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

final class APEARI_Theme_Knowledge
{
    public const VERSION = 1;
    public const PROFILE = 'lite';
    private const INDEX_VERSION = 1;
    private const CACHE = 'apeari_theme_knowledge_v1';

    private static function clean($value, int $depth = 0)
    {
        if ($depth > 7) {
            return null;
        }
        if (is_array($value)) {
            $result = [];
            foreach (array_slice($value, 0, 40, true) as $key => $item) {
                if (in_array($key, ['css', 'src', 'url', 'content', 'fontFace'], true)) {
                    continue;
                }
                $result[$key] = self::clean($item, $depth + 1);
            }
            return $result;
        }
        if (is_string($value)) {
            return sanitize_text_field(mb_substr($value, 0, 300));
        }
        return is_scalar($value) || null === $value ? $value : null;
    }

    private static function files(): array
    {
        $roots = ['theme' => get_stylesheet_directory()];
        if (get_template_directory() !== get_stylesheet_directory()) {
            $roots['parent'] = get_template_directory();
        }
        $files = [];
        $truncated = false;
        $visited = 0;
        foreach ($roots as $origin => $directory) {
            $root = realpath($directory);
            if (!$root || is_link($directory)) {
                continue;
            }
            $walk = static function (string $dir, string $relative = '', int $depth = 0) use (
                &$walk,
                &$files,
                &$truncated,
                &$visited,
                $root,
                $origin,
            ): void {
                if ($depth > 10 || count($files) >= 400 || $visited >= 5000) {
                    $truncated = true;
                    return;
                }
                $names = @scandir($dir);
                $names = is_array($names) ? $names : [];
                usort($names, static function ($a, $b) {
                    $rank = static fn($name) => in_array(
                        $name,
                        ['style.css', 'theme.json', 'functions.php', 'index.php', 'header.php', 'footer.php'],
                        true,
                    )
                        ? 0
                        : (in_array($name, ['templates', 'parts', 'patterns'], true)
                            ? 1
                            : 2);
                    return $rank($a) <=> $rank($b) ?: strcmp($a, $b);
                });
                foreach ($names as $name) {
                    if (++$visited > 5000) {
                        $truncated = true;
                        break;
                    }
                    if ($name[0] === '.' || in_array($name, ['node_modules', 'vendor', 'cache'], true)) {
                        continue;
                    }
                    $path = $dir . DIRECTORY_SEPARATOR . $name;
                    if (is_link($path)) {
                        continue;
                    }
                    $next = $relative . $name;
                    if (is_dir($path)) {
                        $walk($path, $next . '/', $depth + 1);
                        continue;
                    }
                    $ext = strtolower(pathinfo($name, PATHINFO_EXTENSION));
                    if (
                        !in_array($ext, ['css', 'scss', 'sass', 'less', 'json', 'php', 'html', 'htm'], true)
                    ) {
                        continue;
                    }
                    if (count($files) >= 400) {
                        $truncated = true;
                        break;
                    }
                    clearstatcache(true, $path);
                    $stat = @stat($path);
                    if (!$stat || $stat['nlink'] > 1 || $stat['size'] > 262144 || !is_file($path)) {
                        $truncated = true;
                        continue;
                    }
                    $resolved = realpath($path);
                    if (!$resolved || !str_starts_with($resolved, $root . DIRECTORY_SEPARATOR)) {
                        continue;
                    }
                    $files[$origin . '/' . $next] = [
                        'absolute' => $resolved,
                        'origin' => $origin,
                        'path' => $next,
                        'kind' => $ext,
                        'size' => $stat['size'],
                    ];
                }
            };
            $walk($root);
        }
        ksort($files);
        return [$files, $truncated];
    }

