tmgmt_smartling-8.x-4.11/src/Context/HtmlAssetInliner.php

<?php

namespace Drupal\tmgmt_smartling\Context;

//set_time_limit(300);

use Drupal;
use GuzzleHttp\ClientInterface;
use GuzzleHttp\Cookie\CookieJar;
use GuzzleHttp\Cookie\FileCookieJar;
use GuzzleHttp\Exception\RequestException;
use GuzzleHttp\Pool;
use GuzzleHttp\Psr7\Request;
use GuzzleHttp\Psr7\Response;

class HtmlAssetInliner {

  private CookieJar $cookieJar;

  protected static $uriMissingError = array(
    "response" => array(
      "code" => "VALIDATION_ERROR",
      "data" => array("baseUrl" => NULL, "body" => NULL, "headers" => NULL),
      "messages" => array("fileUri parameter is missing."),
    ),
  );


  public function __construct(private ClientInterface $client) {
    $this->cookieJar = new FileCookieJar(tempnam(sys_get_temp_dir(), 'smartling_cookies_'));
  }

  /**
   * Gets complete page data and returns generated string
   *
   * @param string $url - url to retrieve
   * @param bool $keepjs - whether to keep javascript
   * @param bool $compress - whether to remove extra whitespaces
   * @param array $settings
   * @param bool $debug
   *
   * @return string|void
   * @throws \Exception - throws an exception if provided url isn't in proper format
   */
  public function getCompletePage($url, $compress = FALSE, array $settings = [], $debug = FALSE) {
    # validate the URL
    if (!filter_var($url, FILTER_VALIDATE_URL)) {
      throw new \Exception('Invalid URL. Make sure to specify http(s) part.');
    }

    if (empty($url)) {
      if ($debug) {
        Drupal::logger('tmgmt_smartling_context_debug')->info('Url is missing.');
      }

      return self::$uriMissingError;
    }

    $html = $this->getUrlContents($url,
      0,
      'Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10_5_8; en-US) AppleWebKit/534.10 (KHTML, like Gecko) Chrome/8.0.552.215 Safari/534.10',
      $settings,
      $debug
    );

    if (strlen($html) <= 300) {
      if ($debug) {
        Drupal::logger('tmgmt_smartling_context_debug')->info('Response is too small.');
      }

      return '';
    }

    return ($compress) ? $this->compress($html) : $html;
  }

  /**
   * Checks whether or not remote file exists
   *
   * @param $url
   *
   * @param $proj_settings
   * @param float $connection_timeout
   * @param float $timeout
   *
   * @return bool
   */
  public function remoteFileExists($url, $proj_settings, $connection_timeout = 0.5, $timeout = 5) {
    try {
      $options = [
        'timeout' => $timeout,
        'connect_timeout' => $connection_timeout,
        'cookies' => $this->cookieJar,
      ];
      $this->applySettingsToOptions($proj_settings, $options);

      $response = $this->client->head($url, $options);
      return $response->getStatusCode() < 400;
    }
    catch (RequestException $e) {
      return FALSE;
    }
  }

  /**
   * Compresses generated page by removing extra whitespace
   */
  private function compress($string) {
    # remove whitespace
    return str_replace(array(
      "\r\n",
      "\r",
      "\n",
      "\t",
      '  ',
      '    ',
      '    '
    ), ' ', $string);
  }

  /**
   * Gets content for given url using Guzzle and optionally using user agent
   *
   * @param $url
   * @param float $timeout
   * @param string $user_agent
   * @param array $settings
   * @param bool $debug
   *
   * @return int|mixed
   */
  public function getUrlContents(
    $url,
    $timeout = 0,
    $user_agent = 'Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10_5_8; en-US) AppleWebKit/534.10 (KHTML, like Gecko) Chrome/8.0.552.215 Safari/534.10',
    array $settings = [],
    $debug = FALSE
  ) {
    $options = [
      'headers' => ['User-Agent' => $user_agent],
      'timeout' => $timeout,
      'connect_timeout' => 5,
      'cookies' => $this->cookieJar,
      'allow_redirects' => TRUE,
    ];

    $this->applySettingsToOptions($settings, $options);

    try {
      $response = $this->client->get($url, $options);
      $body = (string) $response->getBody();

      if ($debug) {
        Drupal::logger('tmgmt_smartling_context_debug')->info('Guzzle request options: @options', [
          '@options' => print_r($options, TRUE),
        ]);
        Drupal::logger('tmgmt_smartling_context_debug')->info('Curl response headers: @response_headers', [
          '@response_headers' => print_r($response->getHeaders(), TRUE),
        ]);
        Drupal::logger('tmgmt_smartling_context_debug')->info('Curl response body: @response_body', [
          '@response_body' => substr($body, 0, 500) . '*****',
        ]);
      }

      return $body;
    }
    catch (RequestException $e) {
      if ($debug) {
        Drupal::logger('tmgmt_smartling_context_debug')->error('Guzzle error: @message', [
          '@message' => $e->getMessage(),
        ]);
      }
      return -1;
    }
  }

  public function getUrlContentsPooled(
    $urls,
    $timeout = 0,
    $user_agent = 'Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10_5_8; en-US) AppleWebKit/534.10 (KHTML, like Gecko) Chrome/8.0.552.215 Safari/534.10',
    array $settings = [],
    $debug = FALSE
  ) {
    $options = [
      'headers' => ['User-Agent' => $user_agent],
      'timeout' => $timeout,
      'connect_timeout' => 5,
      'cookies' => $this->cookieJar,
      'allow_redirects' => TRUE,
    ];

    $this->applySettingsToOptions($settings, $options);

    $requests = function ($urls_to_fetch) use ($options) {
      foreach ($urls_to_fetch as $resourceId => $url) {
        yield $resourceId => new Request('GET', $url, $options['headers'] ?? []);
      }
    };

    $results = [];
    $pool = new Pool($this->client, $requests($urls), [
      'concurrency' => 5,
      'fulfilled' => function (Response $response, $resourceId) use (&$results) {
        $results[$resourceId] = (string) $response->getBody();
      },
      'rejected' => function (RequestException $e, $resourceId) use (&$results, $debug) {
        if ($debug) {
          Drupal::logger('tmgmt_smartling_context_debug')->error('Guzzle error: @message', [
            '@message' => $e->getMessage(),
          ]);
        }
        $results[$resourceId] = -1;
      },
    ]);

    $promise = $pool->promise();
    $promise->wait();
    return $results;
  }

  /**
   * @param $proj_settings
   * @param $options
   */
  private function applySettingsToOptions(array $proj_settings, array &$options): void {
    if (!empty($proj_settings['enable_basic_auth'])) {
      $options['auth'] = [
        $proj_settings['basic_auth']['login'] ?? '',
        $proj_settings['basic_auth']['password'] ?? '',
      ];
    }

    if (!empty($proj_settings['context_skip_host_verifying'])) {
      $options['verify'] = FALSE;
    }
  }

}