X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=mimedecode.py;h=36cc340c5556f2c07a088d87a78f4f4388280b3f;hb=563d8bfe689ee8f0efab714d57550bb6ccaf323b;hp=ce4b288627ca1b565b0134614f9b97cfcf6349de;hpb=e6aea39b9faf08f3c8edc95d4367f8f828346852;p=mimedecode.git diff --git a/mimedecode.py b/mimedecode.py index ce4b288..36cc340 100755 --- a/mimedecode.py +++ b/mimedecode.py @@ -2,8 +2,13 @@ """Decode MIME message""" import sys, os +import subprocess from mimedecode_version import __version__, \ __author__, __copyright__, __license__ +if sys.version_info[0] >= 3: + # Replace email.message._formatparam with _formatparam from Python 2.7 + # to avoid re-encoding non-ascii params. + import formatparam_27 me = os.path.basename(sys.argv[0]) @@ -17,7 +22,7 @@ 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) @@ -27,19 +32,33 @@ 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. """ @@ -52,16 +71,14 @@ def _decode_header(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" @@ -88,7 +105,7 @@ def decode_header_param(msg, header, param): def _get_exceptions(list): - return [x[1:].lower() for x in list[1:] if x[0]=='-'] + return [x[1:].lower() for x in list[1:] if x[0] == '-'] def _decode_headers_params(msg, header, decode_all_params, param_list): if decode_all_params: @@ -206,6 +223,10 @@ 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 @@ -223,17 +244,28 @@ def decode_body(msg, s): return s outfile = open(filename, 'wb') + if charset and bytes is not str and isinstance(s, bytes): # Python3 + 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() + pipe = subprocess.Popen(command, shell=True, stdout=subprocess.PIPE) + new_s = pipe.stdout.read() + pipe.stdout.close() + if pipe.wait() == 0: # result=0, Ok + s = new_s + if bytes is not str and isinstance(s, bytes): # Python3 + s = s.decode(g.default_encoding, "replace") + if charset and not isinstance(s, bytes): + s = s.encode(charset, "replace") + 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]) + else: + msg["X-MIME-Autoconverted"] = "failed conversion from %s to text/plain by %s id %s" % (content_type, g.host_name, command.split()[0]) 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]) - return s @@ -302,7 +334,11 @@ 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: @@ -349,7 +385,7 @@ def decode_part(msg): outstring = totext(msg, outstring) break elif content_type in g.binary_mask or \ - content_type in g.decoded_binary_mask: + content_type in g.decoded_binary_mask: output_headers(msg) output(outstring) break @@ -528,7 +564,8 @@ def get_opts(): from getopt import getopt, GetoptError try: - options, arguments = getopt(sys.argv[1:], + options, arguments = getopt( + sys.argv[1:], 'hVcCDPH:f:d:p:r:R:b:B:e:I:i:t:O:o:', ['help', 'version', 'host=', 'save-headers=', 'save-body=', 'save-message=', @@ -647,7 +684,14 @@ 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)