单个 SQL 查询查找数据库中所有列中的空值

我想确定所有表中每列中空值的数量。我有一个数据库，它由大约 250 个表组成。其中大多数都在使用中。问题是几乎所有表都包含为某些短期创建的不需要的列术语使用。现在我们想要识别所有表中具有空值的列。由于表的计数很大并且时间较少。我想知道一种最简单的方法来按列识别每个表上的空记录计数。

我尝试了从互联网上获得的这个查询。但是在这个查询中我必须手动给出每个表名称。

DECLARE @cols1 NVARCHAR(MAX);
DECLARE @sql NVARCHAR(MAX);

SELECT @cols1 = STUFF((
    SELECT ', COUNT(CASE WHEN ISNULL(CONVERT(NVARCHAR(MAX), [' + t1.NAME + ']),'''' ) = '''' THEN 1 END) AS ' + t1.name
    FROM sys.columns AS t1
    WHERE t1.object_id = OBJECT_ID('Area')
    -- ORDER BY ', COUNT([' + t1.name + ']) AS ' + t1.name
    FOR XML PATH('')
), 1, 2, '');

SET @sql = '
SELECT ' + @cols1 + '
FROM Area
'
EXEC(@sql)

请帮助我改进查询获取结果。

Thanku

这是一团糟，但它有效：

DECLARE @SQL nvarchar(MAX),
        @CRLF nchar(2) = NCHAR(13) + NCHAR(10);

CREATE TABLE #NullCounts (SchemaName sysname,
                          TableName sysname,
                          ColumnName sysname,
                          NULLCount bigint);

DECLARE @Delimiter nchar(3) = ',' +@CRLF;

SET @SQL = STUFF((SELECT @CRLF + @CRLF +
                         N'WITH Counts AS(' + @CRLF +
                         N'    SELECT N' + QUOTENAME(s.[name],'''') +N' AS SchemaName,' + @CRLF +
                         N'           N' + QUOTENAME(t.[name],'''') +N' AS TableName,' + @CRLF +
                         STRING_AGG(N'           COUNT_BIG(CASE WHEN ' + QUOTENAME(c.[name]) + N' IS NULL THEN 1 END) AS ' + QUOTENAME(c.[name]),@Delimiter) WITHIN GROUP(ORDER BY c.column_id) + @CRLF +
                         N'    FROM ' + QUOTENAME(s.[name]) + N'.' + QUOTENAME(t.[name]) + N' T)' + @CRLF +
                         N'INSERT INTO #NullCounts(SchemaName, TableName, ColumnName, NULLCount)' + @CRLF +
                         N'SELECT SchemaName,' + @CRLF +
                         N'       TableName,' + @CRLF +
                         N'       V.ColumnName,' + @CRLF +
                         N'       V.NULLCount' + @CRLF +
                         N'FROM Counts C' + @CRLF +
                         N'     CROSS APPLY (VALUES' +
                         STUFF(STRING_AGG(N'                        (N' + QUOTENAME(c.[name], '''') + N', C.' + QUOTENAME(c.[name]) + N')',@Delimiter) WITHIN GROUP (ORDER BY c.column_id),1,24,N'') + N')V(ColumnName,NULLCount);'
                  FROM sys.schemas s
                       JOIN sys.tables t ON s.schema_id = t.schema_id
                       JOIN sys.columns c ON t.object_id = c.object_id
                  GROUP BY s.[name], t.[name]
                  FOR XML PATH(N''),TYPE).value('.','nvarchar(MAX)'),1,4,N'');
     
--PRINT @SQL; --This is gunna be way longer than 4,000 characters, so you'll want SELECT

EXEC sys.sp_executesql @SQL;

GO

SELECT *
FROM #NullCounts
ORDER BY SchemaName,
         TableName,
         ColumnName;

GO

DROP TABLE #NullCounts;

是的，我混合STRING_AGG and FOR XML PATH，是的，它有点碍眼，但是打印的（选定的）SQL 会生成一些非常好的语句。见下文：

WITH Counts AS(
    SELECT N'dbo' AS SchemaName,
           N'PerformanceTest' AS TableName,
           COUNT_BIG(CASE WHEN TestID IS NULL THEN 1 END) AS [TestID],
           COUNT_BIG(CASE WHEN TestTarget IS NULL THEN 1 END) AS [TestTarget],
           COUNT_BIG(CASE WHEN TestName IS NULL THEN 1 END) AS [TestName],
           COUNT_BIG(CASE WHEN TimeStart IS NULL THEN 1 END) AS [TimeStart],
           COUNT_BIG(CASE WHEN TimeEnd IS NULL THEN 1 END) AS [TimeEnd],
           COUNT_BIG(CASE WHEN TimeTaken_ms IS NULL THEN 1 END) AS [TimeTaken_ms],
           COUNT_BIG(CASE WHEN TotalRows IS NULL THEN 1 END) AS [TotalRows],
           COUNT_BIG(CASE WHEN RowSets IS NULL THEN 1 END) AS [RowSets],
           COUNT_BIG(CASE WHEN AvgRowsPerSet IS NULL THEN 1 END) AS [AvgRowsPerSet]
    FROM [dbo].[PerformanceTest] T)
INSERT INTO #NullCounts(SchemaName, TableName, ColumnName, NULLCount)
SELECT SchemaName,
       TableName,
       V.ColumnName,
       V.NULLCount
FROM Counts C
     CROSS APPLY (VALUES(N'TestID', C.[TestID]),
                        (N'TestTarget', C.[TestTarget]),
                        (N'TestName', C.[TestName]),
                        (N'TimeStart', C.[TimeStart]),
                        (N'TimeEnd', C.[TimeEnd]),
                        (N'TimeTaken_ms', C.[TimeTaken_ms]),
                        (N'TotalRows', C.[TotalRows]),
                        (N'RowSets', C.[RowSets]),
                        (N'AvgRowsPerSet', C.[AvgRowsPerSet]))V(ColumnName,NULLCount);

WITH Counts AS(
    SELECT N'dbo' AS SchemaName,
           N'someTable' AS TableName,
           COUNT_BIG(CASE WHEN id IS NULL THEN 1 END) AS [id],
           COUNT_BIG(CASE WHEN SomeCol IS NULL THEN 1 END) AS [SomeCol]
    FROM [dbo].[someTable] T)
INSERT INTO #NullCounts(SchemaName, TableName, ColumnName, NULLCount)
SELECT SchemaName,
       TableName,
       V.ColumnName,
       V.NULLCount
FROM Counts C
     CROSS APPLY (VALUES(N'id', C.[id]),
                        (N'SomeCol', C.[SomeCol]))V(ColumnName,NULLCount);

是的，我真的花了最后 45 分钟写下所有这些……

老实说，这不是入门级的，如果你不理解它，你就不应该使用它；而且，我非常怀疑您是否会找到一个不同的入门级解决方案，并且性能与此一样。A CURSOR，例如，虽然可能更容易理解，但是这样做会很慢.

Caveat:如果您的数据库中有任何已弃用的数据类型（即text）这将会失败。如果是这种情况，您需要确保从查询中消除它们WHERE。但是，我建议您修复数据类型（text，例如，已被弃用 15 年）。