def test_initialization(spec, expected, tmp_sample_tasks, tmp_imports):
meta = Meta.default_meta({
'extract_product': False,
'extract_upstream': True
})
spec = TaskSpec(spec, meta=meta, project_root='.')
# check values after initialization
assert spec['class'] == expected
assert isinstance(spec['source'], Path)
# check we can convert it to a Task
spec.to_task(dag=DAG())
def test_error_if_client_dotted_path_returns_none(tmp_sample_tasks,
add_current_to_sys_path,
no_sys_modules_cache, key):
Path('client_dotted_path_returns_none.py').write_text("""
def get():
return None
""")
meta = Meta.default_meta({
'extract_product': False,
'extract_upstream': True,
})
dag = DAG()
spec = {
'source': 'sample.sql',
'product': ['name', 'table'],
}
spec[key] = 'client_dotted_path_returns_none.get'
with pytest.raises(TypeError) as excinfo:
TaskSpec(spec, meta=meta, project_root='.').to_task(dag=dag)
assert (
"Error calling dotted path "
"'client_dotted_path_returns_none.get'. Expected a value but got None"
) in str(excinfo.value)
def test_add_hook(tmp_directory, add_current_to_sys_path):
task = {
'product': 'notebook.ipynb',
'source': 'source.py',
'on_finish': 'hooks.some_hook',
'on_render': 'hooks.some_hook',
'on_failure': 'hooks.some_hook'
}
meta = Meta.default_meta()
meta['extract_product'] = False
Path('source.py').write_text("""
# + tags=["parameters"]
# some code
""")
Path('hooks.py').write_text("""
def some_hook():
pass
""")
dag = DAG()
t, _ = TaskSpec(task, meta, project_root='.').to_task(dag)
assert t.on_finish
assert t.on_render
assert t.on_failure
def test_validate_missing_source(key):
with pytest.raises(KeyError):
TaskSpec({key: None}, {
'extract_product': False,
'extract_upstream': False
},
project_root='.')
def test_grid_with_hook_lazy_import(backup_spec_with_functions_flat,
tmp_imports):
grid_spec = {
'source': 'my_tasks_flat.raw.function',
'name': 'function-',
'product': 'some_file.txt',
'grid': {
'a': [1, 2],
'b': [3, 4]
},
'on_render': 'hooks.on_render',
'on_finish': 'hooks.on_finish',
'on_failure': 'hooks.on_failure',
}
meta = Meta.default_meta()
dag = DAG()
TaskSpec(grid_spec, meta, project_root='.',
lazy_import=True).to_task(dag=dag)
assert all(t.on_render.callable is None for t in dag.values())
assert all(t.on_finish.callable is None for t in dag.values())
assert all(t.on_failure.callable is None for t in dag.values())
assert all(t.on_render._spec.dotted_path == 'hooks.on_render'
for t in dag.values())
assert all(t.on_finish._spec.dotted_path == 'hooks.on_finish'
for t in dag.values())
assert all(t.on_failure._spec.dotted_path == 'hooks.on_failure'
for t in dag.values())
def test_error_if_dotted_path_does_not_return_a_callable(
backup_spec_with_functions_flat, add_current_to_sys_path,
no_sys_modules_cache, key):
Path('test_error_if_dotted_path_does_not_return_a_callable.py').write_text(
"""
some_non_function = 1
""")
meta = Meta.default_meta({'extract_product': False})
spec = {
'source': 'my_tasks_flat.raw.function',
'product': 'some_file.txt',
}
spec[key] = ('test_error_if_dotted_path_does_not_return_a_callable'
'.some_non_function')
with pytest.raises(TypeError) as excinfo:
TaskSpec(spec, meta=meta, project_root='.').to_task(dag=DAG())
expected = ("Error loading dotted path 'test_error_if_dotted_path"
"_does_not_return_a_callable.some_non_function'. Expected a "
"callable object (i.e., some kind of function). Got "
"1 (an object of type: int)")
assert str(excinfo.value) == expected
def test_error_on_invalid_value_for_file_product(backup_online, tmp_imports):
meta = Meta.default_meta()
meta['extract_product'] = False
spec = TaskSpec({
'source': 'online_tasks.square',
'product': 1,
},
meta=meta,
project_root='.')
