Re: FR: Option to terminate page rendering from an Action nicely

Dear Ralf and John,

Thank you both for your replies. I like both proposals, and I find
John's concept elegant. I also agree with Ralf that the handling of
return values of Actions should be documented.

PS. A bit off-topic in this thread, but John's email reminded me that I
do not really like the current implementation of ExportCSVAction in
cgi/actions.py (or the index templates in the sample trackers, it
depends on your point of view): In my opinion, ExportCSVAction and the
index templates _should_ somehow share the code which renders the
"cells of the table" of the index view: We should export precisely
what the user sees. However, instead of code sharing, there are many
hardcoded stuff in ExportCSVAction (see the 'name' vs. 'realname'
handling etc.).

Regards,
Gábor

> Hi Ralf and Nagy:
> 
> In message <[email protected]>,
> Ralf Schlatterbeck writes:
> >Hi Nagy, see below for my fix, it probably is a good idea to put that
> >into some lib shipped with roundup and document the behaviour.  
> 
> I just implmented
> 
>     https://wiki.roundup-tracker.org/QueryResultPieCharts
> 
> using pygal.
> 
> I didn't send a header or write anything at all to the client from the
> handler. I returned the svg image at the end of the handler rather
> than returning '\n'. This called write_html which delivered it
> (compressed with brotli) to the client.
> 
> I think this path should work for any bytestream provided the
> content-type header is properly set.  It may be a side effect of the
> work I did to add compression in the current development code, but I
> don't think so.
> 
> I think the only reason to write directly from the handler would be if
> the returned item is too large to build all at one. In the csv case,
> it looks like one line of the csv output is written as it iterates
> over all the data.
> 
> If you are building the whole returned file in memory, make sure to
> turn it into bytes and try sending is a the handler return value.
> 
> -- rouilj
> 
> >On Fri, Aug 06, 2021 at 09:12:50PM +0200, Nagy Gabor wrote:  
> >> 
> >> Based on ExportCSVAction, I wrote an ExportXLSAction using
> >> xslxwriter. Everything worked fine on Linux systems, but MS Office
> >> always reported corrupted xlsx file, which seemed OK after
> >> applying Office's repair. Mystery...
> >> 
> >> After hours of debugging, I found the bug:
> >> 
> >> First I realized that the problem is that my generated .xlsx file
> >> is correct, but it is served with a '\n' character added to the
> >> end of the file. Analogously to
> >> cgi.actions.ExportCSVAction.handle(), the last line of my handle
> >> method was also   return '\n'. We are getting closer: Yes, the
> >> returned string is appended to the served content, this is
> >> verified for different strings. At this point I believed that I am
> >> done, but neither return '' nor return None worked. In these cases
> >> the issue.index template file is rendered into the xlsx binary.
> >> Hmmm. So I have to looked into Roundup's code deeper, and I found
> >> the relevant code in cgi/client.py (lines 685-696, inner_main()
> >> method of Client class):
> >> 
> >>                 if csrf_ok:
> >>                     # csrf checks pass. Run actions etc.
> >>                     # possibly handle a form submit action (may
> >> change # self.classname and self.template, and may also
> >>                     # append error/ok_messages)
> >>                     html = self.handle_action()
> >>                 else:
> >>                     html = None
> >> 
> >>                 if html:
> >>                     self.write_html(html)
> >>                     return
> >>                 # now render the page
> >>                 # we don't want clients caching our dynamic pages
> >> 
> >> This means that if the action's handle() returns anything which
> >> evaluates to True in the "if html" condition, then it is added to
> >> the page, and the rendering is done; otherwise the rendering goes
> >> on.
> >> 
> >> My feature request is to add a nice possibility for the Action to
> >> terminate the page rendering. (Maybe by raising an exception,
> >> which is handled by the except part of inner_main.) I've just
> >> added an "if html == True:
> >>      return"
> >> test before the "if html:" test, and my Action returns True. (This
> >> is an ugly solution, but it should work, because
> >> self.write_html(True) induces an error, so probably no other
> >> Action returns True, because with the old code, that would lead to
> >> an exception.)  
> >
> >I'm using the following class in my actions:
> >
> >class True_Value (type ("")):
> >    """ A class that evaluates to True but can return a zero-length
> > string. We use this as return value from a handle routine where the
> >        output happend to the file descriptor
> >    """
> >    def __bool__ (self):
> >        return True
> >
> >Note that for Python2 you want to add
> >    __nonzero__ = __bool__
> >
> >I'm returning an instance of that class which appends an empty
> >string.
> >
> >Ralf
> >-- 
> >Dr. Ralf Schlatterbeck                  Tel:   +43/2243/26465-16
> >Open Source Consulting                  www:   www.runtux.com
> >Reichergasse 131, A-3411 Weidling       email: [email protected]
> >
> >
> >_______________________________________________
> >Roundup-users mailing list
> >[email protected]
> >https://lists.sourceforge.net/lists/listinfo/roundup-users  
> 
> --
> 				-- rouilj
> John Rouillard
> ===========================================================================
> My employers don't acknowledge my existence much less my opinions.
> 
> 
> _______________________________________________
> Roundup-users mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/roundup-users


_______________________________________________
Roundup-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/roundup-users