FileBinaryMimeTypeGuesserProxy.php

<?php

declare(strict_types=1);

namespace Infrastructure\Symfony\Mime;

use Symfony\Component\Mime\FileBinaryMimeTypeGuesser;
use Symfony\Component\Mime\MimeTypeGuesserInterface;

/**
 * Proxy to fix weird mimetype guessing for docx file.
 */
final class FileBinaryMimeTypeGuesserProxy implements MimeTypeGuesserInterface
{
    private const DOCX_MIMETYPE = 'application/vnd.openxmlformats-officedocument.wordprocessingml.document';
    private const OCTET_STREAM_MIMETYPE = 'application/octet-stream';
    private const ZIP_MIMETYPE = 'application/zip';
    private const MAIN_PART_CONTENT_TYPE = '[Content_Types].xml';

    private FileBinaryMimeTypeGuesser $fileBinaryMimeTypeGuesser;

    public function __construct()
    {
        // the class we decorate
        $this->fileBinaryMimeTypeGuesser = new FileBinaryMimeTypeGuesser();
    }

    public function isGuesserSupported(): bool
    {
        return $this->fileBinaryMimeTypeGuesser->isGuesserSupported();
    }

    /**
     * In case if the mimetype is "application/octet-stream" or "application/zip" we check deeper if it's a docx otherwise
     * we return the mimetype guessed by FileBinaryMimeTypeGuesser.
     */
    public function guessMimeType(string $path): ?string
    {
        $mimeType = $this->fileBinaryMimeTypeGuesser->guessMimeType($path);
        
        // early return when the mime type the is well guessed 
        // (in case of the MIME type cannot be a docx weird guessed)
        if (true === $this->cannotBeAnUnguessedDocx($mimeType)) {
            return $mimeType;
        }

        try {
            // We inspect the file
            $zipClass = new \ZipArchive();
            $zipClass->open($path);

            // copied from https://github.com/PHPOffice/PHPWord/blob/be0190cd5d8f95b4be08d5853b107aa4e352759a/src/PhpWord/TemplateProcessor.php#L1023
            $contentTypes = $zipClass->getFromName(self::MAIN_PART_CONTENT_TYPE);

            if (false === $contentTypes) {
                return $mimeType;
            }

            // Usually, the name of main part document will be 'document.xml'. However, some .docx files (possibly those from Office 365, experienced also on documents from Word Online created from blank templates) have file 'document22.xml' in their zip archive instead of 'document.xml'.
            // We look for "application/vnd.openxmlformats-officedocument.wordprocessingml.document.main+xml" in content type to determine the document as docx
            // @see http://officeopenxml.com/anatomyofOOXML.php
            $pattern = '~PartName="\/(word\/document.*?\.xml)" ContentType="application\/vnd\.openxmlformats-officedocument\.wordprocessingml\.document\.main\+xml"~';

            return 1 === preg_match($pattern, $contentTypes) ? self::DOCX_MIMETYPE : $mimeType;
        } catch (\Exception $e) {
            return $mimeType;
        }
    }

    /**
     * Only mimetype application/octet-stream or application/zip can be docx.
     */
    private function cannotBeAnUnguessedDocx(?string $mimeType): bool
    {
        return self::OCTET_STREAM_MIMETYPE !== $mimeType
            && self::ZIP_MIMETYPE !== $mimeType
            ;
    }
}