Python Ghost Cursor

Python port of Xetera/ghost-cursor, for use with Pyppeteer and Playwright.

Generate realistic, human-like mouse movement data between coordinates or navigate between elements with Pyppeteer/Playwright like the definitely-not-robot you are.

Installation

pip install python_ghost_cursor

Usage

Generating movement data between 2 coordinates.

from python_ghost_cursor import path

start = {
    "x": 220,
    "y": 402,
}

end = {
    "x": 902,
    "y": 1032,
}

route = path(start, end)

 # [
 #   { "x": 100, "y": 100 },
 #   { "x": 108.75573501957051, "y": 102.83608396351725 },
 #   { "x": 117.54686481838543, "y": 106.20019239793275 },
 #   { "x": 126.3749821408895, "y": 110.08364505509256 },
 #   { "x": 135.24167973152743, "y": 114.47776168684264 }
 #   ... and so on
 # ]

Usage with Pyppeteer:

import asyncio
import pyppeteer
from python_ghost_cursor.pyppeteer import create_cursor

async def main(url):
  selector = "#sign-up button"
  browser = await pyppeteer.launch(headless=False)
  page = await browser.newPage()
  cursor = createCursor(page)
  await page.goto(url)
  await page.waitForSelector(selector)
  await cursor.click(selector)

asyncio.run(main())

Usage with Playwright (async):

import asyncio
from playwright.async_api import async_playwright
from python_ghost_cursor.playwright_async import create_cursor

async def main():
  async with async_playwright() as p:
    selector = "#sign-up button"
    browser = await p.chromium.launch(channel="chrome", headless=False)
    page = await browser.new_page()
    cursor = create_cursor(page)
    await page.goto(url)
    await page.wait_for_selector(selector)
    await cursor.click(selector)

asyncio.run(main())

Usage with Playwright (sync):

from playwright.sync_api import sync_playwright
from python_ghost_cursor.playwright_sync import create_cursor

def main():
  with sync_playwright() as p:
    selector = "#sign-up button"
    browser = p.chromium.launch(channel="chrome", headless=False)
    page = browser.new_page()
    cursor = create_cursor(page)
    page.goto(url)
    page.wait_for_selector(selector)
    cursor.click(selector)

main()

More info

The original repo gives a description of some of the cool features, along with a good explanation of how it works.