如何通过ElasticSearch Update API（Painless）拼接对象而非全量替换

Your current script replaces the entire ctx._source object when the target document exists, which is why you're losing all original fields from Document A. To dynamically append all fields from Document B without hardcoding each one (perfect for scenarios with lots of fields), you can use a Painless script that iterates over the key-value pairs in your parameter and adds them to the existing source.

Step-by-Step Fix #

1. Replace your script logic with a loop that iterates over every entry in your param1 parameter. This will add each field from Document B to Document A's source without overwriting existing fields (if a field with the same name exists in both, it will be updated to the value from Document B—you can adjust this if needed, see note below).
2. Keep the rest of your update request intact (the upsert clause is still useful for handling cases where the document doesn't exist yet).

Updated Request Body #

{"update":{"_index":"test-index","_id":"121423"}}
{"script": {
  "source": "for (entry in params.param1.entrySet()) { ctx._source[entry.getKey()] = entry.getValue(); }",
  "lang":"painless",
  "params":{"param1":{"city": "New York", "job": "programmer"}}
},
"upsert":{"id":121423,"name": "Sample Name","timestamp": "2020-10-01T00:12:00","age":24}}

Key Details #

• The entrySet() method lets us iterate over all key-value pairs in param1. For each pair, we add the key to ctx._source and set its value to the corresponding value from Document B.
• If you want to avoid overwriting existing fields (only add fields that don't already exist in Document A), modify the script to include a check:

  for (entry in params.param1.entrySet()) {
    if (!ctx._source.containsKey(entry.getKey())) {
      ctx._source[entry.getKey()] = entry.getValue();
    }
  }
  

• This approach is scalable—no matter how many fields are in Document B, you won't need to update the script.

Result After Update #

Once you run this request, Document A will be updated to exactly what you're expecting:

{"id": 121423, "name": "Sample Name", "timestamp": "2020-10-01T00:12:00", "age": 24, "city": "New York", "job": "programmer"}

内容的提问来源于stack exchange，提问作者Augusto