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 header] [-p header:param] [-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 decode_headers(msg):
98 "Decode message headers according to global options"
100 for header in gopts.remove_headers:
103 for header in gopts.remove_all_params:
110 msg[header] = value.split(';')[0].strip()
112 for header, param in gopts.remove_header_params:
113 msg.del_param(param, header)
115 for header in gopts.decode_headers:
116 decode_header(msg, header)
118 for header, param in gopts.decode_header_params:
119 decode_header_param(msg, header, param)
122 def set_header(msg, header, value):
125 if msg.has_key(header):
126 msg.replace_header(header, value)
131 def set_content_type(msg, newtype, charset=None):
132 msg.set_type(newtype)
135 msg.set_param("charset", charset, "Content-Type")
138 caps = None # Globally stored mailcap database; initialized only if needed
140 def decode_body(msg, s):
141 "Decode body to plain text using first copiousoutput filter from mailcap"
143 import mailcap, tempfile
147 caps = mailcap.getcaps()
149 content_type = msg.get_content_type()
150 filename = tempfile.mktemp()
153 entries = mailcap.lookup(caps, content_type, "view")
154 for entry in entries:
155 if entry.has_key('copiousoutput'):
156 if entry.has_key('test'):
157 test = mailcap.subst(entry['test'], content_type, filename)
158 if test and os.system(test) != 0:
160 command = mailcap.subst(entry["view"], content_type, filename)
166 file = open(filename, 'w')
170 pipe = os.popen(command, 'r')
175 set_content_type(msg, "text/plain")
176 msg["X-MIME-Autoconverted"] = "from %s to text/plain by %s id %s" % (content_type, gopts.host_name, command.split()[0])
181 def recode_charset(msg, s):
182 "Recode charset of the message to the default charset"
184 save_charset = charset = msg.get_content_charset()
185 if charset and charset.lower() <> gopts.default_encoding:
186 s = recode_if_needed(s, charset)
187 content_type = msg.get_content_type()
188 set_content_type(msg, content_type, gopts.default_encoding)
189 msg["X-MIME-Autoconverted"] = "from %s to %s by %s id %s" % (save_charset, gopts.default_encoding, gopts.host_name, me)
193 def totext(msg, instring):
194 "Convert instring content to text"
196 # Decode body and recode charset
197 s = decode_body(msg, instring)
198 if gopts.recode_charset:
199 s = recode_charset(msg, s)
205 def decode_part(msg):
206 "Decode one part of the message"
209 encoding = msg["Content-Transfer-Encoding"]
211 if encoding in (None, '', '7bit', '8bit', 'binary'):
212 outstring = str(msg.get_payload())
213 else: # Decode from transfer ecoding to text or binary form
214 outstring = str(msg.get_payload(decode=1))
215 set_header(msg, "Content-Transfer-Encoding", "8bit")
216 msg["X-MIME-Autoconverted"] = "from %s to 8bit by %s id %s" % (encoding, gopts.host_name, me)
218 # Test all mask lists and find what to do with this content type
220 ctype = msg.get_content_type()
223 mtype = msg.get_content_maintype()
225 masks.append(mtype + '/*')
228 for content_type in masks:
229 if content_type in gopts.totext_mask:
230 totext(msg, outstring)
232 elif content_type in gopts.binary_mask:
236 elif content_type in gopts.ignore_mask:
238 output("\nMessage body of type `%s' skipped.\n" % content_type)
240 elif content_type in gopts.error_mask:
241 raise ValueError, "content type `%s' prohibited" % content_type
243 # Neither content type nor masks were listed - decode by default
244 totext(msg, outstring)
247 def decode_multipart(msg):
253 if msg.preamble: # Preserve the first part, it is probably not a RFC822-message
254 output(msg.preamble) # Usually it is just a few lines of text (MIME warning)
256 boundary = msg.get_boundary()
258 for subpart in msg.get_payload():
260 output("\n--%s\n" % boundary)
262 # Recursively decode all parts of the subpart
263 decode_message(subpart)
266 output("\n--%s--\n" % boundary)
272 def decode_message(msg):
275 if msg.is_multipart():
276 decode_multipart(msg)
277 elif len(msg): # Simple one-part message (there are headers) - decode it
279 else: # Not a message, just text - copy it literally
280 output(msg.as_string())
284 from m_lib.defenc import default_encoding
285 recode_charset = 1 # recode charset of message body
289 # A list of headers to decode
290 decode_headers = ["From", "To", "Cc", "Reply-To", "Mail-Followup-To",
293 # A list of headers parameters to decode
294 decode_header_params = [
295 ("Content-Type", "name"),
296 ("Content-Disposition", "filename"),
299 # A list of headers to remove
301 # A list of headers parameters to remove
302 remove_header_params = []
303 # A list of headers to be stripped of all parameters
304 remove_all_params = []
306 totext_mask = [] # A list of content-types to decode
307 binary_mask = [] # A list to pass through
308 ignore_mask = [] # Ignore (skip, do not decode and do not include into output)
309 error_mask = [] # Raise error if encounter one of these
311 input_filename = None
312 output_filename = None
314 gopts = GlobalOptions
318 from getopt import getopt, GetoptError
321 options, arguments = getopt(sys.argv[1:],
322 'hVcCDPH:f:d:p:r:R:b:e:i:t:o:',
323 ['help', 'version', 'host=', 'remove-params='])
327 for option, value in options:
328 if option in ('-h', '--help'):
330 elif option in ('-V', '--version'):
333 gopts.recode_charset = 1
335 gopts.recode_charset = 0
336 elif option in ('-H', '--host'):
337 gopts.host_name = value
339 gopts.default_encoding = value
341 gopts.decode_headers.append(value)
343 gopts.decode_headers = []
345 gopts.decode_header_params.append(value.split(':', 1))
347 gopts.decode_header_params = []
349 gopts.remove_headers.append(value)
351 gopts.remove_header_params.append(value.split(':', 1))
352 elif option == '--remove-params':
353 gopts.remove_all_params.append(value)
355 gopts.totext_mask.append(value)
357 gopts.binary_mask.append(value)
359 gopts.ignore_mask.append(value)
361 gopts.error_mask.append(value)
363 gopts.output_filename = value
370 if __name__ == "__main__":
371 arguments = get_opt()
375 gopts.input_filename = '-'
377 if gopts.output_filename:
378 outfile = open(gopts.output_filename, 'w')
380 gopts.output_filename = '-'
383 if (arguments[0] == '-'):
384 gopts.input_filename = '-'
387 gopts.input_filename = arguments[0]
388 infile = open(arguments[0], 'r')
390 if gopts.output_filename:
391 outfile = open(gopts.output_filename, 'w')
393 gopts.output_filename = '-'
396 if gopts.output_filename:
397 usage(1, 'Too many output filenames')
398 if (arguments[1] == '-'):
399 gopts.output_filename = '-'
402 gopts.output_filename = arguments[1]
403 outfile = open(arguments[1], 'w')
405 usage(1, 'Too many arguments')
407 if (infile is sys.stdin) and sys.stdin.isatty():
408 if (outfile is sys.stdout) and sys.stdout.isatty():
410 usage(1, 'Filtering from console is forbidden')
412 if not gopts.host_name:
414 gopts.host_name = socket.gethostname()
416 gopts.outfile = outfile
417 output = outfile.write
420 decode_message(email.message_from_file(infile))