CommentCollection
{
"_id":"5b63f0f23846b70011330889",
"CommentType":"task",
"EntityReferenceId":"6082ef25-6f9a-4874-a832-f72e0f693409",
"CommentLink":null,
"EntityName":"task2",
"participants":[
ObjectId("52ffc4a5d85242602e000000"),
ObjectId("52ffc4a5d85242602e000001")
],
"Threads":[
{
"_id":"69bcef71-3695-4340-bdec-4a6e4c58c490",
"CommentType":"task",
"UserId":ObjectId("52ffc4a5d85242602e000000"),
"CommentByUserType":"Admin",
"EntityReferenceId":"6082ef25-6f9a-4874-a832-f72e0f693409",
"Content":"fdffd",
"ProjectName":null,
"PostedDate":"2018-08-03T13:03:05.939Z",
"Active":true,
"Attachment":[
]
}
another Collection is
userCollection
{
"Id":ObjectId("52ffc4a5d85242602e000000"),
"Name":"Pms Admin",
"Email":"pms#xtrastaff.com",
"Type":"Admin",
"UserId":"6082ef25-6f9a-4874-a832-f72e0f693409",
"UserImage":"6082ef25-6f9a-4874-a832-f72e0f693409"
}
In the CommentCollection there is an array of "participants" which is storing the id's of users (from usercollection).
My requirement is join these two collections for getting user details in my asp.net core project(Linq).Participants contains list of id's
In Mongo shell you would use $lookup which can be used on arrays like in this example and your query could look like this:
db.Comment.aggregate([
{
$lookup: {
from: "user",
localField: "participants",
foreignField: "Id",
as: "participants"
}
}
])
Which simply replaces participants with array of objects from second collection:
{
"_id" : "5b63f0f23846b70011330889",
"CommentType" : "task",
"EntityReferenceId" : "6082ef25-6f9a-4874-a832-f72e0f693409",
"CommentLink" : null,
"EntityName" : "task2",
"participants" : [
{
"_id" : ObjectId("5b6e875b9d52833fbe9879c2"),
"Id" : ObjectId("52ffc4a5d85242602e000000"),
"Name" : "Pms Admin",
"Email" : "pms#xtrastaff.com",
"Type" : "Admin",
"UserId" : "6082ef25-6f9a-4874-a832-f72e0f693409",
"UserImage" : "6082ef25-6f9a-4874-a832-f72e0f693409"
}
],
"Threads" : //...
}
In C# you can express that using Lookup syntax. First option allows you to get a list of BsonDocument type which simply skips type checking:
var collection = db.GetCollection<Comment>("Comment");
List<BsonDocument> result = collection.Aggregate()
.Lookup("user", "participants", "Id", "participants")
.ToList();
The reason why you can't use regular LINQ join here is that actually you're compaing an array with a scalar value (that's what should be in equals part of join). However if you need strongly typed result instead of BsonDocuments you can use different version of Lookup method which takes types and expressions instead of strings. So you need another class for $lookup result which might be solved using inheritance:
public class Comment
{
public string _id { get; set; }
public string CommentType { get; set; }
public string EntityReferenceId { get; set; }
public string CommentLink { get; set; }
public string EntityName { get; set; }
public ObjectId[] participants { get; set; }
public Thread[] Threads { get; set; }
}
public class CommentWithUsers : Comment
{
public User[] Users { get; set; }
}
Then you can get a list of CommentWithUser:
var comments = mydb.GetCollection<Comment>("Comment");
var users = mydb.GetCollection<User>("user");
List<CommentWithUser> result = comments.Aggregate()
.Lookup<Comment, User, CommentWithUsers>(
users,
x => x.participants,
x => x.Id,
x => x.Users).ToList();
Related
I'm trying to allow my entities to be ordered by the Last-Modified property in their metadata, using OData query options.
I tried using a transformer as described in Converting to JSON and accessing metadata, but when I apply ODataQueryOptions to the resulting IQueryable, I get an empty array.
