org.apache.spark.sql.catalyst.parser.ParseException：不匹配的输入“来自”期望<EOF>

Question

我在 pyspark sql 中遇到以下错误。

org.apache.spark.sql.catalyst.parser.ParseException： 不匹配的输入“来自”期望（第 9 行，第 4 位）

select a.appl_sta,
    a.appl_pcnvendorcode,
    a.appl_pcn,
    b.appl_corp_strct_in,
    a.base_dcsn_dt,
    a.pcn_curr_cd,
    a.pcn_mo,
    case when trim(upper(b.appl_corp_strct_in)) = 'S' then 1 else 0 end as sol_ind
    from lod1_aml_p2_413_gp_base a
----^^^
    left join
    (
    with abc as(
    select appl_pcn,appl_corp_strct_in,Row_number () OVER(PARTITION BY appl_pcn ORDER BY gcap_extr_dt_tm DESC) AS Recrd_rank1
    from cstonedb3.gcap_applcnt_data where appl_no =0
    and pcn_mo between string(cast(date_format(add_months('2023-09-26', -2),'yyyyMM') as int)) and string(date_format('2023-09-26','yyyyMM'))
    )select * from abc where Recrd_rank1 =1
    )b
    on a.appl_pcn = b.appl_pcn

有人可以帮忙看看上面的查询出了什么问题吗？该查询在 Hive 中运行良好，但在 Sparlsql 中抛出错误

Answer 1

尝试下面的 SQL 查询。

WITH
    gcap_applcnt_data AS (
        SELECT
            appl_pcn, 
            appl_corp_strct_in, 
            Row_number() OVER (
                PARTITION BY
                    appl_pcn
                ORDER BY gcap_extr_dt_tm DESC
            ) AS rnk, 
            appl_no,
            CAST(pcn_mo AS INT) AS pcn_mo,
            CAST(
                DATE_FORMAT(
                    ADD_MONTHS('2023-09-26', -2), 'yyyyMM'
                ) as int
            ) AS date_from, 
            DATE_FORMAT('2023-09-26', 'yyyyMM') AS date_to,
            (
                CASE
                    WHEN TRIM(UPPER(appl_corp_strct_in)) = 'S' THEN 1
                    ELSE 0
                END
            ) AS sol_ind
        FROM cstonedb3.gcap_applcnt_data
    ),
   
    gp_base AS (
        SELECT
            appl_pcnvendorcode, 
            appl_pcn, 
            base_dcsn_dt, 
            pcn_curr_cd, 
            pcn_mo
        FROM lod1_aml_p2_413_gp_base
    ),
    
    gcap_applcnt_data_cte AS (
        SELECT
            appl_pcn, 
            appl_corp_strct_in, 
            sol_ind
        FROM gcap_applcnt_data
        WHERE
            appl_no = 0
            AND pcn_mo BETWEEN from_date AND to_date
            AND rnk = 1
    )
    
SELECT l.appl_pcnvendorcode, l.appl_pcn, r.appl_corp_strct_in, l.base_dcsn_dt, l.pcn_curr_cd, l.pcn_mo, r.sol_ind
FROM gp_base l
LEFT JOIN gcap_applcnt_data_cte r ON l.appl_pcn = r.appl_pcn