fix thread detection

* complete migration

.envexample

@@ -1,5 +1,6 @@
-TOKEN=
-TOKEN_SECRET=
 CONSUMER_KEY=
 CONSUMER_SECRET=
+BEARER_TOKEN=
+TOKEN=
+TOKEN_SECRET=
 PSEUDO=

README.md

@@ -2,33 +2,26 @@
 
 Bot qui envoie automatiquement des réponses ennuyantes quand les personnes que tu suis finissent leur tweet par un mot spécial.
 
-Certains mots peuvent servir de "trigger" sans être dans la liste, example : `aussi` n'est pas dans la liste, mais en retirant `aus`, on obtient `si`, qui est dans la liste.
-| Mot | Réponse    | ¦ | Mot     | Réponse   | ¦ | Mot  | Réponse                       | ¦ | Mot                     | Réponse    | ¦ | Mot   | Réponse           | ¦ | Mot   | Réponse
-------|---------   |:-:|---------|-----------|:-:|------|-------------------------------|:-:|-------------------------|------------|:-:|-------|------------------ |:-:|-------|-
-quoi  | feur       | ¦ | con     | combre    | ¦ | coup | teau                          | ¦ | ka                      | pitaine    | ¦ | moi   | tié/sson/sissure  | ¦ | ni    | cotine
-oui   | stiti/fi   | ¦ | ok      | sur glace | ¦ | ça   | pristi/perlipopette/von       | ¦ | fais                    | rtile      | ¦ | toi   | lette/ture        | ¦ | quand | dide/tal/didat
-non   | bril       | ¦ | ouais   | stern     | ¦ | bon  | jour/soir (dépend de l'heure) | ¦ | tant (ou autre syntaxe) | gente      | ¦ | top   | inambour          | ¦ | sol   | itaire
-nan   | cy         | ¦ | comment | tateur    | ¦ | qui  | wi/mono                       | ¦ | et                      | eint/ain   | ¦ | jour  | nal               | ¦ | vois  | ture
-hein  | deux       | ¦ | mais    | on        | ¦ | sur  | prise                         | ¦ | la                      | vabo/vande | ¦ | ya/yo | hourt/yo          | ¦ | akhy  | nator
-ci    | tron       | ¦ | fort    | boyard    | ¦ | pas  | nini/steur                    | ¦ | tki                     | la         | ¦ | re    | pas/veil/tourne   | ¦ |
-
 N'hésitez pas à ouvrir un ticket ou faire une merge-request pour contribuer au projet.
 
 ## Lancer le Bot
 
-Donner la permission `Read and Write` (ou `Read + Write + Direct Messages` mais aucun DM n'est envoyé) au bot dans `Settings` puis `App permissions`.
+Donner la permission au minimum `Read and Write` au bot dans `Settings` puis `App permissions`.
+
+Les codes fourni par l'API de Twitter sont généralements disponible dans la page `Keys and tokens`.
 
 Détails des variables d'environnement :
 |     Variable    | Explication et où elle se trouve
-----------------|-
-TOKEN           | Token d'accès disponible dans la section `Authentication Tokens` sous la sous-rubrique `Access Token and Secret`
-TOKEN_SECRET    | Token d'accès secret disponible dans la section `Authentication Tokens` sous la sous-rubrique `Access Token and Secret`
-CONSUMER_KEY    | Clé API disponible dans la section `Consumer Keys`
-CONSUMER_SECRET | Clé secrète API disponible dans la section `Consumer Keys`
-PSEUDOS         | Pseudos du ou des compte.s que vous voulez écouter pour le snipe (a séparer avec une virgule **sans** espaces)
-WHITELIST       | Pseudos des comptes qui ne seront pas touché par le Bot (facultatif, a séparer avec une virgule **sans** espaces, par défaut la liste est vide)
-VERBOSE         | Affiche plus de messages dans la console [False\|True] (facultatif, par défaut sur False)
-FORCELIST       | Force le bot à écouter certains comptes (séparer les comptes avec une virgule **sans** espaces, par défaut la liste est vide)
+| --------------- |-
+| TOKEN           | Token d'accès disponible dans la section `Authentication Tokens` sous la sous-rubrique `Access Token and Secret`
+| TOKEN_SECRET    | Token d'accès secret disponible dans la section `Authentication Tokens` sous la sous-rubrique `Access Token and Secret`
+| CONSUMER_KEY    | Clé API disponible dans la section `Consumer Keys` sous la sous-rubrique `API Key and Secret`
+| CONSUMER_SECRET | Clé secrète API disponible dans la section `Consumer Keys` sous la sous-rubrique `API Key and Secret`
+| BEARER_TOKEN    | Token disponible dans la section `Authentication Tokens` sous la sous-rubrique `Bearer Token`
+| PSEUDOS         | Pseudos du ou des compte.s que vous voulez écouter pour le snipe (a séparer avec une virgule **sans** espaces)
+| WHITELIST       | Pseudos des comptes qui ne seront pas touché par le Bot (facultatif, a séparer avec une virgule **sans** espaces, par défaut la liste est vide)
+| VERBOSE         | Affiche plus de messages dans la console [False\|True] (facultatif, par défaut sur False)
+| FORCELIST       | Force le bot à écouter certains comptes (séparer les comptes avec une virgule **sans** espaces, par défaut la liste est vide)
 
