Group array of string arrays with LINQ-CodePudding

I have array like this, values are string:

var arr1 = new [] { "H", "item1", "item2" };
var arr2 = new [] { "T", "thing1", "thing2" };
var arr3 = new [] { "T", "thing1", "thing2" };
var arr4 = new [] { "END", "something" };
var arr5 = new [] { "H", "item1", "item2" };
var arr6 = new [] { "T", "thing1", "thing2" };
var arr7 = new [] { "T", "thing1", "thing2" };
var arr8 = new [] { "END", "something" };

var allArrays = new [] { arr1, arr2, arr3, arr4, arr5, arr6, arr7, arr8 };

I need to group this in to a new array of arrays, so that one array has arrays that start with H or T. The END records (not included in the results) are the delimiters between each section; each new array starts after an END array.

In the end I would like to have somethng like this:

[
[ [H, item1, item2], [T, thing1, thing2], [T, thing1, thing2] ]
[ [H, item1, item2], [T, thing1, thing2], [T, thing1, thing2] ]
]

I know how I can do this with for each loop, but I'm looking for a cleaner way, possibly using linq. All suggestions are much valued, thank you!

CodePudding user response：

you can try this

List<string[]> list = new List<string[]>();
var newArr = allArrays.Select(a => AddToArr(list, a)).Where(a => a != null);

and helper (this code can be put inline, but it easier to read this way)

private static string[][] AddToArr(List<string[]> list, string[] arr)
{
    if (arr[0] != "END")
    {
        list.Add(arr);
        return null;
    }
    var r = list.ToArray();
    list.Clear();
    return r;
}

result

[
[["H","item1","item2"],["T","thing1","thing2"],["T","thing1","thing2"]],
[["H","item3","item4"],["T","thing3","thing4"],["T","thing5","thing6"]]
]

CodePudding user response：

So arr1, arr2, etc are string[]. allArrays is a string[][].

I hope you gave a meaningful example. From this example it seems that you want all string[] from allArrays, except the string[] that have a [0] that equals the word "END".

If this is what you want, your result is:

string[][] result = allArrays.Where(stringArray => stringArray[0] != "END");

I need to group this in to a new array of arrays, so that one array has arrays that start with H or T. The END records (not included in the results) are the delimiters between each section; each new array starts after an END array.

This is not exactly the same as I see in your example: what if one of the string arrays in allArrays is an empty array, or if it has the value null values. What if one of the the arrays of strings is empty (= length 0), and what if one of the string arrays doesn't start with "H", nor "T", nor "END"?

Literally you say that you only want the string arrays that start with "H" or "T", no other ones. You don't want string arrays that are null, nor empty string arrays. You also don't want string arrays that start with "END", nor the ones that start with String.Empty, or "A" or "B" or anything else than "H" or "T".

If I take your requirement literally, your code should be:

string[] requiredStringAtIndex0 = new string[] {"H", "T"};
string[][] result = allArrays.Where(stringArray => stringArray != null
    && stringArray.Length != 0
    && requiredStringAtIndex0.Contains(stringArray[0]));

In words: from allArrays, keep only those arrays of strings, that are not null, AND that have at least one element AND where the element at index 0 contains either "H" or "T"

CodePudding user response：

Normally I would use an extension method for grouping runs of items based on a predicate, in this case GroupByEndingWith and then throw away the "END" record, like so:

var ans = allArrays.GroupByEndingWith(r => r[0] == "END")
                   .Select(g => g.Drop(1).ToArray())
                   .ToArray();

But, in general, you can use Aggregate to collect items based on a predicate at the expense of comprehension. It often helps to use a tuple to track an overall accumulator and a sub-accumulator. Unfortunately, there is no operator or Append for List<T> that returns the original list (helpful for expression based accumulation) and since C# doesn't yet have a comma operator equivalent, you need an extension method again or you can use ImmutableList.

Using Aggregate and ImmutableList, you can do:

var ans = allArrays.Aggregate(
                (ans: ImmutableList<ImmutableList<string[]>>.Empty, curr: ImmutableList<string[]>.Empty),
                (ac, r) => r[0] == "END"
                                ? (ac.ans.Add(ac.curr), ImmutableList<string[]>.Empty)
                                : (ac.ans, ac.curr.Add(r))
          ).ans
          .Select(l => l.ToArray())
          .ToArray();

NOTE: You can also do this with List if you are willing to create new Lists a lot:

var ans = allArrays.Aggregate(
                (ans: new List<List<string[]>>(), curr: new List<string[]>()),
                (ac, r) => r[0] == "END"
                                ? (ac.ans.Concat(new[] { ac.curr }).ToList(), new List<string[]>())
                                : (ac.ans, ac.curr.Concat(new[] { r }).ToList())
          ).ans
          .Select(l => l.ToArray())
          .ToArray();