searchanddisplace-ingest/app/Ingest/DataJsonConvertor.php

<?php

namespace App\Ingest;

use Illuminate\Support\Facades\Storage;

class DataJsonConvertor extends AbstractConvertor
{
    protected $type;

    public function __construct($path, $type)
    {
        parent::__construct(Storage::disk('local'), $path);

        $this->type = $type;
    }

    /**
     * Convert given document to JSON file which contains the document's data.
     *
     * @throws \Exception
     */
    public function execute()
    {
//        if ($this->type === 'pdf') {
//            $this->convertToDocx();
//        }

        if ($this->type !== 'docx') {
            $this->convertToDocx();
        }

        $json = $this->convertDocxToJson();

        $this->storage->put("$this->directoryPath/document.json", json_encode($json));

        $this->deleteOriginalDocument();
    }

    protected function convertDocxToJson()
    {
        $reader = new DocxReader($this->storage, $this->path);

        return $reader->execute();
    }

    /**
     * Convert document to DOCX format in order to extract data.
     *
     * @throws \Exception
     */
    protected function convertToDocx()
    {
        $office = new Office();

        $convertTo = 'docx';

        if ($this->fileIsPDF($this->path)) {
            $convertTo = 'docx:writer_pdf_Export';
        }

        $success = $office->run(
            $convertTo,
            $this->storage->path($this->path),
            $this->storage->path($this->directoryPath)
        );

        if (! $success) {
            throw new \Exception('Failed when converting from ' . $this->type . ' to DOCX for file: ' . $this->path);
        }

        $this->deleteOriginalDocument();

        $this->setPath(str_replace($this->type, 'docx', $this->path));

        if ( ! $this->storage->exists($this->path)) {
            throw new \Exception('Failed when converting from ' . $this->type . ' to DOCX for file: ' . $this->path . '. The DOCX file doesnt exist.');
        }

        $this->type = 'docx';
    }

    protected function fileIsPDF($filePath)
    {
        $s = '.pdf';

        return substr($filePath, - strlen($s)) === $s;
    }
}