The model and view-model:
public class Foo
{
public int Id { get; set; }
}
public class FooViewModel
{
public string Id { get; set; }
public DateTime LastModified { get; set; }
}
The transformer:
public class Foos_WithLastModified : AbstractTransformerCreationTask<Foo>
{
public Foos_WithLastModified()
{
TransformResults = foos => from foo in foos
let metadata = MetadataFor(foo)
select new
{
Id = foo.Id.ToString(CultureInfo.InvariantCulture),
LastModified = metadata.Value<DateTime>("Last-Modified")
};
}
}
The relevant method in FooController (_session is an IAsyncDocumentSession):
public async Task<ICollection<FooViewModel>> Get(ODataQueryOptions<FooViewModel> options)
{
var settings = new ODataValidationSettings();
settings.AllowedOrderByProperties.Add("LastModified");
options.Validate(settings);
var foos = _session.Query<Foo>()
.TransformWith<Foos_WithLastModified, FooViewModel>();
var odataFoos = (IQueryable<FooViewModel>)options.ApplyTo(foos);
return await odataFoos.ToListAsync();
}
When I hit /api/Foo, the results are as expected:
[
{
"Id": "foos/456",
"LastModified": "2015-11-23T08:43:10.913662Z"
},
{
"Id": "foos/123",
"LastModified": "2015-11-23T08:50:34.0907996Z"
}
]
But when I add OData query options (/api/Foo?$orderby=LastModified), I get an empty array: [].
I also tried changing _session to an IDocumentSession and modifying Get as follows,
[EnableQuery(AllowedOrderByProperties = "LastModified")]
public IQueryable<FooViewModel> Get()
{
return _session.Query<Foo>()
.TransformWith<Foos_WithLastModified, FooViewModel>();
}
but I get the same results.
Are transformers the wrong approach? How can I sort by Last-Modified using OData query options?
I do not know how to handle the OData stuff, never tried that, but in order to query for entities, ordered by the metadata value "Last-Modified" using only RavenDB techniques you can do the following:
Create an index for your entity (in my example a Customer). In this index we add the field LastModified that's using the document's metadata value for Last-Modified.
public class Customer_ByLastModified : AbstractIndexCreationTask<Customer>
{
public class QueryModel
{
public DateTime LastModified { get; set; }
}
public Customer_ByLastModified()
{
Map = customers => from customer in customers
select new
{
LastModified = this.MetadataFor(customer).Value<DateTime>("Last-Modified")
};
}
}
The QueryModel isn't mandatory, but it makes querying via the client API easier, imo. You can then add a Transformer to be able to use the metadata value in your return model:
public class Customers_WithLastModified : AbstractTransformerCreationTask<Customer>
{
public Customers_WithLastModified()
{
TransformResults = results => from customer in results
select new CustomerViewModel
{
Id = customer.Id,
Name = customer.Name,
LastModified = MetadataFor(customer).Value<DateTime>("Last-Modified")
};
}
}
And then query it like this:
using (var session = documentStore.OpenSession())
{
var customers = session.Query<Customer_ByLastModified.QueryModel, Customer_ByLastModified>()
.OrderByDescending(x => x.LastModified)
.TransformWith<Customers_WithLastModified, CustomerViewModel>()
.ToList();
}
Hope this helps!
I have a JSON file like this:
{
"date": "2012-01-05",
"rate": {
"GBP": "5.2549",
"BGN": "2.2189",
"JPY": "0.043985",
"AUD": "3.4648",
"EGP": "0.5601",
"SEK": "0.4898",
"BRL": "1.8444",
"RSD": "0.0412",
"EUR": "4.3398",
"CAD": "3.3165",
"USD": "3.3796",
"HUF": "0.013449",
"MXN": "0.2459",
"MDL": "0.2867",
"NOK": "0.5646",
"CZK": "0.1672",
"KRW": "0.002925",
"RUB": "0.1057",
"PLN": "0.9596",
"CNY": "0.5363",
"NZD": "2.6406",
"UAH": "0.4194",
"XDR": "5.1605",
"TRY": "1.7911",
"DKK": "0.5837",
"INR": "0.0638",
"CHF": "3.5629",
"XAU": "175.2918",
"AED": "0.9201",
"ZAR": "0.4114"
}
}
I need to get the values from rate as a List of Model (List<Model> )
public class Model
{
public string Currency {get; set;}
public double Value {get; set;}
}
Thanks!
