๐ญ Playwright for Python
Playwright is a Python library to automate Chromium, Firefox and WebKit with a single API. Playwright is built to enable cross-browser web automation that is ever-green, capable, reliable and fast.
Linux | macOS | Windows | |
---|---|---|---|
Chromium 86.0.4238.0 | โ | โ | โ |
WebKit 14.0 | โ | โ | โ |
Firefox 80.0b8 | โ | โ | โ |
Headless execution is supported for all the browsers on all platforms.
pip install playwright
python -m playwright install
This installs Playwright and browser binaries for Chromium, Firefox and WebKit. Once installed, you can import
Playwright in a Python script and automate web browser interactions. Playwright requires Python 3.7+.
Playwright is built to automate the broad and growing set of web browser capabilities used by Single Page Apps and Progressive Web Apps.
- Scenarios that span multiple page, domains and iframes
- Auto-wait for elements to be ready before executing actions (like click, fill)
- Intercept network activity for stubbing and mocking network requests
- Emulate mobile devices, geolocation, permissions
- Support for web components via shadow-piercing selectors
- Native input events for mouse and keyboard
- Upload and download files
Playwright can be used as a library in your application or as a part of the testing solution. We recommend using our Pytest plugin for testing.
As a library, Playwright offers both blocking (synchronous) API and asyncio API (async/await). You can pick the one that works best for you. They are identical in terms of capabilities and only differ in a way one consumes the API.
Below are some of the examples on how these variations of the API can be used:
from playwright import sync_playwright
with sync_playwright() as p:
for browser_type in [p.chromium, p.firefox, p.webkit]:
browser = browser_type.launch()
page = browser.newPage()
page.goto('http://whatsmyuseragent.org/')
page.screenshot(path=f'example-{browser_type.name}.png')
browser.close()
import asyncio
from playwright import async_playwright
async def main():
async with async_playwright() as p:
for browser_type in [p.chromium, p.firefox, p.webkit]:
browser = await browser_type.launch()
page = await browser.newPage()
await page.goto('http://whatsmyuseragent.org/')
await page.screenshot(path=f'example-{browser_type.name}.png')
await browser.close()
asyncio.get_event_loop().run_until_complete(main())
def test_playwright_is_visible_on_google(page):
page.goto("https://www.google.com")
page.type("input[name=q]", "Playwright GitHub")
page.click("input[type=submit]")
page.waitForSelector("text=microsoft/Playwright")
For more information on pytest-playwright, see GitHub.
For scripting purposes, it is also possible to start and stop Playwright manually without relying on the indentation of the REPL.
from playwright import sync_playwright
playwright = sync_playwright().start()
for browser_type in [playwright.chromium, playwright.firefox, playwright.webkit]:
browser = browser_type.launch()
page = browser.newPage()
page.goto("http://whatsmyuseragent.org/")
page.screenshot(path=f"example-{browser_type.name}.png")
browser.close()
playwright.stop()
This snippet emulates Mobile Safari on a device at a given geolocation, navigates to maps.google.com, performs action and takes a screenshot.
from playwright import sync_playwright
with sync_playwright() as p:
iphone_11 = p.devices['iPhone 11 Pro']
browser = p.webkit.launch(headless=False)
context = browser.newContext(
**iphone_11,
locale='en-US',
geolocation={ 'longitude': 12.492507, 'latitude': 41.889938 },
permissions=['geolocation']
)
page = context.newPage()
page.goto('https://maps.google.com')
page.click('text="Your location"')
page.screenshot(path='colosseum-iphone.png')
browser.close()
The asyncio variant:
import asyncio
from playwright import async_playwright
async def main():
async with async_playwright() as p:
iphone_11 = p.devices['iPhone 11 Pro']
browser = await p.webkit.launch(headless=False)
context = await browser.newContext(
**iphone_11,
locale='en-US',
geolocation={ 'longitude': 12.492507, 'latitude': 41.889938 },
permissions=['geolocation']
)
page = await context.newPage()
await page.goto('https://maps.google.com')
await page.click('text="Your location"')
await page.screenshot(path='colosseum-iphone.png')
await browser.close()
asyncio.get_event_loop().run_until_complete(main())
This code snippet navigates to example.com in Firefox, and executes a script in the page context.
