[Mailman-Developers] Handling of private gzipped archives

[Matthew Galgoci]

Hi Folks!

I've been bit by a bug in mailman 2.0.1 where mailman assumes that there will always be an
uncompressed version of a monthly private archive file hanging around, which for my setup 
isn't the case. Path truncation in the content_type routine wrongly chomps off .gz extension.

Here is how I fixed the bug:

--- mailman/Mailman/Cgi/private.py.orig Tue Aug 26 15:24:04 2003
+++ mailman/Mailman/Cgi/private.py      Tue Aug 26 16:06:04 2003
@@ -84,7 +84,8 @@
  
 def content_type(path):
     if path[-3:] == '.gz':
-        path = path[:-3]
+        # path = path[:-3]
+        return 'application/x-gzip'
     if path[-4:] == '.txt':
         return 'text/plain'
     return 'text/html'
@@ -186,9 +187,9 @@
             f = open(os.path.join(mlist.archive_directory + '.mbox',
                                   mlist.internal_name() + '.mbox'))
             ctype = 'text/plain'
-        elif true_filename[-3:] == '.gz':
-            import gzip
-            f = gzip.open(true_filename, 'r')
+        #elif true_filename[-3:] == '.gz':
+        #    import gzip
+        #    f = gzip.open(true_filename, 'r')
         else:
             f = open(true_filename, 'r')
     except IOError:


It also occured to me that mailman is bothering to decompress compressed archive files before sending them 
back to the web client. This is wrong in many ways:

1) extra load decompressing the file
2) clients getting a file back with a .gz extention that is really a text file
3) extra bandwidth consumed while sending the uncompressed file back to the client
4) many web browsers will silently decompress application/x-gzip mime data automatically

I checked Mailman 2.1.2 to see if any of this is still done, and apparently the decompression issue still is, at
least in the copy I have here. (haven't checked cvs head). Here's my version of a fix for this:

--- mailman-2.1.2/Mailman/Cgi/private.py.orig   2003-08-26 16:46:16.000000000 -0400
+++ mailman-2.1.2/Mailman/Cgi/private.py        2003-08-26 16:47:21.000000000 -0400
@@ -148,8 +148,8 @@
                                   mlist.internal_name() + '.mbox'))
             ctype = 'text/plain'
         elif true_filename.endswith('.gz'):
-            import gzip
-            f = gzip.open(true_filename, 'r')
+            ctype = 'application/x-gzip'
+            f = open(true_filename, 'r')
         else:
             f = open(true_filename, 'r')
     except IOError:


Most folks on non-Linux-like operating systems have some kind of decompression software available that can handle gz format
even if their browser does not.

Kind Regards,

Matthew Galgoci

-- 

Matthew Galgoci		"Dirty deeds, done dirt cheap"
System Administrator
Red Hat, Inc
919.754.3700 x44155