Upload 6 files

Dockerfile

@@ -0,0 +1,50 @@
+# Use the official Python image as a base image
+FROM python:3.11-slim
+
+# Set environment variables
+ENV PYTHONUNBUFFERED=1 \
+    PYTHONDONTWRITEBYTECODE=1 \
+    GRADIO_SERVER_NAME="0.0.0.0" \
+    REPO_ID="microsoft/Phi-3-mini-4k-instruct-gguf" \
+    MODEL_FILE="Phi-3-mini-4k-instruct-q4.gguf"
+
+# Create a working directory
+WORKDIR /app
+
+# Install dependencies for building C++ extensions and SSL
+RUN apt-get update && \
+    apt-get install -y --no-install-recommends \
+    build-essential \
+    cmake \
+    openssl \
+    wget \
+    && apt-get clean \
+    && rm -rf /var/lib/apt/lists/*
+
+# Copy the requirements file
+COPY requirements.txt /app/
+
+# Install Python dependencies
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Copy SSL certificates
+COPY certificates /app/certificates
+
+# Copy the rest of the application code
+COPY . /app/
+
+# Create the cache directory with appropriate permissions
+RUN mkdir -p /app/hf_cache && chmod -R 777 /app/hf_cache
+
+# Ensure the model is downloaded
+RUN python download_model.py
+
+# Expose the port that Gradio will run on
+EXPOSE 7860
+
+# Set the user to root
+USER root
+
+# Run the application
+CMD ["python", "app.py"]
+

app.py

@@ -0,0 +1,101 @@
+import os
+import gradio as gr
+from langchain.agents import Tool
+from langchain_community.llms import LlamaCpp
+from langchain.agents import initialize_agent
+from functions import get_weather_info, get_forecast, shutdown
+from huggingface_hub import hf_hub_download
+
+# Set custom cache directory
+cache_dir = "/app/hf_cache"
+os.makedirs(cache_dir, exist_ok=True)
+
+# Download the model directly in the app
+model_path = hf_hub_download(
+    repo_id=os.environ.get("REPO_ID", "microsoft/Phi-3-mini-4k-instruct-gguf"),
+    filename=os.environ.get("MODEL_FILE", "Phi-3-mini-4k-instruct-q4.gguf"),
+    cache_dir=cache_dir
+)
+
+# Initialize the LlamaCpp model
+llm = LlamaCpp(
+    model_path=model_path,
+    n_ctx=4096,
+    n_gpu_layers=-1
+)
+
+# Define tools
+weather_tool = Tool(
+    name="WeatherLookup",
+    func=lambda city: get_weather_info(city),
+    description="Useful to get the current weather (Today) information for a city. It includes information on temperature, pressure, humidity, wind, clouds, and rain."
+)
+
+forecast_tool = Tool(
+    name="ForecastLookup",
+    func=lambda city: get_forecast(city),
+    description="Useful to get the weather forecast for the next two days for a city. It includes information on temperature, pressure, humidity, wind, clouds, and rain."
+)
+
+# Tools (Include both Weather and Forecast Tools)
+tools = [weather_tool, forecast_tool]
+
+# Initialize Agent
+agent = initialize_agent(tools, llm, agent="zero-shot-react-description", verbose=True)
+
+def respond(message, history):
+    try:
+        # Create the prompt based on the history
+        prompt = "\n".join([f"{'User' if i % 2 == 0 else 'Assistant'}: {m[0]}" for i, m in enumerate(history)]) + "\nAssistant:"
+
+        # Generate response using LangChain agent
+        response = agent.run(message)
+        
+        # Update history with the assistant's response
+        history.append((message, response))
+
+        return response, history
+    except Exception as e:
+        return f"An error occurred: {e}", history
+
+# Define the Gradio interface
+with gr.Blocks(css="style.css") as demo:
+    gr.Markdown(
+        """
+        # Weather Chatbot
+        Get real-time weather forecasts or chat with our assistant. Type your queries in natural language.
+        """
+    )
+    with gr.Row():
+        with gr.Column():
+            message = gr.Textbox(label="Ask a weather question or chat with the assistant", lines=2, placeholder="Type your question here...")
+            response = gr.Textbox(label="Response", lines=2)
+            state = gr.State([])
+            btn = gr.Button("Submit")
+            btn.click(respond, [message, state], [response, state])
+            shutdown_btn = gr.Button("Shutdown")
+            shutdown_btn.click(shutdown, [], response)
+    
+    gr.Examples(
+        examples=[
+            ["What's the weather in New York?"],
+            ["Tell me the weather forecast for Tokyo."],
+            ["What's the temperature in London?"]
+        ],
+        inputs=message
+    )
+
+# Launch the Gradio interface
+def main():
+    demo.launch(
+        server_name="0.0.0.0", 
+        server_port=7860,
+        ssl_keyfile="/app/certificates/selfsigned.key",
+        ssl_certfile="/app/certificates/selfsigned.crt",
+        ssl_verify=False,  # Disable SSL verification for development
+        share=True,
+    )
+
+if __name__ == "__main__":
+    main()
+

