Fix/hitrate empty topk (#3719)

Fixes #3718 
Description:
- Added validation for empty `top_k` list raising `ValueError` instead
of crashing with `IndexError` in `update()`
- Added support for `top_k: list[int] | int` so users can pass a single
int without wrapping in a list
- Added type validation raising `ValueError` if `top_k` is neither int
nor list
 - Fixed leading space in existing error message
Check list:

- [x] New tests are added (if a new feature is added)
- [x] New doc strings: description and/or example code are in RST format
- [x] Documentation is updated (if required)

---------

Co-authored-by: vfdev <vfdev.5@gmail.com>

Author: rogueslasher

ignite/metrics/rec_sys/hitrate.py

@@ -28,7 +28,9 @@ class HitRate(Metric):
     - returns a list of HitRate ordered by the sorted values of ``top_k``.
 
     Args:
-        top_k: a list of sorted positive integers that specifies `k` for calculating hitrate@top-k.
+        top_k: a single positive integer or a list of positive integers that specifies `k` for
+            calculating hitrate@top-k. If a single int is provided, it will be wrapped in a list.
+            Default is 1.
         ignore_zero_hits: if True, users with no relevant items (ground truth tensor being all zeros)
             are ignored in computation of HitRate. if set False, such users are counted as a miss.
             By default, True.
@@ -97,22 +99,50 @@ class HitRate(Metric):
 
             [0.0, 0.5, 0.5, 0.5]
 
+        int top_k case
+
+        .. testcode:: 3
+
+            metric = HitRate(top_k=2)
+            metric.attach(default_evaluator, "hit_rate")
+            y_pred = torch.Tensor([
+                [4.0, 2.0, 3.0, 1.0],
+            ])
+            y_true = torch.Tensor([
+                [0.0, 0.0, 1.0, 0.0],
+            ])
+            state = default_evaluator.run([(y_pred, y_true)])
+            print(state.metrics["hit_rate"])
+
+        .. testoutput:: 3
+
+            [1.0]
+
     .. versionadded:: 0.5.4
+    .. versionchanged:: 0.5.4
+        `top_k` now accepts a single positive integer in addition to a list of integers.
     """
 
     required_output_keys = ("y_pred", "y")
     _state_dict_all_req_keys = ("_hits_per_k", "_num_examples")
 
     def __init__(
         self,
-        top_k: list[int],
+        top_k: list[int] | int = 1,
         ignore_zero_hits: bool = True,
         output_transform: Callable = lambda x: x,
         device: str | torch.device = torch.device("cpu"),
         skip_unrolling: bool = False,
     ):
+        if not isinstance(top_k, (int, list)):
+            raise ValueError("top_k must be either int or a list[int]")
+
+        top_k = [top_k] if isinstance(top_k, int) else top_k
+
+        if len(top_k) == 0:
+            raise ValueError("top_k must have at least one positive value")
         if any(k <= 0 for k in top_k):
-            raise ValueError(" top_k must be list of positive integers only.")
+            raise ValueError("top_k must be list of positive integers only.")
 
         self.top_k = sorted(top_k)
         self.ignore_zero_hits = ignore_zero_hits

tests/ignite/metrics/rec_sys/test_hitrate_metric.py

@@ -54,6 +54,26 @@ def test_shape_mismatch():
         metric.update((y_pred, y))
 
 
+def test_empty_top_k():
+    with pytest.raises(ValueError, match="top_k must have at least one positive value"):
+        HitRate(top_k=[])
+
+
+def test_invalid_top_k_type():
+    with pytest.raises(ValueError, match="top_k must be either int or a list"):
+        HitRate(top_k="invalid")
+
+
+def test_int_top_k(available_device):
+    metric = HitRate(top_k=2, device=available_device)
+    y_pred = torch.tensor([[4.0, 2.0, 3.0, 1.0]])
+    y_true = torch.tensor([[0.0, 0.0, 1.0, 0.0]])
+    metric.update((y_pred, y_true))
+    res = metric.compute()
+    expected = manual_hit_rate(y_pred.numpy(), y_true.numpy(), [2])
+    np.testing.assert_allclose(res, expected)
+
+
 @pytest.mark.parametrize("top_k", [[1], [1, 2, 4]])
 @pytest.mark.parametrize("ignore_zero_hits", [True, False])
 def test_compute(top_k, ignore_zero_hits, available_device):