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():
37 output(os.linesep) # End of headers
40 def recode_if_needed(s, charset):
41 if bytes is str: # Python2
42 if isinstance(s, bytes) and \
43 charset and charset.lower() != g.default_encoding:
44 s = s.decode(charset, "replace").\
45 encode(g.default_encoding, "replace")
47 if isinstance(s, bytes):
48 s = s.decode(charset, "replace")
52 def _decode_header(s, strip=True):
53 """Return a decoded string according to RFC 2047.
54 NOTE: This is almost the same as email.Utils.decode.
58 L = email.header.decode_header(s)
59 if not isinstance(L, list):
64 for atom, charset in L:
65 atom = recode_if_needed(atom, charset or g.default_encoding)
70 # Now that we've decoded everything, we just need to join all the parts
71 # together into the final string.
74 def decode_header(msg, header):
75 "Decode mail header (if exists) and put it back, if it was encoded"
79 new_value = _decode_header(value)
80 if new_value != value: # do not bother to touch msg if not changed
81 set_header(msg, header, new_value)
84 def decode_header_param(msg, header, param):
85 "Decode mail header's parameter (if exists) and put it back, if it was encoded"
88 value = msg.get_param(param, header=header)
90 if isinstance(value, tuple):
91 new_value = recode_if_needed(value[2], value[0])
93 new_value = _decode_header(value)
94 if new_value != value: # do not bother to touch msg if not changed
95 msg.set_param(param, new_value, header)
98 def _get_exceptions(list):
99 return [x[1:].lower() for x in list[1:] if x[0]=='-']
101 def _decode_headers_params(msg, header, decode_all_params, param_list):
102 if decode_all_params:
103 params = msg.get_params(header=header)
105 for param, value in params:
106 if param not in param_list:
107 decode_header_param(msg, header, param)
109 for param in param_list:
110 decode_header_param(msg, header, param)
112 def _remove_headers_params(msg, header, remove_all_params, param_list):
113 if remove_all_params:
114 params = msg.get_params(header=header)
117 for param, value in params:
118 if param not in param_list:
119 msg.del_param(param, header)
122 if value is None: # No such header
124 if ';' not in value: # There are no parameters
126 del msg[header] # Delete all such headers
127 # Get the value without parameters and set it back
128 msg[header] = value.split(';')[0].strip()
130 for param in param_list:
131 msg.del_param(param, header)
133 def decode_headers(msg):
134 "Decode message headers according to global options"
136 for header_list in g.remove_headers:
137 header_list = header_list.split(',')
138 if header_list[0] == '*': # Remove all headers except listed
139 header_list = _get_exceptions(header_list)
140 for header in msg.keys():
141 if header.lower() not in header_list:
143 else: # Remove listed headers
144 for header in header_list:
147 for header_list, param_list in g.remove_headers_params:
148 header_list = header_list.split(',')
149 param_list = param_list.split(',')
150 remove_all_params = param_list[0] == '*' # Remove all params except listed
151 if remove_all_params:
152 param_list = _get_exceptions(param_list)
153 if header_list[0] == '*': # Remove for all headers except listed
154 header_list = _get_exceptions(header_list)
155 for header in msg.keys():
156 if header.lower() not in header_list:
157 _remove_headers_params(msg, header, remove_all_params, param_list)
158 else: # Decode for listed headers
159 for header in header_list:
160 _remove_headers_params(msg, header, remove_all_params, param_list)
162 for header_list in g.decode_headers:
163 header_list = header_list.split(',')
164 if header_list[0] == '*': # Decode all headers except listed
165 header_list = _get_exceptions(header_list)
166 for header in msg.keys():
167 if header.lower() not in header_list:
168 decode_header(msg, header)
