将Oracle正则表达式查询转换为MSSQL

Question

您好，我需要将此查询从Oracle数据库转换为MSSQL，并获得完全相同的结果：

WHEN REGEXP_LIKE(E.EVENTS, 'selfServe:[^:]*:completed[:]')。

我的以下尝试均失败了：

WHEN EVENTS LIKE '%[:]selfServe[:][^:]%[:]completed[:]%'

EVENTS LIKE '%[:]selfServe[:]%[^:][:]completed[:]%'

WHERE PATINDEX('selfServe:[^:]*:completed[:]', EVENTS) != 0

WHERE PATINDEX('selfServe:[^:]%:completed[:]', EVENTS) != 0。

示例：

这不应该匹配：

OpenQ,
Payment,
Payment:selfServe:Payment-Cancel_Scheduled:initiated::,
Payment:authentication:Authentication:initiated::,
Payment:authentication:Authentication:completed::,
HUP

这应该匹配：

OpenQ2,
Payment,
Payment:selfServe:Payment:initiated::,
Payment:authentication:Authentication:initiated::,
Payment:authentication:Authentication:initiated::,
Payment:authentication:Authentication:completed::,
Payment:selfServe:Payment:completed::,
HUP

在第一种情况下，我有authentication:completed，但没有selfServe:Payment:completed。

Answer 1

我个人将使用字符串拆分功能来解决此问题，该功能在逗号分隔的字符串之一中寻找适当的selfServe:...:completed。如果您使用的是SQL Server的较新版本，则可以使用内置的string_split函数来做到这一点：

string_split

declare @t table(id int,t varchar(1000));
insert into @t values
 (1,'OpenQ,Payment,Payment:selfServe:Payment-Cancel_Scheduled:initiated::,Payment:authentication:Authentication:initiated::,Payment:authentication:Authentication:completed::,HUP')
,(2,'OpenQ2,Payment,Payment:selfServe:Payment:initiated::,Payment:authentication:Authentication:initiated::,Payment:authentication:Authentication:initiated::,Payment:authentication:Authentication:completed::,Payment:selfServe:Payment:completed::,HUP');


select t.id
      ,s.value
      ,t.t
from @t as t
    cross apply string_split(t.t,',') as s
where case when patindex('%:selfServe:%',s.value) > 0
                and patindex('%:completed:%',s.value) > 0
            then 1
            else 0
            end = 1;

输出

+----+---------------------------------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| id |                 value                 |                                                                                                                          t                                                                                                                           |
+----+---------------------------------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
|  2 | Payment:selfServe:Payment:completed:: | OpenQ2,Payment,Payment:selfServe:Payment:initiated::,Payment:authentication:Authentication:initiated::,Payment:authentication:Authentication:initiated::,Payment:authentication:Authentication:completed::,Payment:selfServe:Payment:completed::,HUP |
+----+---------------------------------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+

---

如果没有，则需要滚动自己的字符串分隔符。我将假设您在这里可能有一些较长的字符串（超过4000个字符），因此使用的是基于XML的拆分器，该拆分器与max数据类型配合使用非常好。

当您在BI工具中执行此操作（假设该工具将不允许您在数据库中创建新的表值函数）时，您将需要一个相当复杂的语句来处理数据，其中包含内联字符串拆分器：

滚动自己的

declare @t table(id int,t varchar(1000));
insert into @t values
 (1,'OpenQ,Payment,Payment:selfServe:Payment-Cancel_Scheduled:initiated::,Payment:authentication:Authentication:initiated::,Payment:authentication:Authentication:completed::,HUP')
,(2,'OpenQ2,Payment,Payment:selfServe:Payment:initiated::,Payment:authentication:Authentication:initiated::,Payment:authentication:Authentication:initiated::,Payment:authentication:Authentication:completed::,Payment:selfServe:Payment:completed::,HUP');

with s as
(       -- Convert the string to an XML value, replacing the delimiter with XML tags
    select id
          ,t
          ,convert(xml,'<x>' + replace((select '' + t for xml path('')),',','</x><x>') + '</x>').query('.') as s
    from @t
)
select id
        ,item
        ,t     -- Select the values from the generated XML value by CROSS APPLYing to the XML nodes
from(select id
            ,t
            ,n.x.value('.','nvarchar(max)') as item
    from s
            cross apply s.nodes('x') as n(x)
    ) a
where case when patindex('%:selfServe:%',a.item) > 0
                and patindex('%:completed:%',a.item) > 0
            then 1
            else 0
            end = 1;

输出

+----+-----------------------------------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| id |                  item                   |                                                                                                                          t                                                                                                                           |
+----+-----------------------------------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
|  2 |   Payment:selfServe:Payment:completed:: | OpenQ2,Payment,Payment:selfServe:Payment:initiated::,Payment:authentication:Authentication:initiated::,Payment:authentication:Authentication:initiated::,Payment:authentication:Authentication:completed::,Payment:selfServe:Payment:completed::,HUP |
+----+-----------------------------------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+