[Zope] PDF from external method

Jim Washington jwashin@vt.edu
Thu, 25 Oct 2001 11:52:45 -0400

Jim Penny wrote:

> On Thu, Oct 25, 2001 at 09:23:59AM -0400, Jim Washington wrote:
>>Juergen R. Plasser / HEXAGON wrote:
>>>Ok, sorry, this was a python script error.
>>>Another question: Reportlab examples create files, how can I "RESPONSE" 
>>>them to Zope?
>>What I am using:
>>-- Jim Washington
> This is very nice.
> Yet another way is to simply build a HTML file pointing to the
> .pdf file generated.
> I.e.
> def build_pdf(...):
>    ...
>    # omitted code builds pdf file in variable pdf_filename
>    # referenced as pdf_url
>    return "<a href="%s">%s</a>" % (pdf_url, pdf_filename)
> Advantages:
>   Allows .pdf to be served by static webserver
> Disadvantages:
>   Requires extra click to get to pdf.
>   Requires a policy to reap stale pdf's
> Neutral:
>   Caching generated pdf's may (or may not) be a big win.  It depends
>     on the nature of your task.  I suspect, psychobabblically speaking,
>     that caching will occur more naturally to people who are treating
>     pdf's as semi-static content by delivering a file name, than to 
>     those who purely consider a pdf to be something to be delivered
>     as a result of a method invocation.
> Note:  The disadvantages may not be as large as you think.  Many
> people are disconcerted by long load-times and/or sudden pop-up
> windows.  This splits download/viewer startup time away from
> data gathering/pdf generation time and will typically
> be perceived as more responsive; despite stop watch time showing
> that it is far slower due to the necessity of the extra click.  
> (That is, people are often happier with 2 events of duration t/2 
> than one event of duration t, especially if they have a chance to 
> "walk away" in the middle.)  Thus, for many people, the extra click
> may be perceived as an advantage!
> The policy for reaping pdf's is actually neutral.  Note that Jim
> Washington's method left the pdf file on the hard disk somewhere.
> If you regard the external method as generating files rather than 
> byte-streams, the need for a reaping process simply leaps out at you.
> Jim Penny

Thanks, Jim.  I may use some of this.  I am currently working on the 
user interface for a multitude of reports.

A small caveat to the above.  The PDF's I am generating are small, 
150-200k, and not particularly high-traffic.

I create the PDFs entirely in memory, so I do not need to reap:

from cStringIO import StringIO
pdffile = StringIO()

doc = BaseDocTemplate(pdffile, 
pageTemplates=([coverTemplate,secondPageTemplate, mainTemplate,]))
doclist = []

...use platypus...


then write to RESPONSE as in my note above.

-- Jim Washington