|
1 | 1 | import sys |
2 | 2 | from binascii import unhexlify |
| 3 | +from datetime import datetime |
3 | 4 |
|
| 5 | +import numpy as np |
4 | 6 | import pandas |
5 | 7 | from _snowflake import vectorized |
6 | 8 |
|
|
13 | 15 | ) |
14 | 16 | from feast.value_type import ValueType |
15 | 17 |
|
| 18 | + |
| 19 | +def _cast_array_list_elements_to(array: np.ndarray, mapper): |
| 20 | + for row in array: |
| 21 | + for i, elem in enumerate(row): |
| 22 | + row[i] = mapper(elem) |
| 23 | + |
| 24 | + |
16 | 25 | """ |
17 | 26 | CREATE OR REPLACE FUNCTION feast_snowflake_binary_to_bytes_proto(df BINARY) |
18 | 27 | RETURNS BINARY |
@@ -59,6 +68,185 @@ def feast_snowflake_varchar_to_string_proto(df): |
59 | 68 | return df |
60 | 69 |
|
61 | 70 |
|
| 71 | +""" |
| 72 | +CREATE OR REPLACE FUNCTION feast_snowflake_array_bytes_to_list_bytes_proto(df ARRAY) |
| 73 | + RETURNS BINARY |
| 74 | + LANGUAGE PYTHON |
| 75 | + RUNTIME_VERSION = '3.8' |
| 76 | + PACKAGES = ('protobuf', 'pandas') |
| 77 | + HANDLER = 'feast.infra.utils.snowflake.snowpark.snowflake_udfs.feast_snowflake_array_bytes_to_list_bytes_proto' |
| 78 | + IMPORTS = ('@feast_stage/feast.zip'); |
| 79 | +""" |
| 80 | +# ValueType.STRING_LIST = 12 |
| 81 | +@vectorized(input=pandas.DataFrame) |
| 82 | +def feast_snowflake_array_bytes_to_list_bytes_proto(df): |
| 83 | + sys._xoptions["snowflake_partner_attribution"].append("feast") |
| 84 | + |
| 85 | + df = list( |
| 86 | + map( |
| 87 | + ValueProto.SerializeToString, |
| 88 | + python_values_to_proto_values(df[0].to_numpy(), ValueType.BYTES_LIST), |
| 89 | + ) |
| 90 | + ) |
| 91 | + return df |
| 92 | + |
| 93 | + |
| 94 | +""" |
| 95 | +CREATE OR REPLACE FUNCTION feast_snowflake_array_varchar_to_list_string_proto(df ARRAY) |
| 96 | + RETURNS BINARY |
| 97 | + LANGUAGE PYTHON |
| 98 | + RUNTIME_VERSION = '3.8' |
| 99 | + PACKAGES = ('protobuf', 'pandas') |
| 100 | + HANDLER = 'feast.infra.utils.snowflake.snowpark.snowflake_udfs.feast_snowflake_array_varchar_to_list_string_proto' |
| 101 | + IMPORTS = ('@feast_stage/feast.zip'); |
| 102 | +""" |
| 103 | + |
| 104 | + |
| 105 | +@vectorized(input=pandas.DataFrame) |
| 106 | +def feast_snowflake_array_varchar_to_list_string_proto(df): |
| 107 | + sys._xoptions["snowflake_partner_attribution"].append("feast") |
| 108 | + |
| 109 | + df = list( |
| 110 | + map( |
| 111 | + ValueProto.SerializeToString, |
| 112 | + python_values_to_proto_values(df[0].to_numpy(), ValueType.STRING_LIST), |
| 113 | + ) |
| 114 | + ) |
| 115 | + return df |
| 116 | + |
| 117 | + |
| 118 | +""" |
| 119 | +CREATE OR REPLACE FUNCTION feast_snowflake_array_number_to_list_int32_proto(df ARRAY) |
| 120 | + RETURNS BINARY |
| 121 | + LANGUAGE PYTHON |
| 122 | + RUNTIME_VERSION = '3.8' |
| 123 | + PACKAGES = ('protobuf', 'pandas') |
| 124 | + HANDLER = 'feast.infra.utils.snowflake.snowpark.snowflake_udfs.feast_snowflake_array_number_to_list_int32_proto' |
| 125 | + IMPORTS = ('@feast_stage/feast.zip'); |
| 126 | +""" |
| 127 | + |
| 128 | + |
| 129 | +@vectorized(input=pandas.DataFrame) |
| 130 | +def feast_snowflake_array_number_to_list_int32_proto(df): |
| 131 | + sys._xoptions["snowflake_partner_attribution"].append("feast") |
| 132 | + |
| 133 | + df = list( |
| 134 | + map( |
| 135 | + ValueProto.SerializeToString, |
| 136 | + python_values_to_proto_values(df[0].to_numpy(), ValueType.INT32_LIST), |
| 137 | + ) |
| 138 | + ) |
| 139 | + return df |
| 140 | + |
| 141 | + |
| 142 | +""" |
| 143 | +CREATE OR REPLACE FUNCTION feast_snowflake_array_number_to_list_int64_proto(df ARRAY) |
| 144 | + RETURNS BINARY |
| 145 | + LANGUAGE PYTHON |
| 146 | + RUNTIME_VERSION = '3.8' |
| 147 | + PACKAGES = ('protobuf', 'pandas') |
| 148 | + HANDLER = 'feast.infra.utils.snowflake.snowpark.snowflake_udfs.feast_snowflake_array_number_to_list_int64_proto' |
| 149 | + IMPORTS = ('@feast_stage/feast.zip'); |
