CODEWITHSUNDEEP
sqlsql-serversql-server-2008
Gitz
Gitz

Reputation: 820

Separate words by group wise for each row in SQL

I have a string something like 

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

And I have a table 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

I want to check every comma separated word with table Group_words and find the wrong combination.

For the above string result should be : "No People,Day,side view,looking at camera,snow,mountain,tranquil scene,tranquility,walking,water"

I know its to complicated but simply I want to remove the not matching words from sting which is available into table Group_words.

Upvotes: 1

Views: 163

Answers (1)

Felix Pamittan
Felix Pamittan

Reputation: 31879

Wow, you should be re-designing your tables. Anyway, here is my attempt using Jeff Moden's DelimitedSplit8k.

I believe you now have this function since I answered one of your previous questions that also uses this function.

First, you want to split your @string input into separate rows. You should also split the Group_Words table.

After that you do a LEFT JOIN to get the matching categories. Then you eliminate the invalid words.

See it in action here: 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, '')

OUTPUT:

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

If your @string input has more than 8000 characters, the DelimitedSplit8K will slow down. You can use other splitters instead. Here is one taken for Sir Aaron Bertrands's article.

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

Upvotes: 2

Related Questions