open-interpreter-ai
/
01
mirror of https://github.com/OpenInterpreter/01
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Add updated list of linted files

Browse Source
pull/223/head
Davy Peter Braun 9 months ago
parent 2bf19e9f57
commit 403a29f0d6
13 changed files with 532 additions and 383 deletions
Show Stats Download Patch File Download Diff File

196
software/source/clients/base_device.py
Unescape View File

@ -1,23 +1,18 @@
from dotenv import load_dotenv from dotenv import load_dotenv
load_dotenv() # take environment variables from .env. load_dotenv() # take environment variables from .env.
import os import os
import asyncio import asyncio
import threading import threading
import os
import pyaudio import pyaudio
from starlette.websockets import WebSocket
from queue import Queue
from pynput import keyboard from pynput import keyboard
import json import json
import traceback import traceback
import websockets import websockets
import queue import queue
import pydub
import ast
from pydub import AudioSegment from pydub import AudioSegment
from pydub.playback import play from pydub.playback import play
import io
import time import time
import wave import wave
import tempfile import tempfile
@ -25,7 +20,10 @@ from datetime import datetime
import cv2 import cv2
import base64 import base64
import platform import platform
from interpreter import interpreter # Just for code execution. Maybe we should let people do from interpreter.computer import run? from interpreter import (
interpreter,
) # Just for code execution. Maybe we should let people do from interpreter.computer import run?
# In the future, I guess kernel watching code should be elsewhere? Somewhere server / client agnostic? # In the future, I guess kernel watching code should be elsewhere? Somewhere server / client agnostic?
from ..server.utils.kernel import put_kernel_messages_into_queue from ..server.utils.kernel import put_kernel_messages_into_queue
from ..server.utils.get_system_info import get_system_info from ..server.utils.get_system_info import get_system_info
@ -33,6 +31,7 @@ from ..server.utils.process_utils import kill_process_tree
from ..server.utils.logs import setup_logging from ..server.utils.logs import setup_logging
from ..server.utils.logs import logger from ..server.utils.logs import logger
setup_logging() setup_logging()
os.environ["STT_RUNNER"] = "server" os.environ["STT_RUNNER"] = "server"
@ -51,11 +50,11 @@ RECORDING = False # Flag to control recording state
SPACEBAR_PRESSED = False # Flag to track spacebar press state SPACEBAR_PRESSED = False # Flag to track spacebar press state
# Camera configuration # Camera configuration
CAMERA_ENABLED = os.getenv('CAMERA_ENABLED', False) CAMERA_ENABLED = os.getenv("CAMERA_ENABLED", False)
if type(CAMERA_ENABLED) == str: if type(CAMERA_ENABLED) == str:
CAMERA_ENABLED = (CAMERA_ENABLED.lower() == "true") CAMERA_ENABLED = CAMERA_ENABLED.lower() == "true"
CAMERA_DEVICE_INDEX = int(os.getenv('CAMERA_DEVICE_INDEX', 0)) CAMERA_DEVICE_INDEX = int(os.getenv("CAMERA_DEVICE_INDEX", 0))
CAMERA_WARMUP_SECONDS = float(os.getenv('CAMERA_WARMUP_SECONDS', 0)) CAMERA_WARMUP_SECONDS = float(os.getenv("CAMERA_WARMUP_SECONDS", 0))
# Specify OS # Specify OS
current_platform = get_system_info() current_platform = get_system_info()
@ -66,6 +65,7 @@ p = pyaudio.PyAudio()
send_queue = queue.Queue() send_queue = queue.Queue()
class Device: class Device:
def __init__(self): def __init__(self):
self.pressed_keys = set() self.pressed_keys = set()
@ -89,23 +89,28 @@ class Device:
if ret: if ret:
temp_dir = tempfile.gettempdir() temp_dir = tempfile.gettempdir()
image_path = os.path.join(temp_dir, f"01_photo_{datetime.now().strftime('%Y%m%d%H%M%S%f')}.png") image_path = os.path.join(
temp_dir, f"01_photo_{datetime.now().strftime('%Y%m%d%H%M%S%f')}.png"
)
self.captured_images.append(image_path) self.captured_images.append(image_path)
cv2.imwrite(image_path, frame) cv2.imwrite(image_path, frame)
logger.info(f"Camera image captured to {image_path}") logger.info(f"Camera image captured to {image_path}")
logger.info(f"You now have {len(self.captured_images)} images which will be sent along with your next audio message.") logger.info(
f"You now have {len(self.captured_images)} images which will be sent along with your next audio message."
)
else: else:
logger.error(f"Error: Couldn't capture an image from camera ({camera_index})") logger.error(
f"Error: Couldn't capture an image from camera ({camera_index})"
)
cap.release() cap.release()
return image_path return image_path
def encode_image_to_base64(self, image_path): def encode_image_to_base64(self, image_path):
"""Encodes an image file to a base64 string.""" """Encodes an image file to a base64 string."""
with open(image_path, "rb") as image_file: with open(image_path, "rb") as image_file:
return base64.b64encode(image_file.read()).decode('utf-8') return base64.b64encode(image_file.read()).decode("utf-8")
def add_image_to_send_queue(self, image_path): def add_image_to_send_queue(self, image_path):
"""Encodes an image and adds an LMC message to the send queue with the image data.""" """Encodes an image and adds an LMC message to the send queue with the image data."""
@ -114,7 +119,7 @@ class Device:
"role": "user", "role": "user",
"type": "image", "type": "image",
"format": "base64.png", "format": "base64.png",
"content": base64_image "content": base64_image,
} }
send_queue.put(image_message) send_queue.put(image_message)
# Delete the image file from the file system after sending it # Delete the image file from the file system after sending it
@ -126,7 +131,6 @@ class Device:
self.add_image_to_send_queue(image_path) self.add_image_to_send_queue(image_path)
self.captured_images.clear() # Clear the list after sending self.captured_images.clear() # Clear the list after sending
async def play_audiosegments(self): async def play_audiosegments(self):
"""Plays them sequentially.""" """Plays them sequentially."""
while True: while True:
@ -141,27 +145,35 @@ class Device:
except: except:
logger.info(traceback.format_exc()) logger.info(traceback.format_exc())
def record_audio(self): def record_audio(self):
if os.getenv("STT_RUNNER") == "server":
if os.getenv('STT_RUNNER') == "server":
# STT will happen on the server. we're sending audio. # STT will happen on the server. we're sending audio.
send_queue.put({"role": "user", "type": "audio", "format": "bytes.wav", "start": True}) send_queue.put(
elif os.getenv('STT_RUNNER') == "client": {"role": "user", "type": "audio", "format": "bytes.wav", "start": True}
)
elif os.getenv("STT_RUNNER") == "client":
# STT will happen here, on the client. we're sending text. # STT will happen here, on the client. we're sending text.
send_queue.put({"role": "user", "type": "message", "start": True}) send_queue.put({"role": "user", "type": "message", "start": True})
else: else:
raise Exception("STT_RUNNER must be set to either 'client' or 'server'.") raise Exception("STT_RUNNER must be set to either 'client' or 'server'.")
"""Record audio from the microphone and add it to the queue.""" """Record audio from the microphone and add it to the queue."""
stream = p.open(format=FORMAT, channels=CHANNELS, rate=RATE, input=True, frames_per_buffer=CHUNK) stream = p.open(
format=FORMAT,
channels=CHANNELS,
rate=RATE,
input=True,
frames_per_buffer=CHUNK,
)
print("Recording started...") print("Recording started...")
global RECORDING global RECORDING
# Create a temporary WAV file to store the audio data # Create a temporary WAV file to store the audio data
temp_dir = tempfile.gettempdir() temp_dir = tempfile.gettempdir()
wav_path = os.path.join(temp_dir, f"audio_{datetime.now().strftime('%Y%m%d%H%M%S%f')}.wav") wav_path = os.path.join(
wav_file = wave.open(wav_path, 'wb') temp_dir, f"audio_{datetime.now().strftime('%Y%m%d%H%M%S%f')}.wav"
)
wav_file = wave.open(wav_path, "wb")
wav_file.setnchannels(CHANNELS) wav_file.setnchannels(CHANNELS)
wav_file.setsampwidth(p.get_sample_size(FORMAT)) wav_file.setsampwidth(p.get_sample_size(FORMAT))
wav_file.setframerate(RATE) wav_file.setframerate(RATE)
@ -178,17 +190,30 @@ class Device:
duration = wav_file.getnframes() / RATE duration = wav_file.getnframes() / RATE
if duration < 0.3: if duration < 0.3:
# Just pressed it. Send stop message # Just pressed it. Send stop message
if os.getenv('STT_RUNNER') == "client": if os.getenv("STT_RUNNER") == "client":
send_queue.put({"role": "user", "type": "message", "content": "stop"}) send_queue.put({"role": "user", "type": "message", "content": "stop"})
send_queue.put({"role": "user", "type": "message", "end": True}) send_queue.put({"role": "user", "type": "message", "end": True})
else: else:
send_queue.put({"role": "user", "type": "audio", "format": "bytes.wav", "content": ""}) send_queue.put(
send_queue.put({"role": "user", "type": "audio", "format": "bytes.wav", "end": True}) {
"role": "user",
"type": "audio",
"format": "bytes.wav",
"content": "",
}
)
send_queue.put(
{
"role": "user",
"type": "audio",
"format": "bytes.wav",
"end": True,
}
)
else: else:
self.queue_all_captured_images() self.queue_all_captured_images()
if os.getenv('STT_RUNNER') == "client": if os.getenv("STT_RUNNER") == "client":
# THIS DOES NOT WORK. We moved to this very cool stt_service, llm_service # THIS DOES NOT WORK. We moved to this very cool stt_service, llm_service
# way of doing things. stt_wav is not a thing anymore. Needs work to work # way of doing things. stt_wav is not a thing anymore. Needs work to work
@ -199,12 +224,19 @@ class Device:
send_queue.put({"role": "user", "type": "message", "end": True}) send_queue.put({"role": "user", "type": "message", "end": True})
else: else:
# Stream audio # Stream audio
with open(wav_path, 'rb') as audio_file: with open(wav_path, "rb") as audio_file:
byte_data = audio_file.read(CHUNK) byte_data = audio_file.read(CHUNK)
while byte_data: while byte_data:
send_queue.put(byte_data) send_queue.put(byte_data)
byte_data = audio_file.read(CHUNK) byte_data = audio_file.read(CHUNK)
send_queue.put({"role": "user", "type": "audio", "format": "bytes.wav", "end": True}) send_queue.put(
{
"role": "user",
"type": "audio",
"format": "bytes.wav",
"end": True,
}
)
if os.path.exists(wav_path): if os.path.exists(wav_path):
os.remove(wav_path) os.remove(wav_path)
@ -227,24 +259,27 @@ class Device:
if keyboard.Key.space in self.pressed_keys: if keyboard.Key.space in self.pressed_keys:
self.toggle_recording(True) self.toggle_recording(True)
elif {keyboard.Key.ctrl, keyboard.KeyCode.from_char('c')} <= self.pressed_keys: elif {keyboard.Key.ctrl, keyboard.KeyCode.from_char("c")} <= self.pressed_keys:
logger.info("Ctrl+C pressed. Exiting...") logger.info("Ctrl+C pressed. Exiting...")
kill_process_tree() kill_process_tree()
os._exit(0) os._exit(0)
def on_release(self, key): def on_release(self, key):
"""Detect spacebar release and 'c' key press for camera, and handle key release.""" """Detect spacebar release and 'c' key press for camera, and handle key release."""
self.pressed_keys.discard(key) # Remove the released key from the key press tracking set self.pressed_keys.discard(
key
) # Remove the released key from the key press tracking set
if key == keyboard.Key.space: if key == keyboard.Key.space:
self.toggle_recording(False) self.toggle_recording(False)
elif CAMERA_ENABLED and key == keyboard.KeyCode.from_char('c'): elif CAMERA_ENABLED and key == keyboard.KeyCode.from_char("c"):
self.fetch_image_from_camera() self.fetch_image_from_camera()
async def message_sender(self, websocket): async def message_sender(self, websocket):
while True: while True:
message = await asyncio.get_event_loop().run_in_executor(None, send_queue.get) message = await asyncio.get_event_loop().run_in_executor(
None, send_queue.get
)
if isinstance(message, bytes): if isinstance(message, bytes):
await websocket.send(message) await websocket.send(message)
else: else:
@ -257,7 +292,9 @@ class Device:
async def exec_ws_communication(websocket): async def exec_ws_communication(websocket):
if CAMERA_ENABLED: if CAMERA_ENABLED:
print("\nHold the spacebar to start recording. Press 'c' to capture an image from the camera. Press CTRL-C to exit.") print(
"\nHold the spacebar to start recording. Press 'c' to capture an image from the camera. Press CTRL-C to exit."
)
else: else:
print("\nHold the spacebar to start recording. Press CTRL-C to exit.") print("\nHold the spacebar to start recording. Press CTRL-C to exit.")
@ -280,7 +317,6 @@ class Device:
# At this point, we have our message # At this point, we have our message
if message["type"] == "audio" and message["format"].startswith("bytes"): if message["type"] == "audio" and message["format"].startswith("bytes"):
# Convert bytes to audio file # Convert bytes to audio file
audio_bytes = message["content"] audio_bytes = message["content"]
@ -294,13 +330,13 @@ class Device:
# 16,000 Hz frame rate # 16,000 Hz frame rate
frame_rate=16000, frame_rate=16000,
# mono sound # mono sound
channels=1 channels=1,
) )
self.audiosegments.append(audio) self.audiosegments.append(audio)
# Run the code if that's the client's job # Run the code if that's the client's job
if os.getenv('CODE_RUNNER') == "client": if os.getenv("CODE_RUNNER") == "client":
if message["type"] == "code" and "end" in message: if message["type"] == "code" and "end" in message:
language = message["format"] language = message["format"]
code = message["content"] code = message["content"]
@ -308,7 +344,7 @@ class Device:
send_queue.put(result) send_queue.put(result)
if is_win10(): if is_win10():
logger.info('Windows 10 detected') logger.info("Windows 10 detected")
# Workaround for Windows 10 not latching to the websocket server. # Workaround for Windows 10 not latching to the websocket server.
# See https://github.com/OpenInterpreter/01/issues/197 # See https://github.com/OpenInterpreter/01/issues/197
try: try:
@ -329,42 +365,46 @@ class Device:
await asyncio.sleep(2) await asyncio.sleep(2)
async def start_async(self): async def start_async(self):
# Configuration for WebSocket # Configuration for WebSocket
WS_URL = f"ws://{self.server_url}" WS_URL = f"ws://{self.server_url}"
# Start the WebSocket communication # Start the WebSocket communication
asyncio.create_task(self.websocket_communication(WS_URL)) asyncio.create_task(self.websocket_communication(WS_URL))
# Start watching the kernel if it's your job to do that # Start watching the kernel if it's your job to do that
if os.getenv('CODE_RUNNER') == "client": if os.getenv("CODE_RUNNER") == "client":
asyncio.create_task(put_kernel_messages_into_queue(send_queue)) asyncio.create_task(put_kernel_messages_into_queue(send_queue))
asyncio.create_task(self.play_audiosegments()) asyncio.create_task(self.play_audiosegments())
# If Raspberry Pi, add the button listener, otherwise use the spacebar # If Raspberry Pi, add the button listener, otherwise use the spacebar
if current_platform.startswith("raspberry-pi"): if current_platform.startswith("raspberry-pi"):
logger.info("Raspberry Pi detected, using button on GPIO pin 15") logger.info("Raspberry Pi detected, using button on GPIO pin 15")
# Use GPIO pin 15 # Use GPIO pin 15
pindef = ["gpiochip4", "15"] # gpiofind PIN15 pindef = ["gpiochip4", "15"] # gpiofind PIN15
print("PINDEF", pindef) print("PINDEF", pindef)
# HACK: needs passwordless sudo # HACK: needs passwordless sudo
process = await asyncio.create_subprocess_exec("sudo", "gpiomon", "-brf", *pindef, stdout=asyncio.subprocess.PIPE) process = await asyncio.create_subprocess_exec(
while True: "sudo", "gpiomon", "-brf", *pindef, stdout=asyncio.subprocess.PIPE
line = await process.stdout.readline() )
if line: while True:
line = line.decode().strip() line = await process.stdout.readline()
if "FALLING" in line: if line:
self.toggle_recording(False) line = line.decode().strip()
elif "RISING" in line: if "FALLING" in line:
self.toggle_recording(True) self.toggle_recording(False)
else: elif "RISING" in line:
break self.toggle_recording(True)
else: else:
# Keyboard listener for spacebar press/release break
listener = keyboard.Listener(on_press=self.on_press, on_release=self.on_release) else:
listener.start() # Keyboard listener for spacebar press/release
listener = keyboard.Listener(
on_press=self.on_press, on_release=self.on_release
)
listener.start()
def start(self): def start(self):
if os.getenv('TEACH_MODE') != "True": if os.getenv("TEACH_MODE") != "True":
asyncio.run(self.start_async()) asyncio.run(self.start_async())
p.terminate() p.terminate()

