Scraping with Python Selenium and PhantomJS

In previous posts, I covered scraping using mechanize as the browser. Sometimes though a site uses so much Javascript to dynamically render its pages that using a tool like mechanize (which can't handle Javascript) isn't really feasable. For these cases, we have to use a browser that can run the Javascript required to generate the pages.

Overview

PhantomJS is a headless (non-gui) browser. Selenium is a tool for automating browsers. In this post, we'll use the two together to scrape a Javascript heavy site. First we'll navigate to the site and then, after the HTML has been dynamically generated, we'll feed it into BeautifulSoup for parsing.

First let's set up our environment by installing PhantomJS along with the Selenium bindings for Python:

$ mkdir scraper && cd scraper
$ brew install phantomjs
$ virtualenv venv
$ source venv/bin/activate
$ pip install selenium

Now, let's look at the site we'll use for our example, the job search page for the company L-3 Klein Associates. They use the Taleo Applicant Tracking System and the pages are almost entirely generated via Javascript:

https://l3com.taleo.net/careersection/l3_ext_us/jobsearch.ftl

In this post, we'll develop a script that can scrape, and then print out, all of the jobs listed on their Applicant Tracking System.

Let's get started.

Implementation

First, let's sketch out our class, TaleoJobScraper. In the constructor we'll instantiate a webdriver for PhantomJS. Our main method will be scrape(). It will call scrape_job_links() to iterate through the job listings, and then call driver.quit() once it's complete.

#!/usr/bin/env python

import re, urlparse

from selenium import webdriver
from bs4 import BeautifulSoup
from time import sleep

link = 'https://l3com.taleo.net/careersection/l3_ext_us/jobsearch.ftl'

class TaleoJobScraper(object):
    def __init__(self):
        self.driver = webdriver.PhantomJS()
        self.driver.set_window_size(1120, 550)

    def scrape(self):
        jobs = self.scrape_job_links()
        for job in jobs:
            print job

        self.driver.quit()

if __name__ == '__main__':
    scraper = TaleoJobScraper()
    scraper.scrape()

Now let's take a look at the scrape_job_links() method, which is listed next:

def scrape_job_links(self):
    self.driver.get(link)

    jobs = []
    pageno = 2

    while True:
        s = BeautifulSoup(self.driver.page_source)
        r = re.compile(r'jobdetail\.ftl\?job=\d+$')

        for a in s.findAll('a', href=r):
            tr = a.findParent('tr')
            td = tr.findAll('td')

            job = {}
            job['title'] = a.text
            job['url'] = urlparse.urljoin(link, a['href'])
            job['location'] = td[2].text
            jobs.append(job)

        next_page_elem = self.driver.find_element_by_id('next')
        next_page_link = s.find('a', text='%d' % pageno)

        if next_page_link:
            next_page_elem.click()
            pageno += 1
            sleep(.75)
        else:
            break

    return jobs

First, we open the page with driver.get(). After get() returns, we feed the rendered HTML in driver.page_source into BeautifulSoup. Then we match against the href attribute of the job links. For each job link we extract the title, url, and location.

To get all of the jobs, we also need to handle pagination. There's a pager at the bottom of the jobs listings. Below is a screenshot of the pager. A user can click a page number or the Next link to navigate through the listings.

Form Image

We use the Next link to iterate through every page of the results by first finding the Next element using the driver's find_element_by_id method and then calling click() if we're not on the last page.

next_page_elem = self.driver.find_element_by_id('next')
next_page_link = s.find('a', text='%d' % pageno)

if next_page_link:
    next_page_elem.click()
    pageno += 1
else:
    break

To determine if we're on the last page we search for a link whose text equals the current page number plus one. If no such link exists then we've reached the last page of results and break.

If you'd like to see a working version of the code developed in this post, it's available on github here.

Shameless Plug

Have a scraping project you'd like done? I'm available for hire. Contact me with some details about your project and I'll give you a quote.