169 else: # Decode listed headers
170 for header in header_list:
171 decode_header(msg, header)
173 for header_list, param_list in g.decode_header_params:
174 header_list = header_list.split(',')
175 param_list = param_list.split(',')
176 decode_all_params = param_list[0] == '*' # Decode all params except listed
177 if decode_all_params:
178 param_list = _get_exceptions(param_list)
179 if header_list[0] == '*': # Decode for all headers except listed
180 header_list = _get_exceptions(header_list)
181 for header in msg.keys():
182 if header.lower() not in header_list:
183 _decode_headers_params(msg, header, decode_all_params, param_list)
184 else: # Decode for listed headers
185 for header in header_list:
186 _decode_headers_params(msg, header, decode_all_params, param_list)
189 def set_header(msg, header, value):
193 msg.replace_header(header, value)
198 def set_content_type(msg, newtype, charset=None):
199 msg.set_type(newtype)
202 msg.set_param("charset", charset, "Content-Type")
205 caps = None # Globally stored mailcap database; initialized only if needed
207 def decode_body(msg, s):
208 "Decode body to plain text using first copiousoutput filter from mailcap"
210 import mailcap, tempfile
214 caps = mailcap.getcaps()
216 content_type = msg.get_content_type()
217 filename = tempfile.mktemp()
220 entries = mailcap.lookup(caps, content_type, "view")
221 for entry in entries:
222 if 'copiousoutput' in entry:
224 test = mailcap.subst(entry['test'], content_type, filename)
225 if test and os.system(test) != 0:
227 command = mailcap.subst(entry["view"], content_type, filename)
233 outfile = open(filename, 'wb')
234 if not isinstance(s, bytes):
235 s = s.encode(g.default_encoding, "replace")
239 pipe = os.popen(command, 'r')
244 set_content_type(msg, "text/plain")
245 msg["X-MIME-Autoconverted"] = "from %s to text/plain by %s id %s" % (content_type, g.host_name, command.split()[0])
250 def recode_charset(msg, s):
251 "Recode charset of the message to the default charset"
253 save_charset = charset = msg.get_content_charset()
254 if charset and charset.lower() != g.default_encoding:
255 s = recode_if_needed(s, charset)
256 content_type = msg.get_content_type()
257 set_content_type(msg, content_type, g.default_encoding)
258 msg["X-MIME-Autoconverted"] = "from %s to %s by %s id %s" % (save_charset, g.default_encoding, g.host_name, me)
262 def totext(msg, instring):
263 "Convert instring content to text"
265 # Decode body and recode charset
266 s = decode_body(msg, instring)
268 s = recode_charset(msg, s)
277 def _guess_extension(ctype):
279 if mimetypes is None:
282 user_mime_type = os.path.expanduser('~/.mime.types')
283 if os.path.exists(user_mime_type):
284 mimetypes._db.read(user_mime_type)
285 return mimetypes.guess_extension(ctype)
287 def _save_message(msg, outstring, save_headers=False, save_body=False):
288 for header, param in (
289 ("Content-Disposition", "filename"),
290 ("Content-Type", "name"),
292 fname = msg.get_param(param, header=header)
294 if isinstance(fname, tuple):
295 fname = fname[2] # Do not recode if it isn't recoded yet
297 for forbidden in chr(0), '/', '\\':
298 if forbidden in fname:
307 fname = str(g.save_counter) + fname
309 ext = _guess_extension(msg.get_content_type())
314 outfile = open_output_file(fname)
315 def _output_bytes(s):
316 if not isinstance(s, bytes):
317 s = s.encode(g.default_encoding, "replace")
319 output = _output_bytes
328 def decode_part(msg):
329 "Decode one part of the message"
333 # Test all mask lists and find what to do with this content type
335 ctype = msg.get_content_type()
338 mtype = ctype.split('/')[0]
339 masks.append(mtype + '/*')
343 for content_type in masks:
344 if content_type in g.totext_mask or \
345 content_type in g.decoded_binary_mask:
347 elif content_type in g.binary_mask:
350 elif content_type in g.fully_ignore_mask:
