python/rox/mime.py

   1 """This module provides access to the shared MIME database.
   2
   3 types is a dictionary of all known MIME types, indexed by the type name, e.g.
   4 types['application/x-python']
   5
   6 Applications can install information about MIME types by storing an
   7 XML file as <MIME>/packages/<application>.xml and running the
   8 update-mime-database command, which is provided by the freedesktop.org
   9 shared mime database package.
  10
  11 See http://www.freedesktop.org/standards/shared-mime-info-spec/ for
  12 information about the format of these files."""
  13
  14 import os
  15 import stat
  16 import fnmatch
  17
  18 import rox
  19 import rox.choices
  20 from rox import i18n, _, basedir
  21
  22 from xml.dom import Node, minidom, XML_NAMESPACE
  23
  24 FREE_NS = 'http://www.freedesktop.org/standards/shared-mime-info'
  25
  26 types = {}              # Maps MIME names to type objects
  27
  28 # Icon sizes when requesting MIME type icon
  29 ICON_SIZE_HUGE=96
  30 ICON_SIZE_LARGE=52
  31 ICON_SIZE_SMALL=18
  32 ICON_SIZE_UNSCALED=None
  33
  34 exts = None             # Maps extensions to types
  35 globs = None            # List of (glob, type) pairs
  36 literals = None         # Maps liternal names to types
  37 magic = None
  38
  39 def _get_node_data(node):
  40         """Get text of XML node"""
  41         return ''.join([n.nodeValue for n in node.childNodes]).strip()
  42
  43 def lookup(media, subtype = None):
  44         "Get the MIMEtype object for this type, creating a new one if needed."
  45         if subtype is None and '/' in media:
  46                 media, subtype = media.split('/', 1)
  47         if (media, subtype) not in types:
  48                 types[(media, subtype)] = MIMEtype(media, subtype)
  49         return types[(media, subtype)]
  50
  51 class MIMEtype:
  52         """Type holding data about a MIME type"""
  53         def __init__(self, media, subtype):
  54                 "Don't use this constructor directly; use mime.lookup() instead."
  55                 assert media and '/' not in media
  56                 assert subtype and '/' not in subtype
  57                 assert (media, subtype) not in types
  58
  59                 self.media = media
  60                 self.subtype = subtype
  61                 self._comment = None
  62
  63         def _load(self):
  64                 "Loads comment for current language. Use get_comment() instead."
  65                 resource = os.path.join('mime', self.media, self.subtype + '.xml')
  66                 for path in basedir.load_data_paths(resource):
  67                         doc = minidom.parse(path)
  68                         if doc is None:
  69                                 continue
  70                         for comment in doc.documentElement.getElementsByTagNameNS(FREE_NS, 'comment'):
  71                                 lang = comment.getAttributeNS(XML_NAMESPACE, 'lang') or 'en'
  72                                 goodness = 1 + (lang in i18n.langs)
  73                                 if goodness > self._comment[0]:
  74                                         self._comment = (goodness, _get_node_data(comment))
  75                                 if goodness == 2: return
  76
  77         def get_comment(self):
  78                 """Returns comment for current language, loading it if needed."""
  79                 # Should we ever reload?
  80                 if self._comment is None:
  81                         self._comment = (0, str(self))
  82                         self._load()
  83                 return self._comment[1]
  84
  85         def __str__(self):
  86                 return self.media + '/' + self.subtype
  87
  88         def __repr__(self):
  89                 return '[%s: %s]' % (self, self._comment or '(comment not loaded)')
  90
  91         def get_icon(self, size=None):
  92                 """Return a GdkPixbuf with the icon for this type.  If size
  93                 is None then the image is returned at its natural size,
  94                 otherwise the image is scaled to that width with the height
  95                 at the correct aspect ratio.  The constants
  96                 ICON_SIZE_{HUGE,LARGE,SMALL} match the sizes used by the
  97                 filer."""
  98                 # I suppose it would make more sense to move the code
  99                 # from saving to here...
