Flatten association table to multi-value column? - sql

I have a table with just product ID's and category ID's (products can be in more than one category). How can I flatten the category ID's into a product column so I end us with this:
id | name | desc | categories
1 | test1 | lorem | 1,3,4,23
2 | test2 | ipsom | 4,6,24
It is like I need to loop into a separate table for the categories column. How can I do this or is there a better way?

I created an CLR aggregate function that takes a varchar column and returns all its values separated by commas. In other words, it joins several strings into a comma-separated list. I am sure its performance is way better than any T-Sql trick.
As any aggregate function, it can be used in combination with group by. For example:
SELECT id, name, desc, JoinStrings(CONVERT(VARCHAR(20), category_id))
FROM product p
INNER JOIN category_products c ON p.category_id = c.category_id
GROUP BY id, name, desc
Here's the C# code to create the CLR assembly into Sql Server 2008:
using System;
using System.Data;
using System.Data.SqlClient;
using System.Data.SqlTypes;
using Microsoft.SqlServer.Server;
[Serializable]
[Microsoft.SqlServer.Server.SqlUserDefinedAggregate(Format.UserDefined, IsInvariantToDuplicates=false, IsInvariantToOrder=false, IsInvariantToNulls=true, MaxByteSize=-1)]
public struct JoinStrings : IBinarySerialize
{
private char[] sb;
private int pos;
public void Init()
{
sb = new char[512000];
pos = 0;
}
public void Accumulate(SqlString Value)
{
if (Value.IsNull) return;
char[] src = Value.ToString().ToCharArray();
Array.Copy(src, 0, sb, pos, src.Length);
pos += src.Length;
sb[pos] = ',';
pos++;
}
public void Merge(JoinStrings Group)
{
Accumulate(Group.Terminate());
}
public SqlString Terminate()
{
if (pos <= 0)
return new SqlString();
else
return new SqlString(new String(sb, 0, pos-1));
}
public void Read(System.IO.BinaryReader r)
{
this.Init();
pos = r.ReadInt32();
r.Read(sb, 0, pos);
}
public void Write(System.IO.BinaryWriter w)
{
w.Write(pos);
w.Write(sb, 0, pos);
}
}
Here's the code to create the function (although deploying from Visual Studio should do it automatically):
CREATE AGGREGATE [dbo].[JoinStrings]
(#s [nvarchar](4000))
RETURNS[nvarchar](max)
EXTERNAL NAME [YouAssemblyName].[JoinStrings]

There's no in-built way to do it in MSSQL.
Simulating group_concat MySQL function in Microsoft SQL Server 2005? has a good description of how to go about implementing a workaround.

I would suggest using a Recursive CTE. I believe that it would be something like this:
select productid, categoryid,
row_number() over (partition by id order by categoryid) as rownum
into #tabletorecurse
from TABLENAME
with finaloutput as
(
select productid as id, name, desc, categoryid as categories, rownum
from #tabletorecurse
join PRODUCTTABLE
on PRODUCTTABLE.id = #tabletorecurse.productid
where rownum = 1
union all
select tr.id, tr.name, tr.desc,
finaloutput.categories + ', ' + tr.categoryid, tr.rownum
from #tabletorecurse as tr
join finaloutput
on finaloutput.rownum + 1 = tr.rownum
and finaloutput.id = tr.productid
)
select id, name, desc, categories
from finaloutput
join
(
select max(rownum) as maxrow, id
from finaloutput
group by id
) as maxvalues
on maxvalues.id = finaloutput.id
and maxvalues.maxrow = finaloutput.rownum

Use a function.
This does a lookup to text so you will need to adapt.
The COALESCE is just to put a ,.
This is from a large scale production application - it works and it fast.
Function was questioned by JustinPony as function is slow
I am hitting some tables of million of records but only returning 100 rows.
The function is only applied to the hundred rows.
usage:
select top 5 sID, ( select [dbo].[JoinMVEnum](docSVsys.sID, '140') ) as [Flag Issue]
from docSVsys
function
SET ANSI_NULLS ON
GO
SET QUOTED_IDENTIFIER ON
GO
CREATE FUNCTION [dbo].[JoinMVText]
(
#sID int,
#fieldID tinyint
)
RETURNS VARCHAR(MAX)
AS
BEGIN
DECLARE #MVtextList varchar(max)
SELECT #MVtextList = COALESCE(#MVtextList + '; ', '') + docMVtext.value
FROM docMVtext with (nolock)
WHERE docMVtext.sID = #sID and fieldID = #fieldID
RETURN #MVtextList
END
GO

