zeroinstall/injector/download.py

   1 """
   2 Handles URL downloads.
   3
   4 This is the low-level interface for downloading interfaces, implementations, icons, etc.
   5
   6 @see: L{fetch} higher-level API for downloads that uses this module
   7 """
   8
   9 # Copyright (C) 2009, Thomas Leonard
  10 # See the README file for details, or visit http://0install.net.
  11
  12 import tempfile, os, sys, subprocess
  13
  14 if __name__ == '__main__':
  15         sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.dirname(__file__))))
  16
  17 from zeroinstall import SafeException
  18 from zeroinstall.support import tasks
  19 from logging import info, debug
  20 from zeroinstall import _
  21
  22 download_starting = "starting"  # Waiting for UI to start it
  23 download_fetching = "fetching"  # In progress
  24 download_complete = "complete"  # Downloaded and cached OK
  25 download_failed = "failed"
  26
  27 RESULT_OK = 0
  28 RESULT_FAILED = 1
  29 RESULT_NOT_MODIFIED = 2
  30
  31 class DownloadError(SafeException):
  32         """Download process failed."""
  33         pass
  34
  35 class DownloadAborted(DownloadError):
  36         """Download aborted because of a call to L{Download.abort}"""
  37         def __init__(self, message):
  38                 SafeException.__init__(self, message or _("Download aborted at user's request"))
  39
  40 class Download(object):
  41         """A download of a single resource to a temporary file.
  42         @ivar url: the URL of the resource being fetched
  43         @type url: str
  44         @ivar tempfile: the file storing the downloaded data
  45         @type tempfile: file
  46         @ivar status: the status of the download
  47         @type status: (download_starting | download_fetching | download_failed | download_complete)
  48         @ivar errors: data received from the child's stderr
  49         @type errors: str
  50         @ivar expected_size: the expected final size of the file
  51         @type expected_size: int | None
  52         @ivar downloaded: triggered when the download ends (on success or failure)
  53         @type downloaded: L{tasks.Blocker}
  54         @ivar hint: hint passed by and for caller
  55         @type hint: object
  56         @ivar child: the child process
  57         @type child: subprocess.Popen
  58         @ivar aborted_by_user: whether anyone has called L{abort}
  59         @type aborted_by_user: bool
  60         @ivar unmodified: whether the resource was not modified since the modification_time given at construction
  61         @type unmodified: bool
  62         """
  63         __slots__ = ['url', 'tempfile', 'status', 'errors', 'expected_size', 'downloaded',
  64                      'hint', 'child', '_final_total_size', 'aborted_by_user',
  65                      'modification_time', 'unmodified']
  66
  67         def __init__(self, url, hint = None, modification_time = None):
  68                 """Create a new download object.
  69                 @param url: the resource to download
  70                 @param hint: object with which this download is associated (an optional hint for the GUI)
  71                 @param modification_time: string with HTTP date that indicates last modification time.
  72                   The resource will not be downloaded if it was not modified since that date.
  73                 @postcondition: L{status} == L{download_starting}."""
  74                 self.url = url
  75                 self.status = download_starting
  76                 self.hint = hint
  77                 self.aborted_by_user = False
  78                 self.modification_time = modification_time
  79                 self.unmodified = False
  80
  81                 self.tempfile = None            # Stream for result
  82                 self.errors = None
  83                 self.downloaded = None
  84
  85                 self.expected_size = None       # Final size (excluding skipped bytes)
  86                 self._final_total_size = None   # Set when download is finished
  87
  88                 self.child = None
  89
  90         def start(self):
  91                 """Create a temporary file and begin the download.
  92                 @precondition: L{status} == L{download_starting}"""
  93                 assert self.status == download_starting
  94                 assert self.downloaded is None
  95
  96                 self.tempfile = tempfile.TemporaryFile(prefix = 'injector-dl-data-')
  97
  98                 task = tasks.Task(self._do_download(), "download " + self.url)
  99                 self.downloaded = task.finished
 100
 101         def _do_download(self):
 102                 """Will trigger L{downloaded} when done (on success or failure)."""
 103                 self.errors = ''
 104
 105                 # Can't use fork here, because Windows doesn't have it
 106                 assert self.child is None, self.child
 107                 child_args = [sys.executable, '-u', __file__, self.url]
 108                 if self.modification_time: child_args.append(self.modification_time)
 109                 self.child = subprocess.Popen(child_args, stderr = subprocess.PIPE, stdout = self.tempfile)
 110
 111                 self.status = download_fetching
 112
 113                 # Wait for child to exit, collecting error output as we go
 114
 115                 while True:
 116                         yield tasks.InputBlocker(self.child.stderr, "read data from " + self.url)
 117
 118                         data = os.read(self.child.stderr.fileno(), 100)
 119                         if not data:
 120                                 break
 121                         self.errors += data
 122
 123                 # Download is complete...
