I'm working on an algorithm which can generate 2 types of recommendations, restaurants and dishes. All of this works fine, but I wanted to merge these 2 types of recommendations in a single list, which is where I encountered some issues. From my previous question I concluded that I needed a wrapper class, which I have set up like this:
public class RecommenderItem
{
public Guid Id { get; set; }
public object Entity { get; set; }
}
Now I want to alternate the 2 types of recommendations so the list would look like this:
[Restaurant][Dish][Restaurant][Dish][Restaurant][Dish] //Etc...
Note that these recommendations are completely separate. They are generated purely based on the user's preference, and they have no correlation in between them. My product owner wants to show these recommendations on the home page of our app like this.
These lists are different in length, so if I have added all items from a list, I wanted to just add the remaining objects from the other list. A possible scenario of this could look like this:
/*Other objects before this...*/[Dish][Restaurant][Dish][Dish][Dish] //Etc...
Here did the list of restaurant objects run out and I just wanted to add the remaining dish recommendations at the end of the list.
I have gotten this far, but I'm unsure how I would catch an IndexOutOfBounds exception and add the rest of the remaining objects at the end.
public List<RecommenderItem> GetMergedRecommendationLists(List<Restaurant> restaurantRecommendations,
List<Dish> dishRecommendations)
{
//Setting up the output list.
List<RecommenderItem> output = new List<RecommenderItem>();
int count = 0;
//Check which list is longer and use that count
if (restaurantRecommendations.Count > dishRecommendations.Count)
count = dishRecommendations.Count;
else
count = restaurantRecommendations.Count;
for (int i = 0; i < count; i++)
{
//I'm fully aware this isn't the most optimal way of doing this,
//but I'm only looking at functionality here, optimizing performance comes later.
var restRecommendation = restaurantRecommendations[i];
var dishRecommendation = dishRecommendations[i];
output.Add(new RecommenderItem()
{
Id = restRecommendation.Id,
Entity = restRecommendation
});
output.Add(new RecommenderItem()
{
Id = dishRecommendation.Id,
Entity = dishRecommendation
});
}
return output;
}
Does anyone have an idea how I could do this? Could I just catch an IndexOutOfBounds exception and use .AddRange() for the remaining objects? I'm not sure how I could check which list was out of bounds.
Let me know if I should elaborate more and thanks in advance!
Edit: -removed because it wasn't fair.-
This is a fairly succinct way of doing this.
While not Linq, it works in the spirit of the way Linq works by deferring doing any work until the resulting sequence is enumerated:
public static IEnumerable<RecommenderItem> Merge(IEnumerable<Restaurant> restaurants, IEnumerable<Dish> dishes)
{
using (var r = restaurants.GetEnumerator())
using (var d = dishes.GetEnumerator())
{
while (true)
{
bool rAvailable = r.MoveNext();
bool dAvailable = d.MoveNext();
if (rAvailable)
yield return new RecommenderItem { Id = r.Current.Id, Entity = r.Current };
if (dAvailable)
yield return new RecommenderItem { Id = d.Current.Id, Entity = d.Current };
if (!rAvailable && !dAvailable)
break;
}
}
}
If you happen to be using the MoreLinq NuGet package that includes the ZipLongest extension method, you can use the following simplified implementation instead:
public static IEnumerable<RecommenderItem> Merge(IEnumerable<Restaurant> restaurants, IEnumerable<Dish> dishes)
{
foreach (var item in restaurants.ZipLongest(dishes, (r, d) => new { r, d }))
{
if (item.r != null)
yield return new RecommenderItem { Id = item.r.Id, Entity = item.r };
if (item.d != null)
yield return new RecommenderItem { Id = item.d.Id, Entity = item.d };
}
}
Addendum
As #InBetween posted in his answer, you can put the interleaving logic into an extension method. Here's my version; it's substantially the same, except I've added a small optimisation to avoid calling .MoveNext() when its not necessary:
public static class EnumerableExt
{
public static IEnumerable<T> Interleave<T>(this IEnumerable<T> a, IEnumerable<T> b)