download_model.py

@@ -0,0 +1,15 @@
+import os
+from huggingface_hub import hf_hub_download
+
+def download_model():
+    model_path = "./model/phi-3-gguf/Phi-3-mini-4k-instruct-q4.gguf"
+    if not os.path.exists(model_path):
+        hf_hub_download(
+            repo_id="microsoft/Phi-3-mini-4k-instruct-gguf",
+            filename="Phi-3-mini-4k-instruct-q4.gguf",
+            local_dir="model/phi-3-gguf"
+        )
+
+if __name__ == "__main__":
+    download_model()
+

functions.py

@@ -0,0 +1,280 @@
+import requests
+from collections import defaultdict
+from datetime import datetime, timedelta
+import json
+import pandas as pd
+import threading
+import time
+import os
+import signal
+
+api_key = "c6dfc4d92a8f972d237ef696ec87b37a"
+
+def shutdown():
+    # Wait a bit before shutdown to allow the response to be returned
+    def stop():
+        time.sleep(1)
+        os.kill(os.getpid(), signal.SIGTERM)  # Send SIGTERM to the current process to stop Gradio
+        os._exit(0)
+    threading.Thread(target=stop).start()
+    return "Shutting down and closing the Gradio window..."
+
+def get_weather_info(city):
+    """Fetches current weather information for a city using OpenWeatherMap API."""
+    
+    url_current = f"http://api.openweathermap.org/data/2.5/weather?q={city}&appid={api_key}&units=metric"
+    response_current = requests.get(url_current)
+    if response_current.status_code != 200:
+        return "Error: Could not fetch weather data."
+    data_current = response_current.json()
+
+    weather_description = data_current['weather'][0]['description']
+    temperature_current = data_current['main']['temp']
+    temperature_feels_like = data_current['main']['feels_like']
+    temperature_min = data_current['main']['temp_min']
+    temperature_max = data_current['main']['temp_max']
+    pressure_sea_level = data_current['main'].get('sea_level', data_current['main']['pressure'])
+    pressure_ground_level = data_current['main'].get('grnd_level', data_current['main']['pressure'])
+    humidity = data_current['main']['humidity']
+    visibility = data_current['visibility']
+    wind_speed = data_current['wind']['speed']
+    wind_deg = data_current['wind']['deg']
+    clouds = data_current['clouds']['all']
+    rain = data_current.get('rain', {}).get('1h', 0)
+    dt = datetime.utcfromtimestamp(data_current['dt']).strftime('%Y-%m-%d %H:%M:%S')
+    timezone = data_current['timezone']
+    city_name = data_current['name']
+    response_code = data_current['cod']
+
+    formatted_info = (
+        f"Weather: {weather_description}, "
+        f"Temperature: current {temperature_current}°C, feels like {temperature_feels_like}°C, min {temperature_min}°C, max {temperature_max}°C, "
+        f"Pressure: sea level {pressure_sea_level} hPa, ground level {pressure_ground_level} hPa, "
+        f"Humidity: {humidity}%, "
+        f"Visibility: {visibility} meters, "
+        f"Wind: speed {wind_speed} m/s, deg {wind_deg}, "
+        f"Clouds: {clouds}%, "
+        f"Rain: {rain} mm, "
+        f"Date/Time: {dt}, "
+        f"Timezone: {timezone} seconds, "
+        f"City Name: {city_name}, "
+        f"Response Code: {response_code}"
+    )
+
+    return formatted_info
+    
+def get_forecast(city):
+    """Fetches 2-day weather forecast for a city using OpenWeatherMap API and restructures the data into a table format."""
+    url_forecast = f"http://api.openweathermap.org/data/2.5/forecast?q={city}&appid={api_key}&units=metric"
+    response_forecast = requests.get(url_forecast)
+    if response_forecast.status_code != 200:
+        return "Error: Could not fetch forecast data."
+    forecast_json = response_forecast.json()
+
+    current_date = datetime.now().date()
+    forecast_dates = [current_date + timedelta(days=i) for i in range(1, 3)]
+    important_hours = ['09:00:00', '15:00:00', '21:00:00']
+
+    data = []
+
+    for entry in forecast_json['list']:
+        date, time = entry['dt_txt'].split()
+        date_obj = datetime.strptime(date, '%Y-%m-%d').date()
+        if date_obj in forecast_dates and time in important_hours:
+            data.append({
+                'Date': date,
+                'Time': time,
+                'Temperature': entry['main']['temp'],
+                'Feels Like': entry['main']['feels_like'],
+                'Temp Min': entry['main']['temp_min'],
+                'Temp Max': entry['main']['temp_max'],
+                'Pressure': entry['main']['pressure'],