 100                 import saving
 101                 base=saving.image_for_type(self.media + '/' + self.subtype)
 102                 if not base or not size:
 103                         return base
 104
 105                 h=int(base.get_width()*float(size)/base.get_height())
 106                 return base.scale_simple(size, h, rox.g.gdk.INTERP_BILINEAR)
 107
 108 class MagicRule:
 109         def __init__(self, f):
 110                 self.next=None
 111                 self.prev=None
 112
 113                 #print line
 114                 ind=''
 115                 while True:
 116                         c=f.read(1)
 117                         if c=='>':
 118                                 break
 119                         ind+=c
 120                 if not ind:
 121                         self.nest=0
 122                 else:
 123                         self.nest=int(ind)
 124
 125                 start=''
 126                 while True:
 127                         c=f.read(1)
 128                         if c=='=':
 129                                 break
 130                         start+=c
 131                 self.start=int(start)
 132
 133                 hb=f.read(1)
 134                 lb=f.read(1)
 135                 self.lenvalue=ord(lb)+(ord(hb)<<8)
 136
 137                 self.value=f.read(self.lenvalue)
 138
 139                 c=f.read(1)
 140                 if c=='&':
 141                         self.mask=f.read(self.lenvalue)
 142                         c=f.read(1)
 143                 else:
 144                         self.mask=None
 145
 146                 if c=='~':
 147                         w=''
 148                         while c!='+' and c!='\n':
 149                                 c=f.read(1)
 150                                 if c=='+' or c=='\n':
 151                                         break
 152                                 w+=c
 153
 154                         self.word=int(w)
 155                 else:
 156                         self.word=1
 157
 158                 if c=='+':
 159                         r=''
 160                         while c!='\n':
 161                                 c=f.read(1)
 162                                 if c=='\n':
 163                                         break
 164                                 r+=c
 165                         #print r
 166                         self.range=int(r)
 167                 else:
 168                         self.range=1
 169
 170                 if c!='\n':
 171                         raise 'Malformed MIME magic line'
 172
 173         def getLength(self):
 174                 return self.start+self.lenvalue+self.range
 175
 176         def appendRule(self, rule):
 177                 if self.nest<rule.nest:
 178                         self.next=rule
 179                         rule.prev=self
 180
 181                 elif self.prev:
 182                         self.prev.appendRule(rule)
 183
 184         def match(self, buffer):
 185                 if self.match0(buffer):
 186                         if self.next:
 187                                 return self.next.match(buffer)
 188                         return True
 189
 190         def match0(self, buffer):
 191                 l=len(buffer)
 192                 for o in range(self.range):
 193                         s=self.start+o
 194                         e=s+self.lenvalue
 195                         if l<e:
 196                                 return False
 197                         if self.mask:
 198                                 test=''
 199                                 for i in range(self.lenvalue):
 200                                         c=ord(buffer[s+i]) & ord(self.mask[i])
 201                                         test+=chr(c)
 202                         else:
 203                                 test=buffer[s:e]
 204
 205                         if test==self.value:
 206                                 return True
 207
 208         def __repr__(self):
 209                 return '<MagicRule %d>%d=[%d]%s&%s~%d+%d>' % (self.nest,
 210                                                               self.start,
 211                                                               self.lenvalue,
 212                                                               `self.value`,
 213                                                               `self.mask`,
 214                                                               self.word,
 215                                                               self.range)
 216
 217 class MagicType:
 218         def __init__(self, mtype):
 219                 self.mtype=mtype
 220                 self.top_rules=[]
 221                 self.last_rule=None
 222
 223         def getLine(self, f):
 224                 nrule=MagicRule(f)
 225
 226                 if nrule.nest and self.last_rule:
 227                         self.last_rule.appendRule(nrule)
 228                 else:
 229                         self.top_rules.append(nrule)
 230
 231                 self.last_rule=nrule
 232
 233                 return nrule
 234
 235         def match(self, buffer):
 236                 for rule in self.top_rules:
 237                         if rule.match(buffer):
 238                                 return self.mtype
 239
 240         def __repr__(self):
 241                 return '<MagicType %s>' % self.mtype
 242
 243 class MagicDB:
 244         def __init__(self):
 245                 self.types={}   # Indexed by priority, each entry is a list of type rules
 246                 self.maxlen=0
 247
 248         def mergeFile(self, fname):
 249                 f=file(fname, 'r')
 250                 line=f.readline()
 251                 if line!='MIME-Magic\0\n':
 252                         raise 'Not a MIME magic file'
 253
 254                 while True:
 255                         shead=f.readline()
 256                         #print shead
 257                         if not shead:
 258                                 break
 259                         if shead[0]!='[' or shead[-2:]!=']\n':