{
using (var ae = a.GetEnumerator())
using (var be = b.GetEnumerator())
{
bool aAvailable = true;
bool bAvailable = true;
while (aAvailable || bAvailable)
{
aAvailable = aAvailable && ae.MoveNext();
bAvailable = bAvailable && be.MoveNext();
if (aAvailable)
yield return ae.Current;
if (bAvailable)
yield return be.Current;
}
}
}
}
Once you have that, I realised that you don't need to write an implict operator. Instead, you can just convert the two sequences to the resultant type before calling Interleave() like so:
var restaurantsAsRecommenderItems =
restaurantRecommendations
.Select(r => new RecommenderItem {Id = r.Id, Entity = r});
var dishesAsRecommenderItems =
dishRecommendations
.Select(d => new RecommenderItem {Id = d.Id, Entity = d});
var result =
restaurantsAsRecommenderItems
.Interleave(dishesAsRecommenderItems)
.ToList();
My recommendation would be to just make simple implicit operator :
public static implicit operator RecommenderItem(Restaurant restaurant) {
return new RecommenderItem { Id = restaurant.Id, Entity = restaurant };
}
Then you have possibility to convert these types easily like :
Restaurant rest = //...
RecommenderItem rItem = rest; // here the implicit operator is called
After doing this you can just use one for loop :
int count = Math.Max(restaurantRecommendations.Count, dishRecommendations.Count);
for ( int i = 0; i < count; i++ ) {
if ( i < restRecommendations.Count )
output.Add(restRecommendations[i]);
if ( i < dishRecommendations.Count )
output.Add(dishRecommendations[i]);
}
This will make your work much more easier.
Well, there are probably more elegant LINQ solutions but you have already most, it's also a very efficient approach:
public List<RecommenderItem> GetMergedRecommendationLists(List<Restaurant> restaurantRecommendations, List<Dish> dishRecommendations)
{
//Setting up the output list.
List<RecommenderItem> output = new List<RecommenderItem>();
int count = Math.Min(restaurantRecommendations.Count, dishRecommendations.Count);
for (int i = 0; i < count; i++)
{
var restRecommendation = restaurantRecommendations[i];
var dishRecommendation = dishRecommendations[i];
output.Add(new RecommenderItem()
{
Id = restRecommendation.Id,
Entity = restRecommendation
});
output.Add(new RecommenderItem()
{
Id = dishRecommendation.Id,
Entity = dishRecommendation
});
}
int remainingRestaurant = restaurantRecommendations.Count - count;
int remainingDishes = dishRecommendations.Count - count;
if (remainingRestaurant > 0)
{
for (int i = count; i < restaurantRecommendations.Count; i++)
{
var restRecommendation = restaurantRecommendations[i];
output.Add(new RecommenderItem()
{
Id = restRecommendation.Id,
Entity = restRecommendation
});
}
}
else if (remainingDishes > 0)
{
for (int i = count; i < dishRecommendations.Count; i++)
{
var dishRecommendation = dishRecommendations[i];
output.Add(new RecommenderItem()
{
Id = dishRecommendation.Id,
Entity = dishRecommendation
});
}
}
return output;
}
A simple way of doing it would be:
public static IEnumerable<T> Merge<T>(this IEnumerable<T> first, IEnumerable<T> second)
{
using (var firstEnumerator = first.GetEnumerator())
using (var secondEnumerator = second.GetEnumerator())
{
while (firstEnumerator.MoveNext())
{
yield return firstEnumerator.Current;
if (secondEnumerator.MoveNext())
{
yield return secondEnumerator.Current;
}
}
while (secondEnumerator.MoveNext())
{
yield return secondEnumerator.Current;
}
}
}
After having created two arrays of restaurants and dishes of the same type RecommenderItem, you can use the Zip method like :
var restaurants = restaurantRecommendations.Select(x => new RecommenderItem {
Id = x.Id,
Entity = x
}).ToArray();
var dishes = dishRecommendations.Select(x => new RecommenderItem {
Id = x.Id,
Entity = x
}).ToArray();
var output = restaurants.Zip(dishes, (r, d) => new[] { r, d })
.SelectMany(r => r).Concat(dishes.Skip(restaurants.Length))
.Concat(restaurants.Skip(dishes.Length));
Restaraunt and Dish would have to share a base type:
restaurantRecommendations.Select(item => new RecommenderItem()
{
Id = item.Id,
Entity = item
});
dishRecommendations.Select(item => new RecommenderItem()
