I have a self referencing Categories table. Each Category has a CategoryID, ParentCategoryID, CategoryName, etc. And each category can have any number of sub categories, and each of those sub categories can have any number of sub categories, and so and and so forth. So basically the tree can be X levels deep.
Then Products are associated to leaf (sub) Categories. Is there a way to get all the Products for any given Category (which would be all the products associated to all its leaf descendants) using LINQ to SQL?
This feels like a recursive problem. Is it better to used a Stored Procedure instead?
I don't think linq-to-sql has a good answer to this problem. Since you are using sql server 2005 you can use CTEs to do hierarchical queries. Either a stored procedure or an inline query (using DataContext.ExecuteQuery) will do the trick.
Well here is a terrible rushed implementation using LINQ.
Don't use this :-)
public IQueryable GetCategories(Category parent)
{
var cats = (parent.Categories);
foreach (Category c in cats )
{
cats = cats .Concat(GetCategories(c));
}
return a;
}
The performant approach is to create an insert/modify/delete trigger which maintains an entirely different table which contains node-ancestor pairs for all ancestors of all nodes. This way, the lookup is O(N).
To use it for getting all products belonging to a node and all of its descendants, you can just select all category nodes which have your target node as an ancestor. After this, you simply select any products belonging to any of these categories.
The way I handle this is by using some extension methods (filters). I've written up some sample code from a project I have implemented this on. Look specifically at the lines where I'm populating a ParentPartner object and a SubPartners List.
public IQueryable<Partner> GetPartners()
{
return from p in db.Partners
select new Partner
{
PartnerId = p.PartnerId,
CompanyName = p.CompanyName,
Address1 = p.Address1,
Address2 = p.Address2,
Website = p.Website,
City = p.City,
State = p.State,
County = p.County,
Country = p.Country,
Zip = p.Zip,
ParentPartner = GetPartners().WithPartnerId(p.ParentPartnerId).ToList().SingleOrDefault(),
SubPartners = GetPartners().WithParentPartnerId(p.PartnerId).ToList()
};
}
public static IQueryable<Partner> WithPartnerId(this IQueryable<Partner> qry, int? partnerId)
{
return from t in qry
where t.PartnerId == partnerId
select t;
}
public static IQueryable<Partner> WithParentPartnerId(this IQueryable<Partner> qry, int? parentPartnerId)
{
return from p in qry
where p.ParentPartner.PartnerId == parentPartnerId
select p;
}
Related
I'm trying to combine 3 separate SOQL queries into one, but still end up with 3 separate lists for ease of use and readability later.
List<Object__c> objectList = [SELECT Name, Id, Parent_Object__r.Name, Parent_Object__r.Id,
(SELECT Name, Id FROM Child_Objects__r)
FROM Object__c];
I know I can get a list of child objects thus:
List<Child_Object__c> childObjectList = new List<Child_Object__c>();
for(Object__c object : objectList){
childObjectList.addAll(object.Child_Objects__r);
}
How would I go about adding the Parent_Object__c records to their own list?
I'm assuming a map could be used to deal with duplicates, but how do I get this Parent_Object__c data into that map?
You are basically there.
All lookup fields are available in your example as object.Parent_Object__r. Use a Set to natively avoid duplicates. No deduping required on your part!
Set<Parent_Object__c> parentObjectSet = new Set<Parent_Object__c>();
List<Child_Object__c> childObjectList = new List<Child_Object__c>();
for(Object__c object : objectList){
childObjectList.addAll(object.Child_Objects__r);
parentObjectSet.add(object.Parent_Object__r);
}
Edit:
As per #eyescream (trust him!) you are indeed better off with a map to avoid duplicates.
So the above code would just be slightly different:
Map<Id, Parent_Object__c> parentObjectMap = new Map<Id, Parent_Object__c>();
List<Child_Object__c> childObjectList = new List<Child_Object__c>();
for(Object__c object : objectList){
childObjectList.addAll(object.Child_Objects__r);
if (object.Parent_Object__r != null) {
parentObjectMap.put(object.Parent_Object__r.Id, object.Parent_Object__r);
}
}
The dapper tutorial gives this example to help a user with Multi Mapping (One to Many)
While this works I am curious why they have you store the orders in the dictionary but then in the end they use a linq.Distinct() and return from the list. It seems like it would be cleaner to just return the ordersDictionary.Values as the dictionary logic ensures no duplicates.
//Tutorial
using (var connection = new SqlConnection(FiddleHelper.GetConnectionStringSqlServerW3Schools()))
{
Dictionary<int,Order> orderDictionary = new Dictionary<int, Order>();
List<Order> list = connection.Query<Order, OrderDetail, Order>(sql, (order, orderDetail) =>
{
if (!orderDictionary.TryGetValue(order.OrderID, out Order orderEntry))
{
orderEntry = order;
orderEntry.OrderDetails = new List<OrderDetail>();
orderDictionary.Add(orderEntry.OrderID, orderEntry);
}
orderEntry.OrderDetails.Add(orderDetail);
return orderEntry;
}, splitOn: "OrderID")
.Distinct()
.ToList();
return list;
}
//my suggestion
using (var connection = new SqlConnection(FiddleHelper.GetConnectionStringSqlServerW3Schools()))
{
Dictionary<int,Order> orderDictionary = new Dictionary<int, Order>();
//change 1 no need to store into list here
connection.Query<Order, OrderDetail, Order>(sql, (order, orderDetail) =>
{
if (!orderDictionary.TryGetValue(order.OrderID, out Order orderEntry))
{
orderEntry = order;
orderEntry.OrderDetails = new List<OrderDetail>();
orderDictionary.Add(orderEntry.OrderID, orderEntry);
}
orderEntry.OrderDetails.Add(orderDetail);
return orderEntry;
}, splitOn: "OrderID"); //change 2 remove .Distinct().ToList()
return orderDictionary.Values.ToList(); //change 3 return dictionaryValues
}
I'm the author of this tutorial: https://dapper-tutorial.net/query#example-query-multi-mapping-one-to-many
why they have you store the orders in the dictionary
A row is returned for every OrderDetail. So you want to make sure to add the OrderDetail to the existing Order and not create a new one for every OrderDetail. The dictionary is used for performance to check if the Order has been already created or not.
it would be cleaner to just return the ordersDictionary.Values
How will your query return dictionary values?
Of course, if you are in a method such as yours, you can do
var list = orderDictionary.Values;
return list;
But how to make this Connection.Query return dictionary values? An order is returned for every row/OrderDetail, so the order will be returned multiple times.
Outside the Query, your dictionary solution works great and is even a better solution for performance, but if you want to make your Query return the distinct list of orders without using Distinct or some similar method, it's impossible.
EDIT: Answer comment
my suggestion return orderDictionary.Values.ToList(); //change 3 return dictionaryValues
Thank you for your great feedback, it's always appreciated ;)
It would be weird in a tutorial to use what the query returns when there is no relationship but use the dictionary for one to many relationships
// no relationship
var orders = conn.Query<Order>("", ...).Distinct();
// one to many relationship
conn.Query<Order, OrderDetail>("", ...);
var orders = orderDictionary.Values.ToList();
Your solution is better for performance the way you use it, there is no doubt about this. But this is how people usually use the Query method:
var orders = conn.Query("", ...).Distinct();
var activeOrders = orders.Where(x => x.IsActive).ToList();
var inactiveOrders = orders.Where(x => !x.IsActive).ToList();
They use what the Query method returns.
But again, there is nothing wrong with the way you do it, this is even better if you can do it.
I have a database table which has a bunch of fields including one called Type and another called Code. I also have a list of Type and Code pairs that are encapsulated in a class. Something like this:
public class TypeAndCode
{
public byte Type { get; set; }
public int Code { get; set; }
// overrides for Equals and GetHashCode to compare if Type and Code are equal
}
Now what I need to do is select from the table only those entries who type AND code match an entry in my collection. So, for example, I tried something like this:
var query = myTable.Where(a => myTCList.Contains(new TypeAndCode() { Type = a.Type, Code = a.Code }).ToList();
But it'll give me a NotSupportedException:
Unable to create a constant value of type 'TypeAndCode'. Only primitive types
or enumeration types are supported in this context.
Is there a way to make this work so that I can retrieve from the database only those entries that have a Code and Type that match my list of Code and Type? I'm trying to avoid having to retrieve all the entries (it's a big table) and match them in memory.
I'm aware that I could try something like
var query = myTable.Where(a => listOfTypes.Contains(a.Type) && listOfCodes.Contains(a.Codes))
But that will make some spurious matches where the type and code are from different pairs in my original list.
You can use Any instead:
var query = myTable
.Where(a => myTCList.Any(t => t.Type == a.Type && t.Code == a.Code ))
.ToList();
You should be able to just do this manually without the overloaded methods from your class:
myTCList.Any(x => x.Type == a.Type && x.Code == a.Code)
My ulitmate solution here, in case anybody else encounters a similar problem, was to set up a temporary table that I could write the pairs I wanted to match to which I could them join with the database table. After doing the join and materializing the results, you can delete the temporary table.
Something like:
ctx.myTempTable = (from pair in mypairs
select new myTempTable() { Type = pair.Type, Code = pair.Code }).ToList();
ctx.SaveChanges();
var query = from q in myTable
join t in ctx.myTempTable
on new { q.Type, q.Code } equals new { t.Code, t.Type }
select q;
The whole thing is in a try/catch/finally block with the finally block used to clear up the temporary table(s)
I have the following entities:
class Book():
team = db.StringProperty()
class Entry():
role = db.StringProperty()
I create two Books:
book1 = Book(team='Plants').put()
book2 = Book(team='Zombies').put()
I create three Entrys:
entry1 = Entry(parent=book1, role='Peashooter').put()
entry2 = Entry(parent=book2, role='Gargantuar').put()
entry3 = Entry(parent=book2, role='Flag Zombie').put()
I can make a query for the 'Plants' Book:
query = Entry.all().ancestor(book1).fetch(100)
I'd like to make a query for several books. I can only think of list concatenation:
list1 = Entry.all().ancestor(book1).fetch(100)
list2 = Entry.all().ancestor(book2).fetch(100)
query = list1 + list2
Is there a more elegant way to do this other than list concatenation?
If you want to combine queries for several different ancestors (that don't share a common ancestor), the only way to do it is with separate queries, as you demonstrate. This is no different than if you want to query for several distinct values in a field.
I'm trying to figure out how to execute a custom query with Castle ActiveRecord.
I was able to run simple query that returns my entity, but what I really need is the query like that below (with custom field set):
select count(1) as cnt, data from workstationevent where serverdatetime >= :minDate and serverdatetime < :maxDate and userId = 1 group by data having count(1) > :threshold
Thanks!
In this case what you want is HqlBasedQuery. Your query will be a projection, so what you'll get back will be an ArrayList of tuples containing the results (the content of each element of the ArrayList will depend on the query, but for more than one value will be object[]).
HqlBasedQuery query = new HqlBasedQuery(typeof(WorkStationEvent),
"select count(1) as cnt, data from workstationevent where
serverdatetime >= :minDate and serverdatetime < :maxDate
and userId = 1 group by data having count(1) > :threshold");
var results =
(ArrayList)ActiveRecordMediator.ExecuteQuery(query);
foreach(object[] tuple in results)
{
int count = (int)tuple[0]; // = cnt
string data = (string)tuple[1]; // = data (assuming this is a string)
// do something here with these results
}
You can create an anonymous type to hold the results in a more meaningful fashion. For example:
var results = from summary in
(ArrayList)ActiveRecordMediator.ExecuteQuery(query)
select new {
Count = (int)summary[0], Data = (string)summary[1]
};
Now results will contain a collection of anonymous types with properties Count and Data. Or indeed you could create your own summary type and populate it out this way too.
ActiveRecord also has the ProjectionQuery which does much the same thing but can only return actual mapped properties rather than aggregates or functions as you can with HQL.
Be aware though, if you're using ActiveRecord 1.0.3 (RC3) as I was, this will result in a runtime InvalidCastException. ActiveRecordMediator.ExecuteQuery returns an ArrayList and not a generic ICollection. So in order to make it work, just change this line:
var results = (ICollection<object[]>) ActiveRecordMediator.ExecuteQuery(query);
to
var results = (ArrayList) ActiveRecordMediator.ExecuteQuery(query);
and it should work.
Also note that using count(1) in your hql statement will make the query return an ArrayList of String instead of an ArrayList of object[] (which is what you get when using count(*).)
Just thought I'd point this out for the sake of having it all documented in one place.