def get_root(self):
root = default.find_root_recursively()
if root is None:
raise ValueError('Failed to expand {{root}}, could not '
'find a setup.py in a parent folder')
return root
def _default_dict(include_here):
placeholders = {
'user': '******',
'cwd': '{{cwd}}',
}
if default.find_root_recursively() is not None:
placeholders['root'] = '{{root}}'
if include_here:
placeholders['here'] = '{{here}}'
return placeholders
def __init__(self, path_to_backup_dir, path_to_project_root=None):
self._path_to_backup_dir = Path(path_to_backup_dir)
self._path_to_backup_dir.mkdir(exist_ok=True, parents=True)
if path_to_project_root:
project_root = path_to_project_root
else:
try:
project_root = find_root_recursively()
except Exception as e:
raise DAGSpecInvalidError(
f'Cannot initialize {self!r} because there '
'is not project root. Set one or explicitly pass '
'a value in the path_to_project_root argument') from e
self._path_to_project_root = Path(project_root).resolve()
def __init__(self,
bucket_name,
parent,
json_credentials_path=None,
path_to_project_root=None,
credentials_relative_to_project_root=True,
**kwargs):
if path_to_project_root:
project_root = path_to_project_root
else:
try:
project_root = find_root_recursively()
except Exception as e:
raise DAGSpecInvalidError(
f'Cannot initialize {type(self).__name__} because there '
'is not project root. Set one or explicitly pass '
'a value in the path_to_project_root argument') from e
self._path_to_project_root = Path(project_root).resolve()
if (credentials_relative_to_project_root and json_credentials_path
and not Path(json_credentials_path).is_absolute()):
json_credentials_path = Path(self._path_to_project_root,
json_credentials_path)
self._client_kwargs = kwargs
if json_credentials_path:
c = json.loads(Path(json_credentials_path).read_text())
self._client_kwargs = {
'aws_access_key_id': c['aws_access_key_id'],
'aws_secret_access_key': c['aws_secret_access_key'],
**kwargs
}
self._client = self._init_client()
self._parent = parent
self._bucket_name = bucket_name
def __init__(self,
bucket_name,
parent,
json_credentials_path=None,
path_to_project_root=None,
credentials_relative_to_project_root=True,
**kwargs):
if path_to_project_root:
project_root = path_to_project_root
else:
try:
project_root = find_root_recursively()
except Exception as e:
raise DAGSpecInvalidError(
f'Cannot initialize {type(self).__name__} because there '
'is not project root. Set one or explicitly pass '
'a value in the path_to_project_root argument') from e
self._path_to_project_root = Path(project_root).resolve()
if (credentials_relative_to_project_root and json_credentials_path
and not Path(json_credentials_path).is_absolute()):
json_credentials_path = Path(self._path_to_project_root,
json_credentials_path)
self._from_json = json_credentials_path is not None
if not self._from_json:
self._client_kwargs = kwargs
else:
self._client_kwargs = {
'json_credentials_path': json_credentials_path,
**kwargs
}
storage_client = self._init_client()
self._parent = parent
self._bucket_name = bucket_name
self._bucket = storage_client.bucket(bucket_name)
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()