memind/mindsearch-config.mjs

const CONFIG_TABLE = 'h5_mindsearch_config';
const CONFIG_SCOPE = 'global';

const SERVICE_KINDS = new Set(['provider', 'reader', 'orchestrator']);
const SERVICE_ADAPTERS = new Set(['searxng', 'github', 'reader', 'research-http']);
const ROUTE_KEYS = ['web', 'news', 'code', 'read', 'research'];

const BUILTIN_SERVICES = Object.freeze([
  {
    id: 'searxng',
    name: 'SearXNG',
    kind: 'provider',
    adapter: 'searxng',
    capabilities: ['search.web', 'search.news'],
    endpoint: '',
    healthPath: '/config',
    enabled: false,
    timeoutMs: 8000,
    priority: 100,
  },
  {
    id: 'github',
    name: 'GitHub Code',
    kind: 'provider',
    adapter: 'github',
    capabilities: ['search.code'],
    endpoint: 'https://api.github.com/search/code',
    healthPath: '',
    enabled: false,
    timeoutMs: 8000,
    priority: 90,
  },
  {
    id: 'reader',
    name: 'Safe Reader',
    kind: 'reader',
    adapter: 'reader',
    capabilities: ['search.read'],
    endpoint: '',
    healthPath: '',
    enabled: false,
    timeoutMs: 8000,
    priority: 80,
  },
  {
    id: 'deep-search',
    name: 'TKMind Deep Search',
    kind: 'orchestrator',
    adapter: 'research-http',
    capabilities: ['search.web', 'research.plan', 'research.execute'],
    endpoint: 'http://127.0.0.1:20100',
    healthPath: '/health',
    enabled: false,
    timeoutMs: 120000,
    priority: 100,
    llmProviderKeyId: '',
    llmModel: '',
  },
]);

export const MINDSEARCH_DEFAULT_CONFIG = Object.freeze({
  enabled: false,
  mode: 'off',
  providers: { searxng: false, github: false, reader: false },
  settings: { searxngEndpoint: '', maxResults: 10, timeoutMs: 8000, readerMaxChars: 12000 },
  services: BUILTIN_SERVICES,
  routes: { web: 'searxng', news: 'searxng', code: 'github', read: 'reader', research: 'deep-search' },
});

const clone = (value) => JSON.parse(JSON.stringify(value));
const bool = (value, fallback = false) => {
  if (typeof value === 'boolean') return value;
  if (value == null || value === '') return fallback;
  return /^(1|true|yes|on)$/i.test(String(value));
};

const boundedNumber = (value, fallback, min, max) =>
  Math.max(min, Math.min(max, Number(value ?? fallback) || fallback));

export function isValidSearchServiceEndpoint(value, { allowEmpty = true } = {}) {
  const endpoint = String(value ?? '').trim();
  if (!endpoint) return allowEmpty;
  try {
    const url = new URL(endpoint);
    if (!['http:', 'https:'].includes(url.protocol) || url.username || url.password) return false;
    return !['0.0.0.0', '::', '[::]', '169.254.169.254'].includes(url.hostname);
  } catch {
    return false;
  }
}

function normalizeCapabilities(value, fallback = []) {
  const capabilities = Array.isArray(value) ? value : fallback;
  return [...new Set(capabilities
    .map((item) => String(item ?? '').trim())
    .filter((item) => /^[a-z][a-z0-9_.-]{1,63}$/.test(item)))]
    .slice(0, 20);
}

