2 """Decode MIME message"""
4 from mimedecode_version import __version__, __author__, __copyright__, __license__
10 from cStringIO import StringIO
12 from StringIO import StringIO
14 me = os.path.basename(sys.argv[0])
19 Broytman mimedecode.py version %s, %s
20 """ % (__version__, __copyright__))
23 def usage(code=0, errormsg=''):
26 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]]
29 sys.stderr.write(errormsg + '\n')
34 gopts.outfile.write(s)
36 def output_headers(msg):
37 unix_from = msg.get_unixfrom()
39 output(unix_from + '\n')
40 for key, value in msg.items():
41 output("%s: %s\n" % (key, value))
42 output("\n") # End of headers
45 def recode(s, charset):
46 return unicode(s, charset, "replace").encode(gopts.default_encoding, "replace")
48 def recode_if_needed(s, charset):
49 if charset and charset.lower() <> gopts.default_encoding:
50 s = recode(s, charset)
54 def _decode_header(s):
55 """Return a decoded string according to RFC 2047.
56 NOTE: This is almost the same as email.Utils.decode.
58 from types import ListType
61 L = email.Header.decode_header(s)
62 if not isinstance(L, ListType):
67 for atom, charset in L:
71 rtn.append(recode_if_needed(atom, charset))
73 del rtn[-1] # remove the last space
75 # Now that we've decoded everything, we just need to join all the parts
76 # together into the final string.
79 def decode_header(msg, header):
80 "Decode mail header (if exists) and put it back, if it was encoded"
82 if msg.has_key(header):
84 new_value = _decode_header(value)
85 if new_value <> value: # do not bother to touch msg if not changed
86 set_header(msg, header, new_value)
89 def _decode_header_param(s):
90 return recode_if_needed(s[2], s[0])
92 def decode_header_param(msg, header, param):
93 "Decode mail header's parameter (if exists) and put it back, if it was encoded"
95 if msg.has_key(header):
96 value = msg.get_param(param, header=header)
98 from types import TupleType
99 if isinstance(value, TupleType):
100 new_value = _decode_header_param(value)
102 new_value = _decode_header(value)
103 if new_value <> value: # do not bother to touch msg if not changed
104 msg.set_param(param, new_value, header)
107 def decode_headers(msg):
108 "Decode message headers according to global options"
110 for header in gopts.decode_headers:
111 decode_header(msg, header)
113 for header, param in gopts.decode_header_params:
114 decode_header_param(msg, header, param)
117 def set_header(msg, header, value):
120 if msg.has_key(header):
121 msg.replace_header(header, value)
126 def set_content_type(msg, newtype, charset=None):
127 msg.set_type(newtype)
130 msg.set_param("charset", charset, "Content-Type")
133 caps = None # Globally stored mailcap database; initialized only if needed
135 def decode_body(msg, s):
136 "Decode body to plain text using first copiousoutput filter from mailcap"
138 import mailcap, tempfile
142 caps = mailcap.getcaps()
144 content_type = msg.get_content_type()
145 filename = tempfile.mktemp()
148 entries = mailcap.lookup(caps, content_type, "view")
149 for entry in entries:
150 if entry.has_key('copiousoutput'):
151 if entry.has_key('test'):
152 test = mailcap.subst(entry['test'], content_type, filename)
153 if test and os.system(test) != 0:
155 command = mailcap.subst(entry["view"], content_type, filename)
161 file = open(filename, 'w')
165 pipe = os.popen(command, 'r')
170 set_content_type(msg, "text/plain")
171 msg["X-MIME-Autoconverted"] = "from %s to text/plain by %s id %s" % (content_type, gopts.host_name, command.split()[0])
176 def recode_charset(msg, s):
177 "Recode charset of the message to the default charset"
179 save_charset = charset = msg.get_content_charset()
180 if charset and charset.lower() <> gopts.default_encoding:
181 s = recode_if_needed(s, charset)
182 content_type = msg.get_content_type()
183 set_content_type(msg, content_type, gopts.default_encoding)
184 msg["X-MIME-Autoconverted"] = "from %s to %s by %s id %s" % (save_charset, gopts.default_encoding, gopts.host_name, me)
188 def totext(msg, instring):
189 "Convert instring content to text"
191 if msg.is_multipart(): # Recursively decode all parts of the multipart message
192 newfile = StringIO(msg.as_string())
197 # Decode body and recode charset
198 s = decode_body(msg, instring)
199 if gopts.recode_charset:
200 s = recode_charset(msg, s)
206 def decode_part(msg):
207 "Decode one part of the message"
210 encoding = msg["Content-Transfer-Encoding"]
212 if encoding in (None, '', '7bit', '8bit', 'binary'):
213 outstring = str(msg.get_payload())
214 else: # Decode from transfer ecoding to text or binary form
215 outstring = str(msg.get_payload(decode=1))
216 set_header(msg, "Content-Transfer-Encoding", "8bit")
217 msg["X-MIME-Autoconverted"] = "from %s to 8bit by %s id %s" % (encoding, gopts.host_name, me)
219 # Test all mask lists and find what to do with this content type
221 ctype = msg.get_content_type()
224 mtype = msg.get_content_maintype()
226 masks.append(mtype + '/*')
229 for content_type in masks:
230 if content_type in gopts.totext_mask:
231 totext(msg, outstring)
233 elif content_type in gopts.binary_mask:
237 elif content_type in gopts.ignore_mask:
239 output("\nMessage body of type `%s' skipped.\n" % content_type)
241 elif content_type in gopts.error_mask:
242 raise ValueError, "content type `%s' prohibited" % content_type
244 # Neither content type nor masks were listed - decode by default
245 totext(msg, outstring)
248 def decode_file(infile):
249 "Decode the entire message"
251 msg = email.message_from_file(infile)
252 boundary = msg.get_boundary()
254 if msg.is_multipart():
258 if msg.preamble: # Preserve the first part, it is probably not a RFC822-message
259 output(msg.preamble) # Usually it is just a few lines of text (MIME warning)
261 for subpart in msg.get_payload():
262 output("\n--%s\n" % boundary)
265 output("\n--%s--\n" % boundary)
271 if msg.has_key("Content-Type"): # Simple one-part message - decode it
274 else: # Not a message, just text - copy it literally
275 output(msg.as_string())
279 from m_lib.defenc import default_encoding
280 recode_charset = 1 # recode charset of message body
284 decode_headers = ["From", "Subject"] # A list of headers to decode
285 decode_header_params = [
286 ("Content-Type", "name"),
287 ("Content-Disposition", "filename"),
288 ] # A list of headers' parameters to decode
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: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.totext_mask.append(value)
334 gopts.binary_mask.append(value)
336 gopts.ignore_mask.append(value)
338 gopts.error_mask.append(value)
340 gopts.output_filename = value
347 if __name__ == "__main__":
348 arguments = get_opt()
352 gopts.input_filename = '-'
353 gopts.output_filename = '-'
357 if (arguments[0] == '-'):
358 gopts.input_filename = '-'
361 gopts.input_filename = arguments[0]
362 infile = open(arguments[0], 'r')
364 if gopts.output_filename:
365 outfile = open(gopts.output_filename, 'w')
367 gopts.output_filename = '-'
370 if gopts.output_filename:
371 usage(1, 'Too many output filenames')
372 if (arguments[1] == '-'):
373 gopts.output_filename = '-'
376 gopts.output_filename = arguments[1]
377 outfile = open(arguments[1], 'w')
379 usage(1, 'Too many arguments')
381 if (infile is sys.stdin) and sys.stdin.isatty():
382 if (outfile is sys.stdout) and sys.stdout.isatty():
384 usage(1, 'Filtering from console is forbidden')
386 if not gopts.host_name:
388 gopts.host_name = socket.gethostname()
390 gopts.outfile = outfile