Merge remote-tracking branch 'upstream/main' into u/shivenmian/local

.gitignore

@@ -159,4 +159,5 @@ cython_debug/
 #  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
 #  and can be added to the global gitignore or merged into this file.  For a more nuclear
 #  option (not recommended) you can uncomment the following to ignore the entire idea folder.
-#.idea/
+#.idea/
+OS/01/conversations/user.json

OS/01/conversations/user.json

@@ -1,12 +0,0 @@
-[
-    {
-        "role": "user",
-        "type": "message",
-        "content": " Hello, how are you doing?\n"
-    },
-    {
-        "role": "assistant",
-        "type": "message",
-        "content": "I'm an artificial intelligence, so I don't have feelings, but thank you for asking. How may I assist you today?"
-    }
-]

OS/01/device.py

@@ -22,6 +22,10 @@ from interpreter import interpreter # Just for code execution. Maybe we should l
 from utils.kernel import put_kernel_messages_into_queue
 from stt import stt_wav
 
+from utils.logs import setup_logging
+from utils.logs import logger
+setup_logging()
+
 # Configuration for Audio Recording
 CHUNK = 1024  # Record in chunks of 1024 samples
 FORMAT = pyaudio.paInt16  # 16 bits per sample
@@ -51,7 +55,7 @@ def record_audio():
 
     """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)
-    print("Recording started...")
+    logger.info("Recording started...")
     global RECORDING
 
     # Create a temporary WAV file to store the audio data
@@ -69,7 +73,7 @@ def record_audio():
     wav_file.close()
     stream.stop_stream()
     stream.close()
-    print("Recording stopped.")
+    logger.info("Recording stopped.")
 
     duration = wav_file.getnframes() / RATE
     if duration < 0.3:
@@ -120,7 +124,7 @@ def on_release(key):
     if key == keyboard.Key.space:
         toggle_recording(False)
     elif key == keyboard.Key.esc:
-        print("Exiting...")
+        logger.info("Exiting...")
         os._exit(0)
 
 import asyncio
@@ -137,7 +141,7 @@ async def websocket_communication(WS_URL):
     while True:
         try:
             async with websockets.connect(WS_URL) as websocket:
-                print("Press the spacebar to start/stop recording. Press ESC to exit.")
+                logger.info("Press the spacebar to start/stop recording. Press ESC to exit.")
                 asyncio.create_task(message_sender(websocket))
 
                 initial_message = {"role": None, "type": None, "format": None, "content": None} 
@@ -146,16 +150,18 @@ async def websocket_communication(WS_URL):
                 while True:
                     message = await websocket.recv()
 
-                    print("Got this message from the server:", type(message), message)
+                    logger.debug(f"Got this message from the server: {type(message)} {message}")
 
                     if type(message) == str:
                         message = json.loads(message)
 
                     if message.get("end"):
-                        print(f"Complete message from the server: {message_so_far}")
+                        logger.debug(f"Complete message from the server: {message_so_far}")
+                        logger.info("\n")
                         message_so_far = initial_message
 
                     if "content" in message:
+                        print(message['content'], end="", flush=True)
                         if any(message_so_far[key] != message[key] for key in message_so_far if key != "content"):
                             message_so_far = message
                         else:
@@ -183,7 +189,7 @@ async def websocket_communication(WS_URL):
   
         except:
             # traceback.print_exc()
-            print(f"Connecting to `{WS_URL}`...")
+            logger.info(f"Connecting to `{WS_URL}`...")
             await asyncio.sleep(2)
             
 

OS/01/server.py

@@ -22,6 +22,11 @@ from utils.kernel import put_kernel_messages_into_queue
 from i import configure_interpreter
 from interpreter import interpreter
 
+from utils.logs import setup_logging
+from utils.logs import logger
+setup_logging()
+
+
 app = FastAPI()
 
 conversation_history_path = Path(__file__).parent / 'conversations' / 'user.json'
@@ -64,10 +69,10 @@ if os.getenv('CODE_RUNNER') == "device":
                 to_device.put({"role": "assistant", "type": "code", "format": "python", "end": True})
             
             # Stream the response
-            print("Waiting for the device to respond...")
+            logger.info("Waiting for the device to respond...")
             while True:
                 chunk = from_computer.get()
-                print("Server recieved from device:", chunk)
+                logger.info(f"Server received from device: {chunk}")
                 if "end" in chunk:
                     break
                 yield chunk
@@ -94,7 +99,7 @@ async def websocket_endpoint(websocket: WebSocket):
         await asyncio.gather(receive_task, send_task)
     except Exception as e:
         traceback.print_exc()
-        print(f"Connection lost. Error: {e}")
+        logger.info(f"Connection lost. Error: {e}")
 
 async def receive_messages(websocket: WebSocket):
     while True:
@@ -109,7 +114,7 @@ async def receive_messages(websocket: WebSocket):
 async def send_messages(websocket: WebSocket):
     while True:
         message = await to_device.get()
-        print("Sending to the device:", type(message), message)
+        logger.debug(f"Sending to the device: {type(message)} {message}")
         await websocket.send_json(message)
 
 async def listener():
@@ -159,7 +164,7 @@ async def listener():
         
         for chunk in interpreter.chat(messages, stream=True, display=False):
 
-            print("Got chunk:", chunk)
+            logger.debug("Got chunk:", chunk)
 
             # Send it to the user
             await to_device.put(chunk)
@@ -195,7 +200,7 @@ async def listener():
                 with open(conversation_history_path, 'w') as file:
                     json.dump(interpreter.messages, file, indent=4)
 