function normalizeService(input = {}, fallback = {}) {
  const id = String(input.id ?? fallback.id ?? '').trim().toLowerCase();
  if (!/^[a-z0-9][a-z0-9._-]{0,63}$/.test(id)) return null;
  const kind = SERVICE_KINDS.has(input.kind) ? input.kind : (fallback.kind ?? 'provider');
  const adapter = SERVICE_ADAPTERS.has(input.adapter) ? input.adapter : (fallback.adapter ?? 'searxng');
  const endpoint = String(input.endpoint ?? fallback.endpoint ?? '').trim();
  if (!isValidSearchServiceEndpoint(endpoint)) return null;
  const healthPathValue = String(input.healthPath ?? fallback.healthPath ?? '').trim();
  const healthPath = healthPathValue && /^\/[^\s]{0,255}$/.test(healthPathValue) ? healthPathValue : '';
  const llmProviderKeyIdValue = String(input.llmProviderKeyId ?? fallback.llmProviderKeyId ?? '').trim();
  const llmProviderKeyId = /^[a-zA-Z0-9._:-]{1,128}$/.test(llmProviderKeyIdValue)
    ? llmProviderKeyIdValue
    : '';
  const llmModelValue = String(input.llmModel ?? fallback.llmModel ?? '').trim();
  const llmModel = llmProviderKeyId && !/[\u0000-\u001f\u007f]/.test(llmModelValue)
    ? llmModelValue.slice(0, 200)
    : '';
  return {
    id,
    name: String(input.name ?? fallback.name ?? id).trim().slice(0, 80) || id,
    kind,
    adapter,
    capabilities: normalizeCapabilities(input.capabilities, fallback.capabilities),
    endpoint,
    healthPath,
    enabled: bool(input.enabled, bool(fallback.enabled)),
    timeoutMs: boundedNumber(input.timeoutMs, fallback.timeoutMs ?? 8000, 1000, 120000),
    priority: boundedNumber(input.priority, fallback.priority ?? 100, 1, 1000),
    llmProviderKeyId,
    llmModel,
  };
}

function builtinsFromLegacy(input = {}) {
  const providers = input.providers ?? {};
  const settings = input.settings ?? {};
  return BUILTIN_SERVICES.map((service) => normalizeService({
    ...service,
    enabled: bool(providers[service.adapter], service.enabled),
    endpoint: service.adapter === 'searxng'
      ? String(settings.searxngEndpoint ?? service.endpoint)
      : service.endpoint,
    timeoutMs: service.adapter === 'research-http'
      ? service.timeoutMs
      : (settings.timeoutMs ?? service.timeoutMs),
  }, service));
}

function normalizeServices(input = {}) {
  const source = Array.isArray(input.services) ? input.services : builtinsFromLegacy(input);
  const seen = new Set();
  const services = [];
  for (const item of source) {
    const fallback = BUILTIN_SERVICES.find((candidate) => candidate.id === item?.id) ?? {};
    const service = normalizeService(item, fallback);
    if (!service || seen.has(service.id)) continue;
    seen.add(service.id);
    services.push(service);
  }
  for (const builtin of builtinsFromLegacy(input)) {
    if (!seen.has(builtin.id)) services.push(builtin);
  }
  return services.slice(0, 30);
}

function normalizeRoutes(value, services) {
  const ids = new Set(services.map((service) => service.id));
  const defaults = MINDSEARCH_DEFAULT_CONFIG.routes;
  return Object.fromEntries(ROUTE_KEYS.map((key) => {
    const requested = String(value?.[key] ?? defaults[key] ?? '').trim();
    return [key, !requested || ids.has(requested) ? requested : ''];
  }));
}

export function resolveMindSearchService(config, route) {
  const normalized = normalizeMindSearchConfig(config);
  const serviceId = normalized.routes?.[route] ?? '';
  return normalized.services.find((service) => service.id === serviceId && service.enabled) ?? null;
}

