X-Git-Url: https://git.phdru.name/?p=mimedecode.git;a=blobdiff_plain;f=mimedecode.py;h=2fccbcc7b13f2cece51828769653bc2320257518;hp=01f650563060eb085b6811295a8ee8ce267c2744;hb=972bae8625e655e34170c588561eed078ceb37d9;hpb=d0300ffa1ef1f9afee0e4d53c725f54773d89c13 diff --git a/mimedecode.py b/mimedecode.py index 01f6505..2fccbcc 100755 --- a/mimedecode.py +++ b/mimedecode.py @@ -1,10 +1,14 @@ #! /usr/bin/env python """Decode MIME message""" -from mimedecode_version import __version__, __author__, __copyright__, __license__ - import sys, os -import email +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]) @@ -18,73 +22,85 @@ 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 + '\n') + sys.stderr.write(errormsg + os.linesep) sys.exit(code) def output_headers(msg): unix_from = msg.get_unixfrom() if unix_from: - output(unix_from + '\n') + output(unix_from) + output(os.linesep) for key, value in msg.items(): - output("%s: %s\n" % (key, value)) - output("\n") # End of headers + 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) @@ -182,7 +198,7 @@ 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 @@ -207,13 +223,17 @@ 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 @@ -223,18 +243,29 @@ def decode_body(msg, s): if not command: return s - file = open(filename, 'w') - file.write(s) - file.close() + 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 @@ -242,7 +273,7 @@ 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) @@ -263,6 +294,18 @@ def totext(msg, instring): return s +mimetypes = None + +def _guess_extension(ctype): + global mimetypes + if mimetypes is None: + import mimetypes + mimetypes.init() + user_mime_type = os.path.expanduser('~/.mime.types') + if os.path.exists(user_mime_type): + mimetypes._db.read(user_mime_type) + return mimetypes.guess_extension(ctype) + def _save_message(msg, outstring, save_headers=False, save_body=False): for header, param in ( ("Content-Disposition", "filename"), @@ -270,17 +313,32 @@ def _save_message(msg, outstring, save_headers=False, save_body=False): ): fname = msg.get_param(param, header=header) if fname: + if isinstance(fname, tuple): + fname = fname[2] # Do not recode if it isn't recoded yet + try: + for forbidden in chr(0), '/', '\\': + if forbidden in fname: + raise ValueError + except ValueError: + continue fname = '-' + fname break else: fname = '' g.save_counter += 1 fname = str(g.save_counter) + fname + if '.' not in fname: + ext = _guess_extension(msg.get_content_type()) + if ext: fname += ext global output save_output = output - outfile = open(os.path.join(g.destination_dir, fname), 'w') - output = outfile.write + outfile = open_output_file(fname) + 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: @@ -299,8 +357,7 @@ def decode_part(msg): ctype = msg.get_content_type() if ctype: masks.append(ctype) - mtype = msg.get_content_maintype() - if mtype: + mtype = ctype.split('/')[0] masks.append(mtype + '/*') masks.append('*/*') @@ -312,6 +369,8 @@ def decode_part(msg): elif content_type in g.binary_mask: left_binary = True break + elif content_type in g.fully_ignore_mask: + return encoding = msg["Content-Transfer-Encoding"] if left_binary or encoding in (None, '', '7bit', '8bit', 'binary'): @@ -326,13 +385,13 @@ 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 elif content_type in g.ignore_mask: output_headers(msg) - output("\nMessage body of type %s skipped.\n" % ctype) + output("%sMessage body of type %s skipped.