Merge pull request #92 from opensorceror/feat/python-server-tests

Add tests for Python websocket server

Author: lissyx

python_websocket_server/.pylintrc

@@ -21,7 +21,7 @@ Server configuration is specified in the [`application.conf`](application.conf)
 
 Make sure your model and scorer files are present in the same directory as the `application.conf` file. Then execute:
 
-```
+```sh
 python -m deepspeech_server.app
 ```
 
@@ -70,3 +70,24 @@ DeepSpeech service. The websocket timeout on the ingress is set to 1 hour.
 ## Contributing
 
 Bug reports and merge requests are welcome.
+
+### Running pylint analysis
+
+```sh
+pylint deepspeech_server
+```
+
+### Running tests
+
+To run tests without coverage, execute:
+
+```sh
+python -m pytest tests/test_app.py
+```
+
+To run tests with coverage, and to print coverage to the terminal and write a coverage report, execute:
+
+```sh
+python -m pytest -p pytest_cov --cov=deepspeech_server --cov-report=xml --cov-report=term \
+		  --junitxml=pytest-report.xml tests/test_app.py
+```

python_websocket_server/README.md

@@ -11,4 +11,6 @@ server {
   threadpool {
     count = 5
   }
+
+  stt_endpoint = "/api/v1/stt"
 }

python_websocket_server/application.conf

@@ -1,3 +1,5 @@
+"""Module containing server endpoints and startup functionality"""
+
 import json
 from concurrent.futures import ThreadPoolExecutor
 from pathlib import Path
@@ -8,7 +10,7 @@
 from sanic.log import logger
 
 from deepspeech_server.engine import SpeechToTextEngine
-from deepspeech_server.models import Response, Error
+from deepspeech_server.models import Response, ErrorResponse
 
 # Load app configs and initialize DeepSpeech model
 conf = ConfigFactory.parse_file("application.conf")
@@ -24,26 +26,28 @@
 
 @app.route("/", methods=["GET"])
 async def healthcheck(_):
+    """Route for simple healthcheck that simply returns greeting message"""
     return response.text("Welcome to DeepSpeech Server!")
 
 
-@app.websocket("/api/v1/stt")
-async def stt(request, ws):
+@app.websocket(conf['server.stt_endpoint'])
+async def stt(request, websocket):
+    """Route for requesting speech-to-text transcription"""
     logger.debug(f"Received {request.method} request at {request.path}")
     try:
-        audio = await ws.recv()
+        audio = await websocket.recv()
 
         inference_start = perf_counter()
         text = await app.loop.run_in_executor(executor, lambda: engine.run(audio))
         inference_end = perf_counter() - inference_start
 
-        await ws.send(json.dumps(Response(text, inference_end).__dict__))
+        await websocket.send(json.dumps(Response(text, inference_end).__dict__))
         logger.debug(f"Completed {request.method} request at {request.path} in {inference_end} seconds")
-    except Exception as e:  # pylint: disable=broad-except
-        logger.debug(f"Failed to process {request.method} request at {request.path}. The exception is: {str(e)}.")
-        await ws.send(json.dumps(Error("Something went wrong").__dict__))
+    except Exception as ex:  # pylint: disable=broad-except
+        logger.debug(f"Failed to process {request.method} request at {request.path}. The exception is: {str(ex)}.")
+        await websocket.send(json.dumps(ErrorResponse("Something went wrong").__dict__))
 
-    await ws.close()
+    await websocket.close()
 
 
 if __name__ == "__main__":

python_websocket_server/deepspeech_server/app.py

@@ -1,3 +1,5 @@
+"""Module containing speech-to-text transcription functionality"""
+
 import wave
 from io import BytesIO
 
@@ -7,6 +9,7 @@
 
 
 def normalize_audio(audio):
+    """Normalize the audio into the format required by DeepSpeech"""
     out, err = (
         ffmpeg.input("pipe:0")
         .output(
@@ -26,11 +29,13 @@ def normalize_audio(audio):
 
 
 class SpeechToTextEngine:
+    """Class to perform speech-to-text transcription and related functionality"""
     def __init__(self, model_path, scorer_path):
         self.model = Model(model_path=model_path)
         self.model.enableExternalScorer(scorer_path=scorer_path)
 
     def run(self, audio):
+        """Perform speech-to-text transcription"""
         audio = normalize_audio(audio)
         audio = BytesIO(audio)
         with wave.Wave_read(audio) as wav:

python_websocket_server/deepspeech_server/engine.py

@@ -1,9 +1,13 @@
+"""Module containing models used for server responses"""
+
 class Response:
+    """Response indicating successful operation"""
     def __init__(self, text, time):
         self.text = text
         self.time = time
 
 
-class Error:
+class ErrorResponse:
+    """Response indicating failed operation"""
     def __init__(self, message):
         self.message = message

python_websocket_server/deepspeech_server/models.py

@@ -4,4 +4,9 @@ black>=20.8b1
 ffmpeg-python==0.2.0
 sanic==20.3.0
 pyhocon==0.3.54
-numpy~=1.17.0
+numpy~=1.19.3
+pytest-sanic==1.6.2
+click~=7.1.2
+pytest~=6.1.2
+pylint>=2.6.0
+pytest-cov~=2.10.1

python_websocket_server/requirements.txt

@@ -0,0 +1,37 @@
+import ast
+
+import pytest
+from pyhocon import ConfigFactory
+from sanic.websocket import WebSocketProtocol
+
+from deepspeech_server.app import app
+
+# Load app configs and initialize DeepSpeech model
+conf = ConfigFactory.parse_file("application.conf")
+
+
+@pytest.fixture
+def sanic_server(loop, sanic_client):
+    return loop.run_until_complete(sanic_client(app, protocol=WebSocketProtocol))
+
+
+def test_index_returns_200():
+    request, response = app.test_client.get("/")
+    assert response.status == 200
+
+
+async def test_valid_audio(sanic_server):
+    ws_conn = await sanic_server.ws_connect(conf['server.stt_endpoint'])
+    with open("tests/experience_proves_this.wav", mode="rb") as file:
+        audio = file.read()
+        await ws_conn.send_bytes(audio)
+        result = await ws_conn.receive()
+        assert ast.literal_eval(result.data)["text"] == "experience proves this"
+
+
+async def test_invalid_audio(sanic_server):
+    ws_conn = await sanic_server.ws_connect(conf['server.stt_endpoint'])
+    audio = b"this ain't audio"
+    await ws_conn.send_bytes(audio)
+    result = await ws_conn.receive()
+    assert ast.literal_eval(result.data)["message"] == "Something went wrong"