X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=mimedecode.py;h=637e275e3b127c6598850237fafcec0ca325c856;hb=b47b1f1629b4a4887a422edcf54bf23865c2e2ae;hp=f46165b23936cf100e432c44640d0bf293496ac0;hpb=44fab1065b5918a09093798b2279ebb2852ad169;p=mimedecode.git diff --git a/mimedecode.py b/mimedecode.py index f46165b..637e275 100755 --- a/mimedecode.py +++ b/mimedecode.py @@ -1,22 +1,11 @@ #! /usr/bin/env python """Decode MIME message""" - from mimedecode_version import __version__, __author__, __copyright__, __license__ - import sys, os import email -try: - from cStringIO import StringIO -except ImportError: - from StringIO import StringIO - - -import socket -host_name = socket.gethostname() - me = os.path.basename(sys.argv[0]) @@ -26,20 +15,16 @@ Broytman mimedecode.py version %s, %s """ % (__version__, __copyright__)) if exit: sys.exit(0) - def usage(code=0, errormsg=''): version(0) sys.stdout.write("""\ -Usage: %s [-h|--help] [-V|--version] [-cCDP] [-f charset] [-d header] [-p header:param] [-beit mask] [-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,...]] [-beit mask] [-o output_file] [input_file [output_file]] """ % me) if errormsg: sys.stderr.write(errormsg + '\n') sys.exit(code) -def output(s): - gopts.outfile.write(s) - def output_headers(msg): unix_from = msg.get_unixfrom() if unix_from: @@ -52,8 +37,7 @@ def output_headers(msg): def recode(s, charset): return unicode(s, charset, "replace").encode(gopts.default_encoding, "replace") - -def recode2(s, charset): +def recode_if_needed(s, charset): if charset and charset.lower() <> gopts.default_encoding: s = recode(s, charset) return s @@ -63,11 +47,10 @@ def _decode_header(s): """Return a decoded string according to RFC 2047. NOTE: This is almost the same as email.Utils.decode. """ - from types import ListType import email.Header L = email.Header.decode_header(s) - if not isinstance(L, ListType): + if not isinstance(L, list): # s wasn't decoded return s @@ -76,7 +59,7 @@ def _decode_header(s): if charset is None: rtn.append(atom) else: - rtn.append(recode2(atom, charset)) + rtn.append(recode_if_needed(atom, charset)) rtn.append(' ') del rtn[-1] # remove the last space @@ -84,7 +67,6 @@ def _decode_header(s): # together into the final string. return ''.join(rtn) - def decode_header(msg, header): "Decode mail header (if exists) and put it back, if it was encoded" @@ -96,8 +78,7 @@ def decode_header(msg, header): def _decode_header_param(s): - return recode2(s[2], s[0]) - + return recode_if_needed(s[2], s[0]) def decode_header_param(msg, header, param): "Decode mail header's parameter (if exists) and put it back, if it was encoded" @@ -105,8 +86,7 @@ def decode_header_param(msg, header, param): if msg.has_key(header): value = msg.get_param(param, header=header) if value: - from types import TupleType - if isinstance(value, TupleType): + if isinstance(value, tuple): new_value = _decode_header_param(value) else: new_value = _decode_header(value) @@ -114,14 +94,95 @@ def decode_header_param(msg, header, param): msg.set_param(param, new_value, header) +def _get_exceptions(list): + 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: + params = msg.get_params(header=header) + if params: + for param, value in params: + if param not in param_list: + decode_header_param(msg, header, param) + else: + 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) + if params: + if param_list: + for param, value in params: + if param not in param_list: + msg.del_param(param, header) + else: + value = msg[header] + if value is None: # No such header + return + if ';' not in value: # There are no parameters + return + 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 in gopts.decode_headers: - decode_header(msg, header) - - for header, param in gopts.decode_header_params: - decode_header_param(msg, header, param) + for header_list in gopts.remove_headers: + header_list = header_list.split(',') + 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 + for header in header_list: + del msg[header] + + for header_list, param_list in gopts.remove_headers_params: + header_list = header_list.split(',') + param_list = param_list.split(',') + remove_all_params = param_list[0] == '*' # Remove all params except listed + if remove_all_params: + param_list = _get_exceptions(param_list) + 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 + for header in header_list: + _remove_headers_params(msg, header, remove_all_params, param_list) + + for header_list in gopts.decode_headers: + header_list = header_list.split(',') + 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 + for header in header_list: + decode_header(msg, header) + + for header_list, param_list in gopts.decode_header_params: + header_list = header_list.split(',') + param_list = param_list.split(',') + decode_all_params = param_list[0] == '*' # Decode all params except listed + if decode_all_params: + param_list = _get_exceptions(param_list) + 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 + for header in header_list: + _decode_headers_params(msg, header, decode_all_params, param_list) def set_header(msg, header, value): @@ -140,7 +201,6 @@ 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 def decode_body(msg, s): @@ -179,7 +239,7 @@ def decode_body(msg, 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, host_name, command.split()[0]) + msg["X-MIME-Autoconverted"] = "from %s to text/plain by %s id %s" % (content_type, gopts.host_name, command.split()[0]) return