Allow * and exceptions for -p in the parameters lists

[mimedecode.git] / mimedecode.py

diff --git a/mimedecode.py b/mimedecode.py
index 6de9d1e4db128f7ba4d10e6aa048db30672b95ee..f3362cf824121c218daf2b8a9833f7281b663705 100755 (executable)
--- a/mimedecode.py
+++ b/mimedecode.py
@@ -18,7 +18,7 @@ Broytman mimedecode.py version %s, %s
 def usage(code=0, errormsg=''):
     version(0)
     sys.stdout.write("""\
-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]]
+Usage: %s [-h|--help] [-V|--version] [-cCDP] [-H|--host=hostname] [-f charset] [-d header1[,h2,...]|*[,-h1,...]] [-p header1[,h2,h3,...]:param1[,p2,p3,...]] [-r header] [-R header:param] [--remove-params=header] [-beit mask] [-o output_file] [input_file [output_file]]
 """ % me)
     if errormsg:
         sys.stderr.write(errormsg + '\n')
@@ -47,11 +47,10 @@ def _decode_header(s):
     """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
 
@@ -87,8 +86,7 @@ def decode_header_param(msg, header, param):
     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)
@@ -96,14 +94,69 @@ def decode_header_param(msg, header, param):
                 msg.set_param(param, new_value, header)
 
 
+def _get_exceptions(list):
+    return [x[1:].lower() for x in list[1:] if x[0]=='-']
+
 def decode_headers(msg):
     "Decode message headers according to global options"
 
-    for header in gopts.decode_headers:
-        decode_header(msg, header)
+    for header in gopts.remove_headers:
+        del msg[header]
 
-    for header, param in gopts.decode_header_params:
-        decode_header_param(msg, header, param)
+    for header in gopts.remove_all_params:
+        value = msg[header]
+        if value is None: # No such header
+            continue
+        if ';' not in value: # There are no parameters
+            continue
+        del msg[header] # Delete all such headers
+        # Get the value without parameters and set it back
+        msg[header] = value.split(';')[0].strip()
+
+    for header, param in gopts.remove_header_params:
+        msg.del_param(param, header)
+
+    for header_list in gopts.decode_headers:
+        header_list = header_list.split(',')
+        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
+            for header in header_list:
+                decode_header(msg, header)
+
+    for header_list, param_list in gopts.decode_header_params:
+        header_list = header_list.split(',')
+        param_list = param_list.split(',')
+        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
+            header_list = _get_exceptions(header_list)
+            for header in msg.keys():
+                if header.lower() not in header_list:
+                    if decode_all_params:
+                        params = msg.get_params(header=header)
+                        if params:
+                            for param, value in params:
+                                if param not in param_list:
+                                    decode_header_param(msg, header, param)
+                    else:
+                        for param in param_list:
+                            decode_header_param(msg, header, param)
+        else: # Decode for listed headers
+            for header in header_list:
+                if decode_all_params:
+                    params = msg.get_params(header=header)
+                    if params:
+                        for param, value in params:
+                            if param not in param_list:
+                                decode_header_param(msg, header, param)
+                else:
+                    for param in param_list:
+                        decode_header_param(msg, header, param)
 
 
 def set_header(msg, header, value):
@@ -273,11 +326,22 @@ class GlobalOptions:
 
     host_name = None
 
-    decode_headers = ["From", "To", "Cc", "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 = []
+    # A list of headers parameters to remove
+    remove_header_params = []
+    # A list of headers to be stripped of all parameters
+    remove_all_params = []
 
     totext_mask = [] # A list of content-types to decode
     binary_mask = [] # A list to pass through
@@ -294,8 +358,9 @@ def get_opt():
     from getopt import getopt, GetoptError
 
     try:
-        options, arguments = getopt(sys.argv[1:], 'hVcCDPH:f:d:p:b:e:i:t:o:',
-            ['help', 'version', 'host'])
+        options, arguments = getopt(sys.argv[1:],
+            'hVcCDPH:f:d:p:r:R:b:e:i:t:o:',
+            ['help', 'version', 'host=', 'remove-params='])
     except GetoptError:
         usage(1)
 
@@ -313,6 +378,8 @@ def get_opt():
         elif option == '-f':
             gopts.default_encoding = value
         elif option == '-d':
+            if value.startswith('*'):
+                gopts.decode_headers = []
             gopts.decode_headers.append(value)
         elif option == '-D':
             gopts.decode_headers = []
@@ -320,6 +387,12 @@ def get_opt():
             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 == '-R':
+            gopts.remove_header_params.append(value.split(':', 1))
+        elif option == '--remove-params':
+            gopts.remove_all_params.append(value)
         elif option == '-t':
             gopts.totext_mask.append(value)
         elif option == '-b':