+                'Humidity': entry['main']['humidity'],
+                'Weather': entry['weather'][0]['description'],
+                'Icon': entry['weather'][0]['icon'],
+                'Wind Speed': entry['wind']['speed'],
+                'Wind Deg': entry['wind']['deg'],
+                'Visibility': entry['visibility'],
+                'Pop': entry['pop'],
+                'Rain': entry['rain']['3h'] if 'rain' in entry else 0,
+                'Clouds': entry['clouds']['all']
+            })
+
+    df = pd.DataFrame(data)
+    df.set_index(['Date', 'Time'], inplace=True)
+    return df
+
+def restructure_forecast_00(forecast_json):
+    """Restructures the forecast JSON data into a single-line sentence format."""
+    current_date = datetime.now().date()
+    forecast_dates = [current_date + timedelta(days=i) for i in range(1, 3)]
+    important_hours = ['09:00:00', '12:00:00', '15:00:00', '18:00:00', '21:00:00']
+
+    structured_data = defaultdict(dict)
+    
+    for entry in forecast_json['list']:
+        date, time = entry['dt_txt'].split()
+        date_obj = datetime.strptime(date, '%Y-%m-%d').date()
+        if date_obj in forecast_dates and time in important_hours:
+            structured_data[date][time] = {
+                'temperature': entry['main']['temp'],
+                'feels like': entry['main']['feels_like'],
+                'temp min': entry['main']['temp_min'],
+                'temp max': entry['main']['temp_max'],
+                'pressure': entry['main']['pressure'],
+                'humidity': entry['main']['humidity'],
+                'weather': entry['weather'][0]['description'],
+                'icon': entry['weather'][0]['icon'],
+                'wind speed': entry['wind']['speed'],
+                'wind deg': entry['wind']['deg'],
+                'visibility': entry['visibility'],
+                'pop': entry['pop'],
+                'rain': entry['rain']['3h'] if 'rain' in entry else 0,
+                'clouds': entry['clouds']['all']
+            }
+
+    return format_forecast(structured_data, forecast_dates)
+
+def format_forecast_00(structured_data, forecast_dates):
+    """Formats the structured forecast data into a single-line sentence format."""
+    formatted_forecast = []
+    for date in forecast_dates:
+        date_str = str(date)
+        for time, data in structured_data[date_str].items():
+            formatted_forecast.append(
+                f"{date_str} : {time} ( " +
+                ", ".join(f"{key} - {value}" for key, value in data.items()) +
+                " )"
+            )
+    return "\n".join(formatted_forecast)    
+    
+
+def restructure_forecast2(forecast_json):
+    """Restructures the forecast JSON data into a nested dictionary by date and time, including the next three days."""
+    current_date = datetime.now().date()
+    forecast_dates = [current_date + timedelta(days=i) for i in range(1, 3)]
+
+    structured_data = defaultdict(dict)
+    
+    for entry in forecast_json['list']:
+        date, time = entry['dt_txt'].split()
+        date_obj = datetime.strptime(date, '%Y-%m-%d').date()
+        if date_obj in forecast_dates:
+            structured_data[date][time] = {
+                'temperature': entry['main']['temp'],
+                'feels_like': entry['main']['feels_like'],
+                'temp_min': entry['main']['temp_min'],
+                'temp_max': entry['main']['temp_max'],
+                'pressure': entry['main']['pressure'],
+                'humidity': entry['main']['humidity'],
+                'weather': entry['weather'][0]['description'],
+                'icon': entry['weather'][0]['icon'],
+                'wind_speed': entry['wind']['speed'],
+                'wind_deg': entry['wind']['deg'],
+                'visibility': entry['visibility'],
+                'pop': entry['pop'],
+                'rain': entry['rain']['3h'] if 'rain' in entry else 0,
+                'clouds': entry['clouds']['all']
+            }
+    
+    return {str(date): structured_data[str(date)] for date in forecast_dates}
+
+def restructure_forecast_0(forecast_json):
+    """Restructures the forecast JSON data into a nested dictionary by date and specific times."""
+    current_date = datetime.now().date()
+    forecast_dates = [current_date + timedelta(days=i) for i in range(1, 3)]
+    important_hours = ['09:00:00', '12:00:00', '15:00:00', '18:00:00', '21:00:00']
+
+    structured_data = defaultdict(dict)
+    
+    for entry in forecast_json['list']:
