socli.py 文件源码

def fixGoogleURL(url):
    """
    Fixes the url extracted from HTML when
    performing a google search
    :param url:
    :return: Correctly formatted URL to be used in requests.get
    """
    if "&sa=" in url:
        url=url.split("&")[0]
    if "/url?q=" in url[0:7]:
        url = url[7:] #Removes the "/url?q=" prefix

    if url[:30] == "http://www.google.com/url?url=": #Used to get rid of this header and just retrieve the Stack Overflow link
        url = url[30:]

    if "http" not in url[:4]:
        url = "https://" + url #Add the protocol if it doesn't already exist

    #Makes sure that we stay in the questions section of Stack Overflow
    if not bool(re.search("/questions/[0-9]+", url)) and not bool(re.search("\.com/a/[0-9]", url)):
        return None

    if url[:17] == "https:///url?url=": #Resolves rare bug in which this is a prefix
        url = url[17:]

    return url