How to pass an array into a SQL Server stored procedure?
For example, I have a list of employees. I want to use this list as a table and join it with another table. But the list of employees should be passed as parameter from C#.
SQL Server 2016 (or newer)
You can pass in a delimited list or JSON and use STRING_SPLIT() or OPENJSON().
STRING_SPLIT():
CREATE PROCEDURE dbo.DoSomethingWithEmployees
#List varchar(max)
AS
BEGIN
SET NOCOUNT ON;
SELECT value FROM STRING_SPLIT(#List, ',');
END
GO
EXEC dbo.DoSomethingWithEmployees #List = '1,2,3';
OPENJSON():
CREATE PROCEDURE dbo.DoSomethingWithEmployees
#List varchar(max)
AS
BEGIN
SET NOCOUNT ON;
SELECT value FROM OPENJSON(CONCAT('["',
REPLACE(STRING_ESCAPE(#List, 'JSON'),
',', '","'), '"]')) AS j;
END
GO
EXEC dbo.DoSomethingWithEmployees #List = '1,2,3';
I wrote more about this here:
Handling an unknown number of parameters in SQL Server
Ordered String Splitting in SQL Server with OPENJSON
SQL Server 2008 (or newer)
First, in your database, create the following two objects:
CREATE TYPE dbo.IDList
AS TABLE
(
ID INT
);
GO
CREATE PROCEDURE dbo.DoSomethingWithEmployees
#List AS dbo.IDList READONLY
AS
BEGIN
SET NOCOUNT ON;
SELECT ID FROM #List;
END
GO
Now in your C# code:
// Obtain your list of ids to send, this is just an example call to a helper utility function
int[] employeeIds = GetEmployeeIds();
DataTable tvp = new DataTable();
tvp.Columns.Add(new DataColumn("ID", typeof(int)));
// populate DataTable from your List here
foreach(var id in employeeIds)
tvp.Rows.Add(id);
using (conn)
{
SqlCommand cmd = new SqlCommand("dbo.DoSomethingWithEmployees", conn);
cmd.CommandType = CommandType.StoredProcedure;
SqlParameter tvparam = cmd.Parameters.AddWithValue("#List", tvp);
// these next lines are important to map the C# DataTable object to the correct SQL User Defined Type
tvparam.SqlDbType = SqlDbType.Structured;
tvparam.TypeName = "dbo.IDList";
// execute query, consume results, etc. here
}
SQL Server 2005
If you are using SQL Server 2005, I would still recommend a split function over XML. First, create a function:
CREATE FUNCTION dbo.SplitInts
(
#List VARCHAR(MAX),
#Delimiter VARCHAR(255)
)
RETURNS TABLE
AS
RETURN ( SELECT Item = CONVERT(INT, Item) FROM
( SELECT Item = x.i.value('(./text())[1]', 'varchar(max)')
FROM ( SELECT [XML] = CONVERT(XML, '<i>'
+ REPLACE(#List, #Delimiter, '</i><i>') + '</i>').query('.')
) AS a CROSS APPLY [XML].nodes('i') AS x(i) ) AS y
WHERE Item IS NOT NULL
);
GO
Now your stored procedure can just be:
CREATE PROCEDURE dbo.DoSomethingWithEmployees
#List VARCHAR(MAX)
AS
BEGIN
SET NOCOUNT ON;
SELECT EmployeeID = Item FROM dbo.SplitInts(#List, ',');
END
GO
And in your C# code you just have to pass the list as '1,2,3,12'...
I find the method of passing through table valued parameters simplifies the maintainability of a solution that uses it and often has increased performance compared to other implementations including XML and string splitting.
The inputs are clearly defined (no one has to guess if the delimiter is a comma or a semi-colon) and we do not have dependencies on other processing functions that are not obvious without inspecting the code for the stored procedure.
Compared to solutions involving user defined XML schema instead of UDTs, this involves a similar number of steps but in my experience is far simpler code to manage, maintain and read.
In many solutions you may only need one or a few of these UDTs (User defined Types) that you re-use for many stored procedures. As with this example, the common requirement is to pass through a list of ID pointers, the function name describes what context those Ids should represent, the type name should be generic.
Based on my experience, by creating a delimited expression from the employeeIDs, there is a tricky and nice solution for this problem. You should only create an string expression like ';123;434;365;' in-which 123, 434 and 365 are some employeeIDs. By calling the below procedure and passing this expression to it, you can fetch your desired records. Easily you can join the "another table" into this query. This solution is suitable in all versions of SQL server. Also, in comparison with using table variable or temp table, it is very faster and optimized solution.
