Xen devel - Nov 2007 - [PATCH] [Xend] host.get_log() : clean log of non-printable characters

When retrieving the log via host.get_log() the python parser on the
receiving side gets upset about non-printable characters ("\b"). Those
stem from libxc/xc_domain_restore:xc_domain_restore().

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>




_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

Stefan Berger writes ("[Xen-devel] [PATCH] [Xend] host.get_log() : clean
log of non-printable characters"):
> When retrieving the log via host.get_log() the python parser on the
> receiving side gets upset about non-printable characters ("\b").
Those
> stem from libxc/xc_domain_restore:xc_domain_restore().
It is a shame that we are forced into writing a lossy log retrieval
method by braindamage in XMLRPC and XML 1.0.  Perhaps we should in
future think about a get_log_lossless function which uses a binary
encoding.  It''d have to be base64 :-/.
> -        return xen_api_success(log_buffer)
> +        i = 0
> +        res = ""
> +        while i < len(log_buffer):
> +            c = ord(log_buffer[i])
> +            if (c < 32 or c > 126) and (c < 10 or c > 13):
> +                res += " "
> +            else:
> +                res += log_buffer[i]
> +            i += 1
> +        return xen_api_success(res)
This is a strange way of doing things and will be quite slow.
It''s also wrong in that it replaces tabs.

In theory it would be best to try to map away all Unicode characters
which are not in XML 1.0''s Char.  However, this would involve
explicitly interpreting the logfile as UTF-8 and it''s not clear to me
that it always is.  If it isn''t, it''s probably better to let
the
caller get whatever the logfile byte string is and hope they don''t
choke - at least until we know under what circumstances this arises.

So it''s better just to map away the character we know is causing
problems.  \r will be OK because it''s allowed in XML as an encoding of
newline, which will do.  I''ve chosen below to replace \f as well since
I suspect they may appear at some point.

Ian.

diff -r ba69fe2dce91 tools/python/xen/xend/XendAPI.py
--- a/tools/python/xen/xend/XendAPI.py	Tue Nov 13 20:13:50 2007 +0000
+++ b/tools/python/xen/xend/XendAPI.py	Thu Nov 15 14:32:33 2007 +0000
@@ -994,6 +994,8 @@ class XendAPI(object):
     def host_get_log(self, session, host_ref):
         log_file = open(XendLogging.getLogFilename())
         log_buffer = log_file.read()
+        log_buffer = log_buffer.replace(''\b'',''
'')
+        log_buffer =
log_buffer.replace(''\f'',''\n'')
         return xen_api_success(log_buffer)
 
     def host_send_debug_keys(self, _, host_ref, keys):

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

xen-devel-bounces@lists.xensource.com wrote on 11/15/2007 09:38:37 AM:
> Stefan Berger writes ("[Xen-devel] [PATCH] [Xend] host.get_log() : 
> clean log of non-printable characters"):
> > When retrieving the log via host.get_log() the python parser on the
> > receiving side gets upset about non-printable characters
("\b"). Those
> > stem from libxc/xc_domain_restore:xc_domain_restore().
> 
> It is a shame that we are forced into writing a lossy log retrieval
> method by braindamage in XMLRPC and XML 1.0.  Perhaps we should in
> future think about a get_log_lossless function which uses a binary
> encoding.  It''d have to be base64 :-/.
> 
> > -        return xen_api_success(log_buffer)
> > +        i = 0
> > +        res = ""
> > +        while i < len(log_buffer):
> > +            c = ord(log_buffer[i])
> > +            if (c < 32 or c > 126) and (c < 10 or c >
13):
> > +                res += " "
> > +            else:
> > +                res += log_buffer[i]
> > +            i += 1
> > +        return xen_api_success(res)
> 
> This is a strange way of doing things and will be quite slow.
I agree. log_buffer[i] = '' '' unfortunately does not wokr with
python. The
solution you show below is 
probably the right one.

   Stefan
> It''s also wrong in that it replaces tabs.
> 
> In theory it would be best to try to map away all Unicode characters
> which are not in XML 1.0''s Char.  However, this would involve
> explicitly interpreting the logfile as UTF-8 and it''s not clear to
me
> that it always is.  If it isn''t, it''s probably better to
let the
> caller get whatever the logfile byte string is and hope they don''t
> choke - at least until we know under what circumstances this arises.
> 
> So it''s better just to map away the character we know is causing
> problems.  \r will be OK because it''s allowed in XML as an
encoding of
> newline, which will do.  I''ve chosen below to replace \f as well
since
> I suspect they may appear at some point.
> 
> Ian.
> 
> diff -r ba69fe2dce91 tools/python/xen/xend/XendAPI.py
> --- a/tools/python/xen/xend/XendAPI.py   Tue Nov 13 20:13:50 2007 +0000
> +++ b/tools/python/xen/xend/XendAPI.py   Thu Nov 15 14:32:33 2007 +0000
> @@ -994,6 +994,8 @@ class XendAPI(object):
>      def host_get_log(self, session, host_ref):
>          log_file = open(XendLogging.getLogFilename())
>          log_buffer = log_file.read()
> +        log_buffer = log_buffer.replace(''\b'',''
'')
> +        log_buffer =
log_buffer.replace(''\f'',''\n'')
>          return xen_api_success(log_buffer)
> 
>      def host_send_debug_keys(self, _, host_ref, keys):
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel