<?php
/**
 * Scanner: queries posts in batches, builds snapshots, runs checks, stores one report.
 * All WordPress queries live here; the checks stay pure.
 */
defined('ABSPATH') || exit;

class Postlint_Scanner {

    const SCAN_META = '_postlint_scanned';

    /** Build one snapshot from a WP_Post + its metadata. */
    public static function snapshot($post, array $duplicate_title_ids = []) {
        $content = (string) $post->post_content;
        $plain = trim(wp_strip_all_tags(strip_shortcodes($content)));
        $tags = get_the_terms($post->ID, 'post_tag');
        $cats = get_the_terms($post->ID, 'category');
        return [
            'id' => (int) $post->ID,
            'title' => (string) $post->post_title,
            'type' => (string) $post->post_type,
            'status' => (string) $post->post_status,
            'date' => (string) $post->post_date,
            'excerpt_set' => trim((string) $post->post_excerpt) !== '',
            'tags_count' => is_array($tags) ? count($tags) : 0,
            'categories_count' => self::real_categories($cats),
            'word_count' => $plain === '' ? 0 : count(preg_split('/\s+/', $plain)),
            'has_thumbnail' => has_post_thumbnail($post->ID),
            'images_missing_alt' => self::images_missing_alt($content),
            'duplicate_title' => isset($duplicate_title_ids[self::title_key((string) $post->post_title)]),
            'supports' => [
                'tags' => is_object_in_taxonomy($post->post_type, 'post_tag'),
                'categories' => is_object_in_taxonomy($post->post_type, 'category'),
                'excerpt' => post_type_supports($post->post_type, 'excerpt'),
                'thumbnail' => post_type_supports($post->post_type, 'thumbnail'),
            ],
        ];
    }

    /** The default category ("Uncategorized") is where WordPress puts a post nobody filed. */
    private static function real_categories($cats) {
        if (!is_array($cats)) {
            return 0;
        }
        $default = (int) get_option('default_category', 0);
        return count(array_filter($cats, static function ($c) use ($default) { return (int) $c->term_id !== $default; }));
    }

    /** Bounded, tolerant <img> alt scan (see docs/TESTLOG.md: PREG_UNMATCHED_AS_NULL). */
    public static function images_missing_alt($html, $limit = 200) {
        if ($html === '' || !preg_match_all('/<img\b[^>]*>/i', $html, $m)) {
            return 0;
        }
        $missing = 0;
        foreach (array_slice($m[0], 0, $limit) as $tag) {
            if (preg_match('/\balt\s*=\s*("([^"]*)"|\'([^\']*)\'|([^\s>]+))/i', $tag, $a, PREG_UNMATCHED_AS_NULL)) {
                $value = $a[2] ?? $a[3] ?? $a[4] ?? '';
                if (trim($value) === '') {
                    $missing++;
                }
            } else {
                $missing++;
            }
        }
        return $missing;
    }

    /** One key for "the same title": case-insensitive, and any run of spaces, tabs or no-break spaces counts as one. */
    public static function title_key(string $title): string {
        return mb_strtolower(trim((string) preg_replace('/[\s\x{00A0}]+/u', ' ', $title)));
    }

    /** Title-id map for the duplicate-title check (single pass, O(n)). */
    public static function duplicate_title_map(array $posts) {
        $seen = [];
        foreach ($posts as $p) {
            $key = self::title_key((string) $p->post_title);
            if ($key === '') {
                continue;
            }
            $seen[$key] = ($seen[$key] ?? 0) + 1;
        }
        $dups = [];
        foreach ($seen as $key => $n) {
            if ($n > 1) {
                $dups[$key] = true;
            }
        }
        return $dups;
    }

    /**
     * Scan one batch (offset-based). Returns the number of posts scanned in this batch;
     * 0 means done. The report option is replaced on the first batch and appended after.
     */
    public static function scan_batch(array $settings, $offset, $store = null) {
        $store = $store ?: new Postlint_Store();
        $types = $settings['post_types'];
        $batch = max(1, (int) $settings['batch_size']);
        $posts = $store->published_posts($types, $batch, $offset);
        if (empty($posts)) {
            if ($offset > 0) {
                $report = self::get_report($store);
                $report['complete'] = true;
                $store->write_report($report);
            }
            return 0;
        }
        // Duplicates are judged against every published title in scope, not just this batch,
        // or two posts a batch apart would never meet.
        $dup_map = method_exists($store, 'published_titles')
            ? self::duplicate_title_map(array_map(static function ($t) { return (object) ['post_title' => (string) $t]; }, $store->published_titles($types)))
            : self::duplicate_title_map($posts);
        $report = $offset === 0 ? self::empty_report() : self::get_report($store);
        foreach ($posts as $post) {
            $snapshot = self::snapshot($post, $dup_map);
            $issues = Postlint_Checks::evaluate($snapshot, $settings);
            if ($issues) {
                $report['issues'][] = ['id' => $snapshot['id'], 'title' => $snapshot['title'], 'type' => $snapshot['type'], 'checks' => $issues];
            }
            foreach ($issues as $code) {
                $report['totals'][$code] = ($report['totals'][$code] ?? 0) + 1;
            }
            $store->mark_scanned($post->ID, $store->timestamp());
        }
        $report['scanned'] = ($report['scanned'] ?? 0) + count($posts);
        $report['complete'] = count($posts) < $batch;
        $report['scanned_at'] = $store->now();
        $store->write_report($report);
        return count($posts);
    }

    public static function get_report($store = null) {
        $store = $store ?: new Postlint_Store();
        return array_merge(self::empty_report(), $store->read_report());
    }

    public static function empty_report() {
        $totals = array_fill_keys(array_keys(Postlint_Checks::all_checks()), 0);
        return ['scanned_at' => '', 'scanned' => 0, 'complete' => false, 'totals' => $totals, 'issues' => []];
    }
}