46 lines
1.5 KiB
Python
46 lines
1.5 KiB
Python
import json
|
|
import os
|
|
import re
|
|
|
|
def remove_non_ascii(text):
|
|
return re.sub(r'[^\x00-\x7F]', ' ', text)
|
|
|
|
|
|
|
|
for file in ['train', 'val_unseen', 'val_seen', 'test']:
|
|
print(file)
|
|
if os.path.isfile('data/adversarial/reverie_{}_fnf.json'.format(file)):
|
|
with open('data/adversarial/reverie_{}_fnf.json'.format(file)) as fp:
|
|
data = json.load(fp)
|
|
|
|
|
|
result = {}
|
|
for i in data:
|
|
instruction_id = i['id']
|
|
if instruction_id not in result:
|
|
result[instruction_id] = {
|
|
'distance': float(i['distance']),
|
|
'ix': int(i['ix']),
|
|
'scan': i['scan'],
|
|
'id': i['id'],
|
|
'instructions_l': i['instructions_l'],
|
|
'path_id': int(i['path_id']),
|
|
'objId': i['objId'],
|
|
'path': i['path'],
|
|
'heading': float(i['heading']),
|
|
'instructions': [ remove_non_ascii(i['instruction'])],
|
|
'found': [ i['found'] ],
|
|
}
|
|
else:
|
|
result[instruction_id]['instructions'].append(remove_non_ascii(i['instruction']))
|
|
result[instruction_id]['found'].append( i['found'] )
|
|
|
|
output = []
|
|
for k, item in result.items():
|
|
output.append(item)
|
|
else:
|
|
output = []
|
|
|
|
with open('data/adversarial/REVERIE_{}.json'.format(file), 'w') as fp:
|
|
json.dump(output, fp)
|