We Keep Coding

Making AutoMapper reuse previously mapped instances

I would like AutoMapper to map same object instances Source to the same instances of Target.
I'm sure that AutoMapper can be configured to do this, but how? I need to convert an object graph A to B, keeping the references between the mapped objects.
For example, this fails:
[Fact]
public void Same_instances_are_mapped_to_same_instances()
{
var configuration = new MapperConfiguration(config => config.CreateMap<Source, Target>());
var mapper = configuration.CreateMapper();
var source = new Source();
var list = mapper.Map<IEnumerable<Target>>(new[] { source, source, source });
list.Distinct().Should().HaveCount(1);
}
public class Source
{
}
public class Target
{
}
It would like to make this test pass.
Also, please notice that the mapper should "remember" instances mapped during the current Map call.

That's the nearest you can get:
public static class Program
{
public static void Main()
{
var config = new MapperConfiguration(conf => conf.CreateMap<Source, Target>().PreserveReferences());
var mapper = config.CreateMapper();
var source = new Source();
var list = new[] { source, source, source };
var firstRun = mapper.Map<IEnumerable<Target>>(list);
var secondRun = mapper.Map<IEnumerable<Target>>(list);
// Returns two items
var diffs = firstRun.Concat(secondRun).Distinct();
foreach (var item in diffs)
{
Console.WriteLine(item.Id);
}
}
}
public class Source
{
public Guid Id { get; } = Guid.NewGuid();
}
public class Target
{
public string Id { get; set; }
}
This means, that you only get the same item back in case of each call to mapper.Map(), but if you call the mapper multiple times, you'll get back new items for each call (which makes sense, otherwise the mapper had to hold references to all given and created instances over it's whole lifetime, which could lead to some serious memory problems).

MongoDB Concurrent writing/fetching from multiple processes causes bulk write operation error

I'm currently implementing a MongoDB database for caching.
I've made a very generic client, with the save method working like this:
public virtual void SaveAndOverwriteExistingCollection<T>(string collectionKey, T[] data)
{
if (data == null || !data.Any())
return;
var collection = Connector.MongoDatabase.GetCollection<T>(collectionKey.ToString());
var filter = new FilterDefinitionBuilder<T>().Empty;
var operations = new List<WriteModel<T>>
{
new DeleteManyModel<T>(filter),
};
operations.AddRange(data.Select(t => new InsertOneModel<T>(t)));
try
{
collection.BulkWrite(operations, new BulkWriteOptions { IsOrdered = true});
}
catch (MongoBulkWriteException mongoBulkWriteException)
{
throw mongoBulkWriteException;
}
}
With our other clients, calling this method looking similar to this:
public Person[] Get(bool bypassCache = false)
{
Person[] people = null;
if (!bypassCache)
people = base.Get<Person>(DefaultCollectionKeys.People.CreateCollectionKey());
if (people.SafeAny())
return people;
people = Client<IPeopleService>.Invoke(s => s.Get());
base.SaveAndOverwriteExistingCollection(DefaultCollectionKeys.People.CreateCollectionKey(), people);
return people;
}
After we've persisted data to the backend we reload the cache from MongoDB by calling our Get methods, passing the argument true. So we reload all of the data.
This works fine for most use cases. But considering how we are using a Web-garden solution (multiple processes) for the same application this leads to concurrency issues. If I save and reload the cache while another user is reloading the page, sometimes it throws a E11000 duplicate key error collection.
Command createIndexes failed: E11000 duplicate key error collection:
cache.Person index: Id_1_Name_1_Email_1 dup
key: { : 1, : "John Doe", : "foo#bar.com" }.
Considering how this is a web garden with multiple IIS processes running, locking won't do much good. Considering how bulkwrites should be threadsafe I'm a bit puzzled. I've looked into Upserting the data, but changing our clients to be type specific and updating each field will take too long and feels like unnecessary work. Therefore I'm looking for a very generic solution.
UPDATE
Removed the Insert and Delete. Changed it to a collection of ReplaceOneModel. Currently experiencing issues with only the last element in a collection being persisted.
public virtual void SaveAndOverwriteExistingCollection<T>(string collectionKey, T[] data)
{
if (data == null || !data.Any())
return;
var collection = Connector.MongoDatabase.GetCollection<T>(collectionKey.ToString());
var filter = new FilterDefinitionBuilder<T>().Empty;
var operations = new List<WriteModel<T>>();
operations.AddRange(data.Select(t => new ReplaceOneModel<T>(filter, t) { IsUpsert = true }));
try
{
collection.BulkWrite(operations, new BulkWriteOptions { IsOrdered = true });
}
catch (MongoBulkWriteException mongoBulkWriteException)
{
throw mongoBulkWriteException;
}
}
Just passed in a collection of 811 items and only the last one can be found in the collection after this method has been executed.
Example of a DTO being persisted:
public class TranslationSetting
{
[BsonId(IdGenerator = typeof(GuidGenerator))]
public object ObjectId { get; set; }
public string LanguageCode { get; set; }
public string SettingKey { get; set; }
public string Text { get; set; }
}
With this index:
string TranslationSettings()
{
var indexBuilder = new IndexKeysDefinitionBuilder<TranslationSetting>()
.Ascending(_ => _.SettingKey)
.Ascending(_ => _.LanguageCode);
return MongoDBClient.CreateIndex(DefaultCollectionKeys.TranslationSettings, indexBuilder);
}

