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: """Pretty error message.""" return f"{f'[{code}]' if code else ''}{' ' if len(message) > 0 and code else ''}{message}." def connect(self) -> CloudScraper: """Login to the forum using credentials.""" 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")) # called only if failed at creating the session 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 - params are set by default but its in case forum changing that if reponse.status_code != 200: raise ConnectionRefusedError(self.errorFormat(code = reponse.status_code, message = "Unable to connect")) # called only status code isn't 200 return session def search(self, session) -> list: """Do the research.""" if self.debug: print("Going to search page and check connection...", end = " ") reponse = session.get(f"{self.url}/search.php", params = {"keywords": self.requested_app, "sr": "topics", "sf": "titleonly"}) # fetch results page 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")) # called only if login failed if reponse.status_code != 200: raise ConnectionError(self.errorFormat(code = reponse.status_code, message = "Impossible to make the search")) # called only status code isn't 200 if self.debug: print(f"Connected.") if self.debug: print(f"Fetching results for {self.requested_app}...", end = " ") return self.parse(reponse.text) def parse(self, htmlPage: str) -> list: """Parse HTML reponse to a clean list""" if "No suitable matches were found." in htmlPage: return [] elements = htmlPage.split("