Thanks James for the answer. Yes, s3cmd reads the files names nicely.
I have would have used the Ruby-based tools if it wasn't for the fact
that they require Ruby 1.8.4 or later, which I am not is able to
upgrade to on this specific server. I think I have located the files
with bad file names (with a carriage return at the end?), which seem
to be files generated by a "bad" PHP script. Thanks!
/Clas
On Jul 30, 7:55 am, James Murty <
jamu...@gmail.com> wrote:
> Hi Clas,
>
> I'm afraid you have hit an old and unresolved issue with S3: it is possible
> to store objects in S3 with names that cannot be properly express in the XML
> listing response from the service.
>
> There is a long and ultimately fruitless discussion about the problem here:
http://developer.amazonwebservices.com/connect/thread.jspa?threadID=1...
>
> JetS3t does some XML pre-parsing to avoid some problems like this, but it
> does not handle content such as "".
>
> The only way around this problem may be to try using a ruby-based tool to
> list the objects in this bucket so you can identify and remove the offending
> object. I recall that the ruby XML parser is much less strict than most,
> which actually makes it better in this case. I would recommend the s3cmd
> tool available here:
http://s3sync.net/wiki
>
> Aside from dealing with this troublesome object name, you should find out
> how this item was stored in S3 in the first place. If it was uploaded by
> Synchronize, you should check the name of the file on your system to ensure
> it doesn't contain odd control characters in it, otherwise the same issue
> will recur.
>
> I hope this helps somewhat. It's a tricky problem, unfortunately.
>
> James
>
> ---
http://www.jamesmurty.com