我有下面的查询

set hive.cli.print.header=true;
set hive.query.max.partition=1000;
set hive.mapred.mode=unstrict;

SELECT

dim_lookup("accounts",name,"account_id") = '28016' as company,
dim_lookup("campaigns",name,"campaign_id") in (117649,112311,112319,112313,107799,110743,112559,112557,105191,105231,107377,108675,106587,107325,110671,107329,107181,106565,105123,106569,106579,110835,105127,105243,107185,105211,105215) as campaign_name,
case when is_click_through=0 then "PV" else "PC" end as conv_type,
(SELECT COUNT(1) FROM impressions WHERE ad_info[2] in (117649,112311,112319,112313,107799,110743,112559,112557,105191,105231,107377,108675,106587,107325,110671,107329,107181,106565,105123,106569,106579,110835,105127,105243,107185,105211,105215)) AS impressions

FROM actions

WHERE
data_date>='20170101'
AND data_date<='20171231'
AND conversion_action_id in (20769223,20769214,20769219,20764929,20764932,20764935,20769215,20769216,20764919,20769218,20769217,20769220,20769222)
GROUP BY conv_type

当我执行它时，我得到一个错误

 ERROR ql.Driver: FAILED: ParseException line 8:1 cannot recognize input near 'SELECT' 'COUNT' '(' in expression specification

我正在尝试获取指定的conversion_action_id的每次展示计数。我的查询中可能有什么错误？谢谢您的帮助。

仅供参考：ad_info [2]和campaign_id相同。