<?php
defined( 'ABSPATH' ) || exit;

/**
 * Content Decay Monitor - a feature RankMath doesn't have.
 *
 * Flags posts that are:
 * • More than 12 months old and haven't been updated in 6+ months.
 * • Heavily focused on time-sensitive topics (news, tutorials, comparisons).
 * • Missing date-modified schema.
 *
 * In a full implementation this would integrate with Google Search Console
 * API to pull actual click/impression decay curves. This version gives
 * rule-based decay scores that work without an API key.
 */
class AWBSEO_Content_Decay {

    const DECAY_OPTION = 'awbseo_decay_scores';

    public static function get_decaying_posts( $limit = 50 ) {
        $posts = get_posts( [
            'post_type'      => 'post',
            'post_status'    => 'publish',
            'posts_per_page' => $limit,
            'orderby'        => 'date',
            'order'          => 'ASC', // oldest first
        ] );

        $results = [];
        foreach ( $posts as $post ) {
            $score = self::decay_score( $post );
            if ( $score['risk'] !== 'low' ) {
                $results[] = array_merge( [ 'post' => $post ], $score );
            }
        }

        usort( $results, fn( $a, $b ) => $b['score'] <=> $a['score'] );
        return $results;
    }

    public static function decay_score( $post ) {
        $score    = 0;
        $reasons  = [];
        $now      = time();
        $created  = strtotime( $post->post_date );
        $modified = strtotime( $post->post_modified );

        $age_days      = ( $now - $created ) / DAY_IN_SECONDS;
        $modified_days = ( $now - $modified ) / DAY_IN_SECONDS;

        // Post is older than 12 months.
        if ( $age_days > 365 ) {
            $score += 20;
            /* translators: %.1f = number of years (e.g. 2.3) */
            $reasons[] = sprintf( __( 'Post is over 1 year old (%.1f years).', 'adworksbid-seo' ), round( $age_days / 365, 1 ) );
        }

        // Not updated in 6+ months.
        if ( $modified_days > 180 ) {
            $score += 25;
            /* translators: %d = number of months */
            $reasons[] = sprintf( __( 'Not updated in %d months.', 'adworksbid-seo' ), round( $modified_days / 30 ) );
        } elseif ( $modified_days > 90 ) {
            $score += 10;
            /* translators: %d = number of months */
            $reasons[] = sprintf( __( 'Not updated in %d months.', 'adworksbid-seo' ), round( $modified_days / 30 ) );
        }

        // Time-sensitive keywords in title.
        $time_keywords = [ 'best', 'top', 'review', 'vs', 'versus', 'comparison', 'guide', 'how to', 'tutorial', 'checklist', 'list', 'tips', 'tricks', 'tools', 'plugins', 'apps' ];
        $title_lower   = mb_strtolower( $post->post_title );
        foreach ( $time_keywords as $kw ) {
            if ( false !== mb_strpos( $title_lower, $kw ) ) {
                $score += 10;
                /* translators: %s = the time-sensitive keyword found in the title */
                $reasons[] = sprintf( __( "Title contains '%s' - time-sensitive topic.", 'adworksbid-seo' ), $kw );
                break;
            }
        }

        // Year in title (e.g., "Best Plugins 2022").
        $title_year = self::old_year_in( $post->post_title );
        if ( $title_year ) {
            $score += 30;
            /* translators: %s = the year found in the title (e.g. 2022) */
            $reasons[] = sprintf( __( 'Title contains year %s - almost certainly outdated.', 'adworksbid-seo' ), $title_year );
        }

        // Short content = thin/low-effort = higher decay risk.
        $word_count = str_word_count( wp_strip_all_tags( $post->post_content ) );
        if ( $word_count < 500 ) {
            $score += 15;
            /* translators: %d = word count of the post */
            $reasons[] = sprintf( __( 'Content is only %d words - thin posts decay faster.', 'adworksbid-seo' ), $word_count );
        }

        // No focus keyword = not optimised = higher risk.
        if ( ! AWBSEO_Meta_Manager::get( $post->ID, 'focus_keyword' ) ) {
            $score += 5;
            $reasons[] = __( 'No focus keyword set.', 'adworksbid-seo' );
        }

