def make_train_kbs(scenario_path):
'''
Create trained KBs per sub-fold
'''
kbs_dic = {}
cv_train = 'cv_'+scenario_path.split('/')[-1]
for subdir in os.listdir(scenario_path + '/' + cv_train):
if 'train_' in subdir and 'kb_' not in subdir:
case_subdir = scenario_path + '/'+cv_train+'/'+subdir+'/'
options = ' --discard --feat-timeout +inf '
options += '--root-path '+scenario_path+'/ '
# train scenario
kb_name = case_subdir.split('/')[-1]
input_args = (options + case_subdir).split(' ')
input_args = [x for x in input_args if x]
args,info,lims = train_scenario_mock(input_args)
kbs_dic[case_subdir] = [args,info,lims]
return kbs_dic
def prepare_scenario(scenario_name, tdir, number_insts):
'''
Initial Train and Split
'''
root_arr = os.path.realpath(__file__).split('/')[:-2]
root = '/'.join(root_arr)
src_path = root + '/src/'
scenario_path = root + '/data/oasc_scenarios/train/' + scenario_name
scenario_cv = scenario_path
scenario_t = scenario_path + '/' + tdir
if not os.path.exists(scenario_path):
print 'Scenario Name Err on', scenario_path
sys.exit()
# crea cartella t
if not os.path.exists(scenario_t):
print 'Creating folder t'
copyTrainFiles(scenario_path, tdir)
# split train
#comincia il train che in teoria andrebbe tolto poi
print 'Check and Split Training Data ...'
smart_conf = scenario_path + "/smart.txt"
shouldSplit = False
if not os.path.exists(smart_conf):
shouldSplit = True
else:
with open(smart_conf, 'r') as f:
first_line = f.readline()
if first_line == '' or int(first_line) != number_insts:
shouldSplit = True
subdir = scenario_name
cv_subdir = scenario_cv
if not os.path.exists(cv_subdir + '/cv_' + subdir) or shouldSplit:
case_subdir = cv_subdir + '/t'
options = ' --discard --feat-timeout +inf '
options += '--root-path ' + scenario_path + '/ '
input_args = (options + case_subdir).split(' ')
input_args = [x for x in input_args if x]
kb_arg, kb_info, kb_lims = train_scenario_mock(input_args)
with open(case_subdir + '/kb.args', 'w') as outfile:
json.dump(kb_arg, outfile)
smart_split(cv_subdir, kb_info, number_insts)
if shouldSplit:
with open(smart_conf, 'w') as f:
f.write(str(number_insts))
return scenario_path, scenario_cv, src_path