Automating Android Phones with Scrcpy and UiAutomator2 Using Python

Hello, I'm Jack. In a recent video I demonstrated code‑driven phone automation, and this article details the steps I used.

Scrcpy (screen copy) is an open‑source, cross‑platform tool that mirrors an Android device’s screen to a PC over USB or Wi‑Fi. It works on Linux, Windows, and macOS. The project is available at https://github.com/Genymobile/scrcpy .

On Windows, install Scrcpy by downloading the zip archive, extracting it, and adding the folder (which includes adb ) to the system PATH.

Connect the phone via USB, enable Developer Options , then run the command:

scrcpy

This launches a window showing the phone screen, allowing you to control the device from the PC.

For programmatic control, use UiAutomator2 , a Python wrapper around Google’s UiAutomator framework. Unlike the original Java‑only UiAutomator, UiAutomator2 lets you write automation scripts in Python.

The project is hosted at https://github.com/openatx/uiautomator2 . Install it in a Conda environment:

conda create -n android
conda activate android
python -m pip install uiautomator2 weditor

Launch the web inspector with:

python -m weditor

Use the inspector to locate UI elements (e.g., the like button in BiliBili) and then automate actions such as liking, coin‑paying, and following via Python code. A representative script is:

import uiautomator2 as u2
import time
from PIL import Image
import cv2
import numpy as np

all_videos = []

def get_images(device):
    views = device.xpath('//*[@resource-id="tv.danmaku.bili:id/recycler_view"]/android.view.ViewGroup/android.widget.FrameLayout[1]')
    for idx, view_box in enumerate(views.all()[:5]):
        print("视频{} 封面的中心坐标:".format(idx+1), view_box.center())
        image = view_box.screenshot()
        image.save("{}.jpg".format(idx+1))
        all_videos.append(view_box)

def refresh(device):
    device.swipe_ext("down")

def like_the_video(device):
    like_icon = device.xpath('//*[@resource-id="tv.danmaku.bili:id/recommend_icon"]')
    like_icon.click()
    print("视频点赞成功")

def pay_for_the_video(device):
    coin_icon = device.xpath('//*[@resource-id="tv.danmaku.bili:id/coin_icon"]')
    coin_icon.click()
    time.sleep(0.1)
    pay_icon = device.xpath('//*[@resource-id="tv.danmaku.bili:id/pay_coins"]')
    pay_icon.click()
    print("视频投币成功")

def follow_the_up(device):
    follow_icon = device.xpath('//*[@resource-id="tv.danmaku.bili:id/follow"]')
    follow_icon.click()
    print("关注成功")

def back(device):
    back_icon = device.xpath('//*[@content-desc="转到上一层级"]')
    back_icon.click()
    print("已退出视频")

if __name__ == "__main__":
    _DEVICE_ID = 'da317199'
    d = u2.connect(_DEVICE_ID)  # connect to device
    get_images(d)
    all_videos[3].click()
    print("点进去了！")
    time.sleep(0.1)
    like_the_video(d)
    time.sleep(0.1)
    pay_for_the_video(d)
    time.sleep(0.1)
    follow_the_up(d)

Find the device ID with adb devices after connecting the phone via USB.

This approach can be extended to automate tasks that require a mobile device, such as rapid ticket purchases or other time‑critical actions.

Bottom line: Don’t discard old phones; with a few scripts they become powerful automation tools.