Skip to content
Snippets Groups Projects

Compare revisions

Changes are shown as if the source revision was being merged into the target revision. Learn more about comparing revisions.

Source

Select target project
No results found

Target

Select target project
  • workers/base-worker
1 result
Show changes
Commits on Source (8)
Hide whitespace changes
Inline Side-by-side
Showing
with 410 additions and 294 deletions
VERSION
View file @ d278157b
0.2.4
0.3.0-rc1
arkindex_worker/git.py
View file @ d278157b
......@@ -207,7 +207,6 @@ class GitHelper:
in worker.configure() configure the git helper and start the cloning:
```
gitlab = GitlabHelper(...)
workflow_id = os.environ["ARKINDEX_PROCESS_ID"]
prepare_git_key(...)
self.git_helper = GitHelper(workflow_id=workflow_id, gitlab_helper=gitlab, ...)
self.git_helper.run_clone_in_background()
......
arkindex_worker/worker/__init__.py
View file @ d278157b
......@@ -136,7 +136,14 @@ class ElementsWorker(
return self.process_information.get("activity_state") == "ready"
def configure(self):
super().configure()
# CLI args are stored on the instance so that implementations can access them
self.args = self.parser.parse_args()
if self.is_read_only:
super().configure_for_developers()
else:
super().configure()
super().configure_cache()
# Add report concerning elements
self.report = Reporter(
**self.worker_details, version=getattr(self, "worker_version_id", None)
......@@ -199,7 +206,9 @@ class ElementsWorker(
if isinstance(e, ErrorResponse):
message = f"An API error occurred while processing element {element_id}: {e.title} - {e.content}"
else:
message = f"Failed running worker on element {element_id}: {e}"
message = (
f"Failed running worker on element {element_id}: {repr(e)}"
)
logger.warning(
message,
......
arkindex_worker/worker/base.py
View file @ d278157b
......@@ -59,6 +59,39 @@ class BaseWorker(object):
"""
self.parser = argparse.ArgumentParser(description=description)
self.parser.add_argument(
"-c",
"--config",
help="Alternative configuration file when running without a Worker Version ID",
type=open,
)
self.parser.add_argument(
"-d",
"--database",
help="Alternative SQLite database to use for worker caching",
type=str,
default=None,
)
self.parser.add_argument(
"-v",
"--verbose",
"--debug",
help="Display more information on events and errors",
action="store_true",
default=False,
)
self.parser.add_argument(
"--dev",
help=(
"Run worker in developer mode. "
"Worker will be in read-only state even if a worker_version is supplied. "
),
action="store_true",
default=False,
)
# Call potential extra arguments
self.add_arguments()
# Setup workdir either in Ponos environment or on host's home
if os.environ.get("PONOS_DATA"):
......@@ -77,6 +110,11 @@ class BaseWorker(object):
logger.warning(
"Missing WORKER_VERSION_ID environment variable, worker is in read-only mode"
)
self.worker_run_id = os.environ.get("ARKINDEX_WORKER_RUN_ID")
if not self.worker_run_id:
logger.warning(
"Missing ARKINDEX_WORKER_RUN_ID environment variable, worker is in read-only mode"
)
logger.info(f"Worker will use {self.work_dir} as working directory")
......@@ -96,90 +134,20 @@ class BaseWorker(object):
or when no worker version ID is provided.
:rtype: bool
"""
return self.args.dev or self.worker_version_id is None
def configure(self):
"""
Configure worker using CLI args and environment variables.
"""
self.parser.add_argument(
"-c",
"--config",
help="Alternative configuration file when running without a Worker Version ID",
type=open,
)
self.parser.add_argument(
"-d",
"--database",
help="Alternative SQLite database to use for worker caching",
type=str,
default=None,
return (
self.args.dev
or self.worker_version_id is None
or self.worker_run_id is None
)
self.parser.add_argument(
"-v",
"--verbose",
help="Display more information on events and errors",
action="store_true",
default=False,
)
self.parser.add_argument(
"--dev",
help=(
"Run worker in developer mode. "
"Worker will be in read-only state even if a worker_version is supplied. "
"ARKINDEX_PROCESS_ID environment variable is not required with this flag."
