<?php

declare(strict_types=1);

namespace WordPress\CloudflareWorkersAiProvider\Features;

use function WordPress\CloudflareWorkersAiProvider\get_plugin_settings;

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

/**
 * Settings of the features that use the embeddings index: related posts, the link
 * assistant, member search and search insights. All off until switched on.
 */
final class Settings
{
    /** Setting keys owned by the Content tools tab. */
    public const KEYS = [
        'related_enabled', 'related_auto', 'related_count', 'related_min_score', 'related_daily_queries',
        'related_position', 'related_paragraph', 'related_skip_types', 'related_title', 'related_heading', 'related_excerpt', 'related_thumbnail', 'related_why',
        'links_enabled', 'links_daily_neurons',
        'search_enabled', 'search_capability', 'search_takeover', 'search_daily_neurons', 'search_min_score',
        'insights_enabled', 'insights_retention_days', 'insights_daily_rows', 'insights_daily_neurons', 'insights_gap_threshold',
        'chat_enabled', 'chat_capability', 'chat_floating', 'chat_admin', 'chat_name', 'chat_welcome', 'chat_min_score', 'chat_daily_neurons',
        'chat_rerank', 'chat_min_match', 'search_rerank',
        'chat_subtitle', 'chat_avatar', 'chat_accent', 'chat_position', 'chat_chips', 'chat_placeholder', 'chat_no_match', 'chat_offline',
        'chat_ask_name', 'chat_consent_text', 'chat_store', 'chat_retention_days',
        'chat_public_share', 'chat_visitor_daily', 'chat_turnstile',
        'chat_rules', 'chat_blocked', 'chat_blocked_reply', 'chat_safety', 'chat_safety_answers', 'chat_length', 'chat_disclaimer',
        'agents_enabled', 'agents_capability', 'agents_daily_neurons',
    ];

    /** Who may use the site assistant: the member capabilities, or 'public' for visitors too. */
    /** Where "Add automatically" puts the list. */
    public const RELATED_POSITIONS = ['after', 'before', 'paragraph'];

    public const CHAT_ACCESS = ['public', 'read', 'edit_posts', 'publish_posts', 'edit_others_posts', 'manage_options'];

    /** Who may use member search: a capability, from every member down to administrators. */
    public const SEARCH_CAPABILITIES = ['read', 'edit_posts', 'publish_posts', 'edit_others_posts', 'manage_options'];

    /**
     * The usage task of each feature that spends neurons, and the setting holding its daily limit.
     * Related posts spends none: its lists are worked out from vectors already embedded.
     */
    public const LIMITS = [
        'links'    => 'links_daily_neurons',
        'search'   => 'search_daily_neurons',
        'insights' => 'insights_daily_neurons',
        'chat'     => 'chat_daily_neurons',
        'agents'   => 'agents_daily_neurons',
    ];

