swarmauri_ocr_pytesseract 0.11.0.dev1

Local Tesseract OCR adapter for Swarmauri with sync, async, and batch image-to-text workflows.

Swarmauri OCR Pytesseract

swarmauri_ocr_pytesseract is the Swarmauri OCR adapter for running local Tesseract-powered image-to-text extraction through a consistent Swarmauri OCRBase interface. It accepts file paths, raw bytes, or in-memory PIL images and supports synchronous, asynchronous, and batch OCR workflows.

Why Use Swarmauri OCR Pytesseract

• Use the same OCR component shape across local and pipeline-based Swarmauri workflows.
• Run OCR on local infrastructure without routing images through a hosted API.
• Tune Tesseract language selection and engine flags for receipts, forms, scanned PDFs, screenshots, and other document images.
• Reuse the same component in parsing, ingestion, indexing, and agent workflows.

FAQ

What does this package do?
It wraps PyTesseract and the local Tesseract binary behind Swarmauri's OCR component interface.

Does it require a hosted API key?
No. It runs locally, but the host must have the tesseract executable and any required language packs installed.

What image inputs are supported?
File path strings, raw image bytes, and PIL.Image.Image instances.

Can it process multiple images concurrently?
Yes. Use batch() for synchronous lists and abatch() for async execution with a concurrency limit.

Features

• Local OCR backed by PyTesseract and Tesseract OCR.
• Supports configurable language, config, and explicit tesseract_cmd resolution.
• Accepts paths, bytes, and PIL image objects.
• Includes extract_text, aextract_text, batch, and abatch methods.
• Can report installed OCR languages through get_supported_languages().
• Supports Python 3.10, 3.11, 3.12, 3.13, and 3.14.

Installation

uv add swarmauri_ocr_pytesseract

pip install swarmauri_ocr_pytesseract

System requirement:

• Install the native tesseract binary and ensure it is available on PATH, or set TESSERACT_CMD to the executable location.

Usage

from swarmauri_ocr_pytesseract import PytesseractOCR

ocr = PytesseractOCR(language="eng", config="--psm 6")
text = ocr.extract_text("docs/invoice.png")
print(text)

Examples

OCR from image bytes

from pathlib import Path
from swarmauri_ocr_pytesseract import PytesseractOCR

ocr = PytesseractOCR(language="eng")
image_bytes = Path("receipts/ticket.png").read_bytes()
text = ocr.extract_text(image_bytes)
print(text)

OCR from a PIL image

from PIL import Image
from swarmauri_ocr_pytesseract import PytesseractOCR

image = Image.open("scans/form.png")
ocr = PytesseractOCR(language="eng", config="--oem 3 --psm 4")
print(ocr.extract_text(image))

Async batch OCR

import asyncio
from swarmauri_ocr_pytesseract import PytesseractOCR

ocr = PytesseractOCR(language="eng")

async def run():
    results = await ocr.abatch(
        ["scans/page1.png", "scans/page2.png", "scans/page3.png"],
        max_concurrent=2,
    )
    for index, text in enumerate(results, start=1):
        print(index, text[:120])

asyncio.run(run())

List installed OCR languages

from swarmauri_ocr_pytesseract import PytesseractOCR

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

Related Packages

• swarmauri_parser_fitzpdf
• swarmauri_parser_pypdf2
• swarmauri_parser_pypdftk
• swarmauri_ocr_pytesseract

Swarmauri Foundations

• swarmauri
• swarmauri_core
• swarmauri_base
• swarmauri_standard

More Documentation

• Tesseract OCR documentation
• PyTesseract on PyPI
• Pillow documentation

Best Practices

• Use --psm and --oem options through config to match the page layout you expect.
• Install the correct .traineddata language packs for multilingual OCR.
• Pre-process noisy or skewed scans before OCR to improve extraction quality.
• Use PDF parsers when a PDF already contains embedded text; use OCR when the PDF or image is scan-only.

License

This project is licensed under the Apache-2.0 License.