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):
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:
66 charset = g.default_encoding
67 rtn.append(recode_if_needed(atom, charset))
69 del rtn[-1] # remove the last space
71 # Now that we've decoded everything, we just need to join all the parts
72 # together into the final string.
75 def decode_header(msg, header):
76 "Decode mail header (if exists) and put it back, if it was encoded"
80 new_value = _decode_header(value)
81 if new_value != value: # do not bother to touch msg if not changed
82 set_header(msg, header, new_value)
85 def decode_header_param(msg, header, param):
86 "Decode mail header's parameter (if exists) and put it back, if it was encoded"
89 value = msg.get_param(param, header=header)
91 if isinstance(value, tuple):
92 new_value = recode_if_needed(value[2], value[0])
94 new_value = _decode_header(value)
95 if new_value != value: # do not bother to touch msg if not changed
96 msg.set_param(param, new_value, header)
99 def _get_exceptions(list):
100 return [x[1:].lower() for x in list[1:] if x[0]=='-']
102 def _decode_headers_params(msg, header, decode_all_params, param_list):
103 if decode_all_params:
104 params = msg.get_params(header=header)
106 for param, value in params:
107 if param not in param_list:
108 decode_header_param(msg, header, param)
110 for param in param_list:
111 decode_header_param(msg, header, param)
113 def _remove_headers_params(msg, header, remove_all_params, param_list):
114 if remove_all_params:
115 params = msg.get_params(header=header)
118 for param, value in params:
119 if param not in param_list:
120 msg.del_param(param, header)
123 if value is None: # No such header
125 if ';' not in value: # There are no parameters
127 del msg[header] # Delete all such headers
128 # Get the value without parameters and set it back
129 msg[header] = value.split(';')[0].strip()
131 for param in param_list:
132 msg.del_param(param, header)
134 def decode_headers(msg):
135 "Decode message headers according to global options"
137 for header_list in g.remove_headers:
138 header_list = header_list.split(',')
139 if header_list[0] == '*': # Remove all headers except listed
140 header_list = _get_exceptions(header_list)
141 for header in msg.keys():
142 if header.lower() not in header_list:
144 else: # Remove listed headers
145 for header in header_list:
148 for header_list, param_list in g.remove_headers_params:
149 header_list = header_list.split(',')
150 param_list = param_list.split(',')
151 remove_all_params = param_list[0] == '*' # Remove all params except listed
152 if remove_all_params:
153 param_list = _get_exceptions(param_list)
154 if header_list[0] == '*': # Remove for all headers except listed
155 header_list = _get_exceptions(header_list)
156 for header in msg.keys():
157 if header.lower() not in header_list:
158 _remove_headers_params(msg, header, remove_all_params, param_list)
159 else: # Decode for listed headers
160 for header in header_list:
161 _remove_headers_params(msg, header, remove_all_params, param_list)
163 for header_list in g.decode_headers:
164 header_list = header_list.split(',')
165 if header_list[0] == '*': # Decode all headers except listed
166 header_list = _get_exceptions(header_list)
167 for header in msg.keys():
168 if header.lower() not in header_list:
169 decode_header(msg, header)
170 else: # Decode listed headers
171 for header in header_list:
172 decode_header(msg, header)
174 for header_list, param_list in g.decode_header_params:
175 header_list = header_list.split(',')
176 param_list = param_list.split(',')
177 decode_all_params = param_list[0] == '*' # Decode all params except listed
178 if decode_all_params:
179 param_list = _get_exceptions(param_list)
180 if header_list[0] == '*': # Decode for all headers except listed
181 header_list = _get_exceptions(header_list)
182 for header in msg.keys():
183 if header.lower() not in header_list:
184 _decode_headers_params(msg, header, decode_all_params, param_list)
185 else: # Decode for listed headers
186 for header in header_list:
187 _decode_headers_params(msg, header, decode_all_params, param_list)
190 def set_header(msg, header, value):
194 msg.replace_header(header, value)
199 def set_content_type(msg, newtype, charset=None):
200 msg.set_type(newtype)
203 msg.set_param("charset", charset, "Content-Type")
206 caps = None # Globally stored mailcap database; initialized only if needed
208 def decode_body(msg, s):
209 "Decode body to plain text using first copiousoutput filter from mailcap"
211 import mailcap, tempfile
215 caps = mailcap.getcaps()
217 content_type = msg.get_content_type()
218 filename = tempfile.mktemp()
221 entries = mailcap.lookup(caps, content_type, "view")
222 for entry in entries:
223 if 'copiousoutput' in entry:
225 test = mailcap.subst(entry['test'], content_type, filename)
226 if test and os.system(test) != 0:
228 command = mailcap.subst(entry["view"], content_type, filename)
234 outfile = open(filename, 'wb')
235 if not isinstance(s, bytes):
236 s = s.encode(g.default_encoding, "replace")
240 pipe = os.popen(command, 'r')
245 set_content_type(msg, "text/plain")
246 msg["X-MIME-Autoconverted"] = "from %s to text/plain by %s id %s" % (content_type, g.host_name, command.split()[0])
251 def recode_charset(msg, s):
252 "Recode charset of the message to the default charset"
254 save_charset = charset = msg.get_content_charset()
255 if charset and charset.lower() != g.default_encoding:
256 s = recode_if_needed(s, charset)
257 content_type = msg.get_content_type()
258 set_content_type(msg, content_type, g.default_encoding)
259 msg["X-MIME-Autoconverted"] = "from %s to %s by %s id %s" % (save_charset, g.default_encoding, g.host_name, me)
263 def totext(msg, instring):
264 "Convert instring content to text"
266 # Decode body and recode charset
267 s = decode_body(msg, instring)
269 s = recode_charset(msg, s)
278 def _guess_extension(ctype):
280 if mimetypes is None:
283 user_mime_type = os.path.expanduser('~/.mime.types')
284 if os.path.exists(user_mime_type):
285 mimetypes._db.read(user_mime_type)
286 return mimetypes.guess_extension(ctype)
288 def _save_message(msg, outstring, save_headers=False, save_body=False):
289 for header, param in (
290 ("Content-Disposition", "filename"),
291 ("Content-Type", "name"),
293 fname = msg.get_param(param, header=header)
295 if isinstance(fname, tuple):
296 fname = fname[2] # Do not recode if it isn't recoded yet
298 for forbidden in chr(0), '/', '\\':
299 if forbidden in fname:
308 fname = str(g.save_counter) + fname
310 ext = _guess_extension(msg.get_content_type())
315 outfile = open_output_file(fname)
316 def _output_bytes(s):
317 if not isinstance(s, bytes):
318 s = s.encode(g.default_encoding, "replace")
320 output = _output_bytes
329 def decode_part(msg):
330 "Decode one part of the message"
334 # Test all mask lists and find what to do with this content type
336 ctype = msg.get_content_type()
339 mtype = ctype.split('/')[0]
340 masks.append(mtype + '/*')
344 for content_type in masks:
345 if content_type in g.totext_mask or \
346 content_type in g.decoded_binary_mask:
348 elif content_type in g.binary_mask:
351 elif content_type in g.fully_ignore_mask:
354 encoding = msg["Content-Transfer-Encoding"]
355 if left_binary or encoding in (None, '', '7bit', '8bit', 'binary'):
356 outstring = msg.get_payload()
357 else: # Decode from transfer ecoding to text or binary form