    /** @return array<string, mixed> */
    public static function defaults(): array
    {
        return [
            'related_enabled'         => 0,
            'related_auto'            => 0,
            'related_count'           => 5,
            'related_min_score'       => 0.55,
            // Store queries a day for working out lists (Vectorize free plan: 30M queried dimensions a month).
            'related_daily_queries'   => 500,
            'related_position'        => 'after',
            'related_paragraph'       => 3,
            'related_skip_types'      => [],
            'related_title'           => '',
            'related_heading'         => 'h2',
            'related_excerpt'         => 0,
            'related_thumbnail'       => 0,
            'related_why'             => 0,
            'links_enabled'           => 0,
            'links_daily_neurons'     => 1000,
            'search_enabled'          => 0,
            'search_capability'       => 'read',
            'search_takeover'         => 0,
            'search_daily_neurons'    => 500,
            'search_min_score'        => 0.5,
            'insights_enabled'        => 0,
            'insights_retention_days' => 90,
            'insights_daily_rows'     => 2000,
            'insights_daily_neurons'  => 500,
            'insights_gap_threshold'  => 0.5,
            'chat_enabled'            => 0,
            'chat_capability'         => 'read',
            'chat_floating'           => 1,
            'chat_admin'              => 1,
            'chat_name'               => '',
            'chat_welcome'            => '',
            // bge-m3: a Marathi question about an English post matched at 0.44 on whoisabhi.com;
            // the model says "not found" when weak passages do not answer.
            'chat_min_score'          => 0.35,
            'chat_rerank'             => 1,
            'chat_min_match'          => 0.08,
            'search_rerank'           => 0,
            'agents_enabled'          => 0,
            'agents_capability'       => 'manage_options',
            'agents_daily_neurons'    => 500,
            'chat_daily_neurons'      => 1000,
            'chat_subtitle'           => '',
            'chat_avatar'             => '',
            'chat_accent'             => '#1d4ed8',
            'chat_position'           => 'right',
            'chat_chips'              => [],
            'chat_placeholder'        => '',
            'chat_no_match'           => '',
            'chat_offline'            => '',
            'chat_ask_name'           => 1,
            'chat_consent_text'       => '',
            'chat_store'              => 1,
            'chat_retention_days'     => 90,
            // Public mode: visitors may use this share of the daily limit, and so many questions each a day.
            'chat_public_share'       => 50,
            'chat_visitor_daily'      => 20,
            'chat_turnstile'          => 1,
            // Guardrails the owner adds to the built-in ones (ChatGuard).
            'chat_rules'              => '',
            'chat_blocked'            => [],
            'chat_blocked_reply'      => '',
            'chat_safety'             => 0,
            'chat_safety_answers'     => 0,
            'chat_length'             => 'medium',
            'chat_disclaimer'         => '',
        ];
    }

    /**
     * A setting, falling back to its default.
     *
     * @return mixed
     */
    public static function get(string $key)
    {
        return get_plugin_settings()[$key] ?? self::defaults()[$key] ?? null;
    }

    /**
     * A feature's daily neuron limit; 0 means no limit of its own (the site-wide guard still applies).
     */
    public static function limit(string $task): int
    {
        $key = self::LIMITS[$task] ?? '';
        return $key === '' ? 0 : max(0, (int) self::get($key));
    }

    /**
     * @param array<string, mixed> $clean
     * @param array<string, mixed> $input
     * @return array<string, mixed>
     */
    public static function sanitize(array $clean, array $input): array
    {
        $prev = get_plugin_settings();
        $tab  = isset($input['_tab']) ? sanitize_key((string) $input['_tab']) : '';
        $d    = self::defaults();

        // 2.0: each content tool has its own page, saved as its own tab (related, search,
        // links, insights). It owns the keys with its prefix; everything else is kept.
        $feature = in_array($tab, \WordPress\CloudflareWorkersAiProvider\Admin\SettingsPage::FEATURE_TABS, true);
        if ($tab !== '' && $tab !== 'content' && !$feature) {
            foreach (self::KEYS as $key) {
                $clean[$key] = $prev[$key] ?? $d[$key];
            }
            return $clean;
        }
        if ($feature) {
            foreach (self::KEYS as $key) {
                if (strpos($key, $tab . '_') !== 0) {
                    $input[$key] = $prev[$key] ?? $d[$key];
                }
            }
        }
        // The site assistant has its own page (Cloudflare AI > Site assistant): the Content tools
        // tab does not show its fields, so saving that tab keeps them as they are.
        if ($tab === 'content') {
            foreach (self::KEYS as $key) {
                if (strpos($key, 'chat_') === 0) {
                    $input[$key] = $prev[$key] ?? $d[$key];
                }
            }
        }
        // Saved without a tab (code calling update_option): an absent key keeps its value.
        // Saved from the tab: an absent checkbox means off.
        if ($tab === '') {
            foreach (self::KEYS as $key) {
                if (!array_key_exists($key, $input)) {
                    $input[$key] = $prev[$key] ?? $d[$key];
                }
            }
        }
        $flag  = static fn(string $k): int => empty($input[$k]) ? 0 : 1;
        $int   = static fn(string $k, int $min, int $max): int => max($min, min($max, (int) ($input[$k] ?? $d[$k])));
        $score = static fn(string $k): float => round(max(0.0, min(1.0, (float) ($input[$k] ?? $d[$k]))), 2);

