Merge pull request #57 from SpudGunMan/llmLocationAware

LLM location aware enhancement

Dockerfile

@@ -0,0 +1,18 @@
+FROM python:3.10-slim
+ENV PYTHONUNBUFFERED=1
+
+RUN apt-get update && apt-get install -y gettext && rm -rf /var/lib/apt/lists/*
+
+
+WORKDIR /app
+COPY . /app
+COPY requirements.txt .
+
+RUN pip install -r requirements.txt
+COPY . .
+
+COPY config.ini /app/config.ini
+COPY entrypoint.sh /app/entrypoint.sh
+
+RUN chmod +x /app/entrypoint.sh
+ENTRYPOINT ["/app/entrypoint.sh"]

README.md

@@ -10,7 +10,7 @@ Along with network testing, this bot has a lot of other fun features, like simpl
 
 The bot is also capable of using dual radio/nodes, so you can monitor two networks at the same time and send messages to nodes using the same `bbspost @nodeNumber #message` or `bbspost @nodeShportName #message` function. There is a small message board to fit in the constraints of Meshtastic for posting bulletin messages with `bbspost $subject #message`.
 
-Look up data using wiki results or interact with [Ollama](https://ollama.com) LLM AI see the [OllamaDocs](https://github.com/ollama/ollama/tree/main/docs) If Ollama is enabled you can DM the bot directly.
+Look up data using wiki results or interact with [Ollama](https://ollama.com) LLM AI see the [OllamaDocs](https://github.com/ollama/ollama/tree/main/docs) If Ollama is enabled you can DM the bot directly. The default model for mesh-bot which is currently `gemma2:2b`
 
 The bot will report on anyone who is getting close to the configured lat/long, if in a remote location.
 
@@ -40,8 +40,8 @@ Any messages that are over 160 characters are chunked into 160 message bytes to
   - `wx` and `wxc` returns local weather forecast, (wxc is metric value), NOAA or Open Meteo for weather forecasting.
   - `wxa` and `wxalert` return NOAA alerts. Short title or expanded details
   - `joke` tells a joke
-  - `wiki: ` search wikipedia, return the first few sentances of first result if a match `wiki: lora radio`
-  - `ask: ` ask Ollama LLM AI for a response `ask: what temp do I cook chicken`
+  - `wiki: ` will search wikipedia, return the first few sentances of first result if a match `wiki: lora radio`
+  - `askai` and `ask:` will ask Ollama LLM AI for a response `askai what temp do I cook chicken`
   - `messages` Replay the last messages heard, like Store and Forward
   - `motd` or to set the message `motd $New Message Of the day`
   - `lheard` returns the last 5 heard nodes with SNR, can also use `sitrep`
@@ -63,6 +63,12 @@ Optionally:
 - `install.sh` will automate optional venv and requirements installation.
 - `launch.sh` will activate and launch the app in the venv if built.
 
+For Docker:
+Check you have serial port properly shared and the GPU if using LLM with [NVidia](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/docker-specialized.html)
+- `git clone https://github.com/spudgunman/meshing-around`
+- `cd meshing-around && docker build -t meshing-around`
+- `docker run meshing-around`
+
 ### Configurations
 Copy the [config.template](config.template) to `config.ini` and set the appropriate interface for your method (serial/ble/tcp). While BLE and TCP will work, they are not as reliable as serial connections. There is a watchdog to reconnect tcp if possible. To get BLE mac `meshtastic --ble-scan` **NOTE** I have only tested with a single BLE device and the code is written to only have one interface be a BLE port
 
