chore: fix or noqa remaining mypy issues (now 100% pass)

Author: aaronsteers

airbyte_cdk/cli/source_declarative_manifest/_run.py

@@ -152,7 +152,9 @@ def handle_remote_manifest_command(args: list[str]) -> None:
         )
 
 
-def create_declarative_source(args: list[str]) -> ConcurrentDeclarativeSource:
+def create_declarative_source(
+    args: list[str],
+) -> ConcurrentDeclarativeSource:  # type: ignore [type-arg]
     """Creates the source with the injected config.
 
     This essentially does what other low-code sources do at build time, but at runtime,

airbyte_cdk/connector_builder/message_grouper.py

@@ -232,7 +232,7 @@ def _get_message_groups(
                 current_slice_descriptor = self._parse_slice_description(message.log.message)  # type: ignore[union-attr] # AirbyteMessage with MessageType.LOG has log.message
                 current_slice_pages = []
                 at_least_one_page_in_group = False
-            elif message.type == MessageType.LOG and message.log.message.startswith(
+            elif message.type == MessageType.LOG and message.log.message.startswith(  # type: ignore[union-attr] # None doesn't have 'message'
                 SliceLogger.SLICE_LOG_PREFIX
             ):  # type: ignore[union-attr] # AirbyteMessage with MessageType.LOG has log.message
                 # parsing the first slice
@@ -280,7 +280,7 @@ def _get_message_groups(
                 datetime_format_inferrer.accumulate(message.record)
             elif (
                 message.type == MessageType.CONTROL
-                and message.control.type == OrchestratorType.CONNECTOR_CONFIG
+                and message.control.type == OrchestratorType.CONNECTOR_CONFIG  # type: ignore[union-attr] # None doesn't have 'type'
             ):  # type: ignore[union-attr] # AirbyteMessage with MessageType.CONTROL has control.type
                 yield message.control
             elif message.type == MessageType.STATE:
@@ -310,8 +310,8 @@ def _need_to_close_page(
             and message.type == MessageType.LOG
             and (
                 MessageGrouper._is_page_http_request(json_message)
-                or message.log.message.startswith("slice:")
-            )  # type: ignore[union-attr] # AirbyteMessage with MessageType.LOG has log.message
+                or message.log.message.startswith("slice:")  # type: ignore[union-attr] # AirbyteMessage with MessageType.LOG has log.message
+            )
         )
 
     @staticmethod

airbyte_cdk/destinations/vector_db_based/embedder.py

@@ -107,7 +107,7 @@ def embedding_dimensions(self) -> int:
 class OpenAIEmbedder(BaseOpenAIEmbedder):
     def __init__(self, config: OpenAIEmbeddingConfigModel, chunk_size: int):
         super().__init__(
-            OpenAIEmbeddings(
+            OpenAIEmbeddings(  # type: ignore [call-arg]
                 openai_api_key=config.openai_key, max_retries=15, disallowed_special=()
             ),
             chunk_size,
@@ -118,7 +118,7 @@ class AzureOpenAIEmbedder(BaseOpenAIEmbedder):
     def __init__(self, config: AzureOpenAIEmbeddingConfigModel, chunk_size: int):
         # Azure OpenAI API has — as of 20230927 — a limit of 16 documents per request
         super().__init__(
-            OpenAIEmbeddings(
+            OpenAIEmbeddings(  # type: ignore [call-arg]
                 openai_api_key=config.openai_key,
                 chunk_size=16,
                 max_retries=15,

airbyte_cdk/entrypoint.py

@@ -248,17 +248,18 @@ def handle_record_counts(
             case Type.RECORD:
                 stream_message_count[
                     HashableStreamDescriptor(
-                        name=message.record.stream, namespace=message.record.namespace
+                        name=message.record.stream,  # type: ignore[union-attr] # record has `stream`
+                        namespace=message.record.namespace,  # type: ignore[union-attr] # record has `namespace`
                     )
-                ] += 1.0  # type: ignore[union-attr] # record has `stream` and `namespace`
+                ] += 1.0
             case Type.STATE:
                 stream_descriptor = message_utils.get_stream_descriptor(message)
 
                 # Set record count from the counter onto the state message
                 message.state.sourceStats = message.state.sourceStats or AirbyteStateStats()  # type: ignore[union-attr] # state has `sourceStats`
-                message.state.sourceStats.recordCount = stream_message_count.get(
+                message.state.sourceStats.recordCount = stream_message_count.get(  # type: ignore[union-attr] # state has `sourceStats`
                     stream_descriptor, 0.0
-                )  # type: ignore[union-attr] # state has `sourceStats`
+                )
 
                 # Reset the counter
                 stream_message_count[stream_descriptor] = 0.0

airbyte_cdk/sources/declarative/datetime/min_max_datetime.py

@@ -111,7 +111,7 @@ def create(
         if isinstance(interpolated_string_or_min_max_datetime, InterpolatedString) or isinstance(
             interpolated_string_or_min_max_datetime, str
         ):
-            return MinMaxDatetime(
+            return MinMaxDatetime(  # type: ignore [call-arg]
                 datetime=interpolated_string_or_min_max_datetime, parameters=parameters
             )
         else:

airbyte_cdk/sources/declarative/incremental/datetime_based_cursor.py

@@ -133,7 +133,7 @@ def set_initial_state(self, stream_state: StreamState) -> None:
         :param stream_state: The state of the stream as returned by get_stream_state
         """
         self._cursor = (
-            stream_state.get(self.cursor_field.eval(self.config)) if stream_state else None
+            stream_state.get(self.cursor_field.eval(self.config)) if stream_state else None  # type: ignore [union-attr]
         )  # type: ignore  # cursor_field is converted to an InterpolatedString in __post_init__
 
