Here's the code for my download method which streams from the DB:
[code]
@turbogears.expose()
def download(self, id, **kw):
article = NewsPost.get(id)
if article.mimetype:
cherrypy.response.headerMap["Content-Type"] = article.mimetype
else:
cherrypy.response.headerMap["Content-Type"] = "application/x-download"
cherrypy.response.headerMap["Content-Disposition"] =
"attachment;filename=%s" % article.filename.replace(" ", "_")
cherrypy.response.headerMap["Content-Length"] = "%d" %
len(article.attachment)
return cherrypy.lib.cptools.fileGenerator(StringIO(article.attachment))
[/code]
Hope this helps
Lee
On 12/8/05, Wavy Davy <[EMAIL PROTECTED]> wrote:
>
> On 08/12/05, Sylvain Hellegouarch <[EMAIL PROTECTED]> wrote:
> > Until CP 2.1, CP defauts its content type to text/html, you will be able to
> > change that globally from the config file though a line like:
> >
> > server.defaultContentType = 'application/xml'
>
> Excellent, ready to switch from HTML to XHTML and the flick of a
> config option :) Combined with kids outputformat, you can write in
> validated xhtml and serve it as whatever the hell your customer wants
> :)
>
> > There will be also a new decorator called headers to set HTTP headers per
> > page
> > handler (exposed callable) like this:
> >
> > @cherrypy.expose
> > @cputil.headers([('Content-Type', 'application/atom+xml'),
> > ('Content-Language',
> > 'fr')])
> > def atom(self):
> > return atom doc
>
> Good. But is there a reason behind the cputil packaging? Or is this
> just an example of how it could go? I would prefer cherrypy.headers(),
> for consistency. Everything to do with controlling cherrypy is behind
> the cherrypy namespace.
>
> --
> wavy davy
>
> "True religion confronts earth with heaven
> and brings eternity to bear on time"
> - A. W. Tozer
>