),
action="store_true",
)
# Call potential extra arguments
self.add_arguments()
# CLI args are stored on the instance so that implementations can access them
self.args = self.parser.parse_args()
# Setup logging level
if self.args.verbose:
def configure_for_developers(self):
assert self.is_read_only
# Setup logging level if verbose or if ARKINDEX_DEBUG is set to true
if self.args.verbose or os.environ.get("ARKINDEX_DEBUG"):
logger.setLevel(logging.DEBUG)
logger.debug("Debug output enabled")
# Build Arkindex API client from environment variables
self.api_client = ArkindexClient(**options_from_env())
logger.debug(f"Setup Arkindex API client on {self.api_client.document.url}")
# Load features available on backend, and check authentication
user = self.request("RetrieveUser")
logger.debug(f"Connected as {user['display_name']} - {user['email']}")
self.features = user["features"]
# Load process information except in developer mode
if not self.args.dev:
assert os.environ.get(
"ARKINDEX_PROCESS_ID"
), "ARKINDEX_PROCESS_ID environment variable is not defined"
self.process_information = self.request(
"RetrieveDataImport", id=os.environ["ARKINDEX_PROCESS_ID"]
)
if self.worker_version_id:
# Retrieve initial configuration from API
worker_version = self.request(
"RetrieveWorkerVersion", id=self.worker_version_id
)
logger.info(
f"Loaded worker {worker_version['worker']['name']} revision {worker_version['revision']['hash'][0:7]} from API"
)
self.config = worker_version["configuration"]["configuration"]
if "user_configuration" in worker_version["configuration"]:
# Add default values (if set) to user_configuration
for key, value in worker_version["configuration"][
"user_configuration"
].items():
if "default" in value:
self.user_configuration[key] = value["default"]
self.worker_details = worker_version["worker"]
required_secrets = worker_version["configuration"].get("secrets", [])
elif self.args.config:
if self.args.config:
# Load config from YAML file
self.config = yaml.safe_load(self.args.config)
self.worker_details = {"name": "Local worker"}
......@@ -196,20 +164,58 @@ class BaseWorker(object):
# Load all required secrets
self.secrets = {name: self.load_secret(name) for name in required_secrets}
# Load worker run configuration when available and not in dev mode
if os.environ.get("ARKINDEX_WORKER_RUN_ID") and not self.args.dev:
worker_run = self.request(
"RetrieveWorkerRun", id=os.environ["ARKINDEX_WORKER_RUN_ID"]
)
configuration_id = worker_run.get("configuration_id")
if configuration_id:
worker_configuration = self.request(
"RetrieveWorkerConfiguration", id=configuration_id
)
self.user_configuration = worker_configuration.get("configuration")
if self.user_configuration:
logger.info("Loaded user configuration from WorkerRun")
def configure(self):
"""
Configure worker using CLI args and environment variables.
"""
assert not self.is_read_only
# Setup logging level if verbose or if ARKINDEX_DEBUG is set to true
if self.args.verbose or os.environ.get("ARKINDEX_DEBUG"):
logger.setLevel(logging.DEBUG)
logger.debug("Debug output enabled")
# Build Arkindex API client from environment variables
self.api_client = ArkindexClient(**options_from_env())
logger.debug(f"Setup Arkindex API client on {self.api_client.document.url}")
# Load worker run information
worker_run = self.request("RetrieveWorkerRun", id=self.worker_run_id)
# Load process information
self.process_information = worker_run["process"]
# Load worker version information
worker_version = worker_run["worker_version"]
self.worker_details = worker_version["worker"]
logger.info(
f"Loaded worker {self.worker_details['name']} revision {worker_version['revision']['hash'][0:7]} from API"
)
# Retrieve initial configuration from API
self.config = worker_version["configuration"]["configuration"]
if "user_configuration" in worker_version["configuration"]:
# Add default values (if set) to user_configuration
for key, value in worker_version["configuration"][
"user_configuration"
].items():
if "default" in value:
self.user_configuration[key] = value["default"]
# Load all required secrets
required_secrets = worker_version["configuration"].get("secrets", [])
self.secrets = {name: self.load_secret(name) for name in required_secrets}
# Load worker run configuration when available
worker_configuration = worker_run.get("configuration")
self.user_configuration = worker_configuration.get("configuration")
if self.user_configuration:
logger.info("Loaded user configuration from WorkerRun")
# if debug mode is set to true activate debug mode in logger
if self.user_configuration.get("debug"):
logger.setLevel(logging.DEBUG)
logger.debug("Debug output enabled")
def configure_cache(self):
task_id = os.environ.get("PONOS_TASK")
paths = None
if self.support_cache and self.args.database is not None:
......
