feat(taps,targets): Support the x-singer.decimal JSON Schema extension

docs/guides/sql-tap.md

@@ -80,3 +80,10 @@ class ConfiguredSQLToJSONSchema(SQLToJSONSchema):
 ```
 
 Then, you can use your custom type mapping in your connector as in the previous example.
+
+
+### SQL tap support for Singer Decimal string format
+
+Starting from version `0.45.0`, the Meltano Singer SDK supports the `x-singer.decimal` format for strings. You can configure the tap to use it with the `use_singer_decimal` setting. SQL Targets that support the `x-singer.decimal` format will create an appropriate numeric column in the target database.
+
+Read more about target support for `x-singer.decimal` in the [SQL target guide](./sql-target.md#sql-target-support-for-singer-decimal-string-format).

docs/guides/sql-target.md

@@ -109,3 +109,9 @@ plugins:
 }
 ```
 ````
+
+### SQL target support for Singer Decimal string format
+
+Starting from version `0.45.0`, the Meltano Singer SDK supports the `x-singer.decimal` format for strings. If the source tap is configured to use this format, the SDK will automatically convert the string to a `DECIMAL` type in the target database.
+
+Read more about target support for `x-singer.decimal` in the [SQL tap guide](./sql-tap.md#sql-tap-support-for-singer-decimal-string-format).

singer_sdk/connectors/sql.py

@@ -126,10 +126,21 @@ class SQLToJSONSchema:
     .. versionchanged:: 0.43.0
        Added the :meth:`singer_sdk.connectors.sql.SQLToJSONSchema.from_config` class
        method.
+    .. versionchanged:: 0.45.0
+       Added support for the `use_singer_decimal` option.
     """
 
+    def __init__(self, *, use_singer_decimal: bool = False) -> None:
+        """Initialize the SQL to JSON Schema converter.
+
+        Args:
+            use_singer_decimal: Whether to represent numbers as `string` with
+                the `x-singer.decimal` format instead of as `number`.
+        """
+        self.use_singer_decimal = use_singer_decimal
+
     @classmethod
-    def from_config(cls: type[SQLToJSONSchema], config: dict) -> SQLToJSONSchema:  # noqa: ARG003
+    def from_config(cls: type[SQLToJSONSchema], config: dict) -> SQLToJSONSchema:
         """Create a new instance from a configuration dictionary.
 
         Override this to instantiate this converter with values from the tap's
@@ -148,11 +159,13 @@ def from_config(cls, config):
 
         Args:
             config: The configuration dictionary.
+            use_singer_decimal: Whether to represent numbers as `string` with
+                the `x-singer.decimal` format instead of as `number`.
 
         Returns:
             A new instance of the class.
         """
-        return cls()
+        return cls(use_singer_decimal=config.get("use_singer_decimal", False))
 
     @functools.singledispatchmethod
     def to_jsonschema(self, column_type: sa.types.TypeEngine) -> dict:  # noqa: ARG002, D102, PLR6301
@@ -195,12 +208,14 @@ def integer_to_jsonschema(self, column_type: sa.types.Integer) -> dict:  # noqa:
         return th.IntegerType.type_dict  # type: ignore[no-any-return]
 
     @to_jsonschema.register
-    def float_to_jsonschema(self, column_type: sa.types.Numeric) -> dict:  # noqa: ARG002, PLR6301
+    def float_to_jsonschema(self, column_type: sa.types.Numeric) -> dict:  # noqa: ARG002
         """Return a JSON Schema representation of a generic number type.
 
         Args:
             column_type (:column_type:`Numeric`): The column type.
         """
+        if self.use_singer_decimal:
+            return th.SingerDecimalType.type_dict  # type: ignore[no-any-return]
         return th.NumberType.type_dict  # type: ignore[no-any-return]
 
     @to_jsonschema.register
@@ -278,6 +293,7 @@ def __init__(self, *, max_varchar_length: int | None = None) -> None:
             "hostname": lambda _: sa.types.VARCHAR(253),  # RFC 1035
             "ipv4": lambda _: sa.types.VARCHAR(15),
             "ipv6": lambda _: sa.types.VARCHAR(45),
+            "x-singer.decimal": self._handle_singer_decimal,
         }
 
         self._sql_datatype_mapping: dict[str, JSONtoSQLHandler] = {}
@@ -331,6 +347,17 @@ def _invoke_handler(  # noqa: PLR6301
             return handler()  # type: ignore[no-any-return]
         return handler(schema)
 
+    def _handle_singer_decimal(self, schema: dict) -> sa.types.TypeEngine:  # noqa: PLR6301
+        """Handle a x-singer.decimal format.
+
+        Args:
+            schema: The JSON Schema object.
+
+        Returns:
+            The appropriate SQLAlchemy type.
+        """
+        return sa.types.DECIMAL(schema.get("precision"), schema.get("scale"))
+
     @property
     def fallback_type(self) -> type[sa.types.TypeEngine]:
         """Return the fallback type.

singer_sdk/helpers/capabilities.py

@@ -136,6 +136,18 @@
         ),
     ),
 ).to_dict()
+SQL_TAP_USE_SINGER_DECIMAL = PropertiesList(
+    Property(
+        "use_singer_decimal",
+        BooleanType(),
+        title="Use Singer Decimal",
+        description=(
+            "Whether to use use strings with `x-singer.decimal` format for "
+            "decimals in the discovered schema. "
+            "This is useful to avoid precision loss when working with large numbers."
