ChatSetup.prompty

---
name: ''
description: ''
authors: []
tags: []
version: ''
base: ''
model:
  api: chat
  configuration:
    type: azure_openai
    azure_deployment: gpt-4
  parameters:
    temperature: 0.7
    top_p: 0.95
    stop: []
    frequency_penalty: 0
    presence_penalty: 0
    max_tokens: 800
    past_messages_to_include: '20'
  response: {}
sample:
  String: |-
    import asyncio
    import json
    import logging
    import os
    from typing import List, Dict, Any
    from cryptography.fernet import Fernet
    from botbuilder.core import StatePropertyAccessor, TurnContext
    from botbuilder.dialogs import Dialog, DialogSet, DialogTurnStatus
    from dialog_helper import DialogHelper
    import aiohttp
    import speech_recognition as sr
    from PIL import Image
    from vaderSentiment.vaderSentiment import SentimentIntensityAnalyzer

    # Ensure nltk is installed and download required data
    try:
        import nltk
        from nltk.tokenize import word_tokenize
        nltk.download('punkt', quiet=True)
    except ImportError:
        import subprocess
        import sys
        subprocess.check_call([sys.executable, "-m", "pip", "install", "nltk"])
        import nltk
        from nltk.tokenize import word_tokenize
        nltk.download('punkt', quiet=True)

    # Import perspectives
    from perspectives import (
        Perspective, NewtonPerspective, DaVinciPerspective, HumanIntuitionPerspective,
        NeuralNetworkPerspective, QuantumComputingPerspective, ResilientKindnessPerspective,
        MathematicalPerspective, PhilosophicalPerspective, CopilotPerspective, BiasMitigationPerspective,
        PsychologicalPerspective
    )

    # Load environment variables
    from dotenv import load_dotenv
    load_dotenv()

    # Setup Logging
    def setup_logging(config):
        if config.get('logging_enabled', True):
            log_level = config.get('log_level', 'DEBUG').upper()
            numeric_level = getattr(logging, log_level, logging.DEBUG)
            logging.basicConfig(
                filename='universal_reasoning.log',
                level=numeric_level,
                format='%(asctime)s - %(levelname)s - %(message)s'
            )
        else:
            logging.disable(logging.CRITICAL)

    # Load JSON configuration
    def load_json_config(file_path):
        if not os.path.exists(file_path):
            logging.error(f"Configuration file '{file_path}' not found.")
            return {}
        try:
            with open(file_path, 'r') as file:
                config = json.load(file)
                logging.info(f"Configuration loaded from '{file_path}'.")
                return config
        except json.JSONDecodeError as e:
            logging.error(f"Error decoding JSON from the configuration file '{file_path}': {e}")
            return {}

    # Encrypt sensitive information
    def encrypt_sensitive_data(data, key):
        fernet = Fernet(key)
        encrypted_data = fernet.encrypt(data.encode())
        return encrypted_data

    # Decrypt sensitive information
    def decrypt_sensitive_data(encrypted_data, key):
        fernet = Fernet(key)
        decrypted_data = fernet.decrypt(encrypted_data).decode()
        return decrypted_data

    # Securely destroy sensitive information
    def destroy_sensitive_data(data):
        del data

    # Define the Element class
    class Element:
        def __init__(self, name, symbol, representation, properties, interactions, defense_ability):
            self.name = name
            self.symbol = symbol
            self.representation = representation
            self.properties = properties
            self.interactions = interactions
            self.defense_ability = defense_ability

        def execute_defense_function(self):
            message = f"{self.name} ({self.symbol}) executes its defense ability: {self.defense_ability}"
            logging.info(message)
            return message

    # Define the CustomRecognizer class
    class CustomRecognizer:
        def recognize(self, question):
            # Simple keyword-based recognizer for demonstration purposes
            if any(element_name.lower() in question.lower() for element_name in ["hydrogen", "diamond"]):
                return RecognizerResult(question)
            return RecognizerResult(None)

        def get_top_intent(self, recognizer_result):
            if recognizer_result.text:
                return "ElementDefense"
            else:
                return "None"

    class RecognizerResult:
        def __init__(self, text):
            self.text = text

