remove numpy_helper and some unneeded util functions (pandas-dev#22469)

Author: jbrockmendel

pandas/_libs/index.pyx

@@ -104,10 +104,7 @@ cdef class IndexEngine:
         loc = self.get_loc(key)
         value = convert_scalar(arr, value)
 
-        if PySlice_Check(loc) or util.is_array(loc):
-            arr[loc] = value
-        else:
-            util.set_value_at(arr, loc, value)
+        arr[loc] = value
 
     cpdef get_loc(self, object val):
         if is_definitely_invalid_key(val):

pandas/_libs/lib.pyx

@@ -492,9 +492,7 @@ def astype_intsafe(ndarray[object] arr, new_dtype):
         if is_datelike and checknull(v):
             result[i] = NPY_NAT
         else:
-            # we can use the unsafe version because we know `result` is mutable
-            # since it was created from `np.empty`
-            util.set_value_at_unsafe(result, i, v)
+            result[i] = v
 
     return result
 
@@ -505,9 +503,7 @@ cpdef ndarray[object] astype_unicode(ndarray arr):
         ndarray[object] result = np.empty(n, dtype=object)
 
     for i in range(n):
-        # we can use the unsafe version because we know `result` is mutable
-        # since it was created from `np.empty`
-        util.set_value_at_unsafe(result, i, unicode(arr[i]))
+        result[i] = unicode(arr[i])
 
     return result
 
@@ -518,9 +514,7 @@ cpdef ndarray[object] astype_str(ndarray arr):
         ndarray[object] result = np.empty(n, dtype=object)
 
     for i in range(n):
-        # we can use the unsafe version because we know `result` is mutable
-        # since it was created from `np.empty`
-        util.set_value_at_unsafe(result, i, str(arr[i]))
+        result[i] = str(arr[i])
 
     return result
 

pandas/_libs/reduction.pyx

@@ -280,8 +280,7 @@ cdef class SeriesBinGrouper:
                     result = _get_result_array(res,
                                                self.ngroups,
                                                len(self.dummy_arr))
-
-                util.assign_value_1d(result, i, res)
+                result[i] = res
 
                 islider.advance(group_size)
                 vslider.advance(group_size)
@@ -406,7 +405,7 @@ cdef class SeriesGrouper:
                                                    self.ngroups,
                                                    len(self.dummy_arr))
 
-                    util.assign_value_1d(result, lab, res)
+                    result[lab] = res
                     counts[lab] = group_size
                     islider.advance(group_size)
                     vslider.advance(group_size)

pandas/_libs/src/numpy_helper.h

@@ -30,11 +30,6 @@ cdef extern from *:
     const char *get_c_string(object) except NULL
 
 
-cdef extern from "src/numpy_helper.h":
-    int assign_value_1d(ndarray, Py_ssize_t, object) except -1
-    object get_value_1d(ndarray, Py_ssize_t)
-
-
 cdef extern from "src/headers/stdint.h":
     enum: UINT8_MAX
     enum: UINT16_MAX
@@ -116,26 +111,4 @@ cdef inline object get_value_at(ndarray arr, object loc):
         Py_ssize_t i
 
     i = validate_indexer(arr, loc)
-    return get_value_1d(arr, i)
-
-
-cdef inline set_value_at_unsafe(ndarray arr, object loc, object value):
-    """Sets a value into the array without checking the writeable flag.
-
-    This should be used when setting values in a loop, check the writeable
-    flag above the loop and then eschew the check on each iteration.
-    """
-    cdef:
-        Py_ssize_t i
-
-    i = validate_indexer(arr, loc)
-    assign_value_1d(arr, i, value)
-
-
-cdef inline set_value_at(ndarray arr, object loc, object value):
-    """Sets a value into the array after checking that the array is mutable.
-    """
-    if not cnp.PyArray_ISWRITEABLE(arr):
-        raise ValueError('assignment destination is read-only')
-
-    set_value_at_unsafe(arr, loc, value)
+    return arr[i]

pandas/_libs/util.pxd

@@ -446,7 +446,10 @@ def _get_val_at(self, loc):
         if sp_loc == -1:
             return self.fill_value
         else:
-            return libindex.get_value_at(self, sp_loc)
+            # libindex.get_value_at will end up calling __getitem__,
+            # so to avoid recursing we need to unwrap `self` so the
+            # ndarray.__getitem__ implementation is called.
+            return libindex.get_value_at(np.asarray(self), sp_loc)
 
     @Appender(_index_shared_docs['take'] % _sparray_doc_kwargs)
     def take(self, indices, axis=0, allow_fill=True,

pandas/core/sparse/array.py

@@ -491,8 +491,7 @@ def srcpath(name=None, suffix='.pyx', subdir='src'):
 ts_include = ['pandas/_libs/tslibs/src']
 
 
-lib_depends = ['pandas/_libs/src/numpy_helper.h',
-               'pandas/_libs/src/parse_helper.h',
+lib_depends = ['pandas/_libs/src/parse_helper.h',
                'pandas/_libs/src/compat_helper.h']
 
 np_datetime_headers = [