{
Id = item.Id,
Entity = item
});
Once that's the case you could use something like this slightly modified version of Zip (from System.Linq):
private static IEnumerable<T> ZipThrough<T>(IEnumerable<T> first, IEnumerable<T> second)
{
if (first == null) throw new ArgumentNullException(nameof(first));
if (second == null) throw new ArgumentNullException(nameof(second));
using (var e1 = first.GetEnumerator())
{
using (var e2 = second.GetEnumerator())
{
while (true)
if (e1.MoveNext())
{
yield return e1.Current;
if (e2.MoveNext()) yield return e2.Current;
}
else if (e2.MoveNext())
{
yield return e2.Current;
}
else
{
break;
}
}
}
}
Related
I am working on a function which will modify the input payload properties. The payload contains nodes and each node contains list of features. I need to remove some specific features match condition and also modify each node window property start and end time. I have written the function using traditional nested for loop, but struggling to convert it to Linq function. Anyone has idea how to convert this nested for loop function to a Linq function?
private void ApplyTransformation(InputPayload payload, int startTime = 8, int endTime = 15)
{
var nodes = payload.Nodes;
for (var i = 0; i < nodes.Count(); ++i)
{
var node = nodes[i];
var features = node.Features;
for (var j = 0; j < features.Count(); ++j)
{
var feature = features[j];
if (feature.NodeFeatureTypeID
== FeatureTypeEnum.FEATURE_A
|| feature.FeatureTypeID == FeatureTypeEnum.FEATURE_B
|| feature.FeatureTypeID == FeatureTypeEnum.FEATURE_C
|| feature.FeatureTypeID == FeatureTypeEnum.FEATURE_D
)
{
features.RemoveAt(j);
}
}
var windows = node.Windows;
for (var k = 0; k < windows.Count(); ++k)
{
var window = windows[k];
if (window.NodeFunctionTypeID == FeatureTypeEnum.MM_HOURS) continue;
window.StartHour = new TimeSpan(startTime, 0, 0);
window.EndHour = new TimeSpan(endTime, 0, 0);
}
}
}
Let's do it in parts. This first code of yours removes features that are in a list,
for (var j = 0; j < features.Count(); ++j)
{
var feature = features[j];
if (feature.NodeFeatureTypeID
== FeatureTypeEnum.FEATURE_A
|| feature.FeatureTypeID == FeatureTypeEnum.FEATURE_B
|| feature.FeatureTypeID == FeatureTypeEnum.FEATURE_C
|| feature.FeatureTypeID == FeatureTypeEnum.FEATURE_D
)
{
features.RemoveAt(j);
}
}
We need to convert that to "keep features not in a list"
var discard = new [] {FeatureTypeEnum.FEATURE_A, FeatureTypeEnum.FEATURE_B, FeatureTypeEnum.FEATURE_C, FeatureTypeEnum.FEATURE_D };
node.Features = node.Features.Where(f => !discard.Contains(f)).ToArray();
This part of your code skips if function type is a certain kind, and zeroes timespans if the function type is not a certain kind:
if (window.NodeFunctionTypeID == FeatureTypeEnum.MM_HOURS) continue;
window.StartHour = new TimeSpan(startTime, 0, 0);
window.EndHour = new TimeSpan(endTime, 0, 0);
Which might be better as a loop that operates on just those things we are interested in (we want to modify only those Windows where the nodefuctionTypeId is not Hours):
foreach(var window in node.Windows.Where(w => w.NodeFunctionTypeID != FeatureTypeEnum.MM_HOURS){
window.StartHour = new TimeSpan.FromHous(startTime);
window.EndHour = new TimeSpan.FromHours(endTime);
}
Meaning the whole is:
private void ApplyTransformation(InputPayload payload, int startTime = 8, int endTime = 15)
{
var discard = new [] {FeatureTypeEnum.FEATURE_A, FeatureTypeEnum.FEATURE_B, FeatureTypeEnum.FEATURE_C, FeatureTypeEnum.FEATURE_D };
foreach (var node in payload.Nodes)
{
node.Features = node.Features.Where(f => !discard.Contains(f)).ToArray();
foreach(var window in node.Windows.Where(w => w.NodeFunctionTypeID != FeatureTypeEnum.MM_HOURS){
window.StartHour = TimeSpan.FromHous(startTime);
window.EndHour = TimeSpan.FromHours(endTime);
}
}
}
I don't think I'd convert it all to a Linq form, as it would make a mess; linq queries should not have side effects (modify the objects the query iterates over) so particularly the second part where the time spans are being zeroed would have to become an operation where everything about the window was being copied over to a new Window if you wanted to LINQ it.
If a window is literally just a time span pair then it's not so bad, or if you want to provide a constructor that takes an existing Window and a startTime and endTime:
public Window(Window copyFrom, int startTime, int endTime){
this.X = copyFrom.X;
this.Y = copyFrom.Y;
...
this.StartHour = TimeSpan.FromHours(strtTime);
this.EndHour = TimeSpan.FromHours(endTime);
}
Then maybe your method could become some linq:
foreach (var node in payload.Nodes)
{
node.Features = node.Features.Where(f => !discard.Contains(f)).ToArray();
node.Windows = node.Windows.Select(w => w.NodeFunctionTypeID == FeatureTypeEnum.MM_HOURS ? w : new Window(w, startTime, endTime).ToArray();
}
..but I don't know if I would try for a wholesale replacement of the entire nodes list, for reasons even based on the name of the method: ApplyTransformations sounds like it means "take this list of nodes and change bits of them" not "take this list of nodes and give me a new list with some or all of the nodes replaced or modified" - that sort of behavior in code could wreck something else, if the calling code is expecting a tweak and the object it sends (or objects within it) are swapped out for new ones
Using a linq query for the second part would make things messier, better to just have a for loop
Something like this should work:
var nodes = payload.Nodes;
nodes.Features = nodes.Features.Where(f => !(
f.NodeFeatureTypeID == FeatureTypeEnum.FEATURE_A
|| feature.FeatureTypeID == FeatureTypeEnum.FEATURE_B
|| feature.FeatureTypeID == FeatureTypeEnum.FEATURE_C
|| feature.FeatureTypeID == FeatureTypeEnum.FEATURE_D
)
);
foreach(var window in nodes.Windows)
{
if (window.NodeFunctionTypeID != FeatureTypeEnum.MM_HOURS)
{
window.StartHour = new TimeSpan(startTime, 0, 0);
window.EndHour = new TimeSpan(endTime, 0, 0);
}
}
Pls have look at below code snippet to avoid nested loop using linq.
