SQL/Hive 查询以计算特定值每天的行数

Question

I am currently working on a Python script that uses a query to extract data from our Hive Server. I am expecting an output that will filter the cardnumbers that have x or more transactions per day depending on the input for "TxnCount".

Inputs are: DateTime1, DateTime2, MerchantID, CardNum, terminalID and TxnCount.

My Code (not working):

Query = "SELECT TRIM(i002_number) as CardNum, i004_amt_trxn, TRIM(i042_merch_id) as MerchantID, i043a_merch_name, TRIM(i041_pos_id) as TerminalID, \
i049_cur_trxn, i062v2_trans_id, i003_proc_code, i006_amt_bill, i051_cur_bill, amt_card, cardcurrency, ltimestamp, \
i039_rsp_cd, i018_merch_type, i043b_merch_city, i043c_merch_cnt, i022_pos_entry, i032_acquirer_id, trxntype, reasoncode, \
SUBSTRING(i002_number, 1, 6) AS issuer_bin, COUNT(i002_number) as txncount\
CASE \
    WHEN SUBSTRING(i002_number,1,1) = 5 THEN 'MasterCard' \
    WHEN SUBSTRING(i002_number,1,1) = 4 THEN 'VISA' \
END AS source \
FROM tsys.ods_authorizations \
WHERE ltimestamp >= '"+DateTime1+"' AND ltimestamp <= '"+DateTime2+"' AND i042_merch_id = "+MerchantID+" \
AND i002_number = "+CardNum+" AND i041_pos_id = "+terminalID+""
HAVING txncount >= '"+TxnCount+"'

Sample Expected Data (truncated):

CardNum         TimeStamp           TxnCount
123      2019-06-01 00:00:30.00        2   
123      2019-06-01 05:00:20.00        2
123      2019-06-03 20:00:00.00        1
456      2019-06-04 06:00:00.00        2
456      2019-06-04 00:00:10.91        2
789      2019-06-01 12:00:40.51        1

I think my problem here is that it cannot count per card number since I am having problems with the GROUP BY clause. Also, I have not yet split the date and time and the query would not recognize the difference between dates yet.

Answer 1

Your query is malformed. You have a COUNT() with a bunch of other columns -- and you have no GROUP BY . This is not allowed in SQL.

I would advise you to use parameters as well, rather than munging the query string. So, you presumably want something like this.

Your results appear to want the details of every transaction, rather than a summary (so you have two rows with a count of "2" rather than one). That suggests that you really want window functions:

SELECT a.*
FROM (SELECT a.*,
             (CASE WHEN i002_number LIKE '5%' THEN 'MasterCard'
                   WHEN i002_number LIKE '4%' THEN 'VISA'
              END) AS source
             COUNT(*) OVER (PARTITION BY i002_number, TRUNC(ltimestamp, 'DAY')) as txncount
      FROM FROM tsys.ods_authorizations a
      WHERE ltimestamp >= :timestamp1 AND
            ltimestamp <= :timestamp2 AND
            i042_merch_id = :MerchantID AND
            i002_number = :CardNum AND 
            i041_pos_id = :terminalID
     ) a
WHERE txncount >= :TxnCount