import {z} from 'zod'

export type GetApiApiProjectIdTrainingKerasLearnIdMetadataArgs = {projectId: number, learnId: number, excludeLabels?: boolean | undefined};

export type GetApiApiProjectIdTrainingKerasLearnIdMetadataOptions = {baseUrl?: string | undefined, ApiKeyAuthentication?: string | undefined, fetch?: typeof fetch | undefined};

export const kerasModelLayer = z.object({input: z.object({shape: z.number().int(), name: z.string(), type: z.string()}), output: z.object({shape: z.number().int(), name: z.string(), type: z.string()})});

export const kerasModelTypeEnum = z.enum(["int8", "float32", "akida", "requiresRetrain"]);

export const kerasCustomMetric = z.object({name: z.string(), value: z.string()});

export const boundingBoxWithScore = z.object({label: z.string(), x: z.number(), y: z.number(), width: z.number(), height: z.number(), score: z.number()});

export const modelPrediction = z.object({sampleId: z.number().int(), startMs: z.number(), endMs: z.number(), label: z.string().optional(), prediction: z.string(), predictionCorrect: z.boolean().optional(), expectedAnomalyOutcome: z.string().optional(), f1Score: z.number().optional(), anomalyScores: z.array(z.array(z.number())).optional(), boundingBoxes: z.array(boundingBoxWithScore).optional(), labelMapPredictions: z.record(z.string(), z.string()).optional()});

export const additionalMetric = z.object({name: z.string(), value: z.string(), fullPrecisionValue: z.number(), tooltipText: z.string().optional(), link: z.string().optional()});

export const kerasModelMetadataMetrics = z.object({type: kerasModelTypeEnum, loss: z.number(), accuracy: z.number().optional(), confusionMatrix: z.array(z.array(z.number())), report: z.object({}), onDevicePerformance: z.array(z.object({mcu: z.string(), name: z.string(), isDefault: z.boolean(), latency: z.number(), tflite: z.object({ramRequired: z.number().int(), romRequired: z.number().int(), arenaSize: z.number().int(), modelSize: z.number().int()}), eon: z.object({ramRequired: z.number().int(), romRequired: z.number().int(), arenaSize: z.number().int(), modelSize: z.number().int()}), eon_ram_optimized: z.object({ramRequired: z.number().int(), romRequired: z.number().int(), arenaSize: z.number().int(), modelSize: z.number().int()}).optional(), customMetrics: z.array(kerasCustomMetric).optional(), hasPerformance: z.boolean(), profilingError: z.string().optional()})), predictions: z.array(modelPrediction).optional(), visualization: z.enum(["featureExplorer", "dataExplorer", "none"]), isSupportedOnMcu: z.boolean(), mcuSupportError: z.string().optional(), profilingJobId: z.number().int().optional(), profilingJobFailed: z.boolean().optional(), additionalMetrics: z.array(additionalMetric)});

export const kerasModelMode = z.enum(["classification", "regression", "object-detection", "visual-anomaly", "anomaly-gmm", "freeform", "anomaly"]);

export const objectDetectionLastLayer = z.enum(["mobilenet-ssd", "fomo", "yolov2-akida", "yolov5", "yolov5v5-drpai", "yolox", "yolov7", "yolo-pro", "tao-retinanet", "tao-ssd", "tao-yolov3", "tao-yolov4", "yolov11", "yolov11-abs", "paddleocr-detector", "qc-face-det-lite", "qc-yolox"]);

export const imageInputScaling = z.enum(["0..1", "-1..1", "-128..127", "0..255", "torch", "bgr-subtract-imagenet-mean"]);

export const blockThreshold = z.object({key: z.string(), description: z.string(), helpText: z.string(), suggestedValue: z.number().optional(), suggestedValueText: z.string().optional(), value: z.union([z.number(), z.string()]), dropdownOptions: z.array(z.object({description: z.string(), value: z.string()})).optional()});

export const kerasModelMetadataGraphSeries = z.object({title: z.string(), values: z.array(z.number())});

export const kerasModelMetadataGraph = z.object({title: z.string(), xLabel: z.string().optional(), yLabel: z.string().optional(), description: z.string().optional(), hideInUI: z.boolean().optional(), data: z.array(kerasModelMetadataGraphSeries)});

