import requests
from bs4 import BeautifulSoup
import pandas as pd
headers = {
"Cookie": "..."
}
def main():
global max_count
df = pd.DataFrame(columns=["UserName", "MsCount"])
for page in range(584):
link = f"https://yougame.biz/chat/history/?page={page}"
print(f"-- Working as: {link} --")
responce = requests.get(link, cookies=headers)
if responce.status_code != 200:
print("Invalid request")
break
else:
soup = BeautifulSoup(responce.text, "html.parser")
elements = soup.find_all("div", class_="message")
if elements != None:
for element in elements:
name_user = element.find("div", class_="username")
message_user = element.find("div", class_="bbWrapper")
message_date= element.find("div", class_="message-date")
name = (name_user.text).strip()
message = (message_user.text).strip()
date = (message_date.text).strip()
if name in df["UserName"].values:
df.loc[df["UserName"] == name, "MsCount"] += 1
else:
df = df._append(
{
"UserName": name,
"MsCount": 1,
},
ignore_index=True
)
with open("chat_history.txt", "a", encoding="utf-8") as ch:
ch.write(f"{date}: {name}: {(message)}\n")
else:
print("No elements")
return None
else:
df.to_excel("mainpars.xlsx", index=False)
if __name__ == "__main__":
main()
