有效地处理/解决大型查询

Question

我需要添加另一对左外连接到我的巨大查询。这对我来说越来越困难，因为现在对我来说查询太大了。我想知道是否有一些方法将查询自动转换为维恩图。

是否有工具可以查询并绘制表示查询的venndiagrams？

我正在寻找解决方案在sql server，c＃或python？

是否有自动化的方式来表示查询图形？

WITH cte_biggie 
    AS (SELECT 
    LEFT('I'+Replicate(' ', 2), 2) 
    + LEFT(CONVERT(VARCHAR, Getdate(), 110)+Replicate(' ', 10), 10) 
    + LEFT('MILLENN'+Replicate(' ', 16), 16) 
    + LEFT('TERR'+Replicate(' ', 12), 12) 
    + LEFT(' '+Replicate(' ', 16), 16) 
    + LEFT('IMS'+Replicate(' ', 8), 8) 
    + LEFT(' '+Replicate(' ', 15), 15) 
    + LEFT(LEFT(LEFT(Isnull(MLIS.requestor_first_name, 'XXX'), 
    Charindex( 
    ' ' 
     , Isnull 
    (MLIS.requestor_first_name, 'XXX') + ' ') - 1), Charindex(' ', 
    LEFT(Isnull(MLIS.requestor_first_name, 'XXX'), Charindex(' ', 
    Isnull(MLIS.requestor_first_name, 'XXX') + ' ') - 1) + 
    ' ') - 1) +Replicate(' ', 25), 25) 
    + LEFT(' '+Replicate(' ', 20), 20) 
    + LEFT(LEFT(LEFT(Isnull(MLIS.requestor_last_name, ''), 
    Charindex( 
    ',', 
     Isnull( 
    MLIS.requestor_last_name, '') + ',') - 1), Charindex(',', 
    LEFT(Isnull(MLIS.requestor_last_name, ''), Charindex(',', 
    Isnull(MLIS.requestor_last_name, '') + ',') - 1) + 
    ',') - 1) +Replicate(' ', 25), 25) 
    + LEFT(' '+Replicate(' ', 5), 5) 
    + LEFT(' '+Replicate(' ', 3), 3) 
    + LEFT(' '+Replicate(' ', 3), 3) 
    + LEFT(' '+Replicate(' ', 3), 3) 
    + LEFT(Isnull(MLIS.requestor_npi, '')+Replicate(' ', 12), 12) 
    + LEFT('13'+Replicate(' ', 2), 2) 
    + LEFT(' '+Replicate(' ', 15), 15) 
    + LEFT(' '+Replicate(' ', 30), 30) 
    + LEFT(' '+Replicate(' ', 15), 15) 
    + LEFT(' '+Replicate(' ', 4), 4) 
    + LEFT(' '+Replicate(' ', 2), 2) 
    + LEFT(' '+Replicate(' ', 2), 2) 
    + LEFT(' '+Replicate(' ', 1), 1) 
    + LEFT(' '+Replicate(' ', 10), 10) 
    + LEFT(' '+Replicate(' ', 2), 2) 
    + LEFT(' '+Replicate(' ', 20), 20) 
    + LEFT(' '+Replicate(' ', 10), 10) 
    + LEFT(' '+Replicate(' ', 4), 4) 
    + LEFT(' '+Replicate(' ', 100), 100) 
    + LEFT(' '+Replicate(' ', 20), 20) 
    + LEFT(' '+Replicate(' ', 12), 12) 
    + LEFT(' '+Replicate(' ', 2), 2) 
    + LEFT(' '+Replicate(' ', 1), 1) 
    + LEFT(Isnull(SplitAddresses.address1, '')+Replicate(' ', 50), 50) 
    + LEFT(Isnull(SplitAddresses.address2, '')+Replicate(' ', 50), 50) 
    + LEFT(' '+Replicate(' ', 50), 50) 
    + LEFT(' '+Replicate(' ', 50), 50) 
    + LEFT(Isnull(CLIENT.mlis_client_city, 'XXX')+Replicate(' ', 30), 30) 
    + LEFT(Isnull(CLIENT.mlis_client_state, '')+Replicate(' ', 15), 15) 
    + LEFT(Isnull(CLIENT.mlis_client_zipcode, 'XXX')+Replicate(' ', 10), 10) 
    + LEFT(' '+Replicate(' ', 20), 20) 
    + LEFT(' '+Replicate(' ', 9), 9) 
    + LEFT(' '+Replicate(' ', 7), 7) 
    + LEFT(' '+Replicate(' ', 3), 3) 
    + LEFT(' '+Replicate(' ', 1), 1) 
    + LEFT('MILLENN'+Replicate(' ', 10), 10) 
    + LEFT('1'+Replicate(' ', 1), 1) 
    + LEFT('DrScrub'+Replicate(' ', 15), 15) 
    + LEFT(' '+Replicate(' ', 10), 10) 
    + LEFT(' '+Replicate(' ', 101), 101) PHYSICIAN_INFO 
     --+LEFT(' '+REPLICATE(' ',1),1) 
     FROM (SELECT DISTINCT mlis_practice_id, 
           requestor_npi, 
           requestor_first_name, 
           requestor_last_name 
       FROM complete 
       WHERE 
         Len(requestor_last_name) > 0 
         ) MLIS 
       LEFT JOIN (SELECT * 
          FROM millennium_dw..d_physician 
          WHERE received_ims_information = 1) PHYSICIAN 
         ON MLIS.requestor_npi = Cast(PHYSICIAN.physician_npi AS 
                VARCHAR) 
       --we can remove this cast later 
       LEFT JOIN (SELECT * 
          FROM millennium_dw..d_client 
          WHERE rec_active_flg = 1) CLIENT 
         ON MLIS.mlis_practice_id = CLIENT.client_id 
       LEFT JOIN (SELECT DISTINCT T.mlis_client_address_1, 
              T.client_id, 
              LEFT(T.mlis_client_address_1, Isnull( 
              SplitAddresses.pos - 1, 
              2147483647)) 
              Address1, 
              Substring(T.mlis_client_address_1, 
              SplitAddresses.pos + 1, 
              2147483647) Address2 
          FROM d_client T 

