def get_processor_data(output_dir, return_path=False):
path = osp.join(output_dir, cytokit_io.get_processor_data_path())
proc_data = exec.read_processor_data(path)
if return_path:
return proc_data, path
else:
return proc_data
def processor_data_summary(self):
logging.info('Running processor data summary operation')
processor_data_path = osp.join(self.data_dir, cytokit_io.get_processor_data_path())
nb_name = 'processor_data_analysis.ipynb'
nb_output_path = osp.join(self.data_dir, osp.dirname(processor_data_path), 'processor_data_analysis.ipynb')
nb_params = {'processor_data_path': processor_data_path}
core.run_nb(nb_name, nb_output_path, nb_params)
logging.info('Processor data summary complete; view results with `jupyter notebook %s`', nb_output_path)
def record_processor_data(data, output_dir):
"""Save processor data as json file"""
path = osp.join(output_dir, cytokit_io.get_processor_data_path())
if not osp.exists(osp.dirname(path)):
os.makedirs(osp.dirname(path), exist_ok=True)
# If a data file already exists, merge the two preferring newer entries
if osp.exists(path):
past_data = read_processor_data(path)
past_data.update(data)
data = past_data
# Ensure that values in data dict are lists of dicts (not DataFrames already
# since they will not serialize correctly)
data = {
k: v.to_dict(orient='records') if isinstance(v, pd.DataFrame) else v
for k, v in data.items()
}
# Use pandas for serialization as it has built-in numpy type converters
pd.Series(data).to_json(path, orient='index')
return path