fix: Ensure that Snowflake accounts for number columns that overspeci…

…fy precision (#3306)

* fix: Ensure that Snowflake properly accounts for INT32/INT64 columns with default precision 38 type

Signed-off-by: Danny Chiao <danny@tecton.ai>

* fix: Snowflake infer snowflake OBJECT return type for number inference

Signed-off-by: miles.adkins <miles.adkins@snowflake.com>

Signed-off-by: Danny Chiao <danny@tecton.ai>
Signed-off-by: miles.adkins <miles.adkins@snowflake.com>
Co-authored-by: miles.adkins <miles.adkins@snowflake.com>

sdk/python/feast/infra/offline_stores/snowflake_source.py

@@ -263,6 +263,16 @@ def get_table_column_names_and_types(
                                 result.dtypes[column].name
                             ]
                         else:
+                            if len(result) > 0:
+                                max_value = result.iloc[0][0]
+                                if max_value is not None and len(str(max_value)) <= 9:
+                                    row["snowflake_type"] = "NUMBER32"
+                                    continue
+                                elif (
+                                    max_value is not None and len(str(max_value)) <= 18
+                                ):
+                                    row["snowflake_type"] = "NUMBER64"
+                                    continue
                             raise NotImplementedError(
                                 "NaNs or Numbers larger than INT64 are not supported"
                             )