requirements.txt
View file @ d278157b
arkindex-client==1.0.8
arkindex-client==1.0.9
peewee==3.14.10
Pillow>=9.0
python-gitlab==2.7.1
......
tests/conftest.py
View file @ d278157b
......@@ -104,142 +104,86 @@ def setup_api(responses, monkeypatch, cache_yaml):
def give_env_variable(request, monkeypatch):
"""Defines required environment variables"""
monkeypatch.setenv("WORKER_VERSION_ID", "12341234-1234-1234-1234-123412341234")
monkeypatch.setenv("ARKINDEX_PROCESS_ID", "aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeffff")
monkeypatch.setenv("ARKINDEX_WORKER_RUN_ID", "56785678-5678-5678-5678-567856785678")
monkeypatch.setenv("ARKINDEX_CORPUS_ID", "11111111-1111-1111-1111-111111111111")
@pytest.fixture
def mock_config_api(mock_worker_version_api, mock_process_api, mock_user_api):
"""Mock all API endpoints required to configure a worker"""
pass
@pytest.fixture
def mock_worker_version_api(responses):
"""Provide a mock API response to get worker configuration"""
def mock_worker_run_api(responses):
"""Provide a mock API response to get worker run information"""
payload = {
"id": "12341234-1234-1234-1234-123412341234",
"configuration": {
"docker": {"image": "python:3"},
"configuration": {"someKey": "someValue"},
"secrets": [],
},
"revision": {
"hash": "deadbeef1234",
"name": "some git revision",
},
"docker_image": "python:3",
"docker_image_name": "python:3",
"state": "created",
"id": "56785678-5678-5678-5678-567856785678",
"parents": [],
"worker_version_id": "12341234-1234-1234-1234-123412341234",
"model_version_id": None,
"dataimport_id": "0e6053d5-0d50-41dd-88b6-90907493c433",
"worker": {
"id": "deadbeef-1234-5678-1234-worker",
"name": "Fake worker",
"slug": "fake_worker",
"type": "classifier",
},
}
responses.add(
responses.GET,
"http://testserver/api/v1/workers/versions/12341234-1234-1234-1234-123412341234/",
status=200,
body=json.dumps(payload),
content_type="application/json",
)
@pytest.fixture
def mock_worker_version_user_configuration_api(responses):
"""
Provide a mock API response to get a worker configuration
that includes a `user_configuration`
"""
payload = {
"worker": {"id": "1234", "name": "Workerino", "slug": "workerino"},
"revision": {"hash": "1234lala-lalalalala-lala"},
"configuration": {
"configuration": {"param_1": "/some/path/file.pth", "param_2": 12},
"user_configuration": {
"param_3": {
"title": "A Third Parameter",
"type": "string",
"default": "Animula vagula blandula",
},
"param_4": {"title": "Parameter The Fourth", "type": "int"},
"param_5": {
"title": "Parameter 5 (Five)",
"type": "bool",
"default": True,
},
"configuration_id": "aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeffff",
"worker_version": {
"id": "12341234-1234-1234-1234-123412341234",
"configuration": {
"docker": {"image": "python:3"},
"configuration": {"someKey": "someValue"},
"secrets": [],
},
"revision": {
"hash": "deadbeef1234",
"name": "some git revision",
},
"docker_image": "python:3",
"docker_image_name": "python:3",
"state": "created",
"worker": {
"id": "deadbeef-1234-5678-1234-worker",
"name": "Fake worker",
"slug": "fake_worker",
"type": "classifier",
},
},
}
responses.add(
responses.GET,
"http://testserver/api/v1/workers/versions/12341234-1234-1234-1234-123412341234/",
status=200,
body=json.dumps(payload),
content_type="application/json",
)
@pytest.fixture
def mock_process_api(responses):
"""Provide a mock of the API response to get information on a process. Workers activity is enabled"""
payload = {
"name": None,
"id": "aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeffff",
"state": "running",