     def observe(self, stream_slice: StreamSlice, record: Record) -> None:
@@ -158,7 +158,9 @@ def observe(self, stream_slice: StreamSlice, record: Record) -> None:
         )
         if (
             self._is_within_daterange_boundaries(
-                record, stream_slice.get(start_field), stream_slice.get(end_field)
+                record,
+                stream_slice.get(start_field),  # type: ignore [arg-type]
+                stream_slice.get(end_field),  # type: ignore [arg-type]
             )  # type: ignore # we know that stream_slices for these cursors will use a string representing an unparsed date
             and is_highest_observed_cursor_value
         ):
@@ -368,7 +370,7 @@ def _get_request_options(
                 self._partition_field_start.eval(self.config)
             )
         if self.end_time_option and self.end_time_option.inject_into == option_type:
-            options[self.end_time_option.field_name.eval(config=self.config)] = stream_slice.get(
+            options[self.end_time_option.field_name.eval(config=self.config)] = stream_slice.get(  # type: ignore [union-attr]
                 self._partition_field_end.eval(self.config)
             )  # type: ignore # field_name is always casted to an interpolated string
         return options

airbyte_cdk/sources/declarative/partition_routers/substream_partition_router.py

@@ -130,7 +130,7 @@ def _get_request_option(
                     if value:
                         params.update(
                             {
-                                parent_config.request_option.field_name.eval(
+                                parent_config.request_option.field_name.eval(  # type: ignore [union-attr]
                                     config=self.config
                                 ): value
                             }
@@ -162,7 +162,7 @@ def stream_slices(self) -> Iterable[StreamSlice]:
                 extra_fields = None
                 if parent_stream_config.extra_fields:
                     extra_fields = [
-                        [field_path_part.eval(self.config) for field_path_part in field_path]
+                        [field_path_part.eval(self.config) for field_path_part in field_path]  # type: ignore [union-attr]
                         for field_path in parent_stream_config.extra_fields
                     ]  # type: ignore # extra_fields is always casted to an interpolated string
 
@@ -192,7 +192,10 @@ def stream_slices(self) -> Iterable[StreamSlice]:
                             message=f"Parent stream returned records as invalid type {type(parent_record)}"
                         )
                     try:
-                        partition_value = dpath.get(parent_record, parent_field)
+                        partition_value = dpath.get(
+                            parent_record,  # type: ignore [arg-type]
+                            parent_field,
+                        )
                     except KeyError:
                         continue
 
@@ -228,7 +231,10 @@ def _extract_extra_fields(
         if extra_fields:
             for extra_field_path in extra_fields:
                 try:
-                    extra_field_value = dpath.get(parent_record, extra_field_path)
+                    extra_field_value = dpath.get(
+                        parent_record,  # type: ignore [arg-type]
+                        extra_field_path,
+                    )
                     self.logger.debug(
                         f"Extracted extra_field_path: {extra_field_path} with value: {extra_field_value}"
                     )
@@ -291,7 +297,7 @@ def set_initial_state(self, stream_state: StreamState) -> None:
         if not parent_state and incremental_dependency:
             # Attempt to retrieve child state
             substream_state = list(stream_state.values())
-            substream_state = substream_state[0] if substream_state else {}
+            substream_state = substream_state[0] if substream_state else {}  # type: ignore [assignment]  # Incorrect type for assignment
             parent_state = {}
 
             # Copy child state to parent streams with incremental dependencies

airbyte_cdk/sources/declarative/requesters/error_handlers/http_response_filter.py

@@ -151,16 +151,18 @@ def _create_error_message(self, response: requests.Response) -> Optional[str]:
         :param response: The HTTP response which can be used during interpolation
         :return: The evaluated error message string to be emitted
         """
-        return self.error_message.eval(
+        return self.error_message.eval(  # type: ignore [no-any-return, union-attr]
             self.config, response=self._safe_response_json(response), headers=response.headers
         )  # type: ignore # error_message is always cast to an interpolated string
 
     def _response_matches_predicate(self, response: requests.Response) -> bool:
         return (
             bool(
-                self.predicate.condition
-                and self.predicate.eval(
-                    None, response=self._safe_response_json(response), headers=response.headers
+                self.predicate.condition  # type: ignore [union-attr]
+                and self.predicate.eval(  # type: ignore [union-attr]
+                    None,  # type: ignore [arg-type]
+                    response=self._safe_response_json(response),  # type: ignore [arg-type]
+                    headers=response.headers,
                 )
             )
             if self.predicate

airbyte_cdk/sources/declarative/requesters/paginators/default_paginator.py

@@ -194,7 +194,7 @@ def _get_request_options(self, option_type: RequestOptionType) -> MutableMapping
             and self.pagination_strategy.get_page_size()
             and self.page_size_option.inject_into == option_type
         ):
-            options[self.page_size_option.field_name.eval(config=self.config)] = (
+            options[self.page_size_option.field_name.eval(config=self.config)] = (  # type: ignore [union-attr]
                 self.pagination_strategy.get_page_size()
             )  # type: ignore # field_name is always cast to an interpolated string
         return options

airbyte_cdk/sources/declarative/requesters/request_options/datetime_based_request_options_provider.py

@@ -85,7 +85,7 @@ def _get_request_options(
                 self._partition_field_start.eval(self.config)
             )
         if self.end_time_option and self.end_time_option.inject_into == option_type:
-            options[self.end_time_option.field_name.eval(config=self.config)] = stream_slice.get(
+            options[self.end_time_option.field_name.eval(config=self.config)] = stream_slice.get(  # type: ignore [union-attr]
                 self._partition_field_end.eval(self.config)
             )  # type: ignore # field_name is always casted to an interpolated string
         return options