从 SQL 表中选择前 10 个单词

Question

我需要从 SQL 表字段中选择查询前 10 个单词。

我有字段名称：描述，所以我需要描述字段中的前 10 个单词。

EX： 说明：我对这个答案感觉不好；我不喜欢 GROUP BY，我宁愿不做数字表。希望有人可以接我。

输出：我对这个答案感觉不太好；我不喜欢

Answer 1

另一种选择是使用 CROSS APPLY 和一些 XML。 CROSS APPLY 中的逻辑可以很容易地移植到 UDF（标量或表值）中。

示例

Declare @YourTable table (ID int,LongDesc varchar(max))
Insert into @YourTable values
(1,'I don''t feel good about this answer; I don''t like the GROUP BY and I would rather not do the table of numbers. Hopefully somebody can pick me up on this.')

Select A.ID
      ,ShortDesc = B.S
 From  @YourTable A
 Cross Apply (
                Select S = xDim.value('/x[1]','varchar(100)')+' '
                          +xDim.value('/x[2]','varchar(100)')+' '
                          +xDim.value('/x[3]','varchar(100)')+' '
                          +xDim.value('/x[4]','varchar(100)')+' '
                          +xDim.value('/x[5]','varchar(100)')+' '
                          +xDim.value('/x[6]','varchar(100)')+' '
                          +xDim.value('/x[7]','varchar(100)')+' '
                          +xDim.value('/x[8]','varchar(100)')+' '
                          +xDim.value('/x[9]','varchar(100)')+' '
                          +xDim.value('/x[10]','varchar(100)')
                From  (Select Cast('<x>' + replace((Select replace(A.LongDesc,' ','§§Split§§') as [*] For XML Path('')),'§§Split§§','</x><x>')+'</x>' as xml) as xDim) as A 
             ) B

退货

ID  ShortDesc
1   I don't feel good about this answer; I don't like

Answer 2

left() 函数可以吗？

从 your_table 中选择 left(Description, 10);

Answer 3

您可以查看this 答案以了解如何在SQL Server 的上下文中实现和使用.net 函数。在这里，我使用SQLSCLR实现.netRegex.Match函数。

DECLARE @DataSource TABLE
(
    [Description] NVARCHAR(MAX)
);

INSERT INTO @DataSource ([Description])
VALUES ('word01, word02,         word03, word04, word05, word06')
      ,('word01,word02, word03, word04, word05, word06')
      ,('word01!word02, word03: word04, word05, word06');

SELECT *
FROM @DataSource DS
CROSS APPLY [dbo].[fn_Utils_RegexMatches] ([Description], '^(\w+\b.*?){3}') RM;

这将为您提供以下输出：

这太详细了（并且只提取了第一个three 字）。您的最终查询可能是这样的：

SELECT DS.[Description], RM.[CaptureValue]
FROM @DataSource DS
CROSS APPLY [dbo].[fn_Utils_RegexMatches] ([Description], '^(?n)(\w+\b.*?){3}') RM;

无论如何，使用正则表达式你可以使用任何你想要的分隔符，但更重要的是你可以在 T-SQL 的上下文中执行 .net 代码，这是巨大的。

您可以从这里阅读和学习很多内容。前面的答案实现起来更简单、更快捷。

Answer 4

你可以试试这个

declare @message nvarchar(max) = 'I don''t feel good about this answer; I don''t like the GROUP BY and I would rather not do the table of numbers. Hopefully somebody can pick me up on this'

SELECT left(@message,
CHARINDEX(' ', @message,
  CHARINDEX(' ', @message,
    CHARINDEX(' ',@message,
     CHARINDEX(' ',@message,
      CHARINDEX(' ',@message,
       CHARINDEX(' ',@message,
        CHARINDEX(' ',@message,
         CHARINDEX(' ',@message,
          CHARINDEX(' ',@message,
            CHARINDEX(' ',@message) + 1 ) + 1 ) + 1 ) + 1) + 1) + 1) + 1 )+ 1 ) + 1))

问候

Answer 5

用大数据试试这个。

DECLARE @T TABLE(COL1 NVARCHAR(MAX))
INSERT INTO @T VALUES
('I dont feel good about this answer; I dont like the GROUP BY and I would rather not do the table of numbers. Hopefully somebody can pick me up on this.')

;WITH CTE AS
(
select col1,substring(col1,1,CHARINDEX(' ',col1))Words
,substring(col1,CHARINDEX(' ',col1)+1,len(col1))Residue
, 1 rn from @t

union ALL
select col1,substring(Residue,1,CHARINDEX(' ',Residue))
,substring(Residue,CHARINDEX(' ',Residue)+1,len(Residue))
,rn+1
from cte
where len(Words)>0 and rn<10
)
SELECT Words FROM cte

Answer 6

我知道你已经有了答案，但我只是想换一种方式。

DECLARE @WORD VARCHAR(8000)= 'I don''t feel good about this answer; I don''t like the GROUP BY and I would rather not do the table of numbers. Hopefully somebody can pick me up on this.'
DECLARE @LOOP INT = 1
DECLARE @STR_POS INT = 1
DECLARE @CI INT
DECLARE @RES_STR VARCHAR(8000) = ''

WHILE(@LOOP<=10)
    BEGIN
        SELECT @CI = CHARINDEX(' ',@WORD,@STR_POS)
        SELECT @RES_STR = @RES_STR +' '+ SUBSTRING(@WORD,@STR_POS,(CHARINDEX(' ',@WORD,@STR_POS)-@STR_POS))
        SET @STR_POS = @CI+1
        SET @LOOP=@LOOP+1

    END
PRINT @RES_STR

Answer 7

这并没有给你一个行集解决方案，而是另一种选择，使用类似的技术用更少的代码解析前十个单词，纯粹是为了说明。我想这可以转换为用户定义的函数，但我们知道他们表现不佳。它可能可以使用 CrossApply 组合以消除对 UDF 的需要。我必须将此归功于 xml 语法的帮助：https://www.mssqltips.com/sqlservertip/1771/splitting-delimited-strings-using-xml-in-sql-server/

DECLARE 
@xml as xml,
@str as varchar(1000)= 'I don''t feel good about this answer; I don''t like the GROUP BY and I would rather not do the table of numbers. Hopefully somebody can pick me up on this.',
@delimiter as char(1) = char(32),
@xmlspace  as char(13) = '<y>&#160;</y>'

SET @xml = cast(('<x>'+replace(@str,@delimiter,@xmlspace)+'</x>') as xml)

SELECT TOP 10 N.value('.', 'varchar(50)') FROM @xml.nodes('x') as T(N)