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 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:
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, param in gopts.decode_header_params:
128 decode_header_param(msg, header, param)
131 def set_header(msg, header, value):
134 if msg.has_key(header):
135 msg.replace_header(header, value)
140 def set_content_type(msg, newtype, charset=None):
141 msg.set_type(newtype)
144 msg.set_param("charset", charset, "Content-Type")
147 caps = None # Globally stored mailcap database; initialized only if needed
149 def decode_body(msg, s):
150 "Decode body to plain text using first copiousoutput filter from mailcap"
152 import mailcap, tempfile
156 caps = mailcap.getcaps()
158 content_type = msg.get_content_type()
159 filename = tempfile.mktemp()
162 entries = mailcap.lookup(caps, content_type, "view")
163 for entry in entries:
164 if entry.has_key('copiousoutput'):
165 if entry.has_key('test'):
166 test = mailcap.subst(entry['test'], content_type, filename)
167 if test and os.system(test) != 0:
169 command = mailcap.subst(entry["view"], content_type, filename)
175 file = open(filename, 'w')
179 pipe = os.popen(command, 'r')
184 set_content_type(msg, "text/plain")
185 msg["X-MIME-Autoconverted"] = "from %s to text/plain by %s id %s" % (content_type, gopts.host_name, command.split()[0])
190 def recode_charset(msg, s):
191 "Recode charset of the message to the default charset"
193 save_charset = charset = msg.get_content_charset()
194 if charset and charset.lower() <> gopts.default_encoding:
195 s = recode_if_needed(s, charset)
196 content_type = msg.get_content_type()
197 set_content_type(msg, content_type, gopts.default_encoding)
198 msg["X-MIME-Autoconverted"] = "from %s to %s by %s id %s" % (save_charset, gopts.default_encoding, gopts.host_name, me)
202 def totext(msg, instring):
203 "Convert instring content to text"
205 # Decode body and recode charset
206 s = decode_body(msg, instring)
207 if gopts.recode_charset:
208 s = recode_charset(msg, s)
214 def decode_part(msg):
215 "Decode one part of the message"
218 encoding = msg["Content-Transfer-Encoding"]
220 if encoding in (None, '', '7bit', '8bit', 'binary'):
221 outstring = str(msg.get_payload())
222 else: # Decode from transfer ecoding to text or binary form
223 outstring = str(msg.get_payload(decode=1))
224 set_header(msg, "Content-Transfer-Encoding", "8bit")
225 msg["X-MIME-Autoconverted"] = "from %s to 8bit by %s id %s" % (encoding, gopts.host_name, me)
227 # Test all mask lists and find what to do with this content type
229 ctype = msg.get_content_type()
232 mtype = msg.get_content_maintype()
234 masks.append(mtype + '/*')
237 for content_type in masks:
238 if content_type in gopts.totext_mask:
239 totext(msg, outstring)
241 elif content_type in gopts.binary_mask:
245 elif content_type in gopts.ignore_mask:
247 output("\nMessage body of type `%s' skipped.\n" % content_type)
249 elif content_type in gopts.error_mask:
250 raise ValueError, "content type `%s' prohibited" % content_type
252 # Neither content type nor masks were listed - decode by default
253 totext(msg, outstring)
256 def decode_multipart(msg):
262 if msg.preamble: # Preserve the first part, it is probably not a RFC822-message
263 output(msg.preamble) # Usually it is just a few lines of text (MIME warning)
265 boundary = msg.get_boundary()
267 for subpart in msg.get_payload():
269 output("\n--%s\n" % boundary)
271 # Recursively decode all parts of the subpart
272 decode_message(subpart)
275 output("\n--%s--\n" % boundary)
281 def decode_message(msg):
284 if msg.is_multipart():
285 decode_multipart(msg)
286 elif len(msg): # Simple one-part message (there are headers) - decode it
288 else: # Not a message, just text - copy it literally
289 output(msg.as_string())
293 from m_lib.defenc import default_encoding
294 recode_charset = 1 # recode charset of message body
298 # A list of headers to decode
299 decode_headers = ["From", "To", "Cc", "Reply-To", "Mail-Followup-To",
302 # A list of headers parameters to decode
303 decode_header_params = [
304 ("Content-Type", "name"),
305 ("Content-Disposition", "filename"),
308 # A list of headers to remove
310 # A list of headers parameters to remove
311 remove_header_params = []
312 # A list of headers to be stripped of all parameters
313 remove_all_params = []
315 totext_mask = [] # A list of content-types to decode
316 binary_mask = [] # A list to pass through
317 ignore_mask = [] # Ignore (skip, do not decode and do not include into output)
318 error_mask = [] # Raise error if encounter one of these
320 input_filename = None
321 output_filename = None
323 gopts = GlobalOptions
327 from getopt import getopt, GetoptError
330 options, arguments = getopt(sys.argv[1:],
331 'hVcCDPH:f:d:p:r:R:b:e:i:t:o:',
332 ['help', 'version', 'host=', 'remove-params='])
336 for option, value in options:
337 if option in ('-h', '--help'):
339 elif option in ('-V', '--version'):
342 gopts.recode_charset = 1
344 gopts.recode_charset = 0
345 elif option in ('-H', '--host'):
346 gopts.host_name = value
348 gopts.default_encoding = value
350 if value.startswith('*'):
351 gopts.decode_headers = []
352 gopts.decode_headers.append(value)
354 gopts.decode_headers = []
356 gopts.decode_header_params.append(value.split(':', 1))
358 gopts.decode_header_params = []
360 gopts.remove_headers.append(value)
362 gopts.remove_header_params.append(value.split(':', 1))
363 elif option == '--remove-params':
364 gopts.remove_all_params.append(value)
366 gopts.totext_mask.append(value)
368 gopts.binary_mask.append(value)
370 gopts.ignore_mask.append(value)
372 gopts.error_mask.append(value)
374 gopts.output_filename = value
381 if __name__ == "__main__":
382 arguments = get_opt()
386 gopts.input_filename = '-'
388 if gopts.output_filename:
389 outfile = open(gopts.output_filename, 'w')
391 gopts.output_filename = '-'
394 if (arguments[0] == '-'):
395 gopts.input_filename = '-'
398 gopts.input_filename = arguments[0]
399 infile = open(arguments[0], 'r')
401 if gopts.output_filename:
402 outfile = open(gopts.output_filename, 'w')
404 gopts.output_filename = '-'
407 if gopts.output_filename:
408 usage(1, 'Too many output filenames')
409 if (arguments[1] == '-'):
410 gopts.output_filename = '-'
413 gopts.output_filename = arguments[1]
414 outfile = open(arguments[1], 'w')
416 usage(1, 'Too many arguments')
418 if (infile is sys.stdin) and sys.stdin.isatty():
419 if (outfile is sys.stdout) and sys.stdout.isatty():
421 usage(1, 'Filtering from console is forbidden')
423 if not gopts.host_name:
425 gopts.host_name = socket.gethostname()
427 gopts.outfile = outfile
428 output = outfile.write
431 decode_message(email.message_from_file(infile))