On 14/11/17 17:13, Kieron Taylor wrote:
Hi everyone,
Up to now I've relied on the ruby scripts shipping with Fuseki to load data
onto a server, s-post, s-put etc.
I would prefer to avoid shelling out to these scripts when I need to load data,
but I can't quite figure out what a multipart message body needs to look like
to be accepted. The Ruby script mechanism seems to be described here:
https://www.w3.org/TR/2013/REC-sparql11-http-rdf-update-20130321/#http-post
Sadly the perl libraries that support streamed content can only do so via a
more proper POST body with key:value pairs. I really shouldn't have to write my
own streaming code for this, so can somebody explain what multipart requests
Fuseki can understand? If any?
Kieron
Hi Kieron,
s-post and s-put don't require a multipart request. Just send the RDF
data and set the Content-type header. s-post/s-put guess the
Content-type for you but you can do it directly in a language of your
choice or exec "curl".
Example:
curl -XPOST --header "Content-type: text/turtle" \
--data-binary @DATA.ttl \
http://localhost:3030/ds
However, that may load the file into curl's RAM. (reading from
https://github.com/curl/curl/issues/290)
If you want to do it like an HTML file upload:
curl -F "[email protected];filename=DATA.ttl" http://localhost:3030/ds
which streams a large file which si the multipart form style.
Programmatically, open an HTTP connection with
"Content-type: text/turtle" and push the RDF bytes down it. Fuseki
does not require "Content-length".
Andy