353 encoding = msg["Content-Transfer-Encoding"]
354 if left_binary or encoding in (None, '', '7bit', '8bit', 'binary'):
355 outstring = msg.get_payload()
356 else: # Decode from transfer ecoding to text or binary form
357 outstring = msg.get_payload(decode=1)
358 set_header(msg, "Content-Transfer-Encoding", "8bit")
359 msg["X-MIME-Autoconverted"] = "from %s to 8bit by %s id %s" % (encoding, g.host_name, me)
361 for content_type in masks:
362 if content_type in g.totext_mask:
363 outstring = totext(msg, outstring)
365 elif content_type in g.binary_mask or \
366 content_type in g.decoded_binary_mask:
370 elif content_type in g.ignore_mask:
372 output("%sMessage body of type %s skipped.%s" % (os.linesep, ctype, os.linesep))
374 elif content_type in g.error_mask:
377 # Neither content type nor masks were listed - decode by default
378 outstring = totext(msg, outstring)
380 for content_type in masks:
381 if content_type in g.save_headers_mask:
382 _save_message(msg, outstring, save_headers=True, save_body=False)
383 if content_type in g.save_body_mask:
384 _save_message(msg, outstring, save_headers=False, save_body=True)
385 if content_type in g.save_message_mask:
386 _save_message(msg, outstring, save_headers=True, save_body=True)
388 for content_type in masks:
389 if content_type in g.error_mask:
390 raise ValueError("content type %s prohibited" % ctype)
392 def decode_multipart(msg):
396 boundary = msg.get_boundary()
399 ctype = msg.get_content_type()
402 mtype = ctype.split('/')[0]
403 masks.append(mtype + '/*')
406 for content_type in masks:
407 if content_type in g.fully_ignore_mask:
409 elif content_type in g.ignore_mask:
411 output("%sMessage body of type %s skipped.%s" % (os.linesep, ctype, os.linesep))
413 output("%s--%s--%s" % (os.linesep, boundary, os.linesep))
416 for content_type in masks:
417 if content_type in g.save_body_mask or \
418 content_type in g.save_message_mask:
421 for subpart in msg.get_payload():
423 first_subpart = False
425 _out_l.append(os.linesep)
426 _out_l.append("--%s%s" % (boundary, os.linesep))
427 _out_l.append(subpart.as_string())
428 _out_l.append("%s--%s--%s" % (os.linesep, boundary, os.linesep))
429 outstring = ''.join(_out_l)
434 for content_type in masks:
435 if content_type in g.save_headers_mask:
436 _save_message(msg, outstring, save_headers=True, save_body=False)
437 if content_type in g.save_body_mask:
438 _save_message(msg, outstring, save_headers=False, save_body=True)
439 if content_type in g.save_message_mask:
440 _save_message(msg, outstring, save_headers=True, save_body=True)
442 for content_type in masks:
443 if content_type in g.error_mask:
444 raise ValueError("content type %s prohibited" % ctype)
448 if msg.preamble: # Preserve the first part, it is probably not a RFC822-message
449 output(msg.preamble) # Usually it is just a few lines of text (MIME warning)
450 if msg.preamble is not None:
454 for subpart in msg.get_payload():
457 first_subpart = False
460 output("--%s%s" % (boundary, os.linesep))
462 # Recursively decode all parts of the subpart
463 decode_message(subpart)
466 output("%s--%s--%s" % (os.linesep, boundary, os.linesep))
472 def decode_message(msg):
475 if msg.is_multipart():
476 decode_multipart(msg)
477 elif len(msg): # Simple one-part message (there are headers) - decode it
479 else: # Not a message, just text - copy it literally
480 output(msg.as_string())
483 def open_output_file(filename):
484 fullpath = os.path.abspath(os.path.join(g.destination_dir, filename))
485 full_dir = os.path.dirname(fullpath)
486 create = not os.path.isdir(full_dir)
488 os.makedirs(full_dir)
490 return open(fullpath, 'wb')
493 os.removedirs(full_dir)
497 from m_lib.defenc import default_encoding
498 recode_charset = 1 # recode charset of message body
502 # A list of headers to decode
