<?php

/*
 ****
 * DB
 ****
 */

$dbType = $_ENV['DB_TYPE'] ?? 'sqlite';
$dbPath = $_ENV['DB_PATH'] ?? './searcher.sqlite';
$dbHost = $_ENV['DB_HOST'] ?? '';
$dbPort = $_ENV['DB_PORT'] ?? '';
$dbName = $_ENV['DB_NAME'] ?? '';
$dbUsername = $_ENV['DB_USERNAME'] ?? '';
$dbPassword = $_ENV['DB_PASSWORD'] ?? '';

$dsnParts = [
  'host' => $dbHost,
  'port' => $dbPort,
  'dbname' => $dbName,
];
$dsnParts = array_filter($dsnParts, fn ($value) => ! empty($value));
$dsnParts = array_map(function ($key, $value) {
  return "$key=$value";
}, array_keys($dsnParts), $dsnParts);

$dsn = "$dbType:" . ($dbType == 'sqlite' ? $dbPath : implode(';', $dsnParts));

global $db;
$db = new PDO(
  $dsn,
  $dbUsername,
  $dbPassword,
  [
    PDO::ATTR_DEFAULT_FETCH_MODE => PDO::FETCH_ASSOC,
  ],
);

if (isset($_GET['init--db'])) {
  $db->query(<<<SQL
    create table if not exists search_index (
      id integer primary key,
      url text unqiue,
      content text,
      excerpt text,
      title text,
      timestamp integer
    );
    SQL)
    ->execute();
}


/*
 ****
 * Crawler
 ****
 */

if (! empty($_GET['crawl'])) {
  $targetToCrawl = $_GET['crawl'];
  $targetUrl = parse_url($targetToCrawl);
  $baseUrl = "{$targetUrl['scheme']}://{$targetUrl['host']}";

  $streamContext = stream_context_create([
    'http' => [
      'follow_location' => true,
    ],
  ]);

  $crawledPagesMax = intval($_GET['max-pages'] ?? 10);
  $crawledPagesCount = 0;

  $responseCode = intval(explode(' ', get_headers("$baseUrl/sitemap.xml")[0])[1]);
  if ($responseCode == 200) {
    $dom = new DOMDocument();
    @$dom->loadHTML(file_get_contents("$baseUrl/sitemap.xml", false, $streamContext));
    $xpath = new DomXPath($dom);

    /**@var DOMNodeList $locations*/
    $locations = $xpath->query('//loc');
    foreach ($locations as $location) {
      /*@var DOMNode $location*/
      $locationUrl = $location->textContent;
      if (str_starts_with($locationUrl, '//')) {
        $locationUrl = "{$targetUrl['scheme']}:" . $locationUrl;
      }

      index_page($locationUrl);

      $crawledPagesCount++;
      if ($crawledPagesCount >= $crawledPagesMax) {
        break;
      }
    }
  } else {
    index_page($targetToCrawl);

    $dom = new DOMDocument();
    @$dom->loadHTML(file_get_contents($targetToCrawl, false, $streamContext));
    $xpath = new DomXPath($dom);

    $links = $xpath->query('//a');
    foreach ($links as $link) {
      /*@var DOMNode $link*/
      $href = $link->attributes->getNamedItem('href')->nodeValue;
      $fullPath = str_starts_with($href, 'http') ? $href : "$baseUrl$href";

      $linkUrl = parse_url($fullPath);
      if ($linkUrl['host'] != $targetUrl['host']) {
        continue;
      }

      index_page($fullPath);
    }
  }

  // TODO: check sitemap first, only check those links then
  // TODO: otherwise get every link on site and crawl that
  // TODO: check if link (without query params?) already visited for recursive protection
  // strip html => store text in db
}

function index_page(string $url): bool {
  global $db;

  $url = trim($url, "/ \n\r\t\v\0");

  Logger::log("Indexing $url");

  $streamContext = stream_context_create([
    'http' => [
      'follow_location' => true,
    ],
  ]);

  $html = file_get_contents($url, false, $streamContext);
  $dom = new DOMDocument();
  @$dom->loadHTML($html);
  $xpath = new DomXPath($dom);

