import { TypedField, FieldType } from "react-declarative";

declare var Neuron: any;

interface IData {
    net: {
        binaryThresh: string | number,
        hiddenLayers: {
            [k: string]: string
        } | number[],
        activation: string,
        leakyReluAlpha: string | number,
    },
    train: {
        iterations: string | number,
        errorThresh: string | number,
        log: boolean,
        logPeriod: string | number,
        learningRate: string | number,
        momentum: string | number,
        callbackPeriod: string | number,
        timeout: string | number,
    },
};

const initialData: IData = {
  net: {
    binaryThresh: '0.5',
    hiddenLayers: {
      0: '8',
      1: '6',
      2: '4',
      3: '0',
      4: '0',
      5: '0',
      6: '0',
      7: '0',
    },
    activation: 'leaky-relu',
    leakyReluAlpha: '0.001',
  },
  train: {
    iterations: '30000',
    errorThresh: '0.005',
    log: true,
    logPeriod: '1000',
    learningRate: '0.001',
    momentum: '0.1',
    callbackPeriod: '10',
    timeout: '9999999999',
  },
};

const createNeuronField = (index: number): TypedField => ({
  type: FieldType.Group,
  fields: [
    {
      type: FieldType.Typography,
      columns: "2",
      placeholder: `Number of neurons on layer ${index + 1}`,
    },
    {
      type: FieldType.Slider,
      name: `net.hiddenLayers.${index}`,
      columns: "9",
      minSlider: 0,
      maxSlider: 15,
      stepSlider: 1,
    },
    {
      type: FieldType.Text,
      columns: "1",
      inputType: "number",
      outlined: false,
      compute({ net: { hiddenLayers } }) {
        return hiddenLayers[index].toString();
      },
    },
  ],
});

export const fields: TypedField<IData>[] = [
  {
    type: FieldType.Group,
    phoneColumns: "12",
    tabletColumns: "3",
    desktopColumns: "3",
    fields: [
      {
        type: FieldType.Component,
        sx: { mr: 1 },
        element: ({ net: { hiddenLayers } }) => <Neuron hiddenLayers={hiddenLayers} />,
      },
    ],
  },
  {
    type: FieldType.Group,
    phoneColumns: "12",
    tabletColumns: "9",
    desktopColumns: "9",
    fields: [
      {
        type: FieldType.Line,
        title: "Neural Network",
      },
      {
        type: FieldType.Expansion,
        style: { marginBottom: 16 },
        title: "Layers",
        description: "Number of neurons on each layer",
        fields: [
          createNeuronField(0),
          createNeuronField(1),
          createNeuronField(2),
          createNeuronField(3),
          createNeuronField(4),
          createNeuronField(5),
          createNeuronField(6),
          createNeuronField(7),
        ],
      },
      {
        type: FieldType.Text,
        inputType: "number",
        name: "net.binaryThresh",
        title: "Binary threshhold",
        description: "Transfer function configuration",
        defaultValue: initialData.net.binaryThresh,
      },
      {
        type: FieldType.Combo,
        name: "net.activation",
        title: "Activation function",
        description: "Transfer function itself",
        isInvalid({ net: { activation } }) {
          if (activation === null) {
            return "Transfer function is required";
          } else {
            return null;
          }
        },
        tr(activation) {
          if (activation === "sigmoid") {
            return "Sigmoid function";
          } else if (activation === "relu") {
            return "Relu function";
          } else if (activation === "leaky-relu") {
            return "Leaky-Relu function";
          } else if (activation === "tanh") {
            return "Tanh function";
          } else {
            return "unknown";
          }
        },
        itemList: ["sigmoid", "relu", "leaky-relu", "tanh"],
        defaultValue: initialData.net.activation,
      },
      {
        type: FieldType.Text,
        inputType: "number",
        name: "net.leakyReluAlpha",
        title: "Leaky relu alpha",
        description: "Negative slope coefficient",
        defaultValue: initialData.net.leakyReluAlpha,
      },
    ],
  },
  {
    type: FieldType.Group,
    fields: [
      {
        type: FieldType.Line,
        title: "Training",
      },
      {
        type: FieldType.Expansion,
        style: { marginBottom: 16 },
        title: "Logging",
        description: "Enable logging",
        fields: [
          {
            type: FieldType.Switch,
            title: "Enable logging",
            name: "train.log",
            defaultValue: initialData.train.log,
          },
          {
            type: FieldType.Text,
            inputType: "number",
            name: "train.logPeriod",
            title: "Logging period",
            description: "Trainment logging period",
            isDisabled({ train: { log } }) {
              return !log;
            },
            defaultValue: initialData.train.logPeriod,
          },
        ],
      },
      {
        type: FieldType.Text,
        inputType: "number",
        name: "train.iterations",
        title: "Iterations",
        description: "Number of trainment iterations",
        defaultValue: initialData.train.iterations,
      },
      {
        type: FieldType.Text,
        inputType: "number",
        name: "train.errorThresh",
        title: "Error threshold",
        description: "Trainment error threshold",
        defaultValue: initialData.train.errorThresh,
      },
      {
        type: FieldType.Text,
        inputType: "number",
        name: "train.learningRate",
        title: "Learning rate",
        description: "Trainment learning rate",
        defaultValue: initialData.train.learningRate,
      },
      {
        type: FieldType.Text,
        inputType: "number",
        name: "train.momentum",
        title: "Learning momentum",
        description: "Trainment learning momentum",
        defaultValue: initialData.train.momentum,
      },
      {
        type: FieldType.Text,
        inputType: "number",
        name: "train.callbackPeriod",
        title: "Callback period",
        description: "Callback period",
        defaultValue: initialData.train.callbackPeriod,
      },
      {
        type: FieldType.Text,
        inputType: "number",
        name: "train.timeout",
        title: "Iteration timeout",
        description: "Iteration fallback timeout",
        defaultValue: initialData.train.timeout,
      },
    ],
  },
];