X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=mimedecode.py;h=1e5b18b388dbcb2d6104c12911742181648118fc;hb=fa51f1e417da0f8d66649ebe67c78a58b1b88eae;hp=b6f139b6974ba2cb141d9c6999085cfdce22a02a;hpb=4467e66adb26be6409c68e7ff5408776b224170a;p=mimedecode.git diff --git a/mimedecode.py b/mimedecode.py index b6f139b..1e5b18b 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]) @@ -23,16 +18,13 @@ 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 header] [-p header:param] [-r header] [-R header:param] [--remove-params=header] [-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: @@ -55,11 +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 @@ -95,8 +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) @@ -107,6 +97,21 @@ def decode_header_param(msg, header, param): def decode_headers(msg): "Decode message headers according to global options" + for header in gopts.remove_headers: + del msg[header] + + for header in gopts.remove_all_params: + value = msg[header] + if value is None: + continue + if ';' not in value: + continue + del msg[header] + msg[header] = value.split(';')[0].strip() + + for header, param in gopts.remove_header_params: + msg.del_param(param, header) + for header in gopts.decode_headers: decode_header(msg, header) @@ -188,12 +193,6 @@ def recode_charset(msg, s): def totext(msg, instring): "Convert instring content to text" - if msg.is_multipart(): # Recursively decode all parts of the multipart message - newfile = StringIO(msg.as_string()) - newfile.seek(0) - decode_file(newfile) - return - # Decode body and recode charset s = decode_body(msg, instring) if gopts.recode_charset: @@ -245,34 +244,40 @@ 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) - 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(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: @@ -281,11 +286,22 @@ 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_header_params = [] + # A list of headers to be stripped of all parameters + remove_all_params = [] totext_mask = [] # A list of content-types to decode binary_mask = [] # A list to pass through @@ -302,8 +318,9 @@ 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:e:i:t:o:', + ['help', 'version', 'host=', 'remove-params=']) except GetoptError: usage(1) @@ -328,6 +345,12 @@ 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_header_params.append(value.split(':', 1)) + elif option == '--remove-params': + gopts.remove_all_params.append(value) elif option == '-t': gopts.totext_mask.append(value) elif option == '-b': @@ -391,7 +414,10 @@ if __name__ == "__main__": 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()