 ### En local
 
@@ -46,10 +39,11 @@ docker build https://git.kennel.ml/Anri/feurBot.git#main --tag feurbot:main && \
 docker run -d \
     --name="feurBot" \
     feurbot:main \
-    --TOKEN="" \
-    --TOKEN_SECRET="" \
     --CONSUMER_KEY="" \
     --CONSUMER_SECRET="" \
+    --BEARER_TOKEN="" \
+    --TOKEN="" \
+    --TOKEN_SECRET="" \
     --PSEUDOS=""
 ```
 Ou avec un `docker-compose.yml` :
@@ -60,10 +54,55 @@ services:
     build: https://git.kennel.ml/Anri/feurBot.git#main
     container_name: feurBot
     environment:
-      - TOKEN=
-      - TOKEN_SECRET=
       - CONSUMER_KEY=
       - CONSUMER_SECRET=
+      - BEARER_TOKEN=
+      - TOKEN=
+      - TOKEN_SECRET=
       - PSEUDOS=
     restart: unless-stopped
 ```
+
+## Liste des mots
+Certains mots peuvent servir de déclencheur sans être dans la liste,
+example : `aussi` n'est pas dans la liste, mais en retirant `aus`,
+on obtient `si`, qui est dans la liste.
+
+|           Mot           |             Réponse           |
+| ----------------------- | ----------------------------- |
+| quoi                    | feur/feuse                    |
+| oui                     | stiti/fi                      |
+| non                     | bril                          |
+| nan                     | cy                            |
+| hein                    | deux/bécile                   |
+| ci                      | tron/prine                    |
+| con                     | combre/gelé/pas               |
+| ok                      | sur glace                     |
+| ouais                   | stern                         |
+| comment                 | tateur/tatrice/dant Cousteau  |
+| mais                    | on                            |
+| fort                    | boyard                        |
+| coup                    | teau                          |
+| ça                      | pristi/von/perlipopette       |
+| bon                     | jour/soir (dépend de l'heure) |
+| qui                     | wi/mono                       |
+| sur                     | prise                         |
+| pas                     | nini/steur/trimoine/té/stis   |
+| ka                      | pitaine/pitulation            |
+| fais                    | rtile                         |
+| tant (ou autre syntaxe) | gente/tation                  |
+| et                      | eint/ain                      |
+| la                      | vabo/vande                    |
+| tki                     | la                            |
+| moi                     | tié/sson/sissure              |
+| toi                     | lette/ture                    |
+| top                     | inambour                      |
+| jour                    | nal                           |
+| ya                      | hourt                         |
+| yo                      | yo/ghourt                     |
+| ni                      | cotine                        |
+| re                      | pas/veil/tourne               |
+| quand                   | dide/tal/didat                |
+| sol                     | itaire                        |
+| vois                    | ture                          |
+| akhy                    | nator                         |

main.py

@@ -1,12 +1,11 @@
-from dotenv import load_dotenv
-from os import environ
-from tweepy import OAuth1UserHandler, API, Stream
-from re import sub, findall
-from random import choice
 from datetime import datetime
-from pytz import timezone
-from queue import Queue
-from json import loads
+from os import environ
+from random import choice
+from re import findall, sub
+
+from dotenv import load_dotenv
+from tweepy import Client, StreamingClient, StreamRule, Tweet, User, errors
+
 
