tiktokscraper/app/Libraries/TikTokScraper.php

<?php
namespace App\Libraries;

use \DOMXPath;
use \DOMDocument;
use \DOMNode;
use \DOMNodeList;
use GuzzleHttp\Client;
use \Exception;
use Illuminate\Support\Facades\Log;

class TikTokScraper
{
    /**
     * @var DOMDocument
     */
    private $dom;

    /**
     * @var Client
     */
    private $client;

    public function __construct()
    {
        $this->client = new Client();
    }

    /**
     * Loads specified URL, and loads HTML into scraper
     *
     * @param string $url URL to fetch
     * @return $this
     * @throws Exception
     */
    public function get($url) {
        $response = $this->client->request(
            'GET',
            $url,
            ['headers' => self::composeHeaders()]
        );
        return $this->load($response->getBody()->getContents());
    }

    /**
     * Returns first matching node from DOM.
     *
     * @param string $xpath Xpath to look for in the DOM
     * @param null $parent Optional. Element to use as parent when querying DOM
     *
     * @return DOMNode|false Returns DOMNode on success, or false on failure
     * @throws Exception
     */
    public function getNode($xpath, $parent = null)
    {
        $nodes = $this->getNodes($xpath, $parent);
        if ($nodes->length === 0) {
            throw new \Exception("No nodes found matching < $xpath > xpath");
        }
        return $nodes[0];
    }

    /**
     * Returns all nodes matching specified XPath
     *
     * @param string $xpath Xpath to look for in the DOM
     * @param null $parent Optional. Element to use as parent when querying DOM
     *
     * @return DOMNodeList|false Returns list of DOMNodes on success, or false on failure
     */
    public function getNodes($xpath, $parent = null)
    {
        $domXpath = new DOMXPath($this->dom);
        return $domXpath->query($xpath, $parent);
    }

    /**
     * Extracts user data from page loaded into scraper
     *
     * @return mixed Returns userData object
     * @throws Exception
     */
    public function extractUserData() {
        try {
            $script = $this->getNode('//script[@id="__NEXT_DATA__"]');
            $data = json_decode($script->nodeValue, false);
        } catch (Exception $e) {
            Log::error("Error getting node data. Error thrown: {$e->getMessage()}");
            throw $e;
        }
        if ($data->props->pageProps->statusCode > 0) {
            throw new Exception("Error getting data from TikTok, status code {$data->props->pageProps->statusCode}");
        } else {
            return $data->props->pageProps->userData;
        }
    }

    /**
     * Extracts video data from page loaded into scraper
     *
     * @return mixed Returns videoData object
     * @throws Exception
     */
    public function extractVideoData() {
        try {
            $script = $this->getNode('//script[@id="__NEXT_DATA__"]');
            $data = json_decode($script->nodeValue, false);
        } catch (Exception $e) {
            Log::error("Error getting node data. Error thrown: {$e->getMessage()}");
            throw $e;
        }
        if ($data->props->pageProps->statusCode > 0) {
            throw new Exception("Error getting data from TikTok, status code {$data->props->pageProps->statusCode}");
        } else {
            return $data->props->pageProps->videoData;
        }
    }

    /**
     * Loads supplied HTML into DOMDocument
     *
     * @param string $html HTML to load into DOMDocument
     * @return $this
     */
    private function load($html)
    {
        try {
            $this->dom = new DOMDocument;
            libxml_use_internal_errors(true);
            $this->dom->loadHTML($html);
            libxml_clear_errors();
        } catch (Exception $e) {
            Log::error("There was an error loading HTML document {$e->getMessage()}");
        }
        return $this;
    }

    /**
     * Generates random user-agent string, to use when fetching pages
     *
     * @return array
     * @throws Exception
     */
    private static function composeHeaders() {
        return [
            'User-Agent' => \Campo\UserAgent::random(),
            'Accept'     => '*',
        ];
    }

}