swarmauri_ocr_pytesseract 0.9.3.dev3

Swarmauri Tesseract Image to Text Model

---

Swarmauri OCR Pytesseract

OCR adapter for Swarmauri built on top of PyTesseract. Accepts paths, bytes, or PIL images, and exposes synchronous, async, and batch APIs for extracting text.

Features

• Wraps Tesseract OCR via PyTesseract behind Swarmauri's OCRBase interface.
• Supports multiple languages (language parameter) and custom Tesseract configs.
• Handles individual, async, and batched OCR calls with optional concurrency limits.
• Helper to list installed Tesseract languages through get_supported_languages().

Prerequisites

• Python 3.10 or newer.
• Tesseract OCR installed on the host (tesseract binary reachable on PATH or via TESSERACT_CMD).
• pytesseract, Pillow, and related dependencies (installed automatically with this package).

Installation

# pip
pip install swarmauri_ocr_pytesseract

# poetry
poetry add swarmauri_ocr_pytesseract

# uv (pyproject-based projects)
uv add swarmauri_ocr_pytesseract

Quickstart

from swarmauri_ocr_pytesseract import PytesseractOCR

ocr = PytesseractOCR(language="eng")
text = ocr.extract_text("docs/invoice.png")
print(text)

Processing Image Bytes

from pathlib import Path
from swarmauri_ocr_pytesseract import PytesseractOCR

png_bytes = Path("receipts/ticket.png").read_bytes()
ocr = PytesseractOCR(language="eng", config="--psm 6")
print(ocr.extract_text(png_bytes))

Async and Batch APIs

import asyncio
from swarmauri_ocr_pytesseract import PytesseractOCR

ocr = PytesseractOCR(language="fra")

async def run_async():
    text = await ocr.aextract_text("scans/document_fr.png")
    print(text)

    texts = await ocr.abatch([
        "scans/page1.png",
        "scans/page2.png",
    ], max_concurrent=2)
    for page, content in enumerate(texts, start=1):
        print(f"Page {page}: {content[:80]}")

# asyncio.run(run_async())

List Available Languages

from swarmauri_ocr_pytesseract import PytesseractOCR

ocr = PytesseractOCR()
print(ocr.get_supported_languages())

Tips

• Set TESSERACT_CMD if the binary lives outside standard locations (e.g., Windows installs).
• Use appropriate page segmentation modes (--psm) and OCR engine modes (--oem) through the config parameter to improve quality.
• Pre-process images (grayscale, thresholding) before passing them to the OCR for better accuracy.
• When running in containers, ensure Tesseract language packs (.traineddata) are installed for the languages you plan to use.

Want to help?

If you want to contribute to swarmauri-sdk, read up on our guidelines for contributing that will help you get started.