Merge branch 'main' into bug/df_index_column_pruning_issue

python/xorbits/_mars/tensor/statistics/bincount.py

@@ -286,8 +286,10 @@ def bincount(x, weights=None, minlength=0, chunk_size_limit=None):
  x = astensor(x)
  weights = astensor(weights) if weights is not None else None
 
- if not np.issubdtype(x.dtype, np.int_):
- raise TypeError(f"Cannot cast array data from {x.dtype} to {np.dtype(np.int_)}")
+ if not np.issubdtype(x.dtype, np.int64):
+ raise TypeError(
+ f"Cannot cast array data from {x.dtype} to {np.dtype(np.int64)}"
+ )
  if x.ndim != 1:
  raise ValueError("'x' must be 1 dimension")
  if minlength < 0:

python/xorbits/_mars/utils.py

@@ -489,13 +489,15 @@ def calc_data_size(dt: Any, shape: Tuple[int] = None) -> int:
  return 0
 
  if isinstance(dt, tuple):
- return sum(calc_data_size(c) for c in dt)
+ # int() for windows CI, otherwise may return numpy.int32 by `sum`
+ return int(sum(calc_data_size(c) for c in dt))
 
  shape = getattr(dt, "shape", None) or shape
  if isinstance(dt, (pd.DataFrame, pd.Series)):
  return estimate_pandas_size(dt)
  if hasattr(dt, "estimate_size"):
- return dt.estimate_size()
+ # int() for windows CI, otherwise may return numpy.int32
+ return int(dt.estimate_size()) if dt.estimate_size() is not None else None
  if hasattr(dt, "nbytes"):
  return max(sys.getsizeof(dt), dt.nbytes)
  if hasattr(dt, "shape") and len(dt.shape) == 0: