2 """Decode MIME message"""
5 from mimedecode_version import __version__, \
6 __author__, __copyright__, __license__
8 me = os.path.basename(sys.argv[0])
13 Broytman mimedecode.py version %s, %s
14 """ % (__version__, __copyright__))
17 def usage(code=0, errormsg=''):
20 Usage: %s [-h|--help] [-V|--version] [-cCDP] [-H|--host=hostname] [-f charset] [-d header1[,h2,...]|*[,-h1,...]] [-p header1[,h2,h3,...]:param1[,p2,p3,...]] [-r header1[,h2,...]|*[,-h1,...]] [-R header1[,h2,h3,...]:param1[,p2,p3,...]] [--set-header header:value] [--set-param header:param=value] [-Bbeit mask] [--save-headers|body|message mask] [-O dest_dir] [-o output_file] [input_file [output_file]]
23 sys.stderr.write(errormsg + os.linesep)
27 def output_headers(msg):
28 unix_from = msg.get_unixfrom()
32 for key, value in msg.items():
35 value = value.split(';', 1)
39 output(_decode_header(value[1], strip=False))
41 output(os.linesep) # End of headers
44 def recode_if_needed(s, charset):
45 if bytes is str: # Python2
46 if isinstance(s, bytes) and \
47 charset and charset.lower() != g.default_encoding:
48 s = s.decode(charset, "replace").\
49 encode(g.default_encoding, "replace")
51 if isinstance(s, bytes):
52 s = s.decode(charset, "replace")
56 def _decode_header(s, strip=True):
57 """Return a decoded string according to RFC 2047.
58 NOTE: This is almost the same as email.Utils.decode.
62 L = email.header.decode_header(s)
63 if not isinstance(L, list):
68 for atom, charset in L:
69 atom = recode_if_needed(atom, charset or g.default_encoding)
74 # Now that we've decoded everything, we just need to join all the parts
75 # together into the final string.
78 def decode_header(msg, header):
79 "Decode mail header (if exists) and put it back, if it was encoded"
83 new_value = _decode_header(value)
84 if new_value != value: # do not bother to touch msg if not changed
85 set_header(msg, header, new_value)
88 def decode_header_param(msg, header, param):
89 "Decode mail header's parameter (if exists) and put it back, if it was encoded"
92 value = msg.get_param(param, header=header)
94 if isinstance(value, tuple):
95 new_value = recode_if_needed(value[2], value[0])
97 new_value = _decode_header(value)
98 if new_value != value: # do not bother to touch msg if not changed
99 msg.set_param(param, new_value, header)
102 def _get_exceptions(list):
103 return [x[1:].lower() for x in list[1:] if x[0]=='-']
105 def _decode_headers_params(msg, header, decode_all_params, param_list):
106 if decode_all_params:
107 params = msg.get_params(header=header)
109 for param, value in params:
110 if param not in param_list:
111 decode_header_param(msg, header, param)
113 for param in param_list:
114 decode_header_param(msg, header, param)
116 def _remove_headers_params(msg, header, remove_all_params, param_list):
117 if remove_all_params:
118 params = msg.get_params(header=header)
121 for param, value in params:
122 if param not in param_list:
123 msg.del_param(param, header)
126 if value is None: # No such header
128 if ';' not in value: # There are no parameters
130 del msg[header] # Delete all such headers
131 # Get the value without parameters and set it back
132 msg[header] = value.split(';')[0].strip()
134 for param in param_list:
135 msg.del_param(param, header)
137 def decode_headers(msg):
138 "Decode message headers according to global options"
140 for header_list in g.remove_headers:
141 header_list = header_list.split(',')
142 if header_list[0] == '*': # Remove all headers except listed
143 header_list = _get_exceptions(header_list)
144 for header in msg.keys():
145 if header.lower() not in header_list:
147 else: # Remove listed headers
148 for header in header_list:
151 for header_list, param_list in g.remove_headers_params:
152 header_list = header_list.split(',')
153 param_list = param_list.split(',')
154 remove_all_params = param_list[0] == '*' # Remove all params except listed
155 if remove_all_params:
156 param_list = _get_exceptions(param_list)
157 if header_list[0] == '*': # Remove for all headers except listed