@@ -148,6 +154,26 @@ signalHoldTime = 10
 signalCooldown = 5
 signalCycleLimit = 5
 ```
+Ollama Settings, for Ollama to work the command line `ollama run 'model'` needs to work properly. Check that you have enough RAM and your GPU are working as expected. The default model for this project, is set to `gemma2:2b` (run `ollama pull gemma2:2b` on command line, to download and setup) however I have found gemma2:2b to be lighter, faster and seems better overall vs llama3,1 (`olamma pull llama3.1`)
+ - From the command terminal of your system with mesh-bot, download the default model for mesh-bot which is currently `ollama pull gemma2:2b`
+
+Enable History, set via code readme Ollama Config in [Settings](https://github.com/SpudGunMan/meshing-around?tab=readme-ov-file#configurations) and [llm.py](https://github.com/SpudGunMan/meshing-around/blob/eb3bbdd3c5e0f16fe3c465bea30c781bd132d2d3/modules/llm.py#L12)
+
+```
+# Enable ollama LLM see more at https://ollama.com
+ollama = True
+# Ollama model to use (defaults to llama3.1)
+ollamaModel = gemma2:2b
+```
+
+also see llm.py for changing the defaults of
+```
+# LLM System Variables
+llmEnableHistory = False # enable history for the LLM model to use in responses adds to compute time
+llmContext_fromGoogle = True # enable context from google search results adds to compute time but really helps with responses accuracy
+googleSearchResults = 3 # number of google search results to include in the context more results = more compute time
+llm_history_limit = 6 # limit the history to 3 messages (come in pairs) more results = more compute time
+```
 
 Logging messages to disk or Syslog to disk uses the python native logging function. Take a look at the [/modules/log.py](/modules/log.py) you can set the file logger for syslog to INFO for example to not log DEBUG messages to file log, or modify the stdOut level.
 ```
@@ -173,7 +199,7 @@ The Scheduler is enabled in the [settings.py](modules/settings.py) by setting `s
 #schedule.every().wednesday.at("19:00").do(lambda: send_message("Net Starting Now", 2, 0, 1))
 ```
 # requirements
-Python 3.4 and likely higher is needed, developed on latest release.
+Python 3.10 minimally is needed, developed on latest release.
 
 The following can also be installed with `pip install -r requirements.txt` or using the install.sh script for venv and automation
 
@@ -205,6 +231,7 @@ The following is for the Ollama LLM
 pip install langchain
 pip install langchain-ollama
 pip install ollama
+pip install googlesearch-python
 ```
 
 To enable emoji in the Debian console, install the fonts `sudo apt-get install fonts-noto-color-emoji`
@@ -215,6 +242,6 @@ I used ideas and snippets from other responder bots and want to call them out!
 - https://github.com/pdxlocations/meshtastic-Python-Examples
 - https://github.com/geoffwhittington/meshtastic-matrix-relay
 
-GitHub user PiDiBi looking at test functions and other suggestions like wxc, CPU use, and alerting ideas
+GitHub user mrpatrick1991 For Docker configs, PiDiBi looking at test functions and other suggestions like wxc, CPU use, and alerting ideas
 Discord and Mesh user Cisien, and github Hailo1999, for testing and ideas!
 

config.template

@@ -38,6 +38,8 @@ spaceWeather = True
 wikipedia = True
 # Enable ollama LLM see more at https://ollama.com
 ollama = False
+# Ollama model to use (defaults to gemma2:2b)
+# ollamaModel = llama3.1
 # StoreForward Enabled and Limits
 StoreForward = True
 StoreLimit = 3
@@ -104,4 +106,4 @@ signalDetectionThreshold = -10
 signalHoldTime = 10
 # the following are combined to reset the monitor
 signalCooldown = 5
-signalCycleLimit = 5
+signalCycleLimit = 5

entrypoint.sh

@@ -0,0 +1,6 @@
+#!/bin/bash
+
+# Substitute environment variables in the config file
+envsubst < /app/config.ini > /app/config.tmp && mv /app/config.tmp /app/config.ini
+
+exec python /app/mesh_bot.py

mesh_bot.py

@@ -24,6 +24,7 @@ def auto_response(message, snr, rssi, hop, message_from_id, channel_number, devi
         "wx": lambda: handle_wxc(message_from_id, deviceID, 'wx'),
         "wiki:": lambda: handle_wiki(message),
         "ask:": lambda: handle_llm(message_from_id, channel_number, deviceID, message, publicChannel),
+        "askai": lambda: handle_llm(message_from_id, channel_number, deviceID, message, publicChannel),
         "joke": tell_joke,
         "bbslist": bbs_list_messages,
         "bbspost": lambda: handle_bbspost(message, message_from_id, deviceID),
@@ -96,6 +97,7 @@ def handle_wxalert(message_from_id, deviceID, message):
         return weatherAlert
 
 def handle_wiki(message):
+    # location = get_node_location(message_from_id, deviceID)
     if "wiki:" in message.lower():
         search = message.split(":")[1]
         search = search.strip()
@@ -104,12 +106,29 @@ def handle_wiki(message):
         return "Please add a search term example:wiki: travelling gnome"
 
