COMPAT/PERF move uniques to method scope

Author: mattip

pandas/_libs/hashtable.pyx

@@ -46,12 +46,10 @@ include "hashtable_func_helper.pxi"
 
 cdef class Factorizer:
     cdef public PyObjectHashTable table
-    cdef public ObjectVector uniques
     cdef public Py_ssize_t count
 
     def __init__(self, size_hint):
         self.table = PyObjectHashTable(size_hint)
-        self.uniques = ObjectVector()
         self.count = 0
 
     def get_count(self):
@@ -64,19 +62,22 @@ cdef class Factorizer:
         >>> factorize(np.array([1,2,np.nan], dtype='O'), na_sentinel=20)
         array([ 0,  1, 20])
         """
-        labels = self.table.get_labels(values, self.uniques,
+        uniques = ObjectVector()
+        labels = self.table.get_labels(values, uniques,
                                        self.count, na_sentinel, check_null)
         mask = (labels == na_sentinel)
+        if len(labels) == 0:
+            return labels
         # sort on
         if sort:
             if labels.dtype != np.intp:
                 labels = labels.astype(np.intp)
-            sorter = self.uniques.to_array().argsort()
+            sorter = uniques.to_array().argsort()
             reverse_indexer = np.empty(len(sorter), dtype=np.intp)
             reverse_indexer.put(sorter, np.arange(len(sorter)))
             labels = reverse_indexer.take(labels, mode='clip')
             labels[mask] = na_sentinel
-        self.count = len(self.uniques)
+        self.count = len(uniques)
         return labels
 
     def unique(self, ndarray[object] values):
@@ -86,35 +87,36 @@ cdef class Factorizer:
 
 cdef class Int64Factorizer:
     cdef public Int64HashTable table
-    cdef public Int64Vector uniques
     cdef public Py_ssize_t count
 
     def __init__(self, size_hint):
         self.table = Int64HashTable(size_hint)
-        self.uniques = Int64Vector()
         self.count = 0
 
     def get_count(self):
         return self.count
 
     def factorize(self, int64_t[:] values, sort=False,
                   na_sentinel=-1, check_null=True):
+        uniques = Int64Vector()
         labels = self.table.get_labels(values, self.uniques,
                                        self.count, na_sentinel,
                                        check_null)
 
         # sort on
+        if len(labels) == 0:
+            return labels
         if sort:
             if labels.dtype != np.intp:
                 labels = labels.astype(np.intp)
 
-            sorter = self.uniques.to_array().argsort()
+            sorter = uniques.to_array().argsort()
             reverse_indexer = np.empty(len(sorter), dtype=np.intp)
             reverse_indexer.put(sorter, np.arange(len(sorter)))
 
             labels = reverse_indexer.take(labels)
 
-        self.count = len(self.uniques)
+        self.count = len(uniques)
         return labels
 
 

pandas/tests/test_algos.py

@@ -1098,7 +1098,6 @@ def _test_vector_resize(htable, uniques, dtype, nvals, safely_resizes):
 
         for (tbl, vect, dtype, safely_resizes) in test_cases:
             # resizing to empty is a special case
-            print dtype, safely_resizes
             _test_vector_resize(tbl(), vect(), dtype, 0, safely_resizes)
             _test_vector_resize(tbl(), vect(), dtype, 10, safely_resizes)