C# - How to return an Array to set a Class property

I am trying to create a Class Method which can be called to Query the Database. The function itself works but for some reason, when the Array is returned, they're not set.
My function code is:
public Configuration[] tbl_bus(string type, string match)
{
// Create Obejct Instance
var db = new rkdb_07022016Entities2();
// Create List
List<Configuration> ConfigurationList = new List<Configuration>();
// Allow Query
if (type.ToLower() == "bustype")
{
foreach (var toCheck in db.tblbus_business.Where(b => b.BusType == match))
{
// Create Class Instance
var model = new Configuration { Name = toCheck.Name, BusinessID = toCheck.BusinessID };
// Append to the property
ConfigurationList.Add(model);
}
}
else if (type.ToLower() == "businessid")
{
foreach (var toCheck in db.tblbus_business.Where(b => b.BusinessID == match))
{
// Create Class Instance
var model = new Configuration { Name = toCheck.Name, BusinessID = toCheck.BusinessID };
// Append to the property
ConfigurationList.Add(model);
}
}
return ConfigurationList.ToArray();
}
And my Configuration code is:
public class Configuration
{
// Properties of the Database
public string Name { get; set; }
public string BusinessID { get; set; }
public string Address { get; set; }
}
public Configuration Config { get; set; }
public Controller()
{
this.Config = new Configuration();
}
On my Handler I am doing:
// Inside the NameSpace area
Controller ctrl;
// Inside the Main Void
ctrl = new Controller();
ctrl.tbl_bus("bustype", "CUS");
context.Response.Write(ctrl.Config.Name);
I tried watching the Class function and it does create the Array, only, when I watch the ctrl.Config.Name it is always set to NULL. Could anyone possibly help me in understanding why the return isn't actually setting the properties inside the Configuration class?
Edit: The function does run and it fetches 3006 rows of Data when matching the bus_type to customer. (Its a large Database) - Only, the properties are never set on return.
Edit: Is there a specific way to return an Array to a Class to set the Properties?
Thanks in advance!

Change your Configs in Controller to array
public Configuration[] Configs { get; set; }
Change your tbl_bus function to void, and set the Configs inside the function.
public void tbl_bus(string type, string match)
{
// do your code
// set the configs here
Configs = ConfigurationList.ToArray();
}
Hope it helps.

Although this is not a complete answer to your question, the problem probably lies in the fact that you're not doing anything with the array returned by the method. You're simply discarding it right away. If you change your code to
ctrl = new Controller();
Configuration[] config = ctrl.tbl_bus("bustype", "CUS");
you will be able to reference the array later on.
Console.WriteLine(config.Length);
Now you can use it to set any properties you like.