    private static function entry(array $file, string $hash): array
    {
        $path = $file['path'];
        $entry = ['path' => $path, 'origin' => $file['origin'], 'kind' => $file['kind'], 'sha256' => $hash];
        $entry['role'] = preg_match(
            '~(?:^|/)(header|footer|front-page|home|single|page|archive|search|404)(?:[.-]|$)~',
            $path,
            $role,
        )
            ? $role[1]
            : (preg_match('~^(?:templates|parts|patterns)/~', $path)
                ? explode('/', $path)[0]
                : (str_ends_with($path, '.json')
                    ? 'settings'
                    : (preg_match('/\.(css|scss|sass|less)$/', $path)
                        ? 'styles'
                        : 'render')));
        $content = @file_get_contents($file['absolute']);
        if (!is_string($content) || str_contains($content, "\0")) {
            return $entry;
        }
        if (in_array($file['kind'], ['css', 'scss', 'sass', 'less'], true)) {
            preg_match_all('/([^{}]+)\{([^{}]*)\}/s', $content, $matches, PREG_OFFSET_CAPTURE);
            $selectors = [];
            foreach (array_slice($matches[1], 0, 24) as $index => $match) {
                $selector = trim(preg_replace('~/\*.*?\*/~s', '', $match[0]));
                if (!$selector || str_starts_with($selector, '@')) {
                    continue;
                }
                preg_match_all('/(?:^|;)\s*(--[\w-]+|[a-z-]+)\s*:/i', $matches[2][$index][0], $properties);
                $selectors[] = [
                    'selector' => sanitize_text_field(mb_substr($selector, 0, 220)),
                    'line' => substr_count(substr($content, 0, $match[1]), "\n") + 1,
                    'properties' => array_slice(array_values(array_unique($properties[1])), 0, 16),
                ];
            }
            $entry['selectors'] = $selectors;
        }
        preg_match_all(
            '~(?:@(?:import|use|forward)\s+|get_template_part\s*\(\s*|(?:require|include)(?:_once)?\s*\(?\s*)[\x27"]([a-zA-Z0-9_./-]{1,160})[\x27"]~',
            $content,
            $links,
        );
        $entry['references'] = array_values(array_slice(array_unique($links[1]), 0, 16));
        if (in_array($file['kind'], ['php', 'html', 'htm'], true)) {
            preg_match_all(
                '/\b(the_title|the_content|get_header|get_footer|get_template_part|wp_head|wp_footer|wp_nav_menu)\s*\(|<!--\s+wp:([a-z0-9\/-]+)/i',
                $content,
                $renders,
                PREG_OFFSET_CAPTURE,
            );
            $entry['renders'] = [];
            foreach (array_slice($renders[0], 0, 24) as $index => $match) {
                $entry['renders'][] = [
                    'name' => $renders[1][$index][0] ?: 'wp:' . $renders[2][$index][0],
                    'line' => substr_count(substr($content, 0, $match[1]), "\n") + 1,
                ];
            }
        }
        return $entry;
    }

