X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=mimedecode.py;h=068394761412d2474c7b17301494a960397ee35e;hb=b16160b60897a748415e80bbfeb87af495181b93;hp=64e2184e5b978bc3f70455fe360c98658a8d321a;hpb=11c9d974852d4e643ee63bd4b26f8af6dcea7a41;p=mimedecode.git diff --git a/mimedecode.py b/mimedecode.py index 64e2184..0683947 100755 --- a/mimedecode.py +++ b/mimedecode.py @@ -1,10 +1,18 @@ #! /usr/bin/env python """Decode MIME message""" -import sys, os +import os +import subprocess +import sys + 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]) @@ -12,13 +20,15 @@ def version(exit=1): sys.stdout.write("""\ Broytman mimedecode.py version %s, %s """ % (__version__, __copyright__)) - if exit: sys.exit(0) + if exit: + sys.exit(0) + 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]] -""" % me) +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) # noqa: E501 if errormsg: sys.stderr.write(errormsg + os.linesep) sys.exit(code) @@ -27,19 +37,33 @@ 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(os.linesep) # 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 = s.decode(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 +76,15 @@ 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" @@ -69,13 +92,16 @@ def decode_header(msg, 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" + """Decode mail header's parameter + Decode mail header's parameter (if exists) + and put it back if it was encoded. + """ if header in msg: value = msg.get_param(param, header=header) if value: @@ -83,12 +109,13 @@ def decode_header_param(msg, header, param): 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) 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: @@ -101,6 +128,7 @@ def _decode_headers_params(msg, header, decode_all_params, param_list): for param in param_list: decode_header_param(msg, header, param) + def _remove_headers_params(msg, header, remove_all_params, param_list): if remove_all_params: params = msg.get_params(header=header) @@ -111,71 +139,78 @@ def _remove_headers_params(msg, header, remove_all_params, param_list): msg.del_param(param, header) else: value = msg[header] - if value is None: # No such header + if value is None: # No such header return - if ';' not in value: # There are no parameters + if ';' not in value: # There are no parameters return - del msg[header] # Delete all such headers + del msg[header] # Delete all such headers # Get the value without parameters and set it back msg[header] = value.split(';')[0].strip() else: for param in param_list: msg.del_param(param, header) + def decode_headers(msg): "Decode message headers according to global options" for header_list in g.remove_headers: header_list = header_list.split(',') - if header_list[0] == '*': # Remove all headers except listed + if header_list[0] == '*': # Remove all headers except listed header_list = _get_exceptions(header_list) for header in msg.keys(): if header.lower() not in header_list: del msg[header] - else: # Remove listed headers + else: # Remove listed headers for header in header_list: del msg[header] for header_list, param_list in g.remove_headers_params: header_list = header_list.split(',') param_list = param_list.split(',') - remove_all_params = param_list[0] == '*' # Remove all params except listed + # Remove all params except listed. + remove_all_params = param_list[0] == '*' if remove_all_params: param_list = _get_exceptions(param_list) - if header_list[0] == '*': # Remove for all headers except listed + if header_list[0] == '*': # Remove for all headers except listed header_list = _get_exceptions(header_list) for header in msg.keys(): if header.lower() not in header_list: - _remove_headers_params(msg, header, remove_all_params, param_list) - else: # Decode for listed headers + _remove_headers_params( + msg, header, remove_all_params, param_list) + else: # Decode for listed headers for header in header_list: - _remove_headers_params(msg, header, remove_all_params, param_list) + _remove_headers_params( + msg, header, remove_all_params, param_list) for header_list in g.decode_headers: header_list = header_list.split(',') - if header_list[0] == '*': # Decode all headers except listed + if header_list[0] == '*': # Decode all headers except listed header_list = _get_exceptions(header_list) for header in msg.keys(): if header.lower() not in header_list: decode_header(msg, header) - else: # Decode listed headers + else: # Decode listed headers for header in header_list: decode_header(msg, header) for header_list, param_list in g.decode_header_params: header_list = header_list.split(',') param_list = param_list.split(',') - decode_all_params = param_list[0] == '*' # Decode all params except listed + # Decode all params except listed. + decode_all_params = param_list[0] == '*' if decode_all_params: param_list = _get_exceptions(param_list) - if header_list[0] == '*': # Decode for all headers except listed + if header_list[0] == '*': # Decode for all headers except listed header_list = _get_exceptions(header_list) for header in msg.keys(): if header.lower() not in header_list: - _decode_headers_params(msg, header, decode_all_params, param_list) - else: # Decode for listed headers + _decode_headers_params( + msg, header, decode_all_params, param_list) + else: # Decode for listed headers for header in header_list: - _decode_headers_params(msg, header, decode_all_params, param_list) + _decode_headers_params( + msg, header, decode_all_params, param_list) def set_header(msg, header, value): @@ -194,18 +229,24 @@ def set_content_type(msg, newtype, charset=None): msg.set_param("charset", charset, "Content-Type") -caps = None # Globally stored mailcap database; initialized only if needed +caps = None # Globally stored mailcap database; initialized only if needed + def decode_body(msg, s): "Decode body to plain text using first copiousoutput filter from mailcap" - import mailcap, tempfile + import mailcap + import tempfile global caps if caps is None: 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 +264,32 @@ 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 @@ -245,7 +301,9 @@ def recode_charset(msg, s): s = recode_if_needed(s, charset) content_type = msg.get_content_type() set_content_type(msg, content_type, g.default_encoding) - msg["X-MIME-Autoconverted"] = "from %s to %s by %s id %s" % (save_charset, g.default_encoding, g.host_name, me) + msg["X-MIME-Autoconverted"] = \ + "from %s to %s by %s id %s" \ + % (save_charset, g.default_encoding, g.host_name, me) return s @@ -264,6 +322,7 @@ def totext(msg, instring): mimetypes = None + def _guess_extension(ctype): global mimetypes if mimetypes is None: @@ -274,6 +333,7 @@ def _guess_extension(ctype): 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"), @@ -282,7 +342,7 @@ 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 + fname = fname[2] # Do not recode if it isn't recoded yet try: for forbidden in chr(0), '/', '\\': if forbidden in fname: @@ -297,12 +357,19 @@ def _save_message(msg, outstring, save_headers=False, save_body=False): fname = str(g.save_counter) + fname if '.' not in fname: ext = _guess_extension(msg.get_content_type()) - if ext: fname += ext + if ext: + fname += ext 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: @@ -339,23 +406,25 @@ def decode_part(msg): encoding = msg["Content-Transfer-Encoding"] if left_binary or encoding in (None, '', '7bit', '8bit', 'binary'): outstring = msg.get_payload() - else: # Decode from transfer ecoding to text or binary form + else: # Decode from transfer ecoding to text or binary form outstring = msg.get_payload(decode=1) set_header(msg, "Content-Transfer-Encoding", "8bit") - msg["X-MIME-Autoconverted"] = "from %s to 8bit by %s id %s" % (encoding, g.host_name, me) + msg["X-MIME-Autoconverted"] = \ + "from %s to 8bit by %s id %s" % (encoding, g.host_name, me) for content_type in masks: if content_type in g.totext_mask: 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("%sMessage body of type %s skipped.%s" % (os.linesep, ctype, os.linesep)) + output("%sMessage body of type %s skipped.%s" + % (os.linesep, ctype, os.linesep)) break elif content_type in g.error_mask: break @@ -375,6 +444,7 @@ def decode_part(msg): if content_type in g.error_mask: raise ValueError("content type %s prohibited" % ctype) + def decode_multipart(msg): "Decode multipart" @@ -394,7 +464,8 @@ def decode_multipart(msg): return elif content_type in g.ignore_mask: output_headers(msg) - output("%sMessage body of type %s skipped.%s" % (os.linesep, ctype, os.linesep)) + 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 @@ -431,8 +502,10 @@ def decode_multipart(msg): 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) + # Preserve the first part, it is probably not a RFC822-message. + if msg.preamble: + # Usually it is just a few lines of text (MIME warning). + output(msg.preamble) if msg.preamble is not None: output(os.linesep) @@ -460,9 +533,9 @@ def decode_message(msg): if msg.is_multipart(): decode_multipart(msg) - elif len(msg): # Simple one-part message (there are headers) - decode it + elif len(msg): # Simple one-part message (there are headers) - decode it decode_part(msg) - else: # Not a message, just text - copy it literally + else: # Not a message, just text - copy it literally output(msg.as_string()) @@ -481,7 +554,7 @@ def open_output_file(filename): class GlobalOptions: from m_lib.defenc import default_encoding - recode_charset = 1 # recode charset of message body + recode_charset = 1 # recode charset of message body host_name = None @@ -505,11 +578,15 @@ class GlobalOptions: # A list of header/parameter/value triples to set set_header_param = [] - 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 (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 + totext_mask = [] # A list of content-types to decode + binary_mask = [] # A list of content-types to pass through + # A list of content-types to pass through (content-transfer-decoded). + decoded_binary_mask = [] + # Ignore (do not decode and do not include into output) + # but output a warning instead of the body. + ignore_mask = [] + # Completely ignore - no headers, no body, no warning. + fully_ignore_mask = [] error_mask = [] # Raise error if encounter one of these save_counter = 0 @@ -521,6 +598,7 @@ class GlobalOptions: output_filename = None destination_dir = os.curdir + g = GlobalOptions @@ -528,7 +606,8 @@ 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 +726,14 @@ 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)