示例#1
0
def test_deskew(spoof_tesseract_noop, resources, outdir):
    # Run with deskew
    deskewed_pdf = check_ocrmypdf(resources / 'skew.pdf',
                                  outdir / 'skew.pdf',
                                  '-d',
                                  env=spoof_tesseract_noop)

    # Now render as an image again and use Leptonica to find the skew angle
    # to confirm that it was deskewed
    log = logging.getLogger()

    deskewed_png = outdir / 'deskewed.png'

    ghostscript.rasterize_pdf(deskewed_pdf,
                              deskewed_png,
                              xres=150,
                              yres=150,
                              raster_device='pngmono',
                              log=log,
                              pageno=1)

    from ocrmypdf.leptonica import Pix
    pix = Pix.read(str(deskewed_png))
    skew_angle, skew_confidence = pix.find_skew()

    print(skew_angle)
    assert -0.5 < skew_angle < 0.5, "Deskewing failed"
示例#2
0
def test_deskew(spoof_tesseract_noop):
    # Run with deskew
    deskewed_pdf = check_ocrmypdf(
        'skew.pdf', 'test_deskew.pdf', '-d', '-v', '1', env=spoof_tesseract_noop)

    # Now render as an image again and use Leptonica to find the skew angle
    # to confirm that it was deskewed
    from ocrmypdf.ghostscript import rasterize_pdf
    import logging
    log = logging.getLogger()

    deskewed_png = _outfile('deskewed.png')

    rasterize_pdf(
        deskewed_pdf,
        deskewed_png,
        xres=150,
        yres=150,
        raster_device='pngmono',
        log=log)

    from ocrmypdf.leptonica import Pix
    pix = Pix.read(deskewed_png)
    skew_angle, skew_confidence = pix.find_skew()

    print(skew_angle)
    assert -0.5 < skew_angle < 0.5, "Deskewing failed"
示例#3
0
def test_deskew(resources, outdir):
    # Run with deskew
    deskewed_pdf = check_ocrmypdf(
        resources / 'skew.pdf',
        outdir / 'skew.pdf',
        '-d',
        '--plugin',
        'tests/plugins/tesseract_noop.py',
    )

    # Now render as an image again and use Leptonica to find the skew angle
    # to confirm that it was deskewed
    deskewed_png = outdir / 'deskewed.png'

    ghostscript.rasterize_pdf(
        deskewed_pdf,
        deskewed_png,
        raster_device='pngmono',
        raster_dpi=Resolution(150, 150),
        pageno=1,
    )

    pix = Pix.open(deskewed_png)
    skew_angle, _skew_confidence = pix.find_skew()

    print(skew_angle)
    assert -0.5 < skew_angle < 0.5, "Deskewing failed"
示例#4
0
def test_deskew(spoof_tesseract_noop):
    # Run with deskew
    deskewed_pdf = check_ocrmypdf('skew.pdf',
                                  'test_deskew.pdf',
                                  '-d',
                                  '-v',
                                  '1',
                                  env=spoof_tesseract_noop)

    # Now render as an image again and use Leptonica to find the skew angle
    # to confirm that it was deskewed
    from ocrmypdf.ghostscript import rasterize_pdf
    import logging
    log = logging.getLogger()

    deskewed_png = _outfile('deskewed.png')

    rasterize_pdf(deskewed_pdf,
                  deskewed_png,
                  xres=150,
                  yres=150,
                  raster_device='pngmono',
                  log=log)

    from ocrmypdf.leptonica import Pix
    pix = Pix.read(deskewed_png)
    skew_angle, skew_confidence = pix.find_skew()

    print(skew_angle)
    assert -0.5 < skew_angle < 0.5, "Deskewing failed"
示例#5
0
def test_deskew(spoof_tesseract_noop, resources, outdir):
    # Run with deskew
    deskewed_pdf = check_ocrmypdf(
        resources / 'skew.pdf', outdir / 'skew.pdf', '-d',
        env=spoof_tesseract_noop)

    # Now render as an image again and use Leptonica to find the skew angle
    # to confirm that it was deskewed
    log = logging.getLogger()

    deskewed_png = outdir / 'deskewed.png'

    ghostscript.rasterize_pdf(
        deskewed_pdf,
        deskewed_png,
        xres=150,
        yres=150,
        raster_device='pngmono',
        log=log,
        pageno=1)

    from ocrmypdf.leptonica import Pix
    pix = Pix.read(str(deskewed_png))
    skew_angle, skew_confidence = pix.find_skew()

    print(skew_angle)
    assert -0.5 < skew_angle < 0.5, "Deskewing failed"