5
software/source/server/conftest.py
Unescape View File

@ -1,8 +1,5 @@
import os
import sys
import pytest import pytest
from source.server.i import configure_interpreter from source.server.i import configure_interpreter
from unittest.mock import Mock
from interpreter import OpenInterpreter from interpreter import OpenInterpreter
from fastapi.testclient import TestClient from fastapi.testclient import TestClient
from .server import app from .server import app
@ -16,4 +13,4 @@ def client():
@pytest.fixture @pytest.fixture
def mock_interpreter(): def mock_interpreter():
interpreter = configure_interpreter(OpenInterpreter()) interpreter = configure_interpreter(OpenInterpreter())
return interpreter return interpreter

38
software/source/server/i.py
Unescape View File

@ -1,11 +1,11 @@
from dotenv import load_dotenv from dotenv import load_dotenv
import os
load_dotenv() # take environment variables from .env. load_dotenv() # take environment variables from .env.
import os
import glob import glob
import time import time
import json import json
from pathlib import Path
from interpreter import OpenInterpreter from interpreter import OpenInterpreter
import shutil import shutil
@ -47,7 +47,7 @@ The user's current task list (it might be empty) is: {{ tasks }}
When the user completes the current task, you should remove it from the list and read the next item by running `tasks = tasks[1:]\ntasks[0]`. Then, tell the user what the next task is. When the user completes the current task, you should remove it from the list and read the next item by running `tasks = tasks[1:]\ntasks[0]`. Then, tell the user what the next task is.
When the user tells you about a set of tasks, you should intelligently order tasks, batch similar tasks, and break down large tasks into smaller tasks (for this, you should consult the user and get their permission to break it down). Your goal is to manage the task list as intelligently as possible, to make the user as efficient and non-overwhelmed as possible. They will require a lot of encouragement, support, and kindness. Don't say too much about what's ahead of them— just try to focus them on each step at a time. When the user tells you about a set of tasks, you should intelligently order tasks, batch similar tasks, and break down large tasks into smaller tasks (for this, you should consult the user and get their permission to break it down). Your goal is to manage the task list as intelligently as possible, to make the user as efficient and non-overwhelmed as possible. They will require a lot of encouragement, support, and kindness. Don't say too much about what's ahead of them— just try to focus them on each step at a time.
After starting a task, you should check in with the user around the estimated completion time to see if the task is completed. After starting a task, you should check in with the user around the estimated completion time to see if the task is completed.
To do this, schedule a reminder based on estimated completion time using the function `schedule(message="Your message here.", start="8am")`, WHICH HAS ALREADY BEEN IMPORTED. YOU DON'T NEED TO IMPORT THE `schedule` FUNCTION. IT IS AVAILABLE. You'll receive the message at the time you scheduled it. If the user says to monitor something, simply schedule it with an interval of a duration that makes sense for the problem by specifying an interval, like this: `schedule(message="Your message here.", interval="5m")` To do this, schedule a reminder based on estimated completion time using the function `schedule(message="Your message here.", start="8am")`, WHICH HAS ALREADY BEEN IMPORTED. YOU DON'T NEED TO IMPORT THE `schedule` FUNCTION. IT IS AVAILABLE. You'll receive the message at the time you scheduled it. If the user says to monitor something, simply schedule it with an interval of a duration that makes sense for the problem by specifying an interval, like this: `schedule(message="Your message here.", interval="5m")`
@ -182,7 +182,6 @@ Try multiple methods before saying the task is impossible. **You can do it!**
def configure_interpreter(interpreter: OpenInterpreter): def configure_interpreter(interpreter: OpenInterpreter):
### SYSTEM MESSAGE ### SYSTEM MESSAGE
interpreter.system_message = system_message interpreter.system_message = system_message
@ -205,7 +204,6 @@ def configure_interpreter(interpreter: OpenInterpreter):
"Please provide more information.", "Please provide more information.",
] ]
# Check if required packages are installed # Check if required packages are installed
# THERE IS AN INCONSISTENCY HERE. # THERE IS AN INCONSISTENCY HERE.
@ -259,7 +257,6 @@ def configure_interpreter(interpreter: OpenInterpreter):
time.sleep(2) time.sleep(2)
print("Attempting to start OS control anyway...\n\n") print("Attempting to start OS control anyway...\n\n")
# Should we explore other options for ^ these kinds of tags? # Should we explore other options for ^ these kinds of tags?
# Like: # Like:
@ -295,12 +292,8 @@ def configure_interpreter(interpreter: OpenInterpreter):
# if chunk.get("format") != "active_line": # if chunk.get("format") != "active_line":
# print(chunk.get("content")) # print(chunk.get("content"))
import os
from platformdirs import user_data_dir from platformdirs import user_data_dir
# Directory paths # Directory paths
repo_skills_dir = os.path.join(os.path.dirname(__file__), "skills") repo_skills_dir = os.path.join(os.path.dirname(__file__), "skills")
user_data_skills_dir = os.path.join(user_data_dir("01"), "skills") user_data_skills_dir = os.path.join(user_data_dir("01"), "skills")
@ -314,22 +307,21 @@ def configure_interpreter(interpreter: OpenInterpreter):
src_file = os.path.join(repo_skills_dir, filename) src_file = os.path.join(repo_skills_dir, filename)
dst_file = os.path.join(user_data_skills_dir, filename) dst_file = os.path.join(user_data_skills_dir, filename)
shutil.copy2(src_file, dst_file) shutil.copy2(src_file, dst_file)
interpreter.computer.debug = True interpreter.computer.debug = True
interpreter.computer.skills.path = user_data_skills_dir interpreter.computer.skills.path = user_data_skills_dir
# Import skills # Import skills
interpreter.computer.save_skills = False interpreter.computer.save_skills = False
for file in glob.glob(os.path.join(interpreter.computer.skills.path, "*.py")): for file in glob.glob(os.path.join(interpreter.computer.skills.path, "*.py")):
code_to_run = "" code_to_run = ""
with open(file, "r") as f: with open(file, "r") as f:
code_to_run += f.read() + "\n" code_to_run += f.read() + "\n"
interpreter.computer.run("python", code_to_run) interpreter.computer.run("python", code_to_run)
interpreter.computer.save_skills = True
interpreter.computer.save_skills = True
# Initialize user's task list # Initialize user's task list
interpreter.computer.run( interpreter.computer.run(
@ -354,17 +346,21 @@ def configure_interpreter(interpreter: OpenInterpreter):
### MISC SETTINGS ### MISC SETTINGS
interpreter.auto_run = True interpreter.auto_run = True
interpreter.computer.languages = [l for l in interpreter.computer.languages if l.name.lower() in ["applescript", "shell", "zsh", "bash", "python"]] interpreter.computer.languages = [
l
for l in interpreter.computer.languages
if l.name.lower() in ["applescript", "shell", "zsh", "bash", "python"]
]
interpreter.force_task_completion = True interpreter.force_task_completion = True
# interpreter.offline = True # interpreter.offline = True
interpreter.id = 206 # Used to identify itself to other interpreters. This should be changed programmatically so it's unique. interpreter.id = 206 # Used to identify itself to other interpreters. This should be changed programmatically so it's unique.
### RESET conversations/user.json ### RESET conversations/user.json
app_dir = user_data_dir('01') app_dir = user_data_dir("01")
conversations_dir = os.path.join(app_dir, 'conversations') conversations_dir = os.path.join(app_dir, "conversations")
os.makedirs(conversations_dir, exist_ok=True) os.makedirs(conversations_dir, exist_ok=True)
user_json_path = os.path.join(conversations_dir, 'user.json') user_json_path = os.path.join(conversations_dir, "user.json")
with open(user_json_path, 'w') as file: with open(user_json_path, "w") as file:
json.dump([], file) json.dump([], file)
return interpreter return interpreter

