To search something in Elasticsearch, the request looks like this:
POST https://es-cluster:9200/index-name/_search
The query definition is sent as JSON in the body of the request. Here’s an example search body:
| { | |
| "query": { | |
| "bool": { | |
| "filter": [ | |
| { "term": { "field1": "foo" } }, | |
| { "term": { "field2": "bar" } } | |
| ] | |
| } | |
| } | |
| } |
We have found it’s quite useful to put queries together using the improved object literal syntax of more recent versions of JavaScript:
| const secondThing = { term: { field2: 'bar' } }; | |
| const firstThing = { term: { field1: 'foo' } }; | |
| const filter = [ firstThing, secondThing ]; | |
| const bool = { filter }; | |
| const query = { bool }; | |
| const body = { query }; | |
| const params = { | |
| index: 'your-index', | |
| body | |
| }; | |
| const result = await es.search(params); |
This script shows how we might build the search body for the previous request.
In the script, we’re building the query line-by-line and making use of ES2015’s shorthand property names to link it all together.
I usually find I work backwards from the es.search() statement. First are the params – defining the index and options for the request. Then onto the request body, which itself unravels into its constituent parts.
This technique can be used in Node JS or modern browsers – wherever ES2015 is supported!