 def load(variables) -> dict:
     """Load environment variables"""
@@ -16,7 +15,8 @@ def load(variables) -> dict:
         try:
             if var == "VERBOSE":  # check is VERBOSE is set
                 try:
-                    res = bool(environ[var])
+                    if environ[var].lower() == "true":
+                        res = True
                 except:
                     res = False  # if not its False
             elif var == "WHITELIST":  # check if WHITELIST is set
@@ -32,14 +32,17 @@ def load(variables) -> dict:
             else:
                 res = environ[var]
             if var == "PSEUDOS":
-                res = list(set(res.split(",")) - {""}) # create a list for the channels and remove blank channels and doubles
+                # create a list for the channels and remove blank channels and doubles
+                res = list(set(res.split(",")) - {""})
             keys[var] = res
         except KeyError:
-            print(f"Veuillez définir la variable d'environnement {var} (fichier .env supporté)")
+            print(
+                f"Veuillez définir la variable d'environnement {var} (fichier .env supporté)")
             exit(1)
     return keys
 
-def cleanTweet(tweet: str) -> str:
+
+def cleanTweet(tweet: str) -> str | None:
     """Remove all unwanted elements from the tweet"""
     # Convert to lower case
     tweet = tweet.lower()
@@ -53,7 +56,7 @@ def cleanTweet(tweet: str) -> str:
         tweet = sub(r"#\S+", " ", tweet)
     else:
         # Too much hashtags in the tweet -> so ignore it
-        return ""
+        return None
     # Remove usernames
     tweet = sub(r"@\S+", " ", tweet)
     # Remove everything who isn't a letter/number/space
@@ -64,67 +67,95 @@ def cleanTweet(tweet: str) -> str:
 
     # Remove key smashing in certains words
     #       uiii      naaaan          quoiiii     noooon          heiiin           siiii
-    tweet = sub(r"(?<=ui)i+|(?<=na)a+(?<!n)|(?<=quoi)i+|(?<=no)o+(?<!n)|(?<=hei)i+(?<!n)|(?<=si)i+", "", tweet)
+    tweet = sub(
+        r"(?<=ui)i+|(?<=na)a+(?<!n)|(?<=quoi)i+|(?<=no)o+(?<!n)|(?<=hei)i+(?<!n)|(?<=si)i+", "", tweet)
 
     return tweet.strip()
 
-class Listener(Stream):
+
+class Listener(StreamingClient):
     """Watch for tweets that match criteria in real-time"""
+
     def __init__(
         self,
-        consumer_key,
-        consumer_secret,
-        access_token,
-        access_token_secret,
-        api: API = None,
-        users: list = None,
-        forcelist: list = None,
-        q = Queue()
+        bearer_token,
+        client: Client,
     ):
-        super(Listener, self).__init__(consumer_key, consumer_secret, access_token, access_token_secret)
-        self.q = q
-        self.api = api
-        self.accounts = [users, forcelist]
-        self.listOfFriendsID = getFriendsID(api, users) + getIDs(api, forcelist)
+        super(Listener, self).__init__(bearer_token, wait_on_rate_limit=True)
+        self.client = client
+        self.cache = {}
 
     def on_connect(self):
-        if self.accounts[1] == []:
-            forcelist = "Aucun"
-        else:
-            forcelist = f"@{', @'.join(self.accounts[1])}"
-        print(f"Début du scroll sur Twitter avec les abonnements de @{', @'.join(self.accounts[0])} et ces comptes en plus : {forcelist} comme timeline...")
+        print(f"Début du scroll sur Twitter...")
 
-    def on_disconnect_message(notice):
-        notice = notice["disconnect"]
-        print(f"Déconnexion (code {notice['code']}).", end = " ")
-        if len(notice["reason"]) > 0:
-            print(f"Raison : {notice['reason']}")
+    def _get_user(self, uid: int) -> User:
+        """Return username by ID, with cache support"""
+        # If not cached
+        if not uid in self.cache:
+            # Fetch from Twitter
+            self.cache[uid] = self.client.get_user(
+                id=uid, user_fields="protected", user_auth=True).data
 
-    def on_status(self, status):
-        json = status._json
-        # Verify the author of the tweet
-        if json["user"]["id"] in self.listOfFriendsID and json["user"]["screen_name"] not in keys["WHITELIST"]:
-            # Verify the age of the tweet
-            if seniority(json["created_at"]):
-                # Verify if the tweet isn't a retweet
-                if not hasattr(status, "retweeted_status"):
-                    # Fetch the tweet
-                    if "extended_tweet" in json:
-                        tweet = cleanTweet(status.extended_tweet["full_text"])
-                    else:
-                        tweet = cleanTweet(status.text)
-                    # Fetch the last word of the tweet
-                    lastWord = tweet.split()[-1:][0]
+        # Return the user
+        return self.cache[uid]
+
+    def on_tweet(self, tweet: Tweet):
+        # Check if the tweet is not a retweet
+        if not tweet.text.startswith("RT @"):
+            # Cancel if author of the first tweet in the conversation is in private
+            if tweet.conversation_id and tweet.id != tweet.conversation_id:
                 if keys["VERBOSE"]:
-                        infoLastWord = f"dernier mot : \"{lastWord}\"" if len(lastWord) > 0 else "tweet ignoré car trop de hashtags"
-                        print(f"Tweet trouvé de {json['user']['screen_name']} ({infoLastWord})...", end = " ")
+                    print("Thread...", end=" ")
+                base_tweet = self.client.get_tweet(id=tweet.conversation_id, tweet_fields="author_id", user_auth=True).data
+                # Sometimes Twitter don't give what we want
+                if hasattr(base_tweet, 'author_id'):
+                    base_author = self._get_user(base_tweet.author_id)
+                    # Check if account is in private mode
+                    if base_author.protected:
+                        if keys["VERBOSE"]:
+                            print("Auteur du premier tweet en privé, pas de réponses.")
+                        return
+                    else:
+                        if keys["VERBOSE"]:
+                            print("Auteur du premier tweet en publique...", end=" ")
+                else:
+                    # Can't check the status of the first tweet in the thread, ignoring for safety
+                    if keys["VERBOSE"]:
+                        print("Impossible de vérifier le status de l'auteur du fil.")
+                    return
+            author = self._get_user(tweet.author_id)
+            # Clean the tweet
+            lastWord = cleanTweet(tweet.text)
+
+            # Log
+            if keys["VERBOSE"]:
+                infoLastWord = "dernier mot : "
+                newline = "\n"
+                match lastWord:
+                    case None:
+                        infoLastWord += "tweet ignoré car trop de hashtags"
+                    case w if len(w) == 0:
+                        infoLastWord += "tweet pas intéressant"
+                    case _:
+                        infoLastWord += f"dernier mot : {lastWord.split()[-1:][0]}"
+                        newline = ""
+                print(
+                    f"Tweet trouvé de {author.username} ({infoLastWord})...{newline}", end=" ")
+
+            # Ignore a tweet
+            if lastWord == None or len(lastWord) == 0:
+                return
+
+            # Fetch the last word of the tweet
+            lastWord = lastWord.split()[-1:][0]
+
             # Check if the last word found is a supported word
             if lastWord in universalBase:
                 answer = None
 
                 # Check repetition
                 repetition = findall(r"di(\S+)", lastWord)
-                        if(len(repetition) > 0):
+                if len(repetition) > 0:
                     # We need to repeat something...
                     answer = repeater(repetition[0])
 
@@ -141,118 +172,67 @@ class Listener(Stream):
                         else:
                             # Normal answer
                             answer = answers[mot[0]]
+
+                # If no answer has been found
                 if answer == None:
                     if keys["VERBOSE"]:
                         print(f"{errorMessage} Aucune réponse trouvée.")
+
                 # If an answer has been found
                 else:
                     if keys["VERBOSE"]:
-                                print(f"Envoie d'un {answer[0]}...", end = " ")
+                        print(f"Envoie d'un {answer[0]}...", end=" ")
                     try:
                         # Send the tweet with the answer
-                                self.api.update_status(status = choice(answer), in_reply_to_status_id = json["id"], auto_populate_reply_metadata = True)
-                                print(f"{json['user']['screen_name']} s'est fait {answer[0]} !")
-                            except Exception as error:
-                                error = loads(error.response.text)["errors"][0]
-                                # https://developer.twitter.com/en/support/twitter-api/error-troubleshooting
-                                show_error = True
-                                if error["code"] == 385:
-                                    error["message"] = f"Tweet supprimé ou auteur ({json['user']['screen_name']}) en privé/bloqué."
-                                    show_error = False
-
-                                # Show error only if relevant, always in verbose
-                                if show_error or keys["VERBOSE"]:
-                                    print(f"{errorMessage[:-2]} ({error['code']}) ! {error['message']}")
+                        self.client.create_tweet(
+                            in_reply_to_tweet_id=tweet.id, text=choice(answer))
+                        print(f"{author.username} s'est fait {answer[0]} !")
+                    except errors.Forbidden:
+                        if keys["VERBOSE"]:
+                            print(
+                                f"{errorMessage[:-2]} ! Tweet supprimé ou auteur ({author.username}) en privé/bloqué.")
             else:
                 if keys["VERBOSE"]:
                     print("Annulation car le dernier mot n'est pas intéressant.")
 
