<?php
/**
 * THE LOCAL CHECK: pages whose title or description does not work in Google.
 *
 * Why it exists: the plugin used to be only a receiver for paying customers.
 * A plugin that does nothing without an account is neither approved in the
 * directory nor installed by anyone. This part stands on its own, needs no
 * account and sends nothing anywhere.
 *
 * It changes nothing. It reads the posts already in the database and counts.
 * No external connection, no data leaves the site.
 *
 * Source strings are English (directory convention); Swedish lives in
 * languages/tedrix-sv_SE.po.
 *
 * @package Tedrix
 */

defined( 'ABSPATH' ) || exit;

/** Google cuts titles around 60 characters and descriptions around 155. */
const TEDRIX_TITEL_MAX = 60;
const TEDRIX_BESKR_MAX = 155;
const TEDRIX_BESKR_MIN = 70;

/**
 * Walks published posts and pages and collects the problems that show
 * directly in search results. Cached for an hour so large sites are not
 * scanned every time someone opens the admin page.
 *
 * @param bool $tvinga Skip the cache.
 * @return array{rader: array, antal: int, granskade: int}
 */
function tedrix_kollen( $tvinga = false ) {
	$cache = get_transient( 'tedrix_kollen_v3' );
	if ( ! $tvinga && is_array( $cache ) ) {
		return $cache;
	}

	$q = new WP_Query(
		array(
			'post_type'      => array( 'post', 'page' ),
			'post_status'    => 'publish',
			'posts_per_page' => 300, // cap protects large sites from a heavy run
			'orderby'        => 'modified',
			'order'          => 'DESC',
			'no_found_rows'  => true,
		)
	);

	$rader = array();
	$poster = array_values( array_filter( $q->posts, function ( $p ) { return ! tedrix_ar_noindex( $p ); } ) );

	foreach ( $poster as $post ) {
		$titel   = tedrix_las_seo_titel( $post );
		$beskr   = tedrix_las_seo_beskrivning( $post );
		$problem = array();

		if ( mb_strlen( $titel ) > TEDRIX_TITEL_MAX ) {
			$problem[] = array( 'titel_lang', mb_strlen( $titel ) );
		}
		if ( '' === trim( $beskr ) ) {
			$problem[] = array( 'beskr_saknas', 0 );
		} elseif ( mb_strlen( $beskr ) > TEDRIX_BESKR_MAX ) {
			$problem[] = array( 'beskr_lang', mb_strlen( $beskr ) );
		} elseif ( mb_strlen( $beskr ) < TEDRIX_BESKR_MIN ) {
			$problem[] = array( 'beskr_kort', mb_strlen( $beskr ) );
		}

		if ( $problem ) {
			$rader[] = array(
				'id'      => $post->ID,
				'titel'   => $titel,
				'url'     => get_permalink( $post ),
				'problem' => $problem,
			);
		}
	}

	/* Duplicates: two pages with the same title compete with each other, and
	 * that is invisible when looking at one page at a time. */
	$sedda = array();
	foreach ( $poster as $post ) {
		$t = mb_strtolower( tedrix_las_seo_titel( $post ) );
		if ( '' === $t ) {
			continue;
		}
		$sedda[ $t ][] = $post->ID;
	}
	foreach ( $sedda as $ids ) {
		if ( count( $ids ) < 2 ) {
			continue;
		}
		foreach ( $ids as $id ) {
			$i    = tedrix_rad_index( $rader, $id );
			$text = array( 'dubblett', count( $ids ) - 1 );
			if ( null === $i ) {
				$rader[] = array(
					'id'      => $id,
					'titel'   => tedrix_las_seo_titel( get_post( $id ) ),
					'url'     => get_permalink( $id ),
					'problem' => array( $text ),
				);
			} else {
				$rader[ $i ]['problem'][] = $text;
			}
		}
	}

	$ut = array(
		'rader'     => $rader,
		'antal'     => count( $rader ),
		'granskade' => count( $poster ),
	);
	set_transient( 'tedrix_kollen_v3', $ut, HOUR_IN_SECONDS );
	return $ut;
}

