pytest-pyppeteer

A plugin to run pyppeteer in pytest.

Installation

You can install pytest-pyppeteer via pip:

$ pip install pytest-pyppeteer

or install the latest one on Github:

pip install git+https://github.com/luizyao/pytest-pyppeteer.git

Quickstart

For example, The Shawshank Redemption deserves a 9.0 or higher rating on douban.com.

from dataclasses import dataclass


@dataclass(init=False)
class Elements:
    """Collect locators of page objects, no matter XPath or CSS Selector."""

    # query input
    query = "#inp-query"

    # search button
    apply = ".inp-btn > input:nth-child(1)"

    # the first result
    first_result = "#root > div > div > div > div > div:nth-child(1) > div.item-root a.cover-link"

    # rating
    rating = "#interest_sectl > div.rating_wrap.clearbox > div.rating_self.clearfix > strong"


async def test_lifetimes(browser):
    page = await browser.new_page()
    await page.goto("https://movie.douban.com/")

    await page.type(Elements.query, "The Shawshank Redemption")
    await page.click(Elements.apply)

    await page.waitfor(Elements.first_result)
    await page.click(Elements.first_result)

    await page.waitfor(Elements.rating)
    rating = await page.get_value(Elements.rating)

    assert float(rating) >= 9.0

Usage

Fixtures

browser fixture

Provide an pyppeteer.browser.Browser instance with a new method new_page(), like pyppeteer.browser.Browser.newPage(), new_page() could create a pyppeteer.page.Page instance.

But the pyppeteer.page.Page instance created by new_page() has some new methods:

Method	Type
query_locator	New
waitfor	New
click	Override
type	Override
get_value	New

For example, you can query an element by css or xpath in the same method query_locator instead of original querySelector and xpath.

More details check with page.py in the source code.

browser_factory fixture

Provide to create an pyppeteer.browser.Browser instance.

For example, query the The Shawshank Redemption’s movie and book rating on douban.com at the same time, then compare them.

import asyncio
from dataclasses import dataclass


@dataclass
class Elements:
    query = "#inp-query"
    apply = ".inp-btn > input:nth-child(1)"


@dataclass
class BookElements(Elements):
    url = "https://book.douban.com/"

    result = '(//*[@class="item-root"])[1]/a'
    rating = "#interest_sectl > div > div.rating_self.clearfix > strong"


@dataclass
class MovieElements(Elements):
    url = "https://movie.douban.com/"

    result = "#root > div > div > div > div > div:nth-child(1) > div.item-root a.cover-link"
    rating = "#interest_sectl > div.rating_wrap.clearbox > div.rating_self.clearfix > strong"


async def query_rating(browser, name: str, elements: "Elements"):
    page = await browser.new_page()

    await page.goto(elements.url)

    await page.type(elements.query, name)
    await page.click(elements.apply)

    await page.waitfor(elements.result)
    await page.click(elements.result)

    await page.waitfor(elements.rating)
    rating = await page.get_value(elements.rating)
    return rating


async def test_multiple_browsers(browser_factory):
    browser1 = await browser_factory()
    browser2 = await browser_factory()

    movie_rating, book_rating = await asyncio.gather(
        query_rating(browser1, "The Shawshank Redemption", MovieElements),
        query_rating(browser2, "The Shawshank Redemption", BookElements),
    )

    assert movie_rating == book_rating

Command line options

--executable-path

You can specify the Chromium or Chrome executable path. Otherwise I will use the default install path of Chrome in current platform.

For other platforms, pyppeteer will downloads the recent version of Chromium when called first time. If you don’t prefer this behavior, you can specify an exact path by override this fixture:

@pytest.fixture(scope="session")
def executable_path(executable_path):
    return executable_path or "path/to/Chrome/or/Chromium"

--headless

Run browser in headless mode.

--args

Additional args to pass to the browser instance.

For example, specify a proxy:

$ pytest --args proxy-server "localhost:5555,direct://" --args proxy-bypass-list "192.0.0.1/8;10.0.0.1/8"

Or by override the args fixture:

@pytest.fixture(scope="session")
def args(args) -> List[str]:
    return args + [
        "--proxy-server=localhost:5555,direct://",
        "--proxy-bypass-list=192.0.0.1/8;10.0.0.1/8",
    ]

--window-size

The default browser size is 800*600, you can use this option to change this behavior:

$ pytest --window-size 1200 800

--window-size 0 0 means to starts the browser maximized.

--slow

Slow down the pyppeteer operate in milliseconds. Defaults to 0.0.

Markers

options

You can override some command line options in the specified test.

For example, auto-open a DevTools panel:

import asyncio

import pytest


@pytest.mark.options(devtools=True)
async def test_marker(browser):
    await browser.new_page()
    await asyncio.sleep(2)

License

Distributed under the terms of the MIT license, pytest-pyppeteer is free and open source software.

Issues

If you encounter any problems, please file an issue along with a detailed description.