SQL Server：在特定字符后检查大写或小写

Question

我有像'John is my name; Ram is my name; Adam is my name'的数据。

我的规则是这样的，每个在;之后的第一个字母应该是大写字母。

如何选择所有符合规则的值？

Answer 1

的其他答案显示如何将行转换为与您的模式相匹配的内容。

如果你只是想select符合您所描述的模式的行，你可以使用patindex()或like使用区分大小写的排序规则（或使用collate申请一个）。

这里假定除了规则之外，每个分号后面的字母必须是大写字母，第一个字母也应该是大写字母。如果不是这种情况，只需删除where中的第一个子句即可。

select * 
from t 
where patindex('[ABCDEFGHIJKLMNOPQRSTUVWXYZ]%', val collate latin1_general_cs_as) = 1 
    and patindex('%; [^ABCDEFGHIJKLMNOPQRSTUVWXYZ]%', val collate latin1_general_cs_as) = 0 

---

​​

---

测试设置：

create table t (id int not null identity(1,1),val varchar(256)) 
insert into t values 
('John is my name; Ram is my name; Adam is my name') 
,('john is my name; ram is my name; adam is my name') 

rextester演示：http://rextester.com/DBGIS10645

上述两种返回的：

+----+--------------------------------------------------+ 
| id |      val      | 
+----+--------------------------------------------------+ 
| 1 | John is my name; Ram is my name; Adam is my name | 
+----+--------------------------------------------------+ 

Answer 2

你可能会与XML招这样

DECLARE @YourString VARCHAR(100)='John is my name; Ram is my name; Adam is my name'; 
WITH Splitted AS 
(
    SELECT CAST('<x>' + REPLACE((SELECT REPLACE(@YourString,'; ','$$SplitHere$$') AS [*] FOR XML PATH('')),'$$SplitHere$$','</x><x>')+ '</x>' AS XML) AS Casted 
) 
,DerivedTable AS 
(
    SELECT ROW_NUMBER() OVER(ORDER BY (SELECT NULL)) AS PartNr 
      ,x.value(N'text()[1]',N'nvarchar(max)') AS Part 
    FROM Splitted 
    CROSS APPLY Casted.nodes(N'/x') AS X(x) 
) 
SELECT PartNr 
     ,Part 
     ,CASE WHEN ASCII(LEFT(Part,1)) BETWEEN ASCII('A') AND ASCII('Z') THEN 1 ELSE 0 END AS FirstIsCapital 
FROM DerivedTable; 

Nr Part   FirstLetterIsCaptial 
---------------------------------------- 
1 John is my name  1 
2 Ram is my name  1 
3 Adam is my name  1 

我不知道你的最终目标是什么...找份，其中第一个字母，结果这个分裂不是资本？确保你的规则满员？

但是：
最好的是，以此来纠正你的设计，并将这些部件在1:n相关的边桌。

Answer 3

丑陋的解决方案的点点，但你可以给一个尝试...

Declare @str nvarchar(max) = 'John is my name; Ram is my name; Adam is my name' 

Declare @xml as xml 
Set @xml = cast(('<X>'+replace(@str,';' ,'</X><X>')+'</X>') as xml) 
Select * from (
    Select RowN = Row_Number() over (order by (SELECT NULL)), LTrim(RTrim(N.value('.', 'nvarchar(MAX)'))) as value FROM @xml.nodes('X') as T(N) -- this is to split if you are using sql server 2016 you can use string_Split 
) a 
Where unicode(substring(a.[value],1,1)) = unicode(upper(substring(a.[value],1,1))) 

想法是分割字符串与Unicode值检查，看它是否是上还是不

Answer 4

注意：标准的做法是在应用程序中使用C＃/ VB [.Net]执行此操作。

[1]解决方案：

DECLARE @Source NVARCHAR(100) = N'john is my name; Ram is my name; adam is my name' 

SELECT z.Sentence 
FROM (VALUES (CONVERT(XML, N'<root><i>' + REPLACE(@Source, N';', N'</i><i>;') + N'</i></root>'))) AS x(XmlCol) 
CROSS APPLY x.XmlCol.nodes(N'/root/i') AS y(XmlCol) 
CROSS APPLY (VALUES(y.XmlCol.value('(text())[1]', 'NVARCHAR(100)'))) AS z(Sentence) 
WHERE SUBSTRING(z.Sentence, NULLIF(PATINDEX('%[a-z]%', z.Sentence), 0), 1) LIKE '%[a-z]%' COLLATE Latin1_General_BIN 
ORDER BY ROW_NUMBER() OVER(ORDER BY y.XmlCol) 