358 outstring = msg.get_payload(decode=1)
359 set_header(msg, "Content-Transfer-Encoding", "8bit")
360 msg["X-MIME-Autoconverted"] = "from %s to 8bit by %s id %s" % (encoding, g.host_name, me)
362 for content_type in masks:
363 if content_type in g.totext_mask:
364 outstring = totext(msg, outstring)
366 elif content_type in g.binary_mask or \
367 content_type in g.decoded_binary_mask:
371 elif content_type in g.ignore_mask:
373 output("%sMessage body of type %s skipped.%s" % (os.linesep, ctype, os.linesep))
375 elif content_type in g.error_mask:
378 # Neither content type nor masks were listed - decode by default
379 outstring = totext(msg, outstring)
381 for content_type in masks:
382 if content_type in g.save_headers_mask:
383 _save_message(msg, outstring, save_headers=True, save_body=False)
384 if content_type in g.save_body_mask:
385 _save_message(msg, outstring, save_headers=False, save_body=True)
386 if content_type in g.save_message_mask:
387 _save_message(msg, outstring, save_headers=True, save_body=True)
389 for content_type in masks:
390 if content_type in g.error_mask:
391 raise ValueError("content type %s prohibited" % ctype)
393 def decode_multipart(msg):
397 boundary = msg.get_boundary()
400 ctype = msg.get_content_type()
403 mtype = ctype.split('/')[0]
404 masks.append(mtype + '/*')
407 for content_type in masks:
408 if content_type in g.fully_ignore_mask:
410 elif content_type in g.ignore_mask:
412 output("%sMessage body of type %s skipped.%s" % (os.linesep, ctype, os.linesep))
414 output("%s--%s--%s" % (os.linesep, boundary, os.linesep))
417 for content_type in masks:
418 if content_type in g.save_body_mask or \
419 content_type in g.save_message_mask:
422 for subpart in msg.get_payload():
424 first_subpart = False
426 _out_l.append(os.linesep)
427 _out_l.append("--%s%s" % (boundary, os.linesep))
428 _out_l.append(subpart.as_string())
429 _out_l.append("%s--%s--%s" % (os.linesep, boundary, os.linesep))
430 outstring = ''.join(_out_l)
435 for content_type in masks:
436 if content_type in g.save_headers_mask:
437 _save_message(msg, outstring, save_headers=True, save_body=False)
438 if content_type in g.save_body_mask:
439 _save_message(msg, outstring, save_headers=False, save_body=True)
440 if content_type in g.save_message_mask:
441 _save_message(msg, outstring, save_headers=True, save_body=True)
443 for content_type in masks:
444 if content_type in g.error_mask:
445 raise ValueError("content type %s prohibited" % ctype)
449 if msg.preamble: # Preserve the first part, it is probably not a RFC822-message
450 output(msg.preamble) # Usually it is just a few lines of text (MIME warning)
451 if msg.preamble is not None:
455 for subpart in msg.get_payload():
458 first_subpart = False
461 output("--%s%s" % (boundary, os.linesep))
463 # Recursively decode all parts of the subpart
464 decode_message(subpart)
467 output("%s--%s--%s" % (os.linesep, boundary, os.linesep))
473 def decode_message(msg):
476 if msg.is_multipart():
477 decode_multipart(msg)
478 elif len(msg): # Simple one-part message (there are headers) - decode it
480 else: # Not a message, just text - copy it literally
481 output(msg.as_string())
484 def open_output_file(filename):
485 fullpath = os.path.abspath(os.path.join(g.destination_dir, filename))
486 full_dir = os.path.dirname(fullpath)
487 create = not os.path.isdir(full_dir)
489 os.makedirs(full_dir)
491 return open(fullpath, 'wb')
494 os.removedirs(full_dir)
498 from m_lib.defenc import default_encoding
499 recode_charset = 1 # recode charset of message body
503 # A list of headers to decode
504 decode_headers = ["From", "To", "Cc", "Reply-To", "Mail-Followup-To",
507 # A list of headers parameters to decode
508 decode_header_params = [
