Text Extractor CLI

Overview

This project provides a command‑line tool that extracts structured data from free‑form text using LangChain and Pydantic. It supports two schemas:

1. PersonInfo – name, age, profession, skills.
2. MeetingNotes – title, date, participants, agenda.

The tool automatically detects which schema to use based on the input text.

Installation

# Create a virtual environment
python -m venv .venv
source .venv/bin/activate  # On Windows use .venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

Environment variables

The tool uses OpenAI’s API. Create a .env file in the project root with the following variables:

OPENAI_API_KEY=your_api_key_here
OPENAI_MODEL=gpt-3.5-turbo   # optional, defaults to gpt-3.5-turbo
OPENAI_BASE_URL=               # optional, for local LLMs (e.g. http://localhost:11434/v1)

If you are using a local model (Ollama/LM Studio) provide the OPENAI_BASE_URL and set OPENAI_API_KEY to any non‑empty string (e.g. ollama).

Usage

# Pass text as a command‑line argument
python agent.py "Anna, 28 years old, Python developer. Skills: FastAPI, Docker."

# Or pipe text via stdin
cat <<EOF | python agent.py
Meeting: Sprint Planning
Date: 2024-06-01
Participants: Alice, Bob, Charlie
Agenda: Review backlog, assign tasks, estimate effort
EOF

Output

The tool prints two sections:

1. JSON – a pretty‑printed JSON representation of the Pydantic model.
2. Summary – a human‑readable summary of the key fields.

Example output for a person:

{
  "name": "Anna",
  "age": 28,
  "profession": "Python developer",
  "skills": [
    "FastAPI",
    "Docker"
  ]
}

Summary:
Anna, 28 years old, works as Python developer.
Skills: FastAPI, Docker.

Extending

To add a new schema:

1. Define a new Pydantic model.
2. Create a prompt and parser similar to the existing ones.
3. Update the detect_schema logic or add a new classifier.

License

MIT