campaign-generator 0.1.5

A FastAPI service for campaign generation, competitor research, and audio content.

Campaign Generator

Campaign Generator is a Python FastAPI application for creating content based on competitor research and user-generated audio. It provides a web interface for generating questions, summarizing text, transcribing audio, and fetching the latest news. The app uses local LLMs via Ollama and supports both Mac and Linux environments.

Features

• Audio Transcription: Upload or record audio and transcribe it using Whisper (macOS: whisper-mps, Linux: faster-whisper).
• Text Summarization: Summarize any text using local LLMs (Gemma 3 1B, GPT-OSS 20B) via Ollama.
• Question Generation: Generate questions from text using LLMs.
• Latest News: Fetch recent news on any topic.
• Modern Web UI: Beautiful interface built with Tailwind CSS.
• OS-aware: Automatically selects the best transcription backend and Ollama API endpoint for your platform.

Prerequisites

• Python 3.10+
• Homebrew (for macOS)
• pipx
• ffmpeg (for audio conversion)
• Ollama installed and running with required models (gemma3:1b, gpt-oss:20b)

Install system dependencies (macOS)

brew install pipx
brew install ffmpeg
pipx ensurepath
source ~/.zshrc  # or restart your terminal

Install with pipx

pipx install campaign-generator

Usage

Set your ClickUp API key and team ID in your environment variables:

export CLICKUP_API_KEY=your_clickup_api_key
export CLICKUP_TEAM_ID=your_clickup_team_id

env | grep CLICKUP # to verify they are set

When you start the app for the first time, the following folders will be created in the directory you run the command from:

• research output/: Stores research results generated from RSS feeds.
• question output/: Stores generated questions.
• audio output/: Stores uploaded audio files.
• transcripts output/: Stores transcriptions of audio files.
• content generation/: Stores generated content.

Because of this behavior, it's recommended to run the app from a dedicated project directory.

Start the app with:

campaign-generator

By default, the API will run on http://localhost:8080.

Open your browser and navigate to /frontend to use the web interface:

http://localhost:8080/frontend
http://localhost:8080/frontend/chat # for RAG chat interface
http://localhost:8080/docs # for API docs
http://localhost:8080/graph-viewer # for visualization of the RAG graph

How It Works

• Transcription: Uses whisper-mps on macOS and faster-whisper on Linux for audio transcription.
• LLM Integration: Connects to Ollama API for text summarization and question generation. The API endpoint is chosen automatically based on your OS.
• Templates: The web UI is served from templates/home.html.

Development

• All source code is in the root and routers/ directory.
• The entrypoint is main.py, which runs the FastAPI app.
• You can also run with Docker or Docker Compose (see Dockerfile and docker-compose.yml).

Build and Publish to PyPI

1. Upgrade build and twine:

   python -m pip install --upgrade build twine
   

2. Build your package:

   python -m build
   

3. (Optional) Check your package:

   python -m twine check dist/*
   

4. Upload to PyPI:

   python -m twine upload dist/*
   

After publishing, you (and others) can install globally with pipx:

pipx install campaign-generator

Troubleshooting

• If you see errors about missing templates, ensure you installed with pipx after running pipx ensurepath and that your terminal session is up to date.
• Make sure Ollama is running and the required models are pulled.
• If you see errors about ffmpeg, make sure it is installed and available in your PATH.

License

MIT