X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=mimedecode.py;h=5f3321c22fa73529108230a79516436cfa78e1cc;hb=049f238d8b23e2ab7f543b33d52dd2544ffa7b22;hp=8d50b7d0d56453ab0d148cc0d66311c2d6b13081;hpb=bc6318c1a631d838dfe6bfe1c94638c59ea626bf;p=mimedecode.git diff --git a/mimedecode.py b/mimedecode.py index 8d50b7d..5f3321c 100755 --- a/mimedecode.py +++ b/mimedecode.py @@ -17,7 +17,7 @@ Broytman mimedecode.py version %s, %s def usage(code=0, errormsg=''): version(0) sys.stdout.write("""\ - Usage: %s [-h|--help] [-V|--version] [-cCDP] [-H|--host=hostname] [-f charset] [-d header1[,h2,...]|*[,-h1,...]] [-p header1[,h2,h3,...]:param1[,p2,p3,...]] [-r header1[,h2,...]|*[,-h1,...]] [-R header1[,h2,h3,...]:param1[,p2,p3,...]] [--set-header header:value] [--set-param header:param=value] [-Bbeit mask] [--save-headers|body|message mask] [-O dest_dir] [-o output_file] [input_file [output_file]] +Usage: %s [-h|--help] [-V|--version] [-cCDP] [-H|--host=hostname] [-f charset] [-d header1[,h2,...]|*[,-h1,...]] [-p header1[,h2,h3,...]:param1[,p2,p3,...]] [-r header1[,h2,...]|*[,-h1,...]] [-R header1[,h2,h3,...]:param1[,p2,p3,...]] [--set-header header:value] [--set-param header:param=value] [-Bbeit mask] [--save-headers|body|message mask] [-O dest_dir] [-o output_file] [input_file [output_file]] """ % me) if errormsg: sys.stderr.write(errormsg + os.linesep) @@ -27,63 +27,75 @@ def usage(code=0, errormsg=''): def output_headers(msg): unix_from = msg.get_unixfrom() if unix_from: - output(unix_from + os.linesep) + output(unix_from) + output(os.linesep) for key, value in msg.items(): - output("%s: %s%s" % (key, value, os.linesep)) + output(key) + output(": ") + value = value.split(';', 1) + output(value[0]) + if len(value) == 2: + output(";") + output(_decode_header(value[1], strip=False)) + output(os.linesep) output(os.linesep) # End of headers def recode_if_needed(s, charset): - if charset and charset.lower() <> g.default_encoding: - s = unicode(s, charset, "replace").encode(g.default_encoding, "replace") + if bytes is str: # Python2 + if isinstance(s, bytes) and \ + charset and charset.lower() != g.default_encoding: + s = s.decode(charset, "replace").\ + encode(g.default_encoding, "replace") + else: # Python3 + if isinstance(s, bytes): + s = s.decode(charset, "replace") return s -def _decode_header(s): +def _decode_header(s, strip=True): """Return a decoded string according to RFC 2047. NOTE: This is almost the same as email.Utils.decode. """ - import email.Header + import email.header - L = email.Header.decode_header(s) + L = email.header.decode_header(s) if not isinstance(L, list): # s wasn't decoded return s rtn = [] for atom, charset in L: - if charset is None: - rtn.append(atom) - else: - rtn.append(recode_if_needed(atom, charset)) - rtn.append(' ') - del rtn[-1] # remove the last space + atom = recode_if_needed(atom, charset or g.default_encoding) + if strip: + atom = atom.strip() + rtn.append(atom) # Now that we've decoded everything, we just need to join all the parts # together into the final string. - return ''.join(rtn) + return ' '.join(rtn) def decode_header(msg, header): "Decode mail header (if exists) and put it back, if it was encoded" - if msg.has_key(header): + if header in msg: value = msg[header] new_value = _decode_header(value) - if new_value <> value: # do not bother to touch msg if not changed + if new_value != value: # do not bother to touch msg if not changed set_header(msg, header, new_value) def decode_header_param(msg, header, param): "Decode mail header's parameter (if exists) and put it back, if it was encoded" - if msg.has_key(header): + if header in msg: value = msg.get_param(param, header=header) if value: if isinstance(value, tuple): new_value = recode_if_needed(value[2], value[0]) else: new_value = _decode_header(value) - if new_value <> value: # do not bother to touch msg if not changed + if new_value != value: # do not bother to touch msg if not changed msg.set_param(param, new_value, header) @@ -181,7 +193,7 @@ def decode_headers(msg): def set_header(msg, header, value): "Replace header" - if msg.has_key(header): + if header