public class Employee
{
public string Name { get; set; }
public List<EmployeeProject> EmployeeProject { get; set; }
}
public class EmployeeProject
{
public string ProjectName { get; set; }
public string ClientName { get; set; }
}
/// <summary>
/// Object mapping
/// </summary>
/// <returns></returns>
public static List<Employee> CreateObject()
{
var employeeList = new List<Employee>();
var employee = new Employee();
var employeeProjectList = new List<EmployeeProject>();
employee.Name = "John";
var employeeProject = new EmployeeProject();
employeeProject.ProjectName = "Chrome";
employeeProject.ClientName = "Google";
employeeProjectList.Add(employeeProject);
employeeProject = new EmployeeProject();
employeeProject.ProjectName = "WhatsApp";
employeeProject.ClientName = "Meta";
employeeProjectList.Add(employeeProject);
employee.EmployeeProject = employeeProjectList;
employeeList.Add(employee);
employee.Name = "Alex";
employeeProjectList = new List<EmployeeProject>();
employeeProject = new EmployeeProject();
employeeProject.ProjectName = "Chrome2";
employeeProject.ClientName = "Google2";
employeeProjectList.Add(employeeProject);
employeeProject = new EmployeeProject();
employeeProject.ProjectName = "WhatsApp2";
employeeProject.ClientName = "Meta2";
employeeProjectList.Add(employeeProject);
employee.EmployeeProject = employeeProjectList;
employeeList.Add(employee);
return employeeList;
}
/// <summary>
/// Linq function
/// </summary>
public static void LinqFunctionForNestedQuery()
{
var employeeObject = CreateObject();
var result1 = employeeObject.Select(x =>
{
x.EmployeeProject = x.EmployeeProject.Select(y =>
{
y.ProjectName.Contains("Chrome");
return y;
}).ToList();
return x;
});
}
To maintain readability of the code and to make it simpler , I have modified the code.
Step1 : Replace for with foreach if you can
Step2 : Replace foreach with linq
Important tip. Resharper helps you with code suggestions.
Solution
private void ApplyTransformation(InputPayload payload, int startTime = 8, int endTime = 15)
{
foreach (var node in payload.Nodes)
{
var features = node.Features;
foreach (var feature in features.Where(feature => feature.NodeFeatureTypeID == FeatureTypeEnum.FEATURE_A ||
feature.FeatureTypeID == FeatureTypeEnum.FEATURE_B ||
feature.FeatureTypeID == FeatureTypeEnum.FEATURE_C ||
feature.FeatureTypeID == FeatureTypeEnum.FEATURE_D))
{
features.Remove(feature);
}
var windows = node.Windows;
foreach (var window in windows
.Where(window => window.NodeFunctionTypeID != FeatureTypeEnum.MM_HOURS))
{
window.StartHour = new TimeSpan(startTime, 0, 0);
window.EndHour = new TimeSpan(endTime, 0, 0);
}
}
}
In my C# project with EF Core 5.0, I have several independents tables: Clothes, Hairs, Makeup. Some of their columns are similar, but some are not..`
I need to write a method that will load the rows from these tables. What I have now is:
public async Task<(ClothesDbModel[] clothes, MakeupDbModel[] makeups, HairDbModel[] hairs)> GetDressup(int[] clothesIds, int[] makeupIds, int[] hairIds)
{
ClothesDbModel[] clothes = new ClothesDbModel[0];
if (clothesIds.Length > 0)
{
clothes = await _dbContext.Clothes.Where(c => clothesIds.Contains(c.Id)).ToArrayAsync();
}
MakeupDbModel[] makeups = new MakeupDbModel[0];
if (makeupIds.Length > 0)
{
makeups = await _dbContext.Makeups.Where(c => makeupIds.Contains(c.Id)).ToArrayAsync();
}
HairDbModel[] hairs = new HairDbModel[0];
if (hairIds.Length > 0)
{
hairs = await _dbContext.Hairs.Where(c => hairIds.Contains(c.Id)).ToArrayAsync();
}
return (clothes, makeups, hairs);
}
However, in this case, I have 3 separate queries to the database (3 awaits). I believe that is not the best way to load the data from a performance point of view. Maybe I can load the same data using DbContext only once&
I hope it will work with EF. But I know there are problems with Concat, so it may fail.
class CombinedResult
{
public ClothesDbModel Clothes;
public MakeupDbModel Makeup;
public HairDbModel Hair;
}
...
