<?php
/**
 * Automatic generation: the reason the plugin needs no effort. Everything here works from raw strings so the
 * editor preview, the audit, the REST API and the front end all produce identical results.
 */

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

class SXSEO_Auto {

	/** Meta description: the excerpt if there is one, otherwise the opening of the content, cut at a sentence. */
	public static function description( $content, $excerpt = '' ) {
		$max  = (int) sxseo_opt( 'description_len', 155 );
		$text = sxseo_plain( $excerpt );
		if ( sxseo_len( $text ) < 50 ) {
			$text = sxseo_plain( $content );
		}
		// Drop leading boilerplate like "Posted on ..." and bare URLs.
		$text = preg_replace( '#https?://\S+#i', '', $text );
		$text = trim( preg_replace( '/\s+/u', ' ', $text ) );
		return sxseo_trim( $text, $max );
	}

	/** Focus keyphrase guess from the page title: the title without filler words, at most four words. */
	public static function focus( $title ) {
		$title = preg_replace( '/\s[\-–—|:]\s.*$/u', '', (string) $title ); // drop "- Site name" style tails
		$stop  = sxseo_stopwords();
		$keep  = array();
		foreach ( sxseo_words( $title ) as $w ) {
			if ( ! isset( $stop[ $w ] ) && sxseo_len( $w ) > 1 ) {
				$keep[] = $w;
			}
		}
		if ( ! $keep ) { // a title made only of filler words ("About", "Contact us") is still the best guess
			$keep = array_slice( sxseo_words( $title ), 0, 4 );
		}
		return implode( ' ', array_slice( $keep, 0, 4 ) );
	}

	/**
	 * Keyword list for the legacy keywords meta (and hashtags / schema): focus phrase, tags, categories, then the
	 * most frequent meaningful words of the page. Short, unique, lower-case. Search engines that still read the tag
	 * get relevant terms; none of it is stuffed.
	 */
	public static function keywords( $title, $content, array $terms = array(), $focus = '', $limit = 10 ) {
		$out = array();
		$add = function ( $k ) use ( &$out ) {
			$k = trim( sxseo_lower( wp_strip_all_tags( (string) $k ) ) );
			if ( '' !== $k && ! in_array( $k, $out, true ) ) {
				$out[] = $k;
			}
		};
		if ( '' !== $focus ) {
			$add( $focus );
		}
		foreach ( $terms as $t ) {
			$add( $t );
		}
		$stop  = sxseo_stopwords();
		$freq  = array();
		$words = sxseo_words( $title . ' ' . $title . ' ' . sxseo_plain( $content ) );
		foreach ( $words as $w ) {
			if ( isset( $stop[ $w ] ) || sxseo_len( $w ) < 4 || is_numeric( $w ) ) {
				continue;
			}
			$freq[ $w ] = ( $freq[ $w ] ?? 0 ) + 1;
		}
		arsort( $freq );
		foreach ( array_keys( $freq ) as $w ) {
			if ( count( $out ) >= $limit ) {
				break;
			}
			if ( $freq[ $w ] >= 2 || count( $out ) < 3 ) {
				$add( $w );
			}
		}
		return array_slice( $out, 0, $limit );
	}

	/** First image src in the content (for the social image fallback). */
	public static function first_image( $content ) {
		if ( preg_match( '/<img[^>]+src=["\']([^"\']+)["\']/i', (string) $content, $m ) ) {
			$src = $m[1];
			if ( 0 === strpos( $src, '//' ) ) {
				return set_url_scheme( $src );
			}
			return 0 === strpos( $src, '/' ) ? home_url( $src ) : $src;
		}
		return '';
	}

	/** Fill %vars% in a title template; unknown vars vanish, repeated separators collapse. */
	public static function render_template( $tpl, array $vars ) {
		$vars += array(
			'sitename' => get_bloginfo( 'name' ),
			'tagline'  => get_bloginfo( 'description' ),
			'sep'      => sxseo_opt( 'sep', '–' ),
		);
		$out = preg_replace_callback(
			'/%%?([a-z_]+)%%?/i',
			function ( $m ) use ( $vars ) {
				$k = sxseo_lower( $m[1] );
				return isset( $vars[ $k ] ) ? (string) $vars[ $k ] : '';
			},
			(string) $tpl
		);
		$sep = preg_quote( (string) $vars['sep'], '/' );
		$out = preg_replace( '/\s+/u', ' ', $out );
		$out = preg_replace( '/(\s*' . $sep . '\s*){2,}/u', ' ' . $vars['sep'] . ' ', $out );
		$out = trim( $out, " \t\n\r\0\x0B" );
		$out = preg_replace( '/^' . $sep . '\s*|\s*' . $sep . '$/u', '', $out );
		return trim( $out );
	}
}