def rename(self, rundate=None, tech=None, stage=None, dataset_name=None, dataset_id=None, **kwargs):
"""Rename a dataset
Renames the dataset. In particular, this means that if the dataset is written to file it will be written to a
different file (or different place in the same file). All arguments are optional. If they are not given, they
keep their existing value.
Args:
rundate: Date, the model run date.
tech: String, the technique.
stage: String, the stage.
dataset_name: String, the name of the dataset.
dataset_id: Int, id of the dataset.
"""
# Set rundate
if rundate is not None:
self.rundate = rundate
# Use existing names as default
tech = self.vars["tech"] if tech is None else tech
stage = self.vars["stage"] if stage is None else stage
dataset_name = self.dataset_name if dataset_name is None else dataset_name
if dataset_id is None:
dataset_id = self.dataset_id
else:
dataset_id = _data.parse_dataset_id(self.rundate, tech, stage, dataset_name, dataset_id)
# Update names
self.name = "{name}/{id:04d}".format(name=dataset_name, id=dataset_id)
kwargs.setdefault("session", dataset_name)
self.vars.update(dict(tech=tech, stage=stage, dataset_name=dataset_name, dataset_id=dataset_id, **kwargs))
def delete_from_file(self,
tech=None,
stage=None,
dataset_name=None,
dataset_id=None):
"""Delete this or related datasets from file
Specify arguments relative to this dataset to find datasets which will be deleted.
"""
# Use existing names as default
tech = self.vars["tech"] if tech is None else tech
stage = self.vars["stage"] if stage is None else stage
dataset_name = self.dataset_name if dataset_name is None else dataset_name
if dataset_id is None:
dataset_id = self.dataset_id
else:
dataset_id = _data.parse_dataset_id(self.rundate, tech, stage,
dataset_name, dataset_id)
dataset_id = {dataset_id} if isinstance(dataset_id,
(float,
int)) else set(dataset_id)
ids_to_delete = dataset_id & set(
_data.list_dataset_ids(self.rundate, tech, dataset_name, stage,
dataset_name))
if not ids_to_delete:
return
# Open JSON and HDF5 file and remove datasets
file_vars = dict(self.vars, tech=tech, stage=stage)
json_path = files.path("dataset_json", file_vars=file_vars)
with files.open_path(json_path, mode="rt", write_log=False) as f_json:
json_all = json.load(f_json)
with files.open_datafile("dataset_hdf5",
file_vars=file_vars,
mode="a",
write_log=False) as f_hdf5:
for id_to_delete in ids_to_delete:
name = "{name}/{id:04d}".format(name=dataset_name,
id=id_to_delete)
del json_all[name]
del f_hdf5[name]
log.debug(
"Deleted {name} from dataset {tech}-{stage} at {directory}",
name=name,
tech=tech,
stage=stage,
directory=json_path.parent,
)
with files.open_path(json_path, mode="wt", write_log=False) as f_json:
json.dump(json_all, f_json)
# Delete files if all datasets are deleted
if not any(["/" in k for k in json_all.keys()]):
json_path.unlink()
files.path("dataset_hdf5", file_vars=file_vars).unlink()
def __init__(self,
rundate,
tech,
stage,
dataset_name,
dataset_id,
empty=False,
**kwargs):
"""Create a new Dataset or read an existing one
Note:
Be aware that the implementation is dependent on ``self._fields`` being the first attribute to be set. See
:func:`__setattr__` for more information.
Args:
rundate: Date, the model run date.
tech: String, the technique.
stage: String, the stage.
dataset_name: String, the name of the dataset.
dataset_id: Int, id of the dataset.
empty: Boolean, if False (default) will read dataset from disk if available.
"""
self._fields = dict()
self._data = dict()
self._num_obs = 0
self._default_field_suffix = None
self._kwargs = kwargs
self._kwargs.setdefault("session",
dataset_name) # TODO: Can this be removed?
self.vars = dict(
config.program_vars(**dict(
kwargs,
rundate=rundate,
tech_name=tech,
stage=stage,
dataset_name=dataset_name,
dataset_id=str(dataset_id),
)))
self.vars.update(**kwargs)
self.rundate = rundate
dataset_id = _data.parse_dataset_id(rundate, tech, stage, dataset_name,
dataset_id)
self.name = "{name}/{id:04d}".format(name=dataset_name, id=dataset_id)
self.meta = dict()
# Try to read dataset from disk unless explicitly told to create an empty dataset
if not empty:
try:
self.read()
except FileNotFoundError:
pass