CREATE PROCEDURE dbo.DoSomethingOnSomeEmployees #List AS varchar(max)
AS
BEGIN
SELECT EmployeeID
FROM EmployeesTable
-- inner join AnotherTable on ...
where #List like '%;'+cast(employeeID as varchar(20))+';%'
END
GO
Use a table-valued parameter for your stored procedure.
When you pass it in from C# you'll add the parameter with the data type of SqlDb.Structured.
See here: http://msdn.microsoft.com/en-us/library/bb675163.aspx
Example:
// Assumes connection is an open SqlConnection object.
using (connection)
{
// Create a DataTable with the modified rows.
DataTable addedCategories =
CategoriesDataTable.GetChanges(DataRowState.Added);
// Configure the SqlCommand and SqlParameter.
SqlCommand insertCommand = new SqlCommand(
"usp_InsertCategories", connection);
insertCommand.CommandType = CommandType.StoredProcedure;
SqlParameter tvpParam = insertCommand.Parameters.AddWithValue(
"#tvpNewCategories", addedCategories);
tvpParam.SqlDbType = SqlDbType.Structured;
// Execute the command.
insertCommand.ExecuteNonQuery();
}
You need to pass it as an XML parameter.
Edit: quick code from my project to give you an idea:
CREATE PROCEDURE [dbo].[GetArrivalsReport]
#DateTimeFrom AS DATETIME,
#DateTimeTo AS DATETIME,
#HostIds AS XML(xsdArrayOfULong)
AS
BEGIN
DECLARE #hosts TABLE (HostId BIGINT)
INSERT INTO #hosts
SELECT arrayOfUlong.HostId.value('.','bigint') data
FROM #HostIds.nodes('/arrayOfUlong/u') as arrayOfUlong(HostId)
Then you can use the temp table to join with your tables.
We defined arrayOfUlong as a built in XML schema to maintain data integrity, but you don't have to do that. I'd recommend using it so here's a quick code for to make sure you always get an XML with longs.
IF NOT EXISTS (SELECT * FROM sys.xml_schema_collections WHERE name = 'xsdArrayOfULong')
BEGIN
CREATE XML SCHEMA COLLECTION [dbo].[xsdArrayOfULong]
AS N'<xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema">
<xs:element name="arrayOfUlong">
<xs:complexType>
<xs:sequence>
<xs:element maxOccurs="unbounded"
name="u"
type="xs:unsignedLong" />
</xs:sequence>
</xs:complexType>
</xs:element>
</xs:schema>';
END
GO
Context is always important, such as the size and complexity of the array. For small to mid-size lists, several of the answers posted here are just fine, though some clarifications should be made:
For splitting a delimited list, a SQLCLR-based splitter is the fastest. There are numerous examples around if you want to write your own, or you can just download the free SQL# library of CLR functions (which I wrote, but the String_Split function, and many others, are completely free).
Splitting XML-based arrays can be fast, but you need to use attribute-based XML, not element-based XML (which is the only type shown in the answers here, though #AaronBertrand's XML example is the best as his code is using the text() XML function. For more info (i.e. performance analysis) on using XML to split lists, check out "Using XML to pass lists as parameters in SQL Server" by Phil Factor.
Using TVPs is great (assuming you are using at least SQL Server 2008, or newer) as the data is streamed to the proc and shows up pre-parsed and strongly-typed as a table variable. HOWEVER, in most cases, storing all of the data in DataTable means duplicating the data in memory as it is copied from the original collection. Hence using the DataTable method of passing in TVPs does not work well for larger sets of data (i.e. does not scale well).
XML, unlike simple delimited lists of Ints or Strings, can handle more than one-dimensional arrays, just like TVPs. But also just like the DataTable TVP method, XML does not scale well as it more than doubles the datasize in memory as it needs to additionally account for the overhead of the XML document.
With all of that said, IF the data you are using is large or is not very large yet but consistently growing, then the IEnumerable TVP method is the best choice as it streams the data to SQL Server (like the DataTable method), BUT doesn't require any duplication of the collection in memory (unlike any of the other methods). I posted an example of the SQL and C# code in this answer:
Pass Dictionary to Stored Procedure T-SQL
As others have noted above, one way to do this is to convert your array to a string and then split the string inside SQL Server.
As of SQL Server 2016, there's a built-in way to split strings called
STRING_SPLIT()
It returns a set of rows that you can insert into your temp table (or real table).