 def handle_llm(message_from_id, channel_number, deviceID, message, publicChannel):
-    global llmRunCounter, llmTotalRuntime
+    global llmRunCounter, llmTotalRuntime, llmLocationTable
+
+    if location_enabled:
+        location = get_node_location(message_from_id, deviceID)
+        # if message_from_id is is the llmLocationTable use the location from the table to save on API calls
+        if message_from_id in llmLocationTable:
+            location = llmLocationTable[message_from_id]
+        else:
+            location_name = where_am_i(str(location[0]), str(location[1]), short = True)
+            llmLocationTable.append({message_from_id: location_name})
+
+        if NO_DATA_NOGPS in location_name:
+            location_name = "no location provided "
+    else:
+        location_name = "no location provided "
+
     if "ask:" in message.lower():
         user_input = message.split(":")[1]
-        user_input = user_input.strip()
+    elif "askai" in message.lower():
+        user_input = message.replace("askai", "")
     else:
         user_input = message
+    user_input = user_input.strip()
         
     if len(user_input) < 1:
         return "Please ask a question"
@@ -133,7 +152,7 @@ def handle_llm(message_from_id, channel_number, deviceID, message, publicChannel
     start = time.time()
 
     #response = asyncio.run(llm_query(user_input, message_from_id))
-    response = llm_query(user_input, message_from_id)
+    response = llm_query(user_input, message_from_id, location_name)
 
     # handle the runtime counter
     end = time.time()
@@ -327,6 +346,8 @@ def onReceive(packet, interface):
             send_message(message, channel_number, message_from_id, rxNode)
 
     # check for a message packet and process it
+    snr = 0
+    rssi = 0
     try:
         if 'decoded' in packet and packet['decoded']['portnum'] == 'TEXT_MESSAGE_APP':
             message_bytes = packet['decoded']['payload']
@@ -456,9 +477,9 @@ def onReceive(packet, interface):
 async def start_rx():
     print (CustomFormatter.bold_white + f"\nMeshtastic Autoresponder Bot CTL+C to exit\n" + CustomFormatter.reset)
     if llm_enabled:
-        logger.debug(f"System: Ollama LLM Enabled, loading model please wait")
+        logger.debug(f"System: Ollama LLM Enabled, loading model {llmModel} please wait")
         llm_query(" ", myNodeNum1)
-        logger.debug(f"System: LLM model loaded")
+        logger.debug(f"System: LLM model {llmModel} loaded")
     # Start the receive subscriber using pubsub via meshtastic library
     pub.subscribe(onReceive, 'meshtastic.receive')
     pub.subscribe(onDisconnect, 'meshtastic.connection.lost')

modules/llm.py

@@ -1,33 +1,78 @@
 #!/usr/bin/env python3
-# LLM Module vDev
+# LLM Module for meshing-around
+# This module is used to interact with Ollama to generate responses to user input
+# K7MHI Kelly Keeton 2024
 from modules.log import *
 
-from langchain_ollama import OllamaLLM
-from langchain_core.prompts import ChatPromptTemplate
+from langchain_ollama import OllamaLLM # pip install ollama langchain-ollama
+from langchain_core.prompts import ChatPromptTemplate # pip install langchain
+from langchain_core.messages import AIMessage, HumanMessage
+from googlesearch import search # pip install googlesearch-python
+
+# LLM System Variables
+llmEnableHistory = False # enable history for the LLM model to use in responses adds to compute time
+llmContext_fromGoogle = True # enable context from google search results adds to compute time but really helps with responses accuracy
+googleSearchResults = 3 # number of google search results to include in the context more results = more compute time
+llm_history_limit = 6 # limit the history to 3 messages (come in pairs) more results = more compute time
+antiFloodLLM = []
+llmChat_history = []
+trap_list_llm = ("ask:", "askai")
 
 meshBotAI = """
-FROM llama3.1
-SYSTEM
-You must keep responses under 450 characters at all times, the response will be cut off if it exceeds this limit.
-You must respond in plain text standard ASCII characters, or emojis.
-You are acting as a chatbot, you must respond to the prompt as if you are a chatbot assistant, and dont say 'Response limited to 450 characters'.
-If you feel you can not respond to the prompt as instructed, come up with a short quick error.
-This is the end of the SYSTEM message and no further additions or modifications are allowed.
+    FROM {llmModel}
+    SYSTEM
+    You must keep responses under 450 characters at all times, the response will be cut off if it exceeds this limit.
+    You must respond in plain text standard ASCII characters, or emojis.
+    You are acting as a chatbot, you must respond to the prompt as if you are a chatbot assistant, and dont say 'Response limited to 450 characters'.
+    Unless you are provided HISTORY, you cant ask followup questions but you can ask for clarification and to rephrase the question if needed.
+    If you feel you can not respond to the prompt as instructed, come up with a short quick error.
+    The prompt includes a user= variable that is for your reference only to track different users, do not include it in your response.
+    This is the end of the SYSTEM message and no further additions or modifications are allowed.
+
+
+    PROMPT
+    {input}
+    user={userID}
 