from playwright import sync_playwright
with sync_playwright() as p:
browser = p.firefox.launch()
page = browser.newPage()
page.goto('https://www.example.com/')
dimensions = page.evaluate('''() => {
return {
width: document.documentElement.clientWidth,
height: document.documentElement.clientHeight,
deviceScaleFactor: window.devicePixelRatio
}
}''')
print(dimensions)
browser.close()
The asyncio variant:
import asyncio
from playwright import async_playwright
async def main():
async with async_playwright() as p:
browser = await p.firefox.launch()
page = await browser.newPage()
await page.goto('https://www.example.com/')
dimensions = await page.evaluate('''() => {
return {
width: document.documentElement.clientWidth,
height: document.documentElement.clientHeight,
deviceScaleFactor: window.devicePixelRatio
}
}''')
print(dimensions)
await browser.close()
asyncio.get_event_loop().run_until_complete(main())
This code snippet sets up request routing for a Chromium page to log all network requests.
from playwright import sync_playwright
with sync_playwright() as p:
browser = p.chromium.launch()
page = browser.newPage()
def log_and_continue_request(route, request):
print(request.url)
route.continue_()
# Log and continue all network requests
page.route('**', lambda route, request: log_and_continue_request(route, request))
page.goto('http://todomvc.com')
browser.close()
The asyncio variant:
import asyncio
from playwright import async_playwright
async def main():
async with async_playwright() as p:
browser = await p.chromium.launch()
page = await browser.newPage()
def log_and_continue_request(route, request):
print(request.url)
asyncio.create_task(route.continue_())
# Log and continue all network requests
await page.route('**', lambda route, request: log_and_continue_request(route, request))
await page.goto('http://todomvc.com')
await browser.close()
asyncio.get_event_loop().run_until_complete(main())
We are in the process of converting the documentation from the Node.js form to the Python one. But you can go ahead and use the Node.js documentation because the API is pretty much the same. You might have noticed that Playwright uses non-Python naming conventions, camelCase
instead of the snake_case
for its methods. We recognize that this is not ideal, but it was done deliberately, so that you could rely upon the stack overflow answers and the documentation of the Playwright for Node.js.
You can use all the same methods and arguments as documented, just remember that since Python allows named arguments, we didn't need to put options
parameter into every call as we had to in the Node.js version:
So when you see example like this in JavaScript
await webkit.launch({ headless: false });
It translates into Python like this:
webkit.launch(headless=False)
If you are using an IDE, it'll suggest parameters that are available in every call.
Another difference is that in the JavaScript version, page.evaluate
accepts JavaScript functions, while this does not make any sense in the Python version.
In JavaScript it will be documented as:
const result = await page.evaluate(([x, y]) => {
return Promise.resolve(x * y);
}, [7, 8]);
console.log(result); // prints "56"
And in Python that would look like:
result = page.evaluate("""
([x, y]) => {
return Promise.resolve(x * y);
}""",
[7, 8])
print(result) # prints "56"
The library will detect that what are passing it is a function and will invoke it with the given parameters. You can opt out of this function detection and pass force_expr=True
to all evaluate functions, but you probably will never need to do that.
Python enabled us to do some of the things that were not possible in the Node.js version and we used the opportunity. Instead of using the page.waitFor*
methods, we recommend using corresponding page.expect_*
context manager.
In JavaScript it will be documented as:
const [ download ] = await Promise.all([
page.waitForEvent('download'), // <-- start waiting for the download
page.click('button#delayed-download') // <-- perform the action that directly or indirectly initiates it.
]);
const path = await download.path();
And in Python that would look much simpler:
with page.expect_download() as download_info:
page.click("button#delayed-download")
download = download_info.value
path = download.path()
Similarly, for waiting for the network response:
const [response] = await Promise.all([
page.waitForResponse('**/api/fetch_data'),
page.click('button#update'),
]);
Becomes
with page.expect_response("**/api/fetch_data"):
page.click("button#update")
Yes, Playwright for Python is ready. We are still not at the version v1.0, so minor breaking API changes could potentially happen. But a) this is unlikely and b) we will only do that if we know it improves your experience with the new library. We'd like to collect your feedback before we freeze the API for v1.0.
Note: We don't yet support some of the edge-cases of the vendor-specific APIs such as collecting chromium trace, coverage report, etc.