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 header] [-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(msg):
101 "Decode message headers according to global options"
103 for header in gopts.remove_headers:
106 for header in gopts.remove_all_params:
108 if value is None: # No such header
110 if ';' not in value: # There are no parameters
112 del msg[header] # Delete all such headers
113 # Get the value without parameters and set it back
114 msg[header] = value.split(';')[0].strip()
116 for header, param in gopts.remove_header_params:
117 msg.del_param(param, header)
119 for header_list in gopts.decode_headers:
120 header_list = header_list.split(',')
121 if header_list[0] == '*': # Decode all headers except listed
122 header_list = _get_exceptions(header_list)
123 for header in msg.keys():
124 if header.lower() not in header_list:
125 decode_header(msg, header)
126 else: # Decode listed headers
127 for header in header_list:
128 decode_header(msg, header)
130 for header_list, param_list in gopts.decode_header_params:
131 header_list = header_list.split(',')
132 param_list = param_list.split(',')
133 if header_list[0] == '*': # Decode for all headers except listed
134 header_list = _get_exceptions(header_list)
135 for header in msg.keys():
136 if header.lower() not in header_list:
137 for param in param_list:
138 decode_header_param(msg, header, param)
139 else: # Decode for listed headers
140 for header in header_list:
141 for param in param_list:
142 decode_header_param(msg, header, param)
145 def set_header(msg, header, value):
148 if msg.has_key(header):
149 msg.replace_header(header, value)
154 def set_content_type(msg, newtype, charset=None):
155 msg.set_type(newtype)
158 msg.set_param("charset", charset, "Content-Type")
161 caps = None # Globally stored mailcap database; initialized only if needed
163 def decode_body(msg, s):
164 "Decode body to plain text using first copiousoutput filter from mailcap"
166 import mailcap, tempfile
170 caps = mailcap.getcaps()
172 content_type = msg.get_content_type()
173 filename = tempfile.mktemp()
176 entries = mailcap.lookup(caps, content_type, "view")
177 for entry in entries:
178 if entry.has_key('copiousoutput'):
179 if entry.has_key('test'):
180 test = mailcap.subst(entry['test'], content_type, filename)
181 if test and os.system(test) != 0:
183 command = mailcap.subst(entry["view"], content_type, filename)
189 file = open(filename, 'w')
193 pipe = os.popen(command, 'r')
198 set_content_type(msg, "text/plain")
199 msg["X-MIME-Autoconverted"] = "from %s to text/plain by %s id %s" % (content_type, gopts.host_name, command.split()[0])
204 def recode_charset(msg, s):
205 "Recode charset of the message to the default charset"
207 save_charset = charset = msg.get_content_charset()
208 if charset and charset.lower() <> gopts.default_encoding:
209 s = recode_if_needed(s, charset)
210 content_type = msg.get_content_type()
211 set_content_type(msg, content_type, gopts.default_encoding)
212 msg["X-MIME-Autoconverted"] = "from %s to %s by %s id %s" % (save_charset, gopts.default_encoding, gopts.host_name, me)
216 def totext(msg, instring):
217 "Convert instring content to text"
219 # Decode body and recode charset
220 s = decode_body(msg, instring)
221 if gopts.recode_charset:
222 s = recode_charset(msg, s)
228 def decode_part(msg):
229 "Decode one part of the message"
232 encoding = msg["Content-Transfer-Encoding"]
234 if encoding in (None, '', '7bit', '8bit', 'binary'):
235 outstring = str(msg.get_payload())
236 else: # Decode from transfer ecoding to text or binary form
237 outstring = str(msg.get_payload(decode=1))
238 set_header(msg, "Content-Transfer-Encoding", "8bit")
239 msg["X-MIME-Autoconverted"] = "from %s to 8bit by %s id %s" % (encoding, gopts.host_name, me)
241 # Test all mask lists and find what to do with this content type
243 ctype = msg.get_content_type()
246 mtype = msg.get_content_maintype()
248 masks.append(mtype + '/*')
251 for content_type in masks:
252 if content_type in gopts.totext_mask:
253 totext(msg, outstring)
255 elif content_type in gopts.binary_mask:
259 elif content_type in gopts.ignore_mask:
261 output("\nMessage body of type `%s' skipped.\n" % content_type)
263 elif content_type in gopts.error_mask:
264 raise ValueError, "content type `%s' prohibited" % content_type
266 # Neither content type nor masks were listed - decode by default
267 totext(msg, outstring)
270 def decode_multipart(msg):
276 if msg.preamble: # Preserve the first part, it is probably not a RFC822-message
277 output(msg.preamble) # Usually it is just a few lines of text (MIME warning)
279 boundary = msg.get_boundary()
281 for subpart in msg.get_payload():
283 output("\n--%s\n" % boundary)
285 # Recursively decode all parts of the subpart
286 decode_message(subpart)
289 output("\n--%s--\n" % boundary)
295 def decode_message(msg):
298 if msg.is_multipart():
299 decode_multipart(msg)
300 elif len(msg): # Simple one-part message (there are headers) - decode it
302 else: # Not a message, just text - copy it literally
303 output(msg.as_string())
307 from m_lib.defenc import default_encoding
308 recode_charset = 1 # recode charset of message body
312 # A list of headers to decode
313 decode_headers = ["From", "To", "Cc", "Reply-To", "Mail-Followup-To",
316 # A list of headers parameters to decode
317 decode_header_params = [
318 ("Content-Type", "name"),
319 ("Content-Disposition", "filename"),
322 # A list of headers to remove
324 # A list of headers parameters to remove
325 remove_header_params = []
326 # A list of headers to be stripped of all parameters
327 remove_all_params = []
329 totext_mask = [] # A list of content-types to decode
330 binary_mask = [] # A list to pass through
331 ignore_mask = [] # Ignore (skip, do not decode and do not include into output)
332 error_mask = [] # Raise error if encounter one of these
334 input_filename = None
335 output_filename = None
337 gopts = GlobalOptions
341 from getopt import getopt, GetoptError
344 options, arguments = getopt(sys.argv[1:],
345 'hVcCDPH:f:d:p:r:R:b:e:i:t:o:',
346 ['help', 'version', 'host=', 'remove-params='])
350 for option, value in options:
351 if option in ('-h', '--help'):
353 elif option in ('-V', '--version'):
356 gopts.recode_charset = 1
358 gopts.recode_charset = 0
359 elif option in ('-H', '--host'):
360 gopts.host_name = value
362 gopts.default_encoding = value
364 if value.startswith('*'):
365 gopts.decode_headers = []
366 gopts.decode_headers.append(value)
368 gopts.decode_headers = []
370 gopts.decode_header_params.append(value.split(':', 1))
372 gopts.decode_header_params = []
374 gopts.remove_headers.append(value)
376 gopts.remove_header_params.append(value.split(':', 1))
377 elif option == '--remove-params':
378 gopts.remove_all_params.append(value)
380 gopts.totext_mask.append(value)
382 gopts.binary_mask.append(value)
384 gopts.ignore_mask.append(value)
386 gopts.error_mask.append(value)
388 gopts.output_filename = value
395 if __name__ == "__main__":
396 arguments = get_opt()
400 gopts.input_filename = '-'
402 if gopts.output_filename:
403 outfile = open(gopts.output_filename, 'w')
405 gopts.output_filename = '-'
408 if (arguments[0] == '-'):
409 gopts.input_filename = '-'
412 gopts.input_filename = arguments[0]
413 infile = open(arguments[0], 'r')
415 if gopts.output_filename:
416 outfile = open(gopts.output_filename, 'w')
418 gopts.output_filename = '-'
421 if gopts.output_filename:
422 usage(1, 'Too many output filenames')
423 if (arguments[1] == '-'):
424 gopts.output_filename = '-'
427 gopts.output_filename = arguments[1]
428 outfile = open(arguments[1], 'w')
430 usage(1, 'Too many arguments')
432 if (infile is sys.stdin) and sys.stdin.isatty():
433 if (outfile is sys.stdout) and sys.stdout.isatty():
435 usage(1, 'Filtering from console is forbidden')
437 if not gopts.host_name:
439 gopts.host_name = socket.gethostname()
441 gopts.outfile = outfile
442 output = outfile.write
445 decode_message(email.message_from_file(infile))