unpack.py

   1 # Copyright (C) 2006, Thomas Leonard
   2 # See the README file for details, or visit http://0install.net.
   3 # Copied from the injector, as versions before 0.18 don't have it.
   4
   5 import os
   6 import shutil
   7 import traceback
   8 from tempfile import mkstemp
   9 import re
  10 from logging import debug, info, warn
  11 from zeroinstall import SafeException
  12
  13 _recent_gnu_tar = None
  14 def recent_gnu_tar():
  15         global _recent_gnu_tar
  16         if _recent_gnu_tar is None:
  17                 _recent_gnu_tar = False
  18                 version = os.popen('tar --version 2>&1').next()
  19                 if '(GNU tar)' in version:
  20                         try:
  21                                 version = version.split(')', 1)[1].strip()
  22                                 assert version
  23                                 version = map(int, version.split('.'))
  24                                 _recent_gnu_tar = version > [1, 13, 92]
  25                         except:
  26                                 warn("Failed to extract GNU tar version number")
  27                 debug("Recent GNU tar = %s", _recent_gnu_tar)
  28         return _recent_gnu_tar
  29
  30 def _find_in_path(prog):
  31         for d in os.environ['PATH'].split(':'):
  32                 path = os.path.join(d, prog)
  33                 if os.path.isfile(path):
  34                         return path
  35         return None
  36 _pola_run = _find_in_path('pola-run')
  37 if _pola_run:
  38         info('Found pola-run: %s', _pola_run)
  39 else:
  40         info('pola-run not found; archive extraction will not be sandboxed')
  41
  42 def _exec_maybe_sandboxed(writable, prog, *args):
  43         """execlp prog, with (only) the 'writable' directory writable if sandboxing is available.
  44         If no sandbox is available, run without a sandbox."""
  45         prog_path = _find_in_path(prog)
  46         if _pola_run is None:
  47                 os.execlp(prog_path, prog_path, *args)
  48         # We have pola-shell :-)
  49         pola_args = ['--prog', prog_path, '-f', '/']
  50         for a in args:
  51                 pola_args += ['-a', a]
  52         if writable:
  53                 pola_args += ['-fw', writable]
  54         os.execl(_pola_run, _pola_run, *pola_args)
  55
  56 def unpack_archive(url, data, destdir, extract = None):
  57         """Unpack stream 'data' into directory 'destdir'. If extract is given, extract just
  58         that sub-directory from the archive. Works out the format from the name."""
  59         url = url.lower()
  60         if url.endswith('.tar.bz2'):
  61                 extract_tar(data, destdir, extract, '--bzip2')
  62         elif url.endswith('.deb'):
  63                 extract_deb(data, destdir, extract)
  64         elif url.endswith('.rpm'):
  65                 extract_rpm(data, destdir, extract)
  66         elif url.endswith('.tar.gz') or url.endswith('.tgz'):
  67                 extract_tar(data, destdir, extract, '-z')
  68         else:
  69                 raise SafeException('Unknown extension on "%s"; I only know .tgz, .tar.bz2 and .rpm' % url)
  70
  71 def extract_deb(stream, destdir, extract = None):
  72         if extract:
  73                 raise SafeException('Sorry, but the "extract" attribute is not yet supported for Debs')
  74
  75         stream.seek(0)
  76         # ar can't read from stdin, so make a copy...
  77         deb_copy_name = os.path.join(destdir, 'archive.deb')
  78         deb_copy = file(deb_copy_name, 'w')
  79         shutil.copyfileobj(stream, deb_copy)
  80         deb_copy.close()
  81         _extract(stream, destdir, ('ar', 'x', 'archive.deb', 'data.tar.gz'))
  82         os.unlink(deb_copy_name)
  83         data_name = os.path.join(destdir, 'data.tar.gz')
  84         data_stream = file(data_name)
  85         os.unlink(data_name)
  86         _extract(data_stream, destdir, ('tar', 'xzf', '-'))
  87
  88 def extract_rpm(stream, destdir, extract = None):
  89         if extract:
  90                 raise SafeException('Sorry, but the "extract" attribute is not yet supported for RPMs')
  91         fd, cpiopath = mkstemp('-rpm-tmp')
  92         try:
  93                 child = os.fork()
  94                 if child == 0:
  95                         try:
  96                                 try:
  97                                         os.dup2(stream.fileno(), 0)
  98                                         os.dup2(fd, 1)
  99                                         _exec_maybe_sandboxed(None, 'rpm2cpio', '-')
 100                                 except:
 101                                         traceback.print_exc()
 102                         finally:
 103                                 os._exit(1)
 104                 id, status = os.waitpid(child, 0)
 105                 assert id == child
 106                 if status != 0:
 107                         raise SafeException("rpm2cpio failed; can't unpack RPM archive; exit code %d" % status)
 108                 os.close(fd)
 109                 fd = None
 110                 args = ['cpio', '-mid', '--quiet']
 111                 _extract(file(cpiopath), destdir, args)
 112                 # Set the mtime of every directory under 'tmp' to 0, since cpio doesn't
 113                 # preserve directory mtimes.
 114                 os.path.walk(destdir, lambda arg, dirname, names: os.utime(dirname, (0, 0)), None)
 115         finally:
 116                 if fd is not None:
 117                         os.close(fd)
 118                 os.unlink(cpiopath)
 119
 120 def extract_tar(stream, destdir, extract, decompress):
 121         if extract:
 122                 # Limit the characters we accept, to avoid sending dodgy
 123                 # strings to tar
 124                 if not re.match('^[a-zA-Z0-9][- _a-zA-Z0-9.]*$', extract):
 125                         raise SafeException('Illegal character in extract attribute')
 126
 127         if recent_gnu_tar():
 128                 args = ['tar', decompress, '-x', '--no-same-owner', '--no-same-permissions']
 129         else:
 130                 args = ['tar', decompress, '-xf', '-']
 131
 132         if extract:
 133                 args.append(extract)
 134
 135         _extract(stream, destdir, args)
 136
 137 def _extract(stream, destdir, command):
 138         """Run execvp('command') inside destdir in a child process, with a
 139         rewound stream as stdin."""
 140         child = os.fork()
 141         if child == 0:
 142                 try:
 143                         try:
 144                                 os.chdir(destdir)
 145                                 stream.seek(0)
 146                                 os.dup2(stream.fileno(), 0)
 147                                 _exec_maybe_sandboxed(destdir, *command)
 148                         except:
 149                                 traceback.print_exc()
 150                 finally:
 151                         os._exit(1)
 152         id, status = os.waitpid(child, 0)
 153         assert id == child
 154         if status != 0:
 155                 raise SafeException('Failed to extract archive; exit code %d' % status)