import json
from tqdm import tqdm
import numpy as np
import time

print('start')
start_time = time.time()



inputfile = open('./data/main_sections_170k_metadata.json', "r", encoding='utf-8')

data = json.load(inputfile)

inputfile.close()

dict = {}

for item in tqdm(data):
    key = item['qanon_id']
    child_order = item['child_order']
    content = item['content']
    level =item['other_info']['level']
    if not key in  dict:
        dict[key] = []
    dict[key].append({'content':content, 'child_order':child_order, 'level': level})

    

outputfile = open('./data/main_qanon_170k_metadata.json', "w", encoding='utf-8')
outputfile.write(json.dumps(dict, ensure_ascii=False, indent = 4))
outputfile.close()
end_time = time.time()
print(f"elapsed time:   {end_time-start_time}")
print("end")