It can be done in 2 steps.
Step - 1:
Deserialize the json string into a class like:
public class RootObject
{
public string date { get; set; }
public Dictionary<string, double> rate { get; set; }
}
with the following familiar code:
string json = "YOUR_JSON";
RootObject root = JsonConvert.DeserializeObject<RootObject>(json);
Step - 2:
Now you can extract the Dictionary as list of Models using the following LINQ:
List<Model> models = root.rate.Select(x => new Model {Currency = x.Key, Value = x.Value}).ToList();
When using Json.NET to serialize a MVC view model to json, I have a generic object property on my view model (public object Result { get; set;}) that is getting serialized to key-value pairs instead of an actual json object. Is there a converter I can use to force it to be serialized properly?
This is what is currently being output by Json.NET:
"result": [
{
"key": "AssetId",
"value": "b8d8fb71-2553-485b-91bf-14c6c563d78b"
},
{
"key": "SearchResultType",
"value": "Assets"
},
{
"key": "Name",
"value": "abstract-q-c-1920-1920-8"
}
]
Instead, I would want it to output this:
"result": {
"AssetId": "b8d8fb71-2553-485b-91bf-14c6c563d78b",
"SearchResultType": "Assets",
"Name": "abstract-q-c-1920-1920-8"
}
EDIT:
To answer the question of how that property is getting populated, it is via a RavenDB index:
public class SiteSearchIndexTask : AbstractMultiMapIndexCreationTask<SiteSearchResult>
{
public class Result
{
public object[] Content { get; set; }
}
public override string IndexName
{
get
{
return "SiteSearch/All";
}
}
public SiteSearchIndexTask()
{
AddMap<Asset>(items => from item in items
where item.IsDeleted == false
select new
{
Id = item.Id.ToString(),
ObjectId = item.Id,
ResultType = SearchResultType.Assets,
Title = item.Name.Boost(3),
Tags = item.Tags.Select(x => x.Name).Boost(2),
Result = (object)item,
Query = string.Join(" ", item.Description)
});
AddMap<User>(items => from item in items
where item.IsDeleted == false
select new
{
Id = item.Id,
ObjectId = item.UserId,
ResultType = SearchResultType.Users,
Title = item.Username.Boost(3),
Tags = (BoostedValue) null,
Result = (object)item,
Query = string.Join(" ", item.FullName, item.Email)
});
Store(x => x.ObjectId, FieldStorage.Yes);
Store(x => x.ResultType, FieldStorage.Yes);
Store(x => x.Title, FieldStorage.Yes);
Store(x => x.Tags, FieldStorage.Yes);
Store(x => x.Result, FieldStorage.Yes);
Store(x => x.Query, FieldStorage.Yes);
}
}
Edit 2
Here are the Asset and User models (truncated for brevity, since they're just a bunch of auto properties)
public class Asset : IHasId
{
public string Id { get; set; }
public Guid AssetId
{
get
{
Guid assetId;
Guid.TryParse((Id ?? string.Empty).Replace("assets/", ""), out assetId);
return assetId;
}
set { Id = "assets/" + value; }
}
public string Name { get; set; }
public string Description { get; set; }
}
public class User : IHasId
{
public User()
{
Status = UserStatus.Active;
}
public string Id { get; set; }
public int UserId
{
get
{
int userId;
int.TryParse((Id ?? string.Empty).Replace("users/", ""), out userId);
return userId;
}
set { Id = "users/" + value; }
}
public string Username { get; set; }
public string Password { get; set; }
public string Email { get; set; }
public string FirstName { get; set; }
public string LastName { get; set; }
public UserStatus Status { get; set; }
}
Edit 3
As it turns out, when I step through the code, that object is actually of the type Raven.Abstractions.Linq.DynamicJsonObject, which contains an array of key-value pairs. So I guess this question might be more related to Raven than Json.NET. Unless of course there is a converter to go from key-value pair to json object.
It would help to see how you are querying exactly, but you should take a look at the example on querying unlike documents with a multimap index on the RavenDB site.
When I look at your index, it seems like you are trying to do way too much in terms of field storage and encapsulation of your results. It's sometimes difficult to grasp, but the mapping done in the index is not to define how the results are returned, but rather how the index is built. Unless you are doing a Reduce or TransformResults step, you are still returning the original document.