"mode": "workers",
"corpus": "11111111-1111-1111-1111-111111111111",
"workflow": "http://testserver/ponos/v1/workflow/12341234-1234-1234-1234-123412341234/",
"files": [],
"revision": None,
"element": {
"id": "12341234-1234-1234-1234-123412341234",
"type": "folder",
"name": "Test folder",
"corpus": {
"id": "11111111-1111-1111-1111-111111111111",
"name": "John Doe project",
"public": False,
"configuration": {
"id": "497f6eca-6276-4993-bfeb-53cbbbba6f08",
"name": "string",
"configuration": {},
},
"process": {
"name": None,
"id": "aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeffff",
"state": "running",
"mode": "workers",
"corpus": "11111111-1111-1111-1111-111111111111",
"workflow": "http://testserver/ponos/v1/workflow/12341234-1234-1234-1234-123412341234/",
"files": [],
"revision": None,
"element": {
"id": "1234-deadbeef",
"type": "folder",
"name": "Test folder",
"corpus": {
"id": "11111111-1111-1111-1111-111111111111",
"name": "John Doe project",
"public": False,
},
"thumbnail_url": "http://testserver/thumbnail.png",
"zone": None,
"thumbnail_put_url": "http://testserver/thumbnail.png",
},
"thumbnail_url": "http://testserver/thumbnail.png",
"zone": None,
"thumbnail_put_url": "http://testserver/thumbnail.png",
"folder_type": None,
"element_type": "page",
"element_name_contains": None,
"load_children": True,
"use_cache": False,
"activity_state": "ready",
},
"folder_type": None,
"element_type": "page",
"element_name_contains": None,
"load_children": True,
"use_cache": False,
"activity_state": "ready",
}
responses.add(
responses.GET,
"http://testserver/api/v1/imports/aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeffff/",
status=200,
body=json.dumps(payload),
content_type="application/json",
)
@pytest.fixture
def mock_user_api(responses):
"""
Provide a mock API response to retrieve user details
Signup is disabled in this mock
"""
payload = {
"id": 1,
"email": "bot@teklia.com",
"display_name": "Bender",
"features": {
"signup": False,
},
}
responses.add(
responses.GET,
"http://testserver/api/v1/user/",
"http://testserver/api/v1/imports/workers/56785678-5678-5678-5678-567856785678/",
status=200,
body=json.dumps(payload),
content_type="application/json",
......@@ -259,7 +203,7 @@ def mock_activity_calls(responses):
@pytest.fixture
def mock_elements_worker(monkeypatch, mock_config_api):
def mock_elements_worker(monkeypatch, mock_worker_run_api):
"""Build and configure an ElementsWorker with fixed CLI parameters to avoid issues with pytest"""
monkeypatch.setattr(sys, "argv", ["worker"])
......@@ -290,7 +234,7 @@ def mock_elements_worker_with_list(monkeypatch, responses, mock_elements_worker)
@pytest.fixture
def mock_base_worker_with_cache(mocker, monkeypatch, mock_config_api):
def mock_base_worker_with_cache(mocker, monkeypatch, mock_worker_run_api):
"""Build a BaseWorker using SQLite cache, also mocking a PONOS_TASK"""
monkeypatch.setattr(sys, "argv", ["worker"])
......@@ -300,7 +244,7 @@ def mock_base_worker_with_cache(mocker, monkeypatch, mock_config_api):
@pytest.fixture
def mock_elements_worker_with_cache(monkeypatch, mock_config_api, tmp_path):
def mock_elements_worker_with_cache(monkeypatch, mock_worker_run_api, tmp_path):
"""Build and configure an ElementsWorker using SQLite cache with fixed CLI parameters to avoid issues with pytest"""
cache_path = tmp_path / "db.sqlite"
init_cache_db(cache_path)
......@@ -309,6 +253,7 @@ def mock_elements_worker_with_cache(monkeypatch, mock_config_api, tmp_path):
worker = ElementsWorker(support_cache=True)
worker.configure()
worker.configure_cache()
return worker
......
