LLM Chat Tool

This is a simple command-line tool for interacting with LLM APIs. It allows you to:

• Send text prompts and receive responses.
• Read text from files or URLs and send it to the LLM as a prompt.
• Manage your conversation history.
• Set custom prompts.
• Easily change the chunk size for processing large amounts of text.

Requirements

• Python 3.6 or later
• requests
• filetype
• pypdf
• beautifulsoup4
• prompt_toolkit
• dotenv

You can install the requirements using pip:

pip install requests filetype pypdf beautifulsoup4 prompt-toolkit dotenv

Usage

1. Set up your environment:

   • Create a .env file in the same directory as the script.
   • Add your LLM API credentials to the .env file. Refer to the specific LLM API documentation for the required environment variables.
   • Refer to the example.env file for a sample environment file.

2. Run the script:

   • Gemini

   python gemini.py

   • GPT

   python gpt.py

3. Interact with the LLM:

   • You can provide text prompts directly in the terminal.
   • You can also provide a file path or URL as an argument to the script.
   • Use the following commands to manage your conversation:
     • .q or .quit: Quit the chat tool.
     • .h or .history: Show the chat history.
     • .clear: Clear the chat history.
     • .info: Print information about the current model, chunk size, etc.
     • .g or .goto: Go to the beginning of the text.
     • .goto N: Go to position N in the text.
     • .chunk=N or .chunk N: Set the chunk size to N.
     • .prompt=PROMPT or .prompt PROMPT: Set the default prompt to PROMPT.
     • .o or .open: Open the URL being processed in a web browser.
   • Use the -a or --all flag to process the entire text as a single chunk.
   • Use the -p or --prompt flag to override the default prompt for external data.

Example

# Run the script with a text file
python gemini.py my_file.txt
# or
python gpt.py my_file.txt

# Run the script with a URL
python gemini.py https://www.example.com
# or
python gpt.py https://www.example.com

# Run the script with a direct prompt and override the default prompt
python gemini.py "What is the meaning of life?" -p "Please answer the question in a philosophical way."
# of
python gpt.py "What is the meaning of life?" -p "Please answer the question in a philosophical way."

# Run the script and set the chunk size to 5000
python gemini.py article.txt -c 5000
# or
python gemini.py article.txt -c 5000

Configuration

The following environment variables can be set in the .env file to customize the chat tool:

• DEFAULT_CHUNK_SIZE: The default chunk size for processing text.
• DEFAULT_PROMPT: The default prompt used for external data.
• PROMPT_HISTORY: The path to the file where the input history is stored.
• OUTPUT_HISTORY: The path to the file where the chat history is stored.
• SYSTEM_PROMPT: The system prompt used for the LLM.
• USER_AGENT: The User-Agent header used for HTTP requests.

Contributing

Contributions are welcome! Please open an issue or submit a pull request.

License

This project is licensed under the MIT License. See the LICENSE file for details.

---

Web Search Utility - google_search.py

This Python script provides a command-line interface for web searches using Google Custom Search Engine and a language model for summarizing web pages.

Features

• Uses Google Custom Search Engine API to perform web searches.
• Integrates with either Gemini or GPT models to summarize web pages.
• Provides a user-friendly menu for selecting search results.
• Allows users to navigate through multiple pages of search results.
• Supports user-agent and API key configuration via environment variables.

Requirements

• Python 3.6 or higher
• requests library
• urllib library
• prompt_toolkit library
• dotenv library
• Gemini or GPT model (depending on the chosen helper)

Installation

1. Install the required packages:

   pip install requests urllib3 prompt_toolkit dotenv

2. Install the Gemini or GPT model (depending on the chosen helper) by following their respective installation instructions.

Configuration

1. Create a .env file in the same directory as the script.
2. Set the following environment variables:
   • USER_AGENT: Your desired user agent string (optional).
   • GOOGLE_API_KEY: Your Google Custom Search Engine API key.
   • GOOGLE_CSE_ID: Your Google Custom Search Engine ID.
   • SEARCH_HELPER: The chosen language model (either "gemini" or "gpt").
   • GEMINI_MODEL: The path to your Gemini model (if using Gemini).
   • GPT_MODEL: The path to your GPT model (if using GPT).

Usage

1. Run the script from the command line:

   python google_search.py <query keywords>

   For example:

   python google_search.py "best restaurants in New York"

2. The script will display the search results and prompt you to select one.

3. After selecting a result, the script will use the chosen language model to summarize the web page and display the summary.

Example

Query: best restaurants in New York

Search Results
Please select one search result from the following:

1. Best Restaurants in NYC | 2023 Guide (Michelin, Zagat, More)
2. The 50 Best Restaurants in New York City - Thrillist
3. 100 Best Restaurants in NYC | OpenTable
4. Best Restaurants in New York City - The New York Times
5. Best Restaurants in New York City - Eater
6. Previous
7. Next

License

This project is licensed under the MIT License.