[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
This commit is contained in:
Harry Mellor
2026-09-16 10:57:24 -07:00
committed by GitHub
co-authored by Claude Opus 5
parent 1085b64425
commit 651a88c09a
693 changed files with 2088 additions and 4171 deletions
@@ -12,8 +12,7 @@ from vllm.entrypoints.openai.responses.protocol import ResponsesRequest
def test_function_call_dict_converted_to_object():
"""Test that function_call dictionaries are correctly parsed into
ResponseFunctionToolCall objects.
"""
ResponseFunctionToolCall objects."""
# Create a request with function_call as dict
request_data = {
"model": "gpt-oss",
@@ -330,8 +329,7 @@ def test_validator_handles_empty_iterator():
def test_assistant_string_content_stays_easyinput():
"""EasyInput assistant message with plain string content is not
coerced into a ResponseOutputMessage.
"""
coerced into a ResponseOutputMessage."""
request_data = {
"model": "test-model",
"input": [
@@ -353,8 +351,7 @@ def test_assistant_string_content_stays_easyinput():
def test_assistant_output_style_content_coerced():
"""Assistant message whose content is output-message-shaped (list of
output_text items) should be coerced to ResponseOutputMessage.
"""
output_text items) should be coerced to ResponseOutputMessage."""
request_data = {
"model": "test-model",
"input": [
@@ -67,8 +67,7 @@ class TestResponseInputToHarmonyMessage:
def test_system_message(self):
"""System messages carry developer instructions and must be rendered
as developer messages with DeveloperContent.
"""
as developer messages with DeveloperContent."""
msg = response_input_to_harmony(
{"type": "message", "role": "system", "content": "Be helpful."},
prev_responses=[],
@@ -90,8 +89,7 @@ class TestResponseInputToHarmonyMessage:
def test_developer_message_gets_instructions_prefix(self):
"""Developer messages must use DeveloperContent which adds the
'# Instructions' header the model was trained on.
"""
'# Instructions' header the model was trained on."""
msg = response_input_to_harmony(
{"type": "message", "role": "developer", "content": "Be concise."},
prev_responses=[],
@@ -121,8 +119,7 @@ class TestResponseInputToHarmonyMessage:
def test_developer_message_array_content_concatenated(self):
"""Array content in developer messages is flattened and rendered
via DeveloperContent with the '# Instructions' header.
"""
via DeveloperContent with the '# Instructions' header."""
msg = response_input_to_harmony(
{
"type": "message",
@@ -213,8 +210,7 @@ class TestResponseInputToHarmonyMessage:
def test_function_call_output_uses_most_recent_matching_call(self):
"""When multiple prev_responses share a call_id, the last one wins
because the search is reversed.
"""
because the search is reversed."""
earlier = ResponseFunctionToolCall(
id="fc_old",
call_id="call_test",
@@ -839,13 +839,11 @@ class TestConstructChatMessagesCombinePolicy:
class TestConstructInputMessagesInstructionsLeak:
"""Regression tests for #37697: instructions from a prior response
should NOT leak through previous_response_id.
"""
should NOT leak through previous_response_id."""
def test_old_instructions_stripped_from_prev_msg(self):
"""System message in prev_msg must be dropped so the new request's
instructions are the only system message in the conversation.
"""
instructions are the only system message in the conversation."""
prev = [
{"role": "system", "content": "old instructions"},
{"role": "user", "content": "What is 2+2?"},
@@ -862,8 +860,7 @@ class TestConstructInputMessagesInstructionsLeak:
def test_no_instructions_in_new_request(self):
"""If the new request has no instructions, old ones should still
be stripped -- they must not carry over.
"""
be stripped -- they must not carry over."""
prev = [
{"role": "system", "content": "old instructions"},
{"role": "user", "content": "Hi"},
@@ -1163,8 +1163,7 @@ class TestHarmonyPreambleStreaming:
def test_preamble_done_emits_text_done_events(self) -> None:
"""Completed preamble should emit text done + content_part done +
output_item done, same shape as final channel.
