Elasticsearch嵌套字段阵列聚合脚本

Question

我是ES的新手，所以请忍受我。我在网上到处搜索 +尝试了不同的事情，但无法得到答案。

我有这样的结构映射，

"index_1": { ... 
},
"index_2": { ... 
},
"index_3": {
    "mappings": {
        "dynamic": "strict",
        "properties": { 
...
"keywords": {
                "type": "nested",
                "properties": {
                    "id": {
                        "type": "keyword",
                        "index": false,
                        "ignore_above": 256
                    },
                    "term": {
                        "type": "text",
                        "copy_to": [
                            "keywordsSearchField"
                        ],
                        "term_vector": "with_positions_offsets",
                        "analyzer": "pasc_index_autocomplete_analyzer",
                        "search_analyzer": "pasc_standard_analyzer"
                    },
                    "vocab": {
                        "type": "keyword",
                        "ignore_above": 256,
                        "copy_to": [
                            "keywordsSearchField"
                        ]
                    },
                    "vocabUri": {
                        "type": "keyword",
                        "ignore_above": 256,
                        "copy_to": [
                            "keywordsSearchField"
                        ]
                    }
                }
            },
            "keywordsSearchField": {
                "type": "text",
                "analyzer": "pasc_standard_analyzer"
            },
...
}

所有索引都具有相同的映射。我要做的是计算每个索引中每个文档的嵌套关键字数组大小，然后按类别进行分组，例如： 关键字1-5：500 doc，关键字6-10：1000文档等。

在我发现汇总时无法使用它们之前，我一直在查看Script_fields。这是一个示例，

{
"_source": "*",
"query": {
    "bool": {
        "must": [
            {
                "match_all": {}
            }
        ]
    }
},
"script_fields": {
    "keywords_size": {
        "script": {
            "lang": "painless",
            "source": "params['_source']['keywords'].size() > 1 && params['_source']['keywords'].size() <= 5"
        }
    },
    "keywords_size1": {
        "script": {
            "lang": "painless",
            "source": "params['_source']['keywords'].size() > 6 && params['_source']['keywords'].size() <= 10"
        }
    },
    "keywords_size2": {
        "script": {
            "lang": "painless",
            "source": "params['_source']['keywords'].size() > 11 && params['_source']['keywords'].size() <= 15"
        }
    },
    "size": {
        "script": {
            "lang": "painless",
            "source": "params['_source']['keywords'].size()"
        }
    }
}

足以为每个文档添加一些字段。我也尝试将脚本实现到AGGS，试图为所需的每个类别创建存储桶，但不能使其正常工作。

Answer 1

好的，我设法通过使用脚本来解决此问题。如果对任何人有帮助，我将在此处发布答案，但是我想知道通过无脚本使用嵌套字段聚集的答案。因此，这里进行

{
"query": {
    "bool": {
        "must": [
            {
                "match_all": {}
            }
        ]
    }
},
"size": 0,
  "aggs": {
    "avg_keywords_per_study": {
      "terms": {
        "script": "if (params['_source']['keywords'] != null && (params['_source']['keywords'].length>=1 && params['_source']['keywords'].length<=5)){return '1 to 5'} else if (params['_source']['keywords'] != null && (params['_source']['keywords'].length>=6 && params['_source']['keywords'].length<=10)){return '6 to 10'} else if (params['_source']['keywords'] != null && (params['_source']['keywords'].length>=6 && params['_source']['keywords'].length<=10)){return '6 to 10'} else if (params['_source']['keywords'] != null && (params['_source']['keywords'].length>=11 && params['_source']['keywords'].length<=15)){return '11 to 15'} else if (params['_source']['keywords'] != null && (params['_source']['keywords'].length>=16 && params['_source']['keywords'].length<=20)){return '16 to 20'} else if (params['_source']['keywords'] != null && (params['_source']['keywords'].length>=21 && params['_source']['keywords'].length<=30)){return '16 to 20'} else return '31 or more';"
      }
    }
  }
}

示例响应：

    "aggregations": {
    "avg_keywords_per_study": {
        "doc_count_error_upper_bound": 0,
        "sum_other_doc_count": 0,
        "buckets": [
            {
                "key": "1 to 5",
                "doc_count": 487
            },
            {
                "key": "6 to 10",
                "doc_count": 254
            },
            {
                "key": "11 to 15",
                "doc_count": 28
            },
            {
                "key": "16 to 20",
                "doc_count": 18
            },
            {
                "key": "31 or more",
                "doc_count": 8
            }
        ]
    }
}