PostgreSQL json 提取答案

【问题标题】：PostgreSQL json extractionPostgreSQL json 提取
【发布时间】：2019-12-11 13:05:19
【问题描述】：

我有来自 MongoDB 的航班预订的 json 数据。此数据插入到 PostgreSQL 表中，数据类型为 json。行程信息具有以下结构。

[
  {
    "segments": {
      "1": {
        "airline": "G8",
        "flight_number": "761",
        "arrival-date": "2018-11-02T12:45:00",
        "arrival_airport": "CCU",
        "departure_date": "2018-11-02T11:35:00",
        "departure_airport": "BBI",
        "operating_airline": "G8",
        "stops": 0,
        "index": 1,
        "duration": 4200
      }
    }
  },
  {
    "segments": {
      "2": {
        "airline": "6E",
        "flight_number": "775",
        "arrival-date": "2018-11-03T00:30:00",
        "arrival_airport": "BBI",
        "departure_date": "2018-11-02T23:20:00",
        "departure_airport": "CCU",
        "operating_airline": "6E",
        "stops": 0,
        "index": 2,
        "duration": 4200
      }
    }
  }
]

段数未知。如何将其拆分为多个记录，以便每个段都成为记录？ json_array_elements 是我已经在使用的函数，如果我使用json_array_elements(data->'flights')->'segments'，我会得到这样的输出

{
  "1": {
    "airline": "AI",
    "flight_number": "20",
    "arrival_airport": "CCU",
    "arrival-date": "2018-10-17T16:40:00",
    "departure_date": "2018-10-17T14:25:00",
    "departure_airport": "DEL",
    "operating_airline": "AI",
    "stops": 0,
    "index": 1,
    "duration": 8100
  },
  "2": {
    "airline": "AI",
    "flight_number": "230",
    "arrival_airport": "DAC",
    "arrival-date": "2018-10-17T20:25:00",
    "departure_date": "2018-10-17T19:00:00",
    "departure_airport": "CCU",
    "operating_airline": "AI",
    "stops": 0,
    "index": 2,
    "duration": 3300
  }
}

如何将其转换为单独的记录？我想执行这样的插入 -

insert into stage_itinerary (segment_id,airline,flight_number,arrival_aiport...)  select (....)

【问题讨论】：

标签： sql json postgresql

【解决方案1】：

Click: demo:db<>fiddle

你可以使用json_array_elements():

SELECT
    elems
FROM 
    mytable,
    json_array_elements(my_segments_column) elems

如果要使用拆分数据更新表，则需要插入新数据并删除旧数据。这可以使用 CTE（WITH 子句）在一个查询中完成：

Click: demo:db<>fiddle

WITH del AS (
    DELETE FROM mytable
    RETURNING my_segments_column
)
INSERT INTO mytable
SELECT
    elems
FROM 
    del,
    json_array_elements(my_segments_column) elems;

在 CTE 中，所有数据都被删除，但通过 RETURNING 子句返回。这个返回的数据用于计算如上所示的分割记录，之后插入。

【讨论】：

抱歉没有把问题说清楚。我已经编辑了我的问题。

【解决方案2】：

这实际上是一个非常好的堆叠多个横向查询的用例，因为每个横向查询都可以引用它之前定义的所有对象。

SELECT a4.*
FROM test
, LATERAL (SELECT a1->'segments' a2 FROM jsonb_array_elements(a) a1) a1
, LATERAL (SELECT value FROM jsonb_each(a2)) a3                    
, LATERAL (SELECT * 
           FROM jsonb_to_record(a3) a4 
           ("index" int, stops int, airline text, duration int, 
            "arrival-date" timestamptz, flight_number text, departure_date text,
            arrival_airport text, departure_airport text, operating_airline text)
          ) a4

如果你有幸使用了 postgresql 12 或更高版本，那么上面的查询可以更优雅地写成：

SELECT a2.*
FROM test
, LATERAL (SELECT * FROM jsonb_path_query(a, '$.*.*') a1 ) a1
, LATERAL (SELECT * 
           FROM jsonb_to_record(a1) a2 
          ("index" int, stops int, airline text, duration int, 
           "arrival-date" timestamptz, flight_number text, departure_date text,
           arrival_airport text, departure_airport text, operating_airline text)
) a2

这里是设置代码：

CREATE TABLE test (a) AS SELECT 
'[
  {
    "segments": {
      "1": {
        "airline": "G8",
        "flight_number": "761",
        "arrival-date": "2018-11-02T12:45:00",
        "arrival_airport": "CCU",
        "departure_date": "2018-11-02T11:35:00",
        "departure_airport": "BBI",
        "operating_airline": "G8",
        "stops": 0,
        "index": 1,
        "duration": 4200
      }
    }
  },
  {
    "segments": {
      "2": {
        "airline": "6E",
        "flight_number": "775",
        "arrival-date": "2018-11-03T00:30:00",
        "arrival_airport": "BBI",
        "departure_date": "2018-11-02T23:20:00",
        "departure_airport": "CCU",
        "operating_airline": "6E",
        "stops": 0,
        "index": 2,
        "duration": 4200
      }
    }
  }
]'::jsonb

【讨论】：