import sys, os
import email
-try:
- from cStringIO import StringIO
-except ImportError:
- from StringIO import StringIO
-
me = os.path.basename(sys.argv[0])
sys.exit(code)
-def output(s):
- gopts.outfile.write(s)
-
def output_headers(msg):
unix_from = msg.get_unixfrom()
if unix_from:
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
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
# 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"
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"
from types import TupleType
if isinstance(value, TupleType):
new_value = _decode_header_param(value)
+ changed = new_value <> value[2]
else:
new_value = _decode_header(value)
- if new_value <> value: # do not bother to touch msg if not changed
+ changed = new_value <> value
+ if changed: # do not bother to touch msg if not changed
msg.set_param(param, new_value, header)
msg.set_param("charset", charset, "Content-Type")
-
caps = None # Globally stored mailcap database; initialized only if needed
def decode_body(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)
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:
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:
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 = '-'
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()