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)
106 for header in gopts.remove_headers:
110 def set_header(msg, header, value):
113 if msg.has_key(header):
114 msg.replace_header(header, value)
119 def set_content_type(msg, newtype, charset=None):
120 msg.set_type(newtype)
123 msg.set_param("charset", charset, "Content-Type")
126 caps = None # Globally stored mailcap database; initialized only if needed
128 def decode_body(msg, s):
129 "Decode body to plain text using first copiousoutput filter from mailcap"
131 import mailcap, tempfile
135 caps = mailcap.getcaps()
137 content_type = msg.get_content_type()
138 filename = tempfile.mktemp()
141 entries = mailcap.lookup(caps, content_type, "view")
142 for entry in entries:
143 if entry.has_key('copiousoutput'):
144 if entry.has_key('test'):
145 test = mailcap.subst(entry['test'], content_type, filename)
146 if test and os.system(test) != 0:
148 command = mailcap.subst(entry["view"], content_type, filename)
154 file = open(filename, 'w')
158 pipe = os.popen(command, 'r')
163 set_content_type(msg, "text/plain")
164 msg["X-MIME-Autoconverted"] = "from %s to text/plain by %s id %s" % (content_type, gopts.host_name, command.split()[0])
169 def recode_charset(msg, s):
170 "Recode charset of the message to the default charset"
172 save_charset = charset = msg.get_content_charset()
173 if charset and charset.lower() <> gopts.default_encoding:
174 s = recode_if_needed(s, charset)
175 content_type = msg.get_content_type()
176 set_content_type(msg, content_type, gopts.default_encoding)
177 msg["X-MIME-Autoconverted"] = "from %s to %s by %s id %s" % (save_charset, gopts.default_encoding, gopts.host_name, me)
181 def totext(msg, instring):
182 "Convert instring content to text"
184 # Decode body and recode charset
185 s = decode_body(msg, instring)
186 if gopts.recode_charset:
187 s = recode_charset(msg, s)
193 def decode_part(msg):
194 "Decode one part of the message"
197 encoding = msg["Content-Transfer-Encoding"]
199 if encoding in (None, '', '7bit', '8bit', 'binary'):
200 outstring = str(msg.get_payload())
201 else: # Decode from transfer ecoding to text or binary form
202 outstring = str(msg.get_payload(decode=1))
203 set_header(msg, "Content-Transfer-Encoding", "8bit")
204 msg["X-MIME-Autoconverted"] = "from %s to 8bit by %s id %s" % (encoding, gopts.host_name, me)
206 # Test all mask lists and find what to do with this content type
208 ctype = msg.get_content_type()
211 mtype = msg.get_content_maintype()
213 masks.append(mtype + '/*')
216 for content_type in masks:
217 if content_type in gopts.totext_mask:
218 totext(msg, outstring)
220 elif content_type in gopts.binary_mask:
224 elif content_type in gopts.ignore_mask:
226 output("\nMessage body of type `%s' skipped.\n" % content_type)
228 elif content_type in gopts.error_mask:
229 raise ValueError, "content type `%s' prohibited" % content_type
231 # Neither content type nor masks were listed - decode by default
232 totext(msg, outstring)
235 def decode_multipart(msg):
241 if msg.preamble: # Preserve the first part, it is probably not a RFC822-message
242 output(msg.preamble) # Usually it is just a few lines of text (MIME warning)
244 boundary = msg.get_boundary()
246 for subpart in msg.get_payload():
248 output("\n--%s\n" % boundary)
250 # Recursively decode all parts of the subpart
251 decode_message(subpart)
254 output("\n--%s--\n" % boundary)
260 def decode_message(msg):
263 if msg.is_multipart():
264 decode_multipart(msg)
265 elif len(msg): # Simple one-part message (there are headers) - decode it
267 else: # Not a message, just text - copy it literally
268 output(msg.as_string())
272 from m_lib.defenc import default_encoding
273 recode_charset = 1 # recode charset of message body
277 # A list of headers to decode
278 decode_headers = ["From", "To", "Cc", "Reply-To", "Mail-Followup-To",
281 # A list of headers' parameters to decode
282 decode_header_params = [
283 ("Content-Type", "name"),
284 ("Content-Disposition", "filename"),
287 # A list of headers to remove
290 totext_mask = [] # A list of content-types to decode
291 binary_mask = [] # A list to pass through
292 ignore_mask = [] # Ignore (skip, do not decode and do not include into output)
293 error_mask = [] # Raise error if encounter one of these
295 input_filename = None
296 output_filename = None
298 gopts = GlobalOptions
302 from getopt import getopt, GetoptError
305 options, arguments = getopt(sys.argv[1:], 'hVcCDPH:f:d:p:r:b:e:i:t:o:',
306 ['help', 'version', 'host'])
310 for option, value in options:
311 if option in ('-h', '--help'):
313 elif option in ('-V', '--version'):
316 gopts.recode_charset = 1
318 gopts.recode_charset = 0
319 elif option in ('-H', '--host'):
320 gopts.host_name = value
322 gopts.default_encoding = value
324 gopts.decode_headers.append(value)
326 gopts.decode_headers = []
328 gopts.decode_header_params.append(value.split(':', 1))
330 gopts.decode_header_params = []
332 gopts.remove_headers.append(value)
334 gopts.totext_mask.append(value)
336 gopts.binary_mask.append(value)
338 gopts.ignore_mask.append(value)
340 gopts.error_mask.append(value)
342 gopts.output_filename = value
349 if __name__ == "__main__":
350 arguments = get_opt()
354 gopts.input_filename = '-'
356 if gopts.output_filename:
357 outfile = open(gopts.output_filename, 'w')
359 gopts.output_filename = '-'
362 if (arguments[0] == '-'):
363 gopts.input_filename = '-'
366 gopts.input_filename = arguments[0]
367 infile = open(arguments[0], 'r')
369 if gopts.output_filename:
370 outfile = open(gopts.output_filename, 'w')
372 gopts.output_filename = '-'
375 if gopts.output_filename:
376 usage(1, 'Too many output filenames')
377 if (arguments[1] == '-'):
378 gopts.output_filename = '-'
381 gopts.output_filename = arguments[1]
382 outfile = open(arguments[1], 'w')
384 usage(1, 'Too many arguments')
386 if (infile is sys.stdin) and sys.stdin.isatty():
387 if (outfile is sys.stdout) and sys.stdout.isatty():
389 usage(1, 'Filtering from console is forbidden')
391 if not gopts.host_name:
393 gopts.host_name = socket.gethostname()
395 gopts.outfile = outfile
396 output = outfile.write
399 decode_message(email.message_from_file(infile))