HIVE json 列解析为键和值

Question

在具有 2 列和 2 条记录的 table 中：

记录 1：第 1 列 - my_col 值为：{"XXX": ["123","456"],"YYY": ["246","135"]}，第 2 列 - ID 为 A123

记录 2：第 1 列 - my_col 值为：{"ZZZ":["333"]}，第 2 列 - ID 为 B222

期望：

Key	Value	ID
XXX	123	A123
XXX	456	A123
YYY	246	A123
YYY	135	A123
ZZZ	333	B222

下面的查询只检索键 XXX 和 YYY，我的场景有很多记录。所以查询它自己识别键并在没有指定的情况下展平值

SELECT
t.key,
kv.kval as value,
t.ID
FROM (
SELECT 
 e.key, e.value, --columns from lateral view
 t.id  --column from table  
FROM
    input_df t --add alias
    --move explode to lateral view in the FROM
    lateral view explode(map(
         'XXX',
         split(regexp_replace(get_json_object(my_col,'$.XXX'),'"|\[|\]',''),','),
         'YYY',
         split(regexp_replace(get_json_object(my_col,'$.YYY'),'"|\[|\]',''),',')
     )) e as key, value --add alias and col names 
) t LATERAL VIEW explode(t.value) kv as kval

Answer 1

代码见注释：

WITH input_df AS (
    SELECT 'A123' as id,  '{"XXX": ["123","456"],"YYY": ["246","135"]}' my_col union all
    select 'B222' as id, '{"ZZZ":["333"]}'
)

select split(element,':')[0] key,
       e.value,
       id
from
(
SELECT 
 regexp_replace(e.element,'^\{|"| *\[|\]|\}$','') element, --remove extra chars to get element like this XXX:123,456
 t.id 
FROM
    input_df t
    lateral view explode(split(my_col,'(?<=\]) *, *(?=\")')) e as element --split by comma between ] and " with optional spaces
)s lateral view explode(split(split(element,':')[1],',')) e as value

结果：

key value   id  
XXX 123     A123
XXX 456     A123
YYY 246     A123
YYY 135     A123
ZZZ 333     B222

HIVE json 列解析为键和值

HIVE json column parse as key and value

sql

arrays

json

hive

hiveql