So encapsulating the document as Result = (object)item is overkill. So is having a ResultType enum. If you need to know the type of document that matched, simply use .GetType() on the object and you will quickly see if it is a User or an Asset.
Here is how I would define your search index. Note there's some differences because there were properties you showed in your index that aren't in the models you provided. (I'm assuming there is a separate entity on the backend than the models on the front end, but adjust as needed.)
public class SearchIndex : AbstractMultiMapIndexCreationTask<SearchIndex.Result>
{
public class Result
{
public object[] Content { get; set; }
public string ResultType { get; set; }
}
public SearchIndex()
{
AddMap<Asset>(items => from item in items
select new Result
{
Content = new object[] {item.Name, item.Description},
ResultType = MetadataFor(item)["Raven-Entity-Name"].ToString()
});
AddMap<User>(items => from item in items
select new Result
{
Content = new object[] { item.Username, item.FirstName, item.LastName, item.Email },
ResultType = MetadataFor(item)["Raven-Entity-Name"].ToString()
});
Index(x => x.Content, FieldIndexing.Analyzed);
}
}
And then I would query it like so:
var results = session.Advanced
.LuceneQuery<object, SearchIndex>()
.Where("ResultType:" + resultTypeName)
.AndAlso()
.Search("Content", searchTerm);
You can then simply examine your results and find that while cast as object, you can indeed do result.GetType() and see how that object is constructed. If desired, you could also put a common interface on your entities to cast to instead of object, such as the IAmSearchable shown in the Raven example.
When you finally pass your result back through MVC, it should be serialized properly, since it will be coming from the real object and not the raven DynamicJsonObject.
I have an entity like this:
public class Employment
{
public virtual Company Company {get; set;}
public virtual Person Person {get; set;}
public virtual string Description {get; set;}
}
providing a relationship between two other entities. They have corresponding DTO's and I want to return a result set containing all information on persons and companies. The query is performed on the Employment table and my problem is that Hibernate generates one select statement per company and person.
In my database, The Employment table has 1000 rows. Nhibernate generates 2001 select statements, one for the Employment list and one select for each Person and Company as I map them to DTO's.
I would like hibernate to fetch all information at once, in SQL I would have done something like this:
SELECT e.Description, c.A, c.B, c.C, p.D, p.E, p.F
FROM Employment e
JOIN Company c ON e.Company_Id = c.Company_Id
JOIN Person p ON e.Person_Id = p.Person_Id;
Or even
SELECT Description FROM Employment;
SELECT c.A, c.B, c.C FROM Employment e
JOIN Company c ON e.Company_Id = c.Company_Id;
SELECT p.D, p.E, p.F FROM Employment e
JOIN Person p ON e.Person_Id = p.Person_Id;
I am a pretty fresh user of nHibernate, QueryOver.I welcome Linq-To-Entities answers as well but I prefer to avoid LINQ query expressions.
I've looked all over the web, read about JoinQuery, JoinAlias and Fetch and have come as far as something like this:
//This works, but the objects are retrieved as PersonProxy and CompanyProxy,
//generating 2 SELECT statements for each Employment I map to EmploymentDto
var queryOver =
session.QueryOver<Employment>()
.Fetch(x => x.Person).Eager
.Fetch(x => x.Company).Eager
var mapResult = MappingEngine.Map<IList<EmploymentDto>>(queryOver.List());
//This works, but the objects are still retrieved as PersonProxy and CompanyProxy,
var queryOver =
session.QueryOver<Employment>()
.JoinAlias(x => x.Person, () => personAlias, JoinType.InnerJoin)
.JoinAlias(x => x.Company, () => companyAlias, JoinType.InnerJoin);
var mapResult = MappingEngine.Map<IList<EmploymentDto>>(queryOver.List());
JoinQuery provided the same result aswell. I feel I am missing something important here. Something should be done in the query or before .List() to fetch all child entities instead of loading a list with lots of Employment entities loaded with PersonProxy and CompanyProxy. I can, however, not find out how...
EDIT: Added mapping
Database tables:
TABLE Company(
Id,
A,
B,
C)
TABLE Person(
Id,
D,
E,
F);
TABLE Employment(
Person_Id,
Company_Id,
Description);
Entities
public class Company
{
public virtual string Id { get; set; }
public virtual string A { get; set; }
public virtual bool B { get; set; }
public virtual bool C { get; set; }
}
public class Person
{
public virtual string Id { get; set; }
public virtual string D { get; set; }
public virtual string E { get; set; }
public virtual string F { get; set; }
}
public class Employment
{
public virtual Person Person { get; set; }
public virtual Company Company { get; set; }
public virtual string Description { get; set; }
public override bool Equals(object obj)
{
Employment toCompare = obj as Employment;
if (toCompare == null)
return false;
return (this.GetHashCode() != toCompare.GetHashCode());
}
public override int GetHashCode()
{
unchecked
{
int results = Person != null ? Person.GetHashCode() : 0;
results = (results * 397) ^ (Company != null ? Company.GetHashCode() : 0);
results = (results * 397) ^ (Description != null ? Description.GetHashCode() : 0);
return results;
}
}
}
Mapping
public class CompanyMap : SyncableClassMap<Company>
{
public CompanyMap()
{
Table("Company");
Id(x => x.Id).Column("Id").GeneratedBy.Assigned();
Map(x => x.A).Column("A");
Map(x => x.B).Column("B").CustomType<YesNoType>();
Map(x => x.C).Column("C").CustomType<YesNoType>();
}
}
public class PersonMap : SyncableClassMap<Person>
{
public PersonMap()
{
Table("Person");
Id(x => x.Id).Column("Id").GeneratedBy.Assigned();
Map(x => x.D).Column("D");
Map(x => x.E).Column("E");
Map(x => x.F).Column("F");
}
}
public class EmploymentMap : ClassMap<Employment>
{
public EmploymentMap()
{
Table("Employment");
CompositeId()
.KeyReference(x => x.Person, "Person_Id")
.KeyReference(x => x.Company, "Company_Id");
Map(x => x.Description, "Description");
}
}
after your edit i see you have a keyreference instead of a normal many-to-one.
Unfortunatly this seems to be a limitation of QueryOver/Criteria which does not eager load keyreferences even with Fetchmode specified. However Linq to NH does not have this limitation. Change the query to
using NHibernate.Linq;
var results = session.Query<Employment>()
.Fetch(x => x.Person)
.Fetch(x => x.Company)
.ToList();
I have experienced the same problem you're describing here. I'll take your last code snippet as an example, since that's the way I've made it work:
//This works, but the objects are still retrieved as PersonProxy and CompanyProxy,
var queryOver =
session.QueryOver<Employment>()
.JoinAlias(x => x.Person, () => personAlias, JoinType.InnerJoin)
.JoinAlias(x => x.Company, () => companyAlias, JoinType.InnerJoin);
var mapResult = MappingEngine.Map<IList<EmploymentDto>>(queryOver.List());
First of all, you shouldn't have to specify JoinType.InnerJoin since that's the default join type. Same as with you, I've also found that the persons and companies are loaded lazily this way.
However, if you change the join type to JoinType.LeftOuterJoin, you'll see that everything is loaded eagerly. At least that's what I've experienced. So try change your code to the following:
//This works, but the objects are still retrieved as PersonProxy and CompanyProxy,
var queryOver =
session.QueryOver<Employment>()
.JoinAlias(x => x.Person, () => personAlias, JoinType.LeftOuterJoin)
.JoinAlias(x => x.Company, () => companyAlias, JoinType.LeftOuterJoin);
var mapResult = MappingEngine.Map<IList<EmploymentDto>>(queryOver.List());
I can't explain to you why it is this way, I've just found this to work out of own experiences. And if it's problematic for you doing a left outer join, you can instead try to filter accordingly in your code before (or while) doing the mapping.
For web application in development(ASP.Net MVC), I'm using the telerik grid. The grid is bound to an IQueryable of my list, because it's a big table, and I want that telerik apply it's filter on the list, and then executes this result, not dowloading 10'000 rows(with the joined tables), and then with the filter, use only rows.
I'm using(and I really need it for this page, it's one of the key feature) the filter/order of the grid.