tests/test_base_worker.py
View file @ d278157b
# -*- coding: utf-8 -*-
import json
import logging
import os
import sys
......@@ -45,29 +46,40 @@ def test_init_var_ponos_data_given(monkeypatch):
assert worker.worker_version_id == "12341234-1234-1234-1234-123412341234"
def test_init_var_worker_version_id_missing(
monkeypatch, mock_user_api, mock_process_api
):
def test_init_var_worker_version_id_missing(monkeypatch):
monkeypatch.setattr(sys, "argv", ["worker"])
monkeypatch.delenv("WORKER_VERSION_ID")
monkeypatch.delenv("ARKINDEX_WORKER_RUN_ID")
worker = BaseWorker()
worker.configure()
worker.args = worker.parser.parse_args()
worker.configure_for_developers()
assert worker.worker_version_id is None
assert worker.is_read_only is True
assert worker.config == {} # default empty case
def test_init_var_worker_local_file(
monkeypatch, tmp_path, mock_user_api, mock_process_api
):
def test_init_var_worker_run_id_missing(monkeypatch):
monkeypatch.setattr(sys, "argv", ["worker"])
monkeypatch.delenv("ARKINDEX_WORKER_RUN_ID")
worker = BaseWorker()
worker.args = worker.parser.parse_args()
worker.configure_for_developers()
assert worker.worker_run_id is None
assert worker.is_read_only is True
assert worker.config == {} # default empty case
def test_init_var_worker_local_file(monkeypatch, tmp_path):
# Build a dummy yaml config file
config = tmp_path / "config.yml"
config.write_text("---\nlocalKey: abcdef123")
monkeypatch.setattr(sys, "argv", ["worker", "-c", str(config)])
monkeypatch.delenv("WORKER_VERSION_ID")
monkeypatch.delenv("ARKINDEX_WORKER_RUN_ID")
worker = BaseWorker()
worker.configure()
worker.args = worker.parser.parse_args()
worker.configure_for_developers()
assert worker.worker_version_id is None
assert worker.is_read_only is True
assert worker.config == {"localKey": "abcdef123"} # Use a local file for devs
......@@ -75,86 +87,131 @@ def test_init_var_worker_local_file(
config.unlink()
def test_cli_default(mocker, mock_config_api):
def test_cli_default(mocker, mock_worker_run_api):
worker = BaseWorker()
spy = mocker.spy(worker, "add_arguments")
assert not spy.called
assert logger.level == logging.NOTSET
assert not hasattr(worker, "api_client")
mocker.patch.object(sys, "argv", ["worker"])
worker.configure()
worker.args = worker.parser.parse_args()
assert worker.is_read_only is False
assert worker.worker_version_id == "12341234-1234-1234-1234-123412341234"
assert worker.worker_run_id == "56785678-5678-5678-5678-567856785678"
assert spy.called
assert spy.call_count == 1
worker.configure()
assert not worker.args.verbose
assert logger.level == logging.NOTSET
assert worker.api_client
assert worker.worker_version_id == "12341234-1234-1234-1234-123412341234"
assert worker.is_read_only is False
assert worker.config == {"someKey": "someValue"} # from API
logger.setLevel(logging.NOTSET)
def test_cli_arg_verbose_given(mocker, mock_config_api):
def test_cli_arg_verbose_given(mocker, mock_worker_run_api):
worker = BaseWorker()
spy = mocker.spy(worker, "add_arguments")
assert not spy.called
assert logger.level == logging.NOTSET
assert not hasattr(worker, "api_client")
mocker.patch.object(sys, "argv", ["worker", "-v"])
worker.configure()
worker.args = worker.parser.parse_args()
assert worker.is_read_only is False
assert worker.worker_version_id == "12341234-1234-1234-1234-123412341234"
assert worker.worker_run_id == "56785678-5678-5678-5678-567856785678"
assert spy.called
assert spy.call_count == 1
worker.configure()
assert worker.args.verbose
assert logger.level == logging.DEBUG
assert worker.api_client
assert worker.worker_version_id == "12341234-1234-1234-1234-123412341234"
assert worker.config == {"someKey": "someValue"} # from API
logger.setLevel(logging.NOTSET)
def test_cli_envvar_debug_given(mocker, monkeypatch, mock_worker_run_api):
worker = BaseWorker()
assert logger.level == logging.NOTSET
assert not hasattr(worker, "api_client")
mocker.patch.object(sys, "argv", ["worker"])
monkeypatch.setenv("ARKINDEX_DEBUG", True)
worker.args = worker.parser.parse_args()
assert worker.is_read_only is False
assert worker.worker_version_id == "12341234-1234-1234-1234-123412341234"
assert worker.worker_run_id == "56785678-5678-5678-5678-567856785678"
