OpenSearch/docs/painless/painless-contexts/painless-watcher-transform-...

160 lines
4.7 KiB
Plaintext

[[painless-watcher-transform-context]]
=== Watcher transform context
Use a Painless script as a {xpack-ref}/transform-script.html[watch transform]
to transform a payload into a new payload for further use in the watch.
Transform scripts return an Object value of the new payload.
include::painless-watcher-context-variables.asciidoc[]
*Return*
`Object`::
The new payload.
*API*
The standard <<painless-api-reference, Painless API>> is available.
*Example*
[source,js]
----
POST _watcher/watch/_execute
{
"watch" : {
"trigger" : { "schedule" : { "interval" : "24h" } },
"input" : {
"search" : {
"request" : {
"indices" : [ "seats" ],
"body" : {
"query" : { "term": { "sold": "true"} },
"aggs" : {
"theatres" : {
"terms" : { "field" : "play" },
"aggs" : {
"money" : {
"sum": { "field" : "cost" }
}
}
}
}
}
}
}
},
"transform" : {
"script":
"""
return [
'money_makers': ctx.payload.aggregations.theatres.buckets.stream() <1>
.filter(t -> { <2>
return t.money.value > 50000
})
.map(t -> { <3>
return ['play': t.key, 'total_value': t.money.value ]
}).collect(Collectors.toList()), <4>
'duds' : ctx.payload.aggregations.theatres.buckets.stream() <5>
.filter(t -> {
return t.money.value < 15000
})
.map(t -> {
return ['play': t.key, 'total_value': t.money.value ]
}).collect(Collectors.toList())
]
"""
},
"actions" : {
"my_log" : {
"logging" : {
"text" : "The output of the payload was transformed to {{ctx.payload}}"
}
}
}
}
}
----
// CONSOLE
// TEST[skip: requires setup from other pages]
<1> The Java Stream API is used in the transform. This API allows manipulation of
the elements of the list in a pipeline.
<2> The stream filter removes items that do not meet the filter criteria.
<3> The stream map transforms each element into a new object.
<4> The collector reduces the stream to a `java.util.List`.
<5> This is done again for the second set of values in the transform.
The following action transform changes each value in the mod_log action into a `String`.
This transform does not change the values in the unmod_log action.
[source,js]
----
POST _watcher/watch/_execute
{
"watch" : {
"trigger" : { "schedule" : { "interval" : "24h" } },
"input" : {
"search" : {
"request" : {
"indices" : [ "seats" ],
"body" : {
"query" : {
"term": { "sold": "true"}
},
"aggs" : {
"theatres" : {
"terms" : { "field" : "play" },
"aggs" : {
"money" : {
"sum": { "field" : "cost" }
}
}
}
}
}
}
}
},
"actions" : {
"mod_log" : {
"transform": { <1>
"script" :
"""
def formatter = NumberFormat.getCurrencyInstance();
return [
'msg': ctx.payload.aggregations.theatres.buckets.stream()
.map(t-> formatter.format(t.money.value) + ' for the play ' + t.key)
.collect(Collectors.joining(", "))
]
"""
},
"logging" : {
"text" : "The output of the payload was transformed to: {{ctx.payload.msg}}"
}
},
"unmod_log" : { <2>
"logging" : {
"text" : "The output of the payload was not transformed and this value should not exist: {{ctx.payload.msg}}"
}
}
}
}
}
----
// CONSOLE
// TEST[skip: requires setup from other pages]
This example uses the streaming API in a very similar manner. The differences below are
subtle and worth calling out.
<1> The location of the transform is no longer at the top level, but is within
an individual action.
<2> A second action that does not transform the payload is given for reference.
The following example shows scripted watch and action transforms within the
context of a complete watch. This watch also uses a scripted
<<painless-watcher-condition-context, condition>>.
include::painless-watcher-context-example.asciidoc[]