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")
49 def recode_if_needed(s, charset):
50 if charset and charset.lower() <> gopts.default_encoding:
51 s = recode(s, charset)
55 def _decode_header(s):
56 """Return a decoded string according to RFC 2047.
57 NOTE: This is almost the same as email.Utils.decode.
59 from types import ListType
62 L = email.Header.decode_header(s)
63 if not isinstance(L, ListType):
68 for atom, charset in L:
72 rtn.append(recode_if_needed(atom, charset))
74 del rtn[-1] # remove the last space
76 # Now that we've decoded everything, we just need to join all the parts
77 # together into the final string.
81 def decode_header(msg, header):
82 "Decode mail header (if exists) and put it back, if it was encoded"
84 if msg.has_key(header):
86 new_value = _decode_header(value)
87 if new_value <> value: # do not bother to touch msg if not changed
88 set_header(msg, header, new_value)
91 def _decode_header_param(s):
92 return recode_if_needed(s[2], s[0])
95 def decode_header_param(msg, header, param):
96 "Decode mail header's parameter (if exists) and put it back, if it was encoded"
98 if msg.has_key(header):
99 value = msg.get_param(param, header=header)
101 from types import TupleType
102 if isinstance(value, TupleType):
103 new_value = _decode_header_param(value)
105 new_value = _decode_header(value)
106 if new_value <> value: # do not bother to touch msg if not changed
107 msg.set_param(param, new_value, header)
110 def decode_headers(msg):
111 "Decode message headers according to global options"
113 for header in gopts.decode_headers:
114 decode_header(msg, header)
116 for header, param in gopts.decode_header_params:
117 decode_header_param(msg, header, param)
120 def set_header(msg, header, value):
123 if msg.has_key(header):
124 msg.replace_header(header, value)
129 def set_content_type(msg, newtype, charset=None):
130 msg.set_type(newtype)
133 msg.set_param("charset", charset, "Content-Type")
136 caps = None # Globally stored mailcap database; initialized only if needed
138 def decode_body(msg, s):
139 "Decode body to plain text using first copiousoutput filter from mailcap"
141 import mailcap, tempfile
145 caps = mailcap.getcaps()
147 content_type = msg.get_content_type()
148 filename = tempfile.mktemp()
151 entries = mailcap.lookup(caps, content_type, "view")
152 for entry in entries:
153 if entry.has_key('copiousoutput'):
154 if entry.has_key('test'):
155 test = mailcap.subst(entry['test'], content_type, filename)
156 if test and os.system(test) != 0:
158 command = mailcap.subst(entry["view"], content_type, filename)
164 file = open(filename, 'w')
168 pipe = os.popen(command, 'r')
173 set_content_type(msg, "text/plain")
174 msg["X-MIME-Autoconverted"] = "from %s to text/plain by %s id %s" % (content_type, gopts.host_name, command.split()[0])
179 def recode_charset(msg, s):
180 "Recode charset of the message to the default charset"
182 save_charset = charset = msg.get_content_charset()
183 if charset and charset.lower() <> gopts.default_encoding:
184 s = recode_if_needed(s, charset)
185 content_type = msg.get_content_type()
186 set_content_type(msg, content_type, gopts.default_encoding)
187 msg["X-MIME-Autoconverted"] = "from %s to %s by %s id %s" % (save_charset, gopts.default_encoding, gopts.host_name, me)
191 def totext(msg, instring):
192 "Convert instring content to text"
194 if msg.is_multipart(): # Recursively decode all parts of the multipart message
195 newfile = StringIO(msg.as_string())
200 # Decode body and recode charset
201 s = decode_body(msg, instring)
202 if gopts.recode_charset:
203 s = recode_charset(msg, s)
209 def decode_part(msg):
210 "Decode one part of the message"
213 encoding = msg["Content-Transfer-Encoding"]
215 if encoding in (None, '', '7bit', '8bit', 'binary'):
216 outstring = str(msg.get_payload())
217 else: # Decode from transfer ecoding to text or binary form
218 outstring = str(msg.get_payload(decode=1))
219 set_header(msg, "Content-Transfer-Encoding", "8bit")
220 msg["X-MIME-Autoconverted"] = "from %s to 8bit by %s id %s" % (encoding, gopts.host_name, me)
222 # Test all mask lists and find what to do with this content type
224 ctype = msg.get_content_type()
227 mtype = msg.get_content_maintype()
229 masks.append(mtype + '/*')
232 for content_type in masks:
233 if content_type in gopts.totext_mask:
234 totext(msg, outstring)
236 elif content_type in gopts.binary_mask:
240 elif content_type in gopts.ignore_mask:
242 output("\nMessage body of type `%s' skipped.\n" % content_type)
244 elif content_type in gopts.error_mask:
245 raise ValueError, "content type `%s' prohibited" % content_type
247 # Neither content type nor masks were listed - decode by default
248 totext(msg, outstring)
251 def decode_file(infile):
252 "Decode the entire message"
254 msg = email.message_from_file(infile)
255 boundary = msg.get_boundary()
257 if msg.is_multipart():
261 if msg.preamble: # Preserve the first part, it is probably not a RFC822-message
262 output(msg.preamble) # Usually it is just a few lines of text (MIME warning)
264 for subpart in msg.get_payload():
265 output("\n--%s\n" % boundary)
268 output("\n--%s--\n" % boundary)
274 if msg.has_key("Content-Type"): # Simple one-part message - decode it
277 else: # Not a message, just text - copy it literally
278 output(msg.as_string())
282 from m_lib.defenc import default_encoding
283 recode_charset = 1 # recode charset of message body
287 decode_headers = ["From", "Subject"] # A list of headers to decode
288 decode_header_params = [
289 ("Content-Type", "name"),
290 ("Content-Disposition", "filename"),
291 ] # A list of headers' parameters to decode
293 totext_mask = [] # A list of content-types to decode
294 binary_mask = [] # A list to pass through
295 ignore_mask = [] # Ignore (skip, do not decode and do not include into output)
296 error_mask = [] # Raise error if encounter one of these
298 input_filename = None
299 output_filename = None
301 gopts = GlobalOptions
305 from getopt import getopt, GetoptError
308 options, arguments = getopt(sys.argv[1:], 'hVcCDPH:f:d:p:b:e:i:t:o:',
309 ['help', 'version', 'host'])
313 for option, value in options:
314 if option in ('-h', '--help'):
316 elif option in ('-V', '--version'):
319 gopts.recode_charset = 1
321 gopts.recode_charset = 0
322 elif option in ('-H', '--host'):
323 gopts.host_name = value
325 gopts.default_encoding = value
327 gopts.decode_headers.append(value)
329 gopts.decode_headers = []
331 gopts.decode_header_params.append(value.split(':', 1))
333 gopts.decode_header_params = []
335 gopts.totext_mask.append(value)
337 gopts.binary_mask.append(value)
339 gopts.ignore_mask.append(value)
341 gopts.error_mask.append(value)
343 gopts.output_filename = value
350 if __name__ == "__main__":
351 arguments = get_opt()
355 gopts.input_filename = '-'
356 gopts.output_filename = '-'
360 if (arguments[0] == '-'):
361 gopts.input_filename = '-'
364 gopts.input_filename = arguments[0]
365 infile = open(arguments[0], 'r')
367 if gopts.output_filename:
368 outfile = open(gopts.output_filename, 'w')
370 gopts.output_filename = '-'
373 if gopts.output_filename:
374 usage(1, 'Too many output filenames')
375 if (arguments[1] == '-'):
376 gopts.output_filename = '-'
379 gopts.output_filename = arguments[1]
380 outfile = open(arguments[1], 'w')
382 usage(1, 'Too many arguments')
384 if (infile is sys.stdin) and (outfile is sys.stdout) and \
385 sys.stdin.isatty() and sys.stdout.isatty():
386 usage(1, 'Filtering from console to console is forbidden')
388 if not gopts.host_name:
390 gopts.host_name = socket.gethostname()
392 gopts.outfile = outfile