        $risk = 'low';
        if ( $score >= 50 ) $risk = 'high';
        elseif ( $score >= 25 ) $risk = 'medium';

        return [
            'score'   => $score,
            'risk'    => $risk,
            'reasons' => $reasons,
            'age_days'      => round( $age_days ),
            'modified_days' => round( $modified_days ),
            'word_count'    => $word_count,
        ];
    }

    /**
     * First year in $text that is before the current year (2000 or later),
     * or '' if none. Replaces a hardcoded 2010-2023 range that stopped
     * catching titles like "Best Tools 2025" once the calendar moved on.
     */
    public static function old_year_in( $text ) {
        $current = (int) date_i18n( 'Y' );
        if ( preg_match_all( '/\b(20\d\d)\b/', (string) $text, $m ) ) {
            foreach ( $m[1] as $y ) {
                if ( (int) $y >= 2005 && (int) $y < $current ) {
                    return $y;
                }
            }
        }
        return '';
    }

    /** Short snippet of $text around byte offset $pos, for showing where something was found. */
    private static function snippet( $text, $pos, $len ) {
        $start = max( 0, $pos - 40 );
        $chunk = substr( $text, $start, $len + 80 );
        $chunk = wp_check_invalid_utf8( $chunk, true );
        return trim( ( $start > 0 ? '…' : '' ) . preg_replace( '/\s+/', ' ', $chunk ) . '…' );
    }

    /**
     * Refresh checklist for a post. Every item is worked out for this
     * specific post: 'done' => false shows a gray check (still needs doing,
     * listed first), 'done' => true a green check (already fine). 'details'
     * says exactly what was found - the outdated years/prices in the text,
     * the broken URLs, the word count gap, related posts to link to - so
     * the box tells you what to change, not just what kind of thing to check.
     *
     * @return array[] Each item: [ 'text' => string, 'done' => bool, 'details' => array[] ]
     *                 where each detail is [ 'text' => string, 'url' => string (optional) ].
     */
    public static function refresh_suggestions( $post_id ) {
        global $wpdb;

        $post     = get_post( $post_id );
        $score    = self::decay_score( $post );
        $recent   = $score['modified_days'] <= 90;
        $content  = (string) $post->post_content;
        $plain    = trim( preg_replace( '/\s+/', ' ', wp_strip_all_tags( strip_shortcodes( $content ) ) ) );
        $current  = date_i18n( 'Y' );
        $months   = max( 1, (int) round( $score['modified_days'] / 30 ) );
        $items    = [];

        // ── 1. Outdated facts: old years, prices, percentages in the text ──
        $facts = [];
        if ( preg_match_all( '/\b(20\d\d)\b/', $plain, $ym, PREG_OFFSET_CAPTURE ) ) {
            foreach ( $ym[1] as $hit ) {
                if ( (int) $hit[0] >= 2005 && (int) $hit[0] < (int) $current && count( $facts ) < 3 ) {
                    $facts[ $hit[1] ] = [ 'text' => sprintf( __( 'Mentions %1$s: "%2$s"', 'adworksbid-seo' ), $hit[0], self::snippet( $plain, $hit[1], 4 ) ) ];
                }
            }
        }
        if ( count( $facts ) < 3 && preg_match_all( '/(?:[$€£]\s?\d[\d,.]*|\b\d[\d,.]*\s?(?:USD|EUR|AED|SAR|LBP|KWD|QAR|%))/u', $plain, $pm, PREG_OFFSET_CAPTURE ) ) {
            foreach ( $pm[0] as $hit ) {
                if ( count( $facts ) >= 3 ) break;
                $near = false;
                foreach ( array_keys( $facts ) as $off ) {
                    if ( abs( $off - $hit[1] ) < 80 ) { $near = true; break; }
                }
                if ( ! $near ) {
                    $facts[ $hit[1] ] = [ 'text' => sprintf( __( 'Figure %1$s: "%2$s"', 'adworksbid-seo' ), $hit[0], self::snippet( $plain, $hit[1], strlen( $hit[0] ) ) ) ];
                }
            }
        }
        ksort( $facts );
        $items[] = [
            'text'    => $facts
                ? __( 'Double-check these dates and figures - they may be out of date:', 'adworksbid-seo' )
                : __( 'Review and update all statistics, prices, and data.', 'adworksbid-seo' ),
            'done'    => $recent && ! $facts,
            'details' => $facts
                ? array_values( $facts )
                : ( $recent ? [] : [ [ 'text' => sprintf( _n( 'No dated numbers detected, but the post hasn\'t been touched in %d month - skim it for anything that has changed.', 'No dated numbers detected, but the post hasn\'t been touched in %d months - skim it for anything that has changed.', $months, 'adworksbid-seo' ), $months ) ] ] ),
        ];