One of the main column(determining the kind of the data) is an enum.
The problem is that I get a "Specified type member is not supported in linq to entities" as soon as I'm trying to filter/sort it.
I've to bind it on the enum(and not the mapped int) because if I use the id, filters/order by will be on an int, and I can't expect that the user knows the id of the foreign table.
I just cannot implement myself again all grids parameter(located in url)(I assume, it's either I do everything, or nothing) and filter it correctly, order it correctly).
Do you have an idea of workaround?
I don't know how your Entity Model looks like but I'll suppose that you've something like this Model:
public partial class Project
{
public int Id { get; set; }
public string Name { get; set; }
public int Status { get; set; }
}
and the Status property represents your enum value then you've this enum:
public enum ProjectStatuses
{
Current = 1,
Started = 2,
Stopped = 3,
Finished = 4,
}
Then just create new ViewModel like this :
public class ProjectDetails
{
public int Id { get; set; }
public string Name { get; set; }
public int Status { get; set; }
public ProjectStatuses StatusValue { get { return (ProjectStatuses) Status; } }
// This property to display in telerik ClientTemplate
public string StatusName { get { return Enum.GetName(typeof (ProjectStatuses), Status ); } }
}
And because I love Extension Methods I'll add this one :
public static class ModelListExtensions
{
public static IQueryable<ProjectDetails> ToViewModelDetails(this IQueryable<Project> modelList)
{
return modelList.Select(m => new ProjectDetails
{
Id = m.Id,
Name = m.Name,
Status = m.Status,
};
}
}
Update :
Here is the Controller
public ActionResult Index()
{
int total;
var viewModel = getGridList(out total);
ViewBag.Total = total;
return View(viewModel);
}
//this Action to get ajax pages
[GridAction(EnableCustomBinding = true)]
public ActionResult ReGetIndex(GridCommand command, int roleId)
{
int total;
var list = getGridList(out total, roleId, command);
return View(new GridModel {Data = list, Total = total});
}
private IEnumerable<ProjectDetails> getGridList(out int total, GridCommand command = null)
{
command = command ?? new GridCommand {Page = 1};
foreach (var descriptor in command.SortDescriptors)
{
if (descriptor.Member == "StatusValue")
descriptor.Member = "Status";
}
foreach (FilterDescriptor descriptor in command.FilterDescriptors)
{
if (descriptor.Member == "StatusValue")
descriptor.Member = "Status";
}
var list = modelService.AllAsQuery()
.ToViewModelDetails() // To convert it to our ViewModel if we have one
.Where(command.FilterDescriptors);
total = list.Count();
return (IEnumerable<ProjectDetails>) list.Sort(command.SortDescriptors)
.Page(command.Page - 1, command.PageSize)
.GroupBy(command.GroupDescriptors).ToIList();
}
And this is the View
#model IEnumerable<ProjectDetails>
#{
Html.Telerik()
.Grid(Model)
.Name("ProjectsGrid")
.Sortable()
.Filterable()
.EnableCustomBinding(true)
.DataBinding(dataBinding => dataBinding
.Ajax()
.Select("ReGetIndex", "Projects"))
.Pageable(page => page.Style(GridPagerStyles.PageSizeDropDown | GridPagerStyles.NextPreviousAndNumeric).Total(ViewBag.Total))
.Columns(column =>
{
column.Bound(m => m.Id).Hidden(true);
column.Bound(m => m.Name);
column.Bound(m => m.StatusValue).ClientTemplate("<#= StatusName #>");
})
.Render();
}
Update :
If you want to enforce at least one sort order you could use something like this:
if (!command.SortDescriptors.Any())
{
command.SortDescriptors.Add(new SortDescriptor {Member = "YourDefaultProperty"});
}
You don't really have choice (or few annoying choices)
Wether you use a class instead of enum (but if you used an enum, that's because it was better).
Or you "pseudo-sort" your enum, and use the mapped int.
public enum TT
{
Brown = 0,
Green = 1
}
Of course, you'll have to check the actual datas (mapped int) in your DB and update them to conform to the new order (can't change enum order without impact). And you'll have to do that everytime you want to insert a value between existing enum values.
Or you wait for next EF / linq / c# version, which should have enum support in linq2entities