build(typing): migrate from pyrefly to ty for type checking

.github/workflows/python_lint.yml

@@ -69,8 +69,8 @@ jobs:
       run: uv sync --group dev
 
     # Job-specific step(s):
-    - name: Run Pyrefly Check
-      run: uv run pyrefly check
+    - name: Run ty Check
+      run: uv run ty check
 
   public-api-import-guard:
     name: Public API Import Guard

airbyte/_connector_base.py

@@ -283,10 +283,10 @@
             content = json.dumps(self.config_spec, indent=2)
 
         if output_file:
-            output_file.write_text(content)  # pyrefly: ignore[unbound-name]
+            output_file.write_text(content)
             return
 
-        syntax_highlighted = Syntax(content, format)  # pyrefly: ignore[unbound-name]
+        syntax_highlighted = Syntax(content, format)
         rich.print(syntax_highlighted, file=sys.stderr if stderr else None)
 
     @property
@@ -405,32 +405,32 @@
             AirbyteConnectorFailedError: If a TRACE message of type ERROR is emitted.
         """
         if message.type == Type.LOG:
-            self._print_info_message(message.log.message)  # pyrefly: ignore[missing-attribute]
+            self._print_info_message(message.log.message)  # ty: ignore[unresolved-attribute]
             return
 
         if (
             message.type == Type.TRACE
-            and message.trace.type == TraceType.ERROR  # pyrefly: ignore[missing-attribute]
+            and message.trace.type == TraceType.ERROR  # ty: ignore[unresolved-attribute]
         ):
             self._print_error_message(
-                message.trace.error.message  # pyrefly: ignore[missing-attribute]
+                message.trace.error.message  # ty: ignore[unresolved-attribute]
             )
             if raise_on_error:
                 raise exc.AirbyteConnectorFailedError(
                     connector_name=self.name,
-                    message=message.trace.error.message,  # pyrefly: ignore[missing-attribute]
+                    message=message.trace.error.message,  # ty: ignore[unresolved-attribute]
                     log_text=self._last_log_messages,
                 )
             return
 
         if (
             message.type == Type.CONTROL
-            and message.control.type  # pyrefly: ignore[missing-attribute]
+            and message.control.type  # ty: ignore[unresolved-attribute]
             == OrchestratorType.CONNECTOR_CONFIG
             and self.config_change_callback is not None
         ):
             self.config_change_callback(
-                message.control.connectorConfig.config  # pyrefly: ignore[missing-attribute]
+                message.control.connectorConfig.config  # ty: ignore[unresolved-attribute]
             )
             return
 

airbyte/_message_iterators.py

@@ -178,9 +178,7 @@ def generator() -> Generator[AirbyteMessage, None, None]:
                 if current_file_buffer is None:
                     try:
                         current_file = next(file_iterator)
-                        current_file_buffer = file_opener(
-                            current_file  # pyrefly: ignore[bad-argument-type]
-                        )
+                        current_file_buffer = file_opener(current_file)
                     except StopIteration:
                         # No more files to read; Exit the loop
                         break
@@ -194,7 +192,7 @@ def generator() -> Generator[AirbyteMessage, None, None]:
 
                 try:
                     # Let Pydantic handle the JSON decoding from the raw string
-                    yield (  # pyrefly: ignore[invalid-yield]
+                    yield (  # ty: ignore[invalid-yield]
                         AirbyteMessage.model_validate_json(next_line),
                         current_file,
                     )

airbyte/_processors/sql/postgres.py

@@ -71,5 +71,5 @@ class PostgresSqlProcessor(SqlProcessorBase):
     file_writer_class = JsonlWriter
     sql_config: PostgresConfig
 
-    normalizer = PostgresNormalizer  # pyrefly: ignore[bad-override]
+    normalizer = PostgresNormalizer
     """A Postgres-specific name normalizer for table and column name normalization."""

airbyte/_processors/sql/snowflake.py

@@ -204,9 +204,7 @@ class SnowflakeSqlProcessor(SqlProcessorBase):
     """A Snowflake implementation of the cache."""
 
