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:
        """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("<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 />|</strong>)\n\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
            try:
                _date = findall(r"</a> <i class=\"icon-time\"></i> <small>(.*)</small>", elements[i])[0]
            except:
                _date = None
                print("\n" + elements[i] + "\n")
            elements[i] = {"title": _title, "author": _author, "date": _date, "link": f"https://forum.mobilism.org/viewtopic.php?f={_link['f']}&t={_link['t']}", "linkParams": _link}

        return elements

    def work(self) -> str:
        """Call all the others methods."""
        session = self.connect()
        topics = self.search(session)

        return topics

def save(elements):
    """Save all the results parsed to a CSV file."""
    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: # no args
        print("No App to retrieve.")
        exit(1)
    load_dotenv() # load .env file
    try:
        try: # for logs
            debug = environ["DEBUG_MOBILISM"].lower() in ("yes", "true", "1")
        except:
            debug = False
        try: # try to fetch credentials from de .env first
            pseudoMobilism = environ["PSEUDO_MOBILISM"]
            passwordMobilism = environ["PASSWORD_MOBILISM"]
        except: # if it failed try to get from the cli
            if len(argv) >= 3:
                pseudoMobilism = argv[0]
                passwordMobilism = argv[1]
                argv = argv[-2:]
            else: # if it failed again there is a problem
                raise KeyError
        save(Scraper(pseudoMobilism, passwordMobilism, " ".join([n for n in argv]), debug).work()) # call the work() function
    except KeyError:
        print('Please fill in the username and password (with quotes) by args or with .env file and give an app to retrieve.')