%s" % (os.linesep, ctype, os.linesep)) break elif content_type in g.error_mask: break @@ -343,42 +402,90 @@ def decode_part(msg): for content_type in masks: if content_type in g.save_headers_mask: _save_message(msg, outstring, save_headers=True, save_body=False) - elif content_type in g.save_body_mask: + if content_type in g.save_body_mask: _save_message(msg, outstring, save_headers=False, save_body=True) - elif content_type in g.save_message_mask: + 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 + raise ValueError("content type %s prohibited" % ctype) def decode_multipart(msg): "Decode multipart" decode_headers(msg) + boundary = msg.get_boundary() + + masks = [] + ctype = msg.get_content_type() + if ctype: + masks.append(ctype) + mtype = ctype.split('/')[0] + masks.append(mtype + '/*') + masks.append('*/*') + + for content_type in masks: + if content_type in g.fully_ignore_mask: + return + elif content_type in g.ignore_mask: + output_headers(msg) + output("%sMessage body of type %s skipped.%s" % (os.linesep, ctype, os.linesep)) + if boundary: + output("%s--%s--%s" % (os.linesep, boundary, os.linesep)) + return + + 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) if msg.preamble: # Preserve the first part, it is probably not a RFC822-message output(msg.preamble) # Usually it is just a few lines of text (MIME warning) if msg.preamble is not None: - output("\n") + output(os.linesep) first_subpart = True - boundary = msg.get_boundary() - for subpart in msg.get_payload(): if boundary: if first_subpart: first_subpart = False else: - output("\n") - output("--%s\n" % boundary) + output(os.linesep) + output("--%s%s" % (boundary, os.linesep)) # Recursively decode all parts of the subpart decode_message(subpart) if boundary: - output("\n--%s--\n" % boundary) + output("%s--%s--%s" % (os.linesep, boundary, os.linesep)) if msg.epilogue: output(msg.epilogue) @@ -395,6 +502,19 @@ def decode_message(msg): output(msg.as_string()) +def open_output_file(filename): + fullpath = os.path.abspath(os.path.join(g.destination_dir, filename)) + full_dir = os.path.dirname(fullpath) + create = not os.path.isdir(full_dir) + if create: + os.makedirs(full_dir) + try: + return open(fullpath, 'wb') + except: + if create: + os.removedirs(full_dir) + + class GlobalOptions: from m_lib.defenc import default_encoding recode_charset = 1 # recode charset of message body @@ -424,7 +544,8 @@ class GlobalOptions: totext_mask = [] # A list of content-types to decode binary_mask = [] # A list of content-types to pass through decoded_binary_mask = [] # A list of content-types to pass through (content-transfer-decoded) - ignore_mask = [] # Ignore (skip, do not decode and do not include into output) + ignore_mask = [] # Ignore (do not decode and do not include into output) but output a warning instead of the body + fully_ignore_mask = [] # Completely ignore - no headers, no body, no warning error_mask = [] # Raise error if encounter one of these save_counter = 0 @@ -443,8 +564,9 @@ def get_opts(): from getopt import getopt, GetoptError try: - options, arguments = getopt(sys.argv[1:], - 'hVcCDPH:f:d:p:r:R:b:B:e:i:t:O:o:', + 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=', 'set-header=', 'set-param=']) @@ -493,6 +615,8 @@ def get_opts(): g.binary_mask.append(value) elif option == '-b': g.decoded_binary_mask.append(value) + elif option == '-I': + g.fully_ignore_mask.append(value) elif option == '-i': g.ignore_mask.append(value) elif option == '-e': @@ -521,7 +645,7 @@ if __name__ == "__main__": g.input_filename = '-' infile = sys.stdin if g.output_filename: - outfile = open(os.path.join(g.destination_dir, g.output_filename), 'w') + outfile = open_output_file(g.output_filename) else: g.output_filename = '-' outfile = sys.stdout @@ -534,7 +658,7 @@ if __name__ == "__main__": infile = open(arguments[0], 'r') if la == 1: if g.output_filename: - outfile = open(os.path.join(g.destination_dir, g.output_filename), 'w') + outfile = open_output_file(g.output_filename) else: g.output_filename = '-' outfile = sys.stdout @@ -546,7 +670,7 @@ if __name__ == "__main__": outfile = sys.stdout else: g.output_filename = arguments[1] - outfile = open(os.path.join(g.destination_dir, g.output_filename), 'w') + outfile = open_output_file(g.output_filename) else: usage(1, 'Too many arguments') @@ -560,8 +684,16 @@ 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) for header, value in g.set_header_value: