POP-guotai 0.2.2

Reusable, mutable, prompt functions for LLMs.

Prompt Oriented Programming (POP)

Reusable, mutable prompt functions for LLMs. Inspired by object-oriented programming (OOP), POP treats prompts as first-class citizens—defining them like functions with inputs, outputs, and reusable logic. Execution is delegated to a language model (LLM), such as OpenAI’s GPT, enabling you to call, modify, and reuse prompts in a structured, programmatic way.

---

Table of Contents

1. Features
2. Installation
3. Setup & Configuration
4. Usage
   • PromptFunction Class
   • Improving Prompts
5. Example
6. Future Plans
7. Contributing

---

Features

• Prompt as a Function: Declare a prompt once, then execute it any number of times with different inputs.
• Dynamic Placeholders: Insert variables via placeholders (<<<variable>>>), allowing flexible, user-defined parameters.
• System Role Defaults: POP stores the main prompt in the “system” role by default, letting you easily switch between roles or contexts.
• Prompt Enhancement: Leverages patterns from Fabric (MIT licensed) to expand basic prompts into more detailed instructions via pf._improve_prompt().

---

Installation

1. Clone or Download this repository (or add it as a submodule in your project).
2. Install via pip (recommended):

   pip install -e .
   

   This uses the setup.py (and optionally pyproject.toml) so that pop can be globally recognized as a Python package.

For more details on packaging and distribution, see the Installation Steps in the conversation.

---

Setup & Configuration

Environment Variables

Place a .env file in the same directory as POP.py or Embedder.py, containing API keys if you use OpenAI or other APIs. For example:

OPENAI_API_KEY=your_openai_api_key
JINAAI_API_KEY=your_jina_api_key  # Required only if you use Jina features

Note: POP uses OpenAI by default. Make sure your OPENAI_API_KEY is valid, or configure a different LLM in your code.

---

Usage

PromptFunction Class

from pop import PromptFunction

# Create a prompt function
pf = PromptFunction(
    sys_prompt="You are a helpful AI assistant.",
    prompt="""
    Please draw a simple ASCII art of <<<object>>>.
    """
)

# Execute with user input
art = pf.execute(object="a cat")
print(art)

• Placeholders: <<<object>>> is replaced by your chosen argument (object="a cat").
• Multiple Calls: You can call pf.execute(...) repeatedly with different inputs.

Improving Prompts

expanded_prompt = pf._improve_prompt()
print(expanded_prompt)

• _improve_prompt(): Expands or refines your base prompt with more detailed instructions (inspired by Fabric’s approach).

---

Example

from pop import PromptFunction

# Define a prompt function
pf = PromptFunction(
    prompt="""
    Draw a simple ASCII art of an object described by: <<<user_input>>>
    """
)

# First Execution
result = pf.execute(user_input="A cat")
print("ASCII art of a cat:\n", result)

# Second Execution
result = pf.execute(user_input="An apple")
print("ASCII art of an apple:\n", result)

Sample Output:

ASCII art of a cat:
 /\_/\  
( o.o )
 > ^ <  

ASCII art of an apple:
    ,--./,-.
   / #      \
  |          |
   \        /
    `._,._,'

---

Future Plans

• Additional LLM Integrations: Support local models or alternative APIs beyond OpenAI and Jina.
• Enhanced Prompt Composition: Build chains of PromptFunctions for complex workflows.
• Prompt Testing Framework: Automate checks to validate prompt outputs across multiple inputs.

---

Contributing

Contributions are welcome! Feel free to open issues or submit pull requests:

1. Fork the repo
2. Create a feature branch for your changes
3. Open a PR once ready

Please include clear commit messages and relevant tests or examples.

---

Enjoy Prompt Oriented Programming! If you have questions or suggestions, don’t hesitate to reach out or file an issue.