X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=mimedecode.py;h=c12afbb963bcdd60eeaee789d6de8848d8ee1a26;hb=180a1e42c8b840f489a79d0012cb9ee3d2aa08c3;hp=6c29a9680c8bb927fdbfd5f1b7ef7ac53227b16f;hpb=fb1bca196122731a2c21ecd9e37413f19337fcd2;p=mimedecode.git diff --git a/mimedecode.py b/mimedecode.py index 6c29a96..c12afbb 100755 --- a/mimedecode.py +++ b/mimedecode.py @@ -6,11 +6,6 @@ from mimedecode_version import __version__, __author__, __copyright__, __license import sys, os import email -try: - from cStringIO import StringIO -except ImportError: - from StringIO import StringIO - me = os.path.basename(sys.argv[0]) @@ -23,7 +18,7 @@ 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 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,...]] [--set-header header:value] [--set-param header:param=value] [-Bbeit mask] [-o output_file] [input_file [output_file]] """ % me) if errormsg: sys.stderr.write(errormsg + '\n') @@ -52,11 +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 @@ -92,8 +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) @@ -101,14 +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): @@ -198,14 +272,6 @@ def decode_part(msg): "Decode one part of the message" decode_headers(msg) - encoding = msg["Content-Transfer-Encoding"] - - if encoding in (None, '', '7bit', '8bit', 'binary'): - outstring = str(msg.get_payload()) - 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, gopts.host_name, me) # Test all mask lists and find what to do with this content type masks = [] @@ -217,11 +283,26 @@ def decode_part(msg): masks.append(mtype + '/*') masks.append('*/*') + left_binary = False + for content_type in masks: + if content_type in gopts.binary_mask: + left_binary = True + break + + 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 + 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, gopts.host_name, me) + for content_type in masks: if content_type in gopts.totext_mask: totext(msg, outstring) return - elif content_type in gopts.binary_mask: + elif content_type in gopts.binary_mask or \ + content_type in gopts.decoded_binary_mask: output_headers(msg) output(outstring) return @@ -236,41 +317,40 @@ 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) - - if msg.is_multipart(): - decode_headers(msg) - output_headers(msg) + 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() + boundary = msg.get_boundary() - for subpart in msg.get_payload(): + for subpart in msg.get_payload(): + if boundary: output("\n--%s\n" % boundary) - if subpart.is_multipart(): # Recursively decode all parts of the subpart - newfile = StringIO(subpart.as_string()) - newfile.seek(0) - decode_file(newfile) - else: - 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(msg.as_string()) +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: @@ -279,14 +359,29 @@ class GlobalOptions: host_name = None - decode_headers = ["From", "Subject"] # A list of headers to decode + # 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 = [] + + # A list of header/value pairs to set + set_header_value = [] + # 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 to pass through + 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) error_mask = [] # Raise error if encounter one of these @@ -300,8 +395,9 @@ def get_opt(): from getopt import getopt, GetoptError try: - options, arguments = getopt(sys.argv[1:], 'hVcCDPH:f:d:p:b:e:i:t:o:', - ['help', 'version', 'host']) + options, arguments = getopt(sys.argv[1:], + 'hVcCDPH:f:d:p:r:R:b:B:e:i:t:o:', + ['help', 'version', 'host=', 'set-header=', 'set-param=']) except GetoptError: usage(1) @@ -319,6 +415,8 @@ def get_opt(): 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 = [] @@ -326,10 +424,25 @@ 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 == '--set-header': + gopts.set_header_value.append(value.split(':', 1)) + elif option == '--set-param': + header, value = value.split(':', 1) + if '=' in value: + param, value = value.split('=', 1) + else: + param, value = value.split(':', 1) + gopts.set_header_param.append((header, param, value)) elif option == '-t': gopts.totext_mask.append(value) - elif option == '-b': + elif option == '-B': gopts.binary_mask.append(value) + elif option == '-b': + gopts.decoded_binary_mask.append(value) elif option == '-i': gopts.ignore_mask.append(value) elif option == '-e': @@ -391,8 +504,17 @@ if __name__ == "__main__": gopts.outfile = outfile output = outfile.write + msg = email.message_from_file(infile) + + for header, value in gopts.set_header_value: + msg[header] = value + + for header, param, value in gopts.set_header_param: + if header in msg: + msg.set_param(param, value, header) + try: - decode_file(infile) + decode_message(msg) finally: infile.close() outfile.close()