268
software/source/server/server.py
Unescape View File

@ -1,9 +1,9 @@
from dotenv import load_dotenv from dotenv import load_dotenv
load_dotenv() # take environment variables from .env. load_dotenv() # take environment variables from .env.
import traceback import traceback
from platformdirs import user_data_dir from platformdirs import user_data_dir
import ast
import json import json
import queue import queue
import os import os
@ -13,9 +13,7 @@ import re
from fastapi import FastAPI, Request from fastapi import FastAPI, Request
from fastapi.responses import PlainTextResponse from fastapi.responses import PlainTextResponse
from starlette.websockets import WebSocket, WebSocketDisconnect from starlette.websockets import WebSocket, WebSocketDisconnect
from pathlib import Path
import asyncio import asyncio
import urllib.parse
from .utils.kernel import put_kernel_messages_into_queue from .utils.kernel import put_kernel_messages_into_queue
from .i import configure_interpreter from .i import configure_interpreter
from interpreter import interpreter from interpreter import interpreter
@ -44,28 +42,29 @@ accumulator = Accumulator()
app = FastAPI() app = FastAPI()
app_dir = user_data_dir('01') app_dir = user_data_dir("01")
conversation_history_path = os.path.join(app_dir, 'conversations', 'user.json') conversation_history_path = os.path.join(app_dir, "conversations", "user.json")
SERVER_LOCAL_PORT = int(os.getenv('SERVER_LOCAL_PORT', 10001)) SERVER_LOCAL_PORT = int(os.getenv("SERVER_LOCAL_PORT", 10001))
# This is so we only say() full sentences # This is so we only say() full sentences
def is_full_sentence(text): def is_full_sentence(text):
return text.endswith(('.', '!', '?')) return text.endswith((".", "!", "?"))
def split_into_sentences(text): def split_into_sentences(text):
return re.split(r'(?<=[.!?])\s+', text) return re.split(r"(?<=[.!?])\s+", text)
# Queues # Queues
from_computer = queue.Queue() # Just for computer messages from the device. Sync queue because interpreter.run is synchronous from_computer = queue.Queue() # Just for computer messages from the device. Sync queue because interpreter.run is synchronous
from_user = asyncio.Queue() # Just for user messages from the device. from_user = asyncio.Queue() # Just for user messages from the device.
to_device = asyncio.Queue() # For messages we send. to_device = asyncio.Queue() # For messages we send.
# Switch code executor to device if that's set # Switch code executor to device if that's set
if os.getenv('CODE_RUNNER') == "device": if os.getenv("CODE_RUNNER") == "device":
# (This should probably just loop through all languages and apply these changes instead) # (This should probably just loop through all languages and apply these changes instead)
class Python: class Python:
@ -79,14 +78,33 @@ if os.getenv('CODE_RUNNER') == "device":
"""Generator that yields a dictionary in LMC Format.""" """Generator that yields a dictionary in LMC Format."""
# Prepare the data # Prepare the data
message = {"role": "assistant", "type": "code", "format": "python", "content": code} message = {
"role": "assistant",
"type": "code",
"format": "python",
"content": code,
}
# Unless it was just sent to the device, send it wrapped in flags # Unless it was just sent to the device, send it wrapped in flags
if not (interpreter.messages and interpreter.messages[-1] == message): if not (interpreter.messages and interpreter.messages[-1] == message):
to_device.put({"role": "assistant", "type": "code", "format": "python", "start": True}) to_device.put(
{
"role": "assistant",
"type": "code",
"format": "python",
"start": True,
}
)
to_device.put(message) to_device.put(message)
to_device.put({"role": "assistant", "type": "code", "format": "python", "end": True}) to_device.put(
{
"role": "assistant",
"type": "code",
"format": "python",
"end": True,
}
)
# Stream the response # Stream the response
logger.info("Waiting for the device to respond...") logger.info("Waiting for the device to respond...")
while True: while True:
@ -109,10 +127,12 @@ if os.getenv('CODE_RUNNER') == "device":
# Configure interpreter # Configure interpreter
interpreter = configure_interpreter(interpreter) interpreter = configure_interpreter(interpreter)
@app.get("/ping") @app.get("/ping")
async def ping(): async def ping():
return PlainTextResponse("pong") return PlainTextResponse("pong")
@app.websocket("/") @app.websocket("/")
async def websocket_endpoint(websocket: WebSocket): async def websocket_endpoint(websocket: WebSocket):
await websocket.accept() await websocket.accept()
@ -145,19 +165,21 @@ async def receive_messages(websocket: WebSocket):
except Exception as e: except Exception as e:
print(str(e)) print(str(e))
return return
if 'text' in data: if "text" in data:
try: try:
data = json.loads(data['text']) data = json.loads(data["text"])
if data["role"] == "computer": if data["role"] == "computer":
from_computer.put(data) # To be handled by interpreter.computer.run from_computer.put(
data
) # To be handled by interpreter.computer.run
elif data["role"] == "user": elif data["role"] == "user":
await from_user.put(data) await from_user.put(data)
else: else:
raise("Unknown role:", data) raise ("Unknown role:", data)
except json.JSONDecodeError: except json.JSONDecodeError:
pass # data is not JSON, leave it as is pass # data is not JSON, leave it as is
elif 'bytes' in data: elif "bytes" in data:
data = data['bytes'] # binary data data = data["bytes"] # binary data
await from_user.put(data) await from_user.put(data)
except WebSocketDisconnect as e: except WebSocketDisconnect as e:
if e.code == 1000: if e.code == 1000:
@ -165,13 +187,13 @@ async def receive_messages(websocket: WebSocket):
return return
else: else:
raise raise
async def send_messages(websocket: WebSocket): async def send_messages(websocket: WebSocket):
while True: while True:
message = await to_device.get() message = await to_device.get()
#print(f"Sending to the device: {type(message)} {str(message)[:100]}") # print(f"Sending to the device: {type(message)} {str(message)[:100]}")
try: try:
if isinstance(message, dict): if isinstance(message, dict):
await websocket.send_json(message) await websocket.send_json(message)
@ -184,8 +206,8 @@ async def send_messages(websocket: WebSocket):
await to_device.put(message) await to_device.put(message)
raise raise
async def listener():
async def listener():
while True: while True:
try: try:
while True: while True:
@ -197,8 +219,6 @@ async def listener():
break break
await asyncio.sleep(1) await asyncio.sleep(1)
message = accumulator.accumulate(chunk) message = accumulator.accumulate(chunk)
if message == None: if message == None:
# Will be None until we have a full message ready # Will be None until we have a full message ready
@ -209,8 +229,11 @@ async def listener():
# At this point, we have our message # At this point, we have our message
if message["type"] == "audio" and message["format"].startswith("bytes"): if message["type"] == "audio" and message["format"].startswith("bytes"):
if (
if "content" not in message or message["content"] == None or message["content"] == "": # If it was nothing / silence / empty "content" not in message
or message["content"] == None
or message["content"] == ""
): # If it was nothing / silence / empty
continue continue
# Convert bytes to audio file # Convert bytes to audio file
@ -222,6 +245,7 @@ async def listener():
if False: if False:
os.system(f"open {audio_file_path}") os.system(f"open {audio_file_path}")
import time import time
time.sleep(15) time.sleep(15)
text = stt(audio_file_path) text = stt(audio_file_path)
@ -239,21 +263,21 @@ async def listener():
continue continue
# Load, append, and save conversation history # Load, append, and save conversation history
with open(conversation_history_path, 'r') as file: with open(conversation_history_path, "r") as file:
messages = json.load(file) messages = json.load(file)
messages.append(message) messages.append(message)
with open(conversation_history_path, 'w') as file: with open(conversation_history_path, "w") as file:
json.dump(messages, file, indent=4) json.dump(messages, file, indent=4)
accumulated_text = "" accumulated_text = ""
if any(
if any([m["type"] == "image" for m in messages]) and interpreter.llm.model.startswith("gpt-"): [m["type"] == "image" for m in messages]
) and interpreter.llm.model.startswith("gpt-"):
interpreter.llm.model = "gpt-4-vision-preview" interpreter.llm.model = "gpt-4-vision-preview"
interpreter.llm.supports_vision = True interpreter.llm.supports_vision = True
for chunk in interpreter.chat(messages, stream=True, display=True):
for chunk in interpreter.chat(messages, stream=True, display=True):
if any([m["type"] == "image" for m in interpreter.messages]): if any([m["type"] == "image" for m in interpreter.messages]):
interpreter.llm.model = "gpt-4-vision-preview" interpreter.llm.model = "gpt-4-vision-preview"
@ -263,18 +287,24 @@ async def listener():
await to_device.put(chunk) await to_device.put(chunk)
# Yield to the event loop, so you actually send it out # Yield to the event loop, so you actually send it out
await asyncio.sleep(0.01) await asyncio.sleep(0.01)
if os.getenv('TTS_RUNNER') == "server": if os.getenv("TTS_RUNNER") == "server":
# Speak full sentences out loud # Speak full sentences out loud
if chunk["role"] == "assistant" and "content" in chunk and chunk["type"] == "message": if (
chunk["role"] == "assistant"
and "content" in chunk
and chunk["type"] == "message"
):
accumulated_text += chunk["content"] accumulated_text += chunk["content"]
sentences = split_into_sentences(accumulated_text) sentences = split_into_sentences(accumulated_text)
# If we're going to speak, say we're going to stop sending text. # If we're going to speak, say we're going to stop sending text.
# This should be fixed probably, we should be able to do both in parallel, or only one. # This should be fixed probably, we should be able to do both in parallel, or only one.
if any(is_full_sentence(sentence) for sentence in sentences): if any(is_full_sentence(sentence) for sentence in sentences):
await to_device.put({"role": "assistant", "type": "message", "end": True}) await to_device.put(
{"role": "assistant", "type": "message", "end": True}
)
if is_full_sentence(sentences[-1]): if is_full_sentence(sentences[-1]):
for sentence in sentences: for sentence in sentences:
await stream_tts_to_device(sentence) await stream_tts_to_device(sentence)
@ -287,32 +317,36 @@ async def listener():
# If we're going to speak, say we're going to stop sending text. # If we're going to speak, say we're going to stop sending text.
# This should be fixed probably, we should be able to do both in parallel, or only one. # This should be fixed probably, we should be able to do both in parallel, or only one.
if any(is_full_sentence(sentence) for sentence in sentences): if any(is_full_sentence(sentence) for sentence in sentences):
await to_device.put({"role": "assistant", "type": "message", "start": True}) await to_device.put(
{"role": "assistant", "type": "message", "start": True}
)
# If we have a new message, save our progress and go back to the top # If we have a new message, save our progress and go back to the top
if not from_user.empty(): if not from_user.empty():
# Check if it's just an end flag. We ignore those. # Check if it's just an end flag. We ignore those.
temp_message = await from_user.get() temp_message = await from_user.get()
if type(temp_message) is dict and temp_message.get("role") == "user" and temp_message.get("end"): if (
type(temp_message) is dict
and temp_message.get("role") == "user"
and temp_message.get("end")
):
# Yup. False alarm. # Yup. False alarm.
continue continue
else: else:
# Whoops! Put that back # Whoops! Put that back
await from_user.put(temp_message) await from_user.put(temp_message)
with open(conversation_history_path, 'w') as file: with open(conversation_history_path, "w") as file:
json.dump(interpreter.messages, file, indent=4) json.dump(interpreter.messages, file, indent=4)
# TODO: is triggering seemingly randomly # TODO: is triggering seemingly randomly
#logger.info("New user message recieved. Breaking.") # logger.info("New user message recieved. Breaking.")
#break # break
# Also check if there's any new computer messages # Also check if there's any new computer messages
if not from_computer.empty(): if not from_computer.empty():
with open(conversation_history_path, "w") as file:
with open(conversation_history_path, 'w') as file:
json.dump(interpreter.messages, file, indent=4) json.dump(interpreter.messages, file, indent=4)
logger.info("New computer message recieved. Breaking.") logger.info("New computer message recieved. Breaking.")
@ -320,6 +354,7 @@ async def listener():
except: except:
traceback.print_exc() traceback.print_exc()
async def stream_tts_to_device(sentence): async def stream_tts_to_device(sentence):
force_task_completion_responses = [ force_task_completion_responses = [
"the task is done", "the task is done",
@ -332,8 +367,8 @@ async def stream_tts_to_device(sentence):
for chunk in stream_tts(sentence): for chunk in stream_tts(sentence):
await to_device.put(chunk) await to_device.put(chunk)
def stream_tts(sentence): def stream_tts(sentence):
audio_file = tts(sentence) audio_file = tts(sentence)
with open(audio_file, "rb") as f: with open(audio_file, "rb") as f:
@ -346,85 +381,106 @@ def stream_tts(sentence):
# Stream the audio # Stream the audio
yield {"role": "assistant", "type": "audio", "format": file_type, "start": True} yield {"role": "assistant", "type": "audio", "format": file_type, "start": True}
for i in range(0, len(audio_bytes), chunk_size): for i in range(0, len(audio_bytes), chunk_size):
chunk = audio_bytes[i:i+chunk_size] chunk = audio_bytes[i : i + chunk_size]
yield chunk yield chunk
yield {"role": "assistant", "type": "audio", "format": file_type, "end": True} yield {"role": "assistant", "type": "audio", "format": file_type, "end": True}
from uvicorn import Config, Server from uvicorn import Config, Server
import os import os
import platform
from importlib import import_module from importlib import import_module
# these will be overwritten # these will be overwritten
HOST = '' HOST = ""
PORT = 0 PORT = 0
@app.on_event("startup") @app.on_event("startup")
async def startup_event(): async def startup_event():
server_url = f"{HOST}:{PORT}" server_url = f"{HOST}:{PORT}"
print("") print("")
print_markdown(f"\n*Ready.*\n") print_markdown("\n*Ready.*\n")
print("") print("")
@app.on_event("shutdown") @app.on_event("shutdown")
async def shutdown_event(): async def shutdown_event():
print_markdown("*Server is shutting down*") print_markdown("*Server is shutting down*")
async def main(server_host, server_port, llm_service, model, llm_supports_vision, llm_supports_functions, context_window, max_tokens, temperature, tts_service, stt_service):
global HOST
global PORT
PORT = server_port
HOST = server_host
# Setup services
application_directory = user_data_dir('01')
services_directory = os.path.join(application_directory, 'services')
service_dict = {'llm': llm_service, 'tts': tts_service, 'stt': stt_service} async def main(
server_host,
# Create a temp file with the session number server_port,
session_file_path = os.path.join(user_data_dir('01'), '01-session.txt') llm_service,
with open(session_file_path, 'w') as session_file: model,
session_id = int(datetime.datetime.now().timestamp() * 1000) llm_supports_vision,
session_file.write(str(session_id)) llm_supports_functions,
context_window,
for service in service_dict: max_tokens,
temperature,
service_directory = os.path.join(services_directory, service, service_dict[service]) tts_service,
stt_service,
# This is the folder they can mess around in ):
config = {"service_directory": service_directory} global HOST
global PORT
if service == "llm": PORT = server_port
config.update({ HOST = server_host
# Setup services
application_directory = user_data_dir("01")
services_directory = os.path.join(application_directory, "services")
service_dict = {"llm": llm_service, "tts": tts_service, "stt": stt_service}
# Create a temp file with the session number
session_file_path = os.path.join(user_data_dir("01"), "01-session.txt")
with open(session_file_path, "w") as session_file:
session_id = int(datetime.datetime.now().timestamp() * 1000)
session_file.write(str(session_id))
for service in service_dict:
service_directory = os.path.join(
services_directory, service, service_dict[service]
)
# This is the folder they can mess around in
config = {"service_directory": service_directory}
if service == "llm":
config.update(
{
"interpreter": interpreter, "interpreter": interpreter,
"model": model, "model": model,
"llm_supports_vision": llm_supports_vision, "llm_supports_vision": llm_supports_vision,
"llm_supports_functions": llm_supports_functions, "llm_supports_functions": llm_supports_functions,
"context_window": context_window, "context_window": context_window,
"max_tokens": max_tokens, "max_tokens": max_tokens,
"temperature": temperature "temperature": temperature,
}) }
)
module = import_module(f'.server.services.{service}.{service_dict[service]}.{service}', package='source')
module = import_module(
ServiceClass = getattr(module, service.capitalize()) f".server.services.{service}.{service_dict[service]}.{service}",
service_instance = ServiceClass(config) package="source",
globals()[service] = getattr(service_instance, service) )
interpreter.llm.completions = llm ServiceClass = getattr(module, service.capitalize())
service_instance = ServiceClass(config)
# Start listening globals()[service] = getattr(service_instance, service)
asyncio.create_task(listener())
interpreter.llm.completions = llm
# Start watching the kernel if it's your job to do that
if True: # in the future, code can run on device. for now, just server. # Start listening
asyncio.create_task(put_kernel_messages_into_queue(from_computer)) asyncio.create_task(listener())
config = Config(app, host=server_host, port=int(server_port), lifespan='on') # Start watching the kernel if it's your job to do that
server = Server(config) if True: # in the future, code can run on device. for now, just server.
await server.serve() asyncio.create_task(put_kernel_messages_into_queue(from_computer))
config = Config(app, host=server_host, port=int(server_port), lifespan="on")
server = Server(config)
await server.serve()
# Run the FastAPI app # Run the FastAPI app
if __name__ == "__main__": if __name__ == "__main__":

