import type {
  ComplianceProvenance,
  HostingProvenance,
  JurisdictionCountry,
  ModelIdentity,
  ProvenanceProvider,
  WeightsOriginCountry,
} from '../receipts/schema';

export interface ProviderRouteInfo {
  provider: ProvenanceProvider;
  weights_origin_country: WeightsOriginCountry;
  available_jurisdictions: string[];
  baa_path: string | null;
  hipaa_eligible: boolean;
  zdr_available: boolean;
  default_retention_days: number | null;
  foreign_origin?: boolean;
}

export interface ProviderRegistryRequest {
  url?: string;
  provider?: string;
  providerRoute?: string;
  model?: string;
}

export interface ProviderRegistryConfig {
  providerRoute?: string;
  jurisdictionCountry?: JurisdictionCountry;
  jurisdictionRegion?: string;
  baaInForce?: boolean;
  baaVendor?: string | null;
  dataRetentionDays?: number | null;
  dataResidencyAttested?: boolean;
  foreignOriginConsentReceiptId?: string | null;
  inferenceBilling?: 'agentguard_managed' | 'customer_managed';
  inferenceBillingDetail?: ComplianceProvenance['inference_billing_detail'];
}

export const PROVIDER_REGISTRY: Record<string, ProviderRouteInfo> = {
  'anthropic-direct': {
    provider: 'anthropic',
    weights_origin_country: 'US',
    available_jurisdictions: ['US'],
    baa_path: 'Anthropic Enterprise Messages API with BAA eligible routing',
    hipaa_eligible: true,
    zdr_available: true,
    default_retention_days: 0,
  },
  'aws-bedrock': {
    provider: 'anthropic',
    weights_origin_country: 'US',
    available_jurisdictions: ['us-east-1', 'us-east-2', 'us-west-2', 'eu-west-1', 'eu-central-1', 'ap-northeast-1'],
    baa_path: 'AWS BAA for Bedrock HIPAA eligible services',
    hipaa_eligible: true,
    zdr_available: true,
    default_retention_days: 0,
  },
  'azure-openai': {
    provider: 'openai',
    weights_origin_country: 'US',
    available_jurisdictions: ['eastus', 'westus', 'westeurope', 'francecentral', 'switzerlandnorth'],
    baa_path: 'Microsoft Product Terms DPA',
    hipaa_eligible: true,
    zdr_available: true,
    default_retention_days: 0,
  },
  'openai-direct': {
    provider: 'openai',
    weights_origin_country: 'US',
    available_jurisdictions: ['US'],
    baa_path: 'OpenAI Enterprise BAA with eligible API configuration',
    hipaa_eligible: true,
    zdr_available: true,
    default_retention_days: 30,
  },
  'vertex-ai': {
    provider: 'google',
    weights_origin_country: 'US',
    available_jurisdictions: ['us-central1', 'us-east4', 'europe-west4', 'europe-west1', 'asia-northeast1'],
    baa_path: 'Google Cloud BAA',
    hipaa_eligible: true,
    zdr_available: true,
    default_retention_days: 0,
  },
  fireworks: {
    provider: 'fireworks',
    weights_origin_country: 'unknown',
    available_jurisdictions: ['US'],
    baa_path: null,
    hipaa_eligible: false,
    zdr_available: false,
    default_retention_days: 30,
  },
  baseten: {
    provider: 'baseten',
    weights_origin_country: 'unknown',
    available_jurisdictions: ['US'],
    baa_path: null,
    hipaa_eligible: false,
    zdr_available: false,
    default_retention_days: 30,
  },
  together: {
    provider: 'together',
    weights_origin_country: 'unknown',
    available_jurisdictions: ['US'],
    baa_path: null,
    hipaa_eligible: false,
    zdr_available: false,
    default_retention_days: 30,
  },
  'moonshot-direct': {
    provider: 'moonshot',
    weights_origin_country: 'CN',
    available_jurisdictions: ['CN'],
    baa_path: null,
    hipaa_eligible: false,
    zdr_available: false,
    default_retention_days: 30,
    foreign_origin: true,
  },
  'deepseek-direct': {
    provider: 'deepseek',
    weights_origin_country: 'CN',
    available_jurisdictions: ['CN'],
    baa_path: null,
    hipaa_eligible: false,
    zdr_available: false,
    default_retention_days: 30,
    foreign_origin: true,
  },
  'alibaba-direct': {
    provider: 'alibaba',
    weights_origin_country: 'CN',
    available_jurisdictions: ['CN'],
    baa_path: null,
    hipaa_eligible: false,
    zdr_available: false,
    default_retention_days: 30,
    foreign_origin: true,
  },
  'kimi-k2-on-baseten': {
    provider: 'baseten',
    weights_origin_country: 'CN',
    available_jurisdictions: ['US'],
    baa_path: null,
    hipaa_eligible: false,
    zdr_available: false,
    default_retention_days: 30,
    foreign_origin: true,
  },
  'deepseek-on-together': {
    provider: 'together',
    weights_origin_country: 'CN',
    available_jurisdictions: ['US'],
    baa_path: null,
    hipaa_eligible: false,
    zdr_available: false,
    default_retention_days: 30,
    foreign_origin: true,
  },
  'qwen-on-together': {
    provider: 'together',
    weights_origin_country: 'CN',
    available_jurisdictions: ['US'],
    baa_path: null,
    hipaa_eligible: false,
    zdr_available: false,
    default_retention_days: 30,
    foreign_origin: true,
  },
  'self-hosted': {
    provider: 'self_hosted',
    weights_origin_country: 'unknown',
    available_jurisdictions: ['self-hosted'],
    baa_path: null,
    hipaa_eligible: false,
    zdr_available: true,
    default_retention_days: 0,
  },
};