| 150 | +""" |
| 151 | + |
| 152 | + |
| 153 | +@vectorized(input=pandas.DataFrame) |
| 154 | +def feast_snowflake_array_number_to_list_int64_proto(df): |
| 155 | + sys._xoptions["snowflake_partner_attribution"].append("feast") |
| 156 | + |
| 157 | + df = list( |
| 158 | + map( |
| 159 | + ValueProto.SerializeToString, |
| 160 | + python_values_to_proto_values(df[0].to_numpy(), ValueType.INT64_LIST), |
| 161 | + ) |
| 162 | + ) |
| 163 | + return df |
| 164 | + |
| 165 | + |
| 166 | +""" |
| 167 | +CREATE OR REPLACE FUNCTION feast_snowflake_array_float_to_list_double_proto(df ARRAY) |
| 168 | + RETURNS BINARY |
| 169 | + LANGUAGE PYTHON |
| 170 | + RUNTIME_VERSION = '3.8' |
| 171 | + PACKAGES = ('protobuf', 'pandas') |
| 172 | + HANDLER = 'feast.infra.utils.snowflake.snowpark.snowflake_udfs.feast_snowflake_array_float_to_list_double_proto' |
| 173 | + IMPORTS = ('@feast_stage/feast.zip'); |
| 174 | +""" |
| 175 | + |
| 176 | + |
| 177 | +@vectorized(input=pandas.DataFrame) |
| 178 | +def feast_snowflake_array_float_to_list_double_proto(df): |
| 179 | + sys._xoptions["snowflake_partner_attribution"].append("feast") |
| 180 | + |
| 181 | + numpy_arrays = df[0].to_numpy() |
| 182 | + # Sometimes floats come in as ints so we need to convert back to float |
| 183 | + _cast_array_list_elements_to(numpy_arrays, float) |
| 184 | + |
| 185 | + df = list( |
| 186 | + map( |
| 187 | + ValueProto.SerializeToString, |
| 188 | + python_values_to_proto_values(numpy_arrays, ValueType.DOUBLE_LIST), |
| 189 | + ) |
| 190 | + ) |
| 191 | + return df |
| 192 | + |
| 193 | + |
| 194 | +""" |
| 195 | +CREATE OR REPLACE FUNCTION feast_snowflake_array_boolean_to_list_bool_proto(df ARRAY) |
| 196 | + RETURNS BINARY |
| 197 | + LANGUAGE PYTHON |
| 198 | + RUNTIME_VERSION = '3.8' |
| 199 | + PACKAGES = ('protobuf', 'pandas') |
| 200 | + HANDLER = 'feast.infra.utils.snowflake.snowpark.snowflake_udfs.feast_snowflake_array_boolean_to_list_bool_proto' |
| 201 | + IMPORTS = ('@feast_stage/feast.zip'); |
| 202 | +""" |
| 203 | + |
| 204 | + |
| 205 | +@vectorized(input=pandas.DataFrame) |
| 206 | +def feast_snowflake_array_boolean_to_list_bool_proto(df): |
| 207 | + sys._xoptions["snowflake_partner_attribution"].append("feast") |
| 208 | + |
| 209 | + df = list( |
| 210 | + map( |
| 211 | + ValueProto.SerializeToString, |
| 212 | + python_values_to_proto_values(df[0].to_numpy(), ValueType.BOOL_LIST), |
| 213 | + ) |
| 214 | + ) |
| 215 | + return df |
| 216 | + |
| 217 | + |
| 218 | +""" |
| 219 | +CREATE OR REPLACE FUNCTION feast_snowflake_array_timestamp_to_list_unix_timestamp_proto(df ARRAY) |
| 220 | + RETURNS BINARY |
| 221 | + LANGUAGE PYTHON |
| 222 | + RUNTIME_VERSION = '3.8' |
| 223 | + PACKAGES = ('protobuf', 'pandas') |
| 224 | + HANDLER = 'feast.infra.utils.snowflake.snowpark.snowflake_udfs.feast_snowflake_array_timestamp_to_list_unix_timestamp_proto' |
| 225 | + IMPORTS = ('@feast_stage/feast.zip'); |
| 226 | +""" |
| 227 | + |
| 228 | + |
| 229 | +@vectorized(input=pandas.DataFrame) |
| 230 | +def feast_snowflake_array_timestamp_to_list_unix_timestamp_proto(df): |
| 231 | + sys._xoptions["snowflake_partner_attribution"].append("feast") |
| 232 | + |
| 233 | + numpy_arrays = df[0].to_numpy() |
| 234 | + # Timestamps are coming in as strings so we should convert to timestamps |
| 235 | + _cast_array_list_elements_to( |
| 236 | + numpy_arrays, lambda x: datetime.strptime(x, "%Y-%m-%d %H:%M:%S.%f") |
| 237 | + ) |
| 238 | + |
| 239 | + df = list( |
| 240 | + map( |
| 241 | + ValueProto.SerializeToString, |
| 242 | + python_values_to_proto_values( |
| 243 | + df[0].to_numpy(), ValueType.UNIX_TIMESTAMP_LIST |
| 244 | + ), |
| 245 | + ) |
| 246 | + ) |
| 247 | + return df |
| 248 | + |
| 249 | + |
62 | 250 | """ |
63 | 251 | CREATE OR REPLACE FUNCTION feast_snowflake_number_to_int32_proto(df NUMBER) |
64 | 252 | RETURNS BINARY |
|
0 commit comments