Elasticsearch instance with GP profiles data

Running scripts/start will start an Elasticsearch instance, loading data from data/input/gp-data-merged.json into an index named profiles localhost:9200/profiles.

When running the container in this way, the port on which ES is exposed can be overridden by setting the environment variable ES_PORT (e.g. ES_PORT=9201 ./scripts/start).

A basic set of tests can be run from scripts/test

The data in a running instance can be queried using the ElasticSearch REST API and a command line tool like curl.

Full Text Search Example

curl -s -XPOST http://localhost:9200/profiles/_search?pretty -d '
{
  "size": 10,
  "explain": false,
  "_source": [ "name", "alternativeName", "address", "doctors" ],
  "query" : {
    "bool": {
      "must": {
        "multi_match": {
          "query": "Beech House Surgery",
          "fields": [ "name^2", "alternativeName" ],
          "operator": "and"
        }
      },
      "should":  [
        { "match_phrase": {
          "name": {
            "query": "Beech House Surgery" ,
            "boost": 2
          }
        } }
      ]
    }
  },
  "highlight" : {
    "fields" : {
      "name" : {},
      "alternativeName" : {}
    }
  }
}'

Geo-Location Search Example

curl -s -XPOST http://localhost:9200/profiles/_search?pretty -d '
{
  "query": {
    "bool": {
      "must" : {
        "match_all" : {}
      },
      "filter": {
        "geo_distance": {
          "distance": "1km", 
          "location.coordinates": { 
            "lon": -1.46519099452929,
            "lat": 54.0095586395326
          }
        }
      }
    }
  },
  "sort": [
    {
      "_geo_distance": {
        "location.coordinates": {
          "lon": -1.46519099452929,
          "lat": 54.0095586395326
        },
        "order":         "asc",
        "unit":          "km",
        "distance_type": "plane"
      }
    }
  ]
}

Other useful endpoints on the API are:

• Query index settings:
  • curl http://localhost:9200/profiles/_settings?pretty
• Query index mappings:
  • curl http://localhost:9200/profiles/_mappings?pretty
• Interact with an analyzer:
• curl http://localhost:9200/profiles/_analze?pretty -d '{ "analyzer" : "gp_name_analyzer", "text" : "dr andrew jones" }'

See here for more detail on querying.

Optional

An alternative to using curl for ES config, querying and also for visualisation is kibana. For the latest version go here or your OS package manager. You can also find a comprehensive intro here.

Pre-requisites

• Set vm.max_map_count=262144 on the Docker instance to support ES requirement

Architecture Decision Records

This repo uses Architecture Decision Records to record architectural decisions for this project.

They are stored in doc/adr.