DbContext doesn't release SQLite database

First, these are my intentions:
Create a DbContext on SQLite
Read and write from/to it
Close context
Move the file to another location
Points 1-3 work perfectly. The problem starts when I try to move the database. I get an error stating that:
'The process cannot access the file because it is being used by another process.'
How can I resolve this?
First, I create a context. I have to use it in several methods, and I don't want to create it every time I need it. So I am storing it as a member.
_sqliteContext = new SqlLiteContext(sqliteContextName);
Then I want to access a table called sync and get its latest entry.
var sync = _sqliteContext.Syncs.OrderByDescending(s => s.Date);
_lastSync = sync.Any() ? sync.First().Date : new DateTime(0);
That's it. Then I close the context.
_sqliteContext.Dispose();
And try to move the file.
File.Move(sqliteUploadLocation, sqliteDownloadLocation);
This is where I get the exception.
When I replace the selection with an insert, like the following:
var sync = new Sync { Id = Guid.NewGuid().ToString(), Date = DateTime.Now };
_sqliteContext.Syncs.Add(sync);
_sqliteContext.SaveChanges();
This works, and I can move the database. Any ideas why my selection doesn't release its lock?
Update
// Start synchronisation.
new SyncManager(mssqlString, sqliteUploadLocation).Start();
// Move file from upload to download location.
try
{
File.Move(sqliteUploadLocation, sqliteDownloadLocation);
}
catch (Exception ex)
{
Console.WriteLine("Moving failed!");
Console.WriteLine(ex.Message);
}
public void Start()
{
// Create connection string for the sqlite database.
const string sqliteContextName = "SqLiteContext";
var sqliteConnStringSettings = new ConnectionStringSettings
{
Name = sqliteContextName,
ConnectionString = "Data Source=" + _sqliteUploadLocation + ";Version=3;BinaryGUID=False;",
ProviderName = "System.Data.SQLite"
};
// Read configuration, delete available connection strings and add ours.
var conf = ConfigurationManager.OpenMachineConfiguration();
var connStrings = conf.ConnectionStrings;
connStrings.ConnectionStrings.Remove(sqliteContextName);
connStrings.ConnectionStrings.Add(sqliteConnStringSettings);
try
{
conf.Save(ConfigurationSaveMode.Minimal);
}
catch (Exception)
{
// Insufficient rights to save.
return;
}
ConfigurationManager.RefreshSection("connectionStrings");
// Create connection to the sqlite database.
_sqliteContext = new SqlLiteContext(sqliteContextName);
// Create connection to the mssql database.
_mssqlContext = new MsSqlContext(_mssqlConnString);
// Read last sync date.
var sync = _sqliteContext.Syncs.OrderByDescending(s => s.Date);
_lastSync = sync.Any() ? sync.First().Date : new DateTime(0);
// Synchronize tables.
//SyncTablePerson();
//SyncTableAddressAllocation();
// Creates an entry for this synchronisation.
CreateSyncEntry();
// Release resources.
_sqliteContext.Dispose();
_mssqlContext.Dispose();
}
private void CreateSyncEntry()
{
var sync = new Sync { Id = Guid.NewGuid().ToString(), Date = DateTime.Now };
_sqliteContext.Syncs.Add(sync);
_sqliteContext.SaveChanges();
}
Update 2
public class SqlLiteContext : Context
{
public DbSet<Sync> Syncs { get; set; }
public SqlLiteContext(string connectionString)
: base(connectionString)
{
Database.SetInitializer(new NoOperationStrategy<SqlLiteContext>());
}
protected override void OnModelCreating(DbModelBuilder modelBuilder)
{
modelBuilder.Configurations.Add(new PersonConfig());
modelBuilder.Configurations.Add(new AddressAllocationConfig());
modelBuilder.Configurations.Add(new AddressConfig());
modelBuilder.Configurations.Add(new SyncConfig());
}
}
public class NoOperationStrategy<T> : IDatabaseInitializer<T> where T : DbContext
{
public void InitializeDatabase(T context)
{
}
}
public abstract class Context : DbContext
{
public DbSet<Person> Persons { get; set; }
public DbSet<AddressAllocation> AddressAllocations { get; set; }
public DbSet<Address> Addresses { get; set; }
protected Context(string connectionString)
: base(connectionString)
{
}
}
Refactoring with using
using (var sqliteContext = new SqlLiteContext(_sqliteContextName))
{
// Read last sync date.
var sync = sqliteContext.Syncs.Select(s => s).OrderByDescending(s => s.Date);
var lastSync = sync.Any() ? sync.First().Date : new DateTime(1900, 1, 1);
using (var mssqlContext = new MsSqlContext(_mssqlConnString))
{
SyncTablePerson(sqliteContext, mssqlContext, lastSync);
SyncTableAddressAllocation(sqliteContext, mssqlContext, lastSync);
// Save server changes.
mssqlContext.SaveChanges();
}
// Creates an entry for this synchronisation.
sqliteContext.Syncs.Add(new Sync { Id = Guid.NewGuid().ToString(), Date = DateTime.Now });
// Save local changes.
sqliteContext.SaveChanges();
}

Two things jump to mind:
Make sure Visual Studio isn't locking the database file. Open Server Explorer and if there is a connection to the file make sure its closed or removed altogether.
It's likely that connection pooling is what is holding the connection open. Disable pooling in your connection string like this:
Data Source=e:\mydb.db;Version=3;Pooling=False;
As Matt pointed out you should really use a using statement rather than calling dispose manually. That way if there is an exception the resources are always released properly.

