X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=mimedecode.py;h=4776126474cbd4e00b036cdfd368f89d46219108;hb=d6315799230ac62abddd0604498e93600578f7ec;hp=0e15fe015d3bcaaaed707cd50e8f7fb5ae6ae81c;hpb=9c35fa7761509941014db31c46ad7a6e5a6e8817;p=mimedecode.git diff --git a/mimedecode.py b/mimedecode.py index 0e15fe0..4776126 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]) @@ -30,9 +25,6 @@ Usage: %s [-h|--help] [-V|--version] [-cCDP] [-H|--host=hostname] [-f charset] [ sys.exit(code) -def output(s): - gopts.outfile.write(s) - def output_headers(msg): unix_from = msg.get_unixfrom() if unix_from: @@ -45,8 +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 @@ -69,7 +60,7 @@ 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 @@ -77,7 +68,6 @@ 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" @@ -89,8 +79,7 @@ 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" @@ -133,7 +122,6 @@ 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): @@ -182,7 +170,7 @@ 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, gopts.host_name, me) @@ -192,12 +180,6 @@ 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(str(msg)) - newfile.seek(0) - decode_file(newfile) - return - # Decode body and recode charset s = decode_body(msg, instring) if gopts.recode_charset: @@ -249,34 +231,40 @@ 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(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: @@ -285,7 +273,7 @@ class GlobalOptions: host_name = None - decode_headers = ["From", "Subject"] # A list of headers to decode + decode_headers = ["From", "To", "Subject"] # A list of headers to decode decode_header_params = [ ("Content-Type", "name"), ("Content-Disposition", "filename"), @@ -354,9 +342,12 @@ 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 = '-' @@ -382,16 +373,20 @@ if __name__ == "__main__": else: usage(1, 'Too many arguments') - if (infile is sys.stdin) and (outfile is sys.stdout) and \ - sys.stdin.isatty() and sys.stdout.isatty(): - usage(1, 'Filtering from console to console is forbidden') + 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()