browser.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
"""
    BROWSER
    -------
    Get internet pages, send post requests, inspect HTML using BeautifulSoup
    or simply check response codes of an URL.

    Example usage:

        # Create new Browser instance
        browser = Browser()

        # Login (or simply post data) to a website
        browser.post('http://url.to/login', {'username':'john',
                                             'password':'pass123'})

        # Get page contents
        html = browser.get('http://url.to/password-protected-page')

        # Get page contents as BeautifulSoup
        soup = browser.soup('http://url.to/password-protected-page')

        # Get the response code for a given URL
        code = browser.code('http://url.to/password-protected-page')

    Copyright (c) 2013 T. Zengerink
    Licensed under MIT License.
    See: https://gist.github.com/raw/3151357/6806e68cb9cc0042b265f25be9bc25dd39f75267/LICENSE.md
"""
import cookielib, sys, urllib, urllib2

from BeautifulSoup import BeautifulSoup


class Browser:
    def __init__(self):
        self.cookie_jar = cookielib.CookieJar()
        self.cookie_proc = urllib2.HTTPCookieProcessor(self.cookie_jar)
        self.opener = urllib2.build_opener(self.cookie_proc)

    def code(self, url):
        """
        Get the HTTP response code for a given URL.
        url -- URL to fetch and get code from.
        """
        try:
            urllib2.urlopen(urllib2.Request(url))
        except urllib2.URLError, e:
            return e.code
        return 200

    def get(self, url):
        """
        Get the given URL and return the contents.
        url -- URL to fetch.
        """
        return self.opener.open(url).read()

    def post(self, url, data):
        """
        Post data to a given URL.
        url -- URL to post to.
        """
        return self.opener.open(url, urllib.urlencode(data))

    def soup(self, url):
        """
        Get BeatifulSoup from the page contents.
        url -- URL to fetch.
        """
        return BeautifulSoup(self.get(url))


def main(args):
    browser = Browser()
    for arg in args:
        print browser.get(arg)

if __name__ == "__main__":
    main(sys.argv[1:])