在这种情况下，结果将是

john is my name 
; adam is my name 

[2]如果你正在试图利用第一个字母从每一个一句然后我会用下列溶液（见注释广告行结束）：

DECLARE @Source NVARCHAR(100) = N'john is my name; ram is my name; adam is my name' 

SELECT (
    SELECT u.NewSentence AS '*' 
    FROM (VALUES (CONVERT(XML, N'<root><i>' + REPLACE(@Source, N';', N';</i><i>') + N'</i></root>'))) AS x(XmlCol) -- It convert source string into XML. Every ; acct as a delimiter for sentence. End results will be like this <root><i>john...;</i><i> ram ....</i>...</root> 
    CROSS APPLY x.XmlCol.nodes(N'/root/i') AS y(XmlCol) -- It decompose original XML into separate sentences as XML 
    CROSS APPLY (VALUES(y.XmlCol.value('(text())[1]', 'NVARCHAR(100)'))) AS z(Sentence) -- ... AS NVARCHAR(100) 
    CROSS APPLY (VALUES(PATINDEX('%[a-z]%', z.Sentence))) AS t(FirstLetterIndex) -- It finds index of first letter 
    CROSS APPLY (VALUES(IIF(t.FirstLetterIndex > 0, STUFF(z.Sentence, t.FirstLetterIndex, 1, UPPER(SUBSTRING(z.Sentence, t.FirstLetterIndex, 1))), z.Sentence))) AS u(NewSentence) -- It replace every first letter with the capitalized version/UPPER(...) 
    ORDER BY ROW_NUMBER() OVER(ORDER BY y.XmlCol) -- All sentences should be ordered by original position within source string 
    FOR XML PATH('') -- It concatenates all sentences back in one string 
) 

例如，如果源串是N'john is my name; ram is my name; adam is my name'那么结果将是N'John is my name; Ram is my name; Adam is my name'。

Demo

注：该解决方案的工作（以及基于XML切碎所有其他解决方案）如果源字符串不包括一些XML字符保留（如<）。让我知道如果这是你的情况。

Answer 5

你可以创建一个这样的功能。

Create FUNCTION SPLITTER ( 
    @textData NVARCHAR(MAX), 
    @Delimeter NVARCHAR(MAX)) RETURNS @RtnValue TABLE (
    Data NVARCHAR(MAX)) AS BEGIN 
    DECLARE @index INT DECLARE @data nvarchar(1000) DECLARE @firstCharacter char 
    SET @index = CHARINDEX(@Delimeter,@textData) 

    WHILE (@index>0) 
    BEGIN 
       set @data = LTRIM(RTRIM(SUBSTRING(@textData, 1, @index - 1)))  set @firstCharacter = SUBSTRING(@data,1,1); 
       if UNICODE(@firstCharacter) = UNICODE(upper(@firstCharacter))  begin   INSERT INTO @RtnValue (data) SELECT @data  end; 

     SET @textData = SUBSTRING(@textData, @index + DATALENGTH(@Delimeter)/2, LEN(@textData)) 

     SET @index = CHARINDEX(@Delimeter, @textData) 
    END 
     set @data = @textData set @firstCharacter = SUBSTRING(@data,1,1); 
      if UNICODE(@firstCharacter) = UNICODE(upper(@firstCharacter)) begin  INSERT INTO @RtnValue (data) SELECT @data end; 

    RETURN END 

使用这样

SELECT * FROM分路器（ '约翰是我的名字，拉姆是我的名字;亚当是我的名字'， ';'）

Answer 6

你可以抓住的NGrams8K副本，并做到这一点：

-- note that I made the 3rd item start with lower-case 
DECLARE @YourString VARCHAR(100)='John is my name; Ram is my name; adam is my name'; 

WITH D(n) AS 
(
    SELECT 0 UNION ALL SELECT position 
    FROM dbo.NGrams8k(@yourstring,1) WHERE token = ';' 
), 
TOKEN(token) AS 
(
    SELECT LTRIM(SUBSTRING(@YourString, N+1, 
      ISNULL(NULLIF(CHARINDEX(';', @YourString, N+1),0), 101)-(N+1))) 
    FROM D 
) 
SELECT token, 
     FirstLetterIsCaptial = IIF(ASCII(SUBSTRING(token,1,1)) BETWEEN 65 AND 90, 1, 0) 
FROM TOKEN; 

结果

token    FirstLetterIsCaptial 
------------------ -------------------- 
John is my name 1 
Ram is my name  1 
adam is my name 0