public async Task<(ClothesDbModel[] clothes, MakeupDbModel[] makeups, HairDbModel[] hairs)> GetDressup(int[] clothesIds, int[] makeupIds, int[] hairIds)
{
var queries = new List<IQueryable<CombinedResult>>();
if (clothesIds.Length > 0)
{
queries.Add(_dbContext.Clothes.Where(c => clothesIds.Contains(c.Id)).Select(c => new CombinedResult { Clothes = c }));
}
if (makeupIds.Length > 0)
{
queries.Add(_dbContext.Makeups.Where(c => makeupIds.Contains(c.Id)).Select(c => new CombinedResult { Makeup = c }));
}
if (hairIds.Length > 0)
{
queries.Add(_dbContext.Hairs.Where(c => hairIds.Contains(c.Id)).Select(c => new CombinedResult { Hair = c }));
}
var clothes = new List<ClothesDbModel>();
var makeups = new List<MakeupDbModel>();
var hairs = new List<HairDbModel>();
if (queries.Count > 0)
{
var query = queries[0];
for (var i = 1; i < queries.Count; i++)
{
query = query.Concat(queries[i]);
}
var items = await query.ToListAsync();
foreach (var item in items)
{
if (item.Clothes != null)
clothes.Add(item.Clothes);
else if (item.Makeup != null)
makeups.Add(item.Makeup);
else if (item.Hair != null)
hairs.Add(item.Hair);
}
}
return (clothes.ToArray(), makeups.ToArray(), hairs.ToArray());
}
This one should not be too hard but my mind seems to be having a stack overflow (huehue). I have a series of Lists and I want to find all permutations they can be ordered in. All of the lists have different lengths.
For example:
List 1: 1
List 2: 1, 2
All permutations would be:
1, 1
1, 2
In my case I don't switch the numbers around. (For example 2, 1)
What is the easiest way to write this?
I can't say if the following is the easiest way, but IMO it's the most efficient way. It's basically a generalized version of the my answer to the Looking at each combination in jagged array:
public static class Algorithms
{
public static IEnumerable<T[]> GenerateCombinations<T>(this IReadOnlyList<IReadOnlyList<T>> input)
{
var result = new T[input.Count];
var indices = new int[input.Count];
for (int pos = 0, index = 0; ;)
{
for (; pos < result.Length; pos++, index = 0)
{
indices[pos] = index;
result[pos] = input[pos][index];
}
yield return result;
do
{
if (pos == 0) yield break;
index = indices[--pos] + 1;
}
while (index >= input[pos].Count);
}
}
}
You can see the explanation in the linked answer (shortly it's emulating nested loops). Also since for performace reasons it yields the internal buffer w/o cloning it, you need to clone it if you want store the result for later processing.
Sample usage:
var list1 = new List<int> { 1 };
var list2 = new List<int> { 1, 2 };
var lists = new[] { list1, list2 };
// Non caching usage
foreach (var combination in lists.GenerateCombinations())
{
// do something with the combination
}
// Caching usage
var combinations = lists.GenerateCombinations().Select(c => c.ToList()).ToList();
UPDATE: The GenerateCombinations is a standard C# iterator method, and the implementation basically emulates N nested loops (where N is the input.Count) like this (in pseudo code):
for (int i0 = 0; i0 < input[0].Count; i0++)
for (int i1 = 0; i1 < input[1].Count; i1++)
for (int i2 = 0; i2 < input[2].Count; i2++)
...
for (int iN-1 = 0; iN-1 < input[N-1].Count; iN-1++)
yield { input[0][i0], input[1][i1], input[2][i2], ..., input[N-1][iN-1] }
or showing it differently:
for (indices[0] = 0; indices[0] < input[0].Count; indices[0]++)
{
result[0] = input[0][indices[0]];
for (indices[1] = 0; indices[1] < input[1].Count; indices[1]++)
{
result[1] = input[1][indices[1]];
// ...