"""
output_item done, same shape as final channel."""
from vllm.entrypoints.openai.responses.streaming_events import (
emit_previous_item_done_events,
)
@@ -1185,8 +1184,7 @@ class TestHarmonyPreambleStreaming:
def test_commentary_with_recipient_no_preamble_done(self) -> None:
"""Commentary + recipient='functions.X' should route to function call
done, not preamble done.
"""
done, not preamble done."""
from vllm.entrypoints.openai.responses.streaming_events import (
emit_previous_item_done_events,
)
@@ -1332,8 +1330,7 @@ def _mock_parser_with_reasoning(serving, delta_sequence: list[DeltaMessage]):
class TestStreamingReasoningToContentTransition:
"""Tests for _process_simple_streaming_events reasoning-to-content
transition, specifically the fix for mixed deltas that carry both
reasoning and content simultaneously.
"""
reasoning and content simultaneously."""
@pytest.mark.asyncio
async def test_mixed_delta_reasoning_and_content_emits_reasoning_delta(
@@ -1343,8 +1340,7 @@ class TestStreamingReasoningToContentTransition:
and content set (e.g. reasoning end and content start in the same
chunk), the trailing reasoning text must be emitted as a
ResponseReasoningTextDeltaEvent and included in the
ResponseReasoningTextDoneEvent text.
"""
ResponseReasoningTextDoneEvent text."""
monkeypatch.setattr(envs, "VLLM_USE_EXPERIMENTAL_PARSER_CONTEXT", False)
serving = _make_serving_instance_with_reasoning()
@@ -1413,8 +1409,7 @@ class TestStreamingReasoningToContentTransition:
self, monkeypatch
):
"""When the transition from reasoning to content is clean (no mixed
delta), no extra reasoning delta event should be emitted.
"""
delta), no extra reasoning delta event should be emitted."""
monkeypatch.setattr(envs, "VLLM_USE_EXPERIMENTAL_PARSER_CONTEXT", False)
serving = _make_serving_instance_with_reasoning()
@@ -1475,8 +1470,7 @@ class TestStreamingReasoningToContentTransition:
async def test_reasoning_only_stream_no_content(self, monkeypatch):
"""When the stream has only reasoning deltas and no content, the
reasoning done event should be emitted at finalization with the
full accumulated text, and no text delta events should appear.
"""
full accumulated text, and no text delta events should appear."""
monkeypatch.setattr(envs, "VLLM_USE_EXPERIMENTAL_PARSER_CONTEXT", False)
serving = _make_serving_instance_with_reasoning()
@@ -143,8 +143,7 @@ async def test_streaming_output_consistency(client: OpenAI, model_name: str):
@pytest.mark.parametrize("model_name", [MODEL_NAME])
async def test_streaming_logprobs(client: OpenAI, model_name: str):
"""Test that streaming with logprobs returns valid logprob data on
output_text.delta events and that top_logprobs has the requested count.
"""
output_text.delta events and that top_logprobs has the requested count."""
response = await client.responses.create(
model=model_name,
input="Say hello.",
@@ -124,8 +124,7 @@ class TestProcessorCompoundDeltas:
def test_split_name_and_args_same_index(self):
"""Regression: parsers like KimiK2 emit name and args as separate
DeltaToolCalls at the same index within one DeltaMessage.
"""
DeltaToolCalls at the same index within one DeltaMessage."""
tc_name = _make_tool_call(0, name="get_weather")
tc_args = _make_tool_call(0, arguments='{"city":"SF"}')
delta = DeltaMessage(tool_calls=[tc_name, tc_args])
@@ -141,8 +140,7 @@ class TestProcessorCompoundDeltas:
def test_reasoning_to_content_transition(self):
"""Regression: the old special case in emit_delta handled this;
now split_delta handles it generically.
"""
now split_delta handles it generically."""
processor = SimpleStreamingEventProcessor()
_run_through_processor(processor, DeltaMessage(reasoning="think"))
assert processor.state.current_state == _StateType.REASONING