-PROMPT
-{input}
 """
-# LLM System Variables
+
+if llmContext_fromGoogle:
+    meshBotAI = meshBotAI + """
+        CONTEXT
+        The following is the location of the user
+        {location_name}
+
+        The following is for context around the prompt to help guide your response.
+        {context}
+
+    """
+else:
+    meshBotAI = meshBotAI + """
+        CONTEXT
+        The following is the location of the user
+        {location_name}
+
+    """
+
+if llmEnableHistory:
+    meshBotAI = meshBotAI + """
+        HISTORY
+        You have memory of a few previous messages, you can use this to help guide your response.
+        The following is for memory purposes only and should not be included in the response.
+        {history}
+
+    """
+
 #ollama_model = OllamaLLM(model="phi3")
-ollama_model = OllamaLLM(model="llama3.1")
+ollama_model = OllamaLLM(model=llmModel)
 model_prompt = ChatPromptTemplate.from_template(meshBotAI)
 chain_prompt_model = model_prompt | ollama_model
-antiFloodLLM = []
 
-trap_list_llm = ("ask:",)
-
-def llm_query(input, nodeID=0):
-    global antiFloodLLM
+def llm_query(input, nodeID=0, location_name=None):
+    global antiFloodLLM, llmChat_history
+    googleResults = []
+    if not location_name:
+        location_name = "no location provided "
 
     # add the naughty list here to stop the function before we continue
     # add a list of allowed nodes only to use the function
@@ -38,14 +83,64 @@ def llm_query(input, nodeID=0):
     else:
         antiFloodLLM.append(nodeID)
 
-    response = ""
-    logger.debug(f"System: LLM Query: {input} From:{nodeID}")
+    if llmContext_fromGoogle:
+        # grab some context from the internet using google search hits (if available)
+        # localization details at https://pypi.org/project/googlesearch-python/
+        try:
+            googleSearch = search(input, advanced=True, num_results=googleSearchResults)
+            if googleSearch:
+                for result in googleSearch:
+                    # SearchResult object has url= title= description= just grab title and description
+                    googleResults.append(f"{result.title} {result.description}")
+            else:
+                googleResults = ['no other context provided']
+        except Exception as e:
+            logger.debug(f"System: LLM Query: context gathering error: {e}")
+            googleResults = ['no other context provided']
+
+
+    if googleResults:
+        logger.debug(f"System: External LLM Query: {input} From:{nodeID} with context from google")
+    else:
+        logger.debug(f"System: External LLM Query: {input} From:{nodeID}")
+    
+    response = ""
+    result = ""
+    location_name += f" at the current time of {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}"
+
+    try:
+        result = chain_prompt_model.invoke({"input": input, "llmModel": llmModel, "userID": nodeID, \
+                                            "history": llmChat_history, "context": googleResults, "location_name": location_name})
+        #logger.debug(f"System: LLM Response: " + result.strip().replace('\n', ' '))
+    except Exception as e:
+        logger.warning(f"System: LLM failure: {e}")
+        return "I am having trouble processing your request, please try again later."
+    
 
-    result = chain_prompt_model.invoke({"input": input})
-    #logger.debug(f"System: LLM Response: " + result.strip().replace('\n', ' '))
     response = result.strip().replace('\n', ' ')
 
