使用什么类型的 JOIN

2021-12-17 00:00:00 join postgresql mysql sql-server zoho

我将使用什么类型的 JOIN 来使 table1table2 只匹配一次.例如,我有 table1(40 行)和 table2(10000 行).但是当我在 table1.LocationArea = table2.Location

What type of JOIN would I use to get table1 and table2 to be matched only once. For example, I have table1 (40 rows) and table2 (10000 rows). But I get table1 repeated over and over when I use a join on table1.LocationArea = table2.Location

What I get:                         What I wish I could get:
t1.LocationArea,t2.Location         t1.LocationArea,t2.Location
---------------------------         ---------------------------
az,az                               az,az
az,az                               null,az
ca,ca                               ca,ca
il,il                               il,il
tx,tx                               tx,tx
tx,tx                               null,tx
az,az                               null,az
                                    null,il
                                    null,ca

我希望在查询中得到 10000 条记录.

I wish to end up with 10000 records in the query.

我已经尝试过inner joinleft,并且我正在使用不支持外连接的 ZOHO 报告.

I have tried inner join, left, and I am using ZOHO reports which does not support outer join's.

SELECT "table1"."LocationArea", "Location" 
FROM "table2"
left join "table1" on  "Location" = "table1"."LocationArea"

推荐答案

很明显,您的两个连接列都有重复的值.代替 笛卡尔积,[INNER] JOIN 会产生这样,您希望每一行只使用一次.您可以通过为每个重复项添加一个行号 (rn) 并另外加入 rn 来实现这一点.

Obviously, you have duplicate values for both of the joining columns. Instead of the Cartesian product an [INNER] JOIN would produce for this, you want each row to be used only once. You can achieve this by adding a row number (rn) per duplicate and join on rn additionally.

除非您有额外的限制(如 FK 约束),否则每个表可以有更多或更少的相同值的重复项 - 但您的问题没有任何内容.要保留所有行,可以使用 FULL [OUTER] JOIN.但是你想在结果中保留10000条记录,这是table2的基数.所以它必须是 LEFT[OUTER] JOIN on table1(有 40 行) - 并从 table1 中排除可能过多的行.

Each table can have more or fewer dupes for the same value than the other unless you have additional restrictions in place (like a FK constraint) - but there is nothing in your question. To keep all rows one would use a FULL [OUTER] JOIN. But you want to keep 10000 records in the result, which is the cardinality of table2. So it must be a LEFT [OUTER] JOIN on table1 (with 40 rows) - and exclude possible excessive rows from table1.

SELECT t1."LocationArea", t2."Location"
FROM  (
   SELECT "Location"
        , row_number() OVER (PARTITION BY "Location") AS rn
   FROM   table2
   ) t2
LEFT JOIN (
   SELECT "LocationArea"
        , row_number() OVER (PARTITION BY "LocationArea") AS rn
   FROM   table1
   ) t1 ON t1."LocationArea" = t2."Location"
       AND t1.rn = t2.rn;

适用于 Postgres 或 SQL Server.MySQL 不支持窗口函数,您需要一个替代品:

Works for Postgres or SQL Server. MySQL doesn't support window functions, you would need a substitute:

  • SQL SELECT 最后一个条目无限制

需要说明的是:LEFT JOIN 只是LEFT OUTER JOIN 的简写,因此您已经在使用外连接.你的陈述是一个误解:

To be clear: LEFT JOIN is just shorthand for LEFT OUTER JOIN, so you are already using an outer join. Your statement is a misunderstanding:

我使用的 ZOHO 报告不支持外连接.

I am using ZOHO reports which does not support outer join's.

相关文章