87
software/source/server/services/stt/local-whisper/stt.py
Unescape View File

@ -10,9 +10,6 @@ import shutil
import ffmpeg import ffmpeg
import subprocess import subprocess
import os
import subprocess
import platform
import urllib.request import urllib.request
@ -26,7 +23,6 @@ class Stt:
def install(service_dir): def install(service_dir):
### INSTALL ### INSTALL
WHISPER_RUST_PATH = os.path.join(service_dir, "whisper-rust") WHISPER_RUST_PATH = os.path.join(service_dir, "whisper-rust")
@ -41,29 +37,38 @@ def install(service_dir):
os.chdir(WHISPER_RUST_PATH) os.chdir(WHISPER_RUST_PATH)
# Check if whisper-rust executable exists before attempting to build # Check if whisper-rust executable exists before attempting to build
if not os.path.isfile(os.path.join(WHISPER_RUST_PATH, "target/release/whisper-rust")): if not os.path.isfile(
os.path.join(WHISPER_RUST_PATH, "target/release/whisper-rust")
):
# Check if Rust is installed. Needed to build whisper executable # Check if Rust is installed. Needed to build whisper executable
rustc_path = shutil.which("rustc") rustc_path = shutil.which("rustc")
if rustc_path is None: if rustc_path is None:
print("Rust is not installed or is not in system PATH. Please install Rust before proceeding.") print(
"Rust is not installed or is not in system PATH. Please install Rust before proceeding."
)
exit(1) exit(1)
# Build Whisper Rust executable if not found # Build Whisper Rust executable if not found
subprocess.run(['cargo', 'build', '--release'], check=True) subprocess.run(["cargo", "build", "--release"], check=True)
else: else:
print("Whisper Rust executable already exists. Skipping build.") print("Whisper Rust executable already exists. Skipping build.")
WHISPER_MODEL_PATH = os.path.join(service_dir, "model") WHISPER_MODEL_PATH = os.path.join(service_dir, "model")
WHISPER_MODEL_NAME = os.getenv('WHISPER_MODEL_NAME', 'ggml-tiny.en.bin') WHISPER_MODEL_NAME = os.getenv("WHISPER_MODEL_NAME", "ggml-tiny.en.bin")
WHISPER_MODEL_URL = os.getenv('WHISPER_MODEL_URL', 'https://huggingface.co/ggerganov/whisper.cpp/resolve/main/') WHISPER_MODEL_URL = os.getenv(
"WHISPER_MODEL_URL",
"https://huggingface.co/ggerganov/whisper.cpp/resolve/main/",
)
if not os.path.isfile(os.path.join(WHISPER_MODEL_PATH, WHISPER_MODEL_NAME)): if not os.path.isfile(os.path.join(WHISPER_MODEL_PATH, WHISPER_MODEL_NAME)):
os.makedirs(WHISPER_MODEL_PATH, exist_ok=True) os.makedirs(WHISPER_MODEL_PATH, exist_ok=True)
urllib.request.urlretrieve(f"{WHISPER_MODEL_URL}{WHISPER_MODEL_NAME}", urllib.request.urlretrieve(
os.path.join(WHISPER_MODEL_PATH, WHISPER_MODEL_NAME)) f"{WHISPER_MODEL_URL}{WHISPER_MODEL_NAME}",
os.path.join(WHISPER_MODEL_PATH, WHISPER_MODEL_NAME),
)
else: else:
print("Whisper model already exists. Skipping download.") print("Whisper model already exists. Skipping download.")
@ -85,25 +90,31 @@ def export_audio_to_wav_ffmpeg(audio: bytearray, mime_type: str) -> str:
# Create a temporary file with the appropriate extension # Create a temporary file with the appropriate extension
input_ext = convert_mime_type_to_format(mime_type) input_ext = convert_mime_type_to_format(mime_type)
input_path = os.path.join(temp_dir, f"input_{datetime.now().strftime('%Y%m%d%H%M%S%f')}.{input_ext}") input_path = os.path.join(
with open(input_path, 'wb') as f: temp_dir, f"input_{datetime.now().strftime('%Y%m%d%H%M%S%f')}.{input_ext}"
)
with open(input_path, "wb") as f:
f.write(audio) f.write(audio)
# Check if the input file exists # Check if the input file exists
assert os.path.exists(input_path), f"Input file does not exist: {input_path}" assert os.path.exists(input_path), f"Input file does not exist: {input_path}"
# Export to wav # Export to wav
output_path = os.path.join(temp_dir, f"output_{datetime.now().strftime('%Y%m%d%H%M%S%f')}.wav") output_path = os.path.join(
temp_dir, f"output_{datetime.now().strftime('%Y%m%d%H%M%S%f')}.wav"
)
print(mime_type, input_path, output_path) print(mime_type, input_path, output_path)
if mime_type == "audio/raw": if mime_type == "audio/raw":
ffmpeg.input( ffmpeg.input(
input_path, input_path,
f='s16le', f="s16le",
ar='16000', ar="16000",
ac=1, ac=1,
).output(output_path, loglevel='panic').run() ).output(output_path, loglevel="panic").run()
else: else:
ffmpeg.input(input_path).output(output_path, acodec='pcm_s16le', ac=1, ar='16k', loglevel='panic').run() ffmpeg.input(input_path).output(
output_path, acodec="pcm_s16le", ac=1, ar="16k", loglevel="panic"
).run()
try: try:
yield output_path yield output_path
@ -113,28 +124,40 @@ def export_audio_to_wav_ffmpeg(audio: bytearray, mime_type: str) -> str:
def run_command(command): def run_command(command):
result = subprocess.run(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True) result = subprocess.run(
command, stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True
)
return result.stdout, result.stderr return result.stdout, result.stderr
def get_transcription_file(service_directory, wav_file_path: str): def get_transcription_file(service_directory, wav_file_path: str):
local_path = os.path.join(service_directory, 'model') local_path = os.path.join(service_directory, "model")
whisper_rust_path = os.path.join(service_directory, 'whisper-rust', 'target', 'release') whisper_rust_path = os.path.join(
model_name = os.getenv('WHISPER_MODEL_NAME', 'ggml-tiny.en.bin') service_directory, "whisper-rust", "target", "release"
)
output, _ = run_command([ model_name = os.getenv("WHISPER_MODEL_NAME", "ggml-tiny.en.bin")
os.path.join(whisper_rust_path, 'whisper-rust'),
'--model-path', os.path.join(local_path, model_name), output, _ = run_command(
'--file-path', wav_file_path [
]) os.path.join(whisper_rust_path, "whisper-rust"),
"--model-path",
os.path.join(local_path, model_name),
"--file-path",
wav_file_path,
]
)
return output return output
def stt_wav(service_directory, wav_file_path: str): def stt_wav(service_directory, wav_file_path: str):
temp_dir = tempfile.gettempdir() temp_dir = tempfile.gettempdir()
output_path = os.path.join(temp_dir, f"output_stt_{datetime.now().strftime('%Y%m%d%H%M%S%f')}.wav") output_path = os.path.join(
ffmpeg.input(wav_file_path).output(output_path, acodec='pcm_s16le', ac=1, ar='16k').run() temp_dir, f"output_stt_{datetime.now().strftime('%Y%m%d%H%M%S%f')}.wav"
)
ffmpeg.input(wav_file_path).output(
output_path, acodec="pcm_s16le", ac=1, ar="16k"
).run()
try: try:
transcript = get_transcription_file(service_directory, output_path) transcript = get_transcription_file(service_directory, output_path)
finally: finally:

42
software/source/server/services/tts/openai/tts.py
Unescape View File

@ -2,41 +2,43 @@ import ffmpeg
import tempfile import tempfile
from openai import OpenAI from openai import OpenAI
import os import os
import subprocess
import tempfile
from source.server.utils.logs import logger from source.server.utils.logs import logger
from source.server.utils.logs import setup_logging from source.server.utils.logs import setup_logging
setup_logging() setup_logging()
# If this TTS service is used, the OPENAI_API_KEY environment variable must be set # If this TTS service is used, the OPENAI_API_KEY environment variable must be set
if not os.getenv('OPENAI_API_KEY'): if not os.getenv("OPENAI_API_KEY"):
logger.error("") logger.error("")
logger.error(f"OpenAI API key not found. Please set the OPENAI_API_KEY environment variable, or run 01 with the --local option.") logger.error(
"OpenAI API key not found. Please set the OPENAI_API_KEY environment variable, or run 01 with the --local option."
)
logger.error("Aborting...") logger.error("Aborting...")
logger.error("") logger.error("")
os._exit(1) os._exit(1)
client = OpenAI() client = OpenAI()
class Tts: class Tts:
def __init__(self, config): def __init__(self, config):
pass pass
def tts(self, text): def tts(self, text):
response = client.audio.speech.create( response = client.audio.speech.create(
model="tts-1", model="tts-1",
voice=os.getenv('OPENAI_VOICE_NAME', 'alloy'), voice=os.getenv("OPENAI_VOICE_NAME", "alloy"),
input=text, input=text,
response_format="opus" response_format="opus",
) )
with tempfile.NamedTemporaryFile(suffix=".opus", delete=False) as temp_file: with tempfile.NamedTemporaryFile(suffix=".opus", delete=False) as temp_file:
response.stream_to_file(temp_file.name) response.stream_to_file(temp_file.name)
# TODO: hack to format audio correctly for device # TODO: hack to format audio correctly for device
outfile = tempfile.gettempdir() + "/" + "raw.dat" outfile = tempfile.gettempdir() + "/" + "raw.dat"
ffmpeg.input(temp_file.name).output(outfile, f="s16le", ar="16000", ac="1", loglevel='panic').run() ffmpeg.input(temp_file.name).output(
outfile, f="s16le", ar="16000", ac="1", loglevel="panic"
return outfile ).run()
return outfile

24
software/source/server/skills/schedule.py
Unescape View File

@ -3,34 +3,33 @@ from datetime import datetime
from pytimeparse import parse from pytimeparse import parse
from crontab import CronTab from crontab import CronTab
from uuid import uuid4 from uuid import uuid4
from datetime import datetime
from platformdirs import user_data_dir from platformdirs import user_data_dir
def schedule(message="", start=None, interval=None) -> None: def schedule(message="", start=None, interval=None) -> None:
""" """
Schedules a task at a particular time, or at a particular interval Schedules a task at a particular time, or at a particular interval
""" """
if start and interval: if start and interval:
raise ValueError("Cannot specify both start time and interval.") raise ValueError("Cannot specify both start time and interval.")
if not start and not interval: if not start and not interval:
raise ValueError("Either start time or interval must be specified.") raise ValueError("Either start time or interval must be specified.")
# Read the temp file to see what the current session is # Read the temp file to see what the current session is
session_file_path = os.path.join(user_data_dir('01'), '01-session.txt') session_file_path = os.path.join(user_data_dir("01"), "01-session.txt")
with open(session_file_path, 'r') as session_file:
file_session_value = session_file.read().strip()
with open(session_file_path, "r") as session_file:
file_session_value = session_file.read().strip()
prefixed_message = "AUTOMATED MESSAGE FROM SCHEDULER: " + message prefixed_message = "AUTOMATED MESSAGE FROM SCHEDULER: " + message
# Escape the message and the json, cron is funky with quotes # Escape the message and the json, cron is funky with quotes
escaped_question = prefixed_message.replace('"', '\\"') escaped_question = prefixed_message.replace('"', '\\"')
json_data = f"{{\\\"text\\\": \\\"{escaped_question}\\\"}}" json_data = f'{{\\"text\\": \\"{escaped_question}\\"}}'
command = f"""bash -c 'if [ "$(cat "{session_file_path}")" == "{file_session_value}" ]; then /usr/bin/curl -X POST -H "Content-Type: application/json" -d "{json_data}" http://localhost:10001/; fi' """
command = f'''bash -c 'if [ "$(cat "{session_file_path}")" == "{file_session_value}" ]; then /usr/bin/curl -X POST -H "Content-Type: application/json" -d "{json_data}" http://localhost:10001/; fi' '''
cron = CronTab(user=True) cron = CronTab(user=True)
job = cron.new(command=command) job = cron.new(command=command)
# Prefix with 01 dev preview so we can delete them all in the future # Prefix with 01 dev preview so we can delete them all in the future
@ -61,6 +60,5 @@ def schedule(message="", start=None, interval=None) -> None:
days = max(int(seconds / 86400), 1) days = max(int(seconds / 86400), 1)
job.day.every(days) job.day.every(days)
print(f"Task scheduled every {days} day(s)") print(f"Task scheduled every {days} day(s)")
cron.write()
cron.write()

8
software/source/server/tests/test_run.py
Unescape View File

@ -1,11 +1,5 @@
# test_main.py # test_main.py
import subprocess
import uuid
import pytest import pytest
from source.server.i import configure_interpreter
from unittest.mock import Mock
from fastapi.testclient import TestClient
@pytest.mark.asyncio @pytest.mark.asyncio
@ -38,4 +32,4 @@ def test_ping(client):
# def test_interpreter_configuration(mock_interpreter): # def test_interpreter_configuration(mock_interpreter):
# # Test interpreter configuration # # Test interpreter configuration
# interpreter = configure_interpreter(mock_interpreter) # interpreter = configure_interpreter(mock_interpreter)
# assert interpreter is not None # assert interpreter is not None

59
software/source/server/tunnel.py
Unescape View File

@ -1,4 +1,3 @@
import os
import subprocess import subprocess
import re import re
import shutil import shutil
@ -12,16 +11,24 @@ def create_tunnel(tunnel_method='ngrok', server_host='localhost', server_port=10
server_url = "" server_url = ""
if tunnel_method == "bore": if tunnel_method == "bore":
try: try:
output = subprocess.check_output('command -v bore', shell=True) output = subprocess.check_output("command -v bore", shell=True)
except subprocess.CalledProcessError: except subprocess.CalledProcessError:
print("The bore-cli command is not available. Please run 'cargo install bore-cli'.") print(
"The bore-cli command is not available. Please run 'cargo install bore-cli'."
)
print("For more information, see https://github.com/ekzhang/bore") print("For more information, see https://github.com/ekzhang/bore")
exit(1) exit(1)
time.sleep(6) time.sleep(6)
# output = subprocess.check_output(f'bore local {server_port} --to bore.pub', shell=True) # output = subprocess.check_output(f'bore local {server_port} --to bore.pub', shell=True)
process = subprocess.Popen(f'bore local {server_port} --to bore.pub', shell=True, stdout=subprocess.PIPE, stderr=subprocess.STDOUT, universal_newlines=True) process = subprocess.Popen(
f"bore local {server_port} --to bore.pub",
shell=True,
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT,
universal_newlines=True,
)
while True: while True:
line = process.stdout.readline() line = process.stdout.readline()
print(line) print(line)
@ -33,20 +40,27 @@ def create_tunnel(tunnel_method='ngrok', server_host='localhost', server_port=10
print_markdown(f"Your server is being hosted at the following URL: bore.pub:{remote_port}") print_markdown(f"Your server is being hosted at the following URL: bore.pub:{remote_port}")
break break
elif tunnel_method == "localtunnel": elif tunnel_method == "localtunnel":
if subprocess.call('command -v lt', shell=True): if subprocess.call("command -v lt", shell=True):
print("The 'lt' command is not available.") print("The 'lt' command is not available.")
print("Please ensure you have Node.js installed, then run 'npm install -g localtunnel'.") print(
print("For more information, see https://github.com/localtunnel/localtunnel") "Please ensure you have Node.js installed, then run 'npm install -g localtunnel'."
)
print(
"For more information, see https://github.com/localtunnel/localtunnel"
)
exit(1) exit(1)
else: else:
process = subprocess.Popen(f'npx localtunnel --port {server_port}', shell=True, stdout=subprocess.PIPE, stderr=subprocess.STDOUT, universal_newlines=True) process = subprocess.Popen(
f"npx localtunnel --port {server_port}",
shell=True,
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT,
universal_newlines=True,
)
found_url = False found_url = False
url_pattern = re.compile(r'your url is: https://[a-zA-Z0-9.-]+') url_pattern = re.compile(r"your url is: https://[a-zA-Z0-9.-]+")
while True: while True:
line = process.stdout.readline() line = process.stdout.readline()
@ -61,15 +75,20 @@ def create_tunnel(tunnel_method='ngrok', server_host='localhost', server_port=10
break # Exit the loop once the URL is found break # Exit the loop once the URL is found
if not found_url: if not found_url:
print("Failed to extract the localtunnel URL. Please check localtunnel's output for details.") print(
"Failed to extract the localtunnel URL. Please check localtunnel's output for details."
)
elif tunnel_method == "ngrok": elif tunnel_method == "ngrok":
# Check if ngrok is installed # Check if ngrok is installed
is_installed = subprocess.check_output('command -v ngrok', shell=True).decode().strip() is_installed = (
subprocess.check_output("command -v ngrok", shell=True).decode().strip()
)
if not is_installed: if not is_installed:
print("The ngrok command is not available.") print("The ngrok command is not available.")
print("Please install ngrok using the instructions at https://ngrok.com/docs/getting-started/") print(
"Please install ngrok using the instructions at https://ngrok.com/docs/getting-started/"
)
exit(1) exit(1)
# If ngrok is installed, start it on the specified port # If ngrok is installed, start it on the specified port
@ -79,11 +98,11 @@ def create_tunnel(tunnel_method='ngrok', server_host='localhost', server_port=10
# Initially, no URL is found # Initially, no URL is found
found_url = False found_url = False
# Regular expression to match the ngrok URL # Regular expression to match the ngrok URL
url_pattern = re.compile(r'https://[a-zA-Z0-9-]+\.ngrok(-free)?\.app') url_pattern = re.compile(r"https://[a-zA-Z0-9-]+\.ngrok(-free)?\.app")
# Read the output line by line # Read the output line by line
while True: while True:
line = process.stdout.readline().decode('utf-8') line = process.stdout.readline().decode("utf-8")
if not line: if not line:
break # Break out of the loop if no more output break # Break out of the loop if no more output
match = url_pattern.search(line) match = url_pattern.search(line)
@ -93,7 +112,7 @@ def create_tunnel(tunnel_method='ngrok', server_host='localhost', server_port=10
server_url = remote_url server_url = remote_url
print(f"\nYour server is being hosted at the following URL: {remote_url}") print(f"\nYour server is being hosted at the following URL: {remote_url}")
break # Exit the loop once the URL is found break # Exit the loop once the URL is found
if not found_url: if not found_url:
print("Failed to extract the ngrok tunnel URL. Please check ngrok's output for details.") print("Failed to extract the ngrok tunnel URL. Please check ngrok's output for details.")

