Using Scripted Conditionals

[spupapi (Mariam Shestapulova)]

Hello,

so I've got an index called Hamlet. And in Hamlet there is a field "Speaker" with value of the name of the speaker. So it looks like this:

{
"speaker": "Hamlet"
}

Now I want to make script which will:
Create a script named set_is_hamlet and save it into the cluster state. The script (i) adds a field named is_hamlet to each document, (ii) sets the field to "true" if the document has speaker equals to "HAMLET", (iii) sets the field to "false" otherwise.

How would the script look like?

[Bernt_Rostad]

@spupapi

Create a script named set_is_hamlet

Not sure what you mean by naming a script, a pipeline can be named but an inline script used in an update_by_query has no name. For the sake of simplicity I'll write the script as for use by the latter. But first a comment:

@spupapi

I've got an index called Hamlet.

Index names in Elasticsearch must be lower case (at least up until version 6.8).

So I opened Kibana in my test cluster and created an index named hamlet by adding four small documents of the format you mentioned. E.g:

PUT /hamlet/_doc/1
{
    "speaker": "Ophelia"
}
... etc

The result looks like this:

GET /hamlet/_search
...
  "hits" : {
    "total" : 4,
    "max_score" : 1.0,
    "hits" : [
      {
        "_index" : "hamlet",
        "_type" : "_doc",
        "_id" : "2",
        "_score" : 1.0,
        "_source" : {
          "speaker" : "Hamlet"
        }
      },
      {
        "_index" : "hamlet",
        "_type" : "_doc",
        "_id" : "4",
        "_score" : 1.0,
        "_source" : {
          "speaker" : "HAMLET"
        }
      },
      {
        "_index" : "hamlet",
        "_type" : "_doc",
        "_id" : "1",
        "_score" : 1.0,
        "_source" : {
          "speaker" : "Ophelia"
        }
      },
      {
        "_index" : "hamlet",
        "_type" : "_doc",
        "_id" : "3",
        "_score" : 1.0,
        "_source" : {
          "speaker" : "Laertes"
        }
      }
    ]
  }

The final step is to write a small painless script and use it in an _update_by_query, like this:

POST /hamlet/_update_by_query
{
   "script": {
      "lang": "painless",
      "source": """
         if (ctx._source.speaker == "HAMLET") {
            ctx._source.set_is_hamlet = true
         } else {
            ctx._source.set_is_hamlet = false
         }
               """
    }
 }

This correctly found the documents with speaker "HAMLET" and set its set_is_hamlet to true while the rest were set false (notice that Elasticsearch is case sensitive, so only speaker HAMLET gets set_is_hamlet: true):

GET hamlet/_search
...
  "hits" : {
    "total" : 4,
    "max_score" : 1.0,
    "hits" : [
      {
        "_index" : "hamlet",
        "_type" : "_doc",
        "_id" : "2",
        "_score" : 1.0,
        "_source" : {
          "set_is_hamlet" : false,
          "speaker" : "Hamlet"
        }
      },
      {
        "_index" : "hamlet",
        "_type" : "_doc",
        "_id" : "4",
        "_score" : 1.0,
        "_source" : {
          "set_is_hamlet" : true,
          "speaker" : "HAMLET"
        }
      },
      {
        "_index" : "hamlet",
        "_type" : "_doc",
        "_id" : "1",
        "_score" : 1.0,
        "_source" : {
          "set_is_hamlet" : false,
          "speaker" : "Ophelia"
        }
      },
      {
        "_index" : "hamlet",
        "_type" : "_doc",
        "_id" : "3",
        "_score" : 1.0,
        "_source" : {
          "set_is_hamlet" : false,
          "speaker" : "Laertes"
        }
      }
    ]
  }

I hope this answered your question.

[spupapi (Mariam Shestapulova)]

@Bernt_Rostad

Not sure what you mean by naming a script, a pipeline can be named but an inline script used in an update_by_query has no name.

What I meant was to create a templated script and save it into cluster state, using your code below.

POST _scripts/script_hamlet_example
{
   "script": {
      "lang": "painless",
      "source": """
         if (ctx._source.speaker == "HAMLET") {
            ctx._source.set_is_hamlet = true
         } else {
            ctx._source.set_is_hamlet = false
         }
               """
    }
 }

And yes! Thank you for helping, you made it seem so simple .

[Bernt_Rostad]

@spupapi

What I meant was to create a templated script and save it into cluster state,

Oh, I see. I haven't stored scripts to the cluster before, only used them in pipelines or _update_by_query statements.

@spupapi

Thank you for helping

I'm glad you found your solution