2 """Decode MIME message"""
4 from mimedecode_version import __version__, __author__, __copyright__, __license__
9 me = os.path.basename(sys.argv[0])
14 Broytman mimedecode.py version %s, %s
15 """ % (__version__, __copyright__))
18 def usage(code=0, errormsg=''):
21 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 header:param] [--remove-params=header] [-beit mask] [-o output_file] [input_file [output_file]]
24 sys.stderr.write(errormsg + '\n')
28 def output_headers(msg):
29 unix_from = msg.get_unixfrom()
31 output(unix_from + '\n')
32 for key, value in msg.items():
33 output("%s: %s\n" % (key, value))
34 output("\n") # End of headers
37 def recode(s, charset):
38 return unicode(s, charset, "replace").encode(gopts.default_encoding, "replace")
40 def recode_if_needed(s, charset):
41 if charset and charset.lower() <> gopts.default_encoding:
42 s = recode(s, charset)
46 def _decode_header(s):
47 """Return a decoded string according to RFC 2047.
48 NOTE: This is almost the same as email.Utils.decode.
52 L = email.Header.decode_header(s)
53 if not isinstance(L, list):
58 for atom, charset in L:
62 rtn.append(recode_if_needed(atom, charset))
64 del rtn[-1] # remove the last space
66 # Now that we've decoded everything, we just need to join all the parts
67 # together into the final string.
70 def decode_header(msg, header):
71 "Decode mail header (if exists) and put it back, if it was encoded"
73 if msg.has_key(header):
75 new_value = _decode_header(value)
76 if new_value <> value: # do not bother to touch msg if not changed
77 set_header(msg, header, new_value)
80 def _decode_header_param(s):
81 return recode_if_needed(s[2], s[0])
83 def decode_header_param(msg, header, param):
84 "Decode mail header's parameter (if exists) and put it back, if it was encoded"
86 if msg.has_key(header):
87 value = msg.get_param(param, header=header)
89 if isinstance(value, tuple):
90 new_value = _decode_header_param(value)
92 new_value = _decode_header(value)
93 if new_value <> value: # do not bother to touch msg if not changed
94 msg.set_param(param, new_value, header)
97 def _get_exceptions(list):
98 return [x[1:].lower() for x in list[1:] if x[0]=='-']
100 def _decode_headers_params(msg, header, decode_all_params, param_list):
101 if decode_all_params:
102 params = msg.get_params(header=header)
104 for param, value in params:
105 if param not in param_list:
106 decode_header_param(msg, header, param)
108 for param in param_list:
109 decode_header_param(msg, header, param)
111 def decode_headers(msg):
112 "Decode message headers according to global options"
114 for header_list in gopts.remove_headers:
115 header_list = header_list.split(',')
116 if header_list[0] == '*': # Remove all headers except listed
117 header_list = _get_exceptions(header_list)
118 for header in msg.keys():
119 if header.lower() not in header_list:
121 else: # Remove listed headers
122 for header in header_list:
125 for header in gopts.remove_all_params:
127 if value is None: # No such header
129 if ';' not in value: # There are no parameters
131 del msg[header] # Delete all such headers
132 # Get the value without parameters and set it back
133 msg[header] = value.split(';')[0].strip()
135 for header, param in gopts.remove_header_params:
136 msg.del_param(param, header)
138 for header_list in gopts.decode_headers:
139 header_list = header_list.split(',')
140 if header_list[0] == '*': # Decode all headers except listed
141 header_list = _get_exceptions(header_list)
142 for header in msg.keys():
143 if header.lower() not in header_list:
144 decode_header(msg, header)
145 else: # Decode listed headers
146 for header in header_list:
147 decode_header(msg, header)
149 for header_list, param_list in gopts.decode_header_params:
150 header_list = header_list.split(',')
151 param_list = param_list.split(',')
152 decode_all_params = param_list[0] == '*' # Decode all params except listed
153 if decode_all_params:
154 param_list = _get_exceptions(param_list)
155 if header_list[0] == '*': # Decode for all headers except listed
156 header_list = _get_exceptions(header_list)
157 for header in msg.keys():
158 if header.lower() not in header_list:
159 _decode_headers_params(msg, header, decode_all_params, param_list)
160 else: # Decode for listed headers
161 for header in header_list:
162 _decode_headers_params(msg, header, decode_all_params, param_list)
165 def set_header(msg, header, value):
168 if msg.has_key(header):
169 msg.replace_header(header, value)
174 def set_content_type(msg, newtype, charset=None):
175 msg.set_type(newtype)
178 msg.set_param("charset", charset, "Content-Type")
181 caps = None # Globally stored mailcap database; initialized only if needed
183 def decode_body(msg, s):
184 "Decode body to plain text using first copiousoutput filter from mailcap"
186 import mailcap, tempfile
190 caps = mailcap.getcaps()
192 content_type = msg.get_content_type()
193 filename = tempfile.mktemp()
196 entries = mailcap.lookup(caps, content_type, "view")
197 for entry in entries:
198 if entry.has_key('copiousoutput'):
199 if entry.has_key('test'):
200 test = mailcap.subst(entry['test'], content_type, filename)
201 if test and os.system(test) != 0:
203 command = mailcap.subst(entry["view"], content_type, filename)
209 file = open(filename, 'w')
213 pipe = os.popen(command, 'r')
218 set_content_type(msg, "text/plain")
219 msg["X-MIME-Autoconverted"] = "from %s to text/plain by %s id %s" % (content_type, gopts.host_name, command.split()[0])
