Use custom kh_int_t

WillAyd · WillAyd · commit 76bed7928ad8 · 2024-03-05T20:49:07.000-08:00
diff --git a/pandas/_libs/new_vector.cpp b/pandas/_libs/new_vector.cpp
@@ -661,7 +661,7 @@ template <typename T, bool IsMasked> class PandasHashTable {
     return;
   }
 
-  klib::KHashMap<T, size_t, PandasHashFunction<T>, PandasHashEquality<T>>
+  klib::KHashMap<T, size_t, PandasHashFunction<T>, PandasHashEquality<T>, pd_kh_int_t>
       hash_map_;
   Py_ssize_t na_position_ = -1;
 };
diff --git a/pandas/tests/libs/test_hashtable.py b/pandas/tests/libs/test_hashtable.py
@@ -311,7 +311,7 @@ def test_vector_resize(
     )
     def test_hashtable_large_sizehint(self, hashtable):
         # GH#22729 smoketest for not raising when passing a large size_hint
-        size_hint = np.iinfo(np.uint32).max + 1
+        size_hint = np.iinfo(np.uint32).max
         hashtable(size_hint=size_hint)
 
 

Original file line number	Diff line number	Diff line change
`@@ -661,7 +661,7 @@ template <typename T, bool IsMasked> class PandasHashTable {`
`661`	`661`	`return;`
`662`	`662`	`}`
`663`	`663`
`664`		`- klib::KHashMap<T, size_t, PandasHashFunction<T>, PandasHashEquality<T>>`
	`664`	`+ klib::KHashMap<T, size_t, PandasHashFunction<T>, PandasHashEquality<T>, pd_kh_int_t>`
`665`	`665`	`hash_map_;`
`666`	`666`	`Py_ssize_t na_position_ = -1;`
`667`	`667`	`};`
Original file line number	Diff line number	Diff line change
`@@ -311,7 +311,7 @@ def test_vector_resize(`
`311`	`311`	`)`
`312`	`312`	`def test_hashtable_large_sizehint(self, hashtable):`
`313`	`313`	`# GH#22729 smoketest for not raising when passing a large size_hint`
`314`		`- size_hint = np.iinfo(np.uint32).max + 1`
	`314`	`+ size_hint = np.iinfo(np.uint32).max`
`315`	`315`	`hashtable(size_hint=size_hint)`
`316`	`316`
`317`	`317`