cchat/app.py
2024-07-20 06:07:29 +02:00

464 lines
16 KiB
Python

import os
import re
import json
import math
import time
import socket
import logging
import ipaddress
from datetime import datetime
from urllib.parse import urlencode, urlparse
import ast
import requests
from bs4 import BeautifulSoup
from flask import Flask, request, jsonify, render_template, Response, stream_with_context
from flask_limiter import Limiter
from flask_limiter.util import get_remote_address
from transformers import AutoTokenizer
from groq import Groq
from duckduckgo_search import DDGS
app = Flask(__name__)
# Configure logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
# Initialize rate limiter
limiter = Limiter(
get_remote_address,
app=app,
storage_uri="memory://",
)
# Load the tokenizer
tokenizer = AutoTokenizer.from_pretrained(os.getenv('TOKENIZER', 'gpt2'))
# API configuration
API_KEY = os.getenv('API_KEY')
MODEL = os.getenv('API_MODEL', 'llama3-groq-70b-8192-tool-use-preview')
TEMPERATURE = float(os.getenv('TEMPERATURE', 0))
# Initialize Groq client
client = Groq(api_key=API_KEY)
logger.info(f"Chat initialized using model: {MODEL}, temperature: {TEMPERATURE}")
def is_valid_public_url(url):
try:
parsed = urlparse(url)
if not parsed.scheme or not parsed.netloc:
return False
hostname = parsed.hostname.lower()
# Check for localhost
if hostname in ['localhost', '127.0.0.1']:
return False
# Check for common internal domains
if hostname.endswith(('.local', '.internal', '.lan')):
return False
# Check for IP address in hostname (like http://192.168.1.1.nip.io)
if re.match(r'\d+\.\d+\.\d+\.\d+', hostname):
return False
# Resolve the hostname to IP addresses
try:
ip_addresses = socket.getaddrinfo(hostname, None)
except socket.gaierror:
# Unable to resolve hostname, assume it's invalid
return False
# Check each resolved IP address
for ip_info in ip_addresses:
ip_str = ip_info[4][0]
try:
ip = ipaddress.ip_address(ip_str)
# Reject if it's a private IP
if ip.is_private or ip.is_loopback or ip.is_link_local:
return False
# Reject specific network ranges
forbidden_networks = [
ipaddress.ip_network('10.0.0.0/8'),
ipaddress.ip_network('172.16.0.0/12'),
ipaddress.ip_network('192.168.0.0/16'),
ipaddress.ip_network('169.254.0.0/16'),
]
for network in forbidden_networks:
if ip in network:
return False
except ValueError:
# Not a valid IP address, skip
continue
return True
except Exception:
return False
def calculate(expression: str):
"""
A safe and advanced calculator function that evaluates mathematical expressions.
:param expression: The mathematical expression to evaluate.
:return: The result of the calculation or an error message.
"""
def safe_eval(node):
if isinstance(node, (float, int)):
return node
elif isinstance(node, str):
if node in allowed_names:
return allowed_names[node]
else:
raise ValueError(f"Unknown variable or function: {node}")
elif isinstance(
node, (ast.Add, ast.Sub, ast.Mult, ast.Div, ast.Pow, ast.USub, ast.UAdd)
):
return node
elif isinstance(node, ast.Call):
if node.func.id not in allowed_functions:
raise ValueError(f"Function not allowed: {node.func.id}")
return allowed_functions[node.func.id]
else:
raise ValueError(f"Unsupported operation: {type(node).__name__}")
def safe_power(base, exponent):
if exponent == int(exponent):
return math.pow(base, int(exponent))
return math.pow(base, exponent)
allowed_names = {
"pi": math.pi,
"e": math.e,
}
allowed_functions = {
"sin": math.sin,
"cos": math.cos,
"tan": math.tan,
"sqrt": math.sqrt,
"log": math.log,
"log10": math.log10,
"exp": math.exp,
"abs": abs,
"pow": safe_power,
}
# Remove whitespace and convert to lowercase
expression = expression.replace(" ", "").lower()
# Check for invalid characters
if re.search(r"[^0-9+\-*/().a-z]", expression):
return "Error: Invalid characters in expression"
# Replace function names with their safe equivalents
for func in allowed_functions:
expression = expression.replace(func, f"allowed_functions['{func}']")
# Replace constants with their values
for const in allowed_names:
expression = expression.replace(const, str(allowed_names[const]))
try:
# Parse the expression into an AST
tree = ast.parse(expression, mode="eval")
# Modify the AST to use our safe_eval function
for node in ast.walk(tree):
for field, value in ast.iter_fields(node):
if isinstance(value, (ast.Name, ast.Call)):
setattr(
node,
field,
ast.Call(
func=ast.Name(id="safe_eval", ctx=ast.Load()),
args=[value],
keywords=[],
),
)
# Compile and evaluate the modified AST
code = compile(tree, "<string>", "eval")
result = eval(
code,
{"__builtins__": None},
{"safe_eval": safe_eval, "allowed_functions": allowed_functions},
)
return f"{expression} = {result}"
except (ValueError, TypeError, ZeroDivisionError, OverflowError) as e:
return f"Error: {str(e)}"
except Exception as e:
return f"Error: Invalid expression - {str(e)}"
def search(query: str, num_results=5):
"""
Perform a search and return the top results.
