SummarizeAPI

A FastAPI-based API service for summarizing text and documents using large language models (LLMs).

Supports running models locally via Ollama or remotely via providers like OpenAI.

Ideal if you want to:

• Summarize arbitrary text
• Summarize uploaded documents (e.g. PDFs)
• Switch easily between local and cloud LLMs

---

✨ Features

• Summarize plain text
• Summarize uploaded PDF documents
• Use local Ollama LLMs for private, offline inference
• Or switch to cloud providers (e.g. OpenAI)
• JSON API endpoints
• Built with FastAPI

---

🏗️ Project Structure

summarizeapi/
├── api/
│   └── v1/
│       ├── routes.py
├── core/
│   ├── llm_factory.py
│   └── summarizer.py
├── models/
│   └── summary_schema.py
├── services/
│   └── pdf_reader.py
├── utils/
│   └── config.py
└── main.py

• main.py – entry point for FastAPI
• routes.py – defines API endpoints
• llm_factory.py – handles switching between Ollama / OpenAI / other providers
• pdf_reader.py – extracts text from PDF uploads
• summary_schema.py – request/response models

---

🚀 How it Works

1. Send plain text

   • The API summarizes the text using the configured LLM.

2. Upload a document (PDF)

   • Text is extracted from the PDF.
   • The LLM summarizes the extracted text (or responds to your custom query).

---

🔧 Installation

Clone the repo:

git clone https://github.com/MIDHUNGRAJ/summarizeapi.git
cd summarizeapi

Create a virtual environment and install requirements:

conda create -n ml-engine python=3.11
conda activate ml-engine

pip install -r requirements.txt

---

🔗 Running the API

Start the FastAPI server:

uvicorn summarizeapi.main:app --reload

Then visit:

http://127.0.0.1:8000/docs

to explore the interactive Swagger UI.

---

📝 Example API Call

Summarize Text

curl -X POST "http://127.0.0.1:8000/api/v1/summarize" \
    -H  "accept: application/json" \
    -H  "Content-Type: multipart/form-data" \
    -F "text=Summarize this text quickly"

Summarize PDF

curl -X POST "http://127.0.0.1:8000/api/v1/summarize" \
    -H  "accept: application/json" \
    -H  "Content-Type: multipart/form-data" \
    -F "text=Summarize the document content" \
    -F "file=@yourfile.pdf;type=application/pdf"

---

✅ Todo

• Add support for other document formats (Word, txt, etc.)
• Add support for more LLM providers (Anthropic, Mistral, etc.)
• Implement better error handling
• Add unit tests

---

License

MIT