-                print("New user message recieved. Breaking.")
+                logger.info("New user message recieved. Breaking.")
                 break
 
             # Also check if there's any new computer messages
@@ -204,7 +209,7 @@ async def listener():
                 with open(conversation_history_path, 'w') as file:
                     json.dump(interpreter.messages, file, indent=4)
 
-                print("New computer message recieved. Breaking.")
+                logger.info("New computer message recieved. Breaking.")
                 break
         else:
             with open(conversation_history_path, 'w') as file:
@@ -239,7 +244,7 @@ if __name__ == "__main__":
         if not server_url:
             raise ValueError("The environment variable SERVER_URL is not set. Please set it to proceed.")
         parsed_url = urllib.parse.urlparse(server_url)
-        print("Starting `server.py`...")
+        logger.info("Starting `server.py`...")
 
         config = Config(app, host=parsed_url.hostname, port=parsed_url.port, lifespan='on')
         server = Server(config)

OS/01/start.sh

@@ -25,6 +25,10 @@ export STT_RUNNER=device # If server, audio will be sent over websocket.
 # Will expose the server publically and display that URL.
 export SERVER_EXPOSE_PUBLICALLY=False
 
+# Debug level
+# export LOG_LEVEL=DEBUG
+export LOG_LEVEL="INFO"
+
 ### SETUP
 
 # if using local models, install the models / executables

OS/01/stt.py

@@ -11,6 +11,10 @@ import subprocess
 import openai
 from openai import OpenAI
 
+from utils.logs import setup_logging
+from utils.logs import logger
+setup_logging()
+
 client = OpenAI()
 
 def convert_mime_type_to_format(mime_type: str) -> str:
@@ -81,10 +85,10 @@ def stt_wav(wav_file_path: str):
                 response_format="text"
             )
         except openai.BadRequestError as e:
-            print("openai.BadRequestError:", e)
+            logger.info(f"openai.BadRequestError: {e}")
             return None
 
-        print("Transcription result:", transcript)
+        logger.info(f"Transcription result: {transcript}")
         return transcript
     else:
         temp_dir = tempfile.gettempdir()

OS/01/utils/kernel.py

@@ -2,6 +2,10 @@ import asyncio
 import subprocess
 import platform
 
+from utils.logs import setup_logging
+from utils.logs import logger
+setup_logging()
+
 def get_kernel_messages():
     """
     Is this the way to do this?
@@ -16,7 +20,7 @@ def get_kernel_messages():
         with open('/var/log/dmesg', 'r') as file:
             return file.read()
     else:
-        print("Unsupported platform.")
+        logger.info("Unsupported platform.")
 
 def custom_filter(message):
     # Check for {TO_INTERPRETER{ message here }TO_INTERPRETER} pattern
@@ -28,7 +32,7 @@ def custom_filter(message):
     elif 'USB' in message:
         return message
     # Check for network related keywords
-    elif any(keyword in message for keyword in ['network', 'IP', 'internet', 'LAN', 'WAN', 'router', 'switch']):
+    elif any(keyword in message for keyword in ['network', 'IP', 'internet', 'LAN', 'WAN', 'router', 'switch']) and "networkStatusForFlags" not in message:
         return message
     else:
         return None

OS/01/utils/logs.py

@@ -0,0 +1,22 @@
+import os
+import logging
+
+logger: logging.Logger = logging.getLogger("01")
+root_logger: logging.Logger = logging.getLogger()
+
+
+def _basic_config() -> None:
+    logging.basicConfig(
+        format="%(message)s"
+    )
+
+
+def setup_logging() -> None:
+    env = os.environ.get("LOG_LEVEL", "").upper()
+    if env == "DEBUG":
+        _basic_config()
+        logger.setLevel(logging.DEBUG)
+        root_logger.setLevel(logging.DEBUG)
+    elif env == "INFO":
+        _basic_config()
+        logger.setLevel(logging.INFO)

README.md

@@ -21,6 +21,7 @@ sudo apt-get install portaudio19-dev libav-tools
 ```bash
 python -m pip install -r requirements.txt
 ```
+NB: Depending on your local Python version, you may run into [this issue↗](https://github.com/TaylorSMarks/playsound/issues/150) installing playsound. Workarounds are provided in the issue.
 
 If you want to run local speech-to-text from whisper, download the GGML Whisper model from [Huggingface](https://huggingface.co/ggerganov/whisper.cpp). Then in `OS/01/start.sh`, set `ALL_LOCAL=TRUE` and set `WHISPER_MODEL_PATH` to the path of the model.
 

hardware/devices/jetson-nano/README.md

@@ -0,0 +1,22 @@
+# Development Setup for Jetson Nano
+
+1. Go through the tutorial here: https://developer.nvidia.com/embedded/learn/get-started-jetson-nano-devkit#intro
+
+2. At the end of that guide, you should have a Jetson running off a power supply or micro USB.
+
+3. Get network connectivity. The Jetson does not have a WiFi module so you will need to plug in ethernet.
+    If you have a laptop, you can share internet access over Ethernet.
+
+    To do this with Mac, do the following:
+
+    a. Plug a cable from the Jetson Ethernet port to your Mac (you can use a Ethernet -> USB converter for your Mac).
+
+    b. Go to General->Sharing, then click the little `(i)` icon next to "Internet Sharing", and check all the options.
+
+    ![](mac-share-internet.png)
+
+    c. Go back to General->Sharing, and turn on "Internet Sharing".
+
+    ![](mac-share-internet-v2.png)
+
+    d. Now the Jetson should have connectivity!