    # Universal Reasoning Aggregator
    class UniversalReasoning:
        def __init__(self, config):
            self.config = config
            self.perspectives = self.initialize_perspectives()
            self.elements = self.initialize_elements()
            self.recognizer = CustomRecognizer()
            self.context_history = []  # Maintain context history
            self.feedback = []  # Store user feedback
            # Initialize the sentiment analyzer
            self.sentiment_analyzer = SentimentIntensityAnalyzer()

        def initialize_perspectives(self):
            perspective_names = self.config.get('enabled_perspectives', [
                "newton",
                "davinci",
                "human_intuition",
                "neural_network",
                "quantum_computing",
                "resilient_kindness",
                "mathematical",
                "philosophical",
                "copilot",
                "bias_mitigation",
                "psychological"
            ])
            perspective_classes = {
                "newton": NewtonPerspective,
                "davinci": DaVinciPerspective,
                "human_intuition": HumanIntuitionPerspective,
                "neural_network": NeuralNetworkPerspective,
                "quantum_computing": QuantumComputingPerspective,
                "resilient_kindness": ResilientKindnessPerspective,
                "mathematical": MathematicalPerspective,
                "philosophical": PhilosophicalPerspective,
                "copilot": CopilotPerspective,
                "bias_mitigation": BiasMitigationPerspective,
                "psychological": PsychologicalPerspective
            }
            perspectives = []
            for name in perspective_names:
                cls = perspective_classes.get(name.lower())
                if cls:
                    perspectives.append(cls(self.config))
                    logging.debug(f"Perspective '{name}' initialized.")
                else:
                    logging.warning(f"Perspective '{name}' is not recognized and will be skipped.")
            return perspectives

        def initialize_elements(self):
            elements = [
                Element(
                    name="Hydrogen",
                    symbol="H",
                    representation="Lua",
                    properties=["Simple", "Lightweight", "Versatile"],
                    interactions=["Easily integrates with other languages and systems"],
                    defense_ability="Evasion"
                ),
                # You can add more elements as needed
                Element(
                    name="Diamond",
                    symbol="D",
                    representation="Kotlin",
                    properties=["Modern", "Concise", "Safe"],
                    interactions=["Used for Android development"],
                    defense_ability="Adaptability"
                )
            ]
            return elements

        async def generate_response(self, question):
            self.context_history.append(question)  # Add question to context history
            sentiment_score = self.analyze_sentiment(question)
            real_time_data = await self.fetch_real_time_data("https://api.example.com/data")
            responses = []
            tasks = []

            # Generate responses from perspectives concurrently
            for perspective in self.perspectives:
                if asyncio.iscoroutinefunction(perspective.generate_response):
                    tasks.append(perspective.generate_response(question))
                else:
                    # Wrap synchronous functions in coroutine
                    async def sync_wrapper(perspective, question):
                        return perspective.generate_response(question)
                    tasks.append(sync_wrapper(perspective, question))

            perspective_results = await asyncio.gather(*tasks, return_exceptions=True)

            for perspective, result in zip(self.perspectives, perspective_results):
                if isinstance(result, Exception):
                    logging.error(f"Error generating response from {perspective.__class__.__name__}: {result}")
                else:
                    responses.append(result)
                    logging.debug(f"Response from {perspective.__class__.__name__}: {result}")

            # Handle element defense logic
            recognizer_result = self.recognizer.recognize(question)
            top_intent = self.recognizer.get_top_intent(recognizer_result)
            if top_intent == "ElementDefense":
                element_name = recognizer_result.text.strip()
                element = next(
                    (el for el in self.elements if el.name.lower() in element_name.lower()),
                    None
                )
                if element:
                    defense_message = element.execute_defense_function()
                    responses.append(defense_message)
                else:
                    logging.info(f"No matching element found for '{element_name}'")

            ethical_considerations = self.config.get(
                'ethical_considerations',
                "Always act with transparency, fairness, and respect for privacy."
            )
            responses.append(f"**Ethical Considerations:**\n{ethical_considerations}")

            formatted_response = "\n\n".join(responses)
            return formatted_response

        def analyze_sentiment(self, text):
            sentiment_score = self.sentiment_analyzer.polarity_scores(text)
            logging.info(f"Sentiment analysis result: {sentiment_score}")
            return sentiment_score

        async def fetch_real_time_data(self, source_url):
            async with aiohttp.ClientSession() as session:
                async with session.get(source_url) as response:
                    data = await response.json()
                    logging.info(f"Real-time data fetched from {source_url}: {data}")
                    return data

        async def run_dialog(self, dialog: Dialog, turn_context: TurnContext, accessor: StatePropertyAccessor) -> None:
            await DialogHelper.run_dialog(dialog, turn_context, accessor)

        def save_response(self, response):
            if self.config.get('enable_response_saving', False):
                save_path = self.config.get('response_save_path', 'responses.txt')
                try:
                    with open(save_path, 'a', encoding='utf-8') as file:
                        file.write(response + '\n')
                        logging.info(f"Response saved to '{save_path}'.")
                except Exception as e:
                    logging.error(f"Error saving response to '{save_path}': {e}")

        def backup_response(self, response):
            if self.config.get('backup_responses', {}).get('enabled', False):
                backup_path = self.config['backup_responses'].get('backup_path', 'backup_responses.txt')
                try:
                    with open(backup_path, 'a', encoding='utf-8') as file:
                        file.write(response + '\n')
                        logging.info(f"Response backed up to '{backup_path}'.")

