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 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:
105 if value is None: # No such header
107 if ';' not in value: # There are no parameters
109 del msg[header] # Delete all such headers
110 # Get the value without parameters and set it back
111 msg[header] = value.split(';')[0].strip()
113 for header, param in gopts.remove_header_params:
114 msg.del_param(param, header)
116 for header_list in gopts.decode_headers:
117 header_list = header_list.split(',')
118 if header_list[0] == '*': # Decode all headers except listed
119 header_list = [h[1:].lower() for h in header_list[1:] if h[0]=='-']
120 for header in msg.keys():
121 if header.lower() not in header_list:
122 decode_header(msg, header)
123 else: # Decode listed hiders
124 for header in header_list:
125 decode_header(msg, header)
127 for header_list, param_list in gopts.decode_header_params:
128 header_list = header_list.split(',')
129 param_list = param_list.split(',')
130 for header in header_list:
131 for param in param_list:
132 decode_header_param(msg, header, param)
135 def set_header(msg, header, value):
138 if msg.has_key(header):
139 msg.replace_header(header, value)
144 def set_content_type(msg, newtype, charset=None):
145 msg.set_type(newtype)
148 msg.set_param("charset", charset, "Content-Type")
151 caps = None # Globally stored mailcap database; initialized only if needed
153 def decode_body(msg, s):
154 "Decode body to plain text using first copiousoutput filter from mailcap"
156 import mailcap, tempfile
160 caps = mailcap.getcaps()
162 content_type = msg.get_content_type()
163 filename = tempfile.mktemp()
166 entries = mailcap.lookup(caps, content_type, "view")
167 for entry in entries:
168 if entry.has_key('copiousoutput'):
169 if entry.has_key('test'):
170 test = mailcap.subst(entry['test'], content_type, filename)
171 if test and os.system(test) != 0:
173 command = mailcap.subst(entry["view"], content_type, filename)
179 file = open(filename, 'w')
183 pipe = os.popen(command, 'r')
188 set_content_type(msg, "text/plain")
189 msg["X-MIME-Autoconverted"] = "from %s to text/plain by %s id %s" % (content_type, gopts.host_name, command.split()[0])
194 def recode_charset(msg, s):
195 "Recode charset of the message to the default charset"
197 save_charset = charset = msg.get_content_charset()
198 if charset and charset.lower() <> gopts.default_encoding:
199 s = recode_if_needed(s, charset)
200 content_type = msg.get_content_type()
201 set_content_type(msg, content_type, gopts.default_encoding)
202 msg["X-MIME-Autoconverted"] = "from %s to %s by %s id %s" % (save_charset, gopts.default_encoding, gopts.host_name, me)
206 def totext(msg, instring):
207 "Convert instring content to text"
209 # Decode body and recode charset
210 s = decode_body(msg, instring)
211 if gopts.recode_charset:
212 s = recode_charset(msg, s)
218 def decode_part(msg):
219 "Decode one part of the message"
222 encoding = msg["Content-Transfer-Encoding"]
224 if encoding in (None, '', '7bit', '8bit', 'binary'):
225 outstring = str(msg.get_payload())
226 else: # Decode from transfer ecoding to text or binary form
227 outstring = str(msg.get_payload(decode=1))
228 set_header(msg, "Content-Transfer-Encoding", "8bit")
229 msg["X-MIME-Autoconverted"] = "from %s to 8bit by %s id %s" % (encoding, gopts.host_name, me)
231 # Test all mask lists and find what to do with this content type
233 ctype = msg.get_content_type()
236 mtype = msg.get_content_maintype()
238 masks.append(mtype + '/*')
241 for content_type in masks:
242 if content_type in gopts.totext_mask:
243 totext(msg, outstring)
245 elif content_type in gopts.binary_mask:
249 elif content_type in gopts.ignore_mask:
251 output("\nMessage body of type `%s' skipped.\n" % content_type)
