def _start_dev_server(
saved_bundle_path: str,
api_server_port: int,
enable_swagger: bool,
):
logger.info("Starting BentoML API server in development mode..")
from bentoml.server.api_server import BentoAPIServer
from bentoml.saved_bundle import load_from_dir
bento_service = load_from_dir(saved_bundle_path)
api_server = BentoAPIServer(bento_service, enable_swagger=enable_swagger)
api_server.start(port=api_server_port)
def start_dev_server(saved_bundle_path: str, port: int,
enable_microbatch: bool, run_with_ngrok: bool):
logger.info("Starting BentoML API server in development mode..")
from bentoml import load
from bentoml.server.api_server import BentoAPIServer
from bentoml.marshal.marshal import MarshalService
from bentoml.utils import reserve_free_port
bento_service = load(saved_bundle_path)
if run_with_ngrok:
from bentoml.utils.flask_ngrok import start_ngrok
from threading import Timer
thread = Timer(1, start_ngrok, args=(port, ))
thread.setDaemon(True)
thread.start()
if enable_microbatch:
with reserve_free_port() as api_server_port:
# start server right after port released
# to reduce potential race
marshal_server = MarshalService(
saved_bundle_path,
outbound_host="localhost",
outbound_port=api_server_port,
outbound_workers=1,
)
api_server = BentoAPIServer(bento_service, port=api_server_port)
marshal_server.async_start(port=port)
api_server.start()
else:
api_server = BentoAPIServer(bento_service, port=port)
api_server.start()
def start_dev_server(
saved_bundle_path: str,
port: int = Provide[BentoMLContainer.config.api_server.port],
enable_microbatch: bool = Provide[
BentoMLContainer.config.api_server.enable_microbatch],
mb_max_batch_size: int = Provide[
BentoMLContainer.config.marshal_server.max_batch_size],
mb_max_latency: int = Provide[
BentoMLContainer.config.marshal_server.max_latency],
run_with_ngrok: bool = Provide[
BentoMLContainer.config.api_server.run_with_ngrok],
enable_swagger: bool = Provide[
BentoMLContainer.config.api_server.enable_swagger],
):
logger.info("Starting BentoML API server in development mode..")
import multiprocessing
from bentoml.saved_bundle import load_from_dir
from bentoml.server.api_server import BentoAPIServer
from bentoml.utils import reserve_free_port
if run_with_ngrok:
from threading import Timer
from bentoml.utils.flask_ngrok import start_ngrok
thread = Timer(1, start_ngrok, args=(port, ))
thread.setDaemon(True)
thread.start()
if enable_microbatch:
with reserve_free_port() as api_server_port:
# start server right after port released
# to reduce potential race
marshal_proc = multiprocessing.Process(
target=start_dev_batching_server,
kwargs=dict(
api_server_port=api_server_port,
saved_bundle_path=saved_bundle_path,
port=port,
mb_max_latency=mb_max_latency,
mb_max_batch_size=mb_max_batch_size,
),
daemon=True,
)
marshal_proc.start()
bento_service = load_from_dir(saved_bundle_path)
api_server = BentoAPIServer(bento_service,
port=api_server_port,
enable_swagger=enable_swagger)
api_server.start()
else:
bento_service = load_from_dir(saved_bundle_path)
api_server = BentoAPIServer(bento_service,
port=port,
enable_swagger=enable_swagger)
api_server.start()
def test_api_function_route(bento_service, img_file):
import imageio # noqa # pylint: disable=unused-import
import numpy as np # noqa # pylint: disable=unused-import
rest_server = BentoAPIServer(bento_service)
test_client = rest_server.app.test_client()
index_list = []
for rule in rest_server.app.url_map.iter_rules():
index_list.append(rule.endpoint)
response = test_client.get("/")
assert 200 == response.status_code
response = test_client.get("/healthz")
assert 200 == response.status_code
response = test_client.get("/metadata")
assert 200 == response.status_code
response = test_client.get("/docs.json")
assert 200 == response.status_code
assert "predict_dataframe" in index_list
data = [{"col1": 10}, {"col1": 20}]
response = test_client.post(
"/predict_dataframe", data=json.dumps(data), content_type="application/json"
)
assert response.data.decode().strip() == '[{"col1":20},{"col1":40}]'
assert "predict_dataframe_v1" in index_list
data = [{"col1": 10}, {"col1": 20}]
response = test_client.post(
"/predict_dataframe_v1", data=json.dumps(data), content_type="application/json"
)
assert response.data.decode().strip() == '[{"col1":20},{"col1":40}]'
# Test ImageInput.
with open(str(img_file), "rb") as f:
img = f.read()
response = test_client.post(
"/predict_image", data={'image': (BytesIO(img), 'test_img.png')}
)
assert 200 == response.status_code
assert "[10, 10, 3]" in str(response.data)
response = test_client.post(
"/predict_multi_images",
data={
'original': (BytesIO(img), 'original.jpg'),
'compared': (BytesIO(img), 'compared.jpg'),
},
)
assert 200 == response.status_code
def _start_dev_server(
saved_bundle_path: str,
api_server_port: int,
config: BentoMLConfiguration,
):
logger.info("Starting BentoML API server in development mode..")
