我需要使用T-SQL从以下段落中解析{}
中包含的字符串,然后显示它们。
这是一个{Term1}的测试句子。有时,{Term2}可以是{Phrase Term3}之类的单词或短语。 {Term2}重复。某些术语可能是{Term2}等另一个术语的复数形式。这是一个真正的{Simple} Term。
期望的结果:
Term1
Term2
Phrase Term3
Term2
Term2
Simple
答案 0 :(得分:3)
你可以使用多语句表值函数来做到这一点,但我真的认为这种类型的解析更好地留给更强大的语言。这将处理标记{up to 255 characters}
并输入最多约8,000个字符的字符串,具体取决于SQL Server的版本。如果您需要更多内容,请将sys.all_columns
替换为your own numbers table。请注意,我没有采取任何措施来保护无效的令牌序列......
CREATE FUNCTION dbo.ParseTokens
(
@string NVARCHAR(MAX),
@token1 NVARCHAR(255),
@token2 NVARCHAR(255)
)
RETURNS @t TABLE([Index] INT IDENTITY(1,1), Item NVARCHAR(255))
AS
BEGIN
INSERT @t(Item)
SELECT SUBSTRING(x, 1, COALESCE(NULLIF(CHARINDEX(@token2, x)-1,-1),255))
FROM
(
SELECT Number, x = SUBSTRING(@string, Number,
CHARINDEX(@token1, @string + @token1, Number) - Number)
FROM
(
SELECT ROW_NUMBER() OVER (ORDER BY [object_id])
FROM sys.all_columns
) AS n(Number) WHERE Number <= CONVERT(INT, LEN(@string))
AND SUBSTRING(@token1 + @string, Number, LEN(@token1)) = @token1
) AS y
ORDER BY Number OPTION (MAXDOP 1);
DELETE @t WHERE [Index] = 1;
RETURN;
END
GO
示例用法 - 在独立字符串上:
DECLARE @x NVARCHAR(MAX);
SET @x = N'foo{bar} and think {splunge}';
SELECT Item FROM dbo.ParseTokens(@x, '{', '}') ORDER BY [Index];
结果:
Item
-------
bar
splunge
示例用法 - 针对表:
DECLARE @x TABLE(ID INT IDENTITY(1,1), n NVARCHAR(MAX));
INSERT @x SELECT N'Here is a test sentence with a {Term1}. Sometime, a {Term2}
could be a word or phrase like {Phrase Term3}. {Term2} is repeated. Some Terms
could be a plural form of a another Term like {Term2}s. Here is a real
{Simple} Term.';
INSERT @x SELECT N'Hello {foo} there {bar} ...';
SELECT t.ID, p.Item
FROM @x AS t
CROSS APPLY dbo.ParseTokens(t.n, '{', '}') AS p;
结果:
ID Item
---- ------------
1 Term1
1 Term2
1 Phrase Term3
1 Term2
1 Term2
1 Simple
2 foo
2 bar
答案 1 :(得分:3)
您可以将所有{
替换为start元素,将所有}
替换为closing元素,然后将XML替换为标记,从而将字符串转换为XML。
declare @S nvarchar(max)
set @S = N'Here is a test sentence with a {Term1}. Sometime, a {Term2} could be a word or phrase like {Phrase Term3}. {Term2} is repeated. Some Terms could be a plural form of a another Term like {Term2}s. Here is a real {Simple} Term.'
select T.N.value('text()[1]', 'nvarchar(max)') as Token
from (select cast(replace(replace(@S, N'{', N'<token>'), N'}', N'</token>') as xml)) as S(X)
cross apply S.X.nodes('token') as T(N)