rostaing-ocr

Rostaing OCR, created by Davila Rostaing.

A Python OCR package for extracting text from scanned and native images or PDFs, designed for integration into AI applications such as LLMs and RAG, and to produce clean plain text (.txt) and Markdown (.md) files.

Features

• Converts images (PNG, JPG, etc.) and PDFs into text.
• Includes image preprocessing (grayscale conversion) to improve OCR accuracy.
• Processes multiple files in a single run, consolidating the output.
• Generates output in both plain text (.txt) and Markdown (.md) formats.
• Optional flag to print extracted content directly to the console for quick inspection.

System Prerequisites: Tesseract OCR

Important: This package requires the Tesseract OCR engine to be installed on your system. rostaing-ocr is a Python wrapper that calls the tesseract command-line tool. You must install it and its language packs first.

Windows Installation Guide (Recommended)

• Download the Installer: Go to the official Tesseract at UB Mannheim repository. They provide the most up-to-date and reliable installers for Windows.
• Run the Installer: Start the installation process.
• Crucial Step - Add to PATH: On the "Select Additional Tasks" screen, make sure to check the box for "Add Tesseract to system PATH". This is essential for Python to be able to find and execute Tesseract.
• Select Languages: On the "Select additional language data" screen, expand the list and select the languages you will need for OCR (e.g., check French for fra, English is usually included by default).
• Complete Installation: Finish the installation and, to be safe, restart your command prompt, terminal, or IDE to ensure the system's PATH variable is updated.

macOS (via Homebrew)

brew install tesseract

You can add language packs by installing tesseract-lang.

Linux (Debian/Ubuntu)

sudo apt update
sudo apt install tesseract-ocr

# Also install the language packs you need. For French:
sudo apt install tesseract-ocr-fra

Installation

Best Practice: Use a Virtual Environment

To keep project dependencies isolated and avoid conflicts with other Python projects on your system, it is highly recommended to use a virtual environment.

With a standard Python installation, you can create and activate a new environment using the following commands:

On macOS/Linux:

# Create an environment named '.venv' in your project directory
python3 -m venv .venv

# Activate the environment
source .venv/bin/activate

On Windows:

# Create an environment named '.venv' in your project directory
python -m venv .venv

# Activate the environment
.venv\Scripts\activate

Install the Package

Once Tesseract is set up and your virtual environment is activated, you can install the package from PyPI:

pip install rostaing-ocr

Usage

Here is a basic example of how to use the RostaingOCR class.

from rostaing_ocr import RostaingOCR

# --- Example 1: Process a single file ---
# This will create 'my_result.txt' and 'my_result.md' in the current directory.
extractor = RostaingOCR(
    input_path_or_paths="path/to/my_document.pdf",
    output_basename="my_result", # Optionel
    print_to_console=True # Optionel
)

# You can print to get a summary of the operation.
print(extractor)

# --- Example 2: Process multiple files and print to console ---
# This will process both files, save a consolidated output, and also print the results.
multi_extractor = RostaingOCR(
    input_path_or_paths=["document1.png", "scan_page_2.pdf"],
    output_basename="combined_report", # Optionel
    print_to_console=True, # Optionel
    languages=['fra', 'eng'] # Specify languages for Tesseract # Optionel
)

# You can print the object to get a summary of the operation.
print(multi_extractor)

Application for LLM and RAG Pipelines

Large Language Models (LLMs) like GPT-4 or Llama understand text, not images or scanned documents. A vast amount of valuable knowledge is locked away in unstructured formats such as PDFs of research papers, scanned invoices, or legal contracts.

Rostaing OCR serves as the crucial first step in any data ingestion pipeline for Retrieval-Augmented Generation (RAG) systems. It bridges the gap by converting this inaccessible visual data into clean, structured text that LLMs can process.

By using Rostaing OCR, you can automate the process of building a knowledge base from your documents:

• Input: A directory of Scanned PDFs or Images.
• Extraction (Rostaing OCR): Convert all documents into clean Markdown/Text.
• Processing: The text output can be fed into text splitters and then embedding models.
• Indexing: The resulting vectors are stored in a vector database (e.g., Chroma, Pinecone, FAISS) for efficient retrieval.

In short, Rostaing OCR unlocks your documents, making them ready for any modern AI stack.

License

This project is licensed under the MIT License. See the LICENSE file for more details.

Useful Links

• Github: https://github.com/Rostaing/rostaing-ocr
• Pypi: https://pypi.org/project/rostaing-ocr/
• Linkedin: https://www.linkedin.com/in/davila-rostaing/
• YouTube: youtube.com/@RostaingAI?sub_confirmation=1