Skip to content

Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.

fix(examples): keep standalone client checks out of pytest fixture discovery

zhifu gao04ca7edc95a6
diff --git a/.github/workflows/test-example-collection.yml b/.github/workflows/test-example-collection.yml
index 7461953df..42f25547e 100644
--- a/.github/workflows/test-example-collection.yml
+++ b/.github/workflows/test-example-collection.yml
@@ -5,6 +5,8 @@ on:
paths:
- "examples/mcp_server/**"
- "examples/openai_api/**"
+ - "examples/industrial_data_pretraining/fun_asr_nano/client_test.py"
+ - "tests/test_streaming_client_collection.py"
- "tests/test_example_smoke_collection.py"
- "tests/test_mcp_server_example.py"
- "tests/test_openai_api_docker_validation.py"
@@ -14,6 +16,8 @@ on:
paths:
- "examples/mcp_server/**"
- "examples/openai_api/**"
+ - "examples/industrial_data_pretraining/fun_asr_nano/client_test.py"
+ - "tests/test_streaming_client_collection.py"
- "tests/test_example_smoke_collection.py"
- "tests/test_mcp_server_example.py"
- "tests/test_openai_api_docker_validation.py"
@@ -37,14 +41,15 @@ jobs:
- uses: actions/setup-python@v5
with:
python-version: ${{ matrix.python }}
- - name: Install test runner
- run: python -m pip install pytest
+ - name: Install test runner and client dependencies
+ run: python -m pip install pytest numpy soundfile 'websockets>=10,<16'
- name: Check collection and standalone CLI compatibility
run: |
git rev-parse HEAD > source-commit.txt
python -m pip freeze > resolved-dependencies.txt
python -m pytest -q \
tests/test_example_smoke_collection.py \
+ tests/test_streaming_client_collection.py \
tests/test_mcp_server_example.py \
tests/test_openai_api_docker_validation.py \
--junitxml=example-results.xml
diff --git a/examples/industrial_data_pretraining/fun_asr_nano/client_test.py b/examples/industrial_data_pretraining/fun_asr_nano/client_test.py
index 9efc9f260..70e56be6d 100644
--- a/examples/industrial_data_pretraining/fun_asr_nano/client_test.py
+++ b/examples/industrial_data_pretraining/fun_asr_nano/client_test.py
@@ -45,7 +45,7 @@ def load_audio(file_path):
return audio.astype(np.float32)
-async def test_basic(server, audio, hotwords=None):
+async def check_basic(server, audio, hotwords=None):
"""Test basic streaming ASR flow."""
print("=" * 60)
print("TEST: Basic Streaming ASR")
@@ -136,7 +136,7 @@ async def test_basic(server, audio, hotwords=None):
return True
-async def test_empty_audio(server):
+async def check_empty_audio(server):
"""Test with very short/empty audio."""
print("\n" + "=" * 60)
print("TEST: Empty/Short Audio Handling")
@@ -167,7 +167,7 @@ async def test_empty_audio(server):
return True
-async def test_multiple_sessions(server, audio):
+async def check_multiple_sessions(server, audio):
"""Test multiple consecutive sessions on same connection."""
print("\n" + "=" * 60)
print("TEST: Multiple Sessions")
@@ -210,9 +210,9 @@ async def run_tests(args):
failed = 0
tests = [
- ("Basic Streaming", lambda: test_basic(args.server, audio, args.hotwords)),
- ("Empty Audio", lambda: test_empty_audio(args.server)),
- ("Multiple Sessions", lambda: test_multiple_sessions(args.server, audio)),
+ ("Basic Streaming", lambda: check_basic(args.server, audio, args.hotwords)),
+ ("Empty Audio", lambda: check_empty_audio(args.server)),
+ ("Multiple Sessions", lambda: check_multiple_sessions(args.server, audio)),
]
for name, test_fn in tests:
diff --git a/tests/test_streaming_client_collection.py b/tests/test_streaming_client_collection.py
new file mode 100644
index 000000000..f7be29c54
--- /dev/null
+++ b/tests/test_streaming_client_collection.py
@@ -0,0 +1,106 @@
+"""Exercise the manual client CLI without a speech model or external service."""
+
+import asyncio
+import json
+import os
+from pathlib import Path
+import shutil
+import struct
+import subprocess
+import sys
+import wave
+
+import pytest
+from websockets.legacy.server import serve
+
+
+ROOT = Path(__file__).resolve().parents[1]
+CLIENT = ROOT / "examples/industrial_data_pretraining/fun_asr_nano/client_test.py"
+
+
[email protected]("import_mode", ["prepend", "append", "importlib"])
+def test_manual_client_does_not_request_pytest_fixtures(tmp_path, import_mode):
+ client = tmp_path / "client_test.py"
+ shutil.copyfile(CLIENT, client)
+ sentinel = tmp_path / "test_sentinel.py"
+ sentinel.write_text("def test_sentinel():\n pass\n", encoding="utf-8")
+ env = dict(os.environ, PYTEST_DISABLE_PLUGIN_AUTOLOAD="1")
+ env.pop("PYTEST_ADDOPTS", None)
+ env.pop("PYTEST_PLUGINS", None)
+ result = subprocess.run(
+ [sys.executable, "-m", "pytest", "--collect-only", "-q",
+ "--import-mode=" + import_mode, str(client), str(sentinel)],
+ cwd=tmp_path, env=env, capture_output=True, text=True, timeout=30,
+ )
+ assert result.returncode == 0, result.stdout + result.stderr
+ assert "1 test collected" in result.stdout, result.stdout
+ assert "test_sentinel.py::test_sentinel" in result.stdout
+
+
+def test_manual_client_help_preserves_cli_arguments():
+ result = subprocess.run(
+ [sys.executable, str(CLIENT), "--help"], cwd=CLIENT.parent,
+ capture_output=True, text=True, timeout=30,
+ )
+ assert result.returncode == 0, result.stdout + result.stderr
+ for option in ("--server", "--file", "--hotwords"):
+ assert option in result.stdout
+
+
[email protected]("hotwords", ["", "hello,world"])
[email protected]("missing_final", [False, True])
+def test_manual_client_checks_real_loopback_protocol(tmp_path, hotwords, missing_final):
+ audio_path = tmp_path / "short.wav"
+ pcm = struct.pack("<1600h", *(i % 300 for i in range(1600)))
+ with wave.open(str(audio_path), "wb") as audio:
+ audio.setnchannels(1)
+ audio.setsampwidth(2)
+ audio.setframerate(16000)
+ audio.writeframes(pcm)
+
+ async def run():
+ sessions = []
+ received_hotwords = []
+
+ async def handler(ws, path=None):
+ chunks = []
+ async for message in ws:
+ if message == "START":
+ chunks = []
+ await ws.send(json.dumps({"event": "started"}))
+ elif isinstance(message, bytes):
+ chunks.append(message)
+ elif message.startswith("HOTWORDS:"):
+ received_hotwords.append(message[9:])
+ await ws.send(json.dumps({"event": "hotwords_set", "hotwords": hotwords.split(",")}))
+ elif message == "STOP":
+ sessions.append(b"".join(chunks))
+ if not missing_final:
+ await ws.send(json.dumps({"is_final": True, "sentences": [
+ {"text": "local protocol", "start": 0, "end": 100},
+ ]}))
+ await ws.send(json.dumps({"event": "stopped"}))
+ else:
+ raise AssertionError("Unexpected client command: " + message)
+
+ async with serve(handler, "127.0.0.1", 0) as server:
+ port = server.sockets[0].getsockname()[1]
+ command = [sys.executable, str(CLIENT), "--server", f"ws://127.0.0.1:{port}",
+ "--file", str(audio_path), "--hotwords", hotwords]
+ process = await asyncio.create_subprocess_exec(
+ *command, cwd=CLIENT.parent,
+ stdout=asyncio.subprocess.PIPE, stderr=asyncio.subprocess.PIPE,
+ )
+ try:
+ stdout, stderr = await asyncio.wait_for(process.communicate(), timeout=20)
+ finally:
+ if process.returncode is None:
+ process.kill()
+ await process.wait()
+ return process.returncode, stdout.decode(), stderr.decode(), sessions, received_hotwords
+
+ code, output, errors, sessions, received_hotwords = asyncio.run(run())
+ assert code == (1 if missing_final else 0), output + errors
+ assert ("RESULTS: 0 passed, 3 failed" if missing_final else "RESULTS: 3 passed, 0 failed") in output
+ assert sessions == ([pcm, b"\0" * 1600, pcm] if missing_final else [pcm, b"\0" * 1600, pcm, pcm])
+ assert received_hotwords == ([hotwords] if hotwords else [])