GeminiRequests

GeminiRequests is a Python library (and code example) designed for interacting with Google’s Gemini AI text generation model using the requests library.
This code provides the Gen class, which allows you to store dialogue history, send requests to the API, load/save conversation history, and more.
Note: The last message in the dialogue history must always be from the user when generating text.

Features

1. Easy integration with Gemini AI:
   Uses the standard requests library to send HTTP requests to Google’s Gemini AI API.

2. Dialogue history management:
   The Gen class keeps track of conversation history (a list of messages with roles and text), making it highly useful for context-dependent tasks.
   Important: Ensure the last message in the history is from the user before calling generate().

3. Convenient export and import methods:

   • export_history(filename) / import_history(filename)
   • import_history_anyway(filename) (attempts to load history and creates a file if it doesn’t exist)

4. API key management:
   On the first run, it automatically creates a gemini_api_key file (in the GeminiRequests folder) where you can save your API key.

---

Installation and Setup

Installing from PyPI

pip install GeminiRequests

GitHub

1. Clone the repository (or copy this code into your project):

   git clone https://github.com/Python-shik/GeminiRequests.git

2. Install dependencies:

   pip install -r requirements.txt

   If you’re only using this module, you can simply install:

   pip install requests

Set up the API key:

• On the first run, the script will ask you to enter your Gemini AI API key.
• The key will be saved in the GeminiRequests/gemini_api_key file.

---

Usage

Assume the file with the Gen class is stored in the GeminiRequests folder. Example:

from GeminiRequests import Gen  # Assuming the file is named gen.py

# Create an instance of the class. 
# On the first run, you’ll be prompted to enter your API key if the gemini_api_key file is missing.
ai = Gen()

# Add some messages to the dialogue history
ai.history_add(role="user", content="Hello, Gemini!")
ai.history_add(role="model", content="hey.")

# Ensure the last message in the history is from the user
ai.history_add(role="user", content="What else can you do?")

# Generate a response
response = ai.generate()
print("Response from Gemini AI:", response)

# Save the history
ai.export_history("dialog_history.pkl")

# Clear local history (from memory)
ai.history = []

# Load history again if needed
ai.import_history("dialog_history.pkl")

The Gen Class and Its Methods

• Gen(history=[], system_instructions=None)
  Constructor. On the first run, it creates a gemini_api_key file if it doesn’t exist.

  • history: a list of dictionaries like {"role": ..., "parts": [{"text": ...}]}, describing the current conversation history.
  • system_instructions: a list of instructions sent as part of the request (essentially "prompts" or "context" from the system).

• history_add(role, content)
  Adds a new message (role and text) to the dialogue history.

• generate()
  Sends the dialogue history (and system instructions, if any) to the API and retrieves the generated text.
  Important: Ensure the last message in the history is from the user before calling this method.

• export_history(filename) / import_history(filename)
  Save/load the dialogue history to/from a file using pickle.

• import_history_anyway(filename)
  Attempts to load the history from a file. If the file doesn’t exist, it creates one and retries.

• clear_history(filename)
  Deletes the history file and clears the history from memory.

---

License

This project is distributed under the MIT License.
You are free to modify and use the code in your projects, including commercial ones.

---

Author

Developed by Illya Lazarev
Email: sbdt.israel@gmail.com

If you have ideas or suggestions for improvements, feel free to submit pull requests or open an Issue.