filename = $filePath; } /** * Read text contents from Word files (DOCX) * * @return string|bool */ private function readDocx() { $content = ''; $zip = zip_open($this->filename); if (!$zip || is_numeric($zip)) { return false; } while ($zipEntry = zip_read($zip)) { if (zip_entry_open($zip, $zipEntry) === false) { continue; } if (zip_entry_name($zipEntry) !== 'word/document.xml') { continue; } $content .= zip_entry_read($zipEntry, zip_entry_filesize($zipEntry)); zip_entry_close($zipEntry); } zip_close($zip); $content = str_replace('', ' ', $content); $content = str_replace('', "\r\n", $content); return strip_tags($content); } /** * Read text contents from Excel file (XLSX) * @return string */ private function readXlsx() { $zipHandle = new \ZipArchive(); $outputText = ''; if ($zipHandle->open($this->filename) === true) { if (($xmlIndex = $zipHandle->locateName('xl/sharedStrings.xml')) !== false) { $xmlData = $zipHandle->getFromIndex($xmlIndex); $domDocument = new \DOMDocument(); $domDocument->loadXML($xmlData, LIBXML_NOENT | LIBXML_XINCLUDE | LIBXML_NOERROR | LIBXML_NOWARNING); $outputText = strip_tags(str_replace('', ' ', $domDocument->saveXML())); } else { $outputText .= ''; } $zipHandle->close(); } else { $outputText .= ''; } return $outputText; } /** * Read text contents from Powerpoint file (PPTX) * * @return string */ private function readPptx() { $zipHandle = new \ZipArchive(); $outputText = ''; if ($zipHandle->open($this->filename) === true) { $slideNumber = 1; //loop through slide files while (($xml_index = $zipHandle->locateName('ppt/slides/slide' . $slideNumber . '.xml')) !== false) { $xmlData = $zipHandle->getFromIndex($xml_index); $domDocument = new \DOMDocument(); $domDocument->loadXML($xmlData, LIBXML_NOENT | LIBXML_XINCLUDE | LIBXML_NOERROR | LIBXML_NOWARNING); $outputText .= strip_tags(str_replace('', ' ', $domDocument->saveXML())); $slideNumber++; } if ($slideNumber === 1) { $outputText .= ' '; } $zipHandle->close(); } else { $outputText .= ' '; } return $outputText; } /** * Extract text from given OOXML file * * @return string * @throws \InvalidArgumentException */ public function convertToText() { $pathInfo = pathinfo($this->filename); switch (strtolower($pathInfo['extension'])) { case 'docx': return $this->readDocx(); case 'xlsx': return $this->readXlsx(); case 'pptx': return $this->readPptx(); default: throw new \InvalidArgumentException('File extension "' . $pathInfo['extension'] . '" not supported!'); } } }