def test_exception_invalid_yaml():
cur_dir = os.path.dirname(os.path.abspath(__file__))
yaml = os.path.join(cur_dir, 'invalid.yml')
with pytest.raises(BadConfigSource):
BaseExecutor.load_config(yaml)
with pytest.raises(BadConfigSource):
Flow.load_config(yaml)
def test_runtime_args():
b = BaseExecutor.load_config('BaseExecutor',
metas={'name': 'b123'},
runtime_args={'hello': 'world'})
assert b.runtime_args.hello == 'world'
assert b.metas.name == 'b123'
def test_import_with_new_module_structure_should_pass():
"""
This is a valid and **RECOMMENDED** structure:
- python code for the executor organized in a package structure inside
the ``executor/`` folder
- core logic in ``executor/my_executor.py``
- the ``executor/__init__.py`` contains
``from .my_executor import GoodCrafterNew``, which makes sure the
custom executor class gets registered
- all imports are relative - so in ``executor/my_executor.py`` the ``helper``
module is imported as ``from .helper import foo``
File structure:
my_cust_module/
|- executor/
|- __init__.py
|- my_executor.py
|- helper.py
|- config.yml
|- py_modules
|- executor/__init__.py
"""
b = BaseExecutor.load_config('good_new/crafter.yml')
assert b.__class__.__name__ == 'GoodCrafterNew'
def test_import_with_old_module_structure_should_pass():
"""
This is a valid structure, but not recommended:
- "my_cust_module" is a python module
- all core logic of your customized executor goes to ``__init__.py``
- to import ``foo.py``, you should use relative import, e.g. ``from .foo import bar``
This is not a recommended structure because:
- putting core logic inside ``__init__.py`` is not how python packages
are usually written
- Importing from the workspace disables you from trying out the executor in
the console, or test files at the root of the workspace, making development
more cumbersome
- the main directory is now cluttered with python files
- extracting all python files to a separate directory is how python packages
are usually composed
File structure:
my_cust_module
|- __init__.py
|- helper.py
|- config.yml
|- py_modules
|- __init__.py
"""
b = BaseExecutor.load_config('good_old/crafter.yml')
assert b.__class__.__name__ == 'GoodCrafterOld'
def test_load_from_dict():
# !BaseEncoder
# metas:
# name: ${{BE_TEST_NAME}}
# batch_size: ${{BATCH_SIZE}}
# pod_id: ${{pod_id}}
# workspace: ${{this.name}}-${{this.batch_size}}
d1 = {
'jtype': __default_executor__,
'metas': {
'name': '${{ BE_TEST_NAME }}',
'workspace': '${{this.name}}',
},
}
# !CompoundExecutor
# components:
# - !BinaryPbIndexer
# with:
# index_filename: tmp1
# metas:
# name: test1
# - !BinaryPbIndexer
# with:
# index_filename: tmp2
# metas:
# name: test2
# metas:
# name: compound1
d = {'BE_TEST_NAME': 'hello123'}
b1 = BaseExecutor.load_config(d1, context=d)
assert isinstance(b1, BaseExecutor)
assert b1.metas.name == 'hello123'
def test_default_args_from_load_config():
b = BaseExecutor.load_config('!BaseExecutor {}')
assert isinstance(b.runtime_args, SimpleNamespace)
assert isinstance(b.metas, SimpleNamespace)
# name is always auto-assigned
assert b.metas.name
def test_default_args_from_python():
b = BaseExecutor()
assert isinstance(b.runtime_args, SimpleNamespace)
assert isinstance(b.metas, SimpleNamespace)
# name is always auto-assigned
assert b.metas.name
def _load_executor(self):
"""Load the executor to this runtime, specified by ``uses`` CLI argument."""
