formats.py

   1 import os, sys
   2 from support import PipeThroughCommand, escape, Tmp
   3 import rox
   4
   5 current_command = None
   6
   7 def pipe_through_command(command, src, dst):
   8         global current_command
   9         assert not current_command
  10         try:
  11                 src.seek(0)
  12         except:
  13                 pass
  14         current_command = PipeThroughCommand(command, src, dst)
  15         try:
  16                 current_command.wait()
  17         finally:
  18                 current_command = None
  19
  20 operations = []
  21 class Operation:
  22         add_extension = 0
  23
  24         def __init__(self, extension):
  25                 operations.append(self)
  26                 self.extension = extension
  27
  28         def can_handle(self, data):
  29                 return isinstance(data, FileData)
  30
  31         def save_to_stream(self, data, stream):
  32                 pipe_through_command(self.command, data.source, stream)
  33
  34 class Compress(Operation):
  35         "Compress a stream into another stream."
  36         add_extension = 1
  37
  38         def __init__(self, extension, command, type):
  39                 Operation.__init__(self, extension)
  40                 self.command = command
  41                 self.type = type
  42
  43         def __str__(self):
  44                 return 'Compress as .%s' % self.extension
  45
  46 class Decompress(Operation):
  47         "Decompress a stream into another stream."
  48         type = 'text/plain'
  49
  50         def __init__(self, extension, command):
  51                 Operation.__init__(self, extension)
  52                 self.command = command
  53
  54         def __str__(self):
  55                 return 'Decompress .%s' % self.extension
  56
  57 class Extract(Operation):
  58         "Extract an archive to a directory."
  59         type = 'inode/directory'
  60
  61         def __init__(self, extension, command):
  62                 "If command has a %s then the source path is inserted, else uses stdin."
  63                 Operation.__init__(self, extension)
  64                 self.command = command
  65
  66         def __str__(self):
  67                 return 'Extract from a .%s' % self.extension
  68
  69         def save_to_stream(self, data, stream):
  70                 raise Exception('This operation creates a directory, so you have '
  71                                 'to drag to a filer window on the local machine')
  72
  73         def save_to_file(self, data, path):
  74                 if os.path.exists(path):
  75                         if not os.path.isdir(path):
  76                                 raise Exception("'%s' already exists and is not a directory!" %
  77                                                 path)
  78                 if not os.path.exists(path):
  79                         os.mkdir(path)
  80                 os.chdir(path)
  81                 command = self.command
  82                 source = data.source
  83                 if command.find("'%s'") != -1:
  84                         command = command % escape(source.name)
  85                         source = None
  86                 try:
  87                         pipe_through_command(command, source, None)
  88                 finally:
  89                         try:
  90                                 os.rmdir(path)  # Will only succeed if it's empty
  91                         except:
  92                                 pass
  93                         self.pull_up(path)
  94
  95         def pull_up(self, path):
  96                 # If we created only a single subdirectory, move it up.
  97                 dirs = os.listdir(path)
  98                 if len(dirs) != 1:
  99                         return
 100                 dir = dirs[0]
 101                 unneeded_path = os.path.join(path, dir)
 102                 if not os.path.isdir(unneeded_path):
 103                         return
 104                 import random
 105                 tmp_path = os.path.join(path, 'tmp-' + `random.randint(0, 100000)`)
 106                 os.rename(unneeded_path, tmp_path)
 107                 for file in os.listdir(tmp_path):
 108                         os.rename(os.path.join(tmp_path, file), os.path.join(path, file))
 109                 os.rmdir(tmp_path)
 110
 111 class Archive(Operation):
 112         "Create an archive from a directory."