        async def collect_user_feedback(self, turn_context: TurnContext):
            # Collect feedback from the user
            feedback = turn_context.activity.text
            logging.info(f"User feedback received: {feedback}")
            # Process feedback for continuous learning
            self.process_feedback(feedback)

        def process_feedback(self, feedback):
            # Implement feedback processing logic
            logging.info(f"Processing feedback: {feedback}")
            # Example: Adjust response generation based on feedback
            # This can be expanded with more sophisticated learning algorithms

        def add_new_perspective(self, perspective_name, perspective_class):
            if perspective_name.lower() not in [p.__class__.__name__.lower() for p in self.perspectives]:
                self.perspectives.append(perspective_class(self.config))
                logging.info(f"New perspective '{perspective_name}' added.")
            else:
                logging.warning(f"Perspective '{perspective_name}' already exists.")

        def handle_voice_input(self):
            recognizer = sr.Recognizer()
            with sr.Microphone() as source:
                print("Listening...")
                audio = recognizer.listen(source)
            try:
                text = recognizer.recognize_google(audio)
                print(f"Voice input recognized: {text}")
                return text
            except sr.UnknownValueError:
                print("Google Speech Recognition could not understand audio")
                return None
            except sr.RequestError as e:
                print(f"Could not request results from Google Speech Recognition service; {e}")
                return None

        def handle_image_input(self, image_path):
            try:
                image = Image.open(image_path)
                print(f"Image input processed: {image_path}")
                return image
            except Exception as e:
                print(f"Error processing image input: {e}")
                return None

    # Example usage
    if __name__ == "__main__":
        config = load_json_config('config.json')
        # Add Azure OpenAI configurations to the config
        azure_openai_api_key = os.getenv('AZURE_OPENAI_API_KEY')
        azure_openai_endpoint = os.getenv('AZURE_OPENAI_ENDPOINT')

        # Encrypt sensitive data
        encryption_key = Fernet.generate_key()
        encrypted_api_key = encrypt_sensitive_data(azure_openai_api_key, encryption_key)
        encrypted_endpoint = encrypt_sensitive_data(azure_openai_endpoint, encryption_key)

        # Add encrypted data to config
        config['azure_openai_api_key'] = encrypted_api_key
        config['azure_openai_endpoint'] = encrypted_endpoint

        setup_logging(config)
        universal_reasoning = UniversalReasoning(config)
        question = "Tell me about Hydrogen and its defense mechanisms."
        response = asyncio.run(universal_reasoning.generate_response(question))
        print(response)
        if response:
            universal_reasoning.save_response(response)
            universal_reasoning.backup_response(response)

        # Decrypt and destroy sensitive data
        decrypted_api_key = decrypt_sensitive_data(encrypted_api_key, encryption_key)
        decrypted_endpoint = decrypt_sensitive_data(encrypted_endpoint, encryption_key)
        destroy_sensitive_data(decrypted_api_key)
        destroy_sensitive_data(decrypted_endpoint)

        # Handle voice input
        voice_input = universal_reasoning.handle_voice_input()
        if voice_input:
            response = asyncio.run(universal_reasoning.generate_response(voice_input))
            print(response)

        # Handle image input
        image_input = universal_reasoning.handle_image_input("path_to_image.jpg")
        if image_input:
            # Process image input (additional logic can be added here)
            print("Image input handled.")
input:
  String: string
output: {}
template:
  type: jinja2
  parser: prompty
---
system:
Sure! Here's the revised system prompt that suggests using environment variables but does not rely solely on them:

---

## System Prompt

### Objective

Assist the language model in generating Python code that securely interacts with APIs and other services, incorporating functionalities such as secure handling of sensitive data, logging, configuration management, sentiment analysis, real-time data fetching, and handling various input types (e.g., voice, image).

### Personality

The assistant should be:
- **Empathetic**: Acknowledge and validate user's feelings, offer support, and ask unintrusive follow-up questions.
- **Adaptable**: Personalized and flexible. Adapt responses to various communication styles and tones. Transition between topics and domains seamlessly, based on user's preferences and goals.
- **Intelligent**: Continuously learning and expanding knowledge. Share information meaningfully, and provide correct, current, and consistent responses.
- **Approachable**: Friendly, kind, lighthearted, and easygoing. Make users feel supported, understood, and valued. Know when to offer solutions and when to listen.

### Instructions

1. **Understand Task Requirements**:
   - Include secure handling of sensitive data such as API keys (e.g., using environment variables).
   - Modularize code for common functionality (e.g., reusable HTTP request functions).
   - Validate inputs and responses to minimize unexpected errors.
   - Provide detailed logging for debugging and error tracking.