from bentoml.saved_bundle import load_from_dir
bento_service = load_from_dir(saved_bundle_path)
from bentoml.server.api_server import BentoAPIServer
container = BentoMLContainer()
container.config.from_dict(config.as_dict())
container.wire(packages=[sys.modules[__name__]])
api_server = BentoAPIServer(bento_service)
api_server.start(port=api_server_port)
def test_api_function_route_with_disabled_swagger(bento_service):
rest_server = BentoAPIServer(bento_service, enable_swagger=False)
test_client = rest_server.app.test_client()
response = test_client.get("/")
assert 404 == response.status_code
response = test_client.get("/docs")
assert 404 == response.status_code
response = test_client.get("/healthz")
assert 200 == response.status_code
response = test_client.get("/docs.json")
assert 200 == response.status_code
def test_api_function_route_with_disabled_swagger(bento_service):
rest_server = BentoAPIServer(
bento_service=bento_service,
enable_swagger=False,
enable_metrics=True,
enable_feedback=True,
request_header_flag="BentoML-Is-Merged-Request",
)
test_client = rest_server.app.test_client()
response = test_client.get("/")
assert 404 == response.status_code
response = test_client.get("/docs")
assert 404 == response.status_code
response = test_client.get("/healthz")
assert 200 == response.status_code
response = test_client.get("/docs.json")
assert 200 == response.status_code
# Copyright 2019 Atalaya Tech, Inc.
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
# http://www.apache.org/licenses/LICENSE-2.0
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import os
import azure.functions as func # pylint: disable=import-error
from bentoml.server.api_server import BentoAPIServer
from bentoml.saved_bundle import load_from_dir
bento_path = os.path.abspath(os.path.join(os.path.dirname(__file__), '..'))
svc = load_from_dir(bento_path)
bento_server = BentoAPIServer(svc)
def main(req: func.HttpRequest, context: func.Context) -> func.HttpResponse:
return func.WsgiMiddleware(bento_server.app.wsgi_app).handle(req, context)
def test_api_function_route(bento_service, img_file):
import imageio # noqa # pylint: disable=unused-import
import numpy as np # noqa # pylint: disable=unused-import
rest_server = BentoAPIServer(
bento_service=bento_service,
port=5000,
enable_swagger=True,
enable_metrics=True,
enable_feedback=True,
request_header_flag="BentoML-Is-Merged-Request",
)
test_client = rest_server.app.test_client()
index_list = []
for rule in rest_server.app.url_map.iter_rules():
index_list.append(rule.endpoint)
response = test_client.get("/")
assert 200 == response.status_code
response = test_client.get("/healthz")
assert 200 == response.status_code
response = test_client.get("/metadata")
assert 200 == response.status_code
response = test_client.get("/docs.json")
assert 200 == response.status_code
docs = json.loads(response.data.decode())
assert f"/{CUSTOM_ROUTE}" in docs["paths"]
response = test_client.post(
f"/{CUSTOM_ROUTE}",
data='{"a": 1}',
)
assert 200 == response.status_code
assert '{"a": 1}' == response.data.decode()
assert "predict_dataframe" in index_list
data = [{"col1": 10}, {"col1": 20}]
response = test_client.post("/predict_dataframe",
data=json.dumps(data),
content_type="application/json")
assert response.data.decode().strip() == '[{"col1":20},{"col1":40}]'
assert "predict_dataframe_v1" in index_list
data = [{"col1": 10}, {"col1": 20}]
response = test_client.post("/predict_dataframe_v1",
data=json.dumps(data),
content_type="application/json")
assert response.data.decode().strip() == '[{"col1":20},{"col1":40}]'
# Test ImageInput.
with open(str(img_file), "rb") as f:
img = f.read()
response = test_client.post("/predict_image",
data={'image': (BytesIO(img), 'test_img.png')})
assert 200 == response.status_code
assert "[10, 10, 3]" in str(response.data)
response = test_client.post(
"/predict_multi_images",
data={
'original': (BytesIO(img), 'original.jpg'),
'compared': (BytesIO(img), 'compared.jpg'),
},
)
assert 200 == response.status_code