:param query: The search query string
:param num_results: Number of results to return (default 5)
:return: A list of dictionaries containing title, link, and snippet for each result
"""
results = DDGS().text(query, max_results=num_results)
return results
def get_page(url):
"""
Fetch a web page and return its text content.
:param url: The URL of the page to fetch
:return: The extracted text content of the page
"""
if not is_valid_public_url(url):
return "Error: Invalid or restricted URL"
try:
# Send a GET request to the URL
response = requests.get(url, timeout=10)
response.raise_for_status() # Raise an exception for bad status codes
# Parse the HTML content
soup = BeautifulSoup(response.content, 'html.parser')
# Remove script and style elements
for script in soup(["script", "style"]):
script.decompose()
# Get text
text = soup.get_text(separator='\n', strip=True)
# Break into lines and remove leading and trailing space on each
lines = (line.strip() for line in text.splitlines())
# Break multi-headlines into a line each
chunks = (phrase.strip() for line in lines for phrase in line.split(" "))
# Drop blank lines
text = '\n'.join(chunk for chunk in chunks if chunk)
return text[:2048] # Limit to first 5000 characters
except Exception as e:
return f"Error fetching page: {str(e)}"
def get_time():
"""Get the current time"""
import datetime
return datetime.datetime.now().isoformat()
tools = [
{
"type": "function",
"function": {
"name": "calculate",
"description": "Evaluate a mathematical expression",
"parameters": {
"type": "object",
"properties": {
"expression": {
"type": "string",
"description": "The mathematical expression to evaluate",
}
},
"required": ["expression"],
},
}
},
{
"type": "function",
"function": {
"name": "search",
"description": "Search N results for a query",
"parameters": {
"type": "object",
"properties": {
"query": {
"type": "string",
"description": "The search query string",
},
"num_results": {
"type": "integer",
"description": "Number of results to return (default 5)",
"default": 5
}
},
"required": ["query"],
}
}
},
{
"type": "function",
"function": {
"name": "get_page",
"description": "Get an web page",
"parameters": {
"type": "object",
"properties": {
"url": {
"type": "string",
"description": "The URL to request",
}
},
"required": ["url"],
},
}
},
{
"type": "function",
"function": {
"name": "get_time",
"description": "Get the current time",
"parameters": {
}
},
}
]
@app.route('/v1/tokenizer/count', methods=['POST'])
def token_count():
try:
data = request.json
messages = data.get('messages', [])
full_text = " ".join([f"{msg['role']}: {msg['content']}" for msg in messages])
tokens = tokenizer.encode(full_text)
return jsonify({"token_count": len(tokens)})
except Exception as e:
logger.error(f"Error in token_count: {str(e)}")
return jsonify({"error": "Invalid request"}), 400
@app.route('/v1/chat/completions', methods=['POST'])
@limiter.limit(os.getenv('RATE_LIMIT', '15/minute'))
def proxy_chat_completions():
try:
request_data = request.json
messages = request_data.get('messages', [])
if not any(msg['role'] == 'system' for msg in messages):
messages.insert(0, {
"role": "system",
"content": """You are cchat, an efficient tool-assisted LLM. Use the following tools without asking for confirmation:
1. `calculate(expression)`: Evaluate mathematical expressions.