     file_writer_class = JsonlWriter
-    type_converter_class: type[SnowflakeTypeConverter] = (  # pyrefly: ignore[bad-override]
-        SnowflakeTypeConverter
-    )
+    type_converter_class: type[SnowflakeTypeConverter] = SnowflakeTypeConverter
     supports_merge_insert = True
     sql_config: SnowflakeConfig
 

airbyte/_util/api_util.py

@@ -1318,9 +1318,9 @@ def get_destination(
         }
 
         if destination_type in destination_mapping and raw_configuration is not None:
-            response.destination_response.configuration = destination_mapping[
-                destination_type  # pyrefly: ignore[index-error]
-            ](**raw_configuration)
+            response.destination_response.configuration = destination_mapping[destination_type](
+                **raw_configuration
+            )
         return response.destination_response
 
     raise AirbyteMissingResourceError(

airbyte/_util/meta.py

@@ -135,7 +135,7 @@ def get_notebook_name() -> str | None:
 @lru_cache
 def get_vscode_notebook_name() -> str | None:
     with suppress(Exception):
-        import IPython  # noqa: PLC0415  # pyrefly: ignore[missing-import]
+        import IPython  # noqa: PLC0415
 
         return Path(
             IPython.extract_module_locals()[1]["__vsc_ipynb_file__"],

airbyte/_util/name_normalizers.py

@@ -50,7 +50,7 @@ class LowerCaseNormalizer(NameNormalizerBase):
 
     @staticmethod
     @functools.cache
-    def normalize(name: str) -> str:  # pyrefly: ignore[bad-override]  # pyrefly decorator issue
+    def normalize(name: str) -> str:
         """Return the normalized name.
 
         - All non-alphanumeric characters are replaced with underscores.

airbyte/caches/_utils/_cache_to_dest.py

@@ -109,7 +109,7 @@ def snowflake_cache_to_destination_configuration(
         role=cache.role,
         username=cache.username,
         credentials=UsernameAndPassword(
-            password=cache.password,  # pyrefly: ignore[bad-argument-type]
+            password=cache.password,  # ty: ignore[invalid-argument-type]
         ),
     )
 

airbyte/caches/base.py

@@ -470,7 +470,7 @@ def _write_airbyte_message_stream(
             state_writer=state_writer,
         )
         cache_processor.process_airbyte_messages(
-            messages=stdin,  # pyrefly: ignore[bad-argument-type]
+            messages=stdin,  # ty: ignore[invalid-argument-type]
             write_strategy=write_strategy,
             progress_tracker=progress_tracker,
         )

airbyte/caches/motherduck.py

@@ -42,7 +42,7 @@ class MotherDuckConfig(DuckDBConfig):
 
     database: str = Field()
     api_key: SecretString = Field()
-    db_path: str = Field(default="md:")  # pyrefly: ignore[bad-override]
+    db_path: str = Field(default="md:")
     _paired_destination_name: str = "destination-motherduck"
 
     @overrides

airbyte/cli/pyab.py

@@ -360,10 +360,10 @@ def _resolve_destination_job(
 
 
 cli = App(
-    name="pyab",  # pyrefly: ignore[unexpected-keyword]
-    help="PyAirbyte CLI.",  # pyrefly: ignore[unexpected-keyword]
-    help_format="plaintext",  # pyrefly: ignore[unexpected-keyword]
-    version_flags=[],  # pyrefly: ignore[unexpected-keyword]
+    name="pyab",
+    help="PyAirbyte CLI.",
+    help_format="plaintext",
+    version_flags=[],
 )
 
 

airbyte/cloud/sync_results.py

@@ -202,9 +202,7 @@ def get_full_log_text(self) -> str:
                     timestamp = event.get("timestamp", "")
                     level = event.get("level", "INFO")
                     message = event.get("message", "")
-                    log_lines.append(
-                        f"[{timestamp}] {level}: {message}"  # pyrefly: ignore[bad-argument-type]
-                    )
+                    log_lines.append(f"[{timestamp}] {level}: {message}")
                 result = "\n".join(log_lines)
         elif "logLines" in logs_data:
             log_lines = logs_data["logLines"]
@@ -478,7 +476,7 @@ def stream_names(self) -> list[str]:
     @property
     def streams(
         self,
-    ) -> _SyncResultStreams:  # pyrefly: ignore[unknown-name]
+    ) -> _SyncResultStreams:
         """Return a mapping of stream names to `airbyte.CachedDataset` objects.
 
