We Keep Coding

Postgresql query returns NullReferenceException

I'm using Selenium + postgre integration in order to validate correct data on database. After an action on UI then I'm using Npgsql library in order to receive scalar value from cell and then use it for assertion.
Code looks line:
public static string GetCreditAmount(string orderId, string accountNumber)
{
string findCreditQuery = $#"SELECT ""Credit"" FROM accounting.""AccountLines"" al INNER JOIN accounting.""Accounts"" acc ON al.""Account_FK"" = acc.""Id""
WHERE
""OrderId"" = '{orderId}'
AND ""AccountNumber"" = '{accountNumber}'
ORDER BY al.""CreatedDateUtc"" DESC";
using (var connection = new NpgsqlConnection(Configuration.AdminConnectionString))
{
connection.Open();
var transaction = connection.BeginTransaction();
try
{
var account = connection.ExecuteScalar(findCreditQuery).ToString();
return account;
}
catch (NullReferenceException e)
{
Thread.Sleep(3000);
var account = connection.ExecuteScalar(findCreditQuery).ToString();
return account;
}
catch (Exception e)
{
transaction.Rollback();
throw e;
}
}
}
The problem which I'm receiving is NullReferenceException, however results are really random. In one test I'm firing & asserting few times where in like 85% it does not work (Null reference) and from time to time in like 15% it works (test passed). Strange case is that assertion which was fine in one run fails in another one.
I was trying to add static sleeps (Thread.Sleep) in order to give database few extra second for proceeding but all it still fails.
Is there any proper solution which can be used?

If you do this:
object o = null;
string s = o.ToString();
I believe you will get the same error. I believe the issue is your database object is null.
This will probably fix it:
object o = connection.ExecuteScalar(findCreditQuery);
string account = o == null ? null : o.ToString();
Alternatively, you can implement a DbDataReader and use the IsDbNull method. It will be more work but might also be more scalable if you plan to do more than what's currently being done:
string account;
using (NpgsqlCommand cmd = new NpgsqlCommand(findCreditQuery, connection))
{
using (NpgsqlDataReader reader = cmd.ExecuteReader())
{
reader.Read();
if (!reader.IsDBNull(0))
account = reader.GetString(0);
}
}

Error handling response with Entity Framework when calling stored procedure

Whats the Perfered Method to handle Errors in a SQL Stored Procedure when Using Entity Framework.
I have a chunk of code that uses a database context and EF to call a stored procedure. If everything is perfect, the stored procedure returns a list of keys to tell me which records were modified. If something goes wrong in the stored procedure, the transaction rolls back and I believe the response type is a string and causes this error:
The data reader has more than one field. Multiple fields are not valid for EDM primitive or enumeration types.
After doing so digging I noticed that the SQL developer had put "PRINT: ERROR MESSAGE HERE" statement to pass back a message, However the current call does not allow for this,
Here is how the call is currently being made:
var partIdsCreated = context.Database.SqlQuery<int>("EXEC My_SP #FilePath, #FileName", args).ToList();
The Current Code is expecting a list of Int and if a string is Returned it Errors out with the Error Message listed above, So what is the best way to handle errors coming back from a Stored Procedure in a Transaction Using Entity Framework 6?

Below I have provided two similar ways to handle messages returned by SQL Server. As noted in the comments, both involve the usage of the InfoMessage event on the connection.
To answer your question, I don't really know of a preferred way to do handle PRINT statements as errors, but I am also not really sure of another way to handle the output.
Solution 1: Catch Exception
public class DataLoader_01
{
public int LoadData()
{
_infoMessage = null;
using (var context = new ApplicationDbContext(#"Server=localhost\SQLEXPRESS;Database=StackOverflow;Trusted_Connection=True;"))
{
var sqlConnection = (SqlConnection)context.Database.Connection;
sqlConnection.InfoMessage += InfoMessage;
try
{
var t = context.Database.SqlQuery<int>("EXEC ErrorMessageHandling", new object[] {});
return t.First();
}
catch (EntityCommandExecutionException e)
{
if (string.IsNullOrEmpty(_infoMessage))
{
//do some error handling specific to your application
throw new ApplicationSpecificException(_infoMessage);
}
throw;
}
}
}
private void InfoMessage(object sender, SqlInfoMessageEventArgs e)
{
_infoMessage = e.Message;
}
private string _infoMessage;
}
Solution 2: Check Field Count
public class DataLoader_02
{
public int LoadData()
{
_infoMessage = null;
using (var context = new ApplicationDbContext(#"Server=localhost\SQLEXPRESS;Database=StackOverflow;Trusted_Connection=True;"))
{
var sqlConnection = (SqlConnection)context.Database.Connection;
sqlConnection.InfoMessage += InfoMessage;
var cmd = context.Database.Connection.CreateCommand();
cmd.CommandText = "[dbo].[SomeProc]";
try
{
context.Database.Connection.Open();
var reader = cmd.ExecuteReader();
if (reader.FieldCount == 0)
{
throw new ApplicationSpecificException(_infoMessage);
}
var result = ((IObjectContextAdapter)context).ObjectContext.Translate<int>(reader);
return result.First();
}
finally
{
context.Database.Connection.Close();
}
}
}
private void InfoMessage(object sender, SqlInfoMessageEventArgs e)
{
_infoMessage = e.Message;
}
private string _infoMessage;
}

