X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=mimedecode.py;h=3f5879ef4496d7a64cdc89026dd328007615b6d6;hb=3eb76865c7fd23c300d8d60c42c74039532587c3;hp=cd972210a1dc46d35dd55a3e2a1bb1cd11811863;hpb=c1a5169774327b8f530ed41ff061748e8b10923b;p=mimedecode.git diff --git a/mimedecode.py b/mimedecode.py index cd97221..3f5879e 100755 --- a/mimedecode.py +++ b/mimedecode.py @@ -1,10 +1,13 @@ #! /usr/bin/env python """Decode MIME message""" -import sys, os +import os import subprocess +import sys + from mimedecode_version import __version__, \ __author__, __copyright__, __license__ + if sys.version_info[0] >= 3: # Replace email.message._formatparam with _formatparam from Python 2.7 # to avoid re-encoding non-ascii params. @@ -19,6 +22,7 @@ Broytman mimedecode.py version %s, %s """ % (__version__, __copyright__)) if exit: sys.exit(0) + def usage(code=0, errormsg=''): version(0) sys.stdout.write("""\ @@ -43,16 +47,16 @@ def output_headers(msg): output(";") output(_decode_header(value[1], strip=False)) output(os.linesep) - output(os.linesep) # End of headers + output(os.linesep) # End of headers def recode_if_needed(s, charset): - if bytes is str: # Python2 + if bytes is str: # Python2 if isinstance(s, bytes) and \ charset and charset.lower() != g.default_encoding: s = s.decode(charset, "replace").\ encode(g.default_encoding, "replace") - else: # Python3 + else: # Python3 if isinstance(s, bytes): s = s.decode(charset, "replace") return s @@ -80,13 +84,14 @@ def _decode_header(s, strip=True): # 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" if header in msg: value = msg[header] new_value = _decode_header(value) - if new_value != value: # do not bother to touch msg if not changed + if new_value != value: # do not bother to touch msg if not changed set_header(msg, header, new_value) @@ -100,12 +105,13 @@ def decode_header_param(msg, header, param): new_value = recode_if_needed(value[2], value[0]) else: new_value = _decode_header(value) - if new_value != value: # do not bother to touch msg if not changed + if new_value != value: # do not bother to touch msg if not changed msg.set_param(param, new_value, header) def _get_exceptions(list): - return [x[1:].lower() for x in list[1:] if x[0]=='-'] + return [x[1:].lower() for x in list[1:] if x[0] == '-'] + def _decode_headers_params(msg, header, decode_all_params, param_list): if decode_all_params: @@ -118,6 +124,7 @@ def _decode_headers_params(msg, header, decode_all_params, param_list): for param in param_list: decode_header_param(msg, header, param) + def _remove_headers_params(msg, header, remove_all_params, param_list): if remove_all_params: params = msg.get_params(header=header) @@ -128,69 +135,70 @@ def _remove_headers_params(msg, header, remove_all_params, param_list): msg.del_param(param, header) else: value = msg[header] - if value is None: # No such header + if value is None: # No such header return - if ';' not in value: # There are no parameters + if ';' not in value: # There are no parameters return - del msg[header] # Delete all such headers + del msg[header] # Delete all such headers # Get the value without parameters and set it back msg[header] = value.split(';')[0].strip() else: for param in param_list: msg.del_param(param, header) + def decode_headers(msg): "Decode message headers according to global options" for header_list in g.remove_headers: header_list = header_list.split(',') - if header_list[0] == '*': # Remove all headers except listed + if header_list[0] == '*': # Remove all headers except listed header_list = _get_exceptions(header_list) for header in msg.keys(): if header.lower() not in header_list: del msg[header] - else: # Remove listed headers + else: # Remove listed headers for header in header_list: del msg[header] for header_list, param_list in g.remove_headers_params: header_list = header_list.split(',') param_list = param_list.split(',') - remove_all_params = param_list[0] == '*' # Remove all params except listed + remove_all_params = param_list[0] == '*' # Remove all params except listed if remove_all_params: param_list = _get_exceptions(param_list) - if header_list[0] == '*': # Remove for all headers except listed + if header_list[0] == '*': # Remove for all headers except listed header_list = _get_exceptions(header_list) for header in msg.keys(): if header.lower() not in header_list: _remove_headers_params(msg, header, remove_all_params, param_list) - else: # Decode for listed headers + else: # Decode for listed headers for header in header_list: _remove_headers_params(msg, header, remove_all_params, param_list) for header_list in g.decode_headers: header_list = header_list.split(',') - if header_list[0] == '*': # Decode all headers except listed + if header_list[0] == '*': # Decode all headers except listed header_list = _get_exceptions(header_list) for header in msg.keys(): if header.lower() not in header_list: decode_header(msg, header) - else: # Decode listed headers + else: # Decode listed headers for header in header_list: decode_header(msg, header) for header_list, param_list in g.decode_header_params: header_list = header_list.split(',') param_list = param_list.split(',') - decode_all_params = param_list[0] == '*' # Decode all params except listed + decode_all_params = param_list[0] == '*' # Decode all params except listed if decode_all_params: param_list = _get_exceptions(param_list) - if header_list[0] == '*': # Decode for all headers except listed + if header_list[0] == '*': # Decode for all headers except listed header_list = _get_exceptions(header_list) for header in msg.keys(): if header.lower() not in header_list: _decode_headers_params(msg, header, decode_all_params, param_list) - else: # Decode for listed headers + else: # Decode for listed headers for header in header_list: _decode_headers_params(msg, header, decode_all_params, param_list) @@ -211,12 +219,14 @@ 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 +caps = None # Globally stored mailcap database; initialized only if needed + def decode_body(msg, s): "Decode body to plain text using first copiousoutput filter from mailcap" - import mailcap, tempfile + import mailcap + import tempfile global caps if caps is None: @@ -254,7 +264,7 @@ def decode_body(msg, s): pipe = subprocess.Popen(command, shell=True, stdout=subprocess.PIPE) new_s = pipe.stdout.read() pipe.stdout.close() - if pipe.wait() == 0: # result=0, Ok + if pipe.wait() == 0: # result=0, Ok s = new_s if bytes is not str and isinstance(s, bytes): # Python3 s = s.decode(g.default_encoding, "replace") @@ -296,6 +306,7 @@ def totext(msg, instring): mimetypes = None + def _guess_extension(ctype): global mimetypes if mimetypes is None: @@ -306,6 +317,7 @@ def _guess_extension(ctype): mimetypes._db.read(user_mime_type) return mimetypes.guess_extension(ctype) + def _save_message(msg, outstring, save_headers=False, save_body=False): for header, param in ( ("Content-Disposition", "filename"), @@ -314,7 +326,7 @@ def _save_message(msg, outstring, save_headers=False, save_body=False): fname = msg.get_param(param, header=header) if fname: if isinstance(fname, tuple): - fname = fname[2] # Do not recode if it isn't recoded yet + fname = fname[2] # Do not recode if it isn't recoded yet try: for forbidden in chr(0), '/', '\\': if forbidden in fname: @@ -334,10 +346,12 @@ def _save_message(msg, outstring, save_headers=False, save_body=False): global output save_output = output outfile = open_output_file(fname) + def _output_bytes(s): if not isinstance(s, bytes): s = s.encode(g.default_encoding, "replace") outfile.write(s) + output = _output_bytes if save_headers: output_headers(msg) @@ -375,7 +389,7 @@ def decode_part(msg): encoding = msg["Content-Transfer-Encoding"] if left_binary or encoding in (None, '', '7bit', '8bit', 'binary'): outstring = msg.get_payload() - else: # Decode from transfer ecoding to text or binary form + else: # Decode from transfer ecoding to text or binary form outstring = 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, g.host_name, me) @@ -411,6 +425,7 @@ def decode_part(msg): if content_type in g.error_mask: raise ValueError("content type %s prohibited" % ctype) + def decode_multipart(msg): "Decode multipart" @@ -467,8 +482,8 @@ def decode_multipart(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) if msg.preamble is not None: output(os.linesep) @@ -496,9 +511,9 @@ def decode_message(msg): if msg.is_multipart(): decode_multipart(msg) - elif len(msg): # Simple one-part message (there are headers) - decode it + elif len(msg): # Simple one-part message (there are headers) - decode it decode_part(msg) - else: # Not a message, just text - copy it literally + else: # Not a message, just text - copy it literally output(msg.as_string()) @@ -517,7 +532,7 @@ def open_output_file(filename): class GlobalOptions: from m_lib.defenc import default_encoding - recode_charset = 1 # recode charset of message body + recode_charset = 1 # recode charset of message body host_name = None @@ -541,11 +556,11 @@ class GlobalOptions: # A list of header/parameter/value triples to set set_header_param = [] - totext_mask = [] # A list of content-types to decode - binary_mask = [] # A list of content-types to pass through - decoded_binary_mask = [] # A list of content-types to pass through (content-transfer-decoded) - ignore_mask = [] # Ignore (do not decode and do not include into output) but output a warning instead of the body - fully_ignore_mask = [] # Completely ignore - no headers, no body, no warning + totext_mask = [] # A list of content-types to decode + binary_mask = [] # A list of content-types to pass through + decoded_binary_mask = [] # A list of content-types to pass through (content-transfer-decoded) + ignore_mask = [] # Ignore (do not decode and do not include into output) but output a warning instead of the body + fully_ignore_mask = [] # Completely ignore - no headers, no body, no warning error_mask = [] # Raise error if encounter one of these save_counter = 0 @@ -557,6 +572,7 @@ class GlobalOptions: output_filename = None destination_dir = os.curdir + g = GlobalOptions @@ -564,7 +580,8 @@ def get_opts(): from getopt import getopt, GetoptError try: - options, arguments = getopt(sys.argv[1:], + options, arguments = getopt( + sys.argv[1:], 'hVcCDPH:f:d:p:r:R:b:B:e:I:i:t:O:o:', ['help', 'version', 'host=', 'save-headers=', 'save-body=', 'save-message=',