s @@ -189,22 +249,16 @@ def recode_charset(msg, s): save_charset = charset = msg.get_content_charset() if charset and charset.lower() <> gopts.default_encoding: - s = recode2(s, charset) + s = recode_if_needed(s, charset) content_type = msg.get_content_type() set_content_type(msg, content_type, gopts.default_encoding) - msg["X-MIME-Autoconverted"] = "from %s to %s by %s id %s" % (save_charset, gopts.default_encoding, host_name, me) + msg["X-MIME-Autoconverted"] = "from %s to %s by %s id %s" % (save_charset, gopts.default_encoding, gopts.host_name, me) return s def totext(msg, instring): "Convert instring content to text" - if msg.is_multipart(): # Recursively decode all parts of the multipart message - newfile = StringIO(str(msg)) - newfile.seek(0) - decode_file(newfile) - return - # Decode body and recode charset s = decode_body(msg, instring) if gopts.recode_charset: @@ -225,7 +279,7 @@ def decode_part(msg): else: # Decode from transfer ecoding to text or binary form outstring = str(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, host_name, me) + msg["X-MIME-Autoconverted"] = "from %s to 8bit by %s id %s" % (encoding, gopts.host_name, me) # Test all mask lists and find what to do with this content type masks = [] @@ -256,45 +310,62 @@ def decode_part(msg): totext(msg, outstring) -def decode_file(infile): - "Decode the entire message" +def decode_multipart(msg): + "Decode multipart" - msg = email.message_from_file(infile) - boundary = msg.get_boundary() + decode_headers(msg) + output_headers(msg) - if msg.is_multipart(): - decode_headers(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) - 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) + boundary = msg.get_boundary() - for subpart in msg.get_payload(): + for subpart in msg.get_payload(): + if boundary: output("\n--%s\n" % boundary) - decode_part(subpart) + # Recursively decode all parts of the subpart + decode_message(subpart) + + if boundary: output("\n--%s--\n" % boundary) - if msg.epilogue: - output(msg.epilogue) + if msg.epilogue: + output(msg.epilogue) - else: - if msg.has_key("Content-Type"): # Simple one-part message - decode it - decode_part(msg) - else: # Not a message, just text - copy it literally - output(str(msg)) +def decode_message(msg): + "Decode message" + + if msg.is_multipart(): + decode_multipart(msg) + elif len(msg): # Simple one-part message (there are headers) - decode it + decode_part(msg) + else: # Not a message, just text - copy it literally + output(msg.as_string()) class GlobalOptions: from m_lib.defenc import default_encoding recode_charset = 1 # recode charset of message body - decode_headers = ["From", "Subject"] # A list of headers to decode + host_name = None + + # A list of headers to decode + decode_headers = ["From", "To", "Cc", "Reply-To", "Mail-Followup-To", + "Subject"] + + # A list of headers parameters to decode decode_header_params = [ ("Content-Type", "name"), ("Content-Disposition", "filename"), - ] # A list of headers' parameters to decode + ] + + # A list of headers to remove + remove_headers = [] + # A list of headers parameters to remove + remove_headers_params = [] totext_mask = [] # A list of content-types to decode binary_mask = [] # A list to pass through @@ -311,27 +382,28 @@ def get_opt(): from getopt import getopt, GetoptError try: - options, arguments = getopt(sys.argv[1:], 'hVcCDPf:d:p:b:e:i:t:o:', - ['help', 'version']) + options, arguments = getopt(sys.argv[1:], + 'hVcCDPH:f:d:p:r:R:b:e:i:t:o:', + ['help', 'version', 'host=']) except GetoptError: usage(1) for option, value in options: - if option == '-h': - usage() - elif option == '--help': + if option in ('-h', '--help'): usage() - elif option == '-V': - version() - elif option == '--version': + elif option in ('-V', '--version'): version() elif option == '-c': gopts.recode_charset = 1 elif option == '-C': gopts.recode_charset = 0 + elif option in ('-H', '--host'): + gopts.host_name = value elif option == '-f': gopts.default_encoding = value elif option == '-d': + if value.startswith('*'): + gopts.decode_headers = [] gopts.decode_headers.append(value) elif option == '-D': gopts.decode_headers = [] @@ -339,6 +411,10 @@ def get_opt(): gopts.decode_header_params.append(value.split(':', 1)) elif option == '-P': gopts.decode_header_params = [] + elif option == '-r': + gopts.remove_headers.append(value) + elif option == '-R': + gopts.remove_headers_params.append(value.split(':', 1)) elif option == '-t': gopts.totext_mask.append(value) elif option == '-b': @@ -361,9 +437,12 @@ if __name__ == "__main__": la = len(arguments) if la == 0: gopts.input_filename = '-' - gopts.output_filename = '-' infile = sys.stdin - outfile = sys.stdout + if gopts.output_filename: + outfile = open(gopts.output_filename, 'w') + else: + gopts.output_filename = '-' + outfile = sys.stdout elif la in (1, 2): if (arguments[0] == '-'): gopts.input_filename = '-' @@ -389,8 +468,20 @@ if __name__ == "__main__": else: usage(1, 'Too many arguments') + if (infile is sys.stdin) and sys.stdin.isatty(): + if (outfile is sys.stdout) and sys.stdout.isatty(): + usage() + usage(1, 'Filtering from console is forbidden') + + if not gopts.host_name: + import socket + gopts.host_name = socket.gethostname() + gopts.outfile = outfile - decode_file(infile) + output = outfile.write - infile.close() - outfile.close() + try: + decode_message(email.message_from_file(infile)) + finally: + infile.close() + outfile.close()