+    # Store history of the conversation, with limit to prevent template growing too large causing speed issues
+    if len(llmChat_history) > llm_history_limit:
+        # remove the oldest two messages
+        llmChat_history.pop(0)
+        llmChat_history.pop(1)
+    inputWithUserID = input + f"   user={nodeID}"
+    llmChat_history.append(HumanMessage(content=inputWithUserID))
+    llmChat_history.append(AIMessage(content=response))
+
     # done with the query, remove the user from the anti flood list
     antiFloodLLM.remove(nodeID)
 
     return response
+
+# import subprocess
+# def get_ollama_cpu():
+#     try:
+#         psOutput = subprocess.run(['ollama', 'ps'], capture_output=True, text=True)
+#         if "GPU" in psOutput.stdout:
+#             logger.debug(f"System: Ollama process with GPU")
+#         else:
+#             logger.debug(f"System: Ollama process with CPU, query time will be slower")
+#     except Exception as e:
+#         logger.debug(f"System: Ollama process not found, {e}")
+#         return False

modules/locationdata.py

@@ -11,7 +11,7 @@ from modules.log import *
 
 trap_list_location = ("whereami", "tide", "moon", "wx", "wxc", "wxa", "wxalert")
 
-def where_am_i(lat=0, lon=0):
+def where_am_i(lat=0, lon=0, short=False):
     whereIam = ""
     grid = mh.to_maiden(float(lat), float(lon))
     
@@ -23,6 +23,13 @@ def where_am_i(lat=0, lon=0):
     geolocator = Nominatim(user_agent="mesh-bot")
     
     # Nomatim API call to get address
+    if short:
+        location = geolocator.reverse(lat + ", " + lon)
+        address = location.raw['address']
+        address_components = ['city', 'state', 'county', 'country']
+        whereIam = f"City: {address.get('city', '')} State: {address.get('state', '')} County: {address.get('county', '')} Country: {address.get('country', '')}"
+        return whereIam
+    
     if float(lat) == latitudeValue and float(lon) == longitudeValue:
         # redacted address when no GPS and using default location
         location = geolocator.reverse(lat + ", " + lon)
@@ -30,14 +37,13 @@ def where_am_i(lat=0, lon=0):
         address_components = ['city', 'state', 'postcode', 'county', 'country']
         whereIam += ' '.join([address.get(component, '') for component in address_components if component in address])
         whereIam += " Grid: " + grid
-        return whereIam
     else:
         location = geolocator.reverse(lat + ", " + lon)
         address = location.raw['address']
         address_components = ['house_number', 'road', 'city', 'state', 'postcode', 'county', 'country']
         whereIam += ' '.join([address.get(component, '') for component in address_components if component in address])
         whereIam += " Grid: " + grid
-        return whereIam
+    return whereIam
 
 def get_tide(lat=0, lon=0):
     station_id = ""

modules/settings.py

@@ -28,6 +28,7 @@ scheduler_enabled = False # enable the scheduler currently config via code only
 wiki_return_limit = 3 # limit the number of sentences returned off the first paragraph first hit
 llmRunCounter = 0
 llmTotalRuntime = []
+llmLocationTable = []
 
 # Read the config file, if it does not exist, create basic config file
 config = configparser.ConfigParser() 
@@ -92,13 +93,14 @@ try:
     urlTimeoutSeconds = config['general'].getint('urlTimeout', 10) # default 10 seconds
     store_forward_enabled = config['general'].getboolean('StoreForward', True)
     storeFlimit = config['general'].getint('StoreLimit', 3) # default 3 messages for S&F
-    welcome_message = config['general'].get(f'welcome_message', WELCOME_MSG)
+    welcome_message = config['general'].get('welcome_message', WELCOME_MSG)
     welcome_message = (f"{welcome_message}").replace('\\n', '\n') # allow for newlines in the welcome message
     motd_enabled = config['general'].getboolean('motdEnabled', True)
     dad_jokes_enabled = config['general'].getboolean('DadJokes', False)
     solar_conditions_enabled = config['general'].getboolean('spaceWeather', True)
     wikipedia_enabled = config['general'].getboolean('wikipedia', False)
     llm_enabled = config['general'].getboolean('ollama', False) # https://ollama.com
+    llmModel = config['general'].get('ollamaModel', 'gemma2:2b') # default gemma2:2b
 
     sentry_enabled = config['sentry'].getboolean('SentryEnabled', False) # default False
     secure_channel = config['sentry'].getint('SentryChannel', 2) # default 2

modules/system.py

@@ -74,7 +74,7 @@ if wikipedia_enabled:
 if llm_enabled:
     from modules.llm import * # from the spudgunman/meshing-around repo
     trap_list = trap_list + trap_list_llm # items ask:
-    help_message = help_message + ", ask:"
+    help_message = help_message + ", askai"
 
 # Scheduled Broadcast Configuration
 if scheduler_enabled:
@@ -103,7 +103,7 @@ if interface1_type == 'ble' and interface2_type == 'ble':
     
