hive 獲取 json 中的 key

需求

有一下 json 數組，要求獲取其 key

[{"title":"笑傲江湖","author":"金庸"},{"title":"流星蝴蝶劍","author":"古龍"}]

需求實現

1、首先將 json 數組轉換為 字符串 ；這里使用 字符串替換函數  translate

select translate('[{"title":"笑傲江湖","author":"金庸"},{"title":"流星蝴蝶劍","author":"古龍"}]', '[]{}""', '') json_str;

結果1

title:笑傲江湖,author:金庸,title:流星蝴蝶劍,author:古龍

2、使用 regexp_replace 將中間的 ， 替換成 ；

select regexp_replace(translate('[{"title":"笑傲江湖","author":"金庸"},{"title":"流星蝴蝶劍","author":"古龍"}]', '[]{}""', ''), '\,',
                      '\:') json_str;

結果2

title:笑傲江湖:author:金庸:title:流星蝴蝶劍:author:古龍

3、使用 lateral view posexplode 炸開結果

select *
from (select regexp_replace(translate('[{"title":"笑傲江湖","author":"金庸"},{"title":"流星蝴蝶劍","author":"古龍"}]', '[]{}""', ''),
                            '\,', '\:') json_str) t lateral view posexplode(split(json_str, ':')) t1 as rn, value;

結果3

title:笑傲江湖:author:金庸:title:流星蝴蝶劍:author:古龍    0    title
title:笑傲江湖:author:金庸:title:流星蝴蝶劍:author:古龍    1    笑傲江湖
title:笑傲江湖:author:金庸:title:流星蝴蝶劍:author:古龍    2    author
title:笑傲江湖:author:金庸:title:流星蝴蝶劍:author:古龍    3    金庸
title:笑傲江湖:author:金庸:title:流星蝴蝶劍:author:古龍    4    title
title:笑傲江湖:author:金庸:title:流星蝴蝶劍:author:古龍    5    流星蝴蝶劍
title:笑傲江湖:author:金庸:title:流星蝴蝶劍:author:古龍    6    author
title:笑傲江湖:author:金庸:title:流星蝴蝶劍:author:古龍    7    古龍

4、對以上結果進行過濾

select rn, value
from (select *
      from (select regexp_replace(
                           translate('[{"title":"笑傲江湖","author":"金庸"},{"title":"流星蝴蝶劍","author":"古龍"}]', '[]{}""',
                                     ''),
                           '\,', '\:') json_str) t lateral view posexplode(split(json_str, ':')) t1 as rn, value) t
where rn % 2 = 0;

最終結果

0    title
2    author
4    title
6    author

 另一種實現方式，使用兩層translate 替換后側寫，參考實現

select value
from (select translate(info, ',', ':') info
      from (select translate('[{"title":"笑傲江湖","author":"金庸"},{"title":"流星蝴蝶劍","author":"古龍"}]', '{}[]"', '') info) t) t
         lateral view posexplode(split(info, ':')) t as rn, value
where rn % 2 = 0
group by value;