export const tensorboardGraphs = z.array(kerasModelMetadataGraph);

export const foldModelMetadata = z.object({created: z.string(), layers: z.array(kerasModelLayer), classNames: z.array(z.string()), labels: z.array(z.string()), availableModelTypes: z.array(kerasModelTypeEnum), recommendedModelType: kerasModelTypeEnum, modelValidationMetrics: z.array(kerasModelMetadataMetrics), hasTrainedModel: z.boolean(), mode: kerasModelMode, objectDetectionLastLayer: objectDetectionLastLayer.optional(), imageInputScaling: imageInputScaling, thresholds: z.array(blockThreshold), tensorboardGraphs: tensorboardGraphs.optional(), foldId: z.number().int().gte(0), foldName: z.string()});

export const kFoldsMetadata = z.object({folds: z.array(foldModelMetadata), hasFinalRun: z.boolean().optional(), persistedFoldId: z.number().int(), persistedFoldName: z.string(), persistedFoldMetricType: kerasModelTypeEnum.optional()});

export const kerasModelMetadataResponse = z.object({success: z.boolean(), error: z.string().optional(), created: z.string(), layers: z.array(kerasModelLayer), classNames: z.array(z.string()), labels: z.array(z.string()), availableModelTypes: z.array(kerasModelTypeEnum), recommendedModelType: kerasModelTypeEnum, modelValidationMetrics: z.array(kerasModelMetadataMetrics), hasTrainedModel: z.boolean(), mode: kerasModelMode, objectDetectionLastLayer: objectDetectionLastLayer.optional(), imageInputScaling: imageInputScaling, thresholds: z.array(blockThreshold), tensorboardGraphs: tensorboardGraphs.optional(), kFoldsMetadata: kFoldsMetadata.optional()});

export const validateResponse = async <Output>(
  schema: {
    readonly '~standard': {
      readonly validate: (value: unknown) => ({ readonly value: Output; readonly issues?: undefined } | { readonly issues: ReadonlyArray<{ readonly message: string; readonly path?: ReadonlyArray<PropertyKey | { readonly key: PropertyKey }> | undefined }> }) | Promise<{ readonly value: Output; readonly issues?: undefined } | { readonly issues: ReadonlyArray<{ readonly message: string; readonly path?: ReadonlyArray<PropertyKey | { readonly key: PropertyKey }> | undefined }> }>
    }
  },
  value: unknown
): Promise<Output> => {
  const result = await schema['~standard'].validate(value)

  if (result.issues !== undefined) {
    const detail = result.issues
      .map(({ message, path }) =>
        path?.length
          ? `${path.map((segment) => String(typeof segment === 'object' ? segment.key : segment)).join('.')}: ${message}`
          : message
      )
      .join('; ')

    throw new Error(`Response does not match the schema: ${detail}`, { cause: { issues: result.issues } })
  }

  return result.value
};

/**
 * Keras metadata
 * 
 * GET /api/{projectId}/training/keras/{learnId}/metadata
 * 
 * Also accepts auth: JWTAuthentication, JWTHttpHeaderAuthentication, OAuth2
 */
export const getApiApiProjectIdTrainingKerasLearnIdMetadata = async (args: GetApiApiProjectIdTrainingKerasLearnIdMetadataArgs, options?: GetApiApiProjectIdTrainingKerasLearnIdMetadataOptions) => {
const {baseUrl = "https://studio.edgeimpulse.com/v1", fetch: fetchFn = fetch, ApiKeyAuthentication} = options ?? {}

const query = new URLSearchParams()
const headers = new Headers()
headers.set('accept', "application/json")

if (args.excludeLabels != null) query.append("excludeLabels", String(args.excludeLabels))
if (ApiKeyAuthentication !== undefined) headers.set("x-api-key", ApiKeyAuthentication)

const url = `${baseUrl.replace(/\/+$/, '')}/api/${encodeURIComponent(String(args.projectId))}/training/keras/${encodeURIComponent(String(args.learnId))}/metadata${query.size > 0 ? `?${query}` : ''}`

const res = await fetchFn(url, {method: "GET", headers})

if (!res.ok) throw new Error(`Request failed with status ${res.status}`, {cause: {status: res.status, body: await res.text()}})

return validateResponse(kerasModelMetadataResponse, await res.json())
};