PIP how escape character # in password?

问题

want to continue question How to get pip to work behind a proxy server

I have Windows Server and Python 3.5 (64).

In password my user include #.

I try to use some solve:

"C:\Program Files\Python35\scripts\pip.exe" install --proxy http://proxy_user:pwd#123@proxy.su:1111 TwitterApi

"C:\Program Files\Python35\scripts\pip.exe" install --proxy "http://proxy_user:pwd#123"@proxy.su:1111 TwitterApi

"C:\Program Files\Python35\scripts\pip.exe" install --proxy http://"proxy_user:pwd#123"@proxy.su:1111 TwitterApi

"C:\Program Files\Python35\scripts\pip.exe" install --proxy http://proxy_user:"pwd#123"@proxy.su:1111 TwitterApi

BUT to get error

  File "c:\program files\python35\lib\site-packages\pip\_vendor\requests\package
s\urllib3\util\url.py", line 189, in parse_url
    raise LocationParseError(url)
pip._vendor.requests.packages.urllib3.exceptions.LocationParseError: Failed to p
arse: proxy_user:pwd

How escape character # in this case?

回答1:

Quick way out: Enter it in the encoded form i.e. # -> %23

A better way for pip to handle this might be to add a --proxy-auth flag that takes : and does the encoding for the user before adding it to the Proxy URL.

Issue - This is something not allowed:

Strictly speaking, the literal # character is not valid in the userinfo portion of a URI, according to RFC 3986, and should be percent encoded. However, it's not exactly a surprise that many tools handle this ok: there's clearly no actual ambiguity about that character. Note, however, that if there were an @ symbol in the password you'd definitely have to urlencode it: for that reason, it's a good habit to get into to urlencode your passwords before they go into URIs.

The response to a submitted issue parse_url fails when given credentials in the URL with '/', '#', or '?':

The RFC says specifically:

The authority component is preceded by a double slash ("//") and is terminated by the next slash ("/"), question mark ("?"), or number sign ("#") character, or by the end of the URI. In other words, the current behaviour is correct in expecting the authority to be terminated by the first / (or ? or #) it finds after the preceeding //. Am I sympathetic to people trying to use proxy URIs with pip? Absolutely. I think hacking together something that violates the RFC has the potential for nasty surprises later on.

回答2:

else examples


    $user = str_replace('@', '%40', $user);
    $pass = str_replace('%', '%25', $pass); // don't down! (%)
    $pass = str_replace('#', '%23', $pass);
    $pass = str_replace('@', '%40', $pass);
    $pass = str_replace(':', '%3a', $pass);
    $pass = str_replace(';', '%3b', $pass);
    $pass = str_replace('?', '%3f', $pass);
    $pass = str_replace('$', '%24', $pass);
    $pass = str_replace('!', '%21', $pass);
    $pass = str_replace('/', '%2f', $pass);
    $pass = str_replace('\'', '%27', $pass);
    $pass = str_replace('"', '%22', $pass);

来源：https://stackoverflow.com/questions/39207316/pip-how-escape-character-in-password

标签

python

python-3.x

pip

http-proxy