224 def recode_charset(msg, s):
225 "Recode charset of the message to the default charset"
227 save_charset = charset = msg.get_content_charset()
228 if charset and charset.lower() <> gopts.default_encoding:
229 s = recode_if_needed(s, charset)
230 content_type = msg.get_content_type()
231 set_content_type(msg, content_type, gopts.default_encoding)
232 msg["X-MIME-Autoconverted"] = "from %s to %s by %s id %s" % (save_charset, gopts.default_encoding, gopts.host_name, me)
236 def totext(msg, instring):
237 "Convert instring content to text"
239 # Decode body and recode charset
240 s = decode_body(msg, instring)
241 if gopts.recode_charset:
242 s = recode_charset(msg, s)
248 def decode_part(msg):
249 "Decode one part of the message"
252 encoding = msg["Content-Transfer-Encoding"]
254 if encoding in (None, '', '7bit', '8bit', 'binary'):
255 outstring = str(msg.get_payload())
256 else: # Decode from transfer ecoding to text or binary form
257 outstring = str(msg.get_payload(decode=1))
258 set_header(msg, "Content-Transfer-Encoding", "8bit")
259 msg["X-MIME-Autoconverted"] = "from %s to 8bit by %s id %s" % (encoding, gopts.host_name, me)
261 # Test all mask lists and find what to do with this content type
263 ctype = msg.get_content_type()
266 mtype = msg.get_content_maintype()
268 masks.append(mtype + '/*')
271 for content_type in masks:
272 if content_type in gopts.totext_mask:
273 totext(msg, outstring)
275 elif content_type in gopts.binary_mask:
279 elif content_type in gopts.ignore_mask:
281 output("\nMessage body of type `%s' skipped.\n" % content_type)
283 elif content_type in gopts.error_mask:
284 raise ValueError, "content type `%s' prohibited" % content_type
286 # Neither content type nor masks were listed - decode by default
287 totext(msg, outstring)
290 def decode_multipart(msg):
296 if msg.preamble: # Preserve the first part, it is probably not a RFC822-message
297 output(msg.preamble) # Usually it is just a few lines of text (MIME warning)
299 boundary = msg.get_boundary()
301 for subpart in msg.get_payload():
303 output("\n--%s\n" % boundary)
305 # Recursively decode all parts of the subpart
306 decode_message(subpart)
309 output("\n--%s--\n" % boundary)
315 def decode_message(msg):
318 if msg.is_multipart():
319 decode_multipart(msg)
320 elif len(msg): # Simple one-part message (there are headers) - decode it
322 else: # Not a message, just text - copy it literally
323 output(msg.as_string())
327 from m_lib.defenc import default_encoding
328 recode_charset = 1 # recode charset of message body
332 # A list of headers to decode
333 decode_headers = ["From", "To", "Cc", "Reply-To", "Mail-Followup-To",
336 # A list of headers parameters to decode
337 decode_header_params = [
338 ("Content-Type", "name"),
339 ("Content-Disposition", "filename"),
342 # A list of headers to remove
344 # A list of headers parameters to remove
345 remove_header_params = []
346 # A list of headers to be stripped of all parameters
347 remove_all_params = []
349 totext_mask = [] # A list of content-types to decode
350 binary_mask = [] # A list to pass through
351 ignore_mask = [] # Ignore (skip, do not decode and do not include into output)
352 error_mask = [] # Raise error if encounter one of these
354 input_filename = None
355 output_filename = None
357 gopts = GlobalOptions
361 from getopt import getopt, GetoptError
364 options, arguments = getopt(sys.argv[1:],
365 'hVcCDPH:f:d:p:r:R:b:e:i:t:o:',
366 ['help', 'version', 'host=', 'remove-params='])
370 for option, value in options:
371 if option in ('-h', '--help'):
373 elif option in ('-V', '--version'):
376 gopts.recode_charset = 1
378 gopts.recode_charset = 0
379 elif option in ('-H', '--host'):
380 gopts.host_name = value
382 gopts.default_encoding = value
384 if value.startswith('*'):
385 gopts.decode_headers = []
386 gopts.decode_headers.append(value)
388 gopts.decode_headers = []
390 gopts.decode_header_params.append(value.split(':', 1))
392 gopts.decode_header_params = []
394 gopts.remove_headers.append(value)
396 gopts.remove_header_params.append(value.split(':', 1))
397 elif option == '--remove-params':
398 gopts.remove_all_params.append(value)
400 gopts.totext_mask.append(value)
402 gopts.binary_mask.append(value)
404 gopts.ignore_mask.append(value)
406 gopts.error_mask.append(value)
408 gopts.output_filename = value
415 if __name__ == "__main__":
416 arguments = get_opt()
420 gopts.input_filename = '-'
422 if gopts.output_filename:
423 outfile = open(gopts.output_filename, 'w')
425 gopts.output_filename = '-'
428 if (arguments[0] == '-'):
429 gopts.input_filename = '-'
432 gopts.input_filename = arguments[0]
433 infile = open(arguments[0], 'r')
435 if gopts.output_filename:
436 outfile = open(gopts.output_filename, 'w')
438 gopts.output_filename = '-'
441 if gopts.output_filename:
442 usage(1, 'Too many output filenames')
443 if (arguments[1] == '-'):
444 gopts.output_filename = '-'
447 gopts.output_filename = arguments[1]
448 outfile = open(arguments[1], 'w')
450 usage(1, 'Too many arguments')
452 if (infile is sys.stdin) and sys.stdin.isatty():
453 if (outfile is sys.stdout) and sys.stdout.isatty():
455 usage(1, 'Filtering from console is forbidden')
457 if not gopts.host_name:
459 gopts.host_name = socket.gethostname()
461 gopts.outfile = outfile
462 output = outfile.write
465 decode_message(email.message_from_file(infile))