ExecuteNonQuery: Connection property has not been initialized

I'm receiving an error that says that the connection property has not been initialized (see https://imgur.com/CTHIabz). This is confusing, as it is the same code that is in other methods that work just fine. Here is the code:
public bool AddBirthday(string full_name, string month, int day, int year)
{
if (!string.IsNullOrWhiteSpace(full_name) && month.ToString() != "" && day.ToString() != "" && year.ToString() != "")
{
// all required fields were filled out
// determine which month
values = birthdayMonths.Where(v => v.Value == month).ToList();
using (OleDbConnection dbConn = new OleDbConnection(#"Provider=Microsoft.Jet.OLEDB.4.0;Data Source=|DataDirectory|\qbcdb.mdb"))
{
dbConn.Open();
using (cmd = new OleDbCommand("INSERT INTO birthdays (full_name, birthday_month, birthday_day, birthday_year) VALUES (#full_name, #birthday_month, #birthday_day, #birthday_year)"))
{
foreach (KeyValuePair<int, string> key in values)
{
if (key.Value != "")
{
try
{
cmd.Parameters.AddWithValue("#full_name", full_name);
cmd.Parameters.AddWithValue("#birthday_month", key.Value);
cmd.Parameters.AddWithValue("#birthday_day", day);
cmd.Parameters.AddWithValue("#birthday_year", year);
if (cmd.ExecuteNonQuery() > 0)
{
return true;
}
}
catch (Exception e)
{
MessageBox.Show(e.Message);
}
}
else
{
MessageBox.Show("No values were given.", "Error", MessageBoxButtons.OK, MessageBoxIcon.Warning);
}
}
}
dbConn.Close();
}
}
return false;
}
I've gone over the code and checked to make sure the database was in the right location and everything else I could think of. I'm confused on why this work would on another method that uses the same code (almost) when it comes to the database query and binds it to a dvg.
Any help would be appreciated.
Thanks!
Oh btw, it inserts selected values from a listboxes, as demonstrated here:
insertBirthday.Click += (senders, argss) =>
{
if (btd.AddBirthday(name.Text, birthdays.SelectedItem.ToString(), Convert.ToInt32(dayListBox.SelectedItem.ToString()),
Convert.ToInt32(yearListBox.SelectedItem.ToString())))
{
MessageBox.Show("Birthday Added Successfully", "Success", MessageBoxButtons.OK);
}
};

Use this constructor for OleDbCommand: OleDbCommand Constructor (String, OleDbConnection)

You haven't assigned the connection to your OleDbCommand. There are two ways you could to this:
cmd.Connection = dbConn;
or in the constructor:
using (cmd = new OleDbCommand(sql, dbConn)) { ... }
It's also good practice not to open your connection until you are going to use it, i.e. right before cmd.ExecuteNonQuery(). Since you are using a using block, you also don't need to close the connection, that will be taken care of for you, but it's probably not hurting anything.
I recommend separating your database code from your UI code and validation logic (showing MessageBox and checking values) into a separate "repository" class. All of the data necessary to perform the operation should also be passed in as a parameter to your database code so that you don't have any confusing issues where those values aren't being set somewhere else. Calculate the values to insert before calling the method and then pass those as parameters.

