from PIL import Image
import pytest
import img2pdf
import pikepdf
from ocrmypdf import leptonica
from ocrmypdf.pdfinfo import PdfInfo
from ocrmypdf.exec import ghostscript
from ocrmypdf.helpers import fspath
# pytest.helpers is dynamic
# pylint: disable=no-member
# pylint: disable=w0612
pytestmark = pytest.mark.skipif(
leptonica.get_leptonica_version() < 'leptonica-1.72',
reason="Leptonica is too old, correlation doesn't work")
check_ocrmypdf = pytest.helpers.check_ocrmypdf
run_ocrmypdf = pytest.helpers.run_ocrmypdf
RENDERERS = ['hocr', 'sandwich']
def check_monochrome_correlation(outdir, reference_pdf, reference_pageno,
test_pdf, test_pageno):
gslog = logging.getLogger()
reference_png = outdir / '{}.ref{:04d}.png'.format(reference_pdf.name,
reference_pageno)
test_png = outdir / '{}.test{:04d}.png'.format(test_pdf.name, test_pageno)
def test_pil_conversion(crom_pix):
pix, im = crom_pix
# Check for pixel perfect
assert ImageChops.difference(pix.topil(), im).getbbox() is None
def test_pix_otsu(crom_pix):
pix, _ = crom_pix
im1bpp = pix.otsu_adaptive_threshold()
assert im1bpp.mode == '1'
@pytest.mark.skipif(
lept.get_leptonica_version() < 'leptonica-1.76',
reason="needs new leptonica for API change",
)
def test_crop(resources):
pix = lept.Pix.open(resources / 'linn.png')
foreground = pix.crop_to_foreground()
assert foreground.width < pix.width
def test_clean_bg(resources):
pix = lept.Pix.open(resources / 'congress.jpg')
imbg = pix.clean_background_to_white()
def test_pickle(crom_pix):
pix, _ = crom_pix
def test_pil_conversion(crom_pix):
pix, im = crom_pix
# Check for pixel perfect
assert ImageChops.difference(pix.topil(), im).getbbox() is None
def test_pix_otsu(crom_pix):
pix, _ = crom_pix
im1bpp = pix.otsu_adaptive_threshold()
assert im1bpp.mode == '1'
@pytest.mark.skipif(
lp.get_leptonica_version() < 'leptonica-1.76',
reason="needs new leptonica for API change",
)
def test_crop(resources):
pix = lp.Pix.open(resources / 'linn.png')
foreground = pix.crop_to_foreground()
assert foreground.width < pix.width
def test_clean_bg(resources):
pix = lp.Pix.open(resources / 'congress.jpg')
imbg = pix.clean_background_to_white()
def test_pickle(crom_pix):
pix, _ = crom_pix
import img2pdf
import pytest
from PIL import Image
import pikepdf
from ocrmypdf import leptonica
from ocrmypdf.exec import ghostscript, tesseract
from ocrmypdf.pdfinfo import PdfInfo
# pytest.helpers is dynamic
# pylint: disable=no-member
# pylint: disable=w0612
pytestmark = pytest.mark.skipif(
leptonica.get_leptonica_version() < 'leptonica-1.72',
reason="Leptonica is too old, correlation doesn't work",
)
check_ocrmypdf = pytest.helpers.check_ocrmypdf
run_ocrmypdf = pytest.helpers.run_ocrmypdf
RENDERERS = ['hocr', 'sandwich']
def check_monochrome_correlation(
outdir, reference_pdf, reference_pageno, test_pdf, test_pageno
):
gslog = logging.getLogger()