 260                                 raise 'Malformed section heading'
 261                         pri, tname=shead[1:-2].split(':')
 262                         #print shead[1:-2]
 263                         pri=int(pri)
 264                         mtype=lookup(tname)
 265
 266                         try:
 267                                 ents=self.types[pri]
 268                         except:
 269                                 ents=[]
 270                                 self.types[pri]=ents
 271
 272                         magictype=MagicType(mtype)
 273                         #print tname
 274
 275                         #rline=f.readline()
 276                         c=f.read(1)
 277                         f.seek(-1, 1)
 278                         while c and c!='[':
 279                                 rule=magictype.getLine(f)
 280                                 #print rule
 281                                 if rule and rule.getLength()>self.maxlen:
 282                                         self.maxlen=rule.getLength()
 283
 284                                 c=f.read(1)
 285                                 f.seek(-1, 1)
 286
 287                         ents.append(magictype)
 288                         #self.types[pri]=ents
 289                         if not c:
 290                                 break
 291
 292         def match(self, path, max_pri=100, min_pri=0):
 293                 try:
 294                         buf=file(path, 'r').read(self.maxlen)
 295                         pris=self.types.keys()
 296                         pris.sort(lambda a, b: -cmp(a, b))
 297                         for pri in pris:
 298                                 #print pri, max_pri, min_pri
 299                                 if pri>max_pri:
 300                                         continue
 301                                 if pri<min_pri:
 302                                         break
 303                                 for type in self.types[pri]:
 304                                         m=type.match(buf)
 305                                         if m:
 306                                                 return m
 307                 except:
 308                         pass
 309
 310                 return None
 311
 312         def __repr__(self):
 313                 return '<MagicDB %s>' % self.types
 314
 315
 316 # Some well-known types
 317 text = lookup('text', 'plain')
 318 inode_block = lookup('inode', 'blockdevice')
 319 inode_char = lookup('inode', 'chardevice')
 320 inode_dir = lookup('inode', 'directory')
 321 inode_fifo = lookup('inode', 'fifo')
 322 inode_socket = lookup('inode', 'socket')
 323 inode_symlink = lookup('inode', 'symlink')
 324 inode_door = lookup('inode', 'door')
 325 app_exe = lookup('application', 'executable')
 326
 327 _cache_uptodate = False
 328
 329 def _cache_database():
 330         global exts, globs, literals, magic, _cache_uptodate
 331
 332         _cache_uptodate = True
 333
 334         exts = {}               # Maps extensions to types
 335         globs = []              # List of (glob, type) pairs
 336         literals = {}           # Maps liternal names to types
 337         magic = MagicDB()
 338
 339         def _import_glob_file(path):
 340                 """Loads name matching information from a MIME directory."""
 341                 for line in file(path):
 342                         if line.startswith('#'): continue
 343                         line = line[:-1]
 344
 345                         type_name, pattern = line.split(':', 1)
 346                         mtype = lookup(type_name)
 347
 348                         if pattern.startswith('*.'):
 349                                 rest = pattern[2:]
 350                                 if not ('*' in rest or '[' in rest or '?' in rest):
 351                                         exts[rest] = mtype
 352                                         continue
 353                         if '*' in pattern or '[' in pattern or '?' in pattern:
 354                                 globs.append((pattern, mtype))
 355                         else:
 356                                 literals[pattern] = mtype
 357
 358         for path in basedir.load_data_paths(os.path.join('mime', 'globs')):
 359                 _import_glob_file(path)
 360         for path in basedir.load_data_paths(os.path.join('mime', 'magic')):
 361                 magic.mergeFile(path)
 362
 363         # Sort globs by length
 364         globs.sort(lambda a, b: cmp(len(b[0]), len(a[0])))
 365
 366 def get_type_by_name(path):
 367         """Returns type of file by its name, or None if not known"""
 368         if not _cache_uptodate:
 369                 _cache_database()
 370
 371         leaf = os.path.basename(path)
 372         if leaf in literals:
 373                 return literals[leaf]
 374
 375         lleaf = leaf.lower()
 376         if lleaf in literals:
 377                 return literals[lleaf]
 378
 379         ext = leaf
 380         while 1:
 381                 p = ext.find('.')
 382                 if p < 0: break
 383                 ext = ext[p + 1:]
 384                 if ext in exts:
 385                         return exts[ext]
 386         ext = lleaf
 387         while 1:
 388                 p = ext.find('.')
 389                 if p < 0: break
 390                 ext = ext[p+1:]
 391                 if ext in exts:
 392                         return exts[ext]
 393         for (glob, mime_type) in globs:
 394                 if fnmatch.fnmatch(leaf, glob):
 395                         return mime_type
 396                 if fnmatch.fnmatch(lleaf, glob):
 397                         return mime_type
 398         return None
 399
 400 def get_type_by_contents(path, max_pri=100, min_pri=0):
 401         """Returns type of file by its contents, or None if not known"""
 402         if not _cache_uptodate:
 403                 _cache_database()
 404
 405         return magic.match(path, max_pri, min_pri)
 406
 407 def get_type(path, follow=1, name_pri=100):
 408         """Returns type of file indicated by path.
