def test_en_simple(self):
# Trigger reference-counting bug that was fixed in PyICU 1.0.1:
# https://github.com/ovalhub/pyicu/commit/515e076682e29d806aeb5f6b1016b799d03d92a9
icu = get_icu()
assert_not_equal(icu, None)
t = list(word_break_iterator('eggs', icu.Locale('en')))
assert_equal(t, [4])
def test_en_simple(self):
# Trigger reference-counting bug that was fixed in PyICU 1.0.1:
# https://github.com/ovalhub/pyicu/commit/515e076682e29d806aeb5f6b1016b799d03d92a9
icu = get_icu()
assert_not_equal(icu, None)
t = list(word_break_iterator('eggs', icu.Locale('en')))
assert_equal(t, [4])
def test_en(self):
icu = get_icu()
assert_not_equal(icu, None)
t = list(word_break_iterator(text, icu.Locale('en')))
s = [1, 7, 8, 9, 10, 15, 16, 24, 25, 26, 30, 31, 32, 33, 37, 38, 43, 44, 45, 46, 47, 48, 54, 55, 57, 58, 61, 62, 65, 66, 67]
assert_equal(t, s)
assert_equal(s[-1], len(text))
def test_en(self):
icu = get_icu()
assert_not_equal(icu, None)
t = list(word_break_iterator(text, icu.Locale('en')))
s = [1, 7, 8, 9, 10, 15, 16, 24, 25, 26, 30, 31, 32, 33, 37, 38, 43, 44, 45, 46, 47, 48, 54, 55, 57, 58, 61, 62, 65, 66, 67]
assert_equal(t, s)
assert_equal(s[-1], len(text))
def test_help():
stdout = io.BytesIO()
stderr = io.BytesIO()
with interim(sys, stdout=stdout, stderr=stderr):
rc = try_run(hocr2djvused.main, ['', '--help'])
assert_equal(stderr.getvalue(), '')
assert_equal(rc, 0)
assert_not_equal(stdout.getvalue(), '')
def test_bad_options():
stdout = io.BytesIO()
stderr = io.BytesIO()
with interim(sys, stdout=stdout, stderr=stderr):
rc = try_run(hocr2djvused.main, ['', '--bad-option'])
assert_equal(rc, errors.EXIT_FATAL)
assert_not_equal(stderr.getvalue(), '')
assert_equal(stdout.getvalue(), '')
def _test_list_languages(engine):
stdout = io.BytesIO()
stderr = io.BytesIO()
with interim(sys, stdout=stdout, stderr=stderr):
rc = try_run(ocrodjvu.main, ['', '--engine', engine, '--list-languages'])
assert_equal(stderr.getvalue(), '')
assert_equal(rc, 0)
assert_not_equal(stdout.getvalue(), '')
def test_help():
stdout = io.BytesIO()
stderr = io.BytesIO()
with interim(sys, stdout=stdout, stderr=stderr):
rc = try_run(djvu2hocr.main, ['', '--help'])
assert_equal(rc, 0)
assert_equal(stderr.getvalue(), '')
assert_not_equal(stdout.getvalue(), '')
def test_bad_options():
stdout = io.BytesIO()
stderr = io.BytesIO()
with interim(sys, stdout=stdout, stderr=stderr):
rc = try_run(djvu2hocr.main, [''])
assert_equal(rc, errors.EXIT_FATAL)
assert_not_equal(stderr.getvalue(), '')
assert_equal(stdout.getvalue(), '')
def _test_list_languages(engine):
stdout = io.BytesIO()
stderr = io.BytesIO()
with interim(sys, stdout=stdout, stderr=stderr):
rc = try_run(ocrodjvu.main,
['', '--engine', engine, '--list-languages'])
assert_equal(rc, 0)
assert_equal(stderr.getvalue(), '')
assert_not_equal(stdout.getvalue(), '')
def test_version():
# https://bugs.debian.org/573496
stdout = io.BytesIO()
stderr = io.BytesIO()
with interim(sys, stdout=stdout, stderr=stderr):
rc = try_run(hocr2djvused.main, ['', '--version'])
assert_equal(stderr.getvalue(), '')
assert_equal(rc, 0)
assert_not_equal(stdout.getvalue(), '')
def test_version():
# https://bugs.debian.org/573496
stdout = io.BytesIO()
stderr = io.BytesIO()
with interim(sys, stdout=stdout, stderr=stderr):
rc = try_run(djvu2hocr.main, ['', '--version'])
assert_equal(rc, 0)
assert_equal(stderr.getvalue(), '')
assert_not_equal(stdout.getvalue(), '')
def _rough_test_from_file(base_filename, args):
args = ['#'] + shlex.split(args)
if base_filename.endswith(('cuneiform0.7', 'cuneiform0.8')):
# Add dummy page-size information
args += ['--page-size=1000x1000']
base_filename = os.path.join(here, base_filename)
html_filename = '{base}.html'.format(base=base_filename)
with contextlib.closing(io.BytesIO()) as output_file:
with open(html_filename, 'rb') as html_file:
with interim(sys, stdin=html_file, stdout=output_file):
rc = try_run(hocr2djvused.main, args)
assert_equal(rc, 0)
output = output_file.getvalue()
assert_not_equal(output, '')
def test_nonascii_path():
require_locale_encoding('UTF-8') # djvused breaks otherwise
remove_logging_handlers('ocrodjvu.')
here = os.path.dirname(__file__)
here = os.path.abspath(here)
path = os.path.join(here, '..', 'data', 'empty.djvu')
stdout = io.BytesIO()
stderr = io.BytesIO()
with temporary.directory() as tmpdir:
tmp_path = os.path.join(tmpdir, 'тмп.djvu')
os.symlink(path, tmp_path)
with interim(sys, stdout=stdout, stderr=stderr):
rc = try_run(djvu2hocr.main, ['', tmp_path])
assert_equal(stderr.getvalue(), '')
assert_equal(rc, 0)
assert_not_equal(stdout.getvalue(), '')
def test_nonascii_path():
require_locale_encoding('UTF-8') # djvused breaks otherwise
remove_logging_handlers('ocrodjvu.')
here = os.path.dirname(__file__)
here = os.path.abspath(here)
path = os.path.join(here, '..', 'data', 'empty.djvu')
stdout = io.BytesIO()
stderr = io.BytesIO()
with temporary.directory() as tmpdir:
tmp_path = os.path.join(tmpdir, 'тмп.djvu')
os.symlink(path, tmp_path)
with interim(sys, stdout=stdout, stderr=stderr):
rc = try_run(djvu2hocr.main, ['', tmp_path])
assert_equal(stderr.getvalue(), '')
assert_equal(rc, 0)
assert_not_equal(stdout.getvalue(), '')
def test_en_empty(self):
icu = get_icu()
assert_not_equal(icu, None)
t = list(word_break_iterator('', icu.Locale('en')))
assert_equal(t, [])
def test_en_empty(self):
icu = get_icu()
assert_not_equal(icu, None)
t = list(word_break_iterator('', icu.Locale('en')))
assert_equal(t, [])