        $clean['related_enabled']   = $flag('related_enabled');
        $clean['related_auto']      = $flag('related_auto');
        $clean['related_count']     = $int('related_count', 1, 12);
        $clean['related_min_score'] = $score('related_min_score');
        $clean['related_daily_queries'] = $int('related_daily_queries', 0, 1000000);
        $position                       = (string) ($input['related_position'] ?? $d['related_position']);
        $clean['related_position']      = in_array($position, self::RELATED_POSITIONS, true) ? $position : $d['related_position'];
        $clean['related_paragraph']     = $int('related_paragraph', 1, 50);
        // The tab shows a box for each embedded post type, ticked where the list is added: the
        // unticked ones are kept, so a post type embedded later gets the list.
        if (isset($input['related_types_shown'])) {
            $shown = array_map('sanitize_key', (array) ($input['related_types_shown'] ?? []));
            $on    = array_map('sanitize_key', (array) ($input['related_types'] ?? []));
            $skip  = array_values(array_diff($shown, $on));
        } else {
            $skip = array_map('sanitize_key', (array) ($input['related_skip_types'] ?? $d['related_skip_types']));
        }
        $clean['related_skip_types'] = array_values(array_unique(array_filter($skip)));
        $clean['related_title']      = mb_substr(sanitize_text_field((string) ($input['related_title'] ?? '')), 0, 80);
        $heading                     = (string) ($input['related_heading'] ?? $d['related_heading']);
        $clean['related_heading']    = in_array($heading, ['h2', 'h3', 'h4', 'p'], true) ? $heading : $d['related_heading'];
        $clean['related_excerpt']    = $flag('related_excerpt');
        $clean['related_thumbnail']  = $flag('related_thumbnail');
        $clean['related_why']        = $flag('related_why');

        $clean['links_enabled']       = $flag('links_enabled');
        // At least 1: a tool anyone with an account can use must never be without a limit.
        $clean['links_daily_neurons'] = $int('links_daily_neurons', 1, 10000000);

        $cap                           = (string) ($input['search_capability'] ?? $d['search_capability']);
        $clean['search_enabled']       = $flag('search_enabled');
        $clean['search_capability']    = in_array($cap, self::SEARCH_CAPABILITIES, true) ? $cap : $d['search_capability'];
        $clean['search_takeover']      = $flag('search_takeover');
        $clean['search_daily_neurons'] = $int('search_daily_neurons', 1, 10000000);
        $clean['search_min_score']     = $score('search_min_score');

        $clean['insights_enabled']        = $flag('insights_enabled');
        $clean['insights_retention_days'] = $int('insights_retention_days', 7, 730);
        $clean['insights_daily_rows']     = $int('insights_daily_rows', 100, 100000);
        $clean['insights_daily_neurons']  = $int('insights_daily_neurons', 1, 10000000);
        $clean['insights_gap_threshold']  = $score('insights_gap_threshold');