worker.configure()
assert logger.level == logging.DEBUG
assert worker.api_client
assert worker.config == {"someKey": "someValue"} # from API
logger.setLevel(logging.NOTSET)
def test_configure_dev_mode(
mocker, monkeypatch, mock_user_api, mock_worker_version_api
):
def test_configure_dev_mode(mocker, monkeypatch):
"""
Configuring a worker in developer mode avoid retrieving process information
"""
worker = BaseWorker()
mocker.patch.object(sys, "argv", ["worker", "--dev"])
monkeypatch.setenv(
"ARKINDEX_WORKER_RUN_ID", "aaaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"
)
worker.configure()
worker.args = worker.parser.parse_args()
worker.configure_for_developers()
assert worker.args.dev is True
assert worker.process_information is None
assert worker.worker_version_id == "12341234-1234-1234-1234-123412341234"
assert worker.worker_run_id == "56785678-5678-5678-5678-567856785678"
assert worker.is_read_only is True
assert worker.user_configuration == {}
def test_configure_worker_run(mocker, monkeypatch, responses, mock_config_api):
def test_configure_worker_run(mocker, monkeypatch, responses):
worker = BaseWorker()
mocker.patch.object(sys, "argv", ["worker"])
run_id = "aaaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"
configuration_id = "bbbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb"
monkeypatch.setenv("ARKINDEX_WORKER_RUN_ID", run_id)
responses.add(
responses.GET,
f"http://testserver/api/v1/imports/workers/{run_id}/",
json={"id": run_id, "configuration_id": configuration_id},
)
user_configuration = {
"id": "bbbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb",
"name": "BBB",
"configuration": {"a": "b"},
}
payload = {
"id": "56785678-5678-5678-5678-567856785678",
"parents": [],
"worker_version_id": "12341234-1234-1234-1234-123412341234",
"model_version_id": None,
"dataimport_id": "aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeffff",
"worker": {
"id": "deadbeef-1234-5678-1234-worker",
"name": "Fake worker",
"slug": "fake_worker",
"type": "classifier",
},
"configuration_id": user_configuration["id"],
"worker_version": {
"id": "12341234-1234-1234-1234-123412341234",
"worker": {
"id": "deadbeef-1234-5678-1234-worker",
"name": "Fake worker",
"slug": "fake_worker",
"type": "classifier",
},
"revision": {"hash": "deadbeef1234"},
"configuration": {"configuration": {}},
},
"configuration": user_configuration,
"process": {"id": "aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeffff"},
}
responses.add(
responses.GET,
f"http://testserver/api/v1/workers/configurations/{configuration_id}/",
json={"id": configuration_id, "name": "BBB", "configuration": {"a": "b"}},
"http://testserver/api/v1/imports/workers/56785678-5678-5678-5678-567856785678/",
status=200,
body=json.dumps(payload),
content_type="application/json",
)
worker.args = worker.parser.parse_args()
assert worker.is_read_only is False
assert worker.worker_version_id == "12341234-1234-1234-1234-123412341234"
assert worker.worker_run_id == "56785678-5678-5678-5678-567856785678"
worker.configure()
assert worker.user_configuration == {"a": "b"}
......@@ -163,20 +220,56 @@ def test_configure_worker_run(mocker, monkeypatch, responses, mock_config_api):
def test_configure_user_configuration_defaults(
mocker,
monkeypatch,
mock_worker_version_user_configuration_api,
mock_user_api,
mock_process_api,
responses,
):
worker = BaseWorker()
mocker.patch.object(sys, "argv")
run_id = "aaaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"
monkeypatch.setenv("ARKINDEX_WORKER_RUN_ID", run_id)
worker.args = worker.parser.parse_args()
payload = {
"id": "56785678-5678-5678-5678-567856785678",
"parents": [],
"worker_version_id": "12341234-1234-1234-1234-123412341234",
"model_version_id": None,
"dataimport_id": "aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeffff",
"worker": {
"id": "deadbeef-1234-5678-1234-worker",
"name": "Fake worker",
"slug": "fake_worker",
"type": "classifier",
},
"configuration_id": "af0daaf4-983e-4703-a7ed-a10f146d6684",
"worker_version": {
"id": "12341234-1234-1234-1234-123412341234",
"worker": {
"id": "deadbeef-1234-5678-1234-worker",
"name": "Fake worker",
"slug": "fake_worker",
"type": "classifier",
},
"revision": {"hash": "deadbeef1234"},
