Elasticsearch:深度嵌套聚合下的reverse_nested聚合不起作用

aka*_*kai 3 elasticsearch elasticsearch-aggregation

Elasticsearch 版本:2.3.3

基本上标题说明了一切。如果reverse_nested在第二个嵌套聚合下使用,尽管文档似乎通过reverse_nested(参见结果中的最后一个"doc_count"字段)来限定范围,但它后面的聚合以某种方式不起作用。

这里我准备了一个例子 - 一个文档是一个包含入学日期和考试历史的学生。

映射:

{
    "mappings": {
        "students": {
            "properties": {
                "name": {
                    "type": "string"},
                "enrollment": {
                    "type": "date"},
                "exam_histories": {
                    "type": "nested",
                    "properties": {
                        "date": {
                            "type": "date"},
                        "subjects": {
                            "type": "nested",
                            "properties": {
                                "name": {
                                    "type": "string"},
                                "score": {
                                    "type": "short"}}}}}}}}}
Run Code Online (Sandbox Code Playgroud)

测试文档:

{
    "name": "John",
    "enrollment": "2012-09-01T00:00:00+00:00",
    "exam_histories": [
        {
            "date": "2016-05-05T00:00:00+00:00",
            "subjects": [
                {
                    "name": "math",
                    "score": 90}]}]}
Run Code Online (Sandbox Code Playgroud)

聚合查询(没有实际意义):

{
    "aggs": {
        "nested_exam_histories": {
            "nested": {
                "path": "exam_histories"},
            "aggs": {
                "date_buckets": {
                    "date_histogram": {
                        "field": "exam_histories.date",
                        "interval": "day"},
                    "aggs": {
                        "this_reverse_nested_does_work": {
                            "reverse_nested": {},
                            "aggs": {
                                "newest_enrollment": {
                                    "max": {
                                        "field": "enrollment"}}}},
                        "deep_nested_subjects": {
                            "nested": {
                                "path": "exam_histories.subjects"},
                            "aggs": {
                                "score_buckets": {
                                    "terms": {
                                        "field": "exam_histories.subjects.score"},
                                    "aggs": {
                                        "this_reverse_nested_doesnt_work": {
                                            "reverse_nested": {},
                                            "aggs": {
                                                "newest_exam_date": {
                                                    "max": {
                                                        "field": "exam_histories.date"}}}}}}}}}}}}}}
Run Code Online (Sandbox Code Playgroud)

结果:

...
"aggregations" : {
    "nested_exam_histories" : {
      "doc_count" : 1,
      "date_buckets" : {
        "buckets" : [ {
          "key_as_string" : "2016-05-05T00:00:00.000Z",
          "key" : 1462406400000,
          "doc_count" : 1,
          "this_reverse_nested_does_work" : {
            "doc_count" : 1,
            "newest_enrollment" : {
              "value" : 1.3464576E12,
              "value_as_string" : "2012-09-01T00:00:00.000Z"
            }
          },
          "deep_nested_subjects" : {
            "doc_count" : 1,
            "score_buckets" : {
              "doc_count_error_upper_bound" : 0,
              "sum_other_doc_count" : 0,
              "buckets" : [ {
                "key" : 90,
                "doc_count" : 1,
                "this_reverse_nested_doesnt_work" : {
                  "doc_count" : 1,
                  "newest_exam_date" : {
                    "value" : null
                  }
...
Run Code Online (Sandbox Code Playgroud)

...您可以在其中看到聚合“newest_exam_date”不起作用。这是一个错误还是我做错了什么?

kee*_*ety 5

您需要使用该选项显式指定要“反向聚合”的嵌套对象path,否则它假定该字段位于根级别。

\n\n

文档中:

\n\n
\n

路径 - 定义应连接回哪个嵌套对象字段。默认为空,这意味着它会连接回根\n/主文档级别。该路径不能包含对位于verse_nested 所在的嵌套聚合\xe2\x80\x99s 嵌套\n 结构之外的嵌套\n 对象字段的引用。\n 示例:

\n
\n\n
{\n     "size":0,\n   "aggs": {\n      "nested_exam_histories": {\n         "nested": {\n            "path": "exam_histories"\n         },\n         "aggs": {\n            "date_buckets": {\n               "date_histogram": {\n                  "field": "exam_histories.date",\n                  "interval": "day"\n               },\n               "aggs": {\n                  "this_reverse_nested_does_work": {\n                     "reverse_nested": {},\n                     "aggs": {\n                        "newest_enrollment": {\n                           "max": {\n                              "field": "enrollment"\n                           }\n                        }\n                     }\n                  },\n                  "deep_nested_subjects": {\n                     "nested": {\n                        "path": "exam_histories.subjects"\n                     },\n                     "aggs": {\n                        "score_buckets": {\n                           "terms": {\n                              "field": "exam_histories.subjects.score"\n                           },\n                           "aggs": {\n                              "this_reverse_nested_doesnt_work": {\n                                 "reverse_nested": {\n                                    "path": "exam_histories"\n                                 },\n                                 "aggs": {\n                                    "newest_exam_date": {\n                                       "max": {\n                                          "field": "exam_histories.date"\n                                       }\n                                    }\n                                 }\n                              }\n                           }\n                        }\n                     }\n                  }\n               }\n            }\n         }\n      }\n   }\n}\n
Run Code Online (Sandbox Code Playgroud)\n\n

结果:

\n\n
 {\n   "took": 5,\n   "timed_out": false,\n   "_shards": {\n      "total": 5,\n      "successful": 5,\n      "failed": 0\n   },\n   "hits": {\n      "total": 2,\n      "max_score": 0,\n      "hits": []\n   },\n   "aggregations": {\n      "nested_exam_histories": {\n         "doc_count": 2,\n         "date_buckets": {\n            "buckets": [\n               {\n                  "key_as_string": "2016-05-05T00:00:00.000Z",\n                  "key": 1462406400000,\n                  "doc_count": 2,\n                  "this_reverse_nested_does_work": {\n                     "doc_count": 2,\n                     "newest_enrollment": {\n                        "value": 1377993600000,\n                        "value_as_string": "2013-09-01T00:00:00.000Z"\n                     }\n                  },\n                  "deep_nested_subjects": {\n                     "doc_count": 2,\n                     "score_buckets": {\n                        "doc_count_error_upper_bound": 0,\n                        "sum_other_doc_count": 0,\n                        "buckets": [\n                           {\n                              "key": 90,\n                              "doc_count": 2,\n                              "this_reverse_nested_doesnt_work": {\n                                 "doc_count": 2,\n                                 "newest_exam_date": {\n                                    "value": 1462406400000,\n                                    "value_as_string": "2016-05-05T00:00:00.000Z"\n                                 }\n                              }\n                           }\n                        ]\n                     }\n                  }\n               }\n            ]\n         }\n      }\n   }\n}\n
Run Code Online (Sandbox Code Playgroud)\n\n

注意path第二个“反向聚合”中的选项:

\n\n
reverse_nested": {\n    "path": "exam_histories"\n }\n
Run Code Online (Sandbox Code Playgroud)\n