         This is a convenience wrapper around the `stream_names`

airbyte/datasets/_sql.py

@@ -80,7 +80,7 @@ def __init__(
         # Coalesce False to None
         stream_configuration = stream_configuration or None
 
-        super().__init__(stream_metadata=stream_configuration)  # pyrefly: ignore[bad-argument-type]
+        super().__init__(stream_metadata=stream_configuration)  # ty: ignore[invalid-argument-type]
 
     @property
     def stream_name(self) -> str:

airbyte/mcp/local.py

@@ -559,9 +559,7 @@ def get_stream_previews(
     )
     source.set_config(config_dict)
 
-    streams_param: list[str] | Literal["*"] | None = resolve_list_of_strings(
-        streams
-    )  # pyrefly: ignore[no-matching-overload]
+    streams_param: list[str] | Literal["*"] | None = resolve_list_of_strings(streams)
     if streams_param and len(streams_param) == 1 and streams_param[0] == "*":
         streams_param = "*"
 

airbyte/progress.py

@@ -321,11 +321,11 @@ def tally_pending_writes(
 
         update_period = 1  # Reset the update period to 1 before start.
 
-        for count, message in enumerate(  # pyrefly: ignore[bad-assignment]
-            messages,  # pyrefly: ignore[bad-argument-type]
+        for count, message in enumerate(
+            messages,
             start=1,
         ):
-            yield message  # pyrefly: ignore[invalid-yield]
+            yield message  # ty: ignore[invalid-yield]
             if isinstance(message, str):
                 # This is a string message, not an AirbyteMessage.
                 # For now at least, we don't need to pay the cost of parsing it.
@@ -585,7 +585,7 @@ def _log_read_metrics(self) -> None:
         perf_metrics["stream_metrics"] = stream_metrics
         log_dict["performance_metrics"] = perf_metrics
 
-        self._file_logger.info(json.dumps(log_dict))  # pyrefly: ignore[missing-attribute]
+        self._file_logger.info(json.dumps(log_dict))
 
         perf_logger: BoundLogger = logs.get_global_stats_logger()
         perf_logger.info(**log_dict)

airbyte/secrets/google_gsm.py

@@ -58,7 +58,7 @@ class GSMSecretHandle(SecretHandle):
     labels.
     """
 
-    parent: GoogleGSMSecretManager  # pyrefly: ignore[bad-override]
+    parent: GoogleGSMSecretManager
 
     def _get_gsm_secret_object(self) -> secretmanager.Secret:
         """Get the `Secret` object from GSM."""

airbyte/shared/catalog_providers.py

@@ -49,7 +49,7 @@ def __init__(
         """
         self._catalog: ConfiguredAirbyteCatalog = self.validate_catalog(
             configured_catalog
-        )  # pyrefly: ignore[bad-assignment]
+        )  # ty: ignore[invalid-assignment]
 
     @staticmethod
     def validate_catalog(catalog: ConfiguredAirbyteCatalog) -> None:
@@ -66,7 +66,7 @@ def validate_catalog(catalog: ConfiguredAirbyteCatalog) -> None:
             if stream.sync_id is None:
                 stream.sync_id = 1  # This should ideally increment monotonically with each sync.
 
-        return catalog  # pyrefly: ignore[bad-return]
+        return catalog  # ty: ignore[invalid-return-type]
 
     @property
     def configured_catalog(self) -> ConfiguredAirbyteCatalog:
@@ -178,7 +178,7 @@ def get_cursor_key(
         """Return the cursor key for the given stream."""
         return self.get_configured_stream_info(
             stream_name
-        ).cursor_field  # pyrefly: ignore[bad-return]
+        ).cursor_field  # ty: ignore[invalid-return-type]
 
     def resolve_write_method(
         self,

airbyte/shared/sql_processor.py

@@ -251,12 +251,9 @@ def __init__(
         ] = defaultdict(list, {})
 
