用套接字解析wss协议的主机名？

Question

How can resolve a host name with wss protocol by socket?
I tried this but failed:
import socket
s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
ip = socket.gethostbyname('wss://domain.tld')

Answer 1

wss://domain.tld is not a hostname, it's a URL. You can't resolve a URL with a socket, you have to parse it as a URL to get the hostname out of it, and then you can resolve that. It doesn't matter whether the scheme is wss, http, or rsync; any scheme that has a netloc field will work the same way.

For example, using urllib.parse :

>>> from urllib.parse import urlparse # in 2.x it's from urlparse
>>> url = 'wss://domain.tld'
>>> bits = urllib.parse.urlparse(url)
>>> netloc = bits.netloc
>>> netloc
'domain.tld'

So we're done, right?

Nope. A netloc can be either a host, or a host:port. And you can't just split(':') , because IPv6 addresses can have colons in them—but only if they're enclosed in brackets. So, to get the host part of a netloc , you need to do something like this:

>>> host, _, port = netloc.rpartition(':')
>>> if ']' in port: host = netloc
>>> host
'domain.tld'

And now we are done, we've got a hostname or IP address, which we can pass to socket.gethostbyname .

But a couple notes on that.

First, you don't need to create a socket.socket object to call gethostbyname ; it's a top-level function on the module, that doesn't need any socket objects.

Second, gethostbyname doesn't work on IPv6, and has some limitations even on IPv4, so you may want to use getaddrinfo instead.

So, finishing up:

>>> import socket
>>> addresses = socket.getaddrinfo(host, None) # or host, port if you prefer
gaierror: [Errno 8] nodename nor servname provided, or not known

Well, that's to be expected, since our hostname is domain.tld , and there's no such domain. But if we used, say, www.google.com , we'd get back a nice list of a couple dozen IPv4 addresses and, if your system has IPv6 connectivity, a couple of IPv6 ones as well. You can just use the first one, or prefer IPv4 to IPv6 or vice-versa, or discriminate on some other field. (You can also filter on various fields in the first place by passing more arguments to getaddrinfo .)