Conversion problem in parse_multipart

[Michele Belloli]

On Wed, 01 Oct 2003 20:23:56 -0700, Tim Roberts <timr at probo.com>
wrote:

>Michele Belloli <mkbelloli at libero.it> wrote:
>>
>>I have a problem uploading a file to a web server.
>>
>>I created an html page with a multipart/form-data form and used
>>cgi.parse_multipart function to obtain form data (and obviuosly
>>uploaded file).
>>I noticed that if I upload an excutable file, I obtain a file where
>>0x0a 0x0a, are converted in 0x0a and 0x2e.
>>Obviously this file is corrupted.
>
>Are you sure?  We had something similar reported just a few weeks ago where
>the bug turned out to be in the program displaying the hex results; the
>binary file was just fine
>
>I have used cgi.py to handle large file uploads literally tens of thousands
>of times, and never had one go bad.
>
 
I read cgi.py, in particular 'parse_multipart'; and I repeated each
command in a python shell.
It works fine.
I use the same version (2.2.3) in Python shell and web server but I
obtain two different behaviuors.

>>Analyzing the  file before cgi.parse_multipart, it's possible to
>>verify that it's correct but when this function tries to extract its
>>value, 0x0a 0x0a are converted in 0x0a and 02e.
>
>How are you determining this?  That is, how are you accessing the final
>data?
>

I noticed a bug displaying the hex results using 'print' statement
too, so I created a binary file (using 'wb' flags) and write down
results. Then, I read it using an hex-editor.
Reading file, I tought 'write' method could modify result like 'print'
statement do.
So I used a simple conditional statement.
if result =='\x2e':
  file.write(.....)

Both test in cgi.parse_mulitpart shows the problem.

>What operating system and what web server?

W2K and python web server using SimpleHTTPServer classes


Thanks