How to make an ethical crawler in Python

Web crawling or web scraping is the automation of the access to one or more websites, in order to extract (scrape) data from them. With a high level language like Python, it can even be done easily, but there are some considerations to be taken into account.

Cover image by Chase McBride from Pexels

I assume you know what web crawling is and have basic knowledge about how to make http requests in Python.

Practise ethical web scraping

There's a risk in web scraping: it can be harsh for web servers. Automated requests can saturate them, so it's not rare that administrators put some measures to prevent scraping in their sites, to the point of blocking certain user agent or IP address. For this reason, it is very important to do it carefully.

Our solution will be a get function to wrap requests.get and implement some respectful practices.

Imports

Here are the main libraries that we will use in our ethical web crawler.

import pathlib  # to manage the filesystem
import requests # to make http requests
from time import sleep # for the delays

from urllib.robotparser import RobotFileParser # to parse robots.txt
from urllib.parse import urlparse # to parse urls

False Responses

First of all, we need a new class to replace HTTP responses in some cases. For our needs, we will only give it the fields of status_code and content.

class FalseResponse:
    def __init__(self, code, content):
        self.status_code = code
        self.content = content

Make sure you are allowed

We also need to write a function ask_robots to check the robots.txt file of the site before a petition. This way we know if the admins allow our bot to crawl a specific url. We will keep visited files in _rp in case we check the same robots.txt more than once.

_rp = {}

def ask_robots(url: str, useragent="*": str) -> bool:
    url_struct = urlparse(url)
    base = url_struct.netloc
    # look up in the cache or update it
    if base not in _rp:
        _rp[base] = RobotFileParser()
        _rp[base].set_url(url_struct.scheme + "://" + base + "/robots.txt")
        _rp[base].read()
    return _rp[base].can_fetch(useragent, url)

The default useragent is "*", which looks for rules applied to all crawlers (sometimes specific crawlers have specific rules).

The get function

Now we have what we need to specify our main function.

I will write each chunk of the function followed by a little explanation.

Arguments

def get(url: str, use_cache=True: bool, delay=2: int):
    useragent = "My Crawler"

1. url. The URL you want to get.
2. use_cache. Flag to allow the program to reuse previous responses instead of making a new one.
3. delay. Time in seconds to wait before a request, to avoid saturating the servers if we make more than one.

Check your own cache

    # If a cached answer exists and is acceptable, then return the cached one.
    url_cache = pathlib.Path.cwd() / "cache" / url.replace("/", "_")
    if use_cache and url_cache.exists():
        with open(url_cache, "r") as fh:
            res = fh.read().replace("\\r\\n", "")
        return FalseResponse(200, res)

If the user has allowed the program to use a cached response, then look up a directory called cache where the previous responses are stored and return that.

Ask the robots.txt

    # If the robots.txt doesn't allow the scraping, return forbidden status
    if not ask_robots(url, useragent):
        return FalseResponse(403, "robots.txt forbids it")

Now we use the function we defined earlier. If the robots.txt file excludes the desired URL, we use a status code of 403, to express that although the content exists, the retrieval is not allowed for this user (in this case, the crawler).

Wait and make the request

    # Make the request after the specified delay
    headers = {"User-Agent": useragent}
    sleep(delay)
    res = requests.get(url, timeout=10, headers=headers)

We identify our crawler before making the request using the User-Agent header and wait the specified time before it.

Stop if told so

    # Exit the program if the scraping was penalized
    if res.status_code == 429:  # too many requests
        exit()

Overwhelmed servers may return a code 429 to tell you that you have done too many requests in too little time. Usually these responses include a Retry-After header to tell us how much to wait before a new request. You can use it to try again, but to keep this example simple, we just exit the program.

Store the response and return it

    # Cache the response if allowed by user
    if use_cache:
        url_cache.parent.mkdir(parents=True, exist_ok=True)
        with open(url_cache, "w") as fh:
            fh.write(res.content.decode(res.encoding))

    return res

Again, if the user allowed the use of a cache, we write to it using the same path as before.

Conclusion

This is a generic approach, but it can be used as a base for crawlers with more concrete needs.

This way, we avoid forbidden scraping, unnecessary redundant petitions and optimize traffic. If you practise ethical web-scraping, you will make the work of maintainers easier and will be able to scrape your data with a clean conscience.

Reference

A Guide to Ethical Web Scraping