How to perform elastic search _update_by_query using painless script - for complex condition

Question

Can you suggest how to update documents (with a script - i guess painless) that based on condition fields?

its purpose is to add/or remove values from the document

so if I have those input documents:

doc //1st
{
   "Tags":["foo"],
   "flag":"true"
}


doc //2nd
{
   "flag":"true"
}

doc //3rd
{
   "Tags": ["goo"],
   "flag":"false"
}

And I want to perform something like this:

Update all documents that have "flag=true" with:

Added tags: "me", "one"
Deleted tags: "goo","foo"

so expected result should be something like:

doc //1st
{
   "Tags":["me","one"],
   "flag":"true"
}


doc //2nd
{
   "Tags":["me","one"],
   "flag":"true"
}

doc //3rd
{
   "Tags": ["goo"],
   "flag":"false"
}

score 3 · Accepted Answer · answered May 27 '18 at 18:32

Create mapping:

PUT documents
{
    "mappings": {
        "document": {
            "properties": {
                "tags": {
                    "type": "keyword",
                    "index": "not_analyzed"
                },
                "flag": {
                    "type": "boolean"
                }
            }
        }
    }
}

Insert first doc:

PUT documents/document/1
{
    "tags":["foo"],
    "flag": true
}

Insert second doc (keep in mind that for empty tags I specified empty tags array because if you don't have field at all you will need to check in script does field exists):

PUT documents/document/2
{
    "tags": [],
    "flag": true
}

Add third doc:

PUT documents/document/3
{
    "tags": ["goo"],
    "flag": false
}

And then run _update_by_query which has two arrays as params one for elements to add and one for elements to remove:

POST documents/_update_by_query 
{
    "script": {
        "inline": "for(int i = 0; i < params.add_tags.size(); i++) { if(!ctx._source.tags.contains(params.add_tags[i].value)) { ctx._source.tags.add(params.add_tags[i].value)}} for(int i = 0; i < params.remove_tags.size(); i++) { if(ctx._source.tags.contains(params.remove_tags[i].value)){ctx._source.tags.removeAll(Collections.singleton(params.remove_tags[i].value))}}",
        "params": {
            "add_tags": [
                {"value": "me"},
                {"value": "one"}
            ],
            "remove_tags": [
                {"value": "goo"},
                {"value": "foo"}
            ]
        }
    },
    "query": {
        "bool": {
            "must": [
                {"term": {"flag": true}}
            ]
        }
    }
}

If you then do following search:

GET documents/_search

you will get following result (which I think is what you want):

{
    "took": 1,
    "timed_out": false,
    "_shards": {
        "total": 5,
        "successful": 5,
        "skipped": 0,
        "failed": 0
    },
    "hits": {
        "total": 3,
        "max_score": 1,
        "hits": [{
                "_index": "documents",
                "_type": "document",
                "_id": "2",
                "_score": 1,
                "_source": {
                    "flag": true,
                    "tags": [
                        "me",
                        "one"
                    ]
                }
            },
            {
                "_index": "documents",
                "_type": "document",
                "_id": "1",
                "_score": 1,
                "_source": {
                    "flag": true,
                    "tags": [
                        "me",
                        "one"
                    ]
                }
            },
            {
                "_index": "documents",
                "_type": "document",
                "_id": "3",
                "_score": 1,
                "_source": {
                    "tags": [
                        "goo"
                    ],
                    "flag": false
                }
            }
        ]
    }
}

How to perform elastic search _update_by_query using painless script - for complex condition

1 Answers1