/**
 * A problem as text, in the viewer's language. The cache stores only the
 * kind and the number, so switching language never shows stale text.
 *
 * @param array $p array( kind, number ).
 * @return string
 */
function tedrix_problem_text( $p ) {
	list( $typ, $n ) = $p;
	switch ( $typ ) {
		case 'titel_lang':
			/* translators: %d: number of characters */
			return sprintf( __( 'Title %d characters, cut off in Google.', 'tedrix' ), $n );
		case 'beskr_saknas':
			return __( 'No description, so Google writes its own.', 'tedrix' );
		case 'beskr_lang':
			/* translators: %d: number of characters */
			return sprintf( __( 'Description %d characters, cut off in Google.', 'tedrix' ), $n );
		case 'beskr_kort':
			/* translators: %d: number of characters */
			return sprintf( __( 'Description only %d characters.', 'tedrix' ), $n );
		case 'dubblett':
			/* translators: %d: number of other pages with the same title */
			return sprintf( _n( 'Same title as %d other page.', 'Same title as %d other pages.', $n, 'tedrix' ), $n );
	}
	return '';
}

/**
 * Index of the row for a post, or null.
 *
 * @param array $rader Rows.
 * @param int   $id    Post ID.
 * @return int|null
 */
function tedrix_rad_index( array $rader, $id ) {
	foreach ( $rader as $i => $r ) {
		if ( (int) $r['id'] === (int) $id ) {
			return $i;
		}
	}
	return null;
}

/**
 * Plain text the way Google sees it: no tags, and entities such as &#8217;
 * (which get_the_title() adds for curly quotes) turned back into characters
 * so they count as one character, not seven.
 *
 * @param string $s Text.
 * @return string
 */
function tedrix_ren( $s ) {
	return trim( html_entity_decode( wp_strip_all_tags( (string) $s ), ENT_QUOTES, 'UTF-8' ) );
}

/**
 * SEO plugins let a title or description contain variables such as
 * %%title%% (Yoast), %title% (Rank Math) or #post_title (All in One SEO).
 * Fill in the common ones and drop the rest, so the length is close to what
 * is printed instead of counting the variable names.
 *
 * @param string  $v    Stored value.
 * @param WP_Post $post Post.
 * @return string
 */
function tedrix_fyll_variabler( $v, $post ) {
	if ( false === strpos( $v, '%' ) && false === strpos( $v, '#' ) ) {
		return $v;
	}
	$titel   = tedrix_ren( $post->post_title );
	$sajt    = tedrix_ren( get_bloginfo( 'name' ) );
	$utdrag  = tedrix_ren( has_excerpt( $post ) ? $post->post_excerpt : wp_trim_words( $post->post_content, 25, '' ) );
	$ersatt  = array(
		'%%title%%'          => $titel,
		'%%sitename%%'       => $sajt,
		'%%sep%%'            => '-',
		'%%excerpt%%'        => $utdrag,
		'%%post_title%%'     => $titel,
		'%%sitetitle%%'      => $sajt,
		'%title%'            => $titel,
		'%sitename%'         => $sajt,
		'%sep%'              => '-',
		'%excerpt%'          => $utdrag,
		'#post_title'        => $titel,
		'#site_title'        => $sajt,
		'#separator_sa'      => '-',
		'#post_excerpt'      => $utdrag,
	);
	$v = strtr( $v, $ersatt );
	$v = preg_replace( '/%%[a-z0-9_]+%%|%[a-z0-9_]+%|#(?:post|site|separator|tagline|category|author|current|page|permalink|taxonomy)[a-z0-9_]*/i', '', $v );
	return trim( preg_replace( '/\s{2,}/', ' ', $v ) );
}