        $cap                         = (string) ($input['chat_capability'] ?? $d['chat_capability']);
        $clean['chat_enabled']       = $flag('chat_enabled');
        $clean['chat_capability']    = in_array($cap, self::CHAT_ACCESS, true) ? $cap : $d['chat_capability'];
        $clean['chat_floating']      = $flag('chat_floating');
        $clean['chat_admin']         = $flag('chat_admin');
        $clean['chat_name']          = mb_substr(sanitize_text_field((string) ($input['chat_name'] ?? '')), 0, 40);
        $clean['chat_welcome']       = mb_substr(sanitize_text_field((string) ($input['chat_welcome'] ?? '')), 0, 200);
        $clean['chat_min_score']     = $score('chat_min_score');
        $clean['chat_rerank']        = $flag('chat_rerank');
        $clean['chat_min_match']     = $score('chat_min_match');
        $clean['search_rerank']      = $flag('search_rerank');
        $cap                           = (string) ($input['agents_capability'] ?? $d['agents_capability']);
        $clean['agents_enabled']       = $flag('agents_enabled');
        $clean['agents_capability']    = in_array($cap, self::SEARCH_CAPABILITIES, true) ? $cap : $d['agents_capability'];
        $clean['agents_daily_neurons'] = $int('agents_daily_neurons', 1, 10000000);
        $clean['chat_daily_neurons'] = $int('chat_daily_neurons', 1, 10000000);
        $text                        = static fn(string $k, int $max): string => mb_substr(sanitize_text_field((string) ($input[$k] ?? '')), 0, $max);
        $clean['chat_subtitle']      = $text('chat_subtitle', 80);
        $clean['chat_avatar']        = mb_substr(trim(sanitize_text_field((string) ($input['chat_avatar'] ?? ''))), 0, 4);
        $accent                      = sanitize_hex_color((string) ($input['chat_accent'] ?? ''));
        $clean['chat_accent']        = is_string($accent) && $accent !== '' ? $accent : $d['chat_accent'];
        $clean['chat_position']      = ($input['chat_position'] ?? '') === 'left' ? 'left' : 'right';
        $chips                       = $input['chat_chips'] ?? [];
        $chips                       = is_array($chips) ? $chips : preg_split('/\R/', (string) $chips);
        $clean['chat_chips']         = array_values(array_slice(array_filter(array_map(static fn($c): string => mb_substr(sanitize_text_field((string) $c), 0, 80), (array) $chips), static fn(string $c): bool => $c !== ''), 0, 3));
        $clean['chat_placeholder']   = $text('chat_placeholder', 80);
        $clean['chat_no_match']      = $text('chat_no_match', 300);
        $clean['chat_offline']       = $text('chat_offline', 300);
        $clean['chat_ask_name']      = $flag('chat_ask_name');
        $clean['chat_consent_text']  = $text('chat_consent_text', 300);
        $clean['chat_store']         = $flag('chat_store');
        $clean['chat_retention_days'] = $int('chat_retention_days', 1, 730);
        $clean['chat_public_share']   = $int('chat_public_share', 1, 100);
        $clean['chat_visitor_daily']  = $int('chat_visitor_daily', 1, 500);
        $clean['chat_turnstile']      = $flag('chat_turnstile');
        $clean['chat_rules']          = mb_substr(sanitize_textarea_field((string) ($input['chat_rules'] ?? '')), 0, 1500);
        $blocked                      = $input['chat_blocked'] ?? [];
        $blocked                      = is_array($blocked) ? $blocked : preg_split('/[\r\n,]+/', (string) $blocked);
        $clean['chat_blocked']        = array_values(array_slice(array_unique(array_filter(array_map(static fn($t): string => mb_substr(trim(sanitize_text_field((string) $t)), 0, 60), (array) $blocked), static fn(string $t): bool => $t !== '')), 0, 100));
        $clean['chat_blocked_reply']  = $text('chat_blocked_reply', 300);
        $clean['chat_safety']         = $flag('chat_safety');
        $clean['chat_safety_answers'] = $flag('chat_safety_answers');
        $clean['chat_length']         = in_array($input['chat_length'] ?? '', ['short', 'medium', 'long'], true) ? (string) $input['chat_length'] : 'medium';
        $clean['chat_disclaimer']     = $text('chat_disclaimer', 200);

        return $clean;
    }
}