查找所有表列的MIN和MAX值 [英] Finding MIN and MAX Values for All Table Columns

问题描述

此查询可以正常运行，但速度确实很慢。

This query works as intended, but it's just really slow. Does anyone here have recommendations to improve performance?

我实际上只是创建一个临时表来存储所有表名和列名，并通过WHILE语句在它们之间循环，以达到改善性能的目的。

I essentially just creating a temp table to store all the table and column names, and cycling through them via a WHILE statement, to create dynamic inserts into another table with the details I want.

我最近一次运行耗时约21分钟，这并不十分糟糕（考虑任务），但是我

My latest run took about 21 minutes, which isn't entirely terrible (considering the task), but I'd love to get some input on how/where it can be fine tuned.

USE <DATABASE>;

      IF NOT EXISTS(SELECT *
FROM sys.schemas WHERE name='temp')
BEGIN
EXEC ('CREATE SCHEMA temp');
END;

IF OBJECT_ID('temp.columns') IS NOT NULL
   BEGIN
      DROP TABLE temp.columns
   END;

SELECT [table_name]
    , [column_name]
    , [data_type]
    , [is_nullable]
    , [numeric_scale]
    , [ordinal_position]
INTO [temp].[columns]
FROM information_schema.columns c
WHERE table_schema = 'dbo'
      -- AND table_name = 'CONTACTS'
      ;

IF OBJECT_ID('_TableColumnsUsed') IS NOT NULL
   BEGIN
      DROP TABLE _TableColumnsUsed
   END;

      CREATE TABLE _TableColumnsUsed (Table_Name VARCHAR(255) NULL, Column_Position INT, Column_Name VARCHAR(255) NULL, Min_Value VARCHAR(MAX) NULL, Max_Value VARCHAR(MAX) NULL);


DECLARE
     @CurrentTable      VARCHAR(255)
    , @CurrentColumn     VARCHAR(255)
    , @CurrentIsNullable VARCHAR(3)
    , @CurrentNumeric    BIT
    , @CurrentPosition  INT
    , @SQL               VARCHAR(MAX);

WHILE
(
   SELECT COUNT(1)
   FROM temp.columns
) > 0
   BEGIN
      SELECT TOP 1 @CurrentTable = [Table_Name]
               , @CurrentColumn = [Column_Name]
               , @CurrentIsNullable = [is_nullable]
               , @CurrentNumeric = IIF([numeric_scale] IS NULL, 0, 1)
               , @CurrentPosition = [ordinal_position]
      FROM temp.columns c
      WHERE [table_name] NOT IN ('_TableColumnsUsed')
      ORDER BY [table_name]
            , [ordinal_position];

      SET @SQL = 'INSERT INTO _TableColumnsUsed (Table_Name, Column_Position, Column_Name, Min_Value, Max_Value)
         SELECT Table_Name = '''+@CurrentTable+'''
         , Column_Position = '+CAST(@CurrentPosition AS VARCHAR(3))+'
         , Column_Name = '''+@CurrentColumn+'''
         , Min_Value = MIN(CAST('+@CurrentColumn+' AS VARCHAR(MAX)))
         , Max_Value = MAX(CAST('+@CurrentColumn+' AS VARCHAR(MAX)))
         FROM '+@CurrentTable+'
         WHERE '+IIF(@CurrentIsNullable = 'NO', '1=1',
                                        CASE
                                            WHEN @CurrentNumeric = 0
                                            THEN 'ISNULL(CAST('+@CurrentColumn+' AS VARCHAR(MAX)),'''') <> '''''
                                            WHEN @CurrentNumeric = 1
                                            THEN 'ISNULL('+@CurrentColumn+',0.00) <> 0.00'
                                            ELSE '1=1'
                                        END);

      EXEC (@SQL);
      DELETE c
      FROM [temp].[columns] [c]
      WHERE [c].[table_name] = @CurrentTable
           AND [c].[column_name] = @CurrentColumn;
   END;


      /*  -- Dynamic SQL Output Example

         SELECT Table_Name = 'CONTACTS'
         , Column_Position = 17
         , Column_Name = 'ZIP'
         , Min_Value = MIN(CAST(ZIP AS VARCHAR(MAX)))
         , Max_Value = MAX(CAST(ZIP AS VARCHAR(MAX)))
         FROM CONTACTS
         WHERE 1=1

      */


SELECT Table_Name, Column_Position, Column_Name, Min_Value, Max_Value
FROM _TableColumnsUsed;

推荐答案

尝试一下，它应该在几秒钟内起作用：

Try this, it should work in seconds:

DECLARE @cmd NVARCHAR(MAX)=
(
    SELECT STUFF(
    (
            SELECT ' UNION ALL SELECT ''' + c.TABLE_SCHEMA + ''' AS TableSchema '
                 + ',''' + c.TABLE_NAME + ''' AS TableName '
                 + ',''' + c.COLUMN_NAME + ''' AS ColumnName '
                 + ',''' + c.DATA_TYPE + ''' AS ColumnType '
                 + ',CAST(MIN(' + QUOTENAME(c.COLUMN_NAME)+') AS NVARCHAR(MAX)) AS MinValue ' 
                 + ',CAST(MAX(' + QUOTENAME(c.COLUMN_NAME)+') AS NVARCHAR(MAX)) AS MaxValue ' 
                 + ' FROM ' + QUOTENAME(c.TABLE_SCHEMA) + '.' + QUOTENAME(c.TABLE_NAME)
                 + ' WHERE ' + QUOTENAME(c.COLUMN_NAME) + ' IS NOT NULL'
            FROM INFORMATION_SCHEMA.COLUMNS AS c
            WHERE c.DATA_TYPE IN('bigint','float','int','datetime') --add all types you want to check, be aware of implicit conversions!
            FOR XML PATH(''),TYPE
    ).value('.','nvarchar(max)'),1,10,'')
);
--PRINT @cmd
EXEC(@cmd);

该语句创建一个多合一 EXEC

执行的> UNION ALL 查询您可以取消注释 PRINT 查看已执行的语句。

You can uncomment the PRINT to see the executed statement.

这篇关于查找所有表列的MIN和MAX值的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！