1#@!#!123s

: / proc / 2 / root / proc / 2 / root / opt / imh-python / lib / python2.7 / site-packages / cmu_ded / Libs /
Filename : checker.py
back
"""
Definitions specifically to check things.
By Richard Kostyn.
"""

import ssl
import urllib2
from operator import xor
from urllib2 import HTTPError
from urllib2 import URLError


def check_urls(url1, url2):
    """
    Takes in two URL's and separately goes through and find if the other URL doesn't have it.
    Depending on the response code, we should be good.
    Also finding a way around HTTP 501, 404 for wordpress and 500.
    For the future I want to print in the STR what error code.
    :param url1: First URL.
    :param url2: Second URL.
    :return: An array full of everything missing.  The array will be emtpy if there is nothing.
    """
    mismatch = []
    # Not checking SSL's to validate.
    ctx = ssl.create_default_context()
    ctx.check_hostname = False
    ctx.verify_mode = ssl.CERT_NONE

    # Sending blank auth to go around any password protected website without urllib asking for a password.
    # If it does ask for a password it will stall the process.
    headers = {}
    req1 = urllib2.Request(url1)
    req2 = urllib2.Request(url2)

    content_url1 = None
    content_url2 = None

    try:
        content_url1 = urllib2.urlopen(req1, context=ctx)
        content_url2 = urllib2.urlopen(req2, context=ctx)
    except HTTPError as e:
        if e.code == 401:
            mismatch.append("Website is behind authentication")
            pass

        elif e.code == 403:
            # If we can't access it then we shouldn't do anything.
            pass

        elif e.code == 404:
            wordpress = False
            if 'link' in list(e.hdrs) and 'wp-json' in e.hdrs['link']:
                wordpress = True

            if not wordpress:
                mismatch.append("404 in one of the URL's.")

        elif e.code == 500:
            print "500 on the URL."
            mismatch.append("500 in the URL")
            pass
        else:
            print "Issues in opening %s and %s" % (url1, url2)
            mismatch.append("Issues in opening %s and %s" % (url1, url2))
            pass

    except URLError as e:
        mismatch.append(e.reason)

    except Exception as e:
        mismatch.append("Generic Failure on %s and %s." %(url1, url2))

    if content_url1 is not None and content_url2 is not None:
        # Checking headers for a file being downloaded.
        url1_contents = content_url1.headers
        url2_contents = content_url2.headers
        if (url1_contents.get('content-length') is None) ^ \
        (url2_contents.get('content-length') is None):
            return ["One of the Temp URL's is downloading a file. Please check the .htaccess"]

        # After checking headers we  the content:
        content_url1 = content_url1.readlines()
        content_url2 = content_url2.readlines()

        for url1_line in content_url1:
            if url1_line not in content_url2:
                mismatch.append('[-]: ' + url1_line)

        for url2_line in content_url2:
            if url2_line not in content_url1:
                mismatch.append('[+]: ' + url2_line)

    return mismatch


def validate_temp_urls(from_host, to_host, user):
    """
    Generated temp_url's and checks them.
    :param from_host: The host the customer is coming from.
    :param to_host: The host the customer is going to.
    :param user: The user you want to check.
    :return: The result.  Its in a list full of everything different.
    [+] is something new on the server, while
    [-] is something missing on the new server.
    """
    result = []

    url1 = gen_temp_url(from_host, user)
    url2 = gen_temp_url(to_host, user)

    result = check_urls(url1, url2)

    return result


def gen_temp_url(host, user):
    """
    Temp URL stuff.
    :param host: Host you want to check
    :param user: User you want to check
    :return: The full temp url.
    """
    return 'http://' + host + '/~' + user


if __name__ == '__main__':
    # Personal tests.
    bad_stuff = check_urls('http://vps12736.inmotionhosting.com/~rnk/', 'http://vps30181.inmotionhosting.com/~rnk/')

    for line in bad_stuff:
        print line