Quickstart

This guide walks you through running RAG PDF Highlighter, sending a highlight request, and receiving a highlighted PDF back. Choose the path that fits your use case: start the FastAPI microservice and call it over HTTP, or import the core function directly into your Python code.

As a microservice
As a Python library

Install

Install the package from PyPI:

pip install rag-pdf-highlighter

Start the server

Start the Uvicorn server, binding to all interfaces on port 8000:

uvicorn rag_pdf_highlighter.main:app --host 0.0.0.0 --port 8000

The server is ready when you see Uvicorn reporting Application startup complete. You can verify it is running by visiting http://localhost:8000/ — it returns {"status": "ok the app is running"}.

Send a highlight request

Send a POST /highlight request with a PDF URL and the text chunks you want highlighted. The response body is the annotated PDF file.

curl -X POST http://localhost:8000/highlight \
  -H "Content-Type: application/json" \
  -d '{
    "pdf_url": "https://example.com/report.pdf",
    "documents": [
      {
        "page_content": "Text to highlight in the PDF",
        "metadata": {"page": 0}
      }
    ]
  }' \
  --output highlighted.pdf

The file highlighted.pdf in your current directory is the annotated copy of the original PDF.

Install

Install the package from PyPI:

pip install rag-pdf-highlighter

Import and call highlight_chunks_in_pdf

Import highlight_chunks_in_pdf from the utilities module and pass it a local PDF path with a list of Document objects:

from langchain_core.documents import Document
from rag_pdf_highlighter.utils.pdf_helpers import highlight_chunks_in_pdf

documents = [
    Document(page_content="Text to find", metadata={"page": 0}),
]

output_path = highlight_chunks_in_pdf(
    pdf_path="./report.pdf",
    documents=documents,
)
print(f"Highlighted PDF saved to: {output_path}")

Open the output file

The return value of highlight_chunks_in_pdf is a string containing the path to a new temporary file with all annotations applied. Open that path with any PDF viewer to inspect the highlights.

The output file is written to the system’s temp directory. Copy or move it to a permanent location before your process exits if you need to keep it.

The metadata.page field is zero-indexed. Page 1 of the PDF is {"page": 0}, page 2 is {"page": 1}, and so on. Passing an out-of-range page number causes that chunk to be silently skipped.

Next steps

API reference

Full documentation for the POST /highlight endpoint, request fields, and error responses.

Guides

Deploy with Docker, integrate with LangChain retrieval chains, and handle large PDFs.

Get Started

Guides

Concepts

Next steps

API reference

Guides

Build docs developers (and LLMs) love

Get Started

Guides

Concepts

Documentation Index

​Next steps

API reference

Guides

Build docs developers (and LLMs) love

Next steps