253 elif content_type in gopts.error_mask:
254 raise ValueError, "content type `%s' prohibited" % content_type
256 # Neither content type nor masks were listed - decode by default
257 totext(msg, outstring)
260 def decode_multipart(msg):
266 if msg.preamble: # Preserve the first part, it is probably not a RFC822-message
267 output(msg.preamble) # Usually it is just a few lines of text (MIME warning)
269 boundary = msg.get_boundary()
271 for subpart in msg.get_payload():
273 output("\n--%s\n" % boundary)
275 # Recursively decode all parts of the subpart
276 decode_message(subpart)
279 output("\n--%s--\n" % boundary)
285 def decode_message(msg):
288 if msg.is_multipart():
289 decode_multipart(msg)
290 elif len(msg): # Simple one-part message (there are headers) - decode it
292 else: # Not a message, just text - copy it literally
293 output(msg.as_string())
297 from m_lib.defenc import default_encoding
298 recode_charset = 1 # recode charset of message body
302 # A list of headers to decode
303 decode_headers = ["From", "To", "Cc", "Reply-To", "Mail-Followup-To",
306 # A list of headers parameters to decode
307 decode_header_params = [
308 ("Content-Type", "name"),
309 ("Content-Disposition", "filename"),
312 # A list of headers to remove
314 # A list of headers parameters to remove
315 remove_header_params = []
316 # A list of headers to be stripped of all parameters
317 remove_all_params = []
319 totext_mask = [] # A list of content-types to decode
320 binary_mask = [] # A list to pass through
321 ignore_mask = [] # Ignore (skip, do not decode and do not include into output)
322 error_mask = [] # Raise error if encounter one of these
324 input_filename = None
325 output_filename = None
327 gopts = GlobalOptions
331 from getopt import getopt, GetoptError
334 options, arguments = getopt(sys.argv[1:],
335 'hVcCDPH:f:d:p:r:R:b:e:i:t:o:',
336 ['help', 'version', 'host=', 'remove-params='])
340 for option, value in options:
341 if option in ('-h', '--help'):
343 elif option in ('-V', '--version'):
346 gopts.recode_charset = 1
348 gopts.recode_charset = 0
349 elif option in ('-H', '--host'):
350 gopts.host_name = value
352 gopts.default_encoding = value
354 if value.startswith('*'):
355 gopts.decode_headers = []
356 gopts.decode_headers.append(value)
358 gopts.decode_headers = []
360 gopts.decode_header_params.append(value.split(':', 1))
362 gopts.decode_header_params = []
364 gopts.remove_headers.append(value)
366 gopts.remove_header_params.append(value.split(':', 1))
367 elif option == '--remove-params':
368 gopts.remove_all_params.append(value)
370 gopts.totext_mask.append(value)
372 gopts.binary_mask.append(value)
374 gopts.ignore_mask.append(value)
376 gopts.error_mask.append(value)
378 gopts.output_filename = value
385 if __name__ == "__main__":
386 arguments = get_opt()
390 gopts.input_filename = '-'
392 if gopts.output_filename:
393 outfile = open(gopts.output_filename, 'w')
395 gopts.output_filename = '-'
398 if (arguments[0] == '-'):
399 gopts.input_filename = '-'
402 gopts.input_filename = arguments[0]
403 infile = open(arguments[0], 'r')
405 if gopts.output_filename:
406 outfile = open(gopts.output_filename, 'w')
408 gopts.output_filename = '-'
411 if gopts.output_filename:
412 usage(1, 'Too many output filenames')
413 if (arguments[1] == '-'):
414 gopts.output_filename = '-'
417 gopts.output_filename = arguments[1]
418 outfile = open(arguments[1], 'w')
420 usage(1, 'Too many arguments')
422 if (infile is sys.stdin) and sys.stdin.isatty():
423 if (outfile is sys.stdout) and sys.stdout.isatty():
425 usage(1, 'Filtering from console is forbidden')
427 if not gopts.host_name:
429 gopts.host_name = socket.gethostname()
431 gopts.outfile = outfile
432 output = outfile.write
435 decode_message(email.message_from_file(infile))