Chris@0:         libxml_clear_errors();
Chris@0:         libxml_use_internal_errors($internalErrors);
Chris@0:     }
Chris@0: 
Chris@0:     private function getSchema($xliffVersion)
Chris@0:     {
Chris@0:         if ('1.2' === $xliffVersion) {
Chris@0:             $schemaSource = file_get_contents(__DIR__.'/schema/dic/xliff-core/xliff-core-1.2-strict.xsd');
Chris@0:             $xmlUri = 'http://www.w3.org/2001/xml.xsd';
Chris@0:         } elseif ('2.0' === $xliffVersion) {
Chris@0:             $schemaSource = file_get_contents(__DIR__.'/schema/dic/xliff-core/xliff-core-2.0.xsd');
Chris@0:             $xmlUri = 'informativeCopiesOf3rdPartySchemas/w3c/xml.xsd';
Chris@0:         } else {
Chris@0:             throw new InvalidArgumentException(sprintf('No support implemented for loading XLIFF version "%s".', $xliffVersion));
Chris@0:         }
Chris@0: 
Chris@0:         return $this->fixXmlLocation($schemaSource, $xmlUri);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Internally changes the URI of a dependent xsd to be loaded locally.
Chris@0:      *
Chris@0:      * @param string $schemaSource Current content of schema file
Chris@0:      * @param string $xmlUri       External URI of XML to convert to local
Chris@0:      *
Chris@0:      * @return string
Chris@0:      */
Chris@0:     private function fixXmlLocation($schemaSource, $xmlUri)
Chris@0:     {
Chris@0:         $newPath = str_replace('\\', '/', __DIR__).'/schema/dic/xliff-core/xml.xsd';
Chris@0:         $parts = explode('/', $newPath);
Chris@14:         $locationstart = 'file:///';
Chris@0:         if (0 === stripos($newPath, 'phar://')) {
Chris@14:             $tmpfile = tempnam(sys_get_temp_dir(), 'symfony');
Chris@0:             if ($tmpfile) {
Chris@0:                 copy($newPath, $tmpfile);
Chris@0:                 $parts = explode('/', str_replace('\\', '/', $tmpfile));
Chris@14:             } else {
Chris@14:                 array_shift($parts);
Chris@14:                 $locationstart = 'phar:///';
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@17:         $drive = '\\' === \DIRECTORY_SEPARATOR ? array_shift($parts).'/' : '';
Chris@14:         $newPath = $locationstart.$drive.implode('/', array_map('rawurlencode', $parts));
Chris@0: 
Chris@0:         return str_replace($xmlUri, $newPath, $schemaSource);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns the XML errors of the internal XML parser.
Chris@0:      *
Chris@0:      * @param bool $internalErrors
Chris@0:      *
Chris@0:      * @return array An array of errors
Chris@0:      */
Chris@0:     private function getXmlErrors($internalErrors)
Chris@0:     {
Chris@17:         $errors = [];
Chris@0:         foreach (libxml_get_errors() as $error) {
Chris@0:             $errors[] = sprintf('[%s %s] %s (in %s - line %d, column %d)',
Chris@0:                 LIBXML_ERR_WARNING == $error->level ? 'WARNING' : 'ERROR',
Chris@0:                 $error->code,
Chris@0:                 trim($error->message),
Chris@0:                 $error->file ?: 'n/a',
Chris@0:                 $error->line,
Chris@0:                 $error->column
Chris@0:             );
Chris@0:         }
Chris@0: 
Chris@0:         libxml_clear_errors();
Chris@0:         libxml_use_internal_errors($internalErrors);
Chris@0: 
Chris@0:         return $errors;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Gets xliff file version based on the root "version" attribute.
Chris@0:      * Defaults to 1.2 for backwards compatibility.
Chris@0:      *
Chris@0:      * @param \DOMDocument $dom
Chris@0:      *
Chris@0:      * @throws InvalidArgumentException
Chris@0:      *
Chris@0:      * @return string
Chris@0:      */
Chris@0:     private function getVersionNumber(\DOMDocument $dom)
Chris@0:     {
Chris@0:         /** @var \DOMNode $xliff */
Chris@0:         foreach ($dom->getElementsByTagName('xliff') as $xliff) {
Chris@0:             $version = $xliff->attributes->getNamedItem('version');
Chris@0:             if ($version) {
Chris@0:                 return $version->nodeValue;
Chris@0:             }
Chris@0: 
Chris@0:             $namespace = $xliff->attributes->getNamedItem('xmlns');
Chris@0:             if ($namespace) {
Chris@14:                 if (0 !== substr_compare('urn:oasis:names:tc:xliff:document:', $namespace->nodeValue, 0, 34)) {
Chris@0:                     throw new InvalidArgumentException(sprintf('Not a valid XLIFF namespace "%s"', $namespace));
Chris@0:                 }
Chris@0: 
Chris@0:                 return substr($namespace, 34);
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         // Falls back to v1.2
Chris@0:         return '1.2';
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * @param \SimpleXMLElement|null $noteElement
Chris@0:      * @param string|null            $encoding
Chris@0:      *
Chris@0:      * @return array
Chris@0:      */
Chris@0:     private function parseNotesMetadata(\SimpleXMLElement $noteElement = null, $encoding = null)
Chris@0:     {
Chris@17:         $notes = [];
Chris@0: 
Chris@0:         if (null === $noteElement) {
Chris@0:             return $notes;
Chris@0:         }
Chris@0: 
Chris@0:         /** @var \SimpleXMLElement $xmlNote */
Chris@0:         foreach ($noteElement as $xmlNote) {
Chris@0:             $noteAttributes = $xmlNote->attributes();
Chris@17:             $note = ['content' => $this->utf8ToCharset((string) $xmlNote, $encoding)];
Chris@0:             if (isset($noteAttributes['priority'])) {
Chris@0:                 $note['priority'] = (int) $noteAttributes['priority'];
Chris@0:             }
Chris@0: 
Chris@0:             if (isset($noteAttributes['from'])) {
Chris@0:                 $note['from'] = (string) $noteAttributes['from'];
Chris@0:             }
Chris@0: 
Chris@0:             $notes[] = $note;
Chris@0:         }
Chris@0: 
Chris@0:         return $notes;
Chris@0:     }
Chris@0: }