/**
 * The All in One SEO row for a post (title, description, noindex), read from
 * its own table. Cached per request. Empty array when AIOSEO is not in use.
 *
 * @param int $id Post ID.
 * @return array
 */
function tedrix_aioseo_rad( $id ) {
	static $tabell = null, $rader = array();
	global $wpdb;
	if ( null === $tabell ) {
		$tabell = '';
		if ( tedrix_active_seo_plugins() && in_array( 'aioseo', tedrix_active_seo_plugins(), true ) ) {
			// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching -- one lookup per request, result kept in a static.
			$finns  = $wpdb->get_var( $wpdb->prepare( 'SHOW TABLES LIKE %s', $wpdb->esc_like( $wpdb->prefix . 'aioseo_posts' ) ) );
			$tabell = ( $wpdb->prefix . 'aioseo_posts' === $finns ) ? $finns : '';
		}
	}
	if ( '' === $tabell ) {
		return array();
	}
	$id = (int) $id;
	if ( ! array_key_exists( $id, $rader ) ) {
		// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching,WordPress.DB.PreparedSQL.InterpolatedNotPrepared,PluginCheck.Security.DirectDB.UnescapedDBParameter -- AIOSEO keeps its fields in its own table; the name is the site prefix plus a fixed string, checked above. Kept in a static per request.
		$rad          = $wpdb->get_row( $wpdb->prepare( "SELECT title, description, robots_noindex FROM {$tabell} WHERE post_id = %d", $id ), ARRAY_A );
		$rader[ $id ] = is_array( $rad ) ? $rad : array();
	}
	return $rader[ $id ];
}

/**
 * Whether an SEO plugin tells search engines to skip this page. Such pages
 * never show in Google, so their titles do not matter.
 *
 * @param WP_Post $post Post.
 * @return bool
 */
function tedrix_ar_noindex( $post ) {
	if ( '1' === (string) get_post_meta( $post->ID, '_yoast_wpseo_meta-robots-noindex', true ) ) {
		return true;
	}
	$rm = get_post_meta( $post->ID, 'rank_math_robots', true );
	if ( is_array( $rm ) && in_array( 'noindex', $rm, true ) ) {
		return true;
	}
	if ( 'yes' === (string) get_post_meta( $post->ID, '_seopress_robots_index', true ) ) {
		return true;
	}
	if ( '1' === (string) get_post_meta( $post->ID, '_genesis_noindex', true ) ) {
		return true;
	}
	$aio = tedrix_aioseo_rad( $post->ID );
	return ! empty( $aio['robots_noindex'] );
}

/**
 * The title that actually reaches Google: the SEO plugin's field if set,
 * otherwise the post title.
 *
 * @param WP_Post $post Post.
 * @return string
 */
function tedrix_las_seo_titel( $post ) {
	foreach ( tedrix_seo_meta_keys() as $keys ) {
		if ( empty( $keys['title'] ) ) {
			continue;
		}
		$v = get_post_meta( $post->ID, $keys['title'], true );
		if ( is_string( $v ) && '' !== trim( $v ) ) {
			return tedrix_ren( tedrix_fyll_variabler( $v, $post ) );
		}
	}
	$aio = tedrix_aioseo_rad( $post->ID );
	if ( ! empty( $aio['title'] ) ) {
		$v = tedrix_ren( tedrix_fyll_variabler( (string) $aio['title'], $post ) );
		if ( '' !== $v ) {
			return $v;
		}
	}
	return tedrix_ren( get_the_title( $post ) );
}

/**
 * The description that reaches Google, from whichever SEO plugin is
 * installed. Empty string means none is set.
 *
 * @param WP_Post $post Post.
 * @return string
 */
function tedrix_las_seo_beskrivning( $post ) {
	foreach ( tedrix_seo_meta_keys() as $keys ) {
		if ( empty( $keys['desc'] ) ) {
			continue;
		}
		$v = get_post_meta( $post->ID, $keys['desc'], true );
		if ( is_string( $v ) && '' !== trim( $v ) ) {
			return tedrix_ren( tedrix_fyll_variabler( $v, $post ) );
		}
	}
	$aio = tedrix_aioseo_rad( $post->ID );
	if ( ! empty( $aio['description'] ) ) {
		return tedrix_ren( tedrix_fyll_variabler( (string) $aio['description'], $post ) );
	}
	return '';
}