    public static function get(): array
    {
        $theme = wp_get_theme();
        $previous = json_decode((string) get_option(self::CACHE, ''), true);
        $old =
            is_array($previous) && ($previous['indexVersion'] ?? 0) === self::INDEX_VERSION
                ? array_column($previous['manifest']['files'] ?? [], null, 'key')
                : [];
        [$inventory, $truncated] = self::files();
        $files = [];
        $changed = 0;
        $bytes = 0;
        foreach ($inventory as $key => $file) {
            if ($bytes + $file['size'] > 8 * 1024 * 1024) {
                $truncated = true;
                break;
            }
            $hash = @hash_file('sha256', $file['absolute']);
            if (!is_string($hash)) {
                continue;
            }
            $bytes += $file['size'];
            if (isset($old[$key]) && hash_equals($old[$key]['sha256'], $hash)) {
                $files[] = $old[$key];
            } else {
                $files[] = ['key' => $key] + self::entry($file, $hash);
                ++$changed;
            }
        }
        $settings = function_exists('wp_get_global_settings') ? wp_get_global_settings() : [];
        $styles = function_exists('wp_get_global_styles') ? wp_get_global_styles() : [];
        $blocks = [];
        foreach (APEARI_Plans::BLOCKS as $name) {
            $type = WP_Block_Type_Registry::get_instance()->get_registered($name);
            if (!$type) {
                continue;
            }
            $blocks[] = [
                'name' => $name,
                'allowedStyles' => APEARI_Providers::allowed_styles($name),
                'container' => in_array($name, APEARI_Plans::CONTAINERS, true),
                'parent' => array_values($type->parent ?? []),
            ];
        }
        $manifest = [
            'schemaVersion' => self::VERSION,
            'theme' => [
                'stylesheet' => get_stylesheet(),
                'template' => get_template(),
                'name' => sanitize_text_field($theme->get('Name')),
                'version' => (string) $theme->get('Version'),
                'parentVersion' => $theme->parent() ? (string) $theme->parent()->get('Version') : null,
                'blockTheme' => wp_is_block_theme(),
            ],
            'wordpressVersion' => (string) $GLOBALS['wp_version'],
            'design' => [
                'settings' => self::clean(
                    array_intersect_key($settings, array_flip(['color', 'typography', 'spacing', 'layout'])),
                ),
                'styles' => self::clean(
                    array_intersect_key($styles, array_flip(['color', 'typography', 'spacing', 'elements'])),
                ),
            ],
            'files' => $files,
            'blocks' => $blocks,
            'coverage' => [
                'truncated' => $truncated,
                'maxFiles' => 400,
                'maxFileBytes' => 262144,
                'sourceBytesChecked' => $bytes,
                'selectorSamplePerFile' => 24,
            ],
        ];
        $manifest['design']['hashes'] = [
            'globalSettings' => hash('sha256', wp_json_encode($settings)),
            'globalStyles' => hash('sha256', wp_json_encode($styles)),
            'customCss' => hash('sha256', wp_get_custom_css()),
            'themeMods' => hash('sha256', wp_json_encode(get_theme_mods())),
        ];
        if (strlen(wp_json_encode($manifest['design'])) > 60000) {
            $manifest['design'] = [
                'settings' => self::clean($settings['layout'] ?? []),
                'styles' => [],
                'hashes' => $manifest['design']['hashes'],
            ];
            $manifest['coverage']['truncated'] = true;
        }
        while (strlen(wp_json_encode($manifest)) > 650000 && $manifest['files']) {
            array_pop($manifest['files']);
            $manifest['coverage']['truncated'] = true;
        }
        $revision = hash('sha256', wp_json_encode($manifest));
        $hit = is_array($previous) && hash_equals($previous['revision'] ?? '', $revision);
        if (!$hit || ($previous['indexVersion'] ?? 0) !== self::INDEX_VERSION) {
            update_option(
                self::CACHE,
                wp_json_encode([
                    'revision' => $revision,
                    'manifest' => $manifest,
                    'indexVersion' => self::INDEX_VERSION,
                ]),
                false,
            );
        }
        return [
            'revision' => $revision,
            'manifest' => $manifest,
            'cacheHit' => $hit,
            'parsedFiles' => $changed,
        ];
    }

