<?php

namespace DaftPlug\Rankify\Admin;

use DaftPlug\Rankify\Helpers\Utils;

if (!defined('ABSPATH')) {
  exit();
}

/**
 * Counts what is actually true about the site's content, for the overview.
 *
 * Every figure here is a COUNT with an index-friendly WHERE — nothing loads
 * posts. The whole snapshot is cached because the overview is a page people
 * refresh, and recomputing six aggregates over the posts table on every load
 * would make the plugin the slowest screen in wp-admin on a large site.
 */
class SiteAudit
{
  const CACHE_KEY = 'rankify_site_audit';
  const CACHE_TTL = 900;

  /**
   * Two counted strings the Overview renders.
   *
   * They live here rather than in the view because Prettier's PHP plugin
   * relocates comments between the `<?php ?>` islands of a mixed HTML template,
   * which strands the `translators:` hint away from its gettext call and drops
   * it from the POT. Class files are not affected.
   */
  public static function thinContentNotice($count)
  {
    /* translators: %s: number of pages. */
    $notice = _n('%s page has too little content to generate a description from at all.', '%s pages have too little content to generate a description from at all.', $count, 'crownify');

    return sprintf($notice, number_format_i18n($count));
  }

  public static function pagesLabel($count)
  {
    /* translators: %s: number of pages. */
    $label = _n('%s page', '%s pages', $count, 'crownify');

    return sprintf($label, number_format_i18n($count));
  }

  public static function flush()
  {
    delete_transient(self::CACHE_KEY);
  }

  public static function snapshot()
  {
    $cached = get_transient(self::CACHE_KEY);
    if (is_array($cached)) {
      return $cached;
    }

    $snapshot = self::compute();
    set_transient(self::CACHE_KEY, $snapshot, self::CACHE_TTL);

    return $snapshot;
  }

  /**
   * Post types that can actually appear in search — the denominator for every
   * percentage below. Excludes attachments, whose pages redirect to the file.
   */
  public static function indexablePostTypes()
  {
    $postTypes = get_post_types(['public' => true], 'names');
    unset($postTypes['attachment']);

    foreach ($postTypes as $postType) {
      if ('on' === Utils::getSetting('titles[' . $postType . '][noindex]')) {
        unset($postTypes[$postType]);
      }
    }

    return array_values($postTypes);
  }

  protected static function compute()
  {
    global $wpdb;

    $postTypes = self::indexablePostTypes();
    if (empty($postTypes)) {
      return self::emptySnapshot();
    }

    $total = self::countPublished($postTypes);

    if (0 === $total) {
      return self::emptySnapshot();
    }

    $missingDescription = self::countWithoutMeta($postTypes, '_rankify_description');
    $missingTitle = self::countWithoutMeta($postTypes, '_rankify_title');
    $noindexed = self::countWithMeta($postTypes, '_rankify_noindex', 'on');

    // No featured image means no og:image and no ImageObject in the graph, so
    // shares render as a bare link and the page loses a rich-result signal.
    $missingImage = self::countPublished($postTypes, " AND NOT EXISTS (SELECT 1 FROM %i m WHERE m.post_id = p.ID AND m.meta_key = %s AND m.meta_value != '')", [$wpdb->postmeta, '_thumbnail_id']);

    // A long post title will almost certainly overrun once the template adds
    // the separator and site name. Characters, not pixels — pixel width cannot
    // be measured server-side, and this is a rough triage number, not a verdict.
    $longTitles = self::countPublished($postTypes, ' AND CHAR_LENGTH(p.post_title) > 60');

    // Nothing to summarise from: no override, no excerpt, and no content to
    // derive one from means the page ships with no description at all.
    $noSummarySource = self::countPublished($postTypes, " AND p.post_excerpt = '' AND CHAR_LENGTH(TRIM(p.post_content)) < 50 AND NOT EXISTS (SELECT 1 FROM %i m WHERE m.post_id = p.ID AND m.meta_key = %s AND m.meta_value != '')", [$wpdb->postmeta, '_rankify_description']);

    return [
      'total' => $total,
      'indexable' => max(0, $total - $noindexed),
      'noindexed' => $noindexed,
      'missingDescription' => $missingDescription,
      'missingTitle' => $missingTitle,
      'missingImage' => $missingImage,
      'longTitles' => $longTitles,
      'noSummarySource' => $noSummarySource,
      'postTypes' => $postTypes,
      'generatedAt' => time(),
    ];
  }

  protected static function emptySnapshot()
  {
    return [
      'total' => 0,
      'indexable' => 0,
      'noindexed' => 0,
      'missingDescription' => 0,
      'missingTitle' => 0,
      'missingImage' => 0,
      'longTitles' => 0,
      'noSummarySource' => 0,
      'postTypes' => [],
      'generatedAt' => time(),
    ];
  }