 409         path     - pathname to check (need not exist)
 410         follow   - when reading file, follow symbolic links
 411         name_pri - Priority to do name matches.  100=override magic"""
 412         if not _cache_uptodate:
 413                 _cache_database()
 414
 415         try:
 416                 if follow:
 417                         st = os.stat(path)
 418                 else:
 419                         st = os.lstat(path)
 420         except:
 421                 t = get_type_by_name(path)
 422                 return t or text
 423
 424         if stat.S_ISREG(st.st_mode):
 425                 t = get_type_by_contents(path, min_pri=name_pri)
 426                 if not t: t = get_type_by_name(path)
 427                 if not t: t = get_type_by_contents(path, max_pri=name_pri)
 428                 if t is None:
 429                         if stat.S_IMODE(st.st_mode) & 0111:
 430                                 return app_exe
 431                         else:
 432                                 return text
 433                 return t
 434         elif stat.S_ISDIR(st.st_mode): return inode_dir
 435         elif stat.S_ISCHR(st.st_mode): return inode_char
 436         elif stat.S_ISBLK(st.st_mode): return inode_block
 437         elif stat.S_ISFIFO(st.st_mode): return inode_fifo
 438         elif stat.S_ISLNK(st.st_mode): return inode_symlink
 439         elif stat.S_ISSOCK(st.st_mode): return inode_socket
 440         return inode_door
 441
 442 def install_mime_info(application, package_file = None):
 443         """Copy 'package_file' as ~/.local/share/mime/packages/<application>.xml.
 444         If package_file is None, install <app_dir>/<application>.xml.
 445         If already installed, does nothing. May overwrite an existing
 446         file with the same name (if the contents are different)"""
 447         application += '.xml'
 448         if not package_file:
 449                 package_file = os.path.join(rox.app_dir, application)
 450
 451         new_data = file(package_file).read()
 452
 453         # See if the file is already installed
 454
 455         package_dir = os.path.join('mime', 'packages')
 456         resource = os.path.join(package_dir, application)
 457         for x in basedir.load_data_paths(resource):
 458                 try:
 459                         old_data = file(x).read()
 460                 except:
 461                         continue
 462                 if old_data == new_data:
 463                         return  # Already installed
 464
 465         global _cache_uptodate
 466         _cache_uptodate = False
 467
 468         # Not already installed; add a new copy
 469         try:
 470                 # Create the directory structure...
 471                 new_file = os.path.join(basedir.save_data_path(package_dir), application)
 472
 473                 # Write the file...
 474                 file(new_file, 'w').write(new_data)
 475
 476                 # Update the database...
 477                 if os.path.isdir('/uri/0install/zero-install.sourceforge.net'):
 478                         command = '/uri/0install/zero-install.sourceforge.net/bin/update-mime-database'
 479                 else:
 480                         command = 'update-mime-database'
 481                 if os.spawnlp(os.P_WAIT, command, command, basedir.save_data_path('mime')):
 482                         os.unlink(new_file)
 483                         raise Exception(_("The '%s' command returned an error code!\n" \
 484                                           "Make sure you have the freedesktop.org shared MIME package:\n" \
 485                                           "http://www.freedesktop.org/standards/shared-mime-info.html") % command)
 486         except:
 487                 rox.report_exception()
 488
 489 def get_type_handler(mime_type, handler_type = 'MIME-types'):
 490         """Lookup the ROX-defined run action for a given mime type.
 491         mime_type is an object returned by lookup().
 492         handler_type is a config directory leaf (e.g.'MIME-types')."""
 493         handler = basedir.load_first_config('rox.sourceforge.net', handler_type,
 494                                  mime_type.media + '_' + mime_type.subtype)
 495         if not handler:
 496                 # Fall back to the base handler if no subtype handler exists
 497                 handler = basedir.load_first_config('rox.sourceforge.net', handler_type,
 498                                          mime_type.media)
 499         return handler
 500
 501 def _test(name):
 502         """Print results for name.  Test routine"""
 503         t=get_type(name, name_pri=80)
 504         print name, t, t.get_comment()
 505
 506 if __name__=='__main__':
 507         import sys
 508         if len(sys.argv)<2:
 509                 _test('file.txt')
 510         else:
 511                 for f in sys.argv[1:]:
 512                         _test(f)
 513         #print globs