export async function probeMindSearchService(service, { fetchImpl = fetch } = {}) {
  const normalized = normalizeService(service);
  if (!normalized) throw new Error('Invalid search service configuration');
  if (!normalized.endpoint) {
    return {
      ok: normalized.adapter === 'reader',
      serviceId: normalized.id,
      adapter: normalized.adapter,
      status: null,
      latencyMs: 0,
      message: normalized.adapter === 'reader' ? 'Built-in reader is available' : 'Service endpoint is not configured',
    };
  }
  const startedAt = Date.now();
  let url;
  if (normalized.adapter === 'searxng') {
    url = new URL(normalized.endpoint);
    url.searchParams.set('q', 'OpenAI');
    url.searchParams.set('format', 'json');
  } else if (normalized.healthPath) {
    url = new URL(normalized.healthPath, normalized.endpoint);
  } else {
    return {
      ok: true,
      serviceId: normalized.id,
      adapter: normalized.adapter,
      status: null,
      latencyMs: 0,
      message: 'Configuration is valid; no health path is configured',
    };
  }
  try {
    const response = await fetchImpl(url, {
      signal: AbortSignal.timeout(normalized.timeoutMs),
      headers: {
        accept: 'application/json,text/plain',
        ...(normalized.adapter === 'research-http' && process.env.TKMIND_DEEP_SEARCH_SECRET
          ? { 'x-secret-key': process.env.TKMIND_DEEP_SEARCH_SECRET }
          : {}),
      },
    });
    let resultCount = null;
    if (normalized.adapter === 'searxng' && response.ok) {
      const body = await response.json();
      resultCount = Array.isArray(body.results) ? body.results.length : 0;
    }
    return {
      ok: response.ok && (resultCount == null || resultCount > 0),
      serviceId: normalized.id,
      adapter: normalized.adapter,
      status: response.status,
      latencyMs: Date.now() - startedAt,
      resultCount,
      message: response.ok ? 'Health check completed' : `Service returned HTTP ${response.status}`,
    };
  } catch (error) {
    return {
      ok: false,
      serviceId: normalized.id,
      adapter: normalized.adapter,
      status: null,
      latencyMs: Date.now() - startedAt,
      message: error instanceof Error ? error.message : 'Service health check failed',
    };
  }
}

export function normalizeMindSearchConfig(input = {}) {
  const mode = ['off', 'shadow', 'assist'].includes(input.mode) ? input.mode : 'off';
  const services = normalizeServices(input);
  const serviceByAdapter = (adapter) => services.find((service) => service.adapter === adapter);
  const searxng = serviceByAdapter('searxng');
  const github = serviceByAdapter('github');
  const reader = serviceByAdapter('reader');
  const config = {
    enabled: bool(input.enabled),
    mode,
    providers: {
      searxng: bool(searxng?.enabled),
      github: bool(github?.enabled),
      reader: bool(reader?.enabled),
    },
    settings: {
      searxngEndpoint: String(searxng?.endpoint ?? input.settings?.searxngEndpoint ?? '').trim(),
      maxResults: boundedNumber(input.settings?.maxResults, 10, 1, 20),
      timeoutMs: boundedNumber(input.settings?.timeoutMs, 8000, 1000, 30000),
      readerMaxChars: boundedNumber(input.settings?.readerMaxChars, 12000, 1000, 50000),
    },
    services,
    routes: normalizeRoutes(input.routes, services),
  };
  if (!config.enabled || config.mode === 'off') {
    config.enabled = false;
    config.mode = 'off';
  }
  return config;
}

function envConfig(env = process.env) {
  const enabled = bool(env.TKMIND_SEARCH_ENABLED);
  return normalizeMindSearchConfig({
    enabled,
    mode: env.TKMIND_SEARCH_MODE || (enabled ? 'assist' : 'off'),
    providers: {
      searxng: env.TKMIND_SEARCH_PROVIDER_SEARXNG_ENABLED,
      github: env.TKMIND_SEARCH_PROVIDER_GITHUB_ENABLED,
      reader: env.TKMIND_SEARCH_READER_ENABLED,
    },
    settings: {
      searxngEndpoint: env.TKMIND_SEARCH_SEARXNG_URL,
      maxResults: env.TKMIND_SEARCH_MAX_RESULTS,
      timeoutMs: env.TKMIND_SEARCH_TIMEOUT_MS,
      readerMaxChars: env.TKMIND_SEARCH_READER_MAX_CHARS,
    },
    services: parseJsonArray(env.TKMIND_SEARCH_SERVICES_JSON),
    routes: parseJsonObject(env.TKMIND_SEARCH_ROUTES_JSON),
  });
}

function parseJsonArray(value) {
  if (!value) return undefined;
  try {
    const parsed = JSON.parse(value);
    return Array.isArray(parsed) ? parsed : undefined;
  } catch {
    return undefined;
  }
}

function parseJsonObject(value) {
  if (!value) return undefined;
  try {
    const parsed = JSON.parse(value);
    return parsed && typeof parsed === 'object' && !Array.isArray(parsed) ? parsed : undefined;
  } catch {
    return undefined;
  }
}

