from __future__ import annotations from ..typing import Messages, CreateResult from ..providers.base_provider import AbstractProvider, ProviderModelMixin import time, uuid, random, json from requests import Session from .openai.new import ( get_config, get_answer_token, process_turnstile, get_requirements_token ) def format_conversation(messages: list): conversation = [] for message in messages: conversation.append({ 'id': str(uuid.uuid4()), 'author': { 'role': message['role'], }, 'content': { 'content_type': 'text', 'parts': [ message['content'], ], }, 'metadata': { 'serialization_metadata': { 'custom_symbol_offsets': [], }, }, 'create_time': round(time.time(), 3), }) return conversation def init_session(user_agent): session = Session() cookies = { '_dd_s': '', } headers = { 'accept': '*/*', 'accept-language': 'en-US,en;q=0.8', 'cache-control': 'no-cache', 'pragma': 'no-cache', 'priority': 'u=0, i', 'sec-ch-ua': '"Not)A;Brand";v="99", "Google Chrome";v="127", "Chromium";v="127"', 'sec-ch-ua-arch': '"arm"', 'sec-ch-ua-bitness': '"64"', 'sec-ch-ua-mobile': '?0', 'sec-ch-ua-model': '""', 'sec-ch-ua-platform': '"macOS"', 'sec-ch-ua-platform-version': '"14.4.0"', 'sec-fetch-dest': 'document', 'sec-fetch-mode': 'navigate', 'sec-fetch-site': 'none', 'sec-fetch-user': '?1', 'upgrade-insecure-requests': '1', 'user-agent': user_agent, } session.get('https://chatgpt.com/', cookies=cookies, headers=headers) return session class ChatGpt(AbstractProvider, ProviderModelMixin): label = "ChatGpt" working = True supports_message_history = True supports_system_message = True supports_stream = True models = [ 'gpt-4o', 'gpt-4o-mini', 'gpt-4', 'gpt-4-turbo', 'chatgpt-4o-latest', ] @classmethod def create_completion( cls, model: str, messages: Messages, stream: bool, **kwargs ) -> CreateResult: if model in [ 'gpt-4o', 'gpt-4o-mini', 'gpt-4', 'gpt-4-turbo', 'chatgpt-4o-latest' ]: model = 'auto' elif model in [ 'gpt-3.5-turbo' ]: model = 'text-davinci-002-render-sha' else: raise ValueError(f"Invalid model: {model}") user_agent = 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/127.0.0.0 Safari/537.36' session: Session = init_session(user_agent) config = get_config(user_agent) pow_req = get_requirements_token(config) headers = { 'accept': '*/*', 'accept-language': 'en-US,en;q=0.8', 'content-type': 'application/json', 'oai-device-id': f'{uuid.uuid4()}', 'oai-language': 'en-US', 'origin': 'https://chatgpt.com', 'priority': 'u=1, i', 'referer': 'https://chatgpt.com/', 'sec-ch-ua-mobile': '?0', 'sec-ch-ua-platform': '"Linux"', 'sec-fetch-dest': 'empty', 'sec-fetch-mode': 'cors', 'sec-fetch-site': 'same-origin', 'sec-gpc': '1', 'user-agent': f'{user_agent}' } response = session.post('https://chatgpt.com/backend-anon/sentinel/chat-requirements', headers=headers, json={'p': pow_req}) # Додана перевірка на статус відповідей, якщо "незвична активність" if response.status_code != 200: print(f"Request failed with status: {response.status_code}") print(f"Response content: {response.content}") return response_data = response.json() if "detail" in response_data and "Unusual activity" in response_data["detail"]: print(f"Blocked due to unusual activity: {response_data['detail']}") return turnstile = response_data.get('turnstile', {}) turnstile_required = turnstile.get('required') pow_conf = response_data.get('proofofwork', {}) if turnstile_required: turnstile_dx = turnstile.get('dx') turnstile_token = process_turnstile(turnstile_dx, pow_req) headers = headers | { 'openai-sentinel-turnstile-token' : turnstile_token, 'openai-sentinel-chat-requirements-token': response_data.get('token'), 'openai-sentinel-proof-token' : get_answer_token( pow_conf.get('seed'), pow_conf.get('difficulty'), config ) } json_data = { 'action': 'next', 'messages': format_conversation(messages), 'parent_message_id': str(uuid.uuid4()), 'model': 'auto', 'timezone_offset_min': -120, 'suggestions': [ 'Can you help me create a personalized morning routine that would help increase my productivity throughout the day? Start by asking me about my current habits and what activities energize me in the morning.', 'Could you help me plan a relaxing day that focuses on activities for rejuvenation? To start, can you ask me what my favorite forms of relaxation are?', 'I have a photoshoot tomorrow. Can you recommend me some colors and outfit options that will look good on camera?', 'Make up a 5-sentence story about "Sharky", a tooth-brushing shark superhero. Make each sentence a bullet point.', ], 'history_and_training_disabled': False, 'conversation_mode': { 'kind': 'primary_assistant', }, 'force_paragen': False, 'force_paragen_model_slug': '', 'force_nulligen': False, 'force_rate_limit': False, 'reset_rate_limits': False, 'websocket_request_id': str(uuid.uuid4()), 'system_hints': [], 'force_use_sse': True, 'conversation_origin': None, 'client_contextual_info': { 'is_dark_mode': True, 'time_since_loaded': random.randint(22,33), 'page_height': random.randint(600, 900), 'page_width': random.randint(500, 800), 'pixel_ratio': 2, 'screen_height': random.randint(800, 1200), 'screen_width': random.randint(1200, 2000), }, } # Додав паузу між запитами для уникнення блокувань через частоту time.sleep(2) response = session.post('https://chatgpt.com/backend-anon/conversation', headers=headers, json=json_data, stream=True) replace = '' for line in response.iter_lines(): if line: decoded_line = line.decode() print(f"Received line: {decoded_line}") # Debugging output if decoded_line.startswith('data:'): json_string = decoded_line[6:] # Extract JSON part after 'data:' if json_string.strip(): # Check if there's actual content try: data = json.loads(json_string) except json.JSONDecodeError as e: print(f"Error decoding JSON: {e}, content: {json_string}") continue if data.get('message').get('author').get('role') == 'assistant': tokens = (data.get('message').get('content').get('parts')[0]) yield tokens.replace(replace, '') replace = tokens