I second that.. That approach should never be used.
Imagine altering your webpage to be in that case the following:
c:\SomeTempDirectory\..\boot.ini
Next thing you know.. Important files go missing off the system.
-----Original Message-----
From: Calderon, Juan Carlos (GE Commercial Finance, NonGE)
[mailto:juan.calderon@ge.com]
Sent: Thursday, September 30, 2004 6:15 PM
To: Beckner, Chad A; Shields, Larry; webappsec@securityfocus.com
Subject: RE: Securing file access
Importance: High
woah!
Sending a path in a hidden field to remove a file in server doesn't sound
good to me. be carefull with this approach!
-----Original Message-----
From: Beckner, Chad A [mailto:cbeckner@iupui.edu]
Sent: Miércoles, 29 de Septiembre de 2004 09:09 p.m.
To: Shields, Larry; webappsec@securityfocus.com
Subject: RE: Securing file access
You could also copy the file to a temporary read-only directory, and run an
administrative script to clean out the temp directory every night, hour, or
even after the user has received the file (i.e. on the next page hit using a
hidden variable, such as "remove_file", value="adobe.pdf").
- Chad
-----Original Message-----
From: Shields, Larry [mailto:Larry.Shields@FMR.COM]
Sent: Wednesday, September 29, 2004 1:16 PM
To: webappsec@securityfocus.com
Subject: RE: Securing file access
Be aware that the acrobat reader if using the IE browser plug-in
viewer has been known to have problems when using this method to send PDF
files. We have seen common problems where users ended up with blank
documents or received errors about the PDF file not being correct. From
what I've seen, the two solutions if you can't have this possible failure
are to either a) make customers all change their acrobat reader
configuration to start in "helper" mode (meaning kick off acrobat external
to the browser plug-in) or b) store things on the file system and redirect
the user to the PDF file. Option b does have issues to make sure it is done
securely of course, such as clean-up of old files, making sure that file
names cannot be guessed or predicted by other users, and that there is no
potential for file naming collisions between users.
-Larry
-----Original Message-----
From: Ian [mailto:webappsec2@fishnet.co.uk]
Sent: Tuesday, September 28, 2004 5:05 AM
To: John M. L.; webappsec@securityfocus.com
Subject: Re: Securing file access
On 27 Sep 2004 at 11:57, John M. L. wrote:
I have a project that involves a members only area on web page on IIS.
The members' only area is secured by a database (MS Access) so users
are authenticated by their name and some MD5 hash etc. I need to
allow files (mostly PDFs) for download to authenticated users only.
In my opinion this means that the files can not be stored in any www
accessible folder (regardless of any renaming convention etc, I
absolutely cannot have someone guess a file name to download). In
order to access the files, the database would link a file to a unique
id, so a page that validates the user would then give access to the
file stored outside of the www on the server. Now, this is where the
real question lies. How is this possible since the files are not in a
www accessible path, since a mere link to a file won't due.
Any thoughts would be welcome. If I'm going about this completely
wrong that would be nice to no too :) Forgive me if the answer is
simple, I'm a Linux fan and haven't used IIS etc for years.
One more note: IIS, MS Access and VBScript are not my technologies of
choice, but merely what I was given to work with. I also have very
limited control over administering IIS.
John
www.recaffeinated.com
Hi John,
You are going about this the right way.
Store the PDFs outside the www root, but still give the user ISR_<computer
name> permission to read the files ( or whatever user your site is running
under). Once the web user has authenticated, your script can read the PDF
into memory then stream the file to the user. A simple example is below:
<%
set fs=server.createobject("Scripting.filesystem")
set PDFin=fs.opentextfile(pathtoPDF,1,false)
PDF=PDFin.readall
PDFin.close
set PDFin=nothing
Response.contenttype="application/pdf"
resonse.binarywrite StrToBin(PDF)
response.end
function StrToBin(str)
for x=1 to Len(str)
StrToBin=StrToBin & ChrB(Asc(Mid(str,x,1)))
next
end function
%>
You may not need to use the StrToBin function - I can't remember off the top
of my head ;)
Regards
Ian
--