503 decode_headers = ["From", "To", "Cc", "Reply-To", "Mail-Followup-To",
506 # A list of headers parameters to decode
507 decode_header_params = [
508 ("Content-Type", "name"),
509 ("Content-Disposition", "filename"),
512 # A list of headers to remove
514 # A list of headers parameters to remove
515 remove_headers_params = []
517 # A list of header/value pairs to set
518 set_header_value = []
519 # A list of header/parameter/value triples to set
520 set_header_param = []
522 totext_mask = [] # A list of content-types to decode
523 binary_mask = [] # A list of content-types to pass through
524 decoded_binary_mask = [] # A list of content-types to pass through (content-transfer-decoded)
525 ignore_mask = [] # Ignore (do not decode and do not include into output) but output a warning instead of the body
526 fully_ignore_mask = [] # Completely ignore - no headers, no body, no warning
527 error_mask = [] # Raise error if encounter one of these
530 save_headers_mask = []
532 save_message_mask = []
534 input_filename = None
535 output_filename = None
536 destination_dir = os.curdir
542 from getopt import getopt, GetoptError
545 options, arguments = getopt(sys.argv[1:],
546 'hVcCDPH:f:d:p:r:R:b:B:e:I:i:t:O:o:',
547 ['help', 'version', 'host=',
548 'save-headers=', 'save-body=', 'save-message=',
549 'set-header=', 'set-param='])
553 for option, value in options:
554 if option in ('-h', '--help'):
556 elif option in ('-V', '--version'):
562 elif option in ('-H', '--host'):
565 g.default_encoding = value
567 if value.startswith('*'):
568 g.decode_headers = []
569 g.decode_headers.append(value)
571 g.decode_headers = []
573 g.decode_header_params.append(value.split(':', 1))
575 g.decode_header_params = []
577 g.remove_headers.append(value)
579 g.remove_headers_params.append(value.split(':', 1))
580 elif option == '--set-header':
581 g.set_header_value.append(value.split(':', 1))
582 elif option == '--set-param':
583 header, value = value.split(':', 1)
585 param, value = value.split('=', 1)
587 param, value = value.split(':', 1)
588 g.set_header_param.append((header, param, value))
590 g.totext_mask.append(value)
592 g.binary_mask.append(value)
594 g.decoded_binary_mask.append(value)
596 g.fully_ignore_mask.append(value)
598 g.ignore_mask.append(value)
600 g.error_mask.append(value)
601 elif option == '--save-headers':
602 g.save_headers_mask.append(value)
603 elif option == '--save-body':
604 g.save_body_mask.append(value)
605 elif option == '--save-message':
606 g.save_message_mask.append(value)
608 g.destination_dir = value
610 g.output_filename = value
617 if __name__ == "__main__":
618 arguments = get_opts()
622 g.input_filename = '-'
624 if g.output_filename:
625 outfile = open_output_file(g.output_filename)
627 g.output_filename = '-'
630 if (arguments[0] == '-'):
631 g.input_filename = '-'
634 g.input_filename = arguments[0]
635 infile = open(arguments[0], 'r')
637 if g.output_filename:
638 outfile = open_output_file(g.output_filename)
640 g.output_filename = '-'
643 if g.output_filename:
644 usage(1, 'Too many output filenames')
645 if (arguments[1] == '-'):
646 g.output_filename = '-'
649 g.output_filename = arguments[1]
650 outfile = open_output_file(g.output_filename)
652 usage(1, 'Too many arguments')
654 if (infile is sys.stdin) and sys.stdin.isatty():
655 if (outfile is sys.stdout) and sys.stdout.isatty():
657 usage(1, 'Filtering from console is forbidden')
661 g.host_name = socket.gethostname()
664 if hasattr(outfile, 'buffer'):
666 if not isinstance(s, bytes):
667 s = s.encode(g.default_encoding, "replace")
668 outfile.buffer.write(s)
669 output = output_bytes
671 output = outfile.write
674 msg = email.message_from_file(infile)
676 for header, value in g.set_header_value:
677 set_header(msg, header, value)
679 for header, param, value in g.set_header_param:
681 msg.set_param(param, value, header)