def main(output_folder, narrow_folder):
"""Read in a folder of irregular wide-format files and write as narrow csvs
"""
config = read_packaged_file("config.yaml", "otoole.preprocess")
for parameter, details in config.items():
logger.info("Looking for %s", parameter)
config_details = config[parameter]
filepath = os.path.join(output_folder, parameter + ".csv")
try:
df = pd.read_csv(filepath)
except pd.errors.EmptyDataError:
logger.error("No data found in file for %s", parameter)
expected_columns = config_details["indices"]
default_columns = expected_columns + ["VALUE"]
df = pd.DataFrame(columns=default_columns)
entity_type = config[parameter]["type"]
if entity_type == "param":
narrow = check_parameter(df, config_details, parameter)
if not narrow.empty:
narrow_checked = check_datatypes(narrow, config, parameter)
else:
narrow_checked = narrow
elif entity_type == "set":
narrow = check_set(df, config_details, parameter)
if not narrow.empty:
narrow_checked = check_set_datatype(narrow, config, parameter)
else:
narrow_checked = narrow
write_out_dataframe(narrow_folder, parameter, narrow_checked)
def read_config(path_to_user_config: str = None) -> Dict:
"""Reads the config file holding expected OSeMOSYS set and parameter dimensions
Arguments
---------
path_to_user_config : str, optional, default=None
Optional path to a user defined configuration file
Returns
-------
dict
"""
if path_to_user_config:
config = read_packaged_file(path_to_user_config, None)
else:
config = read_packaged_file("config.yaml", "otoole.preprocess")
return config
def validate_model(args):
file_format = args.format
if args.config:
config = read_packaged_file(args.config)
validate(file_format, args.filepath, config)
else:
validate(file_format, args.filepath)
def __init__(self, datapackage: str, datafilepath: str, sql: bool = False):
self.datapackage = datapackage
self.datafilepath = datafilepath
self.sql = sql
self.package = self._get_package()
self.default_values = self._get_default_values()
self.config = read_packaged_file("config.yaml", "otoole.preprocess")
def write_default_values(filepath):
config = read_packaged_file("config.yaml", "otoole.preprocess")
default_values_path = os.path.join(filepath, "data", "default_values.csv")
with open(default_values_path, "w") as csv_file:
csv_file.write("name,default_value\n")
for name, contents in config.items():
if contents["type"] == "param":
csv_file.write("{},{}\n".format(name, contents["default"]))
def convert_file_to_package(path_to_datafile: str, path_to_datapackage: str):
"""Converts an OSeMOSYS datafile to a Tabular Data Package
Arguments
---------
path_to_datafile: str
path_to_datapackage: str
Path to the folder in which to write the new Tabular Data Package
"""
config = read_packaged_file("config.yaml", "otoole.preprocess")
amply_datafile = read_in_datafile(path_to_datafile, config)
dict_of_dataframes = convert_amply_to_dataframe(amply_datafile, config)
if not os.path.exists(path_to_datapackage):
os.mkdir(path_to_datapackage)
for name, df in dict_of_dataframes.items():
write_out_dataframe(path_to_datapackage, name, df)
datapackage = read_packaged_file("datapackage.json", "otoole.preprocess")
filepath = os.path.join(path_to_datapackage, "datapackage.json")
with open(filepath, "w") as destination:
destination.writelines(datapackage)
write_default_values(path_to_datapackage)
def generate_package(path_to_package):
"""Creates a datapackage in folder ``path_to_package``
[{'fields': 'REGION', 'reference': {'resource': 'REGION', 'fields': 'VALUE'}}]
"""
datapath = os.path.join(path_to_package)
package = Package(base_path=datapath)
package.infer("data/*.csv")
package.descriptor["licenses"] = [{
"name":
"CC-BY-4.0",
"path":
"https://creativecommons.org/licenses/by/4.0/",
"title":
"Creative Commons Attribution 4.0",
}]
package.descriptor["title"] = "The OSeMOSYS Simplicity Example Model"
package.descriptor["name"] = "osemosys_model_simplicity"
package.descriptor["contributors"] = [{
"title": "Will Usher",
"email": "*****@*****.**",
"path": "http://www.kth.se/wusher",
"role": "author",
}]
package.commit()
config = read_packaged_file("config.yaml", "otoole.preprocess")
new_resources = []
for resource in package.resources:
descriptor = resource.descriptor
name = resource.name
if config[name]["type"] == "param":
indices = config[name]["indices"]
logger.debug("Indices of %s are %s", name, indices)
foreign_keys = []
for index in indices:
key = {
"fields": index,
"reference": {
"resource": index,
"fields": "VALUE"
},
}
foreign_keys.append(key)
descriptor["schema"]["foreignKeys"] = foreign_keys
descriptor["schema"]["primaryKey"] = indices
descriptor["schema"]["missingValues"] = [""]
new_resources.append(descriptor)
package.descriptor["resources"] = new_resources
package.commit()
filepath = os.path.join(path_to_package, "datapackage.json")
package.save(filepath)
def read_validation_config():
return read_packaged_file("validate.yaml", "otoole")
