dchadwick/web/modules/custom/ufc/src/Services/FighterImporter.php

<?php

namespace Drupal\ufc\Services;

use Drupal\Core\Entity\EntityTypeManager;
use Drupal\Core\Cache\CacheBackendInterface;
use Drupal\Core\Cache\Cache;
use Drupal\media\Media;
use Drupal\ufc\Fighter;
use GuzzleHttp\Client;
use Symfony\Component\DomCrawler\Crawler;

class FighterImporter {

  /**
   * Guzzle http client service.
   *   @var \GuzzleHttp\Client
   */
  protected $httpClient;

  /**
   * Entity type manager service.
   *   @var \Drupal\Core\Entity\EntityTypeManager
   */
  protected $entityTypeManager;

  /**
   * Array of all fighters.
   *  @var array[]
   */
  public $fighters = [];

  /**
   * The current weight class.
   *   @var string
   */
  protected $weightClass;

  /**
   * The UFC cache bin.
   */
  protected $cache;

  /**
   * The base url for fighter lists.
   *  @var string
   */
  const UFC_BASE = "https://www.ufc.com/athletes/all?filters%5B0%5D=weight_class%3";

  /**
   * All applicable divisions and their keys on UFC.com.
   *   @var array[]
   */
  protected $divisions = [
    'heavyweight' =>  "A11",
    'light_heavyweight' => "A13",
    'middleweight' => "A14",
    'welterweight' => "A15",
    'lightweight' => "A12",
    'featherweight' => "A9",
    'bantamweight' => "A8",
    'flyweight' => "A10",
  ];

  /**
   * Public constructor for Fighter Importer.
   *
   * @param Client $httpClient
   * @param EntityTypeManager $entityTypeManager
   * @param Cache $cache
   */
  public function __construct(
    Client $httpClient,
    EntityTypeManager $entityTypeManager,
    CacheBackendInterface $cache
  ) {
    $this->httpClient = $httpClient;
    $this->entityTypeManager = $entityTypeManager;
    $this->cache = $cache;
  }

  public function importFighters(): void {
    $test_run = FALSE;
    if ($test_run) {
      // Overriding to test!!!
      $fighter_name_text_on_ufc = "
          Georges St-Pierre
        ";
      $fighter = new Fighter($this->httpClient);
      $fighter->first_name = 'scott';
      $fighter->last_name = 'adams';
      $fighter->scrapeDataFromFighterPage();
      $fighter_clone = clone ($fighter);
      unset($fighter_clone->fighter_page);
      unset($fighter_clone->crawler);
      dump($fighter_clone);
    }
    else {
      $fighters_by_div = self::getListOfCurrentFighters();
      // Process each fighter into system.
      foreach ($fighters_by_div as $division => $fighters) {
        $this->processDivision($division, $fighters);
       }
    }
  }
  /**
     * Process a division.
     *
     * @param mixed $div
     * @param mixed $fighters
     */
  public static function processDivision($div, $fighters): void {
    \Drupal::logger('ufc')->notice("Starting to update $div");

    foreach ($fighters as $fighter_data) {
      $fighter = new Fighter(\Drupal::httpClient());
      $fighter->first_name = $fighter_data['firstname'];
      $fighter->last_name = $fighter_data['lastname'];
      $fighter->image = $fighter_data['image'];
      $fighter->class = $div;
      if (!$fighter->scrapeDataFromFighterPage($fighter_data['profile'])) {
        \Drupal::logger('ufc')->alert("FAILED: $fighter->first_name $fighter->last_name to " . $fighter_data['profile']);
      }
      // Check if node exists, by title.
      $fighter->createMediaEntityFromImage();
      $title = $fighter->first_name . " " . $fighter->last_name;
      $node_lookup = reset(\Drupal::entityTypeManager()->getStorage('node')->loadByProperties(['title' => $title]));

      if (!empty($node_lookup)) {
        // Update instead of create.
        $fighter->updatePlayerNode($node_lookup->id());
        \Drupal::logger('ufc')->notice("$title updated successfully.");
      }
      else {
        \Drupal::logger('ufc')->warning("No existing player found for $title...creating");
        $fighter->createPlayerNode();
      }
    }
    $context['results']['processed']++;
  }

