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] [-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.decode_headers:
101 decode_header(msg, header)
103 for header, param in gopts.decode_header_params:
104 decode_header_param(msg, header, param)
107 def set_header(msg, header, value):
110 if msg.has_key(header):
111 msg.replace_header(header, value)
116 def set_content_type(msg, newtype, charset=None):
117 msg.set_type(newtype)
120 msg.set_param("charset", charset, "Content-Type")
123 caps = None # Globally stored mailcap database; initialized only if needed
125 def decode_body(msg, s):
126 "Decode body to plain text using first copiousoutput filter from mailcap"
128 import mailcap, tempfile
132 caps = mailcap.getcaps()
134 content_type = msg.get_content_type()
135 filename = tempfile.mktemp()
138 entries = mailcap.lookup(caps, content_type, "view")
139 for entry in entries:
140 if entry.has_key('copiousoutput'):
141 if entry.has_key('test'):
142 test = mailcap.subst(entry['test'], content_type, filename)
143 if test and os.system(test) != 0:
145 command = mailcap.subst(entry["view"], content_type, filename)
151 file = open(filename, 'w')
155 pipe = os.popen(command, 'r')
160 set_content_type(msg, "text/plain")
161 msg["X-MIME-Autoconverted"] = "from %s to text/plain by %s id %s" % (content_type, gopts.host_name, command.split()[0])
166 def recode_charset(msg, s):
167 "Recode charset of the message to the default charset"
169 save_charset = charset = msg.get_content_charset()
170 if charset and charset.lower() <> gopts.default_encoding:
171 s = recode_if_needed(s, charset)
172 content_type = msg.get_content_type()
173 set_content_type(msg, content_type, gopts.default_encoding)
174 msg["X-MIME-Autoconverted"] = "from %s to %s by %s id %s" % (save_charset, gopts.default_encoding, gopts.host_name, me)
178 def totext(msg, instring):
179 "Convert instring content to text"
181 # Decode body and recode charset
182 s = decode_body(msg, instring)
183 if gopts.recode_charset:
184 s = recode_charset(msg, s)
190 def decode_part(msg):
191 "Decode one part of the message"
194 encoding = msg["Content-Transfer-Encoding"]
196 if encoding in (None, '', '7bit', '8bit', 'binary'):
197 outstring = str(msg.get_payload())
198 else: # Decode from transfer ecoding to text or binary form
199 outstring = str(msg.get_payload(decode=1))
200 set_header(msg, "Content-Transfer-Encoding", "8bit")
201 msg["X-MIME-Autoconverted"] = "from %s to 8bit by %s id %s" % (encoding, gopts.host_name, me)
203 # Test all mask lists and find what to do with this content type
205 ctype = msg.get_content_type()
208 mtype = msg.get_content_maintype()
210 masks.append(mtype + '/*')
213 for content_type in masks:
214 if content_type in gopts.totext_mask:
215 totext(msg, outstring)
217 elif content_type in gopts.binary_mask:
221 elif content_type in gopts.ignore_mask:
223 output("\nMessage body of type `%s' skipped.\n" % content_type)
225 elif content_type in gopts.error_mask:
226 raise ValueError, "content type `%s' prohibited" % content_type
228 # Neither content type nor masks were listed - decode by default
229 totext(msg, outstring)
232 def decode_multipart(msg):
238 if msg.preamble: # Preserve the first part, it is probably not a RFC822-message
239 output(msg.preamble) # Usually it is just a few lines of text (MIME warning)
241 boundary = msg.get_boundary()
243 for subpart in msg.get_payload():
245 output("\n--%s\n" % boundary)
247 # Recursively decode all parts of the subpart
248 decode_message(subpart)
251 output("\n--%s--\n" % boundary)
257 def decode_message(msg):
260 if msg.is_multipart():
261 decode_multipart(msg)
262 elif len(msg): # Simple one-part message (there are headers) - decode it
264 else: # Not a message, just text - copy it literally
265 output(msg.as_string())
269 from m_lib.defenc import default_encoding
270 recode_charset = 1 # recode charset of message body
274 # A list of headers to decode
275 decode_headers = ["From", "To", "Cc", "Reply-To", "Mail-Followup-To",
278 # A list of headers' parameters to decode
279 decode_header_params = [
280 ("Content-Type", "name"),
281 ("Content-Disposition", "filename"),
284 totext_mask = [] # A list of content-types to decode
285 binary_mask = [] # A list to pass through
286 ignore_mask = [] # Ignore (skip, do not decode and do not include into output)
287 error_mask = [] # Raise error if encounter one of these
289 input_filename = None
290 output_filename = None
292 gopts = GlobalOptions
296 from getopt import getopt, GetoptError
299 options, arguments = getopt(sys.argv[1:], 'hVcCDPH:f:d:p:b:e:i:t:o:',
300 ['help', 'version', 'host'])
304 for option, value in options:
305 if option in ('-h', '--help'):
307 elif option in ('-V', '--version'):
310 gopts.recode_charset = 1
312 gopts.recode_charset = 0
313 elif option in ('-H', '--host'):
314 gopts.host_name = value
316 gopts.default_encoding = value
318 gopts.decode_headers.append(value)
320 gopts.decode_headers = []
322 gopts.decode_header_params.append(value.split(':', 1))
324 gopts.decode_header_params = []
326 gopts.totext_mask.append(value)
328 gopts.binary_mask.append(value)
330 gopts.ignore_mask.append(value)
332 gopts.error_mask.append(value)
334 gopts.output_filename = value
341 if __name__ == "__main__":
342 arguments = get_opt()
346 gopts.input_filename = '-'
348 if gopts.output_filename:
349 outfile = open(gopts.output_filename, 'w')
351 gopts.output_filename = '-'
354 if (arguments[0] == '-'):
355 gopts.input_filename = '-'
358 gopts.input_filename = arguments[0]
359 infile = open(arguments[0], 'r')
361 if gopts.output_filename:
362 outfile = open(gopts.output_filename, 'w')
364 gopts.output_filename = '-'
367 if gopts.output_filename:
368 usage(1, 'Too many output filenames')
369 if (arguments[1] == '-'):
370 gopts.output_filename = '-'
373 gopts.output_filename = arguments[1]
374 outfile = open(arguments[1], 'w')
376 usage(1, 'Too many arguments')
378 if (infile is sys.stdin) and sys.stdin.isatty():
379 if (outfile is sys.stdout) and sys.stdout.isatty():
381 usage(1, 'Filtering from console is forbidden')
383 if not gopts.host_name:
385 gopts.host_name = socket.gethostname()
387 gopts.outfile = outfile
388 output = outfile.write
391 decode_message(email.message_from_file(infile))