with pytest.raises(TypeError) as excinfo:
spec.to_task(dag=DAG())
expected = ('Error initializing File with argument 1 '
'(expected str, bytes or os.PathLike object, not int)')
assert expected == str(excinfo.value)
def test_grid_with_missing_name(backup_spec_with_functions_flat,
add_current_to_sys_path, spec):
del spec['name']
with pytest.raises(KeyError) as excinfo:
TaskSpec(spec, Meta.default_meta(),
project_root='.').to_task(dag=DAG())
assert 'Error initializing task with spec' in str(excinfo.value)
def test_grid_and_params(backup_spec_with_functions_flat, tmp_imports,
grid_spec):
grid_spec['params'] = {'a': 1}
with pytest.raises(DAGSpecInitializationError) as excinfo:
TaskSpec(grid_spec, Meta.default_meta(),
project_root='.').to_task(dag=DAG())
assert "'params' is not allowed when using 'grid'" in str(excinfo.value)
def test_grid_with_missing_name(backup_spec_with_functions_flat, tmp_imports,
grid_spec):
del grid_spec['name']
with pytest.raises(DAGSpecInitializationError) as excinfo:
TaskSpec(grid_spec, Meta.default_meta(),
project_root='.').to_task(dag=DAG())
assert 'Error initializing task with source' in str(excinfo.value)
def test_error_when_failing_to_init(spec, tmp_sample_tasks, tmp_imports):
meta = Meta.default_meta({
'extract_product': False,
'extract_upstream': True
})
dag = DAG()
with pytest.raises(DAGSpecInitializationError) as excinfo:
TaskSpec(spec, meta=meta, project_root='.').to_task(dag=dag)
assert 'Error initializing SQLRelation' in str(excinfo.value)
def test_loads_serializer_and_unserializer(backup_online, tmp_imports):
meta = Meta.default_meta()
meta['extract_product'] = False
spec = TaskSpec(
{
'source': 'online_tasks.square',
'product': 'output/square.parquet',
'serializer': 'online_io.serialize',
'unserializer': 'online_io.unserialize',
},
meta=meta,
project_root='.')
dag = DAG()
task, _ = spec.to_task(dag=dag)
from online_io import serialize, unserialize
assert task._serializer.callable is serialize
assert task._unserializer.callable is unserialize
def test_lazy_load(tmp_directory, tmp_imports):
Path('my_module.py').write_text("""
def fn():
pass
""")
meta = Meta.default_meta()
spec = TaskSpec(
{
'source': 'my_module.fn',
'product': 'report.ipynb',
'on_finish': 'not_a_module.not_a_function',
'on_render': 'not_a_module.not_a_function',
'on_failure': 'not_a_module.not_a_function',
'serializer': 'not_a_module.not_a_function',
'unserializer': 'not_a_module.not_a_function',
},
meta,
'.',
lazy_import=True)
assert spec.to_task(dag=DAG())
def test_grid(backup_spec_with_functions_flat, add_current_to_sys_path, spec):
meta = Meta.default_meta()
dag = DAG()
task_group, _ = TaskSpec(spec, meta, project_root='.').to_task(dag=dag)
assert len(task_group) == 4
assert str(dag['function-0'].product) == str(
Path('some_file-0.txt').resolve())
assert str(dag['function-1'].product) == str(
Path('some_file-1.txt').resolve())
assert str(dag['function-2'].product) == str(
Path('some_file-2.txt').resolve())
assert str(dag['function-3'].product) == str(
Path('some_file-3.txt').resolve())
def test_error_on_invalid_class(backup_spec_with_functions_flat, tmp_imports):
meta = Meta.default_meta({'extract_product': False})
spec = {
'source': 'my_tasks_flat.raw.function',
'product': 'some_file.txt',
'class': 'unknown_class'
}
with pytest.raises(ValueError) as excinfo:
TaskSpec(spec, meta=meta, project_root='.').to_task(dag=DAG())
expected = ("Error validating Task spec (class field): "
"'unknown_class' is not a valid Task class name")
assert str(excinfo.value) == expected
def test_constructor_deep_copies_spec_and_meta(tmp_directory, tmp_imports):
prod_default_class = {'SQLScript': 'SQLRelation'}
meta = Meta.default_meta({
'extract_product': False,
'product_default_class': prod_default_class
})
params = {'params': {'a': 1}}
spec = {
'source': 'sample.sql',
'product': 'some_file.txt',
'params': params
}
task_spec = TaskSpec(data=spec, meta=meta, project_root='.')
assert spec is not task_spec.data
assert meta is not task_spec.meta
assert params is not task_spec.data['params']
assert prod_default_class is not task_spec.meta['product_default_class']
def test_skips_source_loader_if_absolute_path(tmp_sample_tasks, tmp_imports):
Path('templates').mkdir()
meta = Meta.default_meta({
'extract_product': False,
'extract_upstream': True,
'source_loader': {
'path': 'templates'
}
})
dag = DAG()
spec = {
'source': str(Path(tmp_sample_tasks, 'sample.sql')),
'product': ['name', 'table'],
'client': 'db.get_client'
}
assert TaskSpec(spec, meta=meta, project_root='.').to_task(dag=dag)
def _init(self, data, env, lazy_import, reload, parent_path,
look_up_project_root_recursively):
