@ui-tars/sdk/dist/utils.d.ts

A powerful cross-platform(ANY device/platform) toolkit for building GUI automation agents for UI-TARS

import { ChatCompletionMessageParam } from 'openai/resources/chat/completions';
import { Conversation, Message } from '@ui-tars/shared/types';
import { type Factors } from './constants';
/**
 * Parse box string to screen coordinates
 *
 *   e.g. '[0.131,0.25,0.131,0.25]' 2560x1440 -> { x: 335.36, y: 360 }
 *
 * @param boxStr box string
 * @param screenWidth screen width
 * @param screenHeight screen height
 * @param factors scaling factor, the training space of the target model.
 * @returns screen coordinates
 */
export declare const parseBoxToScreenCoords: ({ boxStr, screenWidth, screenHeight, factors, }: {
    boxStr: string;
    screenWidth: number;
    screenHeight: number;
    factors?: Factors;
}) => {
    x: null;
    y: null;
} | {
    x: number;
    y: number;
};
export declare const processVlmParams: (conversations: Message[], images: string[]) => {
    images: string[];
    conversations: Message[];
};
export declare const toVlmModelFormat: ({ conversations, systemPrompt, }: {
    conversations: Conversation[];
    systemPrompt: string;
}) => {
    conversations: Message[];
    images: string[];
};
export declare const getSummary: (prediction: string) => string;
/**
 * convert conversations to OpenAI ChatCompletionMessageParam
 * @param conversations conversations
 * @param images images
 * @returns OpenAI ChatCompletionMessageParam
 */
export declare const convertToOpenAIMessages: ({ conversations, images, }: {
    conversations: Message[];
    images: string[];
}) => Array<ChatCompletionMessageParam>;
export declare function replaceBase64Prefix(base64: string): string;
export declare function preprocessResizeImage(image_base64: string, maxPixels: number): Promise<string>;
//# sourceMappingURL=utils.d.ts.map