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