def test_tesseract_presence(self, capsys):
p = P.PyTesseract({})
p.binary = "tesserac" # Misspell it and make sure we get an error
with pytest.raises(SystemExit):
p._is_version_uptodate()
out, err = capsys.readouterr()
assert p.msgs['TS_MISSING'] in out
def test_tesseract_version(self, capsys):
p = P.PyTesseract({})
p.required = "100"
with pytest.raises(SystemExit):
p.make_hocr_from_pnms("")
out, err = capsys.readouterr()
assert p.msgs['TS_VERSION'] in out
def test_version_major_equal(self):
with patch("subprocess.check_output") as mock_subprocess:
p = P.PyTesseract({})
p.required = "3.02.02"
mock_subprocess.return_value = """tesseract 3.02.02"""
uptodate, ver = p._is_version_uptodate()
assert (uptodate)
def test_tesseract_version_nt(self, monkeypatch):
"""
Stupid test because Windows Tesseract only returns 3.02 instead
of 3.02.02
"""
monkeypatch.setattr('os.name', 'nt')
monkeypatch.setattr('subprocess.check_output',
mock.Mock(return_value="tesseract 3.02"))
pyts = pypdfocr_tesseract.PyTesseract({})
pyts.assert_version()
def test_tesseract_version_nt(self, mock_subprocess, mock_os_name):
"""
Stupid test because Windows Tesseract only returns 3.02 instead of 3.02.02
"""
mock_os_name.__str__.return_value = 'nt'
p = P.PyTesseract({})
p.required = "3.02.02"
mock_subprocess.return_value = """tesseract 3.02"""
uptodate, ver = p._is_version_uptodate()
assert (uptodate)
def test_force_Nt(self, monkeypatch, tmpdir):
monkeypatch.setattr('os.name', 'nt')
monkeypatch.setattr('os.path.exists', mock.Mock(return_value=True))
pyts = pypdfocr_tesseract.PyTesseract({})
pyts._ts_version = "4.01"
assert 'tesseract.exe' in pyts.binary
# force a bad tesseract on windows
pyts.binary = "blah"
with pytest.raises(SystemExit):
pyts.make_hocr_from_pnm(str(tmpdir.join('blah.tiff')))
def test_tesseract_old_output(self, version, ext, monkeypatch):
"""Test that correct extension is used based on tesseract version
Old versions of tesseract (before 3.03) created .html files, whereas
more recent versions create .hocr files.
"""
monkeypatch.setattr('os.path.exists', mock.Mock(return_value=True))
monkeypatch.setattr('os.path.isfile', mock.Mock(return_value=True))
monkeypatch.setattr('subprocess.call', mock.Mock(return_value=0))
pyts = pypdfocr_tesseract.PyTesseract({})
pyts._ts_version = version
assert pyts.make_hocr_from_pnm('foo.tiff') == 'foo.{}'.format(ext)
def test_force_Nt(self, mock_os_path_exists, mock_os_name, mock_uptodate,
capsys):
mock_os_name.__str__.return_value = 'nt'
p = P.PyTesseract({})
assert ('tesseract.exe' in p.binary)
mock_os_path_exists.return_value = True
mock_uptodate.return_value = (True, "")
# force a bad tesseract on windows
p.binary = "blah"
print("here")
with pytest.raises(SystemExit):
p.make_hocr_from_pnm('blah.tiff')
def test_tesseract_fail(self, caplog, monkeypatch, tmpdir):
"""
Get all the checks passed and make sure we report the case where
tesseract returns a non-zero status.
"""
monkeypatch.setattr('os.name', 'nt')
monkeypatch.setattr('os.path.exists', mock.Mock(return_value=True))
monkeypatch.setattr('subprocess.check_output', mock.Mock(side_effect=subprocess.CalledProcessError(-1 , 'Boom')))
pyts = pypdfocr_tesseract.PyTesseract({})
pyts._ts_version = "4.01"
assert 'tesseract.exe' in pyts.binary
with pytest.raises(SystemExit):
pyts.make_hocr_from_pnm(str(tmpdir.join('blah.tiff')))
assert pyts.msgs['TS_FAILED'] in caplog.text
def test_tesseract_fail(self, mock_os_path_exists, mock_os_name,
mock_uptodate, mock_subprocess_call, capsys):
"""
Get all the checks past and make sure we report the case where tesseract returns a non-zero status
"""
mock_os_name.__str__.return_value = 'nt'
p = P.PyTesseract({})
assert ('tesseract.exe' in p.binary)
mock_os_path_exists.return_value = True
mock_uptodate.return_value = (True, "")
mock_subprocess_call.return_value = -1
with pytest.raises(SystemExit):
p.make_hocr_from_pnm('blah.tiff')
out, err = capsys.readouterr()
assert p.msgs['TS_FAILED'] in out
def test_tiff_file_check(self, capsys):
p = P.PyTesseract({})
with pytest.raises(SystemExit):
p.make_hocr_from_pnm("DUMMY_NOTPRESENT.tiff")
out, err = capsys.readouterr()
assert p.msgs['TS_img_MISSING'] in out
def test_override_binary_nt(self, monkeypatch):
monkeypatch.setattr("os.name", "nt")
pyts = pypdfocr_tesseract.PyTesseract({'binary': '\\foo\\bar\\bin'})
assert pyts.binary == '"\\\\foo\\\\bar\\\\bin"'
def test_override_binary(self):
pyts = pypdfocr_tesseract.PyTesseract({'binary': '/foo/bar/bin'})
assert '/foo/bar/bin' in pyts.binary
def pyts(self):
return pypdfocr_tesseract.PyTesseract({})
def test_tesseract_4alpha(self, monkeypatch, tmpdir):
monkeypatch.setattr('subprocess.check_output',
mock.Mock(return_value="tesseract 4.00.00alpha"))
pyts = pypdfocr_tesseract.PyTesseract({})
pyts.assert_version()