-    def do_stuff(self):
-        """Loop for the Listener"""
-        while True:
-            self.q.get()
-            self.q.task_done()
-
     def on_request_error(self, status_code):
-        print(f"{errorMessage[:-2]} ({status_code}) !", end = " ")
-        if status_code == 413:
-            if keys["VERBOSE"]:
-                print("La liste des mots est trop longue (triggerWords).")
-        elif status_code == 420:
+        print(f"{errorMessage[:-2]} ({status_code}) !", end=" ")
+        match status_code:
+            case 420:
                 if keys["VERBOSE"]:
                     print("Déconnecter du flux.")
-        else:
-            print("\n")
+            case 429:
+                if keys["VERBOSE"]:
+                    print("En attente de reconnexion...")
+            case _:
+                # newline
+                print("")
         return False
 
+
 def repeater(word: str) -> str:
     """Formating a word who need to be repeated"""
     # Remove first letter if the first letter is a "S" or a "T"
     # Explanation: Trigger word for the repeater is "di" and sometimes it is
     # "dis", sometimes its "dit", that's why we need to remove this 2 letters
     # from the final answer
-    if word[0] == 's' or word[0] == 't':
+    if word[0] == "s" or word[0] == "t":
         word = word[1:]
 
     # Random format from the base answer
     return createBaseAnswers(word)
 
-def getFriendsID(api: API, users: list[str]) -> list:
+
+def getFriends(client: Client, users: list[str]) -> list:
     """Get all friends of choosen users"""
-    liste = []
+    friends_list = []
     # Get IDs of the user's friends
     for user in users:
-        liste.extend(api.get_friend_ids(screen_name=user))
-    return list(set(liste))
+        user_id = client.get_user(username=user, user_auth=True).data.id
+        friends_list.extend(client.get_users_following(
+            id=user_id, user_auth=True))
+    return friends_list[0]
 
-def getIDs(api: API, users: list[str]) -> list:
-    """Get all the ID of users"""
-    liste = []
-    # Get IDs of the users
-    for user in users:
-        liste.append(api.get_user(screen_name=user)._json["id"])
-    return list(set(liste))
-
-def seniority(date: str) -> bool:
-    """Return True only if the given string date is less than one day old"""
-    # Convert string format to datetime format
-    datetimeObject = datetime.strptime(date, "%a %b %d %H:%M:%S +0000 %Y")
-    # Twitter give us an UTC time
-    datetimeObject = datetimeObject.replace(tzinfo = timezone("UTC"))
-    # time now in UTC minus the time we got to get the age of the date
-    age = datetime.now(timezone("UTC")) - datetimeObject
-    # False if older than a day, else True
-    return False if age.days >= 1 else True
-
-def generateWords(array: list[str]) -> list:
-    """
-    Retrieves all possible combinations for the given list and returns the result as a list
-
-    This is used for the filter in the stream (before calling the Listener::on_status)
-    """
-    quoiListe = []
-
-    for text in array:
-        # Add all combinations
-        # Example for 'oui': ['OUI', 'OUi', 'OuI', 'Oui', 'oUI', 'oUi', 'ouI', 'oui']
-        #
-        # -> Depends on: from itertools import product
-        # -> Problem : Create a too long list (+1000 words, max is 400)
-        # -> Cf. https://developer.twitter.com/en/docs/twitter-api/v1/tweets/filter-realtime/overview
-        #
-        # quoiListe.extend(list(map(''.join, product(*zip(text.upper(), text.lower())))))
-
-        if text.lower() not in quoiListe:
-            # Word in lowercase
-            quoiListe.append(text.lower())
-        if text.upper() not in quoiListe:
-            # Word in uppercase
-            quoiListe.append(text.upper())
-        if text.capitalize() not in quoiListe:
-            # Word capitalized
-            quoiListe.append(text.capitalize())
-
-    return quoiListe
 
