NL2TL-dataset/collect2/getUniqueLTL.py

import json 
import re,os 

def findUniqueLTL(paths:list):
    ret={}
    for path in paths:
        with open(path,'r') as f:
            jsonlists=f.readlines()
            for jsonlist in jsonlists:
                j=json.loads(jsonlist)
                ret[j['raw_ltl']]=1
    return ret

if __name__=='__main__':
    path=['path/to/NL2TL-dataset/collect2/ltl_eng_test_mid_ascii_gptAuged.jsonl','path/to/NL2TL-dataset/collect2/ltl_eng_train_mid_ascii_gptAuged.jsonl']
    LTLs=findUniqueLTL(paths=path)
    with open(os.path.join('path/to/NL2TL-dataset/collect2','NLTLsummary.json'),'w') as f :
        f.write(json.dumps(LTLs,sort_keys=False,indent=4,separators=(',',':')))