Windows下pip中的Urllib2错误

Question

我在Windows上用ActiveState Python 2.7.2运行pip时遇到了一些问题。我们使用了一个代理，这可能是问题的一部分。这个代理是不需要认证的。系统的代理设置在Firefox里手动配置或者用一些简单的Python代码都能正常工作：

这个操作是正常的：

urllib.urlopen('http://www.google.com', proxies={'http': 'http://proxy:port'})

它返回了来自google.com的带有头信息的响应：

<addinfourl at 61539976L whose fp = <socket._fileobject object at 0x00000000042924F8>>

在设置了http_proxy环境变量后，我运行了：

pip install loremipsum

结果是：

    Downloading/unpacking loremipsum
      Could not fetch URL http://pypi.python.org/simple/loremipsum: <urlopen error [Errno 11004] getaddrinfo
failed>
      Will skip URL http://pypi.python.org/simple/loremipsum when looking for download links for loremipsum
      Could not fetch URL http://pypi.python.org/simple/: <urlopen error [Errno 11004] getaddrinfo failed>
      Will skip URL http://pypi.python.org/simple/ when looking for download links for loremipsum
      Cannot fetch index base URL http://pypi.python.org/simple/
      Could not fetch URL http://pypi.python.org/simple/loremipsum/: <urlopen error [Errno 11004] getaddrinfo failed>
      Will skip URL http://pypi.python.org/simple/loremipsum/ when looking for download links for loremipsum
      Could not find any downloads that satisfy the requirement loremipsum
    No distributions at all found for loremipsum
    Exception information:
    Traceback (most recent call last):
      File "C:\ActiveState\ActivePython\lib\site-packages\pip\basecommand.py", line 126,  in main
    self.run(options, args)
  File "C:\ActiveState\ActivePython\lib\site-packages\pip\commands\install.py", line 222, in run
    requirement_set.prepare_files(finder, force_root_egg_info=self.bundle, bundle=self.bundle)
  File "C:\ActiveState\ActivePython\lib\site-packages\pip\req.py", line 954, in prepare_files
    url = finder.find_requirement(req_to_install, upgrade=self.upgrade)
  File "C:\ActiveState\ActivePython\lib\site-packages\pip\index.py", line 152, in find_requirement
    raise DistributionNotFound('No distributions at all found for %s' % req)
DistributionNotFound: No distributions at all found for loremipsum

错误11004似乎表示有名称解析的问题，这在使用代理时应该是与代理访问（或无视）有关的问题。

我可以在Linux上测试一个类似的设置（http_proxy变量、相同的代理、不同的Python）。运行上面的命令效果很好。还有，在Windows机器上的浏览器中访问这个网址也能正常工作（显示一堆egg和zip文件）。

我查看了pip的代码，想找出它出错的地方，并在交互式会话中运行了代码。我发现所有下载的准备和操作基本上都在C:\ActiveState\ActivePython\lib\site-packages\pip\downloads.py中进行。在setup()（第125行）中，准备了ProxyHandler并构建了opener，这个opener被存储在urllib2中，以便后续调用使用。当我在交互式环境中运行时，我发现需要为https代理添加一个条目。我还添加了调试信息的打印。这让我在iPython的交互式运行中得到了：

In [1]: import urllib2

In [2]: proxy='proxy:port'

In [3]: proxy_support = urllib2.ProxyHandler({"http": proxy, "ftp": proxy, "https": proxy})

In [4]: opener = urllib2.build_opener(proxy_support, urllib2.CacheFTPHandler, urllib2.HTTPHandler(debugle
vel=1), urllib2.HTTPSHandler(debuglevel=1))

In [5]: urllib2.install_opener(opener)

对于ftp、http和https，代理确实是相同的。我还通过一些打印检查了所有处理命令行参数的代码，确保它们没有干扰代理。代理在downloads.py中的存储方式与上面简化的方式相同（从http_proxy变量读取）。

在弄清楚从pip获取包的URL后，它会进入第74行的__call__()。

首先构造一个请求，使用：

In [6]: url = urllib2.Request('http://pypi.python.org/simple/loremipsum', headers={'Accept-encoding': 'identity'})

然后这个请求会与urllib2.urlopen(url)一起使用：

In [7]: response = urllib2.urlopen(url)
send: 'GET http://pypi.python.org/simple/loremipsum HTTP/1.1\r\nHost: pypi.python.org\r\nUser-Agent: Pyth
on-urllib/2.7\r\nConnection: close\r\nAccept-Encoding: identity\r\n\r\n'
reply: 'HTTP/1.1 301 Moved Permanently\r\n'
header: Server: Varnish
header: Retry-After: 0
header: Location: https://pypi.python.org/simple/loremipsum
header: Content-Length: 0
header: Accept-Ranges: bytes
header: Date: Tue, 12 Aug 2014 16:41:39 GMT
header: Via: 1.1 varnish
header: X-Served-By: cache-fra1234-FRA
header: X-Cache: MISS
header: X-Cache-Hits: 0
header: X-Timer: S1407861699.491394,VS0,VE0
header: Connection: close
header: Age: 0
send: 'CONNECT pypi.python.org:443 HTTP/1.0\r\n'
send: '\r\n'
send: 'GET /simple/loremipsum HTTP/1.1\r\nHost: pypi.python.org\r\nUser-Agent: Python-urllib/2.7\r\nConne
ction: close\r\nAccept-Encoding: identity\r\n\r\n'
reply: 'HTTP/1.1 301 Moved Permanently\r\n'
header: Date: Tue, 12 Aug 2014 16:41:40 GMT
header: Server: nginx/1.6.0
header: Location: /simple/loremipsum/
header: Cache-Control: max-age=600, public
header: Strict-Transport-Security: max-age=31536000; includeSubDomains
header: Via: 1.1 varnish
header: Content-Length: 0
header: Accept-Ranges: bytes
header: Via: 1.1 varnish
header: Age: 44282
header: X-Served-By: cache-iad2135-IAD, cache-fra1231-FRA
header: X-Cache: MISS, HIT
header: X-Cache-Hits: 0, 1
header: X-Timer: S1407861700.831757,VS0,VE0
header: Connection: close
send: 'CONNECT pypi.python.org:443 HTTP/1.0\r\n'
send: '\r\n'
send: 'GET /simple/loremipsum/ HTTP/1.1\r\nHost: pypi.python.org\r\nUser-Agent: Python-urllib/2.7\r\nConn
ection: close\r\nAccept-Encoding: identity\r\n\r\n'
reply: 'HTTP/1.1 200 OK\r\n'
header: Date: Tue, 12 Aug 2014 16:41:41 GMT
header: Server: nginx/1.6.0
header: Content-Type: text/html; charset=utf-8
header: X-PYPI-LAST-SERIAL: 794358
header: Cache-Control: max-age=600, public
header: Strict-Transport-Security: max-age=31536000; includeSubDomains
header: Via: 1.1 varnish
header: Content-Length: 913
header: Accept-Ranges: bytes
header: Via: 1.1 varnish
header: Age: 67708
header: X-Served-By: cache-iad2121-IAD, cache-fra1231-FRA
header: X-Cache: HIT, HIT
header: X-Cache-Hits: 1, 1
header: X-Timer: S1407861701.174694,VS0,VE0
header: Vary: Accept-Encoding
header: Connection: close

这似乎是一个可以接受的答案。我在pip中有完全相同的代码，但它却失败了。

我漏掉了什么？为什么我的交互式会话可以工作，而pip却不行？

windows urllib2 环境变量调试信息代理设置交互式会话名称解析 pip错误

Windows下pip中的Urllib2错误

1 个回答

撰写回答