         self._setup()
-        self.file_writer = (
-            file_writer
-            or self.file_writer_class(  # pyrefly: ignore[bad-instantiation]
-                cache_dir=cast("Path", temp_dir),
-                cleanup=temp_file_cleanup,
-            )
+        self.file_writer = file_writer or self.file_writer_class(
+            cache_dir=cast("Path", temp_dir),
+            cleanup=temp_file_cleanup,
         )
         self.type_converter = self.type_converter_class()
         self._cached_table_definitions: dict[str, sqlalchemy.Table] = {}
@@ -315,7 +312,7 @@ def process_airbyte_messages(
         This method assumes that the catalog is already registered with the processor.
         """
         if not isinstance(write_strategy, WriteStrategy):
-            raise exc.AirbyteInternalError(  # pyrefly: ignore[missing-attribute]
+            raise exc.AirbyteInternalError(
                 message="Invalid `write_strategy` argument. Expected instance of WriteStrategy.",
                 context={"write_strategy": write_strategy},
             )

airbyte/shared/state_providers.py

@@ -67,7 +67,7 @@ def state_message_artifacts(
         if result is None:
             raise exc.PyAirbyteInternalError(message="No state artifacts were declared.")
 
-        return result  # pyrefly: ignore[bad-return]
+        return result  # ty: ignore[invalid-return-type]
 
     @property
     def known_stream_names(
@@ -101,13 +101,13 @@ def get_stream_state(
         """Return the state message for the specified stream name."""
         for state_message in self.state_message_artifacts:
             if (
-                state_message.stream.stream_descriptor.name  # pyrefly: ignore[missing-attribute]
+                state_message.stream.stream_descriptor.name  # ty: ignore[unresolved-attribute]
                 == stream_name
             ):
-                return state_message  # pyrefly: ignore[bad-return]
+                return state_message  # ty: ignore[invalid-return-type]
 
         if not_found != "raise":
-            return not_found  # pyrefly: ignore[bad-return]
+            return not_found  # ty: ignore[invalid-return-type]
 
         raise exc.AirbyteStateNotFoundError(
             message="State message not found.",

airbyte/sources/base.py

@@ -949,10 +949,7 @@ def _read_to_cache(  # noqa: PLR0913  # Too many arguments
                     message="Invalid strategy",
                     context={
                         "write_strategy": write_strategy,
-                        "available_strategies": [
-                            s.value
-                            for s in WriteStrategy  # pyrefly: ignore[not-iterable]
-                        ],
+                        "available_strategies": [s.value for s in WriteStrategy],
                     },
                 ) from None
 

airbyte/types.py

@@ -135,7 +135,6 @@ def to_sql_type(  # noqa: PLR0911  # Too many return statements
         except SQLTypeConversionError:
             print(f"Could not determine airbyte type from JSON schema: {json_schema_property_def}")
         except KeyError:
-            # pyrefly: ignore[unbound-name]
             print(f"Could not find SQL type for airbyte type: {airbyte_type}")
         else:
             # No exceptions were raised, so we can return the SQL type.

pyproject.toml

@@ -59,7 +59,7 @@ dev = [
     "docker>=7.1.0,<8.0",
     "faker>=21.0.0,<22.0",
     "freezegun>=1.4.0,<2.0",
-    "pyrefly>=0.38.0,<0.39.0",
+    "ty>=0.0.56",
     "pandas-stubs>=2.1.4.231218",
     "pdoc>=16.0.0,<17.0",
     "poethepoet>=0.26.1,<0.32.0",
@@ -140,11 +140,11 @@ filterwarnings = [ # syntax: "action:message_regex:category:module:line"
 # [tool.ruff]
 # Ruff moved to dedicated config file: `.ruff.toml`
 
-# [tool.pyrefly]
-# Pyrefly moved to dedicated config file: `pyrefly.toml`
+# [tool.ty]
+# ty moved to dedicated config file: `ty.toml`
 
 [tool.mypy]
-# Disable all mypy checks for this project since we use pyrefly instead
+# Disable all mypy checks for this project since we use ty instead
 ignore_errors = true
 
 [tool.pyright]
@@ -162,7 +162,7 @@ coverage-report = { shell = "coverage report" }
 coverage-html = { shell = "coverage html -d htmlcov && open htmlcov/index.html" }
 coverage-reset = { shell = "coverage erase" }
 
-check = { shell = "ruff check . && pyrefly check && pytest --collect-only -qq" }
+check = { shell = "ruff check . && ty check && pytest --collect-only -qq" }
 
 docs-generate = {env = {PDOC_ALLOW_EXEC = "1"}, cmd = "python -m docs.generate run"}
 docs-preview = {shell = "poe docs-generate && open docs/generated/index.html"}