/* Clear the cache whenever something is published or edited. */
add_action(
	'save_post',
	function () {
		delete_transient( 'tedrix_kollen_v3' );
	}
);

/**
 * Renders the titles and descriptions section of the admin page.
 */
function tedrix_kollen_sektion() {
	$tvinga = isset( $_GET['tedrix_om'] ) && check_admin_referer( 'tedrix_om' );
	$r      = tedrix_kollen( $tvinga );
	?>
	<div class="tdx-kortrubrik">
		<h2><?php esc_html_e( 'Titles and descriptions', 'tedrix' ); ?></h2>
		<?php if ( $r['antal'] ) : ?>
			<a class="tdx-btn" href="<?php echo esc_url( tedrix_fixa_url( 'fix-all-pages' ) ); ?>">&#10022; <?php echo esc_html( tedrix_fixa_alla_text( (int) $r['antal'] ) ); ?></a>
		<?php endif; ?>
	</div>
	<p>
		<?php
		if ( $r['granskade'] ) {
			printf(
				/* translators: 1: pages with problems, 2: pages checked */
				esc_html( _n( '%1$d of %2$d pages shows up wrong in Google. Read from your own database, nothing is sent.', '%1$d of %2$d pages show up wrong in Google. Read from your own database, nothing is sent.', (int) $r['antal'], 'tedrix' ) ),
			(int) $r['antal'],
			(int) $r['granskade']
			);
		} else {
			esc_html_e( 'Read from your own database, nothing is sent.', 'tedrix' );
		}
		?>
	</p>

	<?php if ( ! $r['granskade'] ) : ?>
		<div class="tdx-larm tdx-gul"><p>
			<?php esc_html_e( 'There are no published posts or pages to check yet.', 'tedrix' ); ?>
		</p></div>
	<?php elseif ( ! $r['antal'] ) : ?>
		<div class="tdx-larm tdx-gron"><p>
			<?php esc_html_e( 'Nothing to fix right now. Titles and descriptions look good on the pages checked.', 'tedrix' ); ?>
		</p></div>
	<?php else : ?>
		<table class="tdx-tabell">
			<thead><tr>
				<th><?php esc_html_e( 'Page', 'tedrix' ); ?></th>
				<th><?php esc_html_e( 'What is wrong', 'tedrix' ); ?></th>
				<th></th>
			</tr></thead>
			<tbody>
			<?php foreach ( $r['rader'] as $rad ) : ?>
				<tr>
					<td>
						<strong><?php echo esc_html( wp_trim_words( $rad['titel'], 12 ) ); ?></strong>
						<a class="tdx-sokvag" href="<?php echo esc_url( $rad['url'] ); ?>" target="_blank" rel="noopener"><?php echo esc_html( wp_make_link_relative( $rad['url'] ) ); ?></a>
					</td>
					<td><?php echo wp_kses( implode( '<br>', array_map( 'esc_html', array_map( 'tedrix_problem_text', $rad['problem'] ) ) ), array( 'br' => array() ) ); ?></td>
					<td class="tdx-knappar">
						<a class="tdx-btn-fixa" href="<?php echo esc_url( tedrix_fixa_url( 'fix-page' ) ); ?>"><?php esc_html_e( 'Fix', 'tedrix' ); ?></a>
						<a class="tdx-btn-2" href="<?php echo esc_url( (string) get_edit_post_link( $rad['id'] ) ); ?>"><?php esc_html_e( 'Edit', 'tedrix' ); ?></a>
					</td>
				</tr>
			<?php endforeach; ?>
			</tbody>
		</table>
	<?php endif; ?>

	<div class="tdx-rad">
		<a class="tdx-btn-2" href="<?php echo esc_url( wp_nonce_url( admin_url( 'tools.php?page=tedrix-kollen&tedrix_om=1' ), 'tedrix_om' ) ); ?>">
			<?php esc_html_e( 'Run again', 'tedrix' ); ?>
		</a>
	</div>
	<?php
}