  // $content = strip_tags($content);
  $content = $html;


  // Excerpt

  /**@var DOMNodeList $excerpt*/
  $excerpt = $xpath->query('//meta[@name="description"]');
  if ($excerpt->count() > 0) {
    $excerpt = $excerpt->item(0)->attributes->getNamedItem('content')->textContent ?? '';
  } else {
    $excerpt = '';
  }

  if (empty($excerpt)) {
    $excerpt = $xpath->query('//p');
    if ($excerpt->count() > 0) {
      $excerpt = $excerpt->item(0)->textContent;
      $excerpt = trim($excerpt);
    } else {
      $excerpt = '';
    }
  }


  // Title

  /**@var DOMNodeList $title*/
  $title = $xpath->query('//title');
  if ($title->count() > 0) {
    $title = $title->item(0)->textContent;
    $title = trim($title);
  } else {
    $title = $url;
  }


  // Insert

  $db->prepare(<<<SQL
    delete from search_index where url=:url
    SQL)
    ->execute([
      'url' => $url,
    ]);

  return $db->prepare(<<<SQL
    insert into search_index (url, content, excerpt, title, timestamp) values (:url, :content, :excerpt, :title, :timestamp)
    SQL)
    ->execute([
      'url' => $url,
      'content' => $content,
      'excerpt' => $excerpt,
      'title' => $title,
      'timestamp' => time(),
    ]);
}


/*
 ****
 * Search
 ****
 */

if (isset($_GET['search'])) {
  ?>
  <!DOCTYPE html>
  <html>
    <body>
      <div style="max-width: 1200px; margin: 0 auto;">
        <form>
          <input type="search" name="search" value="<?php echo $_GET['search'] ?? ''; ?>">
          <input type="submit" value="Search">
        </form>
  <?php

  if (! empty($_GET['search'])) {
    $statement = $db->prepare(<<<SQL
      select * from search_index
      where content like :search
      limit :limit offset :offset
    SQL);
    $statement->execute([
      'search' => "%{$_GET['search']}%",
      'limit' => 10,
      'offset' => (($_GET['page'] ?? 1) - 1) * 10
    ]);
    $result = $statement->fetchAll();

    foreach ($result as $row) {
      $excerpt = $row['excerpt'];
      if (empty($excerpt)) {
        $excerpt = "EXCERPT";
      }

      ?>
      <hr>
      <div>
        <a href="<?php echo $row['url']; ?>"><?php echo $row['title']; ?></a>
        <p><?php echo $row['excerpt']; ?></p>
        <a href="<?php echo $row['url']; ?>"><?php echo $row['url']; ?></a>
      </div>
      <?php
    }

    $query = array_diff_key($_GET, array_flip(['page']));
    ?>
    <form>
      <?php foreach ($query as $key => $value): ?>
        <input type="hidden" name="<?php echo $key; ?>" value="<?php echo $value; ?>">
      <?php endforeach; ?>
      <button name="page" value="<?php echo intval($_GET['page'] ?? 1) - 1; ?>">Previous</button>
      <button name="page" value="<?php echo intval($_GET['page'] ?? 1) + 1; ?>">Next</button>
    </form>
    <?php
  }

  ?>
      </div>
    </body>
  </html>
  <?php
}


/*
 ****
 * API
 ****
 */

if (isset($_GET['api'])) {

  /*
   ****
   * API Search
   ****
   */

  if (! empty($_GET['api_search'])) {
    $statement = $db->prepare(<<<SQL
      select * from search_index
      where content like :search
      SQL);
    $statement->execute(['search' => "%{$_GET['api_search']}%"]);
    $result = $statement->fetchAll();

    header('Content-Type: application/json');

    $json = [];
    foreach ($result as $row) {
      $excerpt = $row['excerpt'];
      if (empty($excerpt)) {
        $excerpt = "EXCERPT";
      }

      $json[] = [
        'url' => $row['url'],
        'title' => $row['title'],
        'excerpt' => $row['excerpt'],
      ];
    }

    echo json_encode($json);
  }
}


/*
 ****
 * Utilities
 ****
 */

class Logger {
  static function log(string $message): void {
    echo nl2br($message . "\n");
  }
}