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.
50 from types import ListType
53 L = email.Header.decode_header(s)
54 if not isinstance(L, ListType):
59 for atom, charset in L:
63 rtn.append(recode_if_needed(atom, charset))
65 del rtn[-1] # remove the last space
67 # Now that we've decoded everything, we just need to join all the parts
68 # together into the final string.
71 def decode_header(msg, header):
72 "Decode mail header (if exists) and put it back, if it was encoded"
74 if msg.has_key(header):
76 new_value = _decode_header(value)
77 if new_value <> value: # do not bother to touch msg if not changed
78 set_header(msg, header, new_value)
81 def _decode_header_param(s):
82 return recode_if_needed(s[2], s[0])
84 def decode_header_param(msg, header, param):
85 "Decode mail header's parameter (if exists) and put it back, if it was encoded"
87 if msg.has_key(header):
88 value = msg.get_param(param, header=header)
90 from types import TupleType
91 if isinstance(value, TupleType):
92 new_value = _decode_header_param(value)
94 new_value = _decode_header(value)
95 if new_value <> value: # do not bother to touch msg if not changed
96 msg.set_param(param, new_value, header)
99 def decode_headers(msg):
100 "Decode message headers according to global options"
102 for header in gopts.decode_headers:
103 decode_header(msg, header)
105 for header, param in gopts.decode_header_params:
106 decode_header_param(msg, header, param)
109 def set_header(msg, header, value):
112 if msg.has_key(header):
113 msg.replace_header(header, value)
118 def set_content_type(msg, newtype, charset=None):
119 msg.set_type(newtype)
122 msg.set_param("charset", charset, "Content-Type")
125 caps = None # Globally stored mailcap database; initialized only if needed
127 def decode_body(msg, s):
128 "Decode body to plain text using first copiousoutput filter from mailcap"
130 import mailcap, tempfile
134 caps = mailcap.getcaps()
136 content_type = msg.get_content_type()
137 filename = tempfile.mktemp()
140 entries = mailcap.lookup(caps, content_type, "view")
141 for entry in entries:
142 if entry.has_key('copiousoutput'):
143 if entry.has_key('test'):
144 test = mailcap.subst(entry['test'], content_type, filename)
145 if test and os.system(test) != 0:
147 command = mailcap.subst(entry["view"], content_type, filename)
153 file = open(filename, 'w')
157 pipe = os.popen(command, 'r')
162 set_content_type(msg, "text/plain")
163 msg["X-MIME-Autoconverted"] = "from %s to text/plain by %s id %s" % (content_type, gopts.host_name, command.split()[0])
168 def recode_charset(msg, s):
169 "Recode charset of the message to the default charset"
171 save_charset = charset = msg.get_content_charset()
172 if charset and charset.lower() <> gopts.default_encoding:
173 s = recode_if_needed(s, charset)
174 content_type = msg.get_content_type()
175 set_content_type(msg, content_type, gopts.default_encoding)
176 msg["X-MIME-Autoconverted"] = "from %s to %s by %s id %s" % (save_charset, gopts.default_encoding, gopts.host_name, me)
180 def totext(msg, instring):
181 "Convert instring content to text"
183 # Decode body and recode charset
184 s = decode_body(msg, instring)
185 if gopts.recode_charset:
186 s = recode_charset(msg, s)
192 def decode_part(msg):
193 "Decode one part of the message"
196 encoding = msg["Content-Transfer-Encoding"]
198 if encoding in (None, '', '7bit', '8bit', 'binary'):
199 outstring = str(msg.get_payload())
200 else: # Decode from transfer ecoding to text or binary form
201 outstring = str(msg.get_payload(decode=1))
202 set_header(msg, "Content-Transfer-Encoding", "8bit")
203 msg["X-MIME-Autoconverted"] = "from %s to 8bit by %s id %s" % (encoding, gopts.host_name, me)
205 # Test all mask lists and find what to do with this content type
207 ctype = msg.get_content_type()
210 mtype = msg.get_content_maintype()
212 masks.append(mtype + '/*')
215 for content_type in masks:
216 if content_type in gopts.totext_mask:
217 totext(msg, outstring)
219 elif content_type in gopts.binary_mask:
223 elif content_type in gopts.ignore_mask:
225 output("\nMessage body of type `%s' skipped.\n" % content_type)
227 elif content_type in gopts.error_mask:
228 raise ValueError, "content type `%s' prohibited" % content_type
230 # Neither content type nor masks were listed - decode by default
231 totext(msg, outstring)
234 def decode_multipart(msg):
240 if msg.preamble: # Preserve the first part, it is probably not a RFC822-message
241 output(msg.preamble) # Usually it is just a few lines of text (MIME warning)
243 boundary = msg.get_boundary()
245 for subpart in msg.get_payload():
247 output("\n--%s\n" % boundary)
249 # Recursively decode all parts of the subpart
250 decode_message(subpart)
253 output("\n--%s--\n" % boundary)
259 def decode_message(msg):
262 if msg.is_multipart():
263 decode_multipart(msg)
264 elif msg.has_key("Content-Type"): # Simple one-part message - decode it
266 else: # Not a message, just text - copy it literally
267 output(msg.as_string())
271 from m_lib.defenc import default_encoding
272 recode_charset = 1 # recode charset of message body
276 decode_headers = ["From", "Subject"] # A list of headers to decode
277 decode_header_params = [
278 ("Content-Type", "name"),
279 ("Content-Disposition", "filename"),
280 ] # A list of headers' parameters to decode
282 totext_mask = [] # A list of content-types to decode
283 binary_mask = [] # A list to pass through
284 ignore_mask = [] # Ignore (skip, do not decode and do not include into output)
285 error_mask = [] # Raise error if encounter one of these
287 input_filename = None
288 output_filename = None
290 gopts = GlobalOptions
294 from getopt import getopt, GetoptError
297 options, arguments = getopt(sys.argv[1:], 'hVcCDPH:f:d:p:b:e:i:t:o:',
298 ['help', 'version', 'host'])
302 for option, value in options:
303 if option in ('-h', '--help'):
305 elif option in ('-V', '--version'):
308 gopts.recode_charset = 1
310 gopts.recode_charset = 0
311 elif option in ('-H', '--host'):
312 gopts.host_name = value
314 gopts.default_encoding = value
316 gopts.decode_headers.append(value)
318 gopts.decode_headers = []
320 gopts.decode_header_params.append(value.split(':', 1))
322 gopts.decode_header_params = []
324 gopts.totext_mask.append(value)
326 gopts.binary_mask.append(value)
328 gopts.ignore_mask.append(value)
330 gopts.error_mask.append(value)
332 gopts.output_filename = value
339 if __name__ == "__main__":
340 arguments = get_opt()
344 gopts.input_filename = '-'
346 if gopts.output_filename:
347 outfile = open(gopts.output_filename, 'w')
349 gopts.output_filename = '-'
352 if (arguments[0] == '-'):
353 gopts.input_filename = '-'
356 gopts.input_filename = arguments[0]
357 infile = open(arguments[0], 'r')
359 if gopts.output_filename:
360 outfile = open(gopts.output_filename, 'w')
362 gopts.output_filename = '-'
365 if gopts.output_filename:
366 usage(1, 'Too many output filenames')
367 if (arguments[1] == '-'):
368 gopts.output_filename = '-'
371 gopts.output_filename = arguments[1]
372 outfile = open(arguments[1], 'w')
374 usage(1, 'Too many arguments')
376 if (infile is sys.stdin) and sys.stdin.isatty():
377 if (outfile is sys.stdout) and sys.stdout.isatty():
379 usage(1, 'Filtering from console is forbidden')
381 if not gopts.host_name:
383 gopts.host_name = socket.gethostname()
385 gopts.outfile = outfile
386 output = outfile.write
389 decode_message(email.message_from_file(infile))