Libmus
======

https://data.libmus.com is the primary linked data and sparql endpoint

https://dev.libmus.com is the web application

RDF
---

Stored in ``/Volumes/Data/libmus``

4store: ``libmus_A`` is the main source, on port ``7005``

Doesn't look like the data copied over from ``hermes`` will run :(

To import:

.. code::

  curl -X PUT --user username:password --data-binary @libmus-test.ttl https://data.libmus.com/sparql/models/libmus-test.ttl


sameAs
------

https://data.libmus.com/services/sameAs/

Database ``libmus`` copied over from old server ``hermes``

Seems to be working OK! :)

Search
------

See :doc:`elasticsearch` ... we now have fancy new ElasticSearch7 and LibmusSearch7 services

Unfortunately data can't be copied over from ``hermes``

To import:

.. code:: 

  curl -X PUT --user username:password --data-binary @libmus-test.ttl https://data.libmus.com/services/search/graph/libmus-test.ttl


Notes from Hugh (on the old setup)
----------------------------------

The raw data wasn’t kept on Bytemark. It would have burnt disk and CPU resources. So I kept it on another machine. Oh wait, it’s the same disk that we now use on M2 :-)

See ``/Volumes/Data/libmus``

Some background you don’t really need to know, I hope.

* The different sources were treated differently.
* MB & Wikidata had Makefiles detailing things.
* But the make was often embedded in a .sh file, because of littel variations in the more complex scripts for converting the sql dumps into ttl. (!).
* Oh, and the ttl files had to be split up so they would load - this could happen either on my machine or at Bytemark.
* Things were often invoked using rsh.

Anyway, looking at the Makefile in /Volumes/Data/libmus/MusicBrainz

You will see, just before ^send.

``sparql=/var/www/vhosts/data.libmus.com/sparql/files``

Because lack of space at Bytemark meant that we couldn’t use the normal way of asserting ttl, where the 4s-assert or whatever it was took a copy of the data coming in.

Instead we put the ttl where the 4store wold have put it.
And now I need to go and hunt down the command to assert that, which is the actual question you asked :-)

Yeah, the line before sparql=
m=/var/www/vhosts/data.libmus.com/libmus/data/mb
seems to be the hint (all paths are for the old machines)
There is even a tiny README there :-)

Yeah, 
'/Volumes/Seagate 8TB/Backups/seme4/mnt/ssd-2/vhosts/data.libmus.com/libmus/data/mb'
seems to have the goods.
Things like loadAll.sh (& indexAll.sh), which uses load in the Makefile.

So.
I think you can find all the files in sparql/files and then use soemthing similar to that load in the Makefile to load them?
and ^index