matchID-backend integration

[rhanka]

matchID-backend could provide a proxy for matchID-validation :

• give acces to raw and filtered datasets
• provide elements for matchID-validation configuration for diplay and validation

matchID-backend could provide the folowing methods/APIs:

• datasets:
  • transparent ES API for search and msearch on top of dataset API (/matchID/api/v0/datasets/rnipp/_search? ...)
  • dataset-dependant displays configuration : columns, view, score, randomID
  • dataset connector (equivalent to elasticsearch.json)

only "lang.json" which is not dataset-dependant should be kept as global configuration, which could still be provided by the backend.

A first step could be a "blind"-driven matchID configuration driven server side from matchID-backend to demonstrate this possibility (=> direct docker integration)

A second step would require some enhancements :

• add a class for configuring connexion (or not) to a backend (could be an evolution of the "elasticsearch.json" conf, splitted into "connector.json" and "dataset.json"
• add a "dataset" selector

[rhanka]

A first integration has been done you just have to :

PUT {machid-backend}/matchID/api/v0/datasets/<dataset>/validation`

then matchID-backend writes elasticsearch.json (etc.) to a matchIdConfig directory. The docker-compose from matchID-backend share from backend to frond end so the work is ok for experimentation. The web-app from matchID-backend demonstrate a web integration of matchID-validation in the dataprep process.

A deeper integration should now be developped, where a matchID-validation could get and update configuration from an API - this one could be

GET {matchid-backend}/matchID/api/v0/conf/validation/elasticsearch.json`

(and so on, or a uniq json).