I found another topic with the same problem. After i refactored my code, i added
GC.Collect();
That removed the file lock and i could move the file.
see: https://stackoverflow.com/a/14151917/2462736

Need help identifying a subtle bug.. in IIS? The Session? Elsewhere..?

I have a very subtle bug that I'm having trouble identifying.
Background:
We have 2 sites running off the same application on the same web server.
SiteA -- accessed by www.SiteA.com
SiteB -- accessed by www.SiteB.com
When the request first comes in, the siteId is identified based on the Host and stored in the Session as follows:
protected void Application_BeginRequest(object sender, EventArgs e)
{
string host = Request.Url.Host;
int siteId = new SiteManager().GetSiteByUrl(host).SiteId;
if (SessionUser.Instance().SiteId != siteId)
{
SessionUser.Instance().SiteId = siteId;
}
}
This ID used later in determining what data to retreive and to determine what style to present:
// this happens during an initialization phase
_siteConfiguration = _siteManager.GetSiteById(SessionUser.Instance().SiteId);
// then later:
private void SetPageTheme()
{
string theme = null;
switch (_siteConfiguration.SiteId)
{
case ConfigurationHelper.SITE.A:
theme = "SiteATheme";
break;
case ConfigurationHelper.SITE.B:
theme = "SiteBTheme";
break;
}
Page.Theme = theme;
}
The problem:
the problem I'm facing is if you load both sites at almost exactly the same time, i.e. within milliseconds, sometimes SiteA will load with SiteB's theme and vice versa. This doesn't happen very often but it has come to the attention of the client so it's now a problem.. Something is happening somewhere within those few milliseconds in the difference between SiteA loading and SiteB loading, and I don't know how to identify what that is.
The question:
Does anyone have any ideas what could be going wrong here? Something is getting confused somewhere. Is it IIS mixing up the requests? Is it the Session mixing up the site it's supposed to return the SiteId for?
If any more info is needed, I'll supply it.
Update:
For reference, this is the definition of SessionUser (basically, create a static instance of an object to get the SiteId value from the Session):
public class SessionUser
{
private static SessionUser _instance;
public int SiteId { get; set; }
/// <summary>
///
/// </summary>
/// <returns></returns>
public static SessionUser Instance()
{
if (null == _instance)
{
_instance = new SessionUser();
if (null != HttpContext.Current.Session)
{
if (null == HttpContext.Current.Session["User"])
{
if (HttpContext.Current.Request.QueryString["sid"] != null)
{
int nSiteId = int.Parse(HttpContext.Current.Request.QueryString["sid"]);
_instance.SiteId = nSiteId;
HttpContext.Current.Session["User"] = _instance;
}
}
else
{
_instance = (SessionUser) HttpContext.Current.Session["User"];
}
}
}
return _instance;
}
}

Without knowing what the 'SessionUser' class looks like, I can only speculate.
I will assume that SessionUser.Instance() returns a 'static' instance (or member rather).
Now, these will be shared across the entire application. So it makes sense that this cannot be shared between 2 sites.
I suggest you rather use HttpContext to store the setting at BeginRequest.
The code will then look like the following:
class SessionUser
{
public static SessionUser Instance()
{
var ctx = HttpContext.Current;
var su = ctx["SessionUser"] as SessionUser;
if (su == null)
{
ctx["SessionUser"] = su = new SessionUser();
}
return su;
}
}

I guess you could put the code that stores the current Site ID inside a lock block, but this may hamper performance. It makes more sense to use something not shared by both sites, as leppie says.
For the lock example:
if (SessionUser.Instance().SiteId != siteId)
{
lock(somePrivateStaticObject)
{
if (SessionUser.Instance().SiteId != siteId)
{
SessionUser.Instance().SiteId = siteId;
}
}
}