158 header_list = _get_exceptions(header_list)
159 for header in msg.keys():
160 if header.lower() not in header_list:
161 _remove_headers_params(msg, header, remove_all_params, param_list)
162 else: # Decode for listed headers
163 for header in header_list:
164 _remove_headers_params(msg, header, remove_all_params, param_list)
166 for header_list in g.decode_headers:
167 header_list = header_list.split(',')
168 if header_list[0] == '*': # Decode all headers except listed
169 header_list = _get_exceptions(header_list)
170 for header in msg.keys():
171 if header.lower() not in header_list:
172 decode_header(msg, header)
173 else: # Decode listed headers
174 for header in header_list:
175 decode_header(msg, header)
177 for header_list, param_list in g.decode_header_params:
178 header_list = header_list.split(',')
179 param_list = param_list.split(',')
180 decode_all_params = param_list[0] == '*' # Decode all params except listed
181 if decode_all_params:
182 param_list = _get_exceptions(param_list)
183 if header_list[0] == '*': # Decode for all headers except listed
184 header_list = _get_exceptions(header_list)
185 for header in msg.keys():
186 if header.lower() not in header_list:
187 _decode_headers_params(msg, header, decode_all_params, param_list)
188 else: # Decode for listed headers
189 for header in header_list:
190 _decode_headers_params(msg, header, decode_all_params, param_list)
193 def set_header(msg, header, value):
197 msg.replace_header(header, value)
202 def set_content_type(msg, newtype, charset=None):
203 msg.set_type(newtype)
206 msg.set_param("charset", charset, "Content-Type")
209 caps = None # Globally stored mailcap database; initialized only if needed
211 def decode_body(msg, s):
212 "Decode body to plain text using first copiousoutput filter from mailcap"
214 import mailcap, tempfile
218 caps = mailcap.getcaps()
220 content_type = msg.get_content_type()
221 filename = tempfile.mktemp()
224 entries = mailcap.lookup(caps, content_type, "view")
225 for entry in entries:
226 if 'copiousoutput' in entry:
228 test = mailcap.subst(entry['test'], content_type, filename)
229 if test and os.system(test) != 0:
231 command = mailcap.subst(entry["view"], content_type, filename)
237 outfile = open(filename, 'wb')
238 if not isinstance(s, bytes):
239 s = s.encode(g.default_encoding, "replace")
243 pipe = os.popen(command, 'r')
248 set_content_type(msg, "text/plain")
249 msg["X-MIME-Autoconverted"] = "from %s to text/plain by %s id %s" % (content_type, g.host_name, command.split()[0])
254 def recode_charset(msg, s):
255 "Recode charset of the message to the default charset"
257 save_charset = charset = msg.get_content_charset()
258 if charset and charset.lower() != g.default_encoding:
259 s = recode_if_needed(s, charset)
260 content_type = msg.get_content_type()
261 set_content_type(msg, content_type, g.default_encoding)
262 msg["X-MIME-Autoconverted"] = "from %s to %s by %s id %s" % (save_charset, g.default_encoding, g.host_name, me)
266 def totext(msg, instring):
267 "Convert instring content to text"
269 # Decode body and recode charset
270 s = decode_body(msg, instring)
272 s = recode_charset(msg, s)
281 def _guess_extension(ctype):
283 if mimetypes is None:
286 user_mime_type = os.path.expanduser('~/.mime.types')
287 if os.path.exists(user_mime_type):
288 mimetypes._db.read(user_mime_type)
289 return mimetypes.guess_extension(ctype)
291 def _save_message(msg, outstring, save_headers=False, save_body=False):
292 for header, param in (
293 ("Content-Disposition", "filename"),
294 ("Content-Type", "name"),
296 fname = msg.get_param(param, header=header)
298 if isinstance(fname, tuple):
299 fname = fname[2] # Do not recode if it isn't recoded yet
301 for forbidden in chr(0), '/', '\\':
302 if forbidden in fname:
311 fname = str(g.save_counter) + fname
313 ext = _guess_extension(msg.get_content_type())
318 outfile = open_output_file(fname)
319 def _output_bytes(s):
320 if not isinstance(s, bytes):
321 s = s.encode(g.default_encoding, "replace")
323 output = _output_bytes