Related

sql server, replace chars in string with values in table

how can i replace values in string with values that are in a table?
for example
select *
into #t
from
(
select 'bla'c1,'' c2 union all
select 'table'c1,'TABLE' c2 union all
select 'value'c1,'000' c2 union all
select '...'c1,'' c2
)t1
declare #s nvarchaR(max)='this my string and i want to replace all values that are in table #t'
i have some values in my table and i want to replace C1 with C2 in my string.
the results should be
this my string and i want to replace all 000 that are in TABLE #t
UPDATE:
i solved with a CLR
using System;
using Microsoft.SqlServer.Server;
using System.Data.SqlTypes;
using System.Data.Linq;
namespace ReplaceValues
{
public partial class Functions
{
[SqlFunction
(
//DataAccess = DataAccessKind.Read,
SystemDataAccess = SystemDataAccessKind.Read
)
]
public static string ReplaceValues(string row, string delimitator, string values, string replace/*, bool CaseSensitive*/)
{
//return row;
string[] tmp_values = values.Split(new string[] { delimitator }, StringSplitOptions.None);
string[] tmp_replace = replace.Split(new string[] { delimitator }, StringSplitOptions.None);
row = row.ToUpper();
for (int i = 0; i < Math.Min(tmp_values.Length, tmp_replace.Length); i++)
{
row = row.Replace(tmp_values[i].ToUpper(), tmp_replace[i]);
}
return row;
}
}
}
and then
select *
into #t
from
(
select 'value1'OldValue,'one'NewValue union all
select 'value2'OldValue,'two'NewValue union all
select 'value3'OldValue,'three'NewValue union all
select 'value4'OldValue,'four'NewValue
)t1
select dbo.ReplaceValues(t1.column,'|',t2.v,t2.r)
from MyTable t1
cross apply
(
select dbo.inlineaggr(i1.OldValue,'|',1,1)v,
dbo.inlineaggr(i1.NewValue,'|',1,1)r
from #t i1
)t2
i have to improved it to manage better the case sensitive, but performance are not bad.
(also 'inlineaggr' is a CLR i wrote years ago)
You can do this via recursion. Assuming you have a table of find-replace pairs, you can number the rows and then use recursive cte:
create table #t(c1 nvarchar(100), c2 nvarchar(100));
insert into #t(c1, c2) values
('bla', ''),
('table', 'table'),
('value', '000'),
('...', '');
declare #s nvarchar(max) = 'this my string and i want to replace all values that are in table #t';
with ncte as (
select row_number() over (order by (select null)) as rn, *
from #t
), rcte as (
select rn, replace(#s, c1, c2) as newstr
from ncte
where rn = 1
union all
select ncte.rn, replace(rcte.newstr, ncte.c1, ncte.c2)
from ncte
join rcte on ncte.rn = rcte.rn + 1
)
select *
from rcte
where rn = 4

Convert SQL to Linq select in where

I have three table below:
TABLE_PRODUCT (IdProduct, ProductName, ProductUnit)
TABLE_STORE_HOUSE (IdContain, IdProduct, ProductNumber, TimeInput)
TABLE_SELL (IdSell, IdContain, ProductNumberSell, TimeSell)
Current, How to using LinQ query get TABLE_STORE_HOUSE.IdProduct witch condition TABLE_STORE_HOUSE.ProductNumber - Sum(TABLE_SELL.ProductNumberSell) > 0 and TABLE_STORE_HOUSE.TimeInput is smallest
Help me convert Sql to Linq..............
select top 1 IdContain
from
TABLE_STORE_HOUSE
where IdProduct = '6'
and
ProductNumber - (select sum(ProductNumber)
from TABLE_SELL
Where TABLE_SELL.IdContain = IdContain)> 0
order by TimeInput desc;
Can you try this?
from t in TABLE_STORE_HOUSEs
let TSell = (
from s in TABLE_SELLs
where s.IdContain == t.IdContain
orderby s.ProductNumber
select new {
s.ProductNumber
}
)
where t.IdProduct == 6 && (t.ProductNumber - TSell.Sum(si => si.ProductNumber)) > 0
select new { t.IdContain }
for top 1 you can use Take() function.

