import { option } from '@typebot.io/forge'
import { z } from '@typebot.io/forge/zod'
import { baseOptions } from '../baseOptions'

const extractInfoBaseShape = {
  variableId: option.string.layout({
    inputType: 'variableDropdown',
  }),
  description: option.string.layout({
    label: 'Description',
    accordion: 'Advanced',
  }),
  isRequired: option.boolean.layout({
    label: 'Is required',
    moreInfoTooltip:
      'If set to false, there is a chance the variable will be empty',
    accordion: 'Advanced',
    defaultValue: true,
  }),
}

export const toolParametersSchema = option
  .array(
    option.discriminatedUnion('type', [
      option
        .object({
          type: option.literal('string'),
        })
        .extend(extractInfoBaseShape),
      option
        .object({
          type: option.literal('number'),
        })
        .extend(extractInfoBaseShape),
      option
        .object({
          type: option.literal('boolean'),
        })
        .extend(extractInfoBaseShape),
      option
        .object({
          type: option.literal('enum'),
          values: option
            .array(option.string)
            .layout({ itemLabel: 'possible value', mergeWithLastField: true }),
        })
        .extend(extractInfoBaseShape),
    ])
  )
  .layout({
    itemLabel: 'variable mapping',
    accordion: 'Schema',
  })

type Props = {
  defaultModel?: string
  modelFetch: string | readonly [string, ...string[]]
  modelHelperText?: string
}

export const parseGenerateVariablesOptions = ({
  defaultModel,
  modelFetch,
  modelHelperText,
}: Props) =>
  option.object({
    model:
      typeof modelFetch === 'string'
        ? option.string.layout({
            placeholder: 'Select a model',
            label: 'Model',
            fetcher: modelFetch,
            helperText: modelHelperText,
          })
        : option.enum(modelFetch).layout({
            placeholder: 'Select a model',
            label: 'Model',
            helperText: modelHelperText,
          }),
    prompt: option.string.layout({
      label: 'Prompt',
      placeholder: 'Type your text here',
      inputType: 'textarea',
      isRequired: true,
      moreInfoTooltip:
        'Meant to guide the model on what to generate. i.e. "Generate a role-playing game character", "Extract the company name from this text", etc.',
    }),
    variablesToExtract: toolParametersSchema,
  })

export type GenerateVariablesOptions = z.infer<
  ReturnType<typeof parseGenerateVariablesOptions>
> &
  z.infer<typeof baseOptions>