“=”/like 子句中存储过程中的 REGEX

REGEX in stored procedure in "="/like clause

我创建了一个包含 3 个内部联接的存储过程(请不要评判我 :D):

SELECT 
    T.Name, A.Value 
FROM
    Table AS T
INNER JOIN
    TableAnotherTable AS TA ON TA.ID_1 = T.ID_1
INNER JOIN
    AnotherTable AS A ON A.ID_2 = TA.ID_2
INNER JOIN
    EndTable AS ET ON ET.Value = A.Value

TableAnotherTableTableAnotherTable 之间的关联 table。因此,此存储过程在某些情况下有效(如果 ET.Value 等于 等于 A.Value)。

但是如果:

ET.Value = 'SOME\THING\RIGHT\<ABC>\THERE'
A.Value= 'SOME\THING\RIGHT\{DEFGHILM}\THERE'

存储过程也必须有效。

如何将内部连接子句(带有“=”)的最后一个 ON 更改为:

Regex("pattern1", E.Value) like Regex("pattern", A.Value)

其中 patternpattern1({.*?})(<.*?>)?

谢谢

示例:

ET.Value = 'HI'
A.Value = 'HI'
true

ET.Value = 'SOME\THING\RIGHT\<ABC>\THERE'
A.Value = 'SOME\THING\RIGHT\{DEFGHILMNOP}\THERE'
true

ET.Value = 'HOME\SWEET\HOME\<12345>\'
A.Value = 'HOME\SWEET\HOME\{4875928346}\'
true

ET.Value = 'EXAMPLE\<1234>'
A.Value = 'EG\{1234}'
false

ET.Value = 'dog'
A.Value = 'cat'
false

您可以比较没有 <> 和 {} 之间部分的文本:

这里是连接条件:

LEFT(ET.VALUE,CHARINDEX('<', ET.VALUE)-1) +  RIGHT(ET.VALUE,LEN(ET.VALUE)- CHARINDEX('>', ET.VALUE))
=
LEFT(A.Value,CHARINDEX('{', A.Value)-1) +  RIGHT(A.Value,LEN(A.Value)- CHARINDEX('}', A.Value))

但它在性能方面可能很糟糕,因为您转换了用于加入 table 的密钥。

很遗憾,SQL 服务器中没有正则表达式替换功能。您可以获得的最接近的方法是使用 PATINDEX 找到 <{}>' characters and STUFF` 之间的所有内容:

SELECT t.*, CASE WHEN val1_new = val2_new THEN 'match' END
FROM (VALUES
    ('SOMETHING\<1>\SOMETHING',  'SOMETHING\{2}\SOMETHING'),
    ('SOMETHING\BLABLA\<1>\BLA', 'SOMETHING\BLABLA\{2}\BLA'),
    ('SOME\<1>\THING\BLA',       'SOMETHING\{2}\BLABLA'),
    ('dog',                      'dog')
) AS t(val1, val2)
CROSS APPLY (
    SELECT
        PATINDEX('%[<{]%', val1) AS val1_pos1,
        PATINDEX('%[>}]%', val1) AS val1_pos2,
        PATINDEX('%[<{]%', val2) AS val2_pos1,
        PATINDEX('%[>}]%', val2) AS val2_pos2
) AS ca1
CROSS APPLY (
    SELECT
        CASE WHEN val1_pos1 > 0 AND val1_pos2 > 0 THEN STUFF(val1, val1_pos1, val1_pos2 - val1_pos1 + 1, '') ELSE val1 END,
        CASE WHEN val2_pos1 > 0 AND val2_pos2 > 0 THEN STUFF(val2, val2_pos1, val2_pos2 - val2_pos1 + 1, '') ELSE val2 END
) AS ca3(val1_new, val2_new)