Sql Server - Is there any way to `Concat` nvarchar column in Select Such as `Agregate functions`

Hi I Have Table that called Tags, in tag table I have 2 columns (QuestionID int ,Tag nvachar(100))
I want to Select Questions with all Tags in one column like the below
QuestionID Tag
---------- ----
1 Math
1 Integral
2 Physics
QuestionID QuestionText
---------- -----------
1 What is 2*2?
2 What is Quantom roles?
QuestionID QuestionText Tags
---------- ----------- -------
1 What is 2*2? Math, Integral
2 What is Quantom roles? Physics
Can any one help me with out using scalar value function
There are two ways to answer this:
can use a query like in other answer, but this is work for one table only.
create clr aggregate function for this like a below code (my code in C#).
this solution work for all tables and simple for use,
only use: select Concat(column) from Table in sql server
using System;
using System.Data;
using System.Data.SqlClient;
using System.Data.SqlTypes;
using Microsoft.SqlServer.Server;
using System.Text;
[Serializable]
[Microsoft.SqlServer.Server.SqlUserDefinedAggregate(Format.UserDefined, IsInvariantToDuplicates = false, IsInvariantToNulls = true, IsInvariantToOrder = false, IsNullIfEmpty = true, MaxByteSize = -1)]
public struct Concat : IBinarySerialize
{
public void Init()
{
SB = new StringBuilder();
}
public void Accumulate(SqlString Value)
{
if (Value.IsNull)
return;
if (SB.Length > 0)
SB.Append("\n");
SB.Append(Value);
}
public void Merge(Concat Group)
{
if (SB.Length > 0 && Group.SB.Length > 0)
SB.Append("\n");
SB.Append(Group.SB.ToString());
}
public SqlString Terminate()
{
return new SqlString(SB.ToString());
}
// This is a place-holder member field
StringBuilder SB;
public void Read(System.IO.BinaryReader r)
{
SB = new StringBuilder(r.ReadString());
}
public void Write(System.IO.BinaryWriter w)
{
w.Write(SB.ToString());
}
}
CREATE TABLE #temp
(
QuestionID INT,
Tag NVARCHAR(100)
)
INSERT INTO #temp
(QuestionID,Tag)
VALUES (1,N'Math'),
(1,N'Integral'),
(2,N'Physics')
CREATE TABLE #temp1
(
QuestionID INT,
QuestionText NVARCHAR(100)
)
INSERT INTO #temp1
(QuestionID,QuestionText)
VALUES (1,N'What is 2*2?'),
(2,'What is Quantom roles?')
SELECT h.QuestionID,
h.QuestionText,
Stuff((SELECT ', ' + CONVERT(VARCHAR, b.TAG)
FROM #temp b
WHERE b.QuestionID = h.QuestionID
FOR XML PATH('')), 1, 2, '')
FROM #temp t
JOIN #temp1 h
ON t.QuestionID = h.QuestionID
GROUP BY h.QuestionID,
h.QuestionText
SELECT q.QuestionText
,STUFF((
SELECT ', ' + t2.Tag
FROM Tags t2
WHERE t1.QuestionID = t2.QuestionID
ORDER BY t2.Tag
FOR XML PATH('')
,TYPE
).value('.', 'varchar(max)'), 1, 2, '') AS Tag
FROM Questions q
INNER JOIN Tags t1
ON q.QuestionID = t1.QuestionID
GROUP BY q.QuestionText
,t1.QuestionID
Working example : http://sqlfiddle.com/#!3/e8f0f/7
Try this
create function fn_comma (#question_id int)
returns varchar(100)
as
begin
declare #value varchar(100)
set #value=(SELECT top 1 STUFF((SELECT ', ' + CAST(Value AS VARCHAR(10)) [text()]
FROM Tags
WHERE ID = t.ID
FOR XML PATH(''), TYPE)
.value('.','NVARCHAR(MAX)'),1,2,' ') List_Output
FROM Tags
--where id=1
GROUP BY ID)
return #value
end
Try sub query to concat column data in comma separated values like below :
SELECT [QuestionID],
[QuestionText],
STUFF(( SELECT ',' + [Tag]
FROM [dbo].[Tags]
WHERE [QuestionID] = [Question].[QuestionID]
FOR XML PATH ('')), 1, 1, '') AS [Tags]
FROM [dbo].[Question]
SQL Fiddle Demo
Try the below idea. You just need to rewrite it as a function, then it will return all tags for the question id:
declare #function_in_questionid_para as #int
with std as
(select *,ROW_NUMBER() over(partition by QuestionID order by QuestionID,tag) as dd from #temp)
select * #temp3 into from std
declare #counter as int
set #counter = (select count(*) from #temp where QuestionID = #function_in_questionid_para as #int)
declare #c as int = 1
declare #tags as varchar(200) = ''
while (#c <= #counter)
begin
if (#c > 1) set #tags = #tags + ', '
set #tags = #tags + (select tag from #temp3 where QuestionID = #function_in_questionid_para as #int and dd = #c)
set #c = #c + 1
end
print #tags