+        ),
+    ),
+).to_dict()
 TARGET_SCHEMA_CONFIG = PropertiesList(
     Property(
         "default_target_schema",

singer_sdk/tap_base.py

@@ -24,6 +24,7 @@
 from singer_sdk.helpers._util import dump_json, read_json_file
 from singer_sdk.helpers.capabilities import (
     BATCH_CONFIG,
+    SQL_TAP_USE_SINGER_DECIMAL,
     CapabilitiesEnum,
     PluginCapabilities,
     TapCapabilities,
@@ -675,6 +676,16 @@ def __init__(self, *args: t.Any, **kwargs: t.Any) -> None:
         self._catalog_dict: dict | None = None
         super().__init__(*args, **kwargs)
 
+    @classmethod
+    def append_builtin_config(cls: type[SQLTap], config_jsonschema: dict) -> None:
+        """Appends built-in config to `config_jsonschema` if not already set.
+
+        Args:
+            config_jsonschema: [description]
+        """
+        merge_missing_config_jsonschema(SQL_TAP_USE_SINGER_DECIMAL, config_jsonschema)
+        super().append_builtin_config(config_jsonschema)
+
     @property
     def tap_connector(self) -> SQLConnector:
         """The connector object.

singer_sdk/typing.py

@@ -448,6 +448,12 @@ class RegexType(StringType):
     string_format = "regex"
 
 
+class SingerDecimalType(StringType):
+    """Decimal type."""
+
+    string_format = "singer.decimal"
+
+
 class BooleanType(JSONTypeHelper[bool]):
     """Boolean type.
 

tests/core/test_connector_sql.py

@@ -510,6 +510,14 @@ def my_type_to_jsonschema(self, column_type) -> dict:  # noqa: ARG002
     assert m.to_jsonschema(sa.types.BOOLEAN()) == {"type": ["boolean"]}
 
 
+def test_numeric_to_singer_decimal():
+    converter = SQLToJSONSchema(use_singer_decimal=True)
+    assert converter.to_jsonschema(sa.types.NUMERIC()) == {
+        "type": ["string"],
+        "format": "singer.decimal",
+    }
+
+
 class TestJSONSchemaToSQL:  # noqa: PLR0904
     @pytest.fixture
     def json_schema_to_sql(self) -> JSONSchemaToSQL:
@@ -682,7 +690,7 @@ def test_unknown_format(self, json_schema_to_sql: JSONSchemaToSQL):
         assert isinstance(result, sa.types.VARCHAR)
 
     def test_custom_fallback(self):
-        json_schema_to_sql = JSONSchemaToSQL()
+        json_schema_to_sql = JSONSchemaToSQL(max_varchar_length=None)
         json_schema_to_sql.fallback_type = sa.types.CHAR
         jsonschema_type = {"cannot": "compute"}
         result = json_schema_to_sql.to_sql_type(jsonschema_type)
@@ -696,7 +704,7 @@ def handle_raw_string(self, schema):
 
                 return super().handle_raw_string(schema)
 
-        json_schema_to_sql = CustomJSONSchemaToSQL()
+        json_schema_to_sql = CustomJSONSchemaToSQL(max_varchar_length=None)
 
         vanilla = {"type": ["string"]}
         result = json_schema_to_sql.to_sql_type(vanilla)
@@ -717,6 +725,52 @@ def handle_raw_string(self, schema):
         result = json_schema_to_sql.to_sql_type(image_type)
         assert isinstance(result, sa.types.LargeBinary)
 
+    def test_singer_decimal(self):
+        json_schema_to_sql = JSONSchemaToSQL()
+        jsonschema_type = {"type": ["string"], "format": "x-singer.decimal"}
+        result = json_schema_to_sql.to_sql_type(jsonschema_type)
+        assert isinstance(result, sa.types.DECIMAL)
+
+    def test_singer_decimal_with_precision_scale(self):
+        json_schema_to_sql = JSONSchemaToSQL()
+        precision, scale = 12, 3
+        jsonschema_type = {
+            "type": ["string"],
+            "format": "x-singer.decimal",
+            "precision": precision,
+            "scale": scale,
+        }
+        result = json_schema_to_sql.to_sql_type(jsonschema_type)
+        assert isinstance(result, sa.types.DECIMAL)
+        assert result.precision == precision
+        assert result.scale == scale
+
+    def test_handle_singer_decimal_missing_precision(self):
+        json_schema_to_sql = JSONSchemaToSQL(max_varchar_length=None)
+        schema = {
+            "type": ["string"],
+            "format": "x-singer.decimal",
+            # 'precision' is missing
+            "scale": 2,
+        }
+        result = json_schema_to_sql.to_sql_type(schema)
+        assert isinstance(result, sa.types.DECIMAL)
+        assert result.precision is None
+        assert result.scale == 2
+
+    def test_handle_singer_decimal_missing_scale(self):
+        json_schema_to_sql = JSONSchemaToSQL(max_varchar_length=None)
+        schema = {
+            "type": ["string"],
+            "format": "x-singer.decimal",
+            "precision": 10,
+            # 'scale' is missing
+        }
+        result = json_schema_to_sql.to_sql_type(schema)
+        assert isinstance(result, sa.types.DECIMAL)
+        assert result.precision == 10
+        assert result.scale is None
+
     def test_annotation_sql_datatype(self):
         json_schema_to_sql = JSONSchemaToSQL()
         json_schema_to_sql.register_sql_datatype_handler("json", sa.types.JSON)