        // ── 2. Broken outbound links: list the actual URLs ──
        $broken = [];
        $table  = $wpdb->prefix . 'awb_broken_links';
        $has_table = class_exists( 'AWBSEO_Broken_Links' ) && $wpdb->get_var( $wpdb->prepare( 'SHOW TABLES LIKE %s', $table ) ) === $table;
        $checked   = false;
        if ( $has_table ) {
            $broken_sql = AWBSEO_Broken_Links::broken_sql( 'status_code' );
            $broken = $wpdb->get_results( $wpdb->prepare(
                "SELECT url, status_code FROM {$table} WHERE post_id = %d AND {$broken_sql} AND fix_status != 'fixed' ORDER BY id LIMIT 5",
                $post_id
            ) );
            $checked = (bool) $wpdb->get_var( $wpdb->prepare( "SELECT COUNT(*) FROM {$table} WHERE post_id = %d", $post_id ) );
        }
        $broken_details = [];
        foreach ( (array) $broken as $b ) {
            $broken_details[] = [
                'text' => sprintf(
                    /* translators: 1: URL, 2: problem description */
                    __( '%1$s (%2$s)', 'adworksbid-seo' ),
                    $b->url,
                    404 === (int) $b->status_code ? __( 'page not found', 'adworksbid-seo' ) : __( 'no response', 'adworksbid-seo' )
                ),
                'url'  => $b->url,
            ];
        }
        if ( ! $broken && ! $checked ) {
            $broken_details[] = [ 'text' => __( 'This post\'s links haven\'t been checked yet. Run SEO Analysis on the Dashboard to check them.', 'adworksbid-seo' ) ];
        }
        $items[] = [
            'text'    => $broken
                /* translators: %d = number of broken links in this post */
                ? sprintf( _n( 'Fix or remove %d broken link:', 'Fix or remove %d broken links:', count( $broken ), 'adworksbid-seo' ), count( $broken ) )
                : __( 'No broken links found in this post.', 'adworksbid-seo' ),
            'done'    => ! $broken && $checked,
            'details' => $broken_details,
        ];

        // ── 3. Outdated year in the title: show the suggested new title ──
        $title_year = self::old_year_in( $post->post_title );
        $items[] = [
            'text'    => $title_year
                /* translators: 1: old year, 2: current year */
                ? sprintf( __( 'Change %1$s to %2$s in the title (only if you actually refresh the content):', 'adworksbid-seo' ), $title_year, $current )
                : __( 'Title has no outdated year.', 'adworksbid-seo' ),
            'done'    => ! $title_year,
            'details' => $title_year
                ? [ [ 'text' => '"' . preg_replace( '/\b' . preg_quote( $title_year, '/' ) . '\b/', $current, $post->post_title, 1 ) . '"' ] ]
                : [],
        ];

