import asyncio import json import random import re import string import time import uuid from datetime import datetime, timezone from typing import Any, Dict, List, Optional import boto3 import httpx import tiktoken import platform import hashlib from fastapi import HTTPException from api.config import ( MODEL_MAPPING, get_headers_api_chat, get_headers_chat, BASE_URL, AGENT_MODE, TRENDING_AGENT_MODE, MODEL_PREFIXES ) from api.logger import setup_logger from api.models import ChatRequest from api.validate import getHid # Import the asynchronous getHid function logger = setup_logger(__name__) # --------------------------------------------- # CLOUDFLARE R2 CONFIGURATION # --------------------------------------------- R2_ACCESS_KEY_ID = "df9c9eb87e850a8eb27afd3968077b42" R2_SECRET_ACCESS_KEY = "14b08b0855263bb63d2618da3a6537e1b0446d89d51da03a568620b1e5342ea8" R2_ENDPOINT_URL = "https://f2f92ac53fae792c4155f6e93a514989.r2.cloudflarestorage.com" R2_BUCKET_NAME = "snapzion" # We always store replaced URLs in one file named snapzion.txt R2_REPLACED_URLS_KEY = "snapzion.txt" s3 = boto3.client( "s3", endpoint_url=R2_ENDPOINT_URL, aws_access_key_id=R2_ACCESS_KEY_ID, aws_secret_access_key=R2_SECRET_ACCESS_KEY, ) # Example blocked message BLOCKED_MESSAGE = ( "Generated by BLACKBOX.AI, try unlimited chat https://www.blackbox.ai " "and for API requests replace https://www.blackbox.ai with https://api.blackbox.ai" ) # --------------------------------------------- # RANDOM USER-DATA GENERATION # --------------------------------------------- def get_random_name_email_customer(): """ Generate a random name, email, and customer ID. The customer ID keeps the same length format as 'cus_Rldf7IKdNhdhiw'. """ first_names = ["Alice", "Bob", "Carol", "David", "Evelyn", "Frank", "Grace", "Hector", "Ivy", "Jackie"] last_names = ["Smith", "Johnson", "Davis", "Miller", "Thompson", "Garcia", "Brown", "Wilson", "Martin", "Clark"] random_name = f"{random.choice(first_names)} {random.choice(last_names)}" email_username = ''.join(random.choices(string.ascii_lowercase + string.digits, k=8)) random_email = f"{email_username}@gmail.com" suffix_length = len("Rldf7IKdNhdhiw") suffix_chars = string.ascii_letters + string.digits random_suffix = ''.join(random.choice(suffix_chars) for _ in range(suffix_length)) random_customer_id = f"cus_{random_suffix}" return random_name, random_email, random_customer_id # --------------------------------------------- # HELPER FUNCTIONS # --------------------------------------------- def generate_system_fingerprint() -> str: raw_data = f"{platform.node()}-{time.time()}-{uuid.uuid4()}" short_hash = hashlib.md5(raw_data.encode()).hexdigest()[:12] return f"fp_{short_hash}" def get_last_user_prompt(messages: List[Any]) -> str: for msg in reversed(messages): if msg.role == "user": if isinstance(msg.content, str): return msg.content.strip() elif isinstance(msg.content, list): for item in msg.content: if item.get("type") == "text": return item.get("text", "").strip() return "" def upload_replaced_urls_to_r2(urls: List[str], alt_text: str = "") -> None: if not urls: logger.info("No replaced or final Snapzion URLs to store. Skipping snapzion.txt update.") return existing_data = "" try: response = s3.get_object(Bucket=R2_BUCKET_NAME, Key=R2_REPLACED_URLS_KEY) existing_data = response['Body'].read().decode('utf-8') logger.info("Successfully read existing snapzion.txt from R2.") except s3.exceptions.NoSuchKey: logger.info("snapzion.txt does not exist yet. Will create a new one.") except Exception as e: logger.error(f"Error reading snapzion.txt from R2: {e}") alt_text = alt_text.strip() markdown_lines = [f"![{alt_text}]({url})" for url in urls] to_append = "\n".join(markdown_lines) if existing_data.strip(): updated_content = existing_data + "\n" + to_append else: updated_content = to_append try: s3.put_object( Bucket=R2_BUCKET_NAME, Key=R2_REPLACED_URLS_KEY, Body=updated_content.encode("utf-8"), ContentType="text/plain", ) logger.info(f"Appended {len(urls)} new URLs to snapzion.txt in R2 (in Markdown format).") except Exception as e: logger.error(f"Failed to upload replaced URLs to R2: {e}") def calculate_tokens(text: str, model: str) -> int: try: encoding = tiktoken.encoding_for_model(model) tokens = encoding.encode(text) return len(tokens) except KeyError: logger.warning(f"Model '{model}' not supported by tiktoken for token counting. Using a generic method.") return len(text.split()) def create_chat_completion_data( content: str, model: str, timestamp: int, request_id: str, system_fingerprint: str, prompt_tokens: int = 0, completion_tokens: int = 0, finish_reason: Optional[str] = None, function_call: Optional[Dict] = None, ) -> Dict[str, Any]: usage = None if finish_reason == "stop": usage = { "prompt_tokens": prompt_tokens, "completion_tokens": completion_tokens, "total_tokens": prompt_tokens + completion_tokens, } return { "id": request_id, "object": "chat.completion.chunk", "created": timestamp, "model": model, "system_fingerprint": system_fingerprint, "choices": [{ "index": 0, "delta": { "content": content if not function_call else None, "role": "assistant", "function_call": function_call }, "finish_reason": finish_reason }], "usage": usage, } def message_to_dict(message, model_prefix: Optional[str] = None, tools: Optional[List[Dict]] = None) -> Dict[str, Any]: """ Convert a ChatRequest message to a dict for the request payload. Supports function calling, images, and model prefixes. """ content = "" images_data = [] image_urls = [] # Handle content based on type if isinstance(message.content, list): for item in message.content: if item.get("type") == "text": content = item.get("text", "").strip() elif item.get("type") == "image_url" and len(images_data) < 3: image_url = item.get("image_url", {}).get("url", "") if image_url: # Generate unique file path (assuming .jpg, adjust if needed) file_path = f"MultipleFiles/{uuid.uuid4().hex}.jpg" images_data.append({"filePath": file_path, "contents": image_url}) image_urls.append({"image_url": {"url": image_url}}) elif isinstance(message.content, str): content = message.content.strip() # Apply model prefix to text content if model_prefix and content: content = f"{model_prefix} {content}" # Create payload with both formats base_message = {"role": message.role, "content": content} if images_data: base_message["data"] = { "imageBase64": images_data[0]["contents"] if images_data else "", "fileText": "", "title": "snapshot", "imagesData": images_data } # Add additional image_url entries for testing for img in image_urls[1:]: # Skip the first image (already in imageBase64) base_message["content"] = base_message.get("content", "") # Preserve text base_message.setdefault("content", []).append(img) return base_message if images_data else {"role": message.role, "content": content} def strip_model_prefix(content: str, model_prefix: Optional[str] = None) -> str: if model_prefix and content.startswith(model_prefix): logger.debug(f"Stripping prefix '{model_prefix}' from content.") return content[len(model_prefix):].strip() return content # --------------------------------------------- # STREAMING RESPONSE HANDLER # --------------------------------------------- async def process_streaming_response(request: ChatRequest): system_fingerprint = generate_system_fingerprint() random_name, random_email, random_customer_id = get_random_name_email_customer() request_id = f"chatcmpl-{uuid.uuid4()}" logger.info(f"Processing request (stream) {request_id} - Model: {request.model}") agent_mode = AGENT_MODE.get(request.model, {}) trending_agent_mode = TRENDING_AGENT_MODE.get(request.model, {}) model_prefix = MODEL_PREFIXES.get(request.model, "") headers_api_chat = get_headers_api_chat(BASE_URL) if request.model == "o1-preview": delay_seconds = random.randint(1, 60) logger.info(f"Delay {delay_seconds}s for model 'o1-preview' (Request: {request_id})") await asyncio.sleep(delay_seconds) h_value = await getHid() if not h_value: logger.error("No h-value for validation.") raise HTTPException(status_code=500, detail="Missing h-value.") messages = [message_to_dict(msg, model_prefix=model_prefix, tools=request.tools) for msg in request.messages] json_data = { "agentMode": agent_mode, "clickedAnswer2": False, "clickedAnswer3": False, "clickedForceWebSearch": False, "codeInterpreterMode": False, "codeModelMode": True, "githubToken": "", "deepSearchMode": False, "domains": None, "id": request_id, "imageGenerationMode": False, "isChromeExt": False, "isMicMode": False, "isPremium": True, "isMemoryEnabled": False, "maxTokens": request.max_tokens, "messages": messages, "mobileClient": False, "playgroundTemperature": request.temperature, "playgroundTopP": request.top_p, "previewToken": None, "trendingAgentMode": trending_agent_mode, "userId": None, "userSelectedModel": MODEL_MAPPING.get(request.model, request.model), "userSystemPrompt": None, "validated": h_value, "visitFromDelta": False, "webSearchModePrompt": False, "vscodeClient": False, "customProfile": {"name": "", "occupation": "", "traits": [], "additionalInfo": "", "enableNewChats": False}, "webSearchModeOption": {"autoMode": False, "webMode": False, "offlineMode": True}, "session": { "user": {"name": random_name, "email": random_email, "image": "https://lh3.googleusercontent.com/a/...=s96-c", "subscriptionStatus": "PREMIUM"}, "expires": datetime.now(timezone.utc).isoformat(timespec='milliseconds').replace('+00:00', 'Z'), "subscriptionCache": {"customerId": random_customer_id, "status": "PREMIUM", "isTrialSubscription": "False", "expiryTimestamp": 1744652408, "lastChecked": int(time.time() * 1000)}, "beastMode": False, "reasoningMode": False, "designerMode": False, "workspaceId": "", }, } prompt_tokens = 0 for msg in messages: if "content" in msg: prompt_tokens += calculate_tokens(msg["content"], request.model) if "data" in msg and "imagesData" in msg["data"]: for image_data in msg["data"]["imagesData"]: prompt_tokens += calculate_tokens(image_data["contents"], request.model) completion_tokens = 0 final_snapzion_links = [] async with httpx.AsyncClient() as client: try: async with client.stream("POST", f"{BASE_URL}/api/chat", headers=headers_api_chat, json=json_data, timeout=100) as response: response.raise_for_status() async for chunk in response.aiter_text(): timestamp = int(datetime.now().timestamp()) if not chunk: continue if chunk.startswith("$@$v=undefined-rv1$@$"): chunk = chunk[21:] if BLOCKED_MESSAGE in chunk: logger.info(f"Blocked message found in chunk (Request: {request_id}).") chunk = chunk.replace(BLOCKED_MESSAGE, "").strip() if not chunk: continue if "https://storage.googleapis.com" in chunk: chunk = chunk.replace("https://storage.googleapis.com", "https://cdn.snapzion.com") snapzion_urls = re.findall(r"(https://cdn\.snapzion\.com[^\s\)]+)", chunk) if snapzion_urls: final_snapzion_links.extend(snapzion_urls) cleaned_content = strip_model_prefix(chunk, model_prefix) completion_tokens += calculate_tokens(cleaned_content, request.model) # Handle function call responses function_call = None if cleaned_content and cleaned_content.startswith("{"): try: function_call = json.loads(cleaned_content) cleaned_content = None # Content must be null for function calls except json.JSONDecodeError: pass yield "data: " + json.dumps(create_chat_completion_data( cleaned_content, request.model, timestamp, request_id, system_fingerprint, prompt_tokens, completion_tokens, finish_reason=None, function_call=function_call )) + "\n\n" yield "data: " + json.dumps(create_chat_completion_data("", request.model, timestamp, request_id, system_fingerprint, prompt_tokens, completion_tokens, "stop")) + "\n\n" yield "data: [DONE]\n\n" except httpx.HTTPStatusError as e: logger.error(f"HTTP error (stream) {request_id}: {e}") error_message = f"HTTP error occurred: {e}" try: error_details = e.response.json() error_message += f" Details: {error_details}" except ValueError: error_message += f" Response body: {e.response.text}" yield "data: " + json.dumps(create_chat_completion_data(error_message, request.model, int(datetime.now().timestamp()), request_id, system_fingerprint, prompt_tokens, completion_tokens, "error")) + "\n\n" yield "data: [DONE]\n\n" except httpx.RequestError as e: logger.error(f"Request error (stream) {request_id}: {e}") error_message = f"Request error occurred: {e}" yield "data: " + json.dumps(create_chat_completion_data(error_message, request.model, int(datetime.now().timestamp()), request_id, system_fingerprint, prompt_tokens, completion_tokens, "error")) + "\n\n" yield "data: [DONE]\n\n" except Exception as e: logger.error(f"Unhandled error (stream) {request_id}: {e}") error_message = f"An unexpected error occurred: {e}" yield "data: " + json.dumps(create_chat_completion_data(error_message, request.model, int(datetime.now().timestamp()), request_id, system_fingerprint, prompt_tokens, completion_tokens, "error")) + "\n\n" yield "data: [DONE]\n\n" last_user_prompt = get_last_user_prompt(request.messages) upload_replaced_urls_to_r2(final_snapzion_links, alt_text=last_user_prompt) # --------------------------------------------- # NON-STREAMING RESPONSE HANDLER # --------------------------------------------- async def process_non_streaming_response(request: ChatRequest): system_fingerprint = generate_system_fingerprint() random_name, random_email, random_customer_id = get_random_name_email_customer() request_id = f"chatcmpl-{uuid.uuid4()}" logger.info(f"Processing request (non-stream) {request_id} - Model: {request.model}") agent_mode = AGENT_MODE.get(request.model, {}) trending_agent_mode = TRENDING_AGENT_MODE.get(request.model, {}) model_prefix = MODEL_PREFIXES.get(request.model, "") headers_api_chat = get_headers_api_chat(BASE_URL) headers_chat = get_headers_chat(BASE_URL, next_action=str(uuid.uuid4()), next_router_state_tree=json.dumps([""])) if request.model == "o1-preview": delay_seconds = random.randint(20, 60) logger.info(f"Delay {delay_seconds}s for 'o1-preview' (Request: {request_id})") await asyncio.sleep(delay_seconds) h_value = "00f37b34-a166-4efb-bce5-1312d87f2f94" if not h_value: logger.error("Failed to retrieve h-value.") raise HTTPException(status_code=500, detail="Missing h-value.") messages = [message_to_dict(msg, model_prefix=model_prefix, tools=request.tools) for msg in request.messages] json_data = { "agentMode": agent_mode, "clickedAnswer2": False, "clickedAnswer3": False, "clickedForceWebSearch": False, "codeInterpreterMode": False, "codeModelMode": True, "githubToken": "", "deepSearchMode": False, "domains": None, "id": request_id, "imageGenerationMode": False, "isChromeExt": False, "isMicMode": False, "isPremium": True, "isMemoryEnabled": False, "maxTokens": request.max_tokens, "messages": messages, "mobileClient": False, "playgroundTemperature": request.temperature, "playgroundTopP": request.top_p, "previewToken": None, "trendingAgentMode": trending_agent_mode, "userId": None, "userSelectedModel": MODEL_MAPPING.get(request.model, request.model), "userSystemPrompt": None, "validated": h_value, "visitFromDelta": False, "webSearchModePrompt": False, "vscodeClient": False, "customProfile": {"name": "", "occupation": "", "traits": [], "additionalInfo": "", "enableNewChats": False}, "webSearchModeOption": {"autoMode": False, "webMode": False, "offlineMode": True}, "session": { "user": {"name": random_name, "email": random_email, "image": "https://lh3.googleusercontent.com/a/...