strings_udf lowering shall not import cudf, and functions shall be at…

…trs of string_view

python/cudf/cudf/core/udf/masked_typing.py

@@ -54,12 +54,6 @@ def __init__(self, value):
         if isinstance(value, (types.PyObject, StringView)):
             self.value_type = string_view
 
-            from strings_udf import lowering
-            from strings_udf._lib.cudf_jit_udf import get_character_flags_table_ptr
-
-            character_flags_table_ptr = get_character_flags_table_ptr()
-            lowering.character_flags_table_ptr = character_flags_table_ptr
-
         elif isinstance(value, SUPPORTED_NUMBA_TYPES):
             self.value_type = value
         else:

python/strings_udf/strings_udf/_lib/cudf_jit_udf.pyx

@@ -25,11 +25,8 @@ from strings_udf._lib.cpp.strings_udf cimport (
     from_dstring_array as cpp_from_dstring_array,
     to_string_view_array as cpp_to_string_view_array,
     udf_module as cpp_udf_module,
-    get_character_flags_table as cpp_get_character_flags_table
 )
 
-from libc.stdint cimport uintptr_t, uint8_t
-
 import numpy as np
 
 def process_udf( udf, name, cols ):
@@ -86,7 +83,3 @@ def from_dstring_array(DeviceBuffer d_buffer):
     c_result = move(cpp_from_dstring_array(data, size))
 
     return Column.from_unique_ptr(move(c_result))
-
-def get_character_flags_table_ptr():
-    cdef const uint8_t* tbl_ptr = cpp_get_character_flags_table()
-    return np.int64(<uintptr_t>tbl_ptr)

python/strings_udf/strings_udf/_lib/tables.pyx

@@ -0,0 +1,13 @@
+# Copyright (c) 2022, NVIDIA CORPORATION.
+#
+# distutils: language = c++
+# cython: c_string_type=unicode, c_string_encoding=utf8
+
+from strings_udf._lib.cpp.strings_udf cimport get_character_flags_table as cpp_get_character_flags_table
+from libc.stdint cimport uintptr_t, uint8_t
+import numpy as np
+
+
+def get_character_flags_table_ptr():
+    cdef const uint8_t* tbl_ptr = cpp_get_character_flags_table()
+    return np.int64(<uintptr_t>tbl_ptr)