ifhub-telegram/bot.php

<?php
require_once ("vendor/autoload.php");

use Colorfield\Mastodon\MastodonAPI;
use Symfony\Component\Yaml\Yaml;

$config = Yaml::parse(file_get_contents('config.yml'));
$seen = [];
if (file_exists('.seen.yml')) {
  $seen = Yaml::parse(file_get_contents('.seen.yml'));
}

function get_text($url) {
  $curl = curl_init();
  curl_setopt_array($curl, array(
    CURLOPT_RETURNTRANSFER => 1,
    CURLOPT_URL => $url,
  ));
  $resp = curl_exec($curl);
  curl_close($curl);
  return $resp;
}

function ellipse($str,$n_chars,$crop_str='[…]')
{
  $buff = $str;
  if(strlen($buff) > $n_chars)
  {
    $cut_index=strpos($buff,' ',$n_chars);
    $buff=substr($buff,0,($cut_index===false? $n_chars: $cut_index+1)).$crop_str;
    // then cut for last newline or dot
    $cut_index_a=strpos($buff,'.',$n_chars);
    $cut_index_b=strpos($buff,PHP_EOL,$n_chars);
    $cut_index = max((int) $cut_index_a, (int) $cut_index_b);
    $buff=substr($buff,0,($cut_index===false? $n_chars: $cut_index+1)).$crop_str;
  }
  return trim($buff);
}

function formatdsc($description) {
  global $pandoc;
  $description = strip_tags($description, 'img');
  $description = $pandoc->runWith($description, [
    'from' => 'html',
    'to' => 'plain',
    'wrap' => 'none'
  ]);
  $description = str_replace('Читать дальше', '', $description);
  $description = trim($description);
  return $description;
}

function download($url, $outFile) {
  $options = array(
    CURLOPT_FILE    => fopen($outFile, 'w'),
    CURLOPT_TIMEOUT =>  28800, // set this to 8 hours so we dont timeout on big files
    CURLOPT_URL     => $url
  );

  $ch = curl_init();
  curl_setopt_array($ch, $options);
  curl_exec($ch);
  curl_close($ch);
}

$string = get_text('https://ifhub.club/rss/full/');
$service = new \Sabre\Xml\Service();
$service->elementMap = [
  '{}item' => function(\Sabre\Xml\Reader $reader) {
    return \Sabre\Xml\Deserializer\keyValue($reader, '');
  },
  '{}channel' => function(\Sabre\Xml\Reader $reader) {
    return \Sabre\Xml\Deserializer\repeatingElements($reader, '{}item');
  },
];
$articles = $service->parse($string)[0]['value'];
unset($string);
$pandoc = new \Pandoc\Pandoc();
try {
$telegram = new Longman\TelegramBot\Telegram(
  $config['TELEGRAM_API_KEY'],
  $config['TELEGRAM_BOT_NAME']
);
} catch (\Exception $e) {
  if (!$config['DRY_RUN']) {
    echo 'Wrong Telegram API key.';
    return;
  };
}
try {
  $oAuth = new Colorfield\Mastodon\MastodonOAuth('IFHub.club', $config['MASTODON_SERVER']);
  $oAuth->config->setAuthorizationCode($config['MASTODON_ACCESS_TOKEN']);
  $mastodon = new Colorfield\Mastodon\MastodonAPI($oAuth->config);
} catch (\Exception $e) {
  echo 'WARNING: could not start Mastodon client';
}

foreach ($articles as $article) {
  if (in_array($article['link'], $seen)) {
    continue;
  }
  $title = $article['title'];
  $link_mastodon = $article['link'];
  $description = $article['description'];
  $image = NULL;
  preg_match('/<img.+src=[\'"](?P<src>.+?)[\'"].*>/i', $description, $image);
  if (isset($image[1])) {
    $image = $image[1];
  }
  $long_description = $description;
  $description = substr( $description, 0, strpos($description, '<a name="cut"'));
  $description = formatdsc($description);
  /*
  if (strlen($description) < 50) { // description is too small
    $description = $long_description;
    $description = formatdsc($description);
  }
   */
  $limit = 500 - strlen($link_mastodon) - strlen($title) - 20;
  $title = trim($title);
  $description = "$title\n\n".ellipse($description, $limit);
  $description = $description."\n$link_mastodon";

  if (!$config['DRY_RUN']) {
    if ($config['TELEGRAM'] === true) {
      try {
        $result = \Longman\TelegramBot\Request::sendMessage([
          'chat_id' => $config['TELEGRAM_CHAT_ID'],
          'text' => $description,
          'parse_mode' => 'Markdown'
        ]);
        if (!$config['DRY_RUN']) {
          file_put_contents('.lastrun', time());
        }
      } catch (Longman\TelegramBot\Exception\TelegramException $e) {
        echo $e;
      }
    }
    if ($config['MASTODON'] === true && isset($mastodon)) {
      /*
      if ($image) {
        download('https://ifhub.club'.$image, './'.basename($image));
        $attachment = $mastodon->post('/media', [
          'file' => file_get_contents('./'.basename($image))
        ]);
        var_dump($attachment);
        unlink('./'.basename($image));
        $mdescription .= $attachment->url;
      }*/
      $mastodon->post('statuses', [
        'content' => $description,
        'language' => 'ru',
      ]);
      $seen[] = $article['link'];
    }
    if (!$config['DRY_RUN'] && !empty($config['INDEXNOW'])) {
      $client = new \GuzzleHttp\Client();
      $client->request('GET', 'https://www.bing.com/indexnow?url='.$link.'&key='.$config['INDEXNOW']);
    }
  } else {
    echo $description."\n---\n";
  }
}
if (!$config['DRY_RUN']) {
  file_put_contents('.seen.yml', Yaml::dump($seen));
}