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_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_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_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_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_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_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_list_engines():
global engines
stdout = io.BytesIO()
stderr = io.BytesIO()
with interim(sys, stdout=stdout, stderr=stderr):
rc = try_run(ocrodjvu.main, ['', '--list-engines'])
assert_equal(rc, 0)
assert_equal(stderr.getvalue(), '')
engines = stdout.getvalue().splitlines()
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 test_list_engines():
global engines
stdout = io.BytesIO()
stderr = io.BytesIO()
with interim(sys, stdout=stdout, stderr=stderr):
rc = try_run(ocrodjvu.main, ['', '--list-engines'])
assert_equal(stderr.getvalue(), '')
assert_equal(rc, 0)
engines = stdout.getvalue().splitlines()
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_ocr(engine, layers):
if not distutils.spawn.find_executable(engine):
raise SkipTest('{cmd} not found'.format(cmd=engine))
remove_logging_handlers('ocrodjvu.')
here = os.path.dirname(__file__)
here = os.path.abspath(here)
path = os.path.join(here, '..', 'data', 'alice.djvu')
stdout = io.BytesIO()
stderr = io.BytesIO()
with temporary.directory() as tmpdir:
tmp_path = os.path.join(tmpdir, 'tmp.djvu')
with interim(sys, stdout=stdout, stderr=stderr):
rc = try_run(ocrodjvu.main, ['', '--engine', engine, '--render', layers, '--save-bundled', tmp_path, path])
assert_multi_line_equal(stderr.getvalue(), '')
assert_equal(rc, 0)
assert_multi_line_equal(stdout.getvalue(), '')
def _test_from_file(base_filename, index, extra_args):
base_filename = os.path.join(here, base_filename)
test_filename = '{base}.test{i}'.format(base=base_filename, i=index)
html_filename = '{base}.html'.format(base=base_filename)
with open(test_filename, 'rb') as file:
commandline = file.readline()
expected_output = file.read()
args = shlex.split(commandline) + shlex.split(extra_args)
assert_equal(args[0], '#')
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_multi_line_equal(normalize_djvused(expected_output),
normalize_djvused(output))
def test_bad_page_id():
remove_logging_handlers('ocrodjvu.')
here = os.path.dirname(__file__)
here = os.path.abspath(here)
path = os.path.join(here, '..', 'data', 'bad-page-id.djvu')
stdout = io.BytesIO()
stderr = io.BytesIO()
with temporary.directory() as tmpdir:
out_path = os.path.join(tmpdