SQL 大幅增加来自多个 JOINS 的总和答案

【问题标题】：SQL drastically increasing SUMS from multiple JOINSSQL 大幅增加来自多个 JOINS 的总和
【发布时间】：2021-05-26 02:29:19
【问题描述】：

我需要在下面的 SQL 语句中加入多个表。与原始语句相比，当我添加这些连接时，总和大大增加了计数：

LEFT outer join pp_sku_pattern skupattern 
ON skupattern.sp_sku_no = item.sku_no

LEFT outer join si_pi_shrinkages_hist pishretail 
ON pish_inventory_date = :inventoryDate
AND
pish_sku_no = item.sku_no
AND
ROWNUM = 1 

Left outer join si_store_cycle_counts cyclecounts 
ON cyclecounts.scc_str_no = :storeNbr
AND
scc_sku = item.sku_no

带有仪表板数据视图的原始 SQL 语句： Original SQL statement with dashboard data view image (accurate counts)

带有仪表板数据视图的新 SQL 语句： New SQL statement with dashboard data view image (inaccurate counts)

具有正确总和计数的原始 SQL 语句：

SELECT dept.dept_code,
       dept.dept_desc,
       var.sprvh_inventory_date,
       var.sprvh_store,
       var.sprvh_sku,
       brand.br_name,
       item.sku_style,
       SUM(var.sprvh_phys_qty),
       SUM(var.sprvh_book_qty),
       SUM(var.sprvh_curr_variance),
       SUM(var.sprvh_hist_variance),
       SUM(var.sprvh_revision_number)
FROM   si_pi_research_var_hist var
       left outer join pp_stock_keeping_units item
                    ON var.sprvh_sku = item.sku_no
                       AND var.sprvh_company_code = item.sku_company_code
       left outer join pp_departments dept
                    ON dept.dept_company_code = item.sku_company_code
                       AND dept.dept_code = item.sku_dept_code
       left outer join pp_brands brand
                    ON brand.br_id = item.sku_br_id
WHERE  var.sprvh_inventory_date = :inventoryDate
       AND VAR.sprvh_store = :storeNbr
       AND ( ( var.sprvh_phys_qty IS NOT NULL
               AND var.sprvh_phys_qty <> 0 )
              OR ( var.sprvh_book_qty IS NOT NULL
                   AND var.sprvh_book_qty <> 0 )
              OR ( var.sprvh_curr_variance IS NOT NULL
                   AND var.sprvh_curr_variance <> 0 )
              OR ( var.sprvh_hist_variance IS NOT NULL
                   AND var.sprvh_hist_variance <> 0 ) )
GROUP  BY dept.dept_code,
          dept.dept_desc,
          var.sprvh_inventory_date,
          var.sprvh_store,
          var.sprvh_sku,
          brand.br_name,
          item.sku_style
HAVING SUM(var.sprvh_curr_variance) <> 0
        OR SUM(var.sprvh_hist_variance) <> 0
ORDER  BY Abs(SUM(var.sprvh_curr_variance))

产生错误总和计数的全新 SQL 语句：

SELECT dept.dept_code,
       dept.dept_desc,
       var.sprvh_inventory_date,
       var.sprvh_store,
       var.sprvh_sku,
       brand.br_name,
       item.sku_style,
       skupattern.sp_pattern,
       pishretail.pish_retail,
       cyclecounts.scc_reason_process_flag,
       SUM(var.sprvh_phys_qty),
       SUM(var.sprvh_book_qty),
       SUM(var.sprvh_curr_variance),
       SUM(var.sprvh_hist_variance),
       SUM(var.sprvh_revision_number)
FROM   si_pi_research_var_hist var
       left outer join pp_stock_keeping_units item
                    ON var.sprvh_sku = item.sku_no
                       AND var.sprvh_company_code = item.sku_company_code
       left outer join pp_departments dept
                    ON dept.dept_company_code = item.sku_company_code
                       AND dept.dept_code = item.sku_dept_code
       left outer join pp_brands brand
                    ON brand.br_id = item.sku_br_id
       left outer join pp_sku_pattern skupattern
                    ON skupattern.sp_sku_no = item.sku_no
       left outer join si_pi_shrinkages_hist pishretail
                    ON pishretail.pish_inventory_date = :inventoryDate
                       AND pish_sku_no = item.sku_no
                       AND ROWNUM = 1
       left outer join si_store_cycle_counts cyclecounts
                    ON cyclecounts.scc_str_no = :storeNbr
                       AND scc_sku = item.sku_no
WHERE  var.sprvh_inventory_date = :inventoryDate
       AND VAR.sprvh_store = :storeNbr
       AND ( ( var.sprvh_phys_qty IS NOT NULL
               AND var.sprvh_phys_qty <> 0 )
              OR ( var.sprvh_book_qty IS NOT NULL
                   AND var.sprvh_book_qty <> 0 )
              OR ( var.sprvh_curr_variance IS NOT NULL
                   AND var.sprvh_curr_variance <> 0 )
              OR ( var.sprvh_hist_variance IS NOT NULL
                   AND var.sprvh_hist_variance <> 0 ) )
GROUP  BY dept.dept_code,
          dept.dept_desc,
          var.sprvh_inventory_date,
          var.sprvh_store,
          var.sprvh_sku,
          brand.br_name,
          item.sku_style,
          skupattern.sp_pattern,
          pishretail.pish_retail,
          cyclecounts.scc_reason_process_flag
HAVING SUM(var.sprvh_curr_variance) <> 0
        OR SUM(var.sprvh_hist_variance) <> 0
ORDER  BY Abs(SUM(var.sprvh_curr_variance))

问题：如何在新的 SQL 语句中保持原始语句总和与 3 个新连接相同？

【问题讨论】：

那么问题是什么？您知道，如果您连接到一个连接键不唯一的表，您会乘以当前记录，从而更改总和。所以你必须在新的连接中禁止它，例如将rownum = 1 谓词添加到缺少它的另外两个连接中。
您真的希望任何人都这样阅读查询吗？至少格式化一下。你可以格式化here
@MarmiteBomber 问题是：如何保持原始语句总和相同？感谢您的回答，我将尝试将 rownum = 1 谓词添加到缺少它的另外两个连接中，并让您知道:)
@OldProgrammer 请告诉我如何更好地格式化它，我是新手 :)
@OldProgrammer 谢谢，我已经更新了格式:)

标签： java sql oracle join

【解决方案1】：

您的问题来自您用于连接的非唯一值。例如：

 left outer join si_store_cycle_counts cyclecounts
                    ON cyclecounts.scc_str_no = :storeNbr
                       AND scc_sku = item.sku_no

首先。尝试为每个表字段使用表别名。

 left outer join si_store_cycle_counts cyclecounts
                    ON cyclecounts.scc_str_no = :storeNbr
                       AND cyclecounts.scc_sku = item.sku_no

第二。检查每个连接条件。举个例子。如果 join 从 si_pi_research_var_hist 中返回多于一行，则从连接表中返回一行。

【讨论】：

现在试试