function applyLegacyPatchToServices(current, patch) {
  if (Array.isArray(patch?.services)) return patch.services;
  const providerPatch = patch?.providers ?? {};
  const settingPatch = patch?.settings ?? {};
  return current.services.map((service) => {
    if (service.adapter === 'searxng') {
      return {
        ...service,
        ...(providerPatch.searxng == null ? {} : { enabled: providerPatch.searxng }),
        ...(settingPatch.searxngEndpoint == null ? {} : { endpoint: settingPatch.searxngEndpoint }),
        ...(settingPatch.timeoutMs == null ? {} : { timeoutMs: settingPatch.timeoutMs }),
      };
    }
    if (service.adapter === 'github' && providerPatch.github != null) {
      return { ...service, enabled: providerPatch.github };
    }
    if (service.adapter === 'reader' && providerPatch.reader != null) {
      return { ...service, enabled: providerPatch.reader };
    }
    return service;
  });
}

export async function ensureMindSearchConfigSchema(pool) {
  await pool.query(`CREATE TABLE IF NOT EXISTS ${CONFIG_TABLE} (
    config_scope VARCHAR(32) PRIMARY KEY,
    config_json JSON NOT NULL,
    updated_by CHAR(36) NULL,
    updated_at BIGINT NOT NULL
  ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci`);
}

export function createMindSearchConfigService(pool, { env = process.env, fetchImpl = fetch } = {}) {
  let ready = null;
  const ensure = async () => {
    if (!pool) return;
    ready ??= ensureMindSearchConfigSchema(pool);
    await ready;
  };
  const load = async () => {
    const fallback = envConfig(env);
    if (!pool) return { config: fallback, source: 'env' };
    await ensure();
    const [rows] = await pool.query(`SELECT config_json, updated_by, updated_at FROM ${CONFIG_TABLE} WHERE config_scope = ? LIMIT 1`, [CONFIG_SCOPE]);
    if (!rows?.[0]) return { config: fallback, source: 'env' };
    let stored = rows[0].config_json;
    if (typeof stored === 'string') { try { stored = JSON.parse(stored); } catch { stored = {}; } }
    return { config: normalizeMindSearchConfig(stored), source: 'admin', updatedBy: rows[0].updated_by ?? null, updatedAt: Number(rows[0].updated_at ?? 0) || null };
  };
  return {
    ensureSchema: ensure,
    getEffectiveConfig: async () => (await load()).config,
    getAdminConfig: async () => load(),
    testService: async (serviceId) => {
      const current = await load();
      const service = current.config.services.find((item) => item.id === String(serviceId ?? '').trim());
      if (!service) throw Object.assign(new Error('Search service not found'), { code: 'SEARCH_SERVICE_NOT_FOUND' });
      return probeMindSearchService(service, { fetchImpl });
    },
    updateAdminConfig: async (patch, { updatedBy = null } = {}) => {
      const current = await load();
      const next = normalizeMindSearchConfig({
        ...current.config,
        ...patch,
        providers: { ...current.config.providers, ...(patch?.providers ?? {}) },
        settings: { ...current.config.settings, ...(patch?.settings ?? {}) },
        services: applyLegacyPatchToServices(current.config, patch),
        routes: { ...current.config.routes, ...(patch?.routes ?? {}) },
      });
      if (!pool) return { config: next, source: 'env' };
      await ensure();
      const now = Date.now();
      await pool.query(`INSERT INTO ${CONFIG_TABLE} (config_scope, config_json, updated_by, updated_at) VALUES (?, ?, ?, ?) ON DUPLICATE KEY UPDATE config_json = VALUES(config_json), updated_by = VALUES(updated_by), updated_at = VALUES(updated_at)`, [CONFIG_SCOPE, JSON.stringify(next), updatedBy, now]);
      return { config: next, source: 'admin', updatedBy, updatedAt: now };
    },
    getRuntimeState: async () => {
      const config = await (async () => (await load()).config)();
      return {
        enabled: config.enabled,
        mode: config.mode,
        providers: config.providers,
        services: config.services.map(({ id, name, kind, adapter, capabilities, enabled, timeoutMs, priority }) => ({
          id, name, kind, adapter, capabilities, enabled, timeoutMs, priority,
        })),
        routes: config.routes,
        effective: config.enabled && config.mode !== 'off',
      };
    },
  };
}