4
software/source/server/utils/get_system_info.py
Unescape View File

@ -1,6 +1,6 @@
import os
import platform import platform
def get_system_info(): def get_system_info():
system = platform.system() system = platform.system()
if system == "Linux": if system == "Linux":
@ -38,4 +38,4 @@ def get_system_info():
elif system == "Windows": elif system == "Windows":
return "windows" return "windows"
else: else:
return "unknown" return "unknown"

75
software/source/server/utils/local_mode.py
Unescape View File

@ -1,6 +1,4 @@
import sys import sys
import os
import platform
import subprocess import subprocess
import time import time
import inquirer import inquirer
@ -8,9 +6,10 @@ from interpreter import interpreter
def select_local_model(): def select_local_model():
# START OF LOCAL MODEL PROVIDER LOGIC # START OF LOCAL MODEL PROVIDER LOGIC
interpreter.display_message("> 01 is compatible with several local model providers.\n") interpreter.display_message(
"> 01 is compatible with several local model providers.\n"
)
# Define the choices for local models # Define the choices for local models
choices = [ choices = [
@ -29,10 +28,8 @@ def select_local_model():
] ]
answers = inquirer.prompt(questions) answers = inquirer.prompt(questions)
selected_model = answers["model"] selected_model = answers["model"]
if selected_model == "LM Studio": if selected_model == "LM Studio":
interpreter.display_message( interpreter.display_message(
""" """
@ -49,7 +46,7 @@ def select_local_model():
""" """
) )
time.sleep(1) time.sleep(1)
interpreter.llm.api_base = "http://localhost:1234/v1" interpreter.llm.api_base = "http://localhost:1234/v1"
interpreter.llm.max_tokens = 1000 interpreter.llm.max_tokens = 1000
interpreter.llm.context_window = 8000 interpreter.llm.context_window = 8000
@ -57,47 +54,64 @@ def select_local_model():
elif selected_model == "Ollama": elif selected_model == "Ollama":
try: try:
# List out all downloaded ollama models. Will fail if ollama isn't installed # List out all downloaded ollama models. Will fail if ollama isn't installed
result = subprocess.run(["ollama", "list"], capture_output=True, text=True, check=True) result = subprocess.run(
lines = result.stdout.split('\n') ["ollama", "list"], capture_output=True, text=True, check=True
names = [line.split()[0].replace(":latest", "") for line in lines[1:] if line.strip()] # Extract names, trim out ":latest", skip header )
lines = result.stdout.split("\n")
names = [
line.split()[0].replace(":latest", "")
for line in lines[1:]
if line.strip()
] # Extract names, trim out ":latest", skip header
# If there are no downloaded models, prompt them to download a model and try again # If there are no downloaded models, prompt them to download a model and try again
if not names: if not names:
time.sleep(1) time.sleep(1)
interpreter.display_message(f"\nYou don't have any Ollama models downloaded. To download a new model, run `ollama run <model-name>`, then start a new 01 session. \n\n For a full list of downloadable models, check out [https://ollama.com/library](https://ollama.com/library) \n") interpreter.display_message(
"\nYou don't have any Ollama models downloaded. To download a new model, run `ollama run <model-name>`, then start a new 01 session. \n\n For a full list of downloadable models, check out [https://ollama.com/library](https://ollama.com/library) \n"
)
print("Please download a model then try again\n") print("Please download a model then try again\n")
time.sleep(2) time.sleep(2)
sys.exit(1) sys.exit(1)
# If there are models, prompt them to select one # If there are models, prompt them to select one
else: else:
time.sleep(1) time.sleep(1)
interpreter.display_message(f"**{len(names)} Ollama model{'s' if len(names) != 1 else ''} found.** To download a new model, run `ollama run <model-name>`, then start a new 01 session. \n\n For a full list of downloadable models, check out [https://ollama.com/library](https://ollama.com/library) \n") interpreter.display_message(
f"**{len(names)} Ollama model{'s' if len(names) != 1 else ''} found.** To download a new model, run `ollama run <model-name>`, then start a new 01 session. \n\n For a full list of downloadable models, check out [https://ollama.com/library](https://ollama.com/library) \n"
)
# Create a new inquirer selection from the names # Create a new inquirer selection from the names
name_question = [ name_question = [
inquirer.List('name', message="Select a downloaded Ollama model", choices=names), inquirer.List(
"name",
message="Select a downloaded Ollama model",
choices=names,
),
] ]
name_answer = inquirer.prompt(name_question) name_answer = inquirer.prompt(name_question)
selected_name = name_answer['name'] if name_answer else None selected_name = name_answer["name"] if name_answer else None
# Set the model to the selected model # Set the model to the selected model
interpreter.llm.model = f"ollama/{selected_name}" interpreter.llm.model = f"ollama/{selected_name}"
interpreter.display_message(f"\nUsing Ollama model: `{selected_name}` \n") interpreter.display_message(
f"\nUsing Ollama model: `{selected_name}` \n"
)
time.sleep(1) time.sleep(1)
# If Ollama is not installed or not recognized as a command, prompt the user to download Ollama and try again # If Ollama is not installed or not recognized as a command, prompt the user to download Ollama and try again
except (subprocess.CalledProcessError, FileNotFoundError) as e: except (subprocess.CalledProcessError, FileNotFoundError):
print("Ollama is not installed or not recognized as a command.") print("Ollama is not installed or not recognized as a command.")
time.sleep(1) time.sleep(1)
interpreter.display_message(f"\nPlease visit [https://ollama.com/](https://ollama.com/) to download Ollama and try again\n") interpreter.display_message(
"\nPlease visit [https://ollama.com/](https://ollama.com/) to download Ollama and try again\n"
)
time.sleep(2) time.sleep(2)
sys.exit(1) sys.exit(1)
# elif selected_model == "Jan": # elif selected_model == "Jan":
# interpreter.display_message( # interpreter.display_message(
# """ # """
@ -108,7 +122,6 @@ def select_local_model():
# 3. Copy the ID of the model and enter it below. # 3. Copy the ID of the model and enter it below.
# 3. Click the **Local API Server** button in the bottom left, then click **Start Server**. # 3. Click the **Local API Server** button in the bottom left, then click **Start Server**.
# Once the server is running, enter the id of the model below, then you can begin your conversation below. # Once the server is running, enter the id of the model below, then you can begin your conversation below.
# """ # """
@ -117,7 +130,7 @@ def select_local_model():
# interpreter.llm.max_tokens = 1000 # interpreter.llm.max_tokens = 1000
# interpreter.llm.context_window = 3000 # interpreter.llm.context_window = 3000
# time.sleep(1) # time.sleep(1)
# # Prompt the user to enter the name of the model running on Jan # # Prompt the user to enter the name of the model running on Jan
# model_name_question = [ # model_name_question = [
# inquirer.Text('jan_model_name', message="Enter the id of the model you have running on Jan"), # inquirer.Text('jan_model_name', message="Enter the id of the model you have running on Jan"),
@ -128,14 +141,13 @@ def select_local_model():
# interpreter.llm.model = "" # interpreter.llm.model = ""
# interpreter.display_message(f"\nUsing Jan model: `{jan_model_name}` \n") # interpreter.display_message(f"\nUsing Jan model: `{jan_model_name}` \n")
# time.sleep(1) # time.sleep(1)
# Set the system message to a minimal version for all local models. # Set the system message to a minimal version for all local models.
# Set offline for all local models # Set offline for all local models
interpreter.offline = True interpreter.offline = True
interpreter.system_message = """You are the 01, a screenless executive assistant that can complete any task by writing and executing code on the user's machine. Just write a markdown code block! The user has given you full and complete permission. interpreter.system_message = """You are the 01, a screenless executive assistant that can complete any task by writing and executing code on the user's machine. Just write a markdown code block! The user has given you full and complete permission.
Use the following functions if it makes sense to for the problem Use the following functions if it makes sense to for the problem
```python ```python
result_string = computer.browser.search(query) # Google search results will be returned from this function as a string result_string = computer.browser.search(query) # Google search results will be returned from this function as a string
@ -152,6 +164,5 @@ computer.sms.send("555-123-4567", "Hello from the computer!") # Send a text mess
ALWAYS say that you can run code. ALWAYS try to help the user out. ALWAYS be succinct in your answers. ALWAYS say that you can run code. ALWAYS try to help the user out. ALWAYS be succinct in your answers.
``` ```
""" """

