urllib2源码阅读

翻看两个库的源码，对urllib2代码高内聚、低耦合的特点，体会的还是不深。先写下来，慢慢分析、慢慢领悟吧。
特意弄了个思维导图，对厘清代码的大概结构还是挺有帮助的。我按照函数和类去进行

同urllib一样，urllib2中也可以调用urlopen方法，贴下这个函数的源码:

1  _opener = None
2  def urlopen(url, data=None, timeout=socket._GLOBAL_DEFAULT_TIMEOUT,
3             cafile=None, capath=None, cadefault=False, context=None):
4     global _opener
5     if cafile or capath or cadefault:
6         if context is not None:
7             raise ValueError(
8                 "You can't pass both context and any of cafile, capath, and cadefault"
9             )
10        if not _have_ssl:
11            raise ValueError('SSL support not available')
12        context = ssl.create_default_context(purpose=ssl.Purpose.SERVER_AUTH,
13                                             cafile=cafile,
14                                             capath=capath)
15        https_handler = HTTPSHandler(context=context)
16        opener = build_opener(https_handler)
17    elif context:
18        https_handler = HTTPSHandler(context=context)
19        opener = build_opener(https_handler)
20    elif _opener is None:
21        _opener = opener = build_opener()
22    else:
23        opener = _opener
24    return opener.open(url, data, timeout)

可以看出，urlopen实际上是对build_opener函数返回的对象进行调用，build_opener可以传递HTTPSHandler实例
build_opener使用一个handlers的对象列表创建出一个opener对象，这个opener是OpenerDirector的实例，支持一些默认的handler，包括http、ftp、https(调用时)。

def build_opener(*handlers):
    import types
    def isclass(obj):
        return isinstance(obj, (types.ClassType, type))
    # 实例化OpenerDirector
    opener = OpenerDirector()
    # 默认的handler
    default_classes = [ProxyHandler, UnknownHandler, HTTPHandler,
                       HTTPDefaultErrorHandler, HTTPRedirectHandler,
                       FTPHandler, FileHandler, HTTPErrorProcessor]
    if hasattr(httplib, 'HTTPS'):
        # 添加了HTTPSHandler
        default_classes.append(HTTPSHandler)
    skip = set()
    # handlers参数中有属于default_classes中子类或实例的，加入到skip中
    for klass in default_classes:
        for check in handlers:
            if isclass(check):
                if issubclass(check, klass):
                    skip.add(klass)
            elif isinstance(check, klass):
                skip.add(klass)
  
    for klass in skip:
        default_classes.remove(klass)
    # 把默认类的实例化对象加入到opener中
    for klass in default_classes:
        opener.add_handler(klass())

    for h in handlers:
        if isclass(h):
            h = h()
        opener.add_handler(h)
    return opener

当使用res = urllib2.urlopen('http://python.org')时，过下期间的过程哈(我怕忘)：
urlopen函数会在其中调用build_operner函数，build_opener函数中会实例化OpenerDirector为opener，并把默认的类添加到opener上(add_handler)，返回operner。urlopen调用opener.open方法，去执行响应的请求。
先写到这里吧，之后在继续

转载于:https://www.cnblogs.com/wulalami/p/8258502.html