            OUTER apply (SELECT TOP 1 NULLIF( 
               Patindex(delimiter, 
               T.mlis_client_address_1), 0 
                  ) Pos 
               FROM 
            (VALUES ('% Suite %'), 
              ('% Ste %'), 
              ('% Bldg %'), 
              ('% Building %'), 
              ('%#%')) SplitAddresses 
            ( 
            delimiter) 
               WHERE T.mlis_client_address_1 
                 LIKE 
                 splitaddresses.delimiter 
               ORDER BY pos) SplitAddresses) 
          SplitAddresses 
         ON SplitAddresses.client_id = CLIENT.client_id 
     WHERE PHYSICIAN.physician_npi IS NULL 
       AND CLIENT.received_mlis_information = 1 
       AND LEN(address1)>0 
       AND LEN(CLIENT.mlis_client_state)>0 
       AND address1!='NO ADDRESS' 
       ) 
--AND MLIS.REQUESTOR_NPI IS NOT NULL) 
SELECT * 
FROM cte_biggie 

Answer 1

的CTE可以在同一时间内建立起来的查询：

DECLARE @start_date AS DATETIME = '2012-02-02'; 
WITH L0 AS (SELECT 1 AS C UNION ALL SELECT 1),  --  2 
    L1 AS (SELECT L0.C FROM L0 CROSS JOIN L0 AS B), --  4 
    L2 AS (SELECT L1.C FROM L1 CROSS JOIN L1 AS B), --  16 
    L3 AS (SELECT L2.C FROM L2 CROSS JOIN L2 AS B), -- 256 
    L4 AS (SELECT L3.C FROM L3 CROSS JOIN L3 AS B), -- 65536 
     N AS (SELECT ROW_NUMBER() OVER(ORDER BY L4.C) AS n FROM L4), 
     D AS (SELECT DATEADD(SECOND, n - 1, @start_date) AS d, n FROM N) 
SELECT d from D where n <= (60 * 60 * 24) 

注意，最终SELECT可以使用任何或全部CTE中间结果。这可以大大简化开发和测试。

编辑：在你的情况是这样的。

WITH 
    MLIS as (-- Description of what this piece is supposed to mean. 
    SELECT DISTINCT mlis_practice_id, requestor_npi, requestor_first_name, requestor_last_name 
     FROM complete 
     WHERE Len(requestor_last_name) > 0), 
    Physician as (-- Description of what this piece is supposed to mean. 
    SELECT * 
     FROM millennium_dw..d_physician 
     WHERE received_ims_information = 1), 
    Client as (-- Description of what this piece is supposed to mean. 
    SELECT * 
     FROM millennium_dw..d_client 
     WHERE rec_active_flg = 1), 
    cte_biggie as (-- Something I've totally bodged that should use the above CTEs. 
    SELECT LEFT('I'+Replicate(' ', 2), 2), ...) 
SELECT * from CTE_Biggie