=s96-c", "subscriptionStatus": "PREMIUM"}, "expires": datetime.now(timezone.utc).isoformat(timespec='milliseconds').replace('+00:00', 'Z'), "subscriptionCache": {"customerId": random_customer_id, "status": "PREMIUM", "isTrialSubscription": "False", "expiryTimestamp": 1744652408, "lastChecked": int(time.time() * 1000)}, "beastMode": False, "reasoningMode": False, "designerMode": False, "workspaceId": "", }, } prompt_tokens = 0 for msg in messages: if "content" in msg: prompt_tokens += calculate_tokens(msg["content"], request.model) if "data" in msg and "imagesData" in msg["data"]: for image_data in msg["data"]["imagesData"]: prompt_tokens += calculate_tokens(image_data["contents"], request.model) full_response = "" final_snapzion_links = [] async with httpx.AsyncClient() as client: try: async with client.stream("POST", f"{BASE_URL}/api/chat", headers=headers_api_chat, json=json_data) as response: response.raise_for_status() async for chunk in response.aiter_text(): full_response += chunk except httpx.HTTPStatusError as e: logger.error(f"HTTP error (non-stream) {request_id}: {e}") error_message = f"HTTP error occurred: {e}" try: error_details = e.response.json() error_message += f" Details: {error_details}" except ValueError: error_message += f" Response body: {e.response.text}" return { "id": request_id, "object": "chat.completion", "created": int(datetime.now().timestamp()), "model": request.model, "system_fingerprint": system_fingerprint, "choices": [{"index": 0, "message": {"role": "assistant", "content": error_message}, "finish_reason": "error"}], "usage": {"prompt_tokens": prompt_tokens, "completion_tokens": 0, "total_tokens": prompt_tokens}, } except httpx.RequestError as e: logger.error(f"Request error (non-stream) {request_id}: {e}") error_message = f"Request error occurred: {e}" return { "id": request_id, "object": "chat.completion", "created": int(datetime.now().timestamp()), "model": request.model, "system_fingerprint": system_fingerprint, "choices": [{"index": 0, "message": {"role": "assistant", "content": error_message}, "finish_reason": "error"}], "usage": {"prompt_tokens": prompt_tokens, "completion_tokens": 0, "total_tokens": prompt_tokens}, } except Exception as e: logger.error(f"Unexpected error (non-stream) {request_id}: {e}") error_message = f"An unexpected error occurred: {e}" return { "id": request_id, "object": "chat.completion", "created": int(datetime.now().timestamp()), "model": request.model, "system_fingerprint": system_fingerprint, "choices": [{"index": 0, "message": {"role": "assistant", "content": error_message}, "finish_reason": "error"}], "usage": {"prompt_tokens": prompt_tokens, "completion_tokens": 0, "total_tokens": prompt_tokens}, } if full_response.startswith("$@$v=undefined-rv1$@$"): full_response = full_response[21:] if BLOCKED_MESSAGE in full_response: full_response = full_response.replace(BLOCKED_MESSAGE, "").strip() if not full_response: raise HTTPException(status_code=500, detail="Blocked message in response.") if "https://storage.googleapis.com" in full_response: full_response = full_response.replace("https://storage.googleapis.com", "https://cdn.snapzion.com") snapzion_urls = re.findall(r"(https://cdn\.snapzion\.com[^\s\)]+)", full_response) for link in snapzion_urls: final_snapzion_links.append(link) cleaned_full_response = strip_model_prefix(full_response, model_prefix) completion_tokens = calculate_tokens(cleaned_full_response, request.model) last_user_prompt = get_last_user_prompt(request.messages) upload_replaced_urls_to_r2(final_snapzion_links, alt_text=last_user_prompt) return { "id": request_id, "object": "chat.completion", "created": int(datetime.now().timestamp()), "model": request.model, "system_fingerprint": system_fingerprint, "choices": [{"index": 0, "message": {"role": "assistant", "content": cleaned_full_response}, "finish_reason": "stop"}], "usage": {"prompt_tokens": prompt_tokens, "completion_tokens": completion_tokens, "total_tokens": prompt_tokens + completion_tokens}, }