UrlSplitResult：无法_replace 字段

Question

我正在尝试通过 split-inject-join 将基本身份验证注入 url:

    url = urllib.parse.urlsplit(url)
    new_url = url._replace(username=user, password=password)

但是我对从 urllib.parse.urlsplit 方法得到的 SplitResult 的行为感到惊讶：

>>> v = urlsplit('http://a.b/c/d')
>>> v.username is None and v.password is None  # None, but accessible
True
>>> v._replace(scheme='ftp')  # beautiful!  Just like expected.
SplitResult(scheme='ftp', netloc='a.b', path='/c/d', query='', fragment='')

# however...
>>> v._replace(scheme='ftp', username='u', password='p')
...ValueError: Got unexpected field names: ['username', 'password']

这个SplitResult中的None字段好像不能替换。这很奇怪，因为 the documentation 声称它是是命名元组。

当我用一个自建的命名元组做同样的事情时，'None' 字段可以毫无问题地被替换。

>>> T = namedtuple("T", ("a", "b", "c"))
>>> t = T(a=1, b=2, c=None)
>>> t
T(a=1, b=2, c=None)
>>> t._replace(c=3)
T(a=1, b=2, c=3)

但是替换不存在的字段也会触发同样的异常。

>>> t._replace(d=4)  # should raise, please
...ValueError: Got unexpected field names: ['d']
# I was expecting an AttributeError, but hey...

然而，在这种情况下，意外的字段确实无法访问：

>>> t.d is None
...AttributeError: 'T' object has no attribute 'd'

知道 SplitResult 的不同之处吗？

Answer 1

文档未列出字段 username、password、hostname 或 port 的索引。这表明它们实际上并不是 namedtuple 的一部分，而只是 SplitResult 对象的属性。查看 CPython implementation, it seems like SplitResult inherits from both the namedtuple containing the first few fields which have indices, as well as a mixin which adds the other fields as properties.

UrlSplitResult：无法_replace 字段

UrlSplitResult: can't _replace fields

python

urllib

namedtuple