DECLARE #str varchar(200)
SET #str = "123;456;789;246;22;33;44;55;66"
SELECT value FROM STRING_SPLIT(#str, ';')
would yield:
value
-----
123
456
789
246
22
33
44
55
66
If you want to get fancier:
DECLARE #tt TABLE (
thenumber int
)
DECLARE #str varchar(200)
SET #str = "123;456;789;246;22;33;44;55;66"
INSERT INTO #tt
SELECT value FROM STRING_SPLIT(#str, ';')
SELECT * FROM #tt
ORDER BY thenumber
would give you the same results as above (except the column name is "thenumber"), but sorted. You can use the table variable like any other table, so you can easily join it with other tables in the DB if you want.
Note that your SQL Server install has to be at compatibility level 130 or higher in order for the STRING_SPLIT() function to be recognized. You can check your compatibility level with the following query:
SELECT compatibility_level
FROM sys.databases WHERE name = 'yourdatabasename';
Most languages (including C#) have a "join" function you can use to create a string from an array.
int[] myarray = {22, 33, 44};
string sqlparam = string.Join(";", myarray);
Then you pass sqlparam as your parameter to the stored procedure above.
This will help you. :) Follow the next steps,
Open the Query Editor
Copy Paste the following code as it is, it will create the Function which converts the String to Int
CREATE FUNCTION dbo.SplitInts
(
#List VARCHAR(MAX),
#Delimiter VARCHAR(255)
)
RETURNS TABLE
AS
RETURN ( SELECT Item = CONVERT(INT, Item) FROM
( SELECT Item = x.i.value('(./text())[1]', 'varchar(max)')
FROM ( SELECT [XML] = CONVERT(XML, '<i>'
+ REPLACE(#List, #Delimiter, '</i><i>') + '</i>').query('.')
) AS a CROSS APPLY [XML].nodes('i') AS x(i) ) AS y
WHERE Item IS NOT NULL
);
GO
Create the Following stored procedure
CREATE PROCEDURE dbo.sp_DeleteMultipleId
#List VARCHAR(MAX)
AS
BEGIN
SET NOCOUNT ON;
DELETE FROM TableName WHERE Id IN( SELECT Id = Item FROM dbo.SplitInts(#List, ','));
END
GO
Execute this SP Using exec sp_DeleteId '1,2,3,12' this is a string of Id's which you want to delete,
You can convert your array to string in C# and pass it as a Stored Procedure parameter as below,
int[] intarray = { 1, 2, 3, 4, 5 };
string[] result = intarray.Select(x=>x.ToString()).ToArray();
SqlCommand command = new SqlCommand();
command.Connection = connection;
command.CommandText = "sp_DeleteMultipleId";
command.CommandType = CommandType.StoredProcedure;
command.Parameters.Add("#Id",SqlDbType.VARCHAR).Value=result ;
This will delete multiple rows in a single stored proc call. All the best.
There is no support for array in sql server but there are several ways by which you can pass collection to a stored proc .
By using datatable
By using XML.Try converting your collection in an xml format and then pass it as an input to a stored procedure
The below link may help you
passing collection to a stored procedure
Starting in SQL Server 2016 you can bring the list in as an NVARCHAR() and use OPENJSON
DECLARE #EmployeeList nvarchar(500) = '[1,2,15]'
SELECT *
FROM Employees
WHERE ID IN (SELECT VALUE FROM OPENJSON(#EmployeeList ))
I've been searching through all the examples and answers of how to pass any array to sql server without the hassle of creating new Table type,till i found this linK, below is how I applied it to my project:
--The following code is going to get an Array as Parameter and insert the values of that
--array into another table
Create Procedure Proc1
#UserId int, //just an Id param
#s nvarchar(max) //this is the array your going to pass from C# code to your Sproc
AS
declare #xml xml
set #xml = N'<root><r>' + replace(#s,',','</r><r>') + '</r></root>'
Insert into UserRole (UserID,RoleID)
select
#UserId [UserId], t.value('.','varchar(max)') as [RoleId]
from #xml.nodes('//root/r') as a(t)
END
Hope you enjoy it
Starting in SQL Server 2016 you can simply use split string
Example:
WHERE (#LocationId IS NULL OR Id IN (SELECT items from Split_String(#LocationId, ',')))
CREATE TYPE dumyTable
AS TABLE
(
RateCodeId int,
RateLowerRange int,
RateHigherRange int,
RateRangeValue int
);
GO
CREATE PROCEDURE spInsertRateRanges
#dt AS dumyTable READONLY
AS
BEGIN
SET NOCOUNT ON;
INSERT tblRateCodeRange(RateCodeId,RateLowerRange,RateHigherRange,RateRangeValue)
SELECT *
FROM #dt
END
It took me a long time to figure this out, so in case anyone needs it...
This is based on the SQL 2005 method in Aaron's answer, and using his SplitInts function (I just removed the delim param since I'll always use commas). I'm using SQL 2008 but I wanted something that works with typed datasets (XSD, TableAdapters) and I know string params work with those.
I was trying to get his function to work in a "where in (1,2,3)" type clause, and having no luck the straight-forward way. So I created a temp table first, and then did an inner join instead of the "where in". Here is my example usage, in my case I wanted to get a list of recipes that don't contain certain ingredients:
CREATE PROCEDURE dbo.SOExample1
(
#excludeIngredientsString varchar(MAX) = ''
)
AS
/* Convert string to table of ints */
DECLARE #excludeIngredients TABLE (ID int)
insert into #excludeIngredients
select ID = Item from dbo.SplitInts(#excludeIngredientsString)
/* Select recipies that don't contain any ingredients in our excluded table */
SELECT r.Name, r.Slug
FROM Recipes AS r LEFT OUTER JOIN
RecipeIngredients as ri inner join
#excludeIngredients as ei on ri.IngredientID = ei.ID
ON r.ID = ri.RecipeID
WHERE (ri.RecipeID IS NULL)
I am trying to insert > 8000 characters (submit from a web page) via ExecuteNonQuery (and DatabaseFactory.CreateDatabase() from MS Practices Enterprise Library). The stored procedure defines the parameter as VARCHAR(MAX). The column is VARCHAR(MAX). In theory, 2GB of data should be able to be passed.
What can I do to pass data > 8000? I set a breakpoint and the string.Length is indeed > 8K.
public static void UpdateTerms(string terms)
{
Database db = DatabaseFactory.CreateDatabase();
db.ExecuteNonQuery("uspUpdateTerms", terms);
}
Stored procedure:
ALTER PROCEDURE [dbo].[uspUpdateTerms]
#Terms VARCHAR(MAX)
AS
SET NOCOUNT ON
INSERT INTO tblTerms(Terms)
VALUES(#Terms)
Table (just to show that everything is varchar(max)):
CREATE TABLE [dbo].[tblTerms](
[ID] [int] IDENTITY(1,1) NOT NULL,
[Terms] [varchar](max) NULL,
[DateUpdated] [datetime] NULL,
.
Update:
I just changed the code, and this seems to work, though I am not sure what the difference is:
public static void UpdateTerms(string terms)
{
Database db = DatabaseFactory.CreateDatabase();
DbCommand cmd = db.GetStoredProcCommand("uspUpdateTerms");
db.AddInParameter(cmd, "Terms", DbType.String, terms);
db.ExecuteNonQuery(cmd);
}
The issue may not be the storage of the data, it may be the retrieval.
If you are trying to determine whether or not more than 8000 chars were stored in the DB through enterprise manager, then you are out of luck if you just select the contents of the columns and look at the text length: enterprise manager limits the column output.
To determine how much data is actually stored in the column, execute the following query:
SELECT DATALENGTH(Terms) FROM tblTerms
This will tell you how much text was stored.
EDIT:
Another thought just occurred: the enterprise library caches stored procedure parameters in order to improve performance. If you changed the stored procedure after testing with the parameter set to nvarchar(8000), then switch the parameter to nvarchar(max) without resetting the application (if IIS-hosted, then iisreset or dirty web.config), then you will still be using the old stored proc parameter.
REPLICATE returns the input type irrespective of later assignment. It's annoying, but to avoid silent truncation, try:
SET #x = REPLICATE(CONVERT(VARCHAR(MAX), 'a'), 10000);
This is because SQL Server performs the REPLICATE operation before it considers what you're assigning it to or how many characters you're trying to expand it to. It only cares about the input expression to determine what it should return, and if the input is not a max type, it assumes it is meant to fit within 8,000 bytes. This is explained in Books Online:
If string_expression is not of type varchar(max) or nvarchar(max), REPLICATE truncates the return value at 8,000 bytes. To return values greater than 8,000 bytes, string_expression must be explicitly cast to the appropriate large-value data type.
Your sample code can be fixed by doing:
declare #x varchar(max)
set #x = replicate (cast('a' as varchar(max)), 10000)
select #x, len(#x)
You haven't shown the code where you are trying to use ExecutenonQuery. Note that you should use parameters.
using(var con = new SqlConnection(conString))
using(var cmd = new SqlCommand("storedProcedureName", con))
{
cmd.CommandType = CommandType.StoredProcedure;
cmd.Parameters.Add("#text", SqlDbType.NVarChar, -1);
cmd.Parameters["#text"].Value = yourVeryLongText;
cmd.ExecuteNonQuery();
}
Do you see something blatantly wrong with this line:
commaSepString1 = com.Parameters["#CommaSepString"].Value.ToString();
It shows an error like:
Object reference not set to an instance of an object.
For short, the C# method looks like:
public static DataTable GetFilmDetails(string filmId, out string
commaSepString1)
It takes the following stored procedure (that executes correctly on its own):
com.CommandText = "CatalogGetFilmDetails2";
And since #CommaSepString is an output parameter in the stored procedure, in C# I have this:
param = com.CreateParameter();
param.ParameterName = "#CommaSepString";
param.Direction = ParameterDirection.Output;
param.DbType = DbType.String;
com.Parameters.Add(param);
commaSepString1 = com.Parameters["#CommaSepString"].Value.ToString();
If you are curious about the Stored Procedure, it starts like this:
CREATE PROCEDURE CatalogGetFilmDetails2
(#FilmID int, #CommaSepString VARCHAR(50) OUTPUT) AS
The SP inserts some rows into a table variable based on some joins, it puts all values from column x from that table into a string with comma separated values:
SELECT #CommaSepString
= STUFF((SELECT ', ' + Categ FROM #Filme1 FOR XML PATH('')), 1,1,'')
then it selects the table.
Thank you for the patience to read this. Would it be more clear if I published the entire code for the stored procedure and the C# method? Not sure if I am allowed that and worried it might be too lengthy to read.
PS: There must be something wrong about the commaSepString! The code worked perfectly before I added it in both places.
SQL Server doesn't know output, it only knows input/output. So when you specify
param.Direction = ParameterDirection.Output;
SQL Server will actually read the string as well. Since you haven't supplied a value, SQL Server determines that that its length is zero.
Incredibly, if you specify the magic number -1, SQL Server will determine the size for you:
param.Size = -1;
This is perhaps the second worst gotcha about ADO.NET.
Just behind the fact that SQL NULL translates to DBNull.Value instead of null.
You should set a breakpoint on the offending line, then you should examine your parameters collection to check if the parameter exists.
If it exists, then check if its value is null.
In this case the .ToString() will give you the error described.
In a SQL stored proc i'm inserting a row and hoping to return the IDENTITY INT for the new row, then get it in my C# code.
ALTER PROCEDURE the_proc #val_2 VARCHAR(10), #val_3 VARCHAR(10)
AS
BEGIN
SET NOCOUNT ON
INSERT INTO the_table (field_2, field_3)
OUTPUT INSERTED.field_1
VALUES (#val_2, #val_3)
END
In C# i'm using LINQ but am fuzzy on how to retrieve that OUTPUT value. I tried including it as an OUTPUT parameter in the SQL proc declaration, but couldn't get that working either (exceptions complaining about it not being supplied in the call). The closest i've gotten is in walking into the Db.designer.cs code, where IExecuteResult result ReturnValue contains 0 (not correct) but inspecting the contained Results View (result.ReturnValue Results View) DOES have the outputed value.
key = (int)(db.TheProc(val2,val3).ReturnValue);
key is coming back as 0. I want the IDENTITY INT value from the INSERT.
OUTPUT INSERTED.*
is basically the same thing as doing a select. So this isn't going to show up as an output parameter but rather come back as a result set.
BTW, the ReturnValue should actually be zero which is what you are seeing in this case.
You'll need to change your linq statement so that you capture that result set.
Try this instead (assuming SQL Server) :
ALTER PROCEDURE the_proc
#val_2 VARCHAR(10),
#val_3 VARCHAR(10),
#newKey int OUTPUT
AS
BEGIN
SET NOCOUNT ON
INSERT INTO the_table (field_2, field_3) VALUES (#val_2, #val_3)
SET #newKey = SCOPE_IDENTITY()
END
Once you define your the_proc stored procedure to your LINQ to SQL dbml you can do this:
int? newKey = null;
dataContext.the_proc("val1", "val2", ref newKey);
// newKey.Value now contains your new IDENTITY value
Chris Lively nudged me in the right direction, which is to say re-examining the C# Linq code. The following pulls Field_1 out of the results. Maybe it's a weird way to get there and not the normal idiom, so if anyone has suggestions for something more "correct" please add a comment or answer.
var o = from res in db.MyProc( Val1, Val2 )
select res.Field_1;
int key = o.ToArray()[0];
Thanks.