  /**
   * Get list of current fighters.
   *
   * @return array $fighters
   */
  public function getListOfCurrentFighters(): array {
    foreach ($this->divisions as $division => $div_base_url) {
      $division_url = self::UFC_BASE . $div_base_url;
      $this->weightClass = $division;
      echo "Starting import for " . $division . "\n";
      self::loopThroughFighterPages($division_url);
    }
    return $this->fighters;
  }

  /**
   * There is a pager, loop through to get all fighters.
   *
   * @param string $base_url
   */
  public function loopThroughFighterPages($base_url): void {
    // Implement caching to store instead of needing fresh requests.
    for ($i=0; $i<=100; $i++) {
      $ua = 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/17.4.1 Safari/605.1.15';
      $headers = [
        'referer' => true,
        'verify' => false,
        'headers' => [
          'User-Agent' => $ua,
          'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8',
          'Accept-Encoding' => 'gzip, deflate, br',
        ]
      ];
      $url = $base_url . "&page=$i";
      /* $cid = "ufc:" . $url; */
      $request = $this->httpClient->request('GET', $url, $headers);
      $content = $request->getBody()->getContents();
      $invalid_page = strpos($content, "No Result Found For");
      if (!$invalid_page) {
        \Drupal::logger('ufc')->notice("Extracting fighters from page $i.");
        self::extractFighters($content);
        continue;
      }
      break;
    }
  }

  public function quickTest($str) {
    return "hello$str";
  }

  /**
   * Extract fighters from an html string.
   */
  public function extractFighters(string $input): void {
    $fighter_list = [];
    $crawler = new Crawler($input);
    $athlete_flipcards = $crawler->filter('.c-listing-athlete-flipcard');
    $fighter_names = $athlete_flipcards->each(function (Crawler $crawler, $i) {
      return $crawler->filter('.c-listing-athlete__name')->text();
    });
    $fighter_profile_urls = $athlete_flipcards->each(function (Crawler $crawler, $i) {
      return $crawler->filter('.e-button--black')->attr('href');
    });

    $fighter_images = $athlete_flipcards->each(function (Crawler $crawler, $i) {
      $imgs = $crawler->filter('img')->each(function ($i) {
        return $i->attr('src');
      });
      return $imgs;
    });

    $count_fighter_names = count($fighter_names);
    $count_profile_urls = count($fighter_profile_urls);
    $count_images = count($fighter_images);
    // Make sure the arrays are all the same size.
    assert((
      ($count_fighter_names == $count_profile_urls) &&
      ($count_profile_urls == $count_images)
    ));

    foreach ($fighter_names as $key => $fighter) {
      $fighter_names[$key] = [
        'name' => $fighter,
        'profile' => $fighter_profile_urls[$key],
        'images' => $fighter_images[$key],
      ];
    }

    foreach ($fighter_names as $fighter_data) {
      $name_no_spaces = str_replace(" ", "", $fighter_data['name']);
      $weight_class_exists = array_key_exists($this->weightClass, $this->fighters);
      $fighter_exists = NULL;
      if ($weight_class_exists) {
        $fighter_exists = array_key_exists($name_no_spaces, $this->fighters[$this->weightClass]);
      }
      if (!$fighter_exists) {
        $split_name = explode(" ", $fighter_data['name'], 2);
        $this->fighters[$this->weightClass][$name_no_spaces]['firstname'] = $split_name[0] ?? " ";
        $this->fighters[$this->weightClass][$name_no_spaces]['lastname'] = $split_name[1] ?? " ";
        $this->fighters[$this->weightClass][$name_no_spaces]['profile'] = $fighter_data['profile'];
      }

      // here you are dan, set the profile url of the fighter, then use that in parsing later on

      if (!empty($fighter_data['images']) && count($fighter_data['images']) == 2) {
        $this->fighters[$this->weightClass][$name_no_spaces]['image'] = $fighter_data['images'][0];
      }
      else {
        $this->fighters[$this->weightClass][$name_no_spaces]['image'] = FALSE;
      }
    }
  }

}