为什么这在 Microsoft SQL Server Management Studio 下有效，但在 R 下无效？

Question

我收到此 SQL 声明：

use MYDATABASE; 
if object_id('tempdb..#mapDT') is not null drop table #mapDT; 
create table #mapDT (SqlDatatype varchar (64), MyNewDataType varchar(64));
insert into #mapDT SELECT 'varchar','type text' UNION ALL SELECT 'datetime','type datetime'UNION ALL SELECT 'tinyint','int64.Type' UNION ALL SELECT 'int','int64.Type' UNION ALL SELECT 'float','type number';
SELECT COLUMN_NAME, DATA_TYPE, 'MyString1' + COLUMN_NAME + 'MyString2' + m.MyNewDataType + 'MyString3' FROM INFORMATION_SCHEMA.COLUMNS C JOIN #mapDT m on m.SqlDatatype = C.DATA_TYPE WHERE TABLE_NAME = 'MYTABLE';

并且它运行在 MS SSMS 下运行良好。结果的相关部分是：

MyString1COLUMN01MyString2type textMyString3
MyString1COLUMN02MyString2type datetimeMyString3
MyString1COLUMN03MyString2type textMyString3
MyString1COLUMN04MyString2type textMyString3
MyString1COLUMN05MyString2int64.TypeMyString3
MyString1COLUMN06MyString2type datetimeMyString3
MyString1COLUMN07MyString2type datetimeMyString3
    ...     ...    ...

当我运行它在 R 下时，我必须省略 use MYDATABASE; 并确保我连接到正确的数据库，就是这种情况。但是 SQL 语句是有问题的。的输出：

> library(odbc)
> conn <- dbConnect(odbc(), Driver = "SQL Server", Server = "MYSERVER\MYINSTANCE", Database = "MYDATABASE", Trusted_Connection = "True")
> dbGetQuery(conn, "
+                   if object_id('tempdb..#mapDT') is not null drop table #mapDT; create table #mapDT (SqlDatatype varchar (64), MyNewDataType varchar(64));
+                   insert into #mapDT SELECT 'varchar', 'type text' UNION ALL SELECT 'datetime', 'type datetime' UNION ALL SELECT 'tinyint', 'int64.Type' UNION ALL SELECT 'int', 'int64.Type' UNION ALL SELECT 'float', 'type number';
+                   SELECT COLUMN_NAME, DATA_TYPE, 'MyString1' + COLUMN_NAME + 'MyString2' + m.MyNewDataType + 'MyString3' FROM INFORMATION_SCHEMA.COLUMNS C JOIN #mapDT m on m.SqlDatatype = C.DATA_TYPE WHERE TABLE_NAME = 'MYTABLE';
+                  ")

是data frame with 0 columns and 0 rows。

是不是使用了"temporal tables"（即#mapDT）导致空数据框？如果是这样，有没有办法让 R and/or Python 到 understand/use 他们？

Answer 1

正如 Gord Thompson 所知，代码应包含 SET NOCOUNT ON;。因此，

> library(odbc)
> conn <- dbConnect(odbc(), Driver = "SQL Server", Server = "MYSERVER\MYINSTANCE", Database = "MYDATABASE", Trusted_Connection = "True")
> dbGetQuery(conn, "
+                   SET NOCOUNT ON; if object_id('tempdb..#mapDT') is not null drop table #mapDT; create table #mapDT (SqlDatatype varchar (64), MyNewDataType varchar(64));
+                   insert into #mapDT SELECT 'varchar', 'type text' UNION ALL SELECT 'datetime', 'type datetime' UNION ALL SELECT 'tinyint', 'int64.Type' UNION ALL SELECT 'int', 'int64.Type' UNION ALL SELECT 'float', 'type number';
+                   SELECT COLUMN_NAME, DATA_TYPE, 'MyString1' + COLUMN_NAME + 'MyString2' + m.MyNewDataType + '},' FROM INFORMATION_SCHEMA.COLUMNS C JOIN #mapDT m on m.SqlDatatype = C.DATA_TYPE WHERE TABLE_NAME = 'MYTABLE';
+                  ")

在我的例子中不需要行号，所以我也使用了 formals(print.data.frame)$row.names <- FALSE:

> library(odbc)
> conn <- dbConnect(odbc(), Driver = "SQL Server", Server = "MYSERVER\MYINSTANCE", Database = "MYDATABASE", Trusted_Connection = "True")
> formals(print.data.frame)$row.names <- FALSE
> dbGetQuery(conn, "
+                   SET NOCOUNT ON; if object_id('tempdb..#mapDT') is not null drop table #mapDT; create table #mapDT (SqlDatatype varchar (64), MyNewDataType varchar(64));
+                   insert into #mapDT SELECT 'varchar', 'type text' UNION ALL SELECT 'datetime', 'type datetime' UNION ALL SELECT 'tinyint', 'int64.Type' UNION ALL SELECT 'int', 'int64.Type' UNION ALL SELECT 'float', 'type number';
+                   SELECT COLUMN_NAME, DATA_TYPE, 'MyString1' + COLUMN_NAME + 'MyString2' + m.MyNewDataType + 'MyString3' FROM INFORMATION_SCHEMA.COLUMNS C JOIN #mapDT m on m.SqlDatatype = C.DATA_TYPE WHERE TABLE_NAME = 'MYTABLE';
+                  ")
> formals(print.data.frame)$row.names <- TRUE

Answer 2

尽管如此，请考虑避免任何触发 NOCOUNT 的 DML 语句，因为您的临时 table 是不必要的。只需运行子查询 JOIN 即可获得更简单、可读、可维护的查询。下面 CONCAT 将 varchar 字段和名称计算列上的 + 替换为别名，OUTPUT:

SELECT COLUMN_NAME, DATA_TYPE, 
       CONCAT('MyString1', COLUMN_NAME,
              'MyString2', m.MyNewDataType, 
              'MyString3') AS OUTPUT
FROM INFORMATION_SCHEMA.COLUMNS C 
INNER JOIN 
   ( SELECT 'varchar' AS SqlDatatype, 'type text' AS MyNewDataType 
     UNION ALL SELECT 'datetime', 'type datetime'
     UNION ALL SELECT 'tinyint', 'int64.Type' 
     UNION ALL SELECT 'int', 'int64.Type'
     UNION ALL SELECT 'float', 'type number'
   ) m 

ON m.SqlDatatype = C.DATA_TYPE 
WHERE TABLE_NAME = 'MYTABLE';

为什么这在 Microsoft SQL Server Management Studio 下有效，但在 R 下无效？

Why does this work under Microsoft SQL Server Management Studio but not under R?

sql

odbc

ssms

r

rodbc