I've written a script in python to parse different names and phone numbers of various restaurant names from yelp.com. The scraper is doing its job just fine. The most important feature of this scraper is that it can handle pagination on the fly (if there is any) no matter how many pages it traverses. I tried to create it following the guidelines of OOP. However, I suppose there are still some options to make it better, as in isolating the while True loop by storing it in another function.

This is the script:

import requests
from urllib.parse import quote_plus
from bs4 import BeautifulSoup

class YelpScraper:
    link = 'https://www.yelp.com/search?find_desc={}&find_loc={}&start={}'

    def __init__(self, name, location, num="0"):
        self.name = quote_plus(name)
        self.location = quote_plus(location)
        self.num = quote_plus(num)
        self.base_url = self.link.format(self.name,self.location,self.num)
        self.session = requests.Session()

    def get_info(self):
        s = self.session
        s.headers = {'User-Agent': 'Mozilla/5.0'}

        while True:
            res = s.get(self.base_url)
            soup = BeautifulSoup(res.text, "lxml")
            for items in soup.select("div.biz-listing-large"):
                name = items.select_one(".biz-name span").get_text(strip=True)
                    phone = items.select_one("span.biz-phone").get_text(strip=True)
                except AttributeError: phone = ""
                print("Name: {}\nPhone: {}\n".format(name,phone))

            link = soup.select_one(".pagination-links .next")
            if not link:break
            self.base_url = "https://www.yelp.com" + link.get("href")

if __name__ == '__main__':
    scrape = YelpScraper("Restaurants","San Francisco, CA")

1 Answer 1

  1. You don't need to quote parameters yourself, requests can do it for you;
  2. You don't need a class for that, a simple function will suffice; I’d extract retrieving content from a URL as another function though;
  3. Separate logic from presentation: have your function return a list of name/phone pairs and have the calling code responsible of printing it. Better, turn the function into a generator and yield the pairs as you go;
  4. There is no need to decode the content before parsing it: the lxml parser work best with a sequence of bytes as it can inspect the <head> to use the appropriate encoding.

Proposed improvements:

import requests
from bs4 import BeautifulSoup

def url_fetcher(session, route, base_url='https://www.yelp.com', **kwargs):
    params = kwargs if kwargs else None

    return session.get(base_url + route, params=params)

def yelp_scraper(name, location, num=0):
    session = requests.Session()
    session.headers = {'User-Agent': 'Mozilla/5.0'}

    response = url_fetcher(session, '/search', find_desc=name, find_loc=location, start=num)
    while True:
        soup = BeautifulSoup(response.content, 'lxml')
        for items in soup.select('div.biz-listing-large'):
            name = items.select_one('.biz-name span').get_text(strip=True)
                phone = items.select_one('span.biz-phone').get_text(strip=True)
            except AttributeError:
                phone = ''
            yield name, phone

        link = soup.select_one('.pagination-links .next')
        if not link:
        response = url_fetcher(session, link.get('href'))

if __name__ == '__main__':
    for name, phone in yelp_scraper('Restaurants', 'San Francisco, CA'):
        print('Name:', name)
        print('Phone:', phone)
  • \$\begingroup\$ I did not know that the lxml parser can directly take the response.content and does not need response.text! \$\endgroup\$
    – Graipher
    Commented Jun 11, 2018 at 10:05
  • \$\begingroup\$ @Graipher reading from the docs, there might even be a possibility that the raw response object can be used instead: lxml.de/tutorial.html#the-parse-function \$\endgroup\$ Commented Jun 11, 2018 at 17:01
  • \$\begingroup\$ Thanks @Mathias Ettinger for showing a new way of doing things. My intention was to do the same using class as I'm a new to OOP. If it were not for class I could have done the same using a single function. Povided +1. Thanks. \$\endgroup\$
    – SIM
    Commented Jun 12, 2018 at 17:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.