|
import json
|
|
import os
|
|
import re
|
|
from datetime import datetime
|
|
|
|
def extract_line(line):
|
|
'''
|
|
flag == 0: empty line
|
|
flag == 1: timestamp and user name
|
|
flag == 2: content
|
|
'''
|
|
flag = 0
|
|
if line == '\n':
|
|
return (""), flag
|
|
timestamp = re.findall(r"\d+-\d+-\d+ \d+:\d+:\d+", line)
|
|
if len(timestamp) > 0:
|
|
flag = 1
|
|
timestamp = datetime.strptime(timestamp[0], "%Y-%m-%d %H:%M:%S")
|
|
user_name = line.rsplit(" ", 1)[-1]
|
|
return (timestamp, user_name), flag
|
|
else:
|
|
flag = 2
|
|
return (line,), flag
|
|
|
|
def timestamp_distance(time1, time2):
|
|
if time2 is None:
|
|
return False
|
|
days = (time1 - time2).days
|
|
if days < 0:
|
|
days = (time2 - time1).days
|
|
seconds = (time2 - time1).seconds
|
|
else:
|
|
days = (time1 - time2).days
|
|
seconds = (time1 - time2).seconds
|
|
seconds += days * 3600 * 24
|
|
if seconds > 3600 * 3:
|
|
return True
|
|
else:
|
|
return False
|
|
|
|
|
|
for d in os.listdir("data_raw"):
|
|
target_name = d.split("(")[0]
|
|
with open(os.path.join("data_raw", d, f"{target_name}.txt"), encoding='utf-8') as f:
|
|
lines = f.readlines()
|
|
|
|
records = []
|
|
convs = []
|
|
timestamp_last = None
|
|
text = ""
|
|
for line in lines:
|
|
content, flag = extract_line(line)
|
|
if flag == 0:
|
|
conv = {"from": user_name[:-1], "value": text[:-1], "timestamp": str(timestamp)}
|
|
if timestamp_distance(timestamp, timestamp_last):
|
|
text = ""
|
|
records.append({"id": f"conv_{len(records)}_{timestamp_last}", "conversations": convs})
|
|
convs = [conv]
|
|
else:
|
|
convs.append(conv)
|
|
text = ""
|
|
timestamp_last = timestamp
|
|
elif flag == 1:
|
|
timestamp, user_name = content
|
|
else:
|
|
text += content[0]
|
|
|
|
if len(convs) != 0:
|
|
records.append({"id": f"conv_{len(records)}_{timestamp_last}", "conversations": convs})
|
|
|
|
os.makedirs("data", exist_ok=True)
|
|
json.dump(records, open(os.path.join("data", f"{target_name}.json"), "w", encoding='utf-8'), ensure_ascii=False, indent=4)
|
|
|