self._lazy_import = lazy_import
# initialized with a path to a yaml file...
if isinstance(data, (str, Path)):
# TODO: test this
if parent_path is not None:
raise ValueError('parent_path must be None when '
f'initializing {type(self).__name__} with '
'a path to a YAML spec')
# resolve the parent path to make sources and products unambiguous
# even if the current working directory changes
self._path = Path(data).resolve()
self._parent_path = str(self._path.parent)
if not Path(data).is_file():
raise FileNotFoundError(
'Error initializing DAGSpec with argument '
f'{data!r}: Expected it to be a path to a YAML file, but '
'such file does not exist')
content = Path(data).read_text()
try:
data = yaml.safe_load(content)
except (yaml.parser.ParserError,
yaml.constructor.ConstructorError) as e:
error = e
else:
error = None
if error:
if '{{' in content or '}}' in content:
raise DAGSpecInitializationError(
'Failed to initialize spec. It looks like '
'you\'re using placeholders (i.e. {{placeholder}}). '
'Make sure values are enclosed in parentheses '
'(e.g. key: "{{placeholder}}"). Original '
'parser error:\n\n'
f'{error}')
else:
raise error
# initialized with a dictionary...
else:
self._path = None
# FIXME: add test cases, some of those features wont work if
# _parent_path is None. We should make sure that we either raise
# an error if _parent_path is needed or use the current working
# directory if it's appropriate - this is mostly to make relative
# paths consistent: they should be relative to the file that
# contains them
self._parent_path = (None if not parent_path else str(
Path(parent_path).resolve()))
self.data = data
if isinstance(self.data, list):
self.data = {'tasks': self.data}
# validate keys defined at the top (nested keys are not validated here)
self._validate_top_keys(self.data, self._path)
logger.debug('DAGSpec enviroment:\n%s', pp.pformat(env))
env = env or dict()
path_to_defaults = default.path_to_env_from_spec(
path_to_spec=self._path)
if path_to_defaults:
defaults = yaml.safe_load(Path(path_to_defaults).read_text())
self.env = EnvDict(env,
path_to_here=self._parent_path,
defaults=defaults)
else:
self.env = EnvDict(env, path_to_here=self._parent_path)
self.data, tags = expand_raw_dictionary_and_extract_tags(
self.data, self.env)
logger.debug('Expanded DAGSpec:\n%s', pp.pformat(data))
# if there is a "location" top key, we don't have to do anything else
# as we will just load the dotted path when .to_dag() is called
if 'location' not in self.data:
Meta.initialize_inplace(self.data)
import_tasks_from = self.data['meta']['import_tasks_from']
if import_tasks_from is not None:
# when using a relative path in "import_tasks_from", we must
# make it absolute...
if not Path(import_tasks_from).is_absolute():
# use _parent_path if there is one
if self._parent_path:
self.data['meta']['import_tasks_from'] = str(
Path(self._parent_path, import_tasks_from))
# otherwise just make it absolute
else:
self.data['meta']['import_tasks_from'] = str(
Path(import_tasks_from).resolve())
imported = yaml.safe_load(
Path(self.data['meta']['import_tasks_from']).read_text())
if self.env is not None:
(imported,
tags_other) = expand_raw_dictionaries_and_extract_tags(
imported, self.env)
tags = tags | tags_other
# relative paths here are relative to the file where they
# are declared
base_path = Path(self.data['meta']['import_tasks_from']).parent
for task in imported:
add_base_path_to_source_if_relative(task,
base_path=base_path)
self.data['tasks'].extend(imported)
# check if there are any params declared in env, not used in
# in the pipeline
extra = set(self.env) - self.env.default_keys - tags
if extra:
warnings.warn('The following placeholders are declared in the '
'environment but '
f'unused in the spec: {extra}')
self.data['tasks'] = [
normalize_task(task) for task in self.data['tasks']
]
# NOTE: for simple projects, project root is the parent folder
# of pipeline.yaml, for package projects is the parent folder
# of setup.py
if look_up_project_root_recursively:
project_root = (
None if not self._parent_path else
default.find_root_recursively(
starting_dir=self._parent_path,
filename=None if not self._path else self._path.name))
else:
project_root = self._parent_path
# make sure the folder where the pipeline is located is in sys.path
# otherwise dynamic imports needed by TaskSpec will fail
with add_to_sys_path(self._parent_path, chdir=False):
self.data['tasks'] = [
TaskSpec(t,
self.data['meta'],
project_root=project_root,
lazy_import=lazy_import,
reload=reload) for t in self.data['tasks']
]
else:
self.data['meta'] = Meta.empty()
def test_error_if_extract_but_keys_declared(task, meta):
with pytest.raises(DAGSpecInitializationError):
TaskSpec(task, meta, project_root='.')
def test_error_if_extract_but_keys_declared(task, meta):
with pytest.raises(ValueError):
TaskSpec(task, meta, project_root='.')