 124
 125                 assert self.status is download_fetching
 126                 assert self.tempfile is not None
 127                 assert self.child is not None
 128
 129                 status = self.child.wait()
 130                 self.child = None
 131
 132                 errors = self.errors
 133                 self.errors = None
 134
 135                 if status == RESULT_NOT_MODIFIED:
 136                         debug("%s not modified", self.url)
 137                         self.tempfile = None
 138                         self.unmodified = True
 139                         self.status = download_complete
 140                         self._final_total_size = 0
 141                         self.downloaded.trigger()
 142                         return
 143
 144                 if status and not self.aborted_by_user and not errors:
 145                         errors = _('Download process exited with error status '
 146                                            'code %s') % hex(status)
 147
 148                 self._final_total_size = self.get_bytes_downloaded_so_far()
 149
 150                 stream = self.tempfile
 151                 self.tempfile = None
 152
 153                 try:
 154                         if self.aborted_by_user:
 155                                 raise DownloadAborted(errors)
 156
 157                         if errors:
 158                                 raise DownloadError(errors.strip())
 159
 160                         # Check that the download has the correct size, if we know what it should be.
 161                         if self.expected_size is not None:
 162                                 size = os.fstat(stream.fileno()).st_size
 163                                 if size != self.expected_size:
 164                                         raise SafeException(_('Downloaded archive has incorrect size.\n'
 165                                                         'URL: %(url)s\n'
 166                                                         'Expected: %(expected_size)d bytes\n'
 167                                                         'Received: %(size)d bytes') % {'url': self.url, 'expected_size': self.expected_size, 'size': size})
 168                 except:
 169                         self.status = download_failed
 170                         _unused, ex, tb = sys.exc_info()
 171                         self.downloaded.trigger(exception = (ex, tb))
 172                 else:
 173                         self.status = download_complete
 174                         self.downloaded.trigger()
 175
 176         def abort(self):
 177                 """Signal the current download to stop.
 178                 @postcondition: L{aborted_by_user}"""
 179                 if self.child is not None:
 180                         info(_("Killing download process %s"), self.child.pid)
 181                         import signal
 182                         os.kill(self.child.pid, signal.SIGTERM)
 183                         self.aborted_by_user = True
 184                 else:
 185                         self.status = download_failed
 186
 187         def get_current_fraction(self):
 188                 """Returns the current fraction of this download that has been fetched (from 0 to 1),
 189                 or None if the total size isn't known.
 190                 @return: fraction downloaded
 191                 @rtype: int | None"""
 192                 if self.status is download_starting:
 193                         return 0
 194                 if self.tempfile is None:
 195                         return 1
 196                 if self.expected_size is None:
 197                         return None             # Unknown
 198                 current_size = self.get_bytes_downloaded_so_far()
 199                 return float(current_size) / self.expected_size
 200
 201         def get_bytes_downloaded_so_far(self):
 202                 """Get the download progress. Will be zero if the download has not yet started.
 203                 @rtype: int"""
 204                 if self.status is download_starting:
 205                         return 0
 206                 elif self.status is download_fetching:
 207                         return os.fstat(self.tempfile.fileno()).st_size
 208                 else:
 209                         return self._final_total_size
 210
 211         def __str__(self):
 212                 return _("<Download from %s>") % self.url
 213
 214 if __name__ == '__main__':
 215         def _download_as_child(url, if_modified_since):
 216                 from httplib import HTTPException
 217                 from urllib2 import urlopen, Request, HTTPError, URLError
 218                 try:
 219                         #print "Child downloading", url
 220                         if url.startswith('http:') or url.startswith('https:') or url.startswith('ftp:'):
 221                                 req = Request(url)
 222                                 if url.startswith('http:') and if_modified_since:
 223                                         req.add_header('If-Modified-Since', if_modified_since)
 224                                 src = urlopen(req)
 225                         else:
 226                                 raise Exception(_('Unsupported URL protocol in: %s') % url)
 227
 228                         try:
 229                                 sock = src.fp._sock
 230                         except AttributeError:
 231                                 sock = src.fp.fp._sock  # Python 2.5 on FreeBSD
 232                         while True:
 233                                 data = sock.recv(256)
 234                                 if not data: break
 235                                 os.write(1, data)
 236
 237                         sys.exit(RESULT_OK)
 238                 except (HTTPError, URLError, HTTPException), ex:
 239                         if isinstance(ex, HTTPError) and ex.code == 304: # Not modified
 240                                 sys.exit(RESULT_NOT_MODIFIED)
 241                         print >>sys.stderr, "Error downloading '" + url + "': " + (str(ex) or str(ex.__class__.__name__))
 242                         sys.exit(RESULT_FAILED)
 243         assert (len(sys.argv) == 2) or (len(sys.argv) == 3), "Usage: download URL [If-Modified-Since-Date], not %s" % sys.argv
 244         if len(sys.argv) >= 3:
 245                 if_modified_since_date = sys.argv[2]
 246         else:
 247                 if_modified_since_date = None
 248         _download_as_child(sys.argv[1], if_modified_since_date)