如何在 Oracle 18c 中高效解析 json?

How to parse json efficiently in Oracle 18c?

我正在尝试使用 JSON_OBJECT_TJSON_ARRAY_T API 解析大型 json 并且它工作正常,但我需要专家建议它是否有效?

我正在添加我的 json 文件和解析代码如下

SampleJson

代码

SET SERVEROUTPUT ON;
DECLARE
   l_clob clob;
   l_time timestamp;

   l_json json_object_t;
   l_stops_array json_array_t;
   l_stops_arr json_array_t;

   routeInfoObj json_object_t;
   routeStopArr json_array_t;
BEGIN
   SELECT LOG_CLOB INTO l_clob FROM ITV_DEV_LOGS WHERE LOG_ID = 1435334;
   l_time := systimestamp;

   l_json := json_object_t.parse( l_clob );

   dbms_output.put_line( 'Parsing Time: ' || extract(second from( systimestamp - l_time ) ) );
   l_stops_array := l_json.get_array('data');
   DBMS_OUTPUT.PUT_LINE('Data array: '||l_stops_array.get_size);
   FOR i in 0..l_stops_array.get_size-1 loop
      l_stops_arr := TREAT(l_stops_array.get(i) AS JSON_OBJECT_T).get_array('routedStops');
      DBMS_OUTPUT.PUT_LINE('stops array: '||l_stops_arr.get_size);
      FOR j in 0..l_stops_arr.get_size - 1 loop
         routeInfoObj := TREAT(l_stops_arr.get(j) AS JSON_OBJECT_T).get_object('routingInfo');
         DBMS_OUTPUT.PUT_LINE('Stop : ' || routeInfoObj.get_number('stop'));
         routeStopArr := TREAT(l_stops_arr.get(j) AS JSON_OBJECT_T).get_array('routedJobs');
         FOR k in 0..routeStopArr.get_size - 1 loop
            DBMS_OUTPUT.PUT_LINE('JobRef : ' || TREAT(routeStopArr.get(k) AS JSON_OBJECT_T).get_string('jobRef'));
         // update query to update stop value to respective jobRef
        end loop;
      end loop;
  end loop;
END;

它工作正常,但是有没有办法改进这个实现,因为这只是一个示例 json,里面的对象数量可能会达到 2000,而不是一个一个地更新记录,有没有办法在一个语句中更新所有记录?

您可以使用 json_table() 将 JSON 值转换为关系表示。这又可以在 MERGE 语句中使用。

例如以下查询:

select j.*
from itv_dev_logs
  cross join json_table(log_clob, '$.data.routedStops[*]'
                        columns stop_id integer path '$.stopId',
                                zone_ltf integer path '$.zoneLTF', 
                                info_stop_nr integer path '$.routingInfo.stop',
                                info_route_ref varchar(20) path '$.routingInfo.routeRef',
                                info_eta varchar(20) path '$.routingInfo.eta',
                                info_eta_dt timestamp path '$.routingInfo.etaDateTime',
                                info_stop_time number path '$.routingInfo.stopTime'
                        ) j
where log_id = 1435334;

Returns 像这样:

STOP_ID       | ZONE_LTF | INFO_STOP_NR | INFO_ROUTE_REF | INFO_ETA | INFO_ETA_DT             | INFO_STOP_TIME | INFO_DIST_PREV_STOP | INFO_BREAK_TIME | INFO_BREAK_DURATION
--------------+----------+--------------+----------------+----------+-------------------------+----------------+---------------------+-----------------+--------------------
1554383571432 |        1 |            1 | R119           | 11:01    | 2019-04-16 11:01:00.000 |           0.08 |                0.27 | 00:00           | 00:00              
1554383571515 |        1 |            2 | R119           | 11:07    | 2019-04-16 11:07:00.000 |           0.08 |                0.34 | 00:00           | 00:00              
1554383571601 |        1 |            3 | R119           | 11:13    | 2019-04-16 11:13:00.000 |           0.08 |                   0 | 00:00           | 00:00              
1554383571671 |        1 |            4 | R119           | 11:19    | 2019-04-16 11:19:00.000 |           0.08 |                   0 | 00:00           | 00:00              
1554383571739 |        1 |            5 | R119           | 11:25    | 2019-04-16 11:25:00.000 |           0.08 |                   0 | 00:00           | 00:00              

可以用作更新目标的 MERGE 语句的来源 table:

merge into your_target_table tg
using (
    select j.*
    from itv_dev_logs
      cross join json_table(log_clob, '$.data.routedStops[*]'
                            columns stop_id integer path '$.stopId',
                                    zone_ltf integer path '$.zoneLTF', 
                                    info_stop_nr integer path '$.routingInfo.stop',
                                    info_route_ref varchar(20) path '$.routingInfo.routeRef',
                                    info_eta varchar(20) path '$.routingInfo.eta',
                                    info_eta_dt timestamp path '$.routingInfo.etaDateTime',
                                    info_stop_time number path '$.routingInfo.stopTime'
                            ) j
    where log_id = 1435334
) t on (t.stop_id = tg.stop_id and ... more join conditions ...)
when matched then update
    set stop_nr = t.info_stop_nr, 
        eta_timestamp = t.eta_dt, 

由于您既没有提供目标的结构,也没有提供哪些 JSON 键应该与哪些 table 列匹配的信息,所有列名都只是猜测,您需要将它们替换为正确的名字。