Turn any document into clean, AI-ready Markdown. Local-first desktop app: reads scanned PDFs, batches folders, runs offline, and uses far fewer tokens than vision models.
Python desktop-appdocument-conversionllmlocal-firstmarkdownmarkitdownocrofflinepdf-to-markdownragrustsveltetauri
from __future__ import annotations
import importlib.util
import os
import sys
import tempfile
import unittest
from pathlib import Path
SIDECAR_DIR = Path(__file__).resolve().parents[1]
if str(SIDECAR_DIR) not in sys.path:
sys.path.insert(0, str(SIDECAR_DIR))
OCR_AVAILABLE = importlib.util.find_spec("rapidocr") is not None
def setUpModule() -> None:
if os.environ.get("MDFLUX_REQUIRE_OCR_TESTS") == "1" and not OCR_AVAILABLE:
raise AssertionError("Locked OCR integration tests require rapidocr")
@unittest.skipUnless(OCR_AVAILABLE, "rapidocr is not installed in the core test environment")
class LockedOcrRuntimeTests(unittest.TestCase):
def test_image_and_scanned_pdf_ocr(self) -> None:
from PIL import Image, ImageDraw, ImageFont
import ocr
with tempfile.TemporaryDirectory(prefix="mdflux-ocr-test-") as tmp:
root = Path(tmp)
image_path = root / "ocr-test.png"
pdf_path = root / "ocr-test-scanned.pdf"
image = Image.new("RGB", (1400, 360), "white")
font = ImageFont.load_default(size=72)
ImageDraw.Draw(image).text(
(60, 120),
"MDFlux OCR integration test 2026",
fill="black",
font=font,
stroke_width=1,
)
image.save(image_path)
image.save(pdf_path, "PDF", resolution=150.0)
image_text = ocr.ocr_image(str(image_path), intra_op_threads=1)
self.assertTrue(image_text.strip(), "Image OCR returned no text")
self.assertTrue(ocr.is_scanned_pdf(str(pdf_path)))
pdf_text = ocr.ocr_pdf(str(pdf_path), intra_op_threads=1)
self.assertTrue(pdf_text.strip(), "Scanned-PDF OCR returned no text")
self.assertIn("## Page 1", pdf_text)
if __name__ == "__main__":
unittest.main()