"configuration": {
"configuration": {"param_1": "/some/path/file.pth", "param_2": 12}
},
},
"configuration": {
"id": "af0daaf4-983e-4703-a7ed-a10f146d6684",
"name": "my-userconfig",
"configuration": {
"param_3": "Animula vagula blandula",
"param_5": True,
},
},
"process": {"id": "aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeffff"},
}
responses.add(
responses.GET,
f"http://testserver/api/v1/imports/workers/{run_id}/",
json={"id": run_id},
"http://testserver/api/v1/imports/workers/56785678-5678-5678-5678-567856785678/",
status=200,
body=json.dumps(payload),
content_type="application/json",
)
worker.configure()
assert worker.config == {"param_1": "/some/path/file.pth", "param_2": 12}
......@@ -186,24 +279,97 @@ def test_configure_user_configuration_defaults(
}
def test_configure_worker_run_missing_conf(
mocker, monkeypatch, responses, mock_config_api
):
@pytest.mark.parametrize("debug", (True, False))
def test_configure_user_config_debug(mocker, monkeypatch, responses, debug):
worker = BaseWorker()
mocker.patch.object(sys, "argv", ["worker"])
run_id = "aaaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"
configuration_id = "bbbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb"
monkeypatch.setenv("ARKINDEX_WORKER_RUN_ID", run_id)
assert logger.level == logging.NOTSET
payload = {
"id": "56785678-5678-5678-5678-567856785678",
"parents": [],
"worker_version_id": "12341234-1234-1234-1234-123412341234",
"model_version_id": None,
"dataimport_id": "aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeffff",
"worker": {
"id": "deadbeef-1234-5678-1234-worker",
"name": "Fake worker",
"slug": "fake_worker",
"type": "classifier",
},
"configuration_id": "af0daaf4-983e-4703-a7ed-a10f146d6684",
"worker_version": {
"id": "12341234-1234-1234-1234-123412341234",
"worker": {
"id": "deadbeef-1234-5678-1234-worker",
"name": "Fake worker",
"slug": "fake_worker",
"type": "classifier",
},
"revision": {"hash": "deadbeef1234"},
"configuration": {"configuration": {}},
},
"configuration": {
"id": "af0daaf4-983e-4703-a7ed-a10f146d6684",
"name": "BBB",
"configuration": {"debug": debug},
},
"process": {"id": "aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeffff"},
}
responses.add(
responses.GET,
f"http://testserver/api/v1/imports/workers/{run_id}/",
json={"id": run_id, "configuration_id": configuration_id},
"http://testserver/api/v1/imports/workers/56785678-5678-5678-5678-567856785678/",
status=200,
body=json.dumps(payload),
content_type="application/json",
)
worker.args = worker.parser.parse_args()
worker.configure()
assert worker.user_configuration == {"debug": debug}
expected_log_level = logging.DEBUG if debug else logging.NOTSET
assert logger.level == expected_log_level
logger.setLevel(logging.NOTSET)
def test_configure_worker_run_missing_conf(mocker, monkeypatch, responses):
worker = BaseWorker()
mocker.patch.object(sys, "argv", ["worker"])
payload = {
"id": "56785678-5678-5678-5678-567856785678",
"parents": [],
"worker_version_id": "12341234-1234-1234-1234-123412341234",
"model_version_id": None,
"dataimport_id": "aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeffff",
"worker": {
"id": "deadbeef-1234-5678-1234-worker",
"name": "Fake worker",
"slug": "fake_worker",
"type": "classifier",
},
"configuration_id": "bbbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb",
"worker_version": {
"id": "12341234-1234-1234-1234-123412341234",
"worker": {
"id": "deadbeef-1234-5678-1234-worker",
"name": "Fake worker",
"slug": "fake_worker",
"type": "classifier",
},
"revision": {"hash": "deadbeef1234"},
"configuration": {"configuration": {}},
},
"configuration": {"id": "bbbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb", "name": "BBB"},
"process": {"id": "aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeffff"},
}
responses.add(
responses.GET,
f"http://testserver/api/v1/workers/configurations/{configuration_id}/",
json={"id": configuration_id, "name": "BBB"},
"http://testserver/api/v1/imports/workers/56785678-5678-5678-5678-567856785678/",
status=200,
body=json.dumps(payload),
content_type="application/json",
)
worker.args = worker.parser.parse_args()
worker.configure()
assert worker.user_configuration is None
......
