scrapy-requests 0.2.0

Scrapy with requests-html

scrapy-requests

Scrapy middleware to asynchronously handle javascript pages using requests-html.

requests-html uses pyppeteer to load javascript pages, and handles user-agent specification for you. Using requests-html is very intuitive and simple. Check out their documentation.

Requirements

• Python >= 3.6
• Scrapy >= 2.0
• requests-html

Installation

 pip install scrapy-requests

Configuration

Make twisted use Asyncio event loop And add RequestsMiddleware to the downloader middleware

settings.py

TWISTED_REACTOR = 'twisted.internet.asyncioreactor.AsyncioSelectorReactor'

DOWNLOADER_MIDDLEWARES = {
    'scrapy_requests.RequestsMiddleware': 800
}

Usage

Use scrapy_requests.HtmlRequest instead of scrapy.Request

from scrapy_requests import HtmlRequest

yield HtmlRequest(url=url, callback=self.parse)

The requests will be handled by requests_html, and the request will add an additional meta varialble page containing the HTML object.

def parse(self, response):
    page = response.request.meta['page']
    page.html.render()

Additional settings

If you would like the page to be rendered by pyppeteer - pass True to the render key paramater.

yield HtmlRequest(url=url, callback=self.parse, render=True)

You could choose a more speific functionality for the HTML object.

For example - You could set up a sleep timer before loading the page, and js script execution when loading the page - doing it this way:

script = "document.body.querySelector('.btn').click();"
yield HtmlRequest(url=url, callback=self.parse, render=True, options={'sleep': 2, 'script': script})

You could pass default settings to requests-html session - specifying header, proxies, auth settings etc... You do this by specifying an addtional variable in settings.py

DEFAULT_SCRAPY_REQUESTS_SETTINGS = {
    'verify': False, # Verifying SSL certificates
    'mock_browser': True, # Mock browser user-agent
    'browser_args': ['--no-sandbox', '--proxy-server=x.x.x.x:xxxx'], 
}

Notes

Please star this repo if you found it useful.

Feel free to contribute and propose issues & additional features.

License is MIT.