        // ── 4. Internal links: count + suggest newer related posts not already linked ──
        preg_match_all( '/<a[^>]+href=["\']' . preg_quote( home_url(), '/' ) . '[^"\']*["\'][^>]*>/i', $content, $int_links );
        $int_count = count( $int_links[0] );
        $related_details = [];
        if ( $int_count < 2 ) {
            $related_args = [
                'post_type'      => $post->post_type,
                'post_status'    => 'publish',
                'posts_per_page' => 8,
                'post__not_in'   => [ $post_id ],
                'orderby'        => 'date',
                'order'          => 'DESC',
                'no_found_rows'  => true,
            ];
            $cats = 'post' === $post->post_type ? wp_get_post_categories( $post_id ) : [];
            if ( $cats ) {
                $related_args['category__in'] = $cats;
            }
            foreach ( get_posts( $related_args ) as $rel ) {
                if ( count( $related_details ) >= 3 ) break;
                $link = get_permalink( $rel );
                if ( false === strpos( $content, untrailingslashit( $link ) ) ) {
                    $related_details[] = [ 'text' => sprintf( __( 'Link to: %s', 'adworksbid-seo' ), get_the_title( $rel ) ), 'url' => $link ];
                }
            }
        }
        $items[] = [
            'text'    => $int_count >= 2
                /* translators: %d = number of internal links */
                ? sprintf( _n( 'Has %d internal link.', 'Has %d internal links.', $int_count, 'adworksbid-seo' ), $int_count )
                /* translators: %d = number of internal links */
                : sprintf( _n( 'Only %d internal link - add at least 2. Newer related posts you could link to:', 'Only %d internal links - add at least 2. Newer related posts you could link to:', $int_count, 'adworksbid-seo' ), $int_count ),
            'done'    => $int_count >= 2,
            'details' => $related_details,
        ];

        // ── 5. Content length: exact gap ──
        $wc = (int) $score['word_count'];
        $items[] = [
            'text'    => $wc >= 500
                /* translators: %s = word count */
                ? sprintf( __( '%s words - long enough.', 'adworksbid-seo' ), number_format_i18n( $wc ) )
                /* translators: 1: current word count, 2: words to add */
                : sprintf( __( 'Only %1$s words - add about %2$s more (aim for 1,000+) with new examples, FAQs, or updated information.', 'adworksbid-seo' ), number_format_i18n( $wc ), number_format_i18n( 500 - $wc ) ),
            'done'    => $wc >= 500,
            'details' => [],
        ];

        // ── 6. Meta description: show what's there now ──
        $desc     = trim( (string) AWBSEO_Meta_Manager::get( $post_id, 'description' ) );
        $desc_len = mb_strlen( $desc );
        $desc_ok  = $desc && $desc_len >= 120 && $desc_len <= 160;
        if ( ! $desc ) {
            $desc_text = __( 'No meta description set - write one (120-160 characters).', 'adworksbid-seo' );
        } elseif ( ! $desc_ok ) {
            /* translators: %d = number of characters */
            $desc_text = sprintf( __( 'Meta description is %d characters - aim for 120-160:', 'adworksbid-seo' ), $desc_len );
        } elseif ( ! $recent ) {
            $desc_text = __( 'Rewrite the meta description to match the refreshed content. Current one:', 'adworksbid-seo' );
        } else {
            $desc_text = __( 'Meta description looks good.', 'adworksbid-seo' );
        }
        $items[] = [
            'text'    => $desc_text,
            'done'    => $desc_ok && $recent,
            'details' => ( $desc && ! ( $desc_ok && $recent ) ) ? [ [ 'text' => '"' . $desc . '"' ] ] : [],
        ];

        // ── 7. Target keyword ──
        $kw = trim( (string) AWBSEO_Meta_Manager::get( $post_id, 'focus_keyword' ) );
        $items[] = [
            'text'    => $kw
                /* translators: %s = focus keyword */
                ? sprintf( __( 'Target keyword set: "%s".', 'adworksbid-seo' ), $kw )
                : __( 'No target keyword set - pick the search phrase this post should rank for.', 'adworksbid-seo' ),
            'done'    => '' !== $kw,
            'details' => [],
        ];

        // ── 8. Freshness date ──
        $items[] = [
            'text'    => $recent
                ? __( 'Updated recently - the dateModified in its schema is current.', 'adworksbid-seo' )
                /* translators: %d = number of days since last update */
                : sprintf( __( 'Last updated %d days ago. Save the post after editing so Google sees a fresh dateModified (CitedSEO updates it automatically).', 'adworksbid-seo' ), (int) $score['modified_days'] ),
            'done'    => $recent,
            'details' => [],
        ];

        return array_merge(
            array_values( array_filter( $items, fn( $i ) => ! $i['done'] ) ),
            array_values( array_filter( $items, fn( $i ) => $i['done'] ) )
        );
    }
}