Internal change: refactoring of Arrow schema handling.

Author: anthony-tuininga

src/oracledb/arrow_impl.pxd

@@ -86,8 +86,8 @@ cdef extern from "nanoarrow.h":
         NANOARROW_TIME_UNIT_MICRO
         NANOARROW_TIME_UNIT_NANO
 
+cdef class ArrowSchemaImpl:
 
-cdef class ArrowArrayImpl:
     cdef:
         int32_t precision
         int32_t scale
@@ -96,17 +96,28 @@ cdef class ArrowArrayImpl:
         ArrowType arrow_type
         ArrowTimeUnit time_unit
         int time_factor
-        ArrowArray *arrow_array
         ArrowSchema *arrow_schema
         ArrowType child_arrow_type
         int child_element_size
 
+    cdef bint _is_sparse_vector(self) except*
+    cdef int _set_child_arrow_type(self, ArrowType child_arrow_type) except -1
+    cdef int _set_time_unit(self, ArrowTimeUnit time_unit) except -1
+    cdef int populate_from_schema(self, ArrowSchema* schema) except -1
+    cdef int populate_from_metadata(self, ArrowType arrow_type, str name,
+                                    int8_t precision, int8_t scale,
+                                    ArrowTimeUnit time_unit,
+                                    ArrowType child_arrow_type) except -1
+
+
+cdef class ArrowArrayImpl:
+    cdef:
+        ArrowArray *arrow_array
+        ArrowSchemaImpl schema_impl
+
     cdef int _get_is_null(self, int64_t index, bint* is_null) except -1
     cdef int _get_list_info(self, int64_t index, ArrowArray* arrow_array,
                             int64_t* offset, int64_t* num_elements) except -1
-    cdef bint _is_sparse_vector(self) except *
-    cdef int _set_child_arrow_type(self, ArrowType child_arrow_type) except -1
-    cdef int _set_time_unit(self, ArrowTimeUnit time_unit) except -1
     cdef int append_bytes(self, void* ptr, int64_t num_bytes) except -1
     cdef int append_decimal(self, void* ptr, int64_t num_bytes) except -1
     cdef int append_double(self, double value) except -1
@@ -135,10 +146,7 @@ cdef class ArrowArrayImpl:
     cdef object get_vector(self, int64_t index, bint* is_null)
     cdef int populate_from_array(self, ArrowSchema* schema,
                                  ArrowArray* array) except -1
-    cdef int populate_from_metadata(self, ArrowType arrow_type, str name,
-                                    int8_t precision, int8_t scale,
-                                    ArrowTimeUnit time_unit,
-                                    ArrowType child_arrow_type) except -1
+    cdef int populate_from_schema(self, ArrowSchemaImpl schema_impl) except -1
 
 
 cdef class DataFrameImpl:

src/oracledb/arrow_impl.pyx

@@ -52,5 +52,6 @@ else:
     uint32_template = array.array("L")
 
 include "impl/arrow/utils.pyx"
+include "impl/arrow/schema.pyx"
 include "impl/arrow/array.pyx"
 include "impl/arrow/dataframe.pyx"

src/oracledb/base_impl.pxd

@@ -43,6 +43,7 @@ from .arrow_impl cimport (
     ArrowTimeUnit,
     ArrowType,
     ArrowArrayImpl,
+    ArrowSchemaImpl
 )
 
 cdef enum:
@@ -444,14 +445,14 @@ cdef class OracleMetadata:
         readonly uint32_t vector_dimensions
         readonly uint8_t vector_format
         readonly uint8_t vector_flags
-        ArrowType _arrow_type
+        ArrowSchemaImpl _schema_impl
         uint8_t _py_type_num
 
     cdef int _finalize_init(self) except -1
-    cdef int _set_arrow_type(self) except -1
+    cdef int _set_arrow_schema(self) except -1
     cdef OracleMetadata copy(self)
     @staticmethod
-    cdef OracleMetadata from_arrow_array(ArrowArrayImpl array)
+    cdef OracleMetadata from_arrow_schema(ArrowSchemaImpl schema_impl)
     @staticmethod
     cdef OracleMetadata from_type(object typ)
     @staticmethod
@@ -750,8 +751,6 @@ cdef class BaseVarImpl:
     cdef object _get_scalar_value(self, uint32_t pos)
     cdef int _on_reset_bind(self, uint32_t num_rows) except -1
     cdef int _resize(self, uint32_t new_size) except -1
-    cdef int _set_metadata_from_arrow_array(self,
-                                            ArrowArrayImpl array) except -1
     cdef int _set_metadata_from_type(self, object typ) except -1
     cdef int _set_metadata_from_value(self, object value,
                                       bint is_plsql) except -1
@@ -986,12 +985,12 @@ cdef struct OracleData:
 
 cdef object convert_arrow_to_oracle_data(OracleMetadata metadata,
                                          OracleData* data,
-                                         ArrowArrayImpl arrow_array,
+                                         ArrowArrayImpl array_impl,
                                          ssize_t array_index)
 cdef int convert_oracle_data_to_arrow(OracleMetadata from_metadata,
                                       OracleMetadata to_metadatda,
                                       OracleData* data,
-                                      ArrowArrayImpl arrow_array) except -1
+                                      ArrowArrayImpl array_impl) except -1
 cdef object convert_oracle_data_to_python(OracleMetadata from_metadata,
                                           OracleMetadata to_metadatda,
                                           OracleData* data,
@@ -1000,7 +999,7 @@ cdef object convert_oracle_data_to_python(OracleMetadata from_metadata,
 cdef object convert_python_to_oracle_data(OracleMetadata metadata,
                                           OracleData* data,
                                           object value)
-cdef int convert_vector_to_arrow(ArrowArrayImpl arrow_array,
+cdef int convert_vector_to_arrow(ArrowArrayImpl array_impl,
                                  object vector) except -1
 cdef cydatetime.datetime convert_date_to_python(OracleDataBuffer *buffer)
 cdef uint16_t decode_uint16be(const char_type *buf)