无法获取行中数组< struct< string,string>>的avro数据

zwghvu4y  于 2021-06-02  发布在  Hadoop
关注(0)|答案(1)|浏览(423)

嗨,我有一个结构数组的avro模式,我可以保存数据为avro。但是在从

array<struct<string, string>>

我不能排成一排。我得到的所有数据都在一行中。
这是表格定义

CREATE EXTERNAL TABLE meterevents ROW FORMAT SERDE org.apache.hadoop.hive.serde2.avro.AvroSerDe' STORED as INPUTFORMAT org.apache.hadoop.hive.ql.io.avro.AvroContainerInputFormat' OUTPUTFORMAT 'org.apache.hadoop.hive.ql.io.avro.AvroContainerOutputFormat' LOCATION '/......' TBLPROPERTIES ('avro.schema.url'='/..../schema.avsc');

配置单元表结构

nametype                struct<nametypedescription:string,nametypename:string,nametypeauthority:struct<nametypeauthorityname:string,nametypeauthoritydescription:string>>       from deserializer
names                   struct<name:string,nametype:struct<nametypedescription:string,nametypename:string,nametypeauthority:struct<nametypeauthorityname:string,nametypeauthoritydescription:string>>>  from deserializer
enddeviceeventdetails   struct<enddeviceeventdetailsname:string,enddeviceeventdetailsvalue:string>      from deserializer
enddeviceevent          struct<mrid:string,createddatetime:string,issuerid:string,issuertrackingid:string,reason:string,severity:string,userid:string,asset:struct<assetmrid:string,assetnames:array<struct<name:string,nametype:struct<nametypedescription:string,nametypename:string,nametypeauthority:struct<nametypeauthorityname:string,nametypeauthoritydescription:string>>>>>,enddeviceeventdetails:array<struct<enddeviceeventdetailsname:string,enddeviceeventdetailsvalue:string>>,enddeviceeventtype:string,enddeviceeventnames:array<struct<name:string,nametype:struct<nametypedescription:string,nametypename:string,nametypeauthority:struct<nametypeauthorityname:string,nametypeauthoritydescription:string>>>>,status:struct<statusdatetime:string,statusreason:string,statusremark:string,statusvalue:string>,usagepoint:struct<usagepointmrid:string,usagepointnames:array<struct<name:string,nametype:struct<nametypedescription:string,nametypename:string,nametypeauthority:struct<nametypeauthorityname:string,nametypeauthoritydescription:string>>>>>>       from deserializer
enddeviceeventtype      struct<enddeviceeventtypemrid:string,enddeviceeventtypedomain:string,enddeviceeventtypeeventoraction:string,enddeviceeventtypesubdomain:string,type:string,enddeviceeventtypenames:array<struct<name:string,nametype:struct<nametypedescription:string,nametypename:string,nametypeauthority:struct<nametypeauthorityname:string,nametypeauthoritydescription:string>>>>>       from deserializer
header                  struct<noun:string,context:string,verb:string,value:string,source:string,timestamp:string,correlationid:string,name:string,messageid:string,property:struct<propertyname:array<string>,propertyvalue:array<string>>>    from deserializer
payload                 struct<enddeviceevents:array<struct<mrid:string,createddatetime:string,issuerid:string,issuertrackingid:string,reason:string,severity:string,userid:string,asset:struct<assetmrid:string,assetnames:array<struct<name:string,nametype:struct<nametypedescription:string,nametypename:string,nametypeauthority:struct<nametypeauthorityname:string,nametypeauthoritydescription:string>>>>>,enddeviceeventdetails:array<struct<enddeviceeventdetailsname:string,enddeviceeventdetailsvalue:string>>,enddeviceeventtype:string,enddeviceeventnames:array<struct<name:string,nametype:struct<nametypedescription:string,nametypename:string,nametypeauthority:struct<nametypeauthorityname:string,nametypeauthoritydescription:string>>>>,status:struct<statusdatetime:string,statusreason:string,statusremark:string,statusvalue:string>,usagepoint:struct<usagepointmrid:string,usagepointnames:array<struct<name:string,nametype:struct<nametypedescription:string,nametypename:string,nametypeauthority:struct<nametypeauthorityname:string,nametypeauthoritydescription:string>>>>>>>,enddeviceeventtype:array<struct<enddeviceeventtypemrid:string,enddeviceeventtypedomain:string,enddeviceeventtypeeventoraction:string,enddeviceeventtypesubdomain:string,type:string,enddeviceeventtypenames:array<struct<name:string,nametype:struct<nametypedescription:string,nametypename:string,nametypeauthority:struct<nametypeauthorityname:string,nametypeauthoritydescription:string>>>>>>>

我在查询中使用“横向视图分解”选项

select eddetails.enddeviceeventdetailsname,	eddetails.enddeviceeventdetailsvalue 
FROM meterevents_tmp 
LATERAL VIEW explode(payload.enddeviceevents.enddeviceeventdetails)  ed AS eddetails 
limit 1;

但我还是在单行中获取数据。

enddeviceeventdetailsname       enddeviceeventdetailsvalue
["EventSequenceNumber","EventSequenceNumber","EventSequenceNumber","EventSequenceNumber"]       ["683","684","685","686"

我想把这些数据作为

enddeviceeventdetailsname       enddeviceeventdetailsvalue
EventSequenceNumber              683
EventSequenceNumber              684
EventSequenceNumber              685
EventSequenceNumber              686

我读过关于stackoverflow的另一个问题:使用hiveql分解struct数组
但无法获得预期的输出。因为在那个post中,它的配置单元外部表不是serde,我不能指定“map keys terminated by”和“collection items terminated by”
非常感谢您的帮助。
谢谢

h79rfbju

h79rfbju1#

我能解决这个问题---
我无法按行获取输出,因为

array<struct<string,string>>

是父数组的一部分

array<struct<array<struct<string, string>>>

我更新了查询并使用了嵌套的explod

select eddetails.enddeviceeventdetailsname, eddetails.enddeviceeventdetailsvalue from (select ede.enddeviceeventdetails FROM meterevents_tmp LATERAL VIEW explode(payload.enddeviceevents)  e AS ede) t LATERAL VIEW explode(t.enddeviceeventdetails)  ed AS eddetails limit 10;

我得到了想要的结果-

enddeviceeventdetailsname       enddeviceeventdetailsvalue
EventSequenceNumber     683
EventSequenceNumber     684
EventSequenceNumber     685
EventSequenceNumber     686

相关问题