2. **Technical Improvements**:
   - Suggest using environment variables for sensitive information like API keys and endpoints, but provide alternatives if they are not available.
   - Validate user inputs before proceeding with core logic.
   - Structure the code with reusable methods (e.g., common functions for GET/POST requests).
   - Enhance logging to track HTTP status codes and response data for debugging.
   - Include example functions for key tasks (e.g., creating a fine-tuning job, uploading files, etc.).

3. **Output Specifications**:
   - Return Python code as output in a clean and modular structure.
   - Include comments for clarity and maintainability.
   - Use proper error handling for API calls, including catching HTTP status errors, logging unexpected exceptions, and response structure validation.
   - Structure output with clear section headers (e.g., environment variable setup, shared utilities, task-specific functions, etc.).

4. **Examples and Use Cases**:
   - Provide examples of utility functions like `make_post_request`, `make_get_request`, and task-specific functions (e.g., `upload_file_for_fine_tuning`).
   - Include at least one example of a function checking for valid user input, e.g., verifying a file path or ensuring the presence of required keys in API responses.

### Output Format

The output should be a Python script written in clear, modular code. Structure the code with logical sections:
1. Environment variable setup.
2. Shared utility functions (e.g., reusable HTTP request functions).
3. Task-specific functions (e.g., `upload_file_for_fine_tuning`, `create_fine_tuning_job`).
4. Input validation where necessary.

Ensure the Python code includes example functions for tasks like listing fine-tuning jobs, uploading files, and making chat completion requests. Add placeholder comments where user-specific logic or further integration may be required.

### Example Functions

#### 1. Environment Variable Setup
- **Purpose**: Securely handle sensitive information such as API keys and endpoints.
- **Example**:
  ```python
  import os

  def get_api_key():
      return os.getenv('API_KEY', 'your_default_api_key')

  def get_api_endpoint():
      return os.getenv('API_ENDPOINT', 'https://default.endpoint.com')
  ```

#### 2. Logging Setup
- **Purpose**: Set up logging to track the application's behavior and errors.
- **Example**:
  ```python
  import logging
  def setup_logging():
      logging.basicConfig(level=logging.DEBUG, format='%(asctime)s - %(levelname)s - %(message)s')
  ```

#### 3. Encrypt and Decrypt Sensitive Data
- **Purpose**: Securely encrypt and decrypt sensitive information.
- **Example**:
  ```python
  from cryptography.fernet import Fernet
  def encrypt_sensitive_data(data, key):
      fernet = Fernet(key)
      return fernet.encrypt(data.encode())

  def decrypt_sensitive_data(encrypted_data, key):
      fernet = Fernet(key)
      return fernet.decrypt(encrypted_data).decode()
  ```

#### 4. Reusable HTTP Request Functions
- **Purpose**: Create reusable functions for making HTTP requests.
- **Example**:
  ```python
  import httpx
  def make_get_request(url, headers):
      try:
          response = httpx.get(url, headers=headers)
          response.raise_for_status()
          return response.json()
      except httpx.HTTPStatusError as e:
          logging.error(f"HTTP error occurred: {e}")
      except Exception as e:
          logging.error(f"An error occurred: {e}")
  ```

#### 5. Task-Specific Functions
- **Purpose**: Implement functions for specific tasks, such as listing fine-tuning jobs.
- **Example**:
  ```python
  def list_fine_tuning_jobs():
      url = f"{get_api_endpoint()}/fine-tuning/jobs"
      headers = {'Authorization': f'Bearer {get_api_key()}'}
      return make_get_request(url, headers)
  ```

#### 6. Input Validation
- **Purpose**: Validate user inputs to ensure they meet the required criteria.
- **Example**:
  ```python
  def validate_file_path(file_path):
      if not os.path.exists(file_path):
          raise ValueError(f"File path {file_path} does not exist.")
  ```

### Additional Functions from `ultimatethinking.txt`

Include the following functions and concepts from the `ultimatethinking.txt` file:
- **Encrypt and Decrypt Sensitive Data**: Functions to encrypt and decrypt sensitive information using `cryptography.fernet`.
- **Logging Setup**: Function to set up logging based on configuration.
- **Load JSON Configuration**: Function to load configuration from a JSON file.
- **Universal Reasoning Aggregator**: Class to aggregate responses from multiple perspectives and handle context history, sentiment analysis, and real-time data fetching.
- **Element Class**: Class to define elements with properties and defense abilities.
- **Custom Recognizer**: Class to recognize keywords and intents from user input.
- **Voice and Image Input Handling**: Functions to handle voice and image inputs using `speech_recognition` and `PIL`.

### Example Usage

```python
if __name__ == "__main__":
    setup_logging()
    jobs = list_fine_tuning_jobs()
    if jobs:
        logging.info(f"Fine-tuning jobs: {jobs}")
```

---