 113         add_extension = 1
 114
 115         def __init__(self, extension, command, type):
 116                 assert command.find("'%s'") != -1
 117
 118                 Operation.__init__(self, extension)
 119                 self.command = command
 120                 self.type = type
 121
 122         def __str__(self):
 123                 return 'Create .%s archive' % self.extension
 124
 125         def can_handle(self, data):
 126                 return isinstance(data, DirData)
 127
 128         def save_to_stream(self, data, stream):
 129                 os.chdir(os.path.dirname(data.path))
 130                 command = self.command % escape(os.path.basename(data.path))
 131                 pipe_through_command(command, None, stream)
 132
 133 tgz = Extract('tgz',    "gunzip -c - | tar xf -")
 134 tbz = Extract('tar.bz2', "bunzip2 -c - | tar xf -")
 135 rar = Extract('rar',    "rar x -")
 136 tar = Extract('tar',    "tar xf -")
 137 rpm = Extract('rpm',    "rpm2cpio - | cpio -id --quiet")
 138 cpio = Extract('cpio',  "cpio -id --quiet")
 139 deb = Extract('deb', "ar x '%s'")
 140 zip = Extract('zip', "unzip -q '%s'")
 141 jar = Extract('jar', "unzip -q '%s'")
 142
 143 make_tgz = Archive('tgz', "tar cf - '%s' | gzip", 'application/x-compressed-tar')
 144 Archive('tar.gz',  "tar cf - '%s' | gzip", 'application/x-compressed-tar')
 145 Archive('tar.bz2', "tar cf - '%s' | bzip2", 'application/x-bzip-compressed-tar')
 146 Archive('zip',     "zip -qr - '%s'", 'application/zip'),
 147 Archive('jar',     "zip -qr - '%s'", 'application/x-jar')
 148 Archive('tar',     "tar cf - '%s'", 'application/x-tar')
 149
 150 # Note: these go afterwards so that .tar.gz matches before .gz
 151 make_gz = Compress('gz', "gzip -c -", 'application/x-gzip')
 152 Compress('bz2', "bzip2 -c -", 'application/x-bzip')
 153 Compress('uue', "uuencode /dev/stdout", 'application/x-uuencoded')
 154
 155 gz = Decompress('gz',   "gunzip -c -")
 156 bz2 = Decompress('bz2',  "bunzip2 -ck -")
 157 uue = Decompress('uue',  "uudecode -o /dev/stdout")
 158
 159
 160 # Can bzip2 read bzip files?
 161
 162 aliases = {
 163         'tar.gz': 'tgz',
 164         'tar.bz': 'tar.bz2',
 165         'bz': 'bz2'
 166 }
 167
 168 known_extensions = {}
 169 for x in operations:
 170         try:
 171                 known_extensions[x.extension] = None
 172         except AttributeError:
 173                 pass
 174
 175 class FileData:
 176         "A file on the local filesystem."
 177         mode = None
 178         def __init__(self, path):
 179                 self.path = path
 180
 181                 if path == '-':
 182                         source = sys.stdin
 183                 else:
 184                         try:
 185                                 source = file(path)
 186                                 self.mode = os.stat(path).st_mode
 187                         except:
 188                                 rox.report_exception()
 189                                 sys.exit(1)
 190
 191                 self.path = path
 192                 start = source.read(300)
 193                 try:
 194                         if source is sys.stdin:
 195                                 raise "Always copy stdin!"
 196                         source.seek(0)
 197                         self.source = source
 198                 except:
 199                         # Input is not a regular, local, seekable file, so copy it
 200                         # to a local temp file.
 201                         import shutil
 202                         tmp = Tmp()
 203                         tmp.write(start)
 204                         tmp.flush()
 205                         shutil.copyfileobj(source, tmp)
 206                         tmp.seek(0)
 207                         tmp.flush()
 208                         self.source = tmp
 209                 self.default = self.guess_format(start)
 210
 211                 if path == '-':
 212                         name = 'Data'
 213                 else:
 214                         name = path
 215                         for ext in known_extensions:
 216                                 if path.endswith('.' + ext):
 217                                         new = path[:-len(ext)-1]
 218                                         if len(new) < len(name):
 219                                                 name = new
 220                 if self.default.add_extension:
 221                         name += '.' + self.default.extension
 222                 self.default_name = name
 223
 224         def guess_format(self, data):
 225                 "Return a good default Operation, judging by the first 300 bytes or so."
