def get_assigned_targets(config):
"""
:param config: config dictionaryn
:returns: (dict (targets_id -> group_id), dict (group_id -> group_label))
"""
print ">> Opening csv file from", config["base_csv_path"]
all_targets_df = pd.DataFrame.from_csv(config["base_csv_path"],
sep=config["sep"], index_col=config["index_cols"], tupleize_cols=True)
print ">> Retriving groups"
index_list_size = len(config["index_cols"])
if index_list_size > 1:
grouped_targets = all_targets_df.groupby(level=range(index_list_size))
else:
grouped_targets = all_targets_df.groupby(level=index_list_size-1)
print ">> Setting multi-layer dictionary for groups"
groups = regroup(grouped_targets)
levels = get_length_per_level(groups)
print ">> Setting groups ids"
groups_label_id = set_groups_ids(groups, levels)
print ">> Registering groups and their ids at"
print config["info_frequency_path"].replace(".txt", ".INFO")
info_file_path = config["info_frequency_path"].replace(".txt", ".INFO")
fileloader.createValidPath(info_file_path)
diversity_info_file = open(info_file_path, "w+")
diversity_info_file.write("#INFO: group_name: 'group_identifier'\n")
diversity_info_file.write(yaml.dump(groups, default_flow_style=False))
diversity_info_file.close()
print ">> Assigning group id for each target"
return set_group_id_for_target(grouped_targets, groups, config), groups_label_id
def set_groups_paths(global_path, groups):
#
"""
SETTING AND CREATING PATHS AS IT IS NEEDED FOR EACH GROUP
"""
dir_path = global_path.split("/")
file_name = dir_path.pop()
dir_path = "/".join(dir_path)
fileloader.createValidPath(global_path)
output_paths = {}
for group_label in groups.keys():
output_paths[group_label] = dir_path + "/" + group_label.replace(" ", "_")
output_paths[group_label] += "/" + file_name
fileloader.createValidPath(output_paths[group_label])
output_paths["global"] = global_path
return output_paths
def set_all_paths(frequency_path, groups_label_id):
#Returns dict(group_id -> writing object)
"""
Arguments:
frequency_path = path which will hold all frequency files for each group
groups_label_id = dict ( group_id -> groub_label )
"""
fileloader.createValidPath(frequency_path)
dir_path = frequency_path.split("/")
file_name = dir_path.pop();
dir_path = '/'.join(dir_path)+'/'
groups_files = {}
print ">> Setting paths for groups files"
depth = len(max(groups_label_id.keys()).split("."))
groups_files["global"] = frequency_path
for group_id,group_label in groups_label_id.iteritems():
if len(group_label.split("/")) < depth:
group_path = dir_path+group_label.replace(" ", "_").decode('utf-8')+"/"
fileloader.createValidPath(group_path)
groups_files[group_id]=group_path+file_name
return groups_files
def adjust_paths(configs, main_paths=None, dir_list=None):
"""
Adjust config paths for
:param CONFIGS: Config dictionary stored at /src/config/
:param main_paths: -- Used for recurstion to pass root paths
:param dir_list: used for recursion to pass root dirs
"""
if not main_paths:
main_paths = configs["paths"]
if not dir_list:
dir_list = configs["dir_names"]
if type(configs) == dict:
for attr, value in configs.iteritems():
dir_name = attr.split("_")[0]
if dir_name in dir_list and attr.rstrip("_path") not in dir_list:
configs[attr] = main_paths[dir_name+"_path"] + value
floader.createValidPath(configs[attr])
if type(value) == list or type(value) == dict:
adjust_paths(value, main_paths, dir_list)
if type(configs) == list:
for inner_config in configs:
if type(inner_config) == list or type(inner_config) == dict:
adjust_paths(inner_config, main_paths, dir_list)