in msg: msg.replace_header(header, value) else: msg[header] = value @@ -206,13 +218,17 @@ def decode_body(msg, s): caps = mailcap.getcaps() content_type = msg.get_content_type() + if content_type.startswith('text/'): + charset = msg.get_content_charset() + else: + charset = None filename = tempfile.mktemp() command = None entries = mailcap.lookup(caps, content_type, "view") for entry in entries: - if entry.has_key('copiousoutput'): - if entry.has_key('test'): + if 'copiousoutput' in entry: + if 'test' in entry: test = mailcap.subst(entry['test'], content_type, filename) if test and os.system(test) != 0: continue @@ -223,16 +239,24 @@ def decode_body(msg, s): return s outfile = open(filename, 'wb') + if charset and isinstance(s, bytes): + s = s.decode(charset, "replace") + if not isinstance(s, bytes): + s = s.encode(g.default_encoding, "replace") outfile.write(s) outfile.close() pipe = os.popen(command, 'r') - s = pipe.read() - pipe.close() + new_s = pipe.read() + if pipe.close() is None: # result=0, Ok + s = new_s os.remove(filename) set_content_type(msg, "text/plain") - msg["X-MIME-Autoconverted"] = "from %s to text/plain by %s id %s" % (content_type, g.host_name, command.split()[0]) + if s is new_s: + msg["X-MIME-Autoconverted"] = "from %s to text/plain by %s id %s" % (content_type, g.host_name, command.split()[0]) + else: + msg["X-MIME-Autoconverted"] = "failed conversion from %s to text/plain by %s id %s" % (content_type, g.host_name, command.split()[0]) return s @@ -241,7 +265,7 @@ def recode_charset(msg, s): "Recode charset of the message to the default charset" save_charset = charset = msg.get_content_charset() - if charset and charset.lower() <> g.default_encoding: + if charset and charset.lower() != g.default_encoding: s = recode_if_needed(s, charset) content_type = msg.get_content_type() set_content_type(msg, content_type, g.default_encoding) @@ -302,7 +326,11 @@ def _save_message(msg, outstring, save_headers=False, save_body=False): global output save_output = output outfile = open_output_file(fname) - output = outfile.write + def _output_bytes(s): + if not isinstance(s, bytes): + s = s.encode(g.default_encoding, "replace") + outfile.write(s) + output = _output_bytes if save_headers: output_headers(msg) if save_body: @@ -373,7 +401,7 @@ def decode_part(msg): for content_type in masks: if content_type in g.error_mask: - raise ValueError, "content type %s prohibited" % ctype + raise ValueError("content type %s prohibited" % ctype) def decode_multipart(msg): "Decode multipart" @@ -398,8 +426,36 @@ def decode_multipart(msg): if boundary: output("%s--%s--%s" % (os.linesep, boundary, os.linesep)) return - elif content_type in g.error_mask: - raise ValueError, "content type %s prohibited" % ctype + + for content_type in masks: + if content_type in g.save_body_mask or \ + content_type in g.save_message_mask: + _out_l = [] + first_subpart = True + for subpart in msg.get_payload(): + if first_subpart: + first_subpart = False + else: + _out_l.append(os.linesep) + _out_l.append("--%s%s" % (boundary, os.linesep)) + _out_l.append(subpart.as_string()) + _out_l.append("%s--%s--%s" % (os.linesep, boundary, os.linesep)) + outstring = ''.join(_out_l) + break + else: + outstring = None + + for content_type in masks: + if content_type in g.save_headers_mask: + _save_message(msg, outstring, save_headers=True, save_body=False) + if content_type in g.save_body_mask: + _save_message(msg, outstring, save_headers=False, save_body=True) + if content_type in g.save_message_mask: + _save_message(msg, outstring, save_headers=True, save_body=True) + + for content_type in masks: + if content_type in g.error_mask: + raise ValueError("content type %s prohibited" % ctype) output_headers(msg) @@ -619,7 +675,14 @@ if __name__ == "__main__": g.host_name = socket.gethostname() g.outfile = outfile - output = outfile.write + if hasattr(outfile, 'buffer'): + def output_bytes(s): + if not isinstance(s, bytes): + s = s.encode(g.default_encoding, "replace") + outfile.buffer.write(s) + output = output_bytes + else: + output = outfile.write import email msg = email.message_from_file(infile)