for (indices[N-1] = 0; indices[N-1] < input[N-1].Count; indices[N-1]++)
{
result[N-1] = input[N-1][indices[N-1]];
yield return result;
}
}
}
Nested loops:
List<int> listA = (whatever), listB = (whatever);
var answers = new List<Tuple<int,int>>;
for(int a in listA)
for(int b in listB)
answers.add(Tuple.create(a,b));
// do whatever with answers
Try this:
Func<IEnumerable<string>, IEnumerable<string>> combine = null;
combine = xs =>
xs.Skip(1).Any()
? xs.First().SelectMany(x => combine(xs.Skip(1)), (x, y) => String.Format("{0}{1}", x, y))
: xs.First().Select(x => x.ToString());
var strings = new [] { "AB", "12", "$%" };
foreach (var x in combine(strings))
{
Console.WriteLine(x);
}
That gives me:
A1$
A1%
A2$
A2%
B1$
B1%
B2$
B2%
I made the following IEnumerable<IEnumerable<TValue>> class to solve this problem which allows use of generic IEnumerable's and whose enumerator returns all permutations of the values, one from each inner list. It can be conventiently used directly in a foreach loop.
It's a variant of Michael Liu's answer to IEnumerable and Recursion using yield return
I've modified it to return lists with the permutations instead of the single values.
using System;
using System.Collections;
using System.Collections.Generic;
using System.Linq;
namespace Permutation
{
public class ListOfListsPermuter<TValue> : IEnumerable<IEnumerable<TValue>>
{
private int count;
private IEnumerable<TValue>[] listOfLists;
public ListOfListsPermuter(IEnumerable<IEnumerable<TValue>> listOfLists_)
{
if (object.ReferenceEquals(listOfLists_, null))
{
throw new ArgumentNullException(nameof(listOfLists_));
}
listOfLists =listOfLists_.ToArray();
count = listOfLists.Count();
for (int i = 0; i < count; i++)
{
if (object.ReferenceEquals(listOfLists[i], null))
{
throw new NullReferenceException(string.Format("{0}[{1}] is null.", nameof(listOfLists_), i));
}
}
}
// A variant of Michael Liu's answer in StackOverflow
// https://stackoverflow.com/questions/2055927/ienumerable-and-recursion-using-yield-return
public IEnumerator<IEnumerable<TValue>> GetEnumerator()
{
TValue[] currentList = new TValue[count];
int level = 0;
var enumerators = new Stack<IEnumerator<TValue>>();
IEnumerator<TValue> enumerator = listOfLists[level].GetEnumerator();
try
{
while (true)
{
if (enumerator.MoveNext())
{
currentList[level] = enumerator.Current;
level++;
if (level >= count)
{
level--;
yield return currentList;
}
else
{
enumerators.Push(enumerator);
enumerator = listOfLists[level].GetEnumerator();
}
}
else
{
if (level == 0)
{
yield break;
}
else
{
enumerator.Dispose();
enumerator = enumerators.Pop();
level--;
}
}
}
}
finally
{
// Clean up in case of an exception.
enumerator?.Dispose();
while (enumerators.Count > 0)
{
enumerator = enumerators.Pop();
enumerator.Dispose();
}
}
}
IEnumerator IEnumerable.GetEnumerator()
{
return GetEnumerator();
}
}
}
You can use it directly in a foreach like this:
public static void Main(string[] args)
{
var listOfLists = new List<List<string>>()
{
{ new List<string>() { "A", "B" } },
{ new List<string>() { "C", "D" } }
};
var permuter = new ListOfListsPermuter<string>(listOfLists);
foreach (IEnumerable<string> item in permuter)
{
Console.WriteLine("{ \"" + string.Join("\", \"", item) + "\" }");
}
}
The output:
{ "A", "C" }
{ "A", "D" }
{ "B", "C" }
{ "B", "D" }
I want to compute rank of element in an IEnumerable list and assign it to the member. But below code works only when called 1st time. 2nd time call starts from last rank value. So instead of output 012 and 012, I'm getting 012 and 345
class MyClass
{
public string Name { get; set; }
public int Rank { get; set; }
}
public void SecondTimeRankEvaluvate()
{
MyClass[] myArray = new MyClass[]
{
new MyClass() { Name = "Foo" },
new MyClass() { Name = "Bar" },
new MyClass() { Name = "Baz" }
};
int r = 0;
var first = myArray.Select(s => { s.Rank = r++; return s; });
foreach (var item in first)
{
Console.Write(item.Rank);
}
// Prints 012
Console.WriteLine("");
foreach (var item in first)
{
Console.Write(item.Rank);
}
// Prints 345
}
I understand that the variable r is being captured (closure) and reused when called 2nd time. I don't want that behavior. Is there any clean way to compute rank and assign it?