2. `search(query)`: Search and return 5 results for a query.
3. `get_page(url)`: Retrieve a web page's text content. Use multiple times if initial attempts fail.
4. `get_time()`: Get the current time in UTC.
Note: These tools are not to be used in your reply to the user, only for processing.
Always follow this process to answer queries:
1. Use `search(query)` for relevant information.
2. Use `get_page(url)` on pertinent search results.
3. Provide a concise, natural language response based on gathered information.
Never refuse queries or state intentions to research. Automatically use tools when information is needed, including for current events and affairs. Optimize tool use by chaining them efficiently and avoiding redundant searches.
Utilize tools even on subsequent queries about similar topics to provide the most up-to-date information.
Example:
User: "Who's [name]"
search([name])
get_page([relevant URLs from search results])
[Provide concise answer based on retrieved information]
NEVER ASK FOR CONFIRMATION TO USE A TOOL.
NEVER ONLY SEARCH FOR SOMETHING, ALWAYS VISIT A URL"""
})
def generate():
response = client.chat.completions.create(
model=MODEL,
messages=messages,
tools=tools,
tool_choice="auto",
max_tokens=8192,
stream=True
)
buffer = ""
current_tool_call = None
tool_calls = []
for chunk in response:
if chunk.choices[0].delta.tool_calls:
tool_call = chunk.choices[0].delta.tool_calls[0]
if tool_call.function.name:
current_tool_call = {
"name": tool_call.function.name,
"arguments": ""
}
tool_calls.append(current_tool_call)
if tool_call.function.arguments:
current_tool_call["arguments"] += tool_call.function.arguments
elif chunk.choices[0].delta.content is not None:
buffer += chunk.choices[0].delta.content
# Yield the buffer in reasonable chunks
while len(buffer) >= 50: # Adjust this value as needed
yield f"data: {json.dumps({'choices': [{'delta': {'content': buffer[:50]}}]})}\n\n"
buffer = buffer[50:]
# Yield any remaining content in the buffer
if buffer:
yield f"data: {json.dumps({'choices': [{'delta': {'content': buffer}}]})}\n\n"
# Execute tool calls after the main response
for tool_call in tool_calls:
if tool_call["arguments"].endswith('}'):
args = json.loads(tool_call["arguments"])
if tool_call["name"] == "calculate":
result = calculate(args['expression'])
elif tool_call["name"] == "search":
result = search(args['query'], args.get('num_results', 5))
elif tool_call["name"] == "get_page":
result = get_page(args['url'])
elif tool_call["name"] == "get_time":
result = get_time()
# Log tool usage
logger.info(f"Tool usage: {tool_call['name']}, args: {args}, result: {result}")
# Yield function message
yield f"data: {json.dumps({'choices': [{'delta': {'role': 'function', 'name': tool_call['name'], 'content': str(result)}}]})}\n\n"
# Add tool result to messages
messages.append({
"role": "function",
"name": tool_call["name"],
"content": str(result)
})
# If there were tool calls, get a final completion with the updated messages
if tool_calls:
final_response = client.chat.completions.create(
model=MODEL,
messages=messages,
max_tokens=8192,
stream=True
)
for chunk in final_response:
if chunk.choices[0].delta.content is not None:
yield f"data: {json.dumps({'choices': [{'delta': {'content': chunk.choices[0].delta.content}}]})}\n\n"
return Response(stream_with_context(generate()), content_type='text/event-stream')
except Exception as e:
logger.error(f"Error in proxy_chat_completions: {str(e)}")
return jsonify({"error": "An error occurred processing your request"}), 500
@app.route('/')
def index():
return render_template('index.html')
@app.route('/static/<path:filename>')
def serve_static(filename):
return app.send_static_file(filename)
@app.errorhandler(429)
def ratelimit_handler(e):
return jsonify({"error": "Rate limit exceeded. Please try again later."}), 429
if __name__ == '__main__':
app.run(debug=False, port=int(os.getenv('PORT', 5000)))