X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=mimedecode.py;h=78783ae6e2e7ffe930d5db226d54851e2433f7fe;hb=75fa3299ddbda6c432999a26d16fccf7bccfb34d;hp=96c3ea57c1132229977f63ee85ed95e7db4a3c7d;hpb=d10dabb4acb56f746624bd697ede70c686c0a174;p=mimedecode.git diff --git a/mimedecode.py b/mimedecode.py index 96c3ea5..78783ae 100755 --- a/mimedecode.py +++ b/mimedecode.py @@ -1,10 +1,8 @@ #! /usr/bin/env python """Decode MIME message""" - from mimedecode_version import __version__, __author__, __copyright__, __license__ - import sys, os import email @@ -13,10 +11,6 @@ try: except ImportError: from StringIO import StringIO - -import socket -host_name = socket.gethostname() - me = os.path.basename(sys.argv[0]) @@ -26,32 +20,29 @@ Broytman mimedecode.py version %s, %s """ % (__version__, __copyright__)) if exit: sys.exit(0) - -def usage(code=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] [filename] +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]] """ % me) + if errormsg: + sys.stderr.write(errormsg + '\n') sys.exit(code) -def output(s, outfile=sys.stdout): - outfile.write(s) - -def output_headers(msg, outfile=sys.stdout): +def output_headers(msg): unix_from = msg.get_unixfrom() if unix_from: output(unix_from + '\n') for key, value in msg.items(): - output("%s: %s\n" % (key, value), outfile) - output("\n", outfile) # End of headers + output("%s: %s\n" % (key, value)) + output("\n") # End of headers 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 @@ -74,7 +65,7 @@ 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 @@ -82,7 +73,6 @@ 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" @@ -94,8 +84,7 @@ 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" @@ -138,7 +127,6 @@ 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): @@ -177,7 +165,7 @@ 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 @@ -187,10 +175,10 @@ 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 @@ -198,7 +186,7 @@ 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 = StringIO(msg.as_string()) newfile.seek(0) decode_file(newfile) return @@ -223,7 +211,7 @@ 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 = [] @@ -281,13 +269,15 @@ def decode_file(infile): decode_part(msg) else: # Not a message, just text - copy it literally - output(str(msg)) + output(msg.as_string()) class GlobalOptions: from m_lib.defenc import default_encoding recode_charset = 1 # recode charset of message body + host_name = None + decode_headers = ["From", "Subject"] # A list of headers to decode decode_header_params = [ ("Content-Type", "name"), @@ -299,31 +289,32 @@ class GlobalOptions: ignore_mask = [] # Ignore (skip, do not decode and do not include into output) error_mask = [] # Raise error if encounter one of these + input_filename = None + output_filename = None + gopts = GlobalOptions -def init(): +def get_opt(): from getopt import getopt, GetoptError try: - options, arguments = getopt(sys.argv[1:], 'hVcCDPf:d:p:b:e:i:t:', - ['help', 'version']) + options, arguments = getopt(sys.argv[1:], 'hVcCDPH:f:d:p:b:e:i:t:o:', + ['help', 'version', 'host']) except GetoptError: usage(1) for option, value in options: - if option == '-h': + if option in ('-h', '--help'): usage() - elif option == '--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': @@ -342,6 +333,8 @@ def init(): gopts.ignore_mask.append(value) elif option == '-e': gopts.error_mask.append(value) + elif option == '-o': + gopts.output_filename = value else: usage(1) @@ -349,14 +342,55 @@ def init(): if __name__ == "__main__": - arguments = init() + arguments = get_opt() la = len(arguments) - if la >= 2: - usage(1) - if (la == 0) or (arguments[0] == '-'): + if la == 0: + gopts.input_filename = '-' infile = sys.stdin + 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 = '-' + infile = sys.stdin + else: + gopts.input_filename = arguments[0] + infile = open(arguments[0], 'r') + if la == 1: + if gopts.output_filename: + outfile = open(gopts.output_filename, 'w') + else: + gopts.output_filename = '-' + outfile = sys.stdout + elif la == 2: + if gopts.output_filename: + usage(1, 'Too many output filenames') + if (arguments[1] == '-'): + gopts.output_filename = '-' + outfile = sys.stdout + else: + gopts.output_filename = arguments[1] + outfile = open(arguments[1], 'w') else: - infile = open(arguments[0], 'r') + 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 + output = outfile.write decode_file(infile) + + infile.close() + outfile.close()