const FOREIGN_ORIGIN_PATTERNS = [/kimi/i, /deepseek/i, /qwen/i, /yi[-_ ]/i, /baichuan/i, /glm/i, /moonshot/i];

export function isForeignOriginModel(model: string): boolean {
  return FOREIGN_ORIGIN_PATTERNS.some((pattern) => pattern.test(model));
}

export function inferProviderRoute(req: ProviderRegistryRequest): string {
  const model = String(req.model || '').toLowerCase();
  const url = String(req.url || '').toLowerCase();
  if ((req.providerRoute === 'baseten' || url.includes('baseten')) && /kimi/.test(model)) return 'kimi-k2-on-baseten';
  if ((req.providerRoute === 'together' || url.includes('together')) && /deepseek/.test(model)) return 'deepseek-on-together';
  if ((req.providerRoute === 'together' || url.includes('together')) && /qwen/.test(model)) return 'qwen-on-together';
  if (req.providerRoute && PROVIDER_REGISTRY[req.providerRoute]) return req.providerRoute;
  if (url.includes('bedrock') || req.provider === 'bedrock') return 'aws-bedrock';
  if (url.includes('anthropic') || req.provider === 'anthropic') return 'anthropic-direct';
  if (url.includes('azure') || url.includes('openai.azure')) return 'azure-openai';
  if (url.includes('aiplatform.googleapis') || url.includes('vertex')) return 'vertex-ai';
  if (url.includes('fireworks') || req.providerRoute === 'fireworks') return 'fireworks';
  if (url.includes('baseten') || req.providerRoute === 'baseten') return 'baseten';
  if (url.includes('together') || req.providerRoute === 'together') return 'together';
  if (url.includes('moonshot') || /kimi/.test(model)) return 'moonshot-direct';
  if (url.includes('deepseek') || /deepseek/.test(model)) return 'deepseek-direct';
  if (url.includes('dashscope') || url.includes('alibaba') || /qwen/.test(model)) return 'alibaba-direct';
  if (req.provider === 'openai') return 'openai-direct';
  if (req.provider === 'gemini') return 'vertex-ai';
  return 'self-hosted';
}

export function inferModelIdentity(model: string, route: string): ModelIdentity {
  const info = PROVIDER_REGISTRY[route] ?? PROVIDER_REGISTRY['self-hosted'];
  const modelId = model || 'unknown';
  return {
    provider: modelProvider(modelId, info.provider),
    model_id: modelId,
    model_version: modelVersion(modelId),
    model_family: modelFamily(modelId),
    weights_origin_country: isForeignOriginModel(modelId) || info.foreign_origin ? 'CN' : info.weights_origin_country,
  };
}

export function inferHosting(route: string, config: ProviderRegistryConfig = {}): HostingProvenance {
  const info = PROVIDER_REGISTRY[route] ?? PROVIDER_REGISTRY['self-hosted'];
  const region = config.jurisdictionRegion || info.available_jurisdictions[0] || 'unknown';
  return {
    provider_route: route,
    jurisdiction_country: config.jurisdictionCountry || countryFromRegion(region),
    jurisdiction_region: region,
  };
}

export function inferCompliance(route: string, model: ModelIdentity, config: ProviderRegistryConfig = {}): ComplianceProvenance {
  const info = PROVIDER_REGISTRY[route] ?? PROVIDER_REGISTRY['self-hosted'];
  const retention = config.dataRetentionDays !== undefined ? config.dataRetentionDays : info.default_retention_days;
  return {
    baa_in_force: config.baaInForce ?? false,
    baa_vendor: config.baaVendor ?? info.baa_path,
    hipaa_eligible: info.hipaa_eligible,
    data_retention_days: retention,
    data_residency_attested: config.dataResidencyAttested ?? false,
    foreign_origin_weight_flag: model.weights_origin_country === 'CN' || info.foreign_origin === true,
    foreign_origin_consent_receipt_id: config.foreignOriginConsentReceiptId ?? null,
    inference_billing: config.inferenceBilling ?? 'customer_managed',
    inference_billing_detail: config.inferenceBilling === 'agentguard_managed' ? config.inferenceBillingDetail ?? null : null,
  };
}

function modelProvider(model: string, fallback: ProvenanceProvider): ProvenanceProvider {
  const lower = model.toLowerCase();
  if (lower.includes('claude')) return 'anthropic';
  if (lower.includes('gpt') || lower.includes('openai')) return 'openai';
  if (lower.includes('gemini')) return 'google';
  if (lower.includes('mistral')) return 'mistral';
  if (lower.includes('llama')) return 'meta';
  if (lower.includes('command')) return 'cohere';
  if (lower.includes('deepseek')) return 'deepseek';
  if (lower.includes('kimi') || lower.includes('moonshot')) return 'moonshot';
  if (lower.includes('qwen') || lower.includes('yi') || lower.includes('baichuan') || lower.includes('glm')) return 'alibaba';
  return fallback;
}

function modelVersion(model: string): string {
  const match = model.match(/(?:^|[-_])(20\d{6}|\d{8})(?:$|[-_])/);
  return match?.[1] ?? 'unknown';
}

function modelFamily(model: string): string {
  const lower = model.toLowerCase();
  if (lower.includes('claude')) return 'claude';
  if (lower.includes('gpt')) return 'gpt';
  if (lower.includes('gemini')) return 'gemini';
  if (lower.includes('llama-4') || lower.includes('llama4')) return 'llama-4';
  if (lower.includes('mistral')) return 'mistral';
  if (lower.includes('kimi')) return 'kimi';
  if (lower.includes('deepseek')) return 'deepseek';
  if (lower.includes('qwen')) return 'qwen';
  return lower.split(/[/:_]/).pop()?.split('-').slice(0, 2).join('-') || 'unknown';
}

function countryFromRegion(region: string): JurisdictionCountry {
  const lower = region.toLowerCase();
  if (lower.startsWith('us') || lower.includes('eastus') || lower.includes('westus')) return 'US';
  if (lower.startsWith('eu') || lower.includes('europe') || lower.includes('france') || lower.includes('switzerland')) return 'EU';
  if (lower.startsWith('uk')) return 'UK';
  if (lower.startsWith('ca')) return 'CA';
  if (lower.startsWith('cn')) return 'CN';
  if (lower.startsWith('ap-northeast') || lower.includes('tokyo') || lower.includes('japan')) return 'JP';
  if (lower.startsWith('ap-southeast-2') || lower.includes('australia')) return 'AU';
  return 'unknown';
}