Elasticsearch highlight on array, how to return entire array

Question

I'm trying to do some highlighting in Elasticsearch and am having difficulty getting the output I'd like for multi-value fields. Here's what I'm doing now:

{ "query" : { "match": { "nameSet": "test" } }, 
  "highlight" : { "fields" : { "*": {"number_of_fragments": 0 } } }
}

This gives me (omitted unnecessary fields):

"hits" : [ {
    ...
    "_source" : {
        "nameSet" : ["TEST", "NAME"]
    },
    "highlight" : {
       "nameSet" : [ "<em>TEST</em>" ]
    }
 }, 
 ...

What I would like to have is the full array, and not just the item that matched. In this example, I would like "TEST" to be emphasized and "NAME" to be present but not emphasized.

"hits" : [ {
    ...
    "_source" : {
        "nameSet" : ["TEST", "NAME"]
    },
    "highlight" : {
       "nameSet" : [ "<em>TEST</em>", "NAME" ] 
    }
 }, 
 ...

Any way to do this purely in ES?

Thanks.

Answer 1

There are currently two open threads about this, the first from over a decade ago:

Highlighting array field - Also return non-matching entries

Use path to indicate highlighted index of array items

The general approach for now seems to be stripping the highlight tags, doing a comparison to the array elements, and then replacing the matching elements with the highlight. Obviously this is not ideal, but it should work in all cases, if done carefully.

// We can use this to generate a function to strip whichever highlights
// we chose to use
function stripHighlights(open, close, highlight) {
    return function(highlight) {
        return highlight.replaceAll(open, "").replaceAll(close, "")
    }
}

// In this case, we use "em" because they are the ones from the example
let stripEmTags = stripHighlights("<em>", "</em>");

// This does the actual work of replacing the source array element
// with the highlighted one, if it was highlighted
function highlightedArrayField(sourceField, highlightField) {
    let strippedHighlights = highlightField.map(stripEmTags);

    return sourceField.map(elem => {
        let matchedIdx = strippedHighlights.indexOf(elem);
        if(matchedIdx == -1) {
            return elem;
        }

        return highlightField[matchedIdx];
    });
}

// This is the general function we can call to go through the hits and
// determine their replaced source values.
// We could use an Object.assign to update the properties of the hit if needed.
function highlightedHit(hit) {
    return Object.keys(hit.highlight).reduce((agg, field) => {
        let highlightField = hit.highlight[field];
        let sourceField = hit._source[field];

        if(Array.isArray(sourceField)) {
            agg[field] = highlightedArrayField(sourceField, highlightField);
        } else {
            agg[field] = highlightField;
        }
        return agg;
    }, {});
}