Also r variable (in actual code) isn't in the same scope where foreach loop is present. It is in a function which returns var first
var first = myArray.Select((s, i) => { s.Rank = i; return s; });
LINQ uses lazy evaluation and runs the Select part every time you use myArray.
You can force evaluation to happen only once by storing the result in a List.
Change
var first = myArray.Select(s => { s.Rank = r++; return s; });
to
var first = myArray.Select(s => { s.Rank = r++; return s; }).ToList();
Another way would be to join myArray with a new sequence using Zip every time, like this
var first = myArray.Zip(Enumerable.Range(0, int.MaxValue), (s, r) =>
{
s.Rank = r;
return s;
});
You shouldn't use LINQ if you're not querying the collection.
To update each item in an array, use a for loop:
for (int i = 0; i < myArray.Length; i++)
{
myArray[i].Rank = i;
}
To update each item in an enumerable, use a foreach loop:
int r = 0;
foreach (var item in myArray)
{
item.Rank = r++;
}
I have CreateDiscountViewByUser discountViewByUser it contains a list of cities that are chosen by the user, but they may already be those cities that have been added.
List<DiscountCity> discountCities = (from city in db.DiscountCities
where city.DiscountId == discountViewByUser.Id
select city).ToList();
for (int y = 0; y < discountCities.Count(); y++)
{
var dc = discountCities[y];
bool flag = false;
for (int i = 0; i < discountViewByUser.DiscountCitys.Length; i++)
{
if (dc.CityId == discountViewByUser.DiscountCitys[i])
{
flag = true;
discountCities.Remove(dc);
y--;
}
}
if (!flag)
{
db.DiscountCities.DeleteObject(dc);
}
}
foreach (var dc in discountCities)
{
DiscountCity discountCity = new DiscountCity
{Id = Guid.NewGuid(),
CityId = dc.CityId,
DiscountId = main.Id};
db.DiscountCities.AddObject(discountCity);
}
how to add only the new city?
My code does not work = (
UPDATE:
discountViewByUser.DiscountCitys type int[].
db.DiscountCities table: Id DiscountId CityId.
example:
in database: Odessa, Kiev
user set: Odessa, Moscow.
I need delete Kiev and add moscow how do this?
What I recommend is adding all the items and then removing duplicates.
// Where uniqueList is a List<T> of unique items:
uniqueList.AddRange(valuesToAdd);
uniqueList = uniqueList.Distinct(new CityEqualityComparer()).ToList();
// Sorry, I don't know how this would fit into your code
Since you are comparing cities by their CityId's, you will probably need to use a custom IEqualityComparer to determine which cities are duplicates.
Here is an example of such a class:
class CityEqualityComparer : IEqualityComparer<City>
{
public bool Equals(City arg1, City arg2)
{
return arg1.CityId == arg2.CityId;
}
public int GetHashCode(City arg)
{
return arg.CityId;
}
}
This question may also be of some help.
I suggest you do this in 2 steps.
1) Find the cities to be deleted
var deleteCities = db.DiscountCities.Where(c => c.DiscountId == discountViewByUser.Id
&& !discountViewByUser.DiscountCitys.Contains(c.CityId));
foreach(deleteCity in deleteCities)
{
db.DiscountCities.DeleteObject(deleteCity);
}
2) Find cityId's to be inserted
var insertCities = discountViewByUser.DiscountCitys.Except(
db.DiscountCities.Where(c => c.DiscountId == discountViewByUser.Id)
.Select(c => c.CityId));
foreach(var insertCity in insertCities)
{
DiscountCity discountCity = new DiscountCity
{Id = Guid.NewGuid(), CityId = insertCity, DiscountId = discountViewByUser.Id};
db.DiscountCities.AddObject(discountCity);
}