tests/test_elements_worker/__init__.py
View file @ d278157b
# -*- coding: utf-8 -*-
# API calls during worker configuration
BASE_API_CALLS = [
("GET", "http://testserver/api/v1/user/"),
("GET", "http://testserver/api/v1/imports/aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeffff/"),
(
"GET",
"http://testserver/api/v1/workers/versions/12341234-1234-1234-1234-123412341234/",
"http://testserver/api/v1/imports/workers/56785678-5678-5678-5678-567856785678/",
),
]
tests/test_elements_worker/test_cli.py
View file @ d278157b
......@@ -10,7 +10,7 @@ import pytest
from arkindex_worker.worker import ElementsWorker
def test_cli_default(monkeypatch, mock_config_api):
def test_cli_default(monkeypatch, mock_worker_run_api):
_, path = tempfile.mkstemp()
with open(path, "w") as f:
json.dump(
......@@ -33,7 +33,7 @@ def test_cli_default(monkeypatch, mock_config_api):
os.unlink(path)
def test_cli_arg_elements_list_given(mocker, mock_config_api):
def test_cli_arg_elements_list_given(mocker, mock_worker_run_api):
_, path = tempfile.mkstemp()
with open(path, "w") as f:
json.dump(
......@@ -62,7 +62,7 @@ def test_cli_arg_element_one_given_not_uuid(mocker):
worker.configure()
def test_cli_arg_element_one_given(mocker, mock_config_api):
def test_cli_arg_element_one_given(mocker, mock_worker_run_api):
mocker.patch.object(
sys, "argv", ["worker", "--element", "12341234-1234-1234-1234-123412341234"]
)
......@@ -74,7 +74,7 @@ def test_cli_arg_element_one_given(mocker, mock_config_api):
assert not worker.args.elements_list
def test_cli_arg_element_many_given(mocker, mock_config_api):
def test_cli_arg_element_many_given(mocker, mock_worker_run_api):
mocker.patch.object(
sys,
"argv",
......
tests/test_elements_worker/test_worker.py
View file @ d278157b
......@@ -79,17 +79,8 @@ def test_readonly(responses, mock_elements_worker):
] == BASE_API_CALLS
def test_activities_disabled(
responses, monkeypatch, mock_worker_version_api, mock_user_api
):
def test_activities_disabled(responses, monkeypatch, mock_worker_run_api):
"""Test worker process elements without updating activities when they are disabled for the process"""
responses.add(
responses.GET,
"http://testserver/api/v1/imports/aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeffff/",
status=200,
body=json.dumps({"activity_state": "disabled"}),
content_type="application/json",
)
monkeypatch.setattr(sys, "argv", ["worker"])
worker = ElementsWorker()
worker.configure()
......@@ -101,7 +92,7 @@ def test_activities_disabled(
] == BASE_API_CALLS
def test_activities_dev_mode(mocker, mock_user_api, mock_worker_version_api):
def test_activities_dev_mode(mocker):
"""
Worker activities are not stored in dev mode
"""
......@@ -115,7 +106,7 @@ def test_activities_dev_mode(mocker, mock_user_api, mock_worker_version_api):
assert worker.store_activity is False
def test_update_call(responses, mock_elements_worker, mock_process_api):
def test_update_call(responses, mock_elements_worker, mock_worker_run_api):
"""Test an update call with feature enabled triggers an API call"""
responses.add(
responses.PUT,
......
tests/test_merge.py
View file @ d278157b
......@@ -181,7 +181,9 @@ def test_merge_from_worker(
monkeypatch.setenv("PONOS_DATA", str(tmpdir))
# Create the task's output dir, so that it can create its own database
(tmpdir / "my_task").mkdir()
mock_base_worker_with_cache.args = mock_base_worker_with_cache.parser.parse_args()
mock_base_worker_with_cache.configure()
mock_base_worker_with_cache.configure_cache()
# Then we have 2 elements and a transcription
assert CachedImage.select().count() == 0
......