mobilismScrap/main.py

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("<tr>\n<td>")[1:]
        elements[-1] = elements[-1].split("</td>\n</tr>")[0]
        for i in range(0, len(elements)):
            try:
                _title = findall(r"class=\"topictitle\">(.*)<\/a>", elements[i])[0]
            except:
                _title = None
            try:
                _author = findall(r"<br />\n<i class=\"icon-user\"></i> by <a href=\"\./memberlist\.php\?mode=viewprofile&amp;u=\d+\"( style=\"color: #.*;\" class=\"username-coloured\")?>(.*)</a>", elements[i])[0][-1]
            except:
                _author = None
            try:
                _link = findall(r"\./viewtopic\.php\?f=(\d*)&amp;t=(\d*)&amp", elements[i])[0]
                _link = {"f": _link[0], "t": _link[1]}
            except:
                _link = None
            elements[i] = {"title": _title, "author": _author, "link": f"https://forum.mobilism.org/viewtopic.php?f={_link['f']}&t={_link['t']}", "linkParams": _link}

        return elements

    def work(self) -> str:
        session = self.connect()
        link = self.search(session)

        return link

def save(elements):
    taille = len(elements)
    if taille == 0:
        print("Aucun élément n'a été trouvé avec la recherche.")
        return
    filename = "results.csv"
    with open(filename, "w") as f:
        f.write(";".join(list(elements[0].keys())[:-1]))
        f.write("\n")
        for element in elements:
            if element != "linkParams":
                f.write(";".join(str(e) for e in list(element.values())[:-1]))
            f.write("\n")
    print(f"{taille} éléments ont étés enrengistés dans le fichier {filename}.")

if __name__ == "__main__":
    argv = argv[1:]
    if len(argv) < 1:
        print("No App to retrieve.")
        exit(1)
    load_dotenv()
    try:
        try:
            debug = environ["DEBUG_MOBILISM"].lower() in ("yes", "true", "1")
        except:
            debug = False
        try:
            pseudoMobilism = environ["PSEUDO_MOBILISM"]
            passwordMobilism = environ["PASSWORD_MOBILISM"]
        except:
            if len(argv) >= 3:
                pseudoMobilism = argv[0]
                passwordMobilism = argv[1]
                argv = argv[-2:]
            else:
                raise KeyError
        save(Scraper(pseudoMobilism, passwordMobilism, " ".join([n for n in argv]), debug).work())
    except KeyError:
        print('Please fill in the username and password (with quotes) by args or with .env file and give an app to retrieve.')