+        date, time = entry['dt_txt'].split()
+        date_obj = datetime.strptime(date, '%Y-%m-%d').date()
+        if date_obj in forecast_dates and time in important_hours:
+            structured_data[date][time] = {
+                'temperature': entry['main']['temp'],
+                'feels_like': entry['main']['feels_like'],
+                'temp_min': entry['main']['temp_min'],
+                'temp_max': entry['main']['temp_max'],
+                'pressure': entry['main']['pressure'],
+                'humidity': entry['main']['humidity'],
+                'weather': entry['weather'][0]['description'],
+                'icon': entry['weather'][0]['icon'],
+                'wind_speed': entry['wind']['speed'],
+                'wind_deg': entry['wind']['deg'],
+                'visibility': entry['visibility'],
+                'pop': entry['pop'],
+                'rain': entry['rain']['3h'] if 'rain' in entry else 0,
+                'clouds': entry['clouds']['all']
+            }
+
+    return {str(date): structured_data[str(date)] for date in forecast_dates}
+
+
+def restructure_forecast3(forecast_json):
+    """Restructures the forecast JSON data into a nested dictionary by date and time, including the next three days."""
+    current_date = datetime.now().date()
+    forecast_dates = [current_date + timedelta(days=i) for i in range(1, 4)]
+
+    structured_data = defaultdict(dict)
+    
+    for entry in forecast_json['list']:
+        date, time = entry['dt_txt'].split()
+        date_obj = datetime.strptime(date, '%Y-%m-%d').date()
+        if date_obj in forecast_dates:
+            structured_data[date][time] = {
+                'temperature': entry['main']['temp'],
+                'feels_like': entry['main']['feels_like'],
+                'temp_min': entry['main']['temp_min'],
+                'temp_max': entry['main']['temp_max'],
+                'pressure': entry['main']['pressure'],
+                'humidity': entry['main']['humidity'],
+                'weather': entry['weather'][0]['description'],
+                'icon': entry['weather'][0]['icon'],
+                'wind_speed': entry['wind']['speed'],
+                'wind_deg': entry['wind']['deg'],
+                'visibility': entry['visibility'],
+                'pop': entry['pop'],
+                'rain': entry['rain']['3h'] if 'rain' in entry else 0,
+                'clouds': entry['clouds']['all']
+            }
+    
+    return {str(date): structured_data[str(date)] for date in forecast_dates}
+    
+    
+def get_weather_info_0(city):
+    """Fetches current weather information for a city using OpenWeatherMap API."""
+    
+    url_current = f"http://api.openweathermap.org/data/2.5/weather?q={city}&appid={api_key}&units=metric"
+    response_current = requests.get(url_current)
+    if response_current.status_code != 200:
+        return "Error: Could not fetch weather data."
+    data_current = response_current.json()
+
+    response = {
+        'coordinates': data_current['coord'],
+        'weather': data_current['weather'][0],
+        'temperature': {
+            'current': data_current['main']['temp'],
+            'feels_like': data_current['main']['feels_like'],
+            'min': data_current['main']['temp_min'],
+            'max': data_current['main']['temp_max']
+        },
+        'pressure': {
+            'sea_level': data_current['main'].get('sea_level', data_current['main']['pressure']),
+            'ground_level': data_current['main'].get('grnd_level', data_current['main']['pressure'])
+        },
+        'humidity': data_current['main']['humidity'],
+        'visibility': data_current['visibility'],
+        'wind': data_current['wind'],
+        'clouds': data_current['clouds'],
+        'rain': data_current.get('rain', {}),
+        'dt': data_current['dt'],
+        'sys': data_current['sys'],
+        'timezone': data_current['timezone'],
+        'id': data_current['id'],
+        'name': data_current['name'],
+        'cod': data_current['cod']
+    }
+
+    return json.dumps(response, indent=2)

requirements.txt

@@ -0,0 +1,8 @@
+gradio
+pandas
+transformers
+llama-cpp-python
+langchain
+langchain-community
+huggingface_hub
+python-dotenv

style.css

@@ -0,0 +1,11 @@
+body {
+    background-color: #f0f4f8;
+}
+.gr-textbox, .gr-output {
+    background-color: #ffffff;
+    border: 1px solid #e0e7ec;
+}
+.gr-button {
+    background-color: #d0d7dc;
+    color: #000000;
+}