from sys import argv from os import environ from dotenv import load_dotenv from cloudscraper import CloudScraper, create_scraper from re import findall class Scraper: def __init__(self, pseudo, password, app, debug = False): self.debug = debug self.url = "https://forum.mobilism.org" self.requested_app = app self.loginData = { "username": pseudo, "password": password, "login": "Login" } def errorFormat(self, code: int = None, message: str = "") -> str: return f"{f'[{code}]' if code else ''}{' ' if len(message) > 0 and code else ''}{message}." def connect(self) -> CloudScraper: session = create_scraper(browser = {"browser": "chrome", "platform": "windows"}) # connect with cloudflare bypasser with a chrome browser on windows if not session: raise SystemError(self.errorFormat(message = "The creation of the session failed")) if self.debug: print("Retrieval of the login SID...", end = " ") reponse = session.get(f"{self.url}/ucp.php", params = {"mode": "login"}) # get login page to get "sid" if reponse.status_code != 200: raise ConnectionError(self.errorFormat(code = reponse.status_code, message = "Login page not available")) try: self.loginData["sid"] = reponse.cookies.get_dict()["ppcw_29d3s_sid"] # register "sid" except: raise ValueError(self.errorFormat(message = "Cookie containing the SID not found.")) if self.debug: print("SID retrieval done,", end = " ") if self.debug: print("Connection attempt...") reponse = session.post(f"{self.url}/ucp.php", data = self.loginData, params = {"mode": "login"}) # connect to the forum using credentials if reponse.status_code != 200: raise ConnectionRefusedError(self.errorFormat(code = reponse.status_code, message = "Unable to connect")) reponse = session.get(f"{self.url}/index.php", cookies = reponse.cookies, params = {"sid": self.loginData["sid"]}) # back to index page if reponse.status_code != 200: raise ConnectionError(self.errorFormat(code = reponse.status_code, message = "Unable to get to the index page")) return session def search(self, session) -> list: if self.debug: print("Going to search page...", end = " ") reponse = session.get(f"{self.url}/search.php", params = {"keywords": self.requested_app, "sr": "topics", "sf": "titleonly"}) if "Sorry but you are not permitted to use the search system. If you're not logged in please" in reponse.text: raise ConnectionError(self.errorFormat(message = "Connection failed, check credentials")) if reponse.status_code != 200: raise ConnectionError(self.errorFormat(code = reponse.status_code, message = "Impossible to make the search")) if self.debug: print(f"Results retrieval for {self.requested_app}...", end = " ") return self.parse(reponse.text) def parse(self, htmlPage: str) -> list: if "No suitable matches were found." in htmlPage: return [] elements = htmlPage.split("