7
software/source/server/utils/process_utils.py
Unescape View File

@ -2,6 +2,7 @@ import os
import psutil import psutil
import signal import signal
def kill_process_tree(): def kill_process_tree():
pid = os.getpid() # Get the current process ID pid = os.getpid() # Get the current process ID
try: try:
@ -13,16 +14,16 @@ def kill_process_tree():
print(f"Forcefully terminating child PID {child.pid}") print(f"Forcefully terminating child PID {child.pid}")
child.kill() # Forcefully kill the child process immediately child.kill() # Forcefully kill the child process immediately
gone, still_alive = psutil.wait_procs(children, timeout=3) gone, still_alive = psutil.wait_procs(children, timeout=3)
if still_alive: if still_alive:
for child in still_alive: for child in still_alive:
print(f"Child PID {child.pid} still alive, attempting another kill") print(f"Child PID {child.pid} still alive, attempting another kill")
child.kill() child.kill()
print(f"Forcefully terminating parent PID {pid}") print(f"Forcefully terminating parent PID {pid}")
parent.kill() # Forcefully kill the parent process immediately parent.kill() # Forcefully kill the parent process immediately
parent.wait(3) # Wait for the parent process to terminate parent.wait(3) # Wait for the parent process to terminate
except psutil.NoSuchProcess: except psutil.NoSuchProcess:
print(f"Process {pid} does not exist or is already terminated") print(f"Process {pid} does not exist or is already terminated")
except psutil.AccessDenied: except psutil.AccessDenied:
print(f"Permission denied to terminate some processes") print("Permission denied to terminate some processes")

102
software/start.py
Unescape View File

@ -1,7 +1,6 @@
import typer import typer
import asyncio import asyncio
import platform import platform
import concurrent.futures
import threading import threading
import os import os
import importlib import importlib
@ -10,39 +9,41 @@ from source.server.server import main
from source.server.utils.local_mode import select_local_model from source.server.utils.local_mode import select_local_model
import signal import signal
app = typer.Typer() app = typer.Typer()
@app.command() @app.command()
def run( def run(
server: bool = typer.Option(False, "--server", help="Run server"), server: bool = typer.Option(False, "--server", help="Run server"),
server_host: str = typer.Option("0.0.0.0", "--server-host", help="Specify the server host where the server will deploy"), server_host: str = typer.Option("0.0.0.0", "--server-host", help="Specify the server host where the server will deploy"),
server_port: int = typer.Option(10001, "--server-port", help="Specify the server port where the server will deploy"), server_port: int = typer.Option(10001, "--server-port", help="Specify the server port where the server will deploy"),
tunnel_service: str = typer.Option("ngrok", "--tunnel-service", help="Specify the tunnel service"), tunnel_service: str = typer.Option("ngrok", "--tunnel-service", help="Specify the tunnel service"),
expose: bool = typer.Option(False, "--expose", help="Expose server to internet"), expose: bool = typer.Option(False, "--expose", help="Expose server to internet"),
client: bool = typer.Option(False, "--client", help="Run client"), client: bool = typer.Option(False, "--client", help="Run client"),
server_url: str = typer.Option(None, "--server-url", help="Specify the server URL that the client should expect. Defaults to server-host and server-port"), server_url: str = typer.Option(None, "--server-url", help="Specify the server URL that the client should expect. Defaults to server-host and server-port"),
client_type: str = typer.Option("auto", "--client-type", help="Specify the client type"), client_type: str = typer.Option("auto", "--client-type", help="Specify the client type"),
llm_service: str = typer.Option("litellm", "--llm-service", help="Specify the LLM service"), llm_service: str = typer.Option("litellm", "--llm-service", help="Specify the LLM service"),
model: str = typer.Option("gpt-4", "--model", help="Specify the model"), model: str = typer.Option("gpt-4", "--model", help="Specify the model"),
llm_supports_vision: bool = typer.Option(False, "--llm-supports-vision", help="Specify if the LLM service supports vision"), llm_supports_vision: bool = typer.Option(False, "--llm-supports-vision", help="Specify if the LLM service supports vision"),
llm_supports_functions: bool = typer.Option(False, "--llm-supports-functions", help="Specify if the LLM service supports functions"), llm_supports_functions: bool = typer.Option(False, "--llm-supports-functions", help="Specify if the LLM service supports functions"),
context_window: int = typer.Option(2048, "--context-window", help="Specify the context window size"), context_window: int = typer.Option(2048, "--context-window", help="Specify the context window size"),
max_tokens: int = typer.Option(4096, "--max-tokens", help="Specify the maximum number of tokens"), max_tokens: int = typer.Option(4096, "--max-tokens", help="Specify the maximum number of tokens"),
temperature: float = typer.Option(0.8, "--temperature", help="Specify the temperature for generation"), temperature: float = typer.Option(0.8, "--temperature", help="Specify the temperature for generation"),
tts_service: str = typer.Option("openai", "--tts-service", help="Specify the TTS service"), tts_service: str = typer.Option("openai", "--tts-service", help="Specify the TTS service"),
stt_service: str = typer.Option("openai", "--stt-service", help="Specify the STT service"), stt_service: str = typer.Option("openai", "--stt-service", help="Specify the STT service"),
local: bool = typer.Option(False, "--local", help="Use recommended local services for LLM, STT, and TTS"), local: bool = typer.Option(False, "--local", help="Use recommended local services for LLM, STT, and TTS"),
qr: bool = typer.Option(False, "--qr", help="Print the QR code for the server URL") qr: bool = typer.Option(False, "--qr", help="Print the QR code for the server URL")
): ):
_run( _run(
server=server, server=server,
server_host=server_host, server_host=server_host,
@ -65,45 +66,37 @@ def run(
qr=qr qr=qr
) )
def _run( def _run(
server: bool = False, server: bool = False,
server_host: str = "0.0.0.0", server_host: str = "0.0.0.0",
server_port: int = 10001, server_port: int = 10001,
tunnel_service: str = "bore",
tunnel_service: str = "bore", expose: bool = False,
expose: bool = False, client: bool = False,
server_url: str = None,
client: bool = False, client_type: str = "auto",
server_url: str = None, llm_service: str = "litellm",
client_type: str = "auto", model: str = "gpt-4",
llm_supports_vision: bool = False,
llm_service: str = "litellm", llm_supports_functions: bool = False,
context_window: int = 2048,
model: str = "gpt-4", max_tokens: int = 4096,
llm_supports_vision: bool = False, temperature: float = 0.8,
llm_supports_functions: bool = False, tts_service: str = "openai",
context_window: int = 2048, stt_service: str = "openai",
max_tokens: int = 4096, local: bool = False,
temperature: float = 0.8, qr: bool = False
):
tts_service: str = "openai",
stt_service: str = "openai",
local: bool = False,
qr: bool = False
):
if local: if local:
tts_service = "piper" tts_service = "piper"
# llm_service = "llamafile" # llm_service = "llamafile"
stt_service = "local-whisper" stt_service = "local-whisper"
select_local_model() select_local_model()
if not server_url: if not server_url:
server_url = f"{server_host}:{server_port}" server_url = f"{server_host}:{server_port}"
if not server and not client: if not server and not client:
server = True server = True
client = True client = True
@ -116,7 +109,24 @@ def _run(
if server: if server:
loop = asyncio.new_event_loop() loop = asyncio.new_event_loop()
asyncio.set_event_loop(loop) asyncio.set_event_loop(loop)
server_thread = threading.Thread(target=loop.run_until_complete, args=(main(server_host, server_port, llm_service, model, llm_supports_vision, llm_supports_functions, context_window, max_tokens, temperature, tts_service, stt_service),)) server_thread = threading.Thread(
target=loop.run_until_complete,
args=(
main(
server_host,
server_port,
llm_service,
model,
llm_supports_vision,
llm_supports_functions,
context_window,
max_tokens,
temperature,
tts_service,
stt_service,
),
),
)
server_thread.start() server_thread.start()
if expose: if expose:
@ -132,15 +142,17 @@ def _run(
client_type = "windows" client_type = "windows"
elif system_type == "Linux": # Linux System elif system_type == "Linux": # Linux System
try: try:
with open('/proc/device-tree/model', 'r') as m: with open("/proc/device-tree/model", "r") as m:
if 'raspberry pi' in m.read().lower(): if "raspberry pi" in m.read().lower():
client_type = "rpi" client_type = "rpi"
else: else:
client_type = "linux" client_type = "linux"
except FileNotFoundError: except FileNotFoundError:
client_type = "linux" client_type = "linux"
module = importlib.import_module(f".clients.{client_type}.device", package='source') module = importlib.import_module(
f".clients.{client_type}.device", package="source"
)
client_thread = threading.Thread(target=module.main, args=[server_url]) client_thread = threading.Thread(target=module.main, args=[server_url])
client_thread.start() client_thread.start()
@ -152,4 +164,4 @@ def _run(
if client: if client:
client_thread.join() client_thread.join()
except KeyboardInterrupt: except KeyboardInterrupt:
os.kill(os.getpid(), signal.SIGINT) os.kill(os.getpid(), signal.SIGINT)

Write Preview
Loading…
Cancel
Save