[Zope] Generating non-existent documents on the fly?

[Kent Polk]

On 20 Oct 1999 14:25:01 -0500, Stefan Hoffmeister wrote:
>
>Is there a way to generate non-existent documents on the fly - some kind
>of "rewriting" the target URL?
>
>Example site structure
>
>  /root (http://localhost/)
>     /content
>        layout_template
>        some_txt
>
>A user queries
>
>  http://localhost/content/some.html
>
>Obviously, "some.html" does not exist. This is where some mechanism (the
>one I am trying to find <g>) kicks in. It will analyze the query string,
>detect that there is "some_txt" and "layout_template". Based on that it
>will generate "some.html" on the fly.
>
>Any ideas?

There are several ways of going about this, depending on what
you are doing. I don't know of a way to do it without using
Python to wrap the object in a class which knows what to do
with the rest of the URL.

For example, say you are extracting a document from a SQL
database but you want the client to save the data with a
specific filename (stored in the database?). You can set
the headers via something like:
<!--#call "RESPONSE.setHeader('Content-Disposition', 
        'attachment; filename='+filename)"-->

But 'filename' isn't standard (AFAIK) and isn't implement in all
browsers. However, if you include the filename as the last component
of the URL, browsers can (usually) deal with that. So what you need
is some sort of __getitem__ or __bobo_traverse which knows how to
deliver the correct file data given a query and the 'filename'.

How you invoke the __bobo_traversee__ is the hard part. :^) If
you are doing a SQL query with a SQL Method, you can use the
Advanced Options to specify the Class name which will wrap the
SQL method results. That class would include a __bobo_traverse__
method which takes the key (the filename part of the URL) and
returns the result.  Note that you might not even use the key
because you might have all the info needed already. The 
__bobo_traverse__ method could set the mime headers, etc
and ship the data out via a str() method or such.

In the above example, the method preceeds the virtual part of
the URL, You can reverse the order and have many, many virtual
URL components preceed the method, again, using __bobo_traverse__.
I do this with my tabular data mining product I am working on.