SilverStripe\TextExtraction\Extractor\SolrCellTextExtractor::getContent()

Complexity

Conditions	9
Paths	19

Size

Total Lines

57

Duplication

Lines	0
Ratio	0 %

Importance

Changes

0

<?php
    public function getContent($file)
    {
        if (!$file || (is_string($file) && !file_exists($file))) {
            // no file
            return '';
        }

        $fileName = $file instanceof File ? $file->getFilename() : basename($file);
        $client = $this->getHttpClient();

        // Get and validate base URL
        $baseUrl = $this->config()->get('base_url');
        if (!$this->config()->get('base_url')) {
            throw new InvalidArgumentException('SolrCellTextExtractor.base_url not specified');
        }

        try {
            $path = $this->getPathFromFile($file);

It seems like $file can also be of type string; however, parameter $file of SilverStripe\TextExtraction\Extractor\FileTextExtractor::getPathFromFile() does only seem to accept SilverStripe\Assets\File, maybe add an additional type check?

            $request = $client
                ->post($baseUrl)
                ->addPostFields(['extractOnly' => 'true', 'extractFormat' => 'text'])

The method addPostFields() does not exist on Psr\Http\Message\ResponseInterface.

                ->addPostFiles(['myfile' => $path]);
            $response = $request->send();
        } catch (InvalidArgumentException $e) {
            $msg = sprintf(
                'Error extracting text from "%s" (message: %s)',
                $fileName,
                $e->getMessage()
            );
            Injector::inst()->get(LoggerInterface::class)->notice($msg);

            return null;
        } catch (Exception $e) {
            // Catch other errors that Tika can throw vai Guzzle but are not caught and break Solr search
            // query in some cases.
            $msg = sprintf(
                'Tika server error attempting to extract from "%s" (message: %s)',
                $path,
                $e->getMessage()
            );

            Injector::inst()->get(LoggerInterface::class)->notice($msg);

            return null;
        }

        // Just initialise it, it doesn't take much.
        $matches = [];

        // Use preg match to avoid SimpleXML running out of memory on large text nodes
        preg_match(
            sprintf('/\<str name\="%s"\>(.*?)\<\/str\>/s', preg_quote($fileName)),
            (string)$response->getBody(),
            $matches
        );

        return $matches ? $matches[1] : null;
    }