try:
self._executor: BaseExecutor = BaseExecutor.load_config(
self.args.uses,
uses_with=self.args.uses_with,
uses_metas=self.args.uses_metas,
uses_requests=self.args.uses_requests,
runtime_args={
'workspace': self.args.workspace,
'shard_id': self.args.shard_id,
'shards': self.args.shards,
'replicas': self.args.replicas,
'name': self.args.name,
'py_modules': self.args.py_modules,
},
extra_search_paths=self.args.extra_search_paths,
)
except BadConfigSource as ex:
self.logger.error(
f'fail to load config from {self.args.uses}, if you are using docker image for --uses, '
f'please use "docker://YOUR_IMAGE_NAME"')
raise ExecutorFailToLoad from ex
except FileNotFoundError as ex:
self.logger.error(f'fail to load file dependency')
raise ExecutorFailToLoad from ex
except Exception as ex:
self.logger.critical(
f'can not load the executor from {self.args.uses}')
raise ExecutorFailToLoad from ex
def test_runtime_args_not_serialisable():
param = NotSerialisable()
b = BaseExecutor.load_config(
'BaseExecutor',
runtime_args={'hello': 'world', 'not_seri': param},
)
assert b.runtime_args.hello == 'world'
assert b.runtime_args.not_seri is param
def test_jtype(tmpdir):
flow_path = os.path.join(tmpdir, 'flow.yml')
exec_path = os.path.join(tmpdir, 'exec.yml')
f = Flow()
f.save_config(flow_path)
with open(flow_path, 'r') as file:
conf = yaml.safe_load(file)
assert 'jtype' in conf
assert conf['jtype'] == 'Flow'
e = BaseExecutor()
e.save_config(exec_path)
with open(exec_path, 'r') as file:
conf = yaml.safe_load(file)
assert 'jtype' in conf
assert conf['jtype'] == 'BaseExecutor'
assert type(BaseExecutor.load_config(exec_path)) == BaseExecutor
assert type(Flow.load_config(flow_path)) == Flow
def test_runtime_args_from_load_config():
y = '''
!BaseExecutor
metas:
name: my-mwu-encoder
workspace: ./
'''
b = BaseExecutor.load_config(y)
assert b.metas.workspace == './'
assert b.metas.name == 'my-mwu-encoder'
def test_load_dataclass_executor():
executor_yaml = '''
jtype: MyDataClassExecutor
with:
my_field: this is my field
metas:
name: test-name-updated
workspace: test-work-space-updated
requests:
/foo: baz
'''
exec = BaseExecutor.load_config(executor_yaml)
assert exec.my_field == 'this is my field'
assert exec.requests['/foo'] == MyDataClassExecutor.baz
assert exec.metas.name == 'test-name-updated'
assert exec.metas.workspace == 'test-work-space-updated'
def _load_executor(self, metrics_registry: Optional['CollectorRegistry'] = None):
"""
Load the executor to this runtime, specified by ``uses`` CLI argument.
:param metrics_registry: Optional prometheus metrics registry that will be passed to the executor so that it can expose metrics
"""
try:
self._executor: BaseExecutor = BaseExecutor.load_config(
self.args.uses,
uses_with=self.args.uses_with,
uses_metas=self.args.uses_metas,
uses_requests=self.args.uses_requests,
runtime_args={ # these are not parsed to the yaml config file but are pass directly during init
'workspace': self.args.workspace,
'shard_id': self.args.shard_id,
'shards': self.args.shards,
'replicas': self.args.replicas,
'name': self.args.name,
'metrics_registry': metrics_registry,
},
py_modules=self.args.py_modules,
extra_search_paths=self.args.extra_search_paths,
)
self.logger.debug(f'{self._executor} is successfully loaded!')
except BadConfigSource:
self.logger.error(
f'fail to load config from {self.args.uses}, if you are using docker image for --uses, '
f'please use `docker://YOUR_IMAGE_NAME`'
)
raise
except FileNotFoundError:
self.logger.error(f'fail to load file dependency')
raise
except Exception:
self.logger.critical(f'can not load the executor from {self.args.uses}')
raise
def test_encoder_name_env_replace():
os.environ['BE_TEST_NAME'] = 'hello123'
with BaseExecutor.load_config('yaml/test-encoder-env.yml') as be:
assert be.metas.name == 'hello123'
def test_exec_from_python():
be = BaseExecutor(metas={'name': 'hello', 'random_name': 'random_value'})
assert be.metas.name == 'hello'
assert be.metas.random_name == 'random_value'
def test_encoder_name_dict_replace():
d = {'BE_TEST_NAME': 'hello123'}
with BaseExecutor.load_config('yaml/test-encoder-env.yml',
context=d) as be:
assert be.metas.name == 'hello123'
assert be.metas.workspace == 'hello123'
def test_encoder_inject_config_via_kwargs():
with BaseExecutor.load_config('yaml/test-encoder-env.yml',
metas={'shard_id': 345}) as be:
assert be.metas.shard_id == 345
def test_import_casual_structure_should_fail():
# this structure is a copy-paste from
# https://github.com/jina-ai/jina/issues/1546#issuecomment-751481422
with pytest.raises(ImportError):
BaseExecutor.load_config('bad1/crafter.yml')
def test_load_external_success():
with BaseExecutor.load_config('yaml/dummy_ext_exec_success.yml') as e:
assert e.__class__.__name__ == 'DummyExternalIndexer'
def test_use_from_local_dir_exe_level():
with BaseExecutor.load_config('dummyhub/config.yml'):
pass
def test_load_external_fail():
with pytest.raises(yaml.constructor.ConstructorError):
BaseExecutor.load_config('yaml/dummy_ext_exec.yml')