509 ("Content-Type", "name"),
510 ("Content-Disposition", "filename"),
513 # A list of headers to remove
515 # A list of headers parameters to remove
516 remove_headers_params = []
518 # A list of header/value pairs to set
519 set_header_value = []
520 # A list of header/parameter/value triples to set
521 set_header_param = []
523 totext_mask = [] # A list of content-types to decode
524 binary_mask = [] # A list of content-types to pass through
525 decoded_binary_mask = [] # A list of content-types to pass through (content-transfer-decoded)
526 ignore_mask = [] # Ignore (do not decode and do not include into output) but output a warning instead of the body
527 fully_ignore_mask = [] # Completely ignore - no headers, no body, no warning
528 error_mask = [] # Raise error if encounter one of these
531 save_headers_mask = []
533 save_message_mask = []
535 input_filename = None
536 output_filename = None
537 destination_dir = os.curdir
543 from getopt import getopt, GetoptError
546 options, arguments = getopt(sys.argv[1:],
547 'hVcCDPH:f:d:p:r:R:b:B:e:I:i:t:O:o:',
548 ['help', 'version', 'host=',
549 'save-headers=', 'save-body=', 'save-message=',
550 'set-header=', 'set-param='])
554 for option, value in options:
555 if option in ('-h', '--help'):
557 elif option in ('-V', '--version'):
563 elif option in ('-H', '--host'):
566 g.default_encoding = value
568 if value.startswith('*'):
569 g.decode_headers = []
570 g.decode_headers.append(value)
572 g.decode_headers = []
574 g.decode_header_params.append(value.split(':', 1))
576 g.decode_header_params = []
578 g.remove_headers.append(value)
580 g.remove_headers_params.append(value.split(':', 1))
581 elif option == '--set-header':
582 g.set_header_value.append(value.split(':', 1))
583 elif option == '--set-param':
584 header, value = value.split(':', 1)
586 param, value = value.split('=', 1)
588 param, value = value.split(':', 1)
589 g.set_header_param.append((header, param, value))
591 g.totext_mask.append(value)
593 g.binary_mask.append(value)
595 g.decoded_binary_mask.append(value)
597 g.fully_ignore_mask.append(value)
599 g.ignore_mask.append(value)
601 g.error_mask.append(value)
602 elif option == '--save-headers':
603 g.save_headers_mask.append(value)
604 elif option == '--save-body':
605 g.save_body_mask.append(value)
606 elif option == '--save-message':
607 g.save_message_mask.append(value)
609 g.destination_dir = value
611 g.output_filename = value
618 if __name__ == "__main__":
619 arguments = get_opts()
623 g.input_filename = '-'
625 if g.output_filename:
626 outfile = open_output_file(g.output_filename)
628 g.output_filename = '-'
631 if (arguments[0] == '-'):
632 g.input_filename = '-'
635 g.input_filename = arguments[0]
636 infile = open(arguments[0], 'r')
638 if g.output_filename:
639 outfile = open_output_file(g.output_filename)
641 g.output_filename = '-'
644 if g.output_filename:
645 usage(1, 'Too many output filenames')
646 if (arguments[1] == '-'):
647 g.output_filename = '-'
650 g.output_filename = arguments[1]
651 outfile = open_output_file(g.output_filename)
653 usage(1, 'Too many arguments')
655 if (infile is sys.stdin) and sys.stdin.isatty():
656 if (outfile is sys.stdout) and sys.stdout.isatty():
658 usage(1, 'Filtering from console is forbidden')
662 g.host_name = socket.gethostname()
665 if hasattr(outfile, 'buffer'):
667 if not isinstance(s, bytes):
668 s = s.encode(g.default_encoding, "replace")
669 outfile.buffer.write(s)
670 output = output_bytes
672 output = outfile.write
675 msg = email.message_from_file(infile)
677 for header, value in g.set_header_value:
678 set_header(msg, header, value)
680 for header, param, value in g.set_header_param:
682 msg.set_param(param, value, header)