332 def decode_part(msg):
333 "Decode one part of the message"
337 # Test all mask lists and find what to do with this content type
339 ctype = msg.get_content_type()
342 mtype = ctype.split('/')[0]
343 masks.append(mtype + '/*')
347 for content_type in masks:
348 if content_type in g.totext_mask or \
349 content_type in g.decoded_binary_mask:
351 elif content_type in g.binary_mask:
354 elif content_type in g.fully_ignore_mask:
357 encoding = msg["Content-Transfer-Encoding"]
358 if left_binary or encoding in (None, '', '7bit', '8bit', 'binary'):
359 outstring = msg.get_payload()
360 else: # Decode from transfer ecoding to text or binary form
361 outstring = msg.get_payload(decode=1)
362 set_header(msg, "Content-Transfer-Encoding", "8bit")
363 msg["X-MIME-Autoconverted"] = "from %s to 8bit by %s id %s" % (encoding, g.host_name, me)
365 for content_type in masks:
366 if content_type in g.totext_mask:
367 outstring = totext(msg, outstring)
369 elif content_type in g.binary_mask or \
370 content_type in g.decoded_binary_mask:
374 elif content_type in g.ignore_mask:
376 output("%sMessage body of type %s skipped.%s" % (os.linesep, ctype, os.linesep))
378 elif content_type in g.error_mask:
381 # Neither content type nor masks were listed - decode by default
382 outstring = totext(msg, outstring)
384 for content_type in masks:
385 if content_type in g.save_headers_mask:
386 _save_message(msg, outstring, save_headers=True, save_body=False)
387 if content_type in g.save_body_mask:
388 _save_message(msg, outstring, save_headers=False, save_body=True)
389 if content_type in g.save_message_mask:
390 _save_message(msg, outstring, save_headers=True, save_body=True)
392 for content_type in masks:
393 if content_type in g.error_mask:
394 raise ValueError("content type %s prohibited" % ctype)
396 def decode_multipart(msg):
400 boundary = msg.get_boundary()
403 ctype = msg.get_content_type()
406 mtype = ctype.split('/')[0]
407 masks.append(mtype + '/*')
410 for content_type in masks:
411 if content_type in g.fully_ignore_mask:
413 elif content_type in g.ignore_mask:
415 output("%sMessage body of type %s skipped.%s" % (os.linesep, ctype, os.linesep))
417 output("%s--%s--%s" % (os.linesep, boundary, os.linesep))
420 for content_type in masks:
421 if content_type in g.save_body_mask or \
422 content_type in g.save_message_mask:
425 for subpart in msg.get_payload():
427 first_subpart = False
429 _out_l.append(os.linesep)
430 _out_l.append("--%s%s" % (boundary, os.linesep))
431 _out_l.append(subpart.as_string())
432 _out_l.append("%s--%s--%s" % (os.linesep, boundary, os.linesep))
433 outstring = ''.join(_out_l)
438 for content_type in masks:
439 if content_type in g.save_headers_mask:
440 _save_message(msg, outstring, save_headers=True, save_body=False)
441 if content_type in g.save_body_mask:
442 _save_message(msg, outstring, save_headers=False, save_body=True)
443 if content_type in g.save_message_mask:
444 _save_message(msg, outstring, save_headers=True, save_body=True)
446 for content_type in masks:
447 if content_type in g.error_mask:
448 raise ValueError("content type %s prohibited" % ctype)
452 if msg.preamble: # Preserve the first part, it is probably not a RFC822-message
453 output(msg.preamble) # Usually it is just a few lines of text (MIME warning)
454 if msg.preamble is not None:
458 for subpart in msg.get_payload():
461 first_subpart = False
464 output("--%s%s" % (boundary, os.linesep))
466 # Recursively decode all parts of the subpart
467 decode_message(subpart)
470 output("%s--%s--%s" % (os.linesep, boundary, os.linesep))
476 def decode_message(msg):
479 if msg.is_multipart():
480 decode_multipart(msg)
481 elif len(msg): # Simple one-part message (there are headers) - decode it
483 else: # Not a message, just text - copy it literally
484 output(msg.as_string())
487 def open_output_file(filename):
488 fullpath = os.path.abspath(os.path.join(g.destination_dir, filename))
489 full_dir = os.path.dirname(fullpath)
490 create = not os.path.isdir(full_dir)
492 os.makedirs(full_dir)
494 return open(fullpath, 'wb')
