对于SQL中的每一行，按组明智地分隔单词

Question

我有一个类似的字符串

No People,Day,side view,looking at camera,snow,mountain,tranquil scene,tranquility,Night,walking,water,Two Person,looking Down

我有一张桌子 Group_words

Group                                                                                                                                                            Category
---------------------------------------------------------------------------------------------------------------------------------------------------------------- --------------------
No People,One Person,Two Person,Three Person,Four Person,five person,medium group of people,large group of people,unrecognizable person,real people              People
Day,dusk,night,dawn,sunset,sunrise                                                                                                                               Weather
looking at camera,looking way,looking sideways,looking down,looking up                                                                                           View Angle

我想用表格检查每个逗号分隔的单词，Group_words并找到错误的组合。

对于上述字符串，结果应为：“ 没有人，天，侧视图，看着相机，雪，山，宁静的景象，宁静，步行，水 ”

• Night已删除，因为Day该字符串中可用。
• Two Person已删除，因为No People该字符串中可用。
• looking Down已删除，因为looking at camera该字符串中可用。

我知道它很复杂，但我只是想从表中可用的字符串中删除不匹配的单词Group_words。

Answer 1

哇，您应该重新设计桌子。无论如何，这是我尝试使用Jeff Moden的
DelimitedSplit8k

。

首先，您想将@string输入拆分为单独的行。您还应该拆分Group_Words表。

之后，您需要执行一次操作LEFT JOIN以获取匹配的类别。然后，您消除无效的单词。

在此处查看其运行情况： SQL Fiddle

DECLARE @string VARCHAR(8000)
SET @string = 'No People,Day,side view,looking at camera,snow,mountain,tranquil scene,tranquility,Night,walking,water,Two Person,looking Down'

-- Split @string variable
DECLARE @tbl_string AS TABLE(ItemNumber INT, Item VARCHAR(8000))
INSERT INTO @tbl_string
SELECT
    ItemNumber, LTRIM(RTRIM(Item))
FROM dbo.DelimitedSplit8K(@string, ',')

-- Normalize Group_Words
DECLARE @tbl_grouping AS TABLE(Category VARCHAR(20), ItemNumber INT, Item VARCHAR(8000))
INSERT INTO @tbl_grouping
SELECT
    w.Category, s.ItemNumber, LTRIM(RTRIM(s.Item))
FROM Group_Words w
CROSS APPLY dbo.DelimitedSplit8K(w.[Group], ',')s

;WITH Cte AS(
    SELECT      
        s.ItemNumber,
        s.Item,
        g.category,     
        RN = ROW_NUMBER() OVER(PARTITION BY g.Category ORDER BY s.ItemNumber)
    FROM @tbl_string s
    LEFT JOIN @tbl_grouping g
        ON g.Item = s.Item
)
SELECT STUFF((
        SELECT ',' + Item
        FROM Cte
        WHERE 
            RN = 1
            OR Category IS NULL
        ORDER BY ItemNumber
        FOR XML PATH(''), TYPE).value('.', 'VARCHAR(MAX)'),
    1, 1, '')

输出 ：

|                                                                                                  |
|--------------------------------------------------------------------------------------------------|
| No People,Day,side view,looking at camera,snow,mountain,tranquil scene,tranquility,walking,water |

如果您@string输入的内容超过8000字符，DelimitedSplit8K将会变慢。您可以改用其他拆分器。这是亚伦·伯特兰兹爵士的文章的摘录。

CREATE FUNCTION dbo.SplitStrings_XML
(
   @List       NVARCHAR(MAX),
   @Delimiter  NVARCHAR(255)
)
RETURNS TABLE
WITH SCHEMABINDING
AS
   RETURN 
   (  
      SELECT Item = y.i.value('(./text())[1]', 'nvarchar(4000)')
      FROM 
      ( 
        SELECT x = CONVERT(XML, '<i>' 
          + REPLACE(@List, @Delimiter, '</i><i>') 
          + '</i>').query('.')
      ) AS a CROSS APPLY x.nodes('i') AS y(i)
   );
GO