Output a comma separated list in T-SQL

I have a table with phone numbers in it. Instead of spitting out a single row for each number I want to return a comma separated list of phone numbers. What's the easiest way to do this in sql? A while loop?
Some of those answers are overly complicated with coalesce and more complex XML queries. I use this all the time:
select #Phones=(
Select PhoneColumn+','
From TableName
For XML Path(''))
-- Remove trailing comma if necessary
select #Phones=left(#Phones,len(#Phones)-1)
You could create a UDF that would do something like this
CREATE FUNCTION dbo.GetBirthdays(#UserId INT)
RETURNS VARCHAR(MAX)
AS
BEGIN
DECLARE #combined VARCHAR(MAX)
SELECT #combined = COALESCE(#combined + ', ' + colName + ', colName)
FROM YourTable
WHERE UserId = #UserId
ORDER BY ColName
END
Basically this just pulls all of the values into a simple list.
FWIW I created a SQL CLR Aggregate function. Works like a champ!
[Serializable]
[SqlUserDefinedAggregate(Format.UserDefined,
Name = "JoinStrings",
IsInvariantToNulls=true,
IsInvariantToDuplicates=false,
IsInvariantToOrder=false,
MaxByteSize=8000)] public struct
JoinStrings : IBinarySerialize {
public string Result;
public void Init()
{
Result = "";
}
public void Accumulate(SqlString value)
{
if (value.IsNull)
return;
Result += value.Value + ",";
}
public void Merge(JoinStrings Group)
{
Result += Group.Result;
}
public SqlString Terminate()
{
return new SqlString(Result.ToString().Trim(new
char[] { ',' }));
}
public void Read(System.IO.BinaryReader r)
{
Result = r.ReadString();
}
public void Write(System.IO.BinaryWriter w)
{
w.Write(Result.ToString());
} }
I can then use it like this:
SELECT dbo.JoinStrings(Phone) FROM Phones Where UserID = XXX
See my answer from this question. There are a couple of other ways to do it listed in that question also. COALESCE or for xml path should do the trick though.
Edit (added my answer from the previous question):
CREATE FUNCTION [dbo].[fn_MyFunction]()RETURNS NVARCHAR(MAX)
AS
BEGIN
DECLARE #str NVARCHAR(MAX)
DECLARE #Delimiter CHAR(2)
SET #Delimiter = ', '
SELECT #str = COALESCE(#str + #Delimiter,'') + AColumn
FROM dbo.myTable
RETURN RTRIM(LTRIM(#str))
END
Assuming you have a Customers table which has a unique ID and another table named PhoneNumbers with multiple phone numbers for each customer sharing the Customer ID field as a Foreign Key this would work using a correlated sub-Query
Select C.ID, C.FirstName, C.LastName,
(select (STUFF(( SELECT ', ' + PhoneNumber from PhoneNumbers P where P.CID = C.ID
FOR XML PATH('')), 1, 2, ''))) as PhoneNumbers
from Customers C
Select Unique ID, Replace(Rtrim(Ltrim(Case when [Phone_Number1] is not null Then [Phone_Number1]+' ' Else '' End +
Case when [Phone_Number2] is not null Then [Phone_Number2]+' ' Else '' End +
Case when [Phone_Number3] is not null Then [Phone_Number3]+' ' Else '' End)),' ',', ') as Phone_numbers
From MYTable
Hope this is what you are looking for and I dont know if this will help you so far after the question.

How to get a distinct list of words used in all Field Records using MS SQL?

