wmt-2020-pl-en/googletrans/conversion.py

32 lines
930 B
Python

"""A conversion module for googletrans"""
import re
import json
def format_json(text):
# save state
states = []
for i, pos in enumerate(re.finditer('"', text)):
p = pos.start() + 1
if i % 2 == 0:
nxt = text.find('"', p)
states.append((p, text[p:nxt]))
# replace all weired characters in text
while text.find(',,') > -1:
text = text.replace(',,', ',null,')
while text.find('[,') > -1:
text = text.replace('[,', '[null,')
# recover state
for i, pos in enumerate(re.finditer('"', text)):
p = pos.start() + 1
if i % 2 == 0:
j = int(i / 2)
nxt = text.find('"', p)
# replacing a portion of a string
# use slicing to extract those parts of the original string to be kept
text = text[:p] + states[j][1] + text[nxt:]
converted = json.loads(text)
return converted