 226                 l = len(data)
 227                 def string(offset, match):
 228                         return data[offset:offset + len(match)] == match
 229                 def short(offset, match):
 230                         if l > offset + 1:
 231                                 a = data[offset]
 232                                 b = data[offset + 1]
 233                                 return ((a == match & 0xff) and (b == (match >> 8))) or \
 234                                         (b == match & 0xff) and (a == (match >> 8))
 235                         return 0
 236
 237                 # Archives
 238                 if string(257, 'ustar\0') or string(257, 'ustar\040\040\0'):
 239                         return tar
 240                 if short(0, 070707) or short(0, 0143561) or string(0, '070707') or \
 241                    string(0, '070701') or string(0, '070702'):
 242                         return cpio
 243                 if string(0, '!<arch>') or string(0, '\\<ar>') or string(0, '<ar>'):
 244                         if string(7, '\ndebian'):
 245                                 return deb
 246                 if string(0, 'Rar!'): return rar
 247                 if string(0, 'PK\003\004'): return zip
 248                 if string(0, '\xed\xab\xee\xdb'): return rpm
 249
 250                 # Compressed streams
 251                 if string(0, '\037\213'):
 252                         if self.path.endswith('.tar.gz') or self.path.endswith('.tgz'):
 253                                 return tgz
 254                         return gz
 255                 if string(0, 'BZh') or string(0, 'BZ'):
 256                         if self.path.endswith('.tar.bz') or self.path.endswith('.tar.bz2') or \
 257                            self.path.endswith('.tbz') or self.path.endswith('.tbz2'):
 258                                 return tbz
 259                         return bz2
 260                 if string(0, 'begin '):
 261                         return uue
 262
 263                 return make_gz
 264
 265 class DirData:
 266         mode = None
 267         def __init__(self, path):
 268                 self.path = path
 269                 self.default = make_tgz
 270                 self.default_name = path + '.' + self.default.extension
 271
 272 def test():
 273         test_data = 'Hello\0World\n'
 274         src = Tmp()
 275         src.write(test_data)
 276         src.flush()
 277         data = FileData(src.name)
 278         for comp in operations:
 279                 if not isinstance(comp, Compress): continue
 280                 dec = [o for o in operations if isinstance(o, Decompress) and
 281                                                 o.extension == comp.extension]
 282                 assert len(dec) == 1
 283                 dec = dec[0]
 284                 print "Test %s / %s" % (comp, dec)
 285                 middle = Tmp()
 286                 comp.save_to_stream(data, middle)
 287                 out = Tmp()
 288                 dec.save_to_stream(FileData(middle.name), out)
 289                 del middle
 290                 assert file(out.name).read() == test_data
 291                 print "Passed"
 292         del src
 293
 294         dir = '/tmp/archive-regression-test'
 295         out = dir + '.out'
 296         if not os.path.exists(dir): os.mkdir(dir)
 297         print >>file(dir + '/test', 'w'), test_data
 298         data = DirData(dir)
 299
 300         for archive in operations:
 301                 if not isinstance(archive, Archive): continue
 302                 extract = [o for o in operations if isinstance(o, Extract) and
 303                                                     o.extension == archive.extension]
 304                 if not extract:
 305                         print "(skipping %s; no extractor)" % archive
 306                         continue
 307
 308                 if os.path.exists(out): os.system("rm -r '%s'" % out)
 309
 310                 assert len(extract) == 1
 311                 extract = extract[0]
 312                 print "Test %s / %s" % (archive, extract)
 313
 314                 middle = Tmp()
 315                 archive.save_to_stream(data, middle)
 316                 extract.save_to_file(FileData(middle.name), dir + '.out')
 317
 318                 assert os.listdir(dir) == os.listdir(out)
 319                 assert file(dir + '/test').read() == file(out + '/test').read()
 320                 print "Passed"
 321
 322         os.unlink(dir + '/test')
 323         os.rmdir(dir)
 324         if os.path.exists(out): os.system("rm -r '%s'" % out)
 325
 326 if __name__ == '__main__': test()