Hi Noble,
I started digging into the code to this and came away a bit confused -
apologize being 'quite a newbie' here.
The solrconfig contains both an "Update Handler" and an "Update
Request Handler "
Should I plug in the former or the latter. I am not sure I want to
replace the 'high performance' updateHandler.
If it is the "Update Request Handler" level then I saw that the
current XMLRequestHandler iterates through a list of content streams
and calls the processor for each of them.
So when you mentioned the ContentStreamBase below, I was wondering if
the advise was for me to add such 'content stream' that gives my
transformed XML to the XmlUpdateRequestHandler and then the logic
flows as before inside XMLRequestHandler. (not sure where to plugin
such a content stream).
I am sorry if these are very trivial questions, but I was unsure of
which part of the architecture to plug-in without 'reinventing the
wheel'. While I want to plug-in on the server side, don't want to do
that redoing/reinventing too much code. I want to plug in at the
highest level so that future enhancements will work as-is for us?
Regards
Guna
On Jan 24, 2009, at 12:42 AM, Noble Paul നോബിള്
नोब्ळ् wrote:
That does not look like a great option. DIH looks like an overkill for
this usecase.
You can write a simple UpdateHandler to do that .
All that you need to do is to extent ContentStreamHandlerBase and
register it as an UpdateHandler
On Sat, Jan 24, 2009 at 12:34 PM, Shalin Shekhar Mangar
<shalinman...@gmail.com> wrote:
There's another option. Using DIH with Solrj. Take a look at:
https://issues.apache.org/jira/browse/SOLR-853
There's a patch there but it hasn't been updated to trunk. A
contribution
would be most welcome.
On Sat, Jan 24, 2009 at 3:11 AM, Gunaranjan Chandraraju <
chandrar...@apple.com> wrote:
Hi
I had earlier described my requirement of needing to 'post XMLs as-
is' to
SOLR and have it handled just as the DIH would do on import using
the
mapping in data-config.xml. I got multiple answers for the 'post
approach'
- the top two being
- Use SOLR CELL
- Use SOLRJ
In general I would like to keep all the 'data conversion' inside
the SOLR
powered search system rather than having clients do the XSL and
transforming
the XML before sending them (CELL approach).
My question is? How should I design this
- Tomcat Servlet that provides this 'post' endpoint. Accepts the
XML over
HTTP, transforms it and calls SOLRJ to update. This is the same
TOMCAT that
houses SOLR.
- SOLR Handler (Is this the right way?)
- Take this a step further and implement it as an extension to
DIH - a
handler that will refer to DIH data-config xml and use the same
transformation. This way I can invoke an import for 'batched
files' or do a
'post 'for the same XML with the same data-config mapping being
applied.
Maybe it can be a separate handler that just refers to the same
data-config.xml and not necessarily bundled with DIH handler code.
Looking for some advise. If the DIH extension is the way to go
then I
would be happy to extend it and contribute that back to SOLR.
Regards,
Guna
--
Regards,
Shalin Shekhar Mangar.
--
--Noble Paul