    public static function local(string $instruction): array
    {
        $snapshot = self::get();
        $catalog = json_decode(
            (string) file_get_contents(APEARI_PATH . 'resources/knowledge/gutenberg-v1.json'),
            true,
        );
        $terms = array_slice(
            array_unique(
                preg_split(
                    '/[^a-z0-9_-]+/',
                    strtolower(remove_accents($instruction)),
                    -1,
                    PREG_SPLIT_NO_EMPTY,
                ),
            ),
            0,
            100,
        );
        $score = static function (string $text) use ($terms): int {
            $words = array_slice(
                array_unique(
                    preg_split('/[^a-z0-9_-]+/', strtolower(remove_accents($text)), -1, PREG_SPLIT_NO_EMPTY),
                ),
                0,
                100,
            );
            $score = 0;
            foreach ($terms as $term) {
                if (strlen($term) < 3) {
                    continue;
                }
                foreach ($words as $word) {
                    if (
                        $word === $term ||
                        (strlen($word) >= 4 &&
                            strlen($term) >= 4 &&
                            (str_starts_with($word, $term) || str_starts_with($term, $word)))
                    ) {
                        ++$score;
                        break;
                    }
                }
            }
            return $score;
        };
        $recipes = array_values(
            array_filter(
                $catalog['recipes'],
                static fn($item) => in_array(self::PROFILE, $item['profiles'], true) &&
                    $score($item['terms']) > 0,
            ),
        );
        usort(
            $recipes,
            static fn($a, $b) => $score($b['terms']) <=> $score($a['terms']) ?: strcmp($a['id'], $b['id']),
        );
        $files = $snapshot['manifest']['files'];
        foreach ($files as &$file) {
            $file['rank'] = $score(wp_json_encode($file));
        }
        unset($file);
        usort($files, static fn($a, $b) => $b['rank'] <=> $a['rank'] ?: strcmp($a['key'], $b['key']));
        $candidates = array_map(static function ($file) {
            unset($file['rank']);
            if (isset($file['selectors'])) {
                $file['selectors'] = array_slice($file['selectors'], 0, 6);
            }
            return $file;
        }, array_slice($files, 0, self::PROFILE === 'lite' ? 3 : 8));
        $result = [
            'catalogVersion' => $catalog['version'],
            'revision' => $snapshot['revision'],
            'rules' => array_values(
                array_filter(
                    $catalog['rules'],
                    static fn($item) => in_array(self::PROFILE, $item['profiles'], true),
                ),
            ),
            'recipes' => array_slice($recipes, 0, 5),
            'theme' => $snapshot['manifest']['theme'],
            'design' => $snapshot['manifest']['design'],
            'sourceCandidates' => $candidates,
            'blocks' => $snapshot['manifest']['blocks'],
            'coverage' => $snapshot['manifest']['coverage'],
        ];
        while (strlen(wp_json_encode($result)) > 24000) {
            if ($result['sourceCandidates']) {
                array_pop($result['sourceCandidates']);
            } elseif (isset($result['design'])) {
                unset($result['design']);
            } else {
                break;
            }
        }
        return $result;
    }

    public static function remote(string $url, string $token, array $tls = [], bool $force = false): ?string
    {
        if ('https' !== wp_parse_url($url, PHP_URL_SCHEME)) {
            return null;
        }
        $unavailable = 'apeari_knowledge_remote_' . hash('sha256', $url . '|' . $token . '|unavailable');
        if (!$force && get_transient($unavailable)) {
            return null;
        }
        $snapshot = self::get();
        $key = 'apeari_knowledge_remote_' . hash('sha256', $url . '|' . $token . '|' . $snapshot['revision']);
        if (!$force && get_transient($key)) {
            return $snapshot['revision'];
        }
        $response = wp_safe_remote_request(
            rtrim($url, '/') . '/v1/knowledge/themes/' . $snapshot['revision'],
            array_merge($tls, [
                'method' => 'PUT',
                'timeout' => 12,
                'redirection' => 0,
                'limit_response_size' => 16000,
                'headers' => ['Authorization' => 'Bearer ' . $token, 'Content-Type' => 'application/json'],
                'body' => wp_json_encode(['siteUrl' => home_url('/'), 'manifest' => $snapshot['manifest']]),
            ]),
        );
        if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) {
            set_transient($unavailable, true, MINUTE_IN_SECONDS);
            return null;
        }
        $data = json_decode(wp_remote_retrieve_body($response), true);
        if (!is_array($data) || ($data['revision'] ?? '') !== $snapshot['revision']) {
            return null;
        }
        set_transient($key, true, 10 * MINUTE_IN_SECONDS);
        return $snapshot['revision'];
    }
}