Cion\TextToSpeech\Sources\WebsiteSource

Complexity

Total Complexity

9

Size/Duplication

Total Lines	71
Duplicated Lines	0 %

Coupling/Cohesion

Components	0
Dependencies	1

Importance

Changes

0

3 Methods

Rating	Name	Duplication	Size	Complexity
A	handle()	0	11	2
A	getDOMDocumentArticle()	0	11	2
A	getTextFromArticle()	0	27	5

<?php

namespace Cion\TextToSpeech\Sources;

use Cion\TextToSpeech\Contracts\Source as SourceContract;
use DOMDocument;
use DOMNodeList;
use RecursiveIteratorIterator;

class WebsiteSource implements SourceContract
{
    /**
     * Handles in getting the text from source.
     *
     * @param  string $data
     * @return string
     */
    public function handle(string $data): string
    {
        $articles = $this->getDOMDocumentArticle($data);

        if ($articles === null)
        {
            return '';
        }

        return $this->getTextFromArticle($articles);
    }

    /**
     * Get the DOM Node List of article tag.
     *
     * @return DOMNodeList|null
     */
    protected function getDOMDocumentArticle(string $url)
    {
        $dom = new DOMDocument();
        @$dom->loadHTML(file_get_contents($url));

It seems like you do not handle an error condition here. This can introduce security issues, and is generally not recommended.

        $element = $dom->getElementsByTagName('article')->item(0);

        if ($element !== null)
        {
            return $element->childNodes;
        }
    }

    /**
     * Get text from the articles DOM Node List.
     *
     * @param DOMNodeList $articles
     * @return string
     */
    protected function getTextFromArticle(DOMNodeList $articles): string
    {
        $text = '';

        for ($i = 0; $i < $articles->length; $i++)
        {
            // Check element if there is a childNodes
            if ($articles->item($i)->childNodes === null)
            {
                continue;
            }

            $dit = new RecursiveIteratorIterator(
                new RecursiveDOMIterator($articles->item($i)),
                RecursiveIteratorIterator::SELF_FIRST
            );
            foreach ($dit as $node)
            {
                if ($node->nodeName === 'p')
                {
                    $text .= $node->textContent;
                }
            }
        }
        
        return $text;
    }
}