package_cache/server.py

   1 # Copyright (C) 2014 W. Trevor King <wking@tremily.us>
   2 #
   3 # This file is part of package-cache.
   4 #
   5 # package-cache is free software: you can redistribute it and/or modify it
   6 # under the terms of the GNU General Public License as published by the Free
   7 # Software Foundation, either version 3 of the License, or (at your option) any
   8 # later version.
   9 #
  10 # package-cache is distributed in the hope that it will be useful, but WITHOUT
  11 # ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
  12 # FOR A PARTICULAR PURPOSE.  See the GNU General Public License for more
  13 # details.
  14 #
  15 # You should have received a copy of the GNU General Public License along with
  16 # package-cache.  If not, see <http://www.gnu.org/licenses/>.
  17
  18 import calendar as _calendar
  19 import email.utils as _email_utils
  20 import mimetypes as _mimetypes
  21 import os as _os
  22 import urllib.error as _urllib_error
  23 import urllib.request as _urllib_request
  24
  25 from . import __version__
  26
  27
  28 class InvalidFile (ValueError):
  29     def __init__(self, url):
  30         super(InvalidFile, self).__init__('invalid file {!r}'.format(url))
  31         self.url = url
  32
  33
  34 class Server (object):
  35     def __init__(self, sources, cache):
  36         self.sources = sources
  37         self.cache = cache
  38         self.opener = _urllib_request.build_opener()
  39         self.opener.addheaders = [
  40             ('User-agent', 'Package-cache/{}'.format(__version__)),
  41             ]
  42
  43     def __call__(self, environ, start_response):
  44         try:
  45             return self._serve_request(
  46                 environ=environ, start_response=start_response)
  47         except InvalidFile:
  48             start_response('404 Not Found', [])
  49         except _urllib_error.HTTPError as e:
  50             print('{} {}'.format(e.code, e.reason))
  51             start_response('{} {}'.format(e.code, e.reason), [])
  52         return [b'']
  53
  54     def _serve_request(self, environ, start_response):
  55         method = environ['REQUEST_METHOD']
  56         url = environ.get('PATH_INFO', None)
  57         if url is None:
  58             raise InvalidFile(url=url)
  59         cache_path = self._get_cache_path(url=url)
  60         if not _os.path.exists(path=cache_path):
  61             self._get_file_from_sources(url=url, path=cache_path)
  62         if not _os.path.isfile(path=cache_path):
  63             raise InvalidFile(url=url)
  64         return self._serve_file(
  65             path=cache_path, environ=environ, start_response=start_response)
  66
  67     def _get_cache_path(self, url):
  68         relative_path = url.lstrip('/').replace('/', _os.path.sep)
  69         cache_path = _os.path.abspath(_os.path.join(self.cache, relative_path))
  70         check_relative_path = _os.path.relpath(
  71             path=cache_path, start=self.cache)
  72         if check_relative_path.startswith(_os.pardir + _os.path.sep):
  73             raise InvalidFile(url=url)
  74         return cache_path
  75
  76     def _get_file_from_sources(self, url, path):
  77         dirname = _os.path.dirname(path)
  78         if not _os.path.isdir(dirname):
  79             _os.makedirs(dirname, exist_ok=True)
  80         for i, source in enumerate(self.sources):
  81             source_url = source.rstrip('/') + url
  82             try:
  83                 self._get_file(url=source_url, path=path)
  84             except _urllib_error.HTTPError:
  85                 if i == len(self.sources) - 1:
  86                     raise
  87             else:
  88                 return
  89
  90     def _get_file(self, url, path):
  91         with self.opener.open(url) as response:
  92             last_modified = response.getheader('Last-Modified', None)
  93             content_length = int(response.getheader('Content-Length'))
  94             with open(path, 'wb') as f:
  95                 block_size = 8192
  96                 while True:
  97                     data = response.read(block_size)
  98                     f.write(data)
  99                     if len(data) < block_size:
 100                         break
 101         if last_modified:
 102             mtime = _calendar.timegm(_email_utils.parsedate(last_modified))
 103             _os.utime(path=path, times=(mtime, mtime))
 104
 105     def _serve_file(self, path, environ, start_response):
 106         headers = {
 107             'Content-Length': self._get_content_length(path=path),
 108             'Content-Type': self._get_content_type(path=path),
 109             'Last-Modified': self._get_last_modified(path=path),
 110             }
 111         f = open(path, 'rb')
 112         if 'wsgi.file_wrapper' in environ:
 113             file_iterator = environ['wsgi.file_wrapper'](f)
 114         else:
 115             file_iterator = iter(lambda: f.read(block_size), '')
 116         start_response('200 OK', list(headers.items()))
 117         return file_iterator
 118
 119     def _get_content_length(self, path):
 120         """Content-Length value per RFC 2616
 121
 122         Content-Length:
 123           https://tools.ietf.org/html/rfc2616#section-14.13
 124         """
 125         return str(_os.path.getsize(path))
 126
 127     def _get_content_type(self, path):
 128         """Content-Type value per RFC 2616
 129
 130         Content-Type:
 131           https://tools.ietf.org/html/rfc2616#section-14.17
 132         Media types:
 133           https://tools.ietf.org/html/rfc2616#section-3.7
 134         """
 135         mimetype, charset = _mimetypes.guess_type(url=path)
 136         if charset:
 137             return '{}; charset={}'.format(mimetype, charset)
 138         else:
 139             return mimetype
 140
 141     def _get_last_modified(self, path):
 142         """Last-Modified value per RFC 2616
 143
 144         Last-Modified:
 145           https://tools.ietf.org/html/rfc2616#section-14.29
 146         Date formats:
 147           https://tools.ietf.org/html/rfc2616#section-3.3.1
 148           https://tools.ietf.org/html/rfc1123#page-55
 149           https://tools.ietf.org/html/rfc822#section-5
 150         """
 151         mtime = _os.path.getmtime(path)
 152         return _email_utils.formatdate(
 153             timeval=mtime, localtime=False, usegmt=True)