  /**
   * Count published, indexable posts.
   *
   * The IN list is one %s per public post type, so the SQL cannot be a single
   * literal. Tables go through %i and every value is bound.
   */
  protected static function countPublished(array $postTypes, $extraSql = '', array $extraArgs = [])
  {
    global $wpdb;

    $marks = implode(',', array_fill(0, count($postTypes), '%s'));
    $sql = 'SELECT COUNT(*) FROM %i p WHERE p.post_type IN (' . $marks . ") AND p.post_status = 'publish' AND p.post_password = ''" . $extraSql;

    // phpcs:ignore WordPress.DB.PreparedSQL.NotPrepared, WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, PluginCheck.Security.DirectDB.UnescapedDBParameter -- IN list length follows the public post types. Identifiers use %i and every value is bound. The count has to be live.
    return (int) $wpdb->get_var($wpdb->prepare($sql, array_merge([$wpdb->posts], $postTypes, $extraArgs)));
  }

  protected static function countWithoutMeta(array $postTypes, $metaKey)
  {
    global $wpdb;

    return self::countPublished($postTypes, " AND NOT EXISTS (SELECT 1 FROM %i m WHERE m.post_id = p.ID AND m.meta_key = %s AND m.meta_value != '')", [$wpdb->postmeta, $metaKey]);
  }

  protected static function countWithMeta(array $postTypes, $metaKey, $value)
  {
    global $wpdb;

    return self::countPublished($postTypes, ' AND EXISTS (SELECT 1 FROM %i m WHERE m.post_id = p.ID AND m.meta_key = %s AND m.meta_value = %s)', [$wpdb->postmeta, $metaKey, $value]);
  }

  /**
   * Specific posts worth fixing, newest first, with an edit link.
   *
   * The overview is only useful if it ends somewhere actionable — a count of
   * 47 missing descriptions is a statistic, and a list of five posts with Edit
   * next to them is a task.
   */
  public static function needsAttention($limit = 5)
  {
    global $wpdb;

    $postTypes = self::indexablePostTypes();
    if (empty($postTypes)) {
      return [];
    }

    $marks = implode(',', array_fill(0, count($postTypes), '%s'));
    $sql = 'SELECT p.ID, p.post_title, p.post_type FROM %i p WHERE p.post_type IN (' . $marks . ") AND p.post_status = 'publish' AND p.post_password = '' AND p.post_excerpt = '' AND NOT EXISTS (SELECT 1 FROM %i m WHERE m.post_id = p.ID AND m.meta_key = %s AND m.meta_value != '') ORDER BY p.post_modified DESC LIMIT %d";

    // phpcs:ignore WordPress.DB.PreparedSQL.NotPrepared, WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, PluginCheck.Security.DirectDB.UnescapedDBParameter -- IN list length follows the public post types. Identifiers use %i and every value is bound.
    $rows = $wpdb->get_results($wpdb->prepare($sql, array_merge([$wpdb->posts], $postTypes, [$wpdb->postmeta, '_rankify_description', (int) $limit])));

    $items = [];
    foreach ($rows as $row) {
      $items[] = [
        'id' => (int) $row->ID,
        'title' => $row->post_title ?: __('(no title)', 'crownify'),
        'type' => $row->post_type,
        'editUrl' => get_edit_post_link((int) $row->ID, 'raw'),
        'reason' => __('No meta description. One will be generated from the content', 'crownify'),
      ];
    }

    return $items;
  }

  /**
   * Answer-engine readiness: the things that decide whether an AI assistant can
   * find, read and cite this site.
   */
  public static function aiReadiness()
  {
    $blocked = class_exists('\DaftPlug\Rankify\Features\Robots\AiCrawlers') ? \DaftPlug\Rankify\Features\Robots\AiCrawlers::blocked() : [];
    $all = class_exists('\DaftPlug\Rankify\Features\Robots\AiCrawlers') ? \DaftPlug\Rankify\Features\Robots\AiCrawlers::all() : [];

    // Blocking a citation crawler is the one that actually costs traffic —
    // it removes the site from AI answers rather than just from training data.
    $blockedCitation = [];
    foreach ($blocked as $token) {
      if (!empty($all[$token]['citation'])) {
        $blockedCitation[] = $all[$token]['label'];
      }
    }

    return [
      'llmsTxt' => 'on' === Utils::getSetting('llmsTxt'),
      'schema' => 'on' === Utils::getSetting('schemaMarkup'),
      'totalCrawlers' => count($all),
      'blockedCount' => count($blocked),
      'blockedCitation' => $blockedCitation,
    ];
  }
}