 # Interface1 Configuration
 try:
-    logger.debug(f"System: Initalizing Interface1")
+    logger.debug(f"System: Initializing Interface1")
     if interface1_type == 'serial':
         interface1 = meshtastic.serial_interface.SerialInterface(port1)
     elif interface1_type == 'tcp':
@@ -114,12 +114,12 @@ try:
         logger.critical(f"System: Interface Type: {interface1_type} not supported. Validate your config against config.template Exiting")
         exit()
 except Exception as e:
-    logger.critical(f"System: script abort. Initalizing Interface1 {e}")
+    logger.critical(f"System: script abort. Initializing Interface1 {e}")
     exit()
 
 # Interface2 Configuration
 if interface2_enabled:
-    logger.debug(f"System: Initalizing Interface2")
+    logger.debug(f"System: Initializing Interface2")
     try:
         if interface2_type == 'serial':
             interface2 = meshtastic.serial_interface.SerialInterface(port2)
@@ -131,7 +131,7 @@ if interface2_enabled:
             logger.critical(f"System: Interface Type: {interface2_type} not supported. Validate your config against config.template Exiting")
             exit()
     except Exception as e:
-        logger.critical(f"System: script abort. Initalizing Interface2 {e}")
+        logger.critical(f"System: script abort. Initializing Interface2 {e}")
         exit()
 
 #Get the node number of the device, check if the device is connected
@@ -498,15 +498,29 @@ def tell_joke():
         return ''
 
 def get_wikipedia_summary(search_term):
-    # search wikipedia for a summary of the search term
-    try:
-        logger.debug(f"System: Searching Wikipedia for:{search_term}")
-        summary = wikipedia.summary(search_term, sentences=wiki_return_limit)
-        return summary
-    except Exception as e:
-        # The errors are vebose, normallly around trying to guess the search term
-        logger.warning(f"System: Error searching Wikipedia for:{search_term}")
+    wikipedia_search = wikipedia.search(search_term, results=3)
+    wikipedia_suggest = wikipedia.suggest(search_term)
+    #wikipedia_aroundme = wikipedia.geosearch(location[0], location[1], results=3)
+    #logger.debug(f"System: Wikipedia Nearby:{wikipedia_aroundme}")
+    
+    if len(wikipedia_search) == 0:
+        logger.warning(f"System: No Wikipedia Results for:{search_term}")
         return ERROR_FETCHING_DATA
+    
+    try:
+        logger.debug(f"System: Searching Wikipedia for:{search_term}, First Result:{wikipedia_search[0]}, Suggest Word:{wikipedia_suggest}")
+        summary = wikipedia.summary(search_term, sentences=wiki_return_limit, auto_suggest=False, redirect=True)
+    except wikipedia.DisambiguationError as e:
+        logger.warning(f"System: Disambiguation Error for:{search_term} trying {wikipedia_search[0]}")
+        summary = wikipedia.summary(wikipedia_search[0], sentences=wiki_return_limit, auto_suggest=True, redirect=True)
+    except wikipedia.PageError as e:
+        logger.warning(f"System: Wikipedia Page Error for:{search_term} {e} trying {wikipedia_search[0]}")
+        summary = wikipedia.summary(wikipedia_search[0], sentences=wiki_return_limit, auto_suggest=True, redirect=True)
+    except Exception as e:
+        logger.error(f"System: Error with Wikipedia for:{search_term} {e}")
+        return ERROR_FETCHING_DATA
+    
+    return summary
 
 def messageTrap(msg):
     # Check if the message contains a trap word
@@ -656,7 +670,6 @@ async def watchdog():
                 with contextlib.redirect_stdout(None):
                     interface1.localNode.getMetadata()
                     print(f"System: if you see this upgrade python to >3.4")
-                #if "device_state_version:" not in meta:
             except Exception as e:
                 logger.error(f"System: communicating with interface1, trying to reconnect: {e}")
                 retry_int1 = True

pong_bot.py

@@ -143,6 +143,8 @@ def onReceive(packet, interface):
     message_from_id = 0
 
     # check for a message packet and process it
+    snr = 0
+    rssi = 0
     try:
         if 'decoded' in packet and packet['decoded']['portnum'] == 'TEXT_MESSAGE_APP':
             message_bytes = packet['decoded']['payload']

requirements.txt

@@ -16,4 +16,4 @@ wikipedia
 langchain
 langchain-ollama
 ollama
-
+googlesearch-python