Example #1
0
 public function testPptx()
 {
     if (!class_exists('ZipArchive')) {
         $this->markTestSkipped('ZipArchive class (Zip extension) is not loaded');
     }
     $pptxDocument = Document\Pptx::loadPptxFile(__DIR__ . '/_openXmlDocuments/test.pptx', true);
     $this->assertTrue($pptxDocument instanceof Document\Pptx);
     $this->assertEquals($pptxDocument->getFieldValue('title'), 'Test document');
     $this->assertEquals($pptxDocument->getFieldValue('description'), 'This is a test document which can be used to demonstrate something.');
     $this->assertTrue($pptxDocument->getFieldValue('body') != '');
 }
Example #2
0
 /**
  * index a file
  *
  * @param File $file the file to be indexed
  * @param bool $commit
  *
  * @return bool true when something was stored in the index, false otherwise (eg, folders are not indexed)
  * @throws NotIndexedException when an unsupported file type is encountered
  */
 public function indexFile(File $file, $commit = true)
 {
     // we decide how to index on mime type or file extension
     $mimeType = $file->getMimeType();
     $fileExtension = strtolower(pathinfo($file->getName(), PATHINFO_EXTENSION));
     // initialize plain lucene document
     $doc = new Document();
     // index content for local files only
     $storage = $file->getStorage();
     if ($storage->isLocal()) {
         $path = $storage->getLocalFile($file->getInternalPath());
         //try to use special lucene document types
         if ('text/html' === $mimeType) {
             //TODO could be indexed, even if not local
             $doc = HTML::loadHTML($file->getContent());
         } else {
             if ('text/' === substr($mimeType, 0, 5) || 'application/x-tex' === $mimeType) {
                 $body = $file->getContent();
                 if ($body != '') {
                     $doc->addField(Document\Field::UnStored('body', $body));
                 }
             } else {
                 if ('application/pdf' === $mimeType) {
                     $doc = Pdf::loadPdf($file->getContent());
                     // the zend classes only understand docx and not doc files
                 } else {
                     if ($fileExtension === 'docx') {
                         $doc = Document\Docx::loadDocxFile($path);
                         //} else if ('application/msexcel' === $mimeType) {
                     } else {
                         if ($fileExtension === 'xlsx') {
                             $doc = Document\Xlsx::loadXlsxFile($path);
                             //} else if ('application/mspowerpoint' === $mimeType) {
                         } else {
                             if ($fileExtension === 'pptx') {
                                 $doc = Document\Pptx::loadPptxFile($path);
                             } else {
                                 if ($fileExtension === 'odt') {
                                     $doc = Odt::loadOdtFile($path);
                                 } else {
                                     if ($fileExtension === 'ods') {
                                         $doc = Ods::loadOdsFile($path);
                                     } else {
                                         throw new NotIndexedException();
                                     }
                                 }
                             }
                         }
                     }
                 }
             }
         }
     }
     // Store filecache id as unique id to lookup by when deleting
     $doc->addField(Document\Field::Keyword('fileId', $file->getId()));
     // Store document path for the search results
     $doc->addField(Document\Field::Text('path', $file->getPath(), 'UTF-8'));
     $doc->addField(Document\Field::unIndexed('mtime', $file->getMTime()));
     $doc->addField(Document\Field::unIndexed('size', $file->getSize()));
     $doc->addField(Document\Field::unIndexed('mimetype', $mimeType));
     $this->index->updateFile($doc, $file->getId(), $commit);
     return true;
 }
 /**
  * Read Pptx files
  *
  * @param $attributeValue
  * @param $value
  * @return null|string
  */
 private function readPptx($attributeValue, $value)
 {
     $filePath = $this->getFilePath($attributeValue['basePath'], $value);
     if (!file_exists($filePath)) {
         return null;
     }
     return Pptx::loadPptxFile($filePath)->body;
 }