feat(taps): SQL taps now emit schemas with maxLength when applicable (

#2651)
meltano · Sep 6, 2024 · 4ce312f · 4ce312f
1 parent ae86c53
commit 4ce312f
Show file tree

Hide file tree

Showing 5 changed files with 15 additions and 14 deletions.
diff --git a/singer_sdk/connectors/sql.py b/singer_sdk/connectors/sql.py
@@ -166,15 +166,14 @@ def float_to_jsonschema(self, column_type: sa.types.Numeric) -> dict:  # noqa: A
         return th.NumberType.type_dict  # type: ignore[no-any-return]
 
     @to_jsonschema.register
-    def string_to_jsonschema(self, column_type: sa.types.String) -> dict:  # noqa: ARG002, PLR6301
+    def string_to_jsonschema(self, column_type: sa.types.String) -> dict:  # noqa: PLR6301
         """Return a JSON Schema representation of a generic string type.
 
         Args:
             column_type (:column_type:`String`): The column type.
         """
-        # TODO: Enable support for maxLength.
-        # if sa_type.length:
-        #     return StringType(max_length=sa_type.length).type_dict  # noqa: ERA001
+        if column_type.length:
+            return th.StringType(max_length=column_type.length).type_dict  # type: ignore[no-any-return]
         return th.StringType.type_dict  # type: ignore[no-any-return]
 
     @to_jsonschema.register

diff --git a/singer_sdk/streams/sql.py b/singer_sdk/streams/sql.py
@@ -210,8 +210,6 @@ def get_records(self, context: Context | None) -> t.Iterable[dict[str, t.Any]]:
 
         with self.connector._connect() as conn:  # noqa: SLF001
             for record in conn.execute(query).mappings():
-                # TODO: Standardize record mapping type
-                # https://github.com/meltano/sdk/issues/2096
                 transformed_record = self.post_process(dict(record))
                 if transformed_record is None:
                     # Record filtered out during post_process()

diff --git a/tests/core/test_connector_sql.py b/tests/core/test_connector_sql.py
@@ -405,7 +405,6 @@ def test_fully_qualified_name_empty_error():
         pytest.param(
             sa.types.VARCHAR(length=127),
             {"type": ["string"], "maxLength": 127},
-            marks=pytest.mark.xfail,
             id="varchar-length",
         ),
         pytest.param(sa.types.TEXT(), {"type": ["string"]}, id="text"),

diff --git a/tests/samples/conftest.py b/tests/samples/conftest.py
@@ -29,14 +29,15 @@ def _sqlite_sample_db(sqlite_connector):
                     f"""
                     CREATE TABLE t{t} (
                         c1 int PRIMARY KEY NOT NULL,
-                        c2 varchar(10) NOT NULL
+                        c2 varchar(10) NOT NULL,
+                        c3 text NOT NULL
                     )
                     """
                 ),
             )
             for x in range(100):
                 conn.execute(
-                    sa.text(f"INSERT INTO t{t} VALUES ({x}, 'x={x}')"),  # noqa: S608
+                    sa.text(f"INSERT INTO t{t} VALUES ({x}, 'x={x}', 'y={x}')"),  # noqa: S608
                 )
 
 

diff --git a/tests/samples/test_tap_sqlite.py b/tests/samples/test_tap_sqlite.py
@@ -70,7 +70,7 @@ def test_sqlite_discovery(sqlite_sample_tap: SQLTap):
     sqlite_sample_tap.sync_all()
     stream = t.cast(SQLStream, sqlite_sample_tap.streams["main-t1"])
     schema = stream.schema
-    assert len(schema["properties"]) == 2
+    assert len(schema["properties"]) == 3
     assert stream.name == stream.tap_stream_id == "main-t1"
 
     md_map = MetadataMapping.from_iterable(stream.catalog_entry["metadata"])
@@ -90,13 +90,17 @@ def test_sqlite_discovery(sqlite_sample_tap: SQLTap):
 def test_sqlite_input_catalog(sqlite_sample_tap: SQLTap):
     sqlite_sample_tap.sync_all()
     stream = t.cast(SQLStream, sqlite_sample_tap.streams["main-t1"])
-    assert len(stream.schema["properties"]) == 2
-    assert len(stream.stream_maps[0].transformed_schema["properties"]) == 2
+    assert len(stream.schema["properties"]) == 3
+    assert len(stream.stream_maps[0].transformed_schema["properties"]) == 3
 
     for schema in [stream.schema, stream.stream_maps[0].transformed_schema]:
-        assert len(schema["properties"]) == 2
+        assert len(schema["properties"]) == 3
         assert schema["properties"]["c1"] == {"type": ["integer"]}
-        assert schema["properties"]["c2"] == {"type": ["string", "null"]}
+        assert schema["properties"]["c2"] == {
+            "type": ["string", "null"],
+            "maxLength": 10,
+        }
+        assert schema["properties"]["c3"] == {"type": ["string", "null"]}
         assert stream.name == stream.tap_stream_id == "main-t1"
 
     md_map = MetadataMapping.from_iterable(stream.catalog_entry["metadata"])