If I have a table field named 'description', what would be the SQL (using MS SQL) to get a list of records of all distinct words used in this field.
For example:
If the table contains the following for the 'description' field:
Record1 "The dog jumped over the fence."
Record2 "The giant tripped on the fence."
...
The SQL record output would be:
"The","giant","dog","jumped","tripped","on","over","fence"
I do not think you can do this with a SELECT. The best chance is to write a user defined function that returns a table with all the words and then do SELECT DISTINCT on it.
Disclaimer: Function dbo.Split is from http://www.sqlteam.com/forums/topic.asp?TOPIC_ID=50648
CREATE TABLE test
(
id int identity(1, 1) not null,
description varchar(50) not null
)
INSERT INTO test VALUES('The dog jumped over the fence')
INSERT INTO test VALUES('The giant tripped on the fence')
CREATE FUNCTION dbo.Split
(
#RowData nvarchar(2000),
#SplitOn nvarchar(5)
)
RETURNS #RtnValue table
(
Id int identity(1,1),
Data nvarchar(100)
)
AS
BEGIN
Declare #Cnt int
Set #Cnt = 1
While (Charindex(#SplitOn,#RowData)>0)
Begin
Insert Into #RtnValue (data)
Select
Data = ltrim(rtrim(Substring(#RowData,1,Charindex(#SplitOn,#RowData)-1)))
Set #RowData = Substring(#RowData,Charindex(#SplitOn,#RowData)+1,len(#RowData))
Set #Cnt = #Cnt + 1
End
Insert Into #RtnValue (data)
Select Data = ltrim(rtrim(#RowData))
Return
END
CREATE FUNCTION dbo.SplitAll(#SplitOn nvarchar(5))
RETURNS #RtnValue table
(
Id int identity(1,1),
Data nvarchar(100)
)
AS
BEGIN
DECLARE My_Cursor CURSOR FOR SELECT Description FROM dbo.test
DECLARE #description varchar(50)
OPEN My_Cursor
FETCH NEXT FROM My_Cursor INTO #description
WHILE ##FETCH_STATUS = 0
BEGIN
INSERT INTO #RtnValue
SELECT Data FROM dbo.Split(#description, #SplitOn)
FETCH NEXT FROM My_Cursor INTO #description
END
CLOSE My_Cursor
DEALLOCATE My_Cursor
RETURN
END
SELECT DISTINCT Data FROM dbo.SplitAll(N' ')
I just had a similar problem and tried using SQL CLR to solve it. Might be handy to someone
using System;
using System.Data;
using System.Data.SqlClient;
using System.Data.SqlTypes;
using Microsoft.SqlServer.Server;
using System.Collections;
using System.Collections.Generic;
public partial class UserDefinedFunctions
{
private class SplitStrings : IEnumerable
{
private List<string> splits;
public SplitStrings(string toSplit, string splitOn)
{
splits = new List<string>();
// nothing, return empty list
if (string.IsNullOrEmpty(toSplit))
{
return;
}
// return one word
if (string.IsNullOrEmpty(splitOn))
{
splits.Add(toSplit);
return;
}
splits.AddRange(
toSplit.Split(new string[] { splitOn }, StringSplitOptions.RemoveEmptyEntries)
);
}
#region IEnumerable Members
public IEnumerator GetEnumerator()
{
return splits.GetEnumerator();
}
#endregion
}
[Microsoft.SqlServer.Server.SqlFunction(FillRowMethodName = "readRow", TableDefinition = "word nvarchar(255)")]
public static IEnumerable fnc_clr_split_string(string toSplit, string splitOn)
{
return new SplitStrings(toSplit, splitOn);
}
public static void readRow(object inWord, out SqlString word)
{
string w = (string)inWord;
if (string.IsNullOrEmpty(w))
{
word = string.Empty;
return;
}
if (w.Length > 255)
{
w = w.Substring(0, 254);
}
word = w;
}
};
It is not the fastest approach but might be used by somebody for a small amount of data:
declare #tmp table(descr varchar(400))
insert into #tmp
select 'The dog jumped over the fence.'
union select 'The giant tripped on the fence.'
/* the actual doing starts here */
update #tmp
set descr = replace(descr, '.', '') --get rid of dots in the ends of sentences.
declare #xml xml
set #xml = '<c>' + replace(
(select ' ' + descr
from #tmp
for xml path('')
), ' ', '</c><c>') + '</c>'
;with
allWords as (
select section.Cols.value('.', 'varchar(250)') words
from #xml.nodes('/c') section(Cols)
)
select words
from allWords
where ltrim(rtrim(words)) <> ''
group by words
In SQL on it's own it would probably need to be a big stored procedure, but if you read all the records out to the scripting language of your choice, you can easily loop over them and split each out into arrays/hashes.
it'd be a messy stored procedure with a temp table and a SELECT DISTINCT at the end.
if you had the words already as records, you would use SELECT DISTINCT [WordsField] from [owner].[tablename]