import sys, os
import email
-try:
- from cStringIO import StringIO
-except ImportError:
- from StringIO import StringIO
-
me = os.path.basename(sys.argv[0])
"""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
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)
for header, param in gopts.decode_header_params:
decode_header_param(msg, header, param)
+ for header in gopts.remove_headers:
+ del msg[header]
+
def set_header(msg, header, value):
"Replace header"
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:
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:
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 = []
totext_mask = [] # A list of content-types to decode
binary_mask = [] # A list to pass through
from getopt import getopt, GetoptError
try:
- options, arguments = getopt(sys.argv[1:], 'hVcCDPH:f:d:p:b:e:i:t:o:',
+ options, arguments = getopt(sys.argv[1:], 'hVcCDPH:f:d:p:r:b:e:i:t:o:',
['help', 'version', 'host'])
except GetoptError:
usage(1)
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 == '-t':
gopts.totext_mask.append(value)
elif option == '-b':
gopts.outfile = outfile
output = outfile.write
- decode_file(infile)
-
- infile.close()
- outfile.close()
+ try:
+ decode_message(email.message_from_file(infile))
+ finally:
+ infile.close()
+ outfile.close()