497 os.removedirs(full_dir)
501 from m_lib.defenc import default_encoding
502 recode_charset = 1 # recode charset of message body
506 # A list of headers to decode
507 decode_headers = ["From", "To", "Cc", "Reply-To", "Mail-Followup-To",
510 # A list of headers parameters to decode
511 decode_header_params = [
512 ("Content-Type", "name"),
513 ("Content-Disposition", "filename"),
516 # A list of headers to remove
518 # A list of headers parameters to remove
519 remove_headers_params = []
521 # A list of header/value pairs to set
522 set_header_value = []
523 # A list of header/parameter/value triples to set
524 set_header_param = []
526 totext_mask = [] # A list of content-types to decode
527 binary_mask = [] # A list of content-types to pass through
528 decoded_binary_mask = [] # A list of content-types to pass through (content-transfer-decoded)
529 ignore_mask = [] # Ignore (do not decode and do not include into output) but output a warning instead of the body
530 fully_ignore_mask = [] # Completely ignore - no headers, no body, no warning
531 error_mask = [] # Raise error if encounter one of these
534 save_headers_mask = []
536 save_message_mask = []
538 input_filename = None
539 output_filename = None
540 destination_dir = os.curdir
546 from getopt import getopt, GetoptError
549 options, arguments = getopt(sys.argv[1:],
550 'hVcCDPH:f:d:p:r:R:b:B:e:I:i:t:O:o:',
551 ['help', 'version', 'host=',
552 'save-headers=', 'save-body=', 'save-message=',
553 'set-header=', 'set-param='])
557 for option, value in options:
558 if option in ('-h', '--help'):
560 elif option in ('-V', '--version'):
566 elif option in ('-H', '--host'):
569 g.default_encoding = value
571 if value.startswith('*'):
572 g.decode_headers = []
573 g.decode_headers.append(value)
575 g.decode_headers = []
577 g.decode_header_params.append(value.split(':', 1))
579 g.decode_header_params = []
581 g.remove_headers.append(value)
583 g.remove_headers_params.append(value.split(':', 1))
584 elif option == '--set-header':
585 g.set_header_value.append(value.split(':', 1))
586 elif option == '--set-param':
587 header, value = value.split(':', 1)
589 param, value = value.split('=', 1)
591 param, value = value.split(':', 1)
592 g.set_header_param.append((header, param, value))
594 g.totext_mask.append(value)
596 g.binary_mask.append(value)
598 g.decoded_binary_mask.append(value)
600 g.fully_ignore_mask.append(value)
602 g.ignore_mask.append(value)
604 g.error_mask.append(value)
605 elif option == '--save-headers':
606 g.save_headers_mask.append(value)
607 elif option == '--save-body':
608 g.save_body_mask.append(value)
609 elif option == '--save-message':
610 g.save_message_mask.append(value)
612 g.destination_dir = value
614 g.output_filename = value
621 if __name__ == "__main__":
622 arguments = get_opts()
626 g.input_filename = '-'
628 if g.output_filename:
629 outfile = open_output_file(g.output_filename)
631 g.output_filename = '-'
634 if (arguments[0] == '-'):
635 g.input_filename = '-'
638 g.input_filename = arguments[0]
639 infile = open(arguments[0], 'r')
641 if g.output_filename:
642 outfile = open_output_file(g.output_filename)
644 g.output_filename = '-'
647 if g.output_filename:
648 usage(1, 'Too many output filenames')
649 if (arguments[1] == '-'):
650 g.output_filename = '-'
653 g.output_filename = arguments[1]
654 outfile = open_output_file(g.output_filename)
656 usage(1, 'Too many arguments')
658 if (infile is sys.stdin) and sys.stdin.isatty():
659 if (outfile is sys.stdout) and sys.stdout.isatty():
661 usage(1, 'Filtering from console is forbidden')
665 g.host_name = socket.gethostname()
668 if hasattr(outfile, 'buffer'):
670 if not isinstance(s, bytes):
671 s = s.encode(g.default_encoding, "replace")
672 outfile.buffer.write(s)
673 output = output_bytes
675 output = outfile.write
678 msg = email.message_from_file(infile)
680 for header, value in g.set_header_value:
681 set_header(msg, header, value)
683 for header, param, value in g.set_header_param:
685 msg.set_param(param, value, header)