In the era of Big Data, the web is an endless source of information. For this reason, there are plenty of good tools/frameworks to perform scraping of web pages.
So, I guess, in an ideal world there should be no need of a new web scraping framework. Nevertheless, there are always subtle differences between theory and practice. The case of web scraping made no exceptions.
Real world web pages are often full of javascript codes that alter the DOM as the user requests/navigates pages. Consequently, scraping javascript intensive web pages could be impossible.
Such considerations were the sparks that gave birth to CHeSF, the Chrome Headless Scraping Framework. To make a long story short, CHeSF relies on both selenium-python and ChromeDriver to perform scraping of webpages also when javascript makes it impossible.
I know that already exists some nice solutions to this problems, but in my point of view CHeSF is simpler: you just create a class that inherits from it, define the parse method and launch it with a start url.
The framework is still very alpha. You should expect that things could change rapidly. Currently, there is no documentation, nor packaging. There is just an example showing how you could use the framework to easily scrape TripAdvisor reviews. Personally, I used it to collect this dataset, i.e. a collection of more than 220k TripAdvisor reviews.
CHeSF borrows its working philosophy (in part) from Scrapy, i.e. making a scraping tool means creating (at least) a python class.
import sys
import os
# the path to the crhome driver executable
path_to_chrome_driver_exe = 'path_to_chromedriver.exe'
# currently, no packages exists for CHeSF, so use this hack until
# I'll have some free time to implement packaging
path_to_chesf = 'path_to_chesf_in_your_system'
sys.path.insert(0, os.path.abspath(path_to_chesf))
from chesf import CHeSF, MAX_ATTEMPTS
class TripAdvisorScraper(CHeSF):
def __init__(self):
super().__init__(path_to_chrome_driver_exe, debug=False)
# this is the core of the Scraper, you must define it since by
# convention is the callback called with the first url passed,
# after, you can define other callbacks
def parse(self):
# the main pro of CHeSF is that you could use directly
# javascript to parse the page
script = """
let urls = [];
let anchors = document.querySelectorAll("a.property_title.prominent");
for (let a of anchors)
urls.push(a.href);
return urls;
"""
# the array returned from the javascript is automagically
# transformed to a python list (this is selenium magic)
links = self.call_js(script)
for link in links:
print(link)
# you could use both xpath and css selectors (just change the
# method you use)
next_page = self.css('a.nav.next.taLnk.ui_button.primary', timeout=1)
if len(next_page) > 0:
# clicks are immediately executed
self.enqueue_click(next_page[0], self.parse)
start_url = 'https://www.tripadvisor.com/Hotels-g187791-c2-Rome_Lazio-Hotels.html'
scraper = TripAdvisorScraper()
try:
scraper.start(start_url)
except:
scraper.quit()
raiseIn case of questions and/or suggestions, write me a note using my GitHub contact email.
Q. Hey man, it absolutely doesn't work! What's wrong? A. Please, check that your ChromeDriver is suitable for the Chrome version you are using.