How to reference SQL connection

I'm just learning asp.net/C# and am building my first application.
In my application, I am rendering data from sql on almost every view.
My controllers are getting large, because every time I make a request, I'm using somthing like:
try
{
sqlConnection = new SqlConnection(dbConnectionString);
SqlCommand command = new SqlCommand("sp_Test", sqlConnection);
command.CommandType = CommandType.StoredProcedure;
sqlConnection.Open();
return command.ExecuteNonQuery();
sqlConnection.Close();
}
catch (SqlException ex)
{
Console.WriteLine("SQL Error" + ex.Message.ToString());
return 0;
}
Is there a way to turn the sql into a simple using block?
Maybe something like:
using(myConnection){
SqlCommand command = new SqlCommand("sp_Test", sqlConnection);
command.CommandType = CommandType.StoredProcedure;
}

There are many better approaches do it. You can create a SqlHelper class that can be used to execute stored procedures and SQL queries and also return DataReader and DataTable/DataSets.
public class SqlHelper
{
public SqlHelper(string connString)
{
}
public DataSet GetDatasetByCommand(string Command);
public SqlDataReader GetReaderBySQL(string strSQL);
public SqlDataReader GetReaderByCmd(string Command);
public SqlConnection GetSqlConnection();
public void CloseConnection();
}
You can see one such sample here:
http://www.nullskull.com/a/1295/sql-helper-class-in-c.aspx
If you want more advanced approach you can go for Enterprise Library Data Access Block
http://msdn.microsoft.com/en-us/magazine/cc163766.aspx

The best thing to do is refactor that statement into a seperate method. It looks like the only thing that could vary is the name of the procedure.
So create an object with two properties, a boolean success and an error message.
Call the function and pass in the name of the sql command. Your function should run your repeated code in the try block based on the given procedure name, then return an object with true/false and an error message if the call failed. This should make your controllers much smaller. Example code for the controller:
var result = MyNewMethod("sp_Test");
if(!result.Success)
{
Console.WriteLine(result.ErrorMessage);
return 0;
}

Can I combine a using() {} block with a method's out parameter?

Given a method
public static bool Connection.TryCreate(out Connection connection) {}
And a piece of calling code:
Connection connection;
if (!Connection.TryCreate(out connection))
// handle failure gracefully.
/*
* work with connection
*
* …
*
*/
connection.Dispose();
I'm using the same pattern as bool.TryParse and friends, i.e. TryCreate returns whether the operation was successful.
I realize the using() variable needs to be read-only within its block, but is there a way to turn the above into a using() {} block (TryCreate only sets it once), like so:
using (Connection connection)
{
if (!Connection.TryCreate(out connection))
// this would leave the using() block prematurely
/*
* work with sconnection
*
* …
*
*/
}
(This doesn't compile:
error CS1657: Cannot pass 'connection' as a ref or out argument because it is a 'using variable'
)

No, that is not possible.
The using (x) {...} construct makes a copy of x when it enters the block, so you can do this:
var x = new FileStream(...);
using (x)
{
x = null;
}
The stream will still be disposed when the using block ends.
The corollary is that this won't work either:
Stream x = null;
using (x)
{
x = new FileStream(...);
}
here the stream you construct inside the using block will not be disposed.
What you can do, however, is this:
Connection connection;
if (Connection.TryCreate(out connection))
using (connection)
{
}
In C# 7.0 and onwards you can combine this with "out variables" to form:
if (Connection.TryCreate(out var connection))
using (connection)
{
}

Looks like a bad use of the Try* pattern (some would argue this is an anti-pattern).
Instead of a TryCreate, just have a Create method that throws an exception if not successful and that returns the created connection.
Then you could do the usual:
using(Connection connection = Connection.Create())
{
}
Alternatively, if you want to avoid an exception being thrown and the required try{}catch{}, have the Create method return null when a connection could not be created and test for that.

You can do it like this:
Connection connection;
if (Connection.TryCreate(out connection))
{
using (connection)
{
…
}
}
But it might be better if you just returned null on failure:
using (Connection connection = Connection.Create())
{
if (connection != null)
{
…
}
}
The finally block that is created by the using checks whether connection is null and doesn't do anything if it is.
Also, if you're not declaring the variable in the using, then it doesn't have to be read-only.

No. If you are concerned about exceptions in the gap between a method call and the using, you could use try/finally:
Connection conn = null;
try {
if(!conn.TryCreate(out conn)) return;
...
} finally {
if(conn != null) conn.Dispose();
}

A step sideways ?
public class ConnectTo : IDisposable
{
public Connection CurrentConnection {get; private set;}
public ConnectTo()
{
CurrentConnection = null;
// Connect up to whatever.
}
#region IDisposable
// Blah blah
#endregion
}
then
using( ConnectedTo conn = new ConnectTo())
{
if (conn.CurrentConnection != null)
{
//Do Stuff
}
}