 def createBaseTrigger(lists: list[list]) -> list:
     """Merges all given lists into one"""
@@ -261,6 +241,7 @@ def createBaseTrigger(lists: list[list]) -> list:
         listing.extend(liste)
     return list(set(listing))
 
+
 def createBaseAnswers(word: str) -> list:
     """Generates default answers for a given word"""
     irritating_word = [
@@ -270,6 +251,7 @@ def createBaseAnswers(word: str) -> list:
     ]
 
     return [
+        # Assuming the first element of this list is always word, don't change it
         word,
         f"({word})",
         word.upper(),
@@ -277,47 +259,152 @@ def createBaseAnswers(word: str) -> list:
         f"{word}...",
     ]
 
-def start():
-    """Start the bot"""
-    auth = OAuth1UserHandler(keys["CONSUMER_KEY"], keys["CONSUMER_SECRET"])
-    auth.set_access_token(keys["TOKEN"], keys["TOKEN_SECRET"])
 
-    api = API(auth)
+def createClient(consumer_key, consumer_secret, access_token, access_token_secret) -> Client:
+    """Create a client for the Twitter API v2"""
+    client = Client(
+        consumer_key=consumer_key,
+        consumer_secret=consumer_secret,
+        access_token=access_token,
+        access_token_secret=access_token_secret
+    )
 
     if keys["VERBOSE"]:
         try:
-            api.verify_credentials()
-            print(f"Authentification réussie en tant que", end = " ")
-        except:
-            print("Erreur d'authentification.")
-            exit(1)
-        print(f"@{api.verify_credentials().screen_name}.")
+            print(
+                f"Authentification réussie en tant que @{client.get_me().data.username}.\n")
 
-    if keys['WHITELIST'] == []:
+            # Compte ignorés
+            if keys["WHITELIST"] == []:
                 whitelist = "Aucun"
             else:
                 whitelist = f"@{', @'.join(keys['WHITELIST'])}"
             print(f"Liste des comptes ignorés : {whitelist}.")
 
-    stream = Listener(
-        consumer_key=keys["CONSUMER_KEY"],
-        consumer_secret=keys["CONSUMER_SECRET"],
-        access_token=keys["TOKEN"],
-        access_token_secret=keys["TOKEN_SECRET"],
-        api=api,
-        users=keys["PSEUDOS"],
-        forcelist=keys["FORCELIST"]
+            # Compte forcés
+            if keys["FORCELIST"] == []:
+                forcelist = "Aucun"
+            else:
+                forcelist = f"@{', @'.join(keys['FORCELIST'])}"
+            print(f"Liste des comptes forcés : {forcelist}.")
+
+            # Compte aux following suivis
+            if keys["PSEUDOS"] == []:
+                pseudos = "Aucun"
+            else:
+                pseudos = f"@{', @'.join(keys['PSEUDOS'])}"
+            print(
+                f"Les comptes suivis par ces comptes sont traqués : {pseudos}.\n")
+
+            print(
+                "Notez que si un compte est dans la whitelist, il sera dans tout les cas ignoré.\n")
+        except:
+            print("Erreur d'authentification.")
+            exit(1)
+
+    return client
+
+
+def create_rules(tracked_users: list[str]) -> list[str]:
+    """Create rules for tracking users, by respecting the twitter API policies"""
+    rules = []
+
+    # Repeating rules
+    repeat = "-is:retweet lang:fr ("
+
+    # Buffer
+    buffer = repeat
+
+    tracked_users.sort()
+
+    # Track users
+    for user in tracked_users:
+        # Check if the rule don't exceeds the maximum length of a rule (512)
+        # 5 is len of "from:"
+        # 1 is len for closing parenthesis
+        if len(buffer) + len(user) + 5 + 1 > 512:
+            rules.append(buffer[:-4] + ")")
+            buffer = repeat
+        buffer += f"from:{user} OR "
+
+    if len(buffer) > 0:
+        rules.append(buffer[:-4] + ")")
+
+    if len(rules) > 25:
+        raise BufferError("Too much rules.")
+
+    return rules
+
+
+def start():
+    """Start the bot"""
+    client = createClient(
+        keys["CONSUMER_KEY"],
+        keys["CONSUMER_SECRET"],
+        keys["TOKEN"],
+        keys["TOKEN_SECRET"],
     )
-    stream.filter(track = triggerWords, languages = ["fr"], stall_warnings = True, threaded = True)
+
+    # Only track specifics users
+    # Including users in forcelist and removing users in whitelist
+    tracked_users = [
+        i for i in [
+            user.data["username"] for user in getFriends(client, keys["PSEUDOS"])
+        ] + keys["FORCELIST"] if i not in keys["WHITELIST"]
+    ]
+
+    stream = Listener(keys["BEARER_TOKEN"], client)
+
+    # Gathering rules
+    rules = [rule for rule in create_rules(tracked_users)]
+
+    # Check if rules already exists
+    old_rules = stream.get_rules().data
+    old_rules_values = []
+    if old_rules:
+        old_rules_values = [rule.value for rule in old_rules]
+    need_changes = False
+    # Same amount of rules
+    if len(old_rules_values) == len(rules):
+        for rule in rules:
+            # Check if Twitter doesn't know the rule and change rules if needed
+            if rule not in old_rules_values:
+                need_changes = True
+                break
+    else:
+        need_changes = True
+
+    if need_changes:
+        if keys["VERBOSE"]:
+            print("Changes needed... ", end=" ")
+        # Clean old rules
+        if old_rules:
+            if keys["VERBOSE"]:
+                print("deleting old rules... ", end=" ")
+            stream.delete_rules([rule.id for rule in old_rules])
+
+        # Add new rules
+        if keys["VERBOSE"]:
+            print("sending new filter to Twitter.")
+        stream.add_rules([StreamRule(rule) for rule in rules])
+
+    # Apply the filter
+    stream.filter(tweet_fields=["author_id", "conversation_id"])
+
 
 if __name__ == "__main__":
     """
-    TOKEN is the Access Token available in the Authentication Tokens section under Access Token and Secret sub-heading.
-    TOKEN_SECRET is the Access Token Secret available in the Authentication Tokens section under Access Token and Secret sub-heading.
-    CONSUMER_KEY is the API Key available in the Consumer Keys section.
-    CONSUMER_SECRET is the API Secret Key available in the Consumer Keys section.
+    TOKEN is the Access Token available in the Authentication Tokens section under the Access Token and Secret sub-heading
+    TOKEN_SECRET is the Access Token Secret available in the Authentication Tokens section under the Access Token and Secret sub-heading
+    CONSUMER_KEY is the API Key available in the Consumer Keys section under the API Key and Secret sub-heading
+    CONSUMER_SECRET is the API Secret Key available in the Consumer Keys section under the API Key and Secret sub-heading
+    BEARER_TOKEN is the Bearer Token available in the Authentication Tokens section under the Bearer Token sub-heading
     --
-    PSEUDO is the PSEUDO of the account you want to listen to snipe.
+    PSEUDOS is a list of account you want to listen, all of his·er following (guys followed by PSEUDO) will be sniped
+    WHITELSIT is a list of account who are protected from the bot
+    FORCELIST is a list of account who are targeted by the bot, if user is in the whitelist, he·r will be ignored
+    ---
+    VERBOSE enable some debugs log
     """
     # Error message
     errorMessage = "Une erreur survient !"
@@ -490,11 +577,9 @@ if __name__ == "__main__":
     # List of all the trigger words
     universalBase = createBaseTrigger(list(base.values()))
 
-    # List of all the triggers words's variations
-    triggerWords = generateWords(universalBase)
-
     # Loading environment variables
-    keys = load(["TOKEN", "TOKEN_SECRET", "CONSUMER_KEY", "CONSUMER_SECRET", "PSEUDOS", "VERBOSE", "WHITELIST", "FORCELIST"])
+    keys = load(["TOKEN", "TOKEN_SECRET", "CONSUMER_KEY", "CONSUMER_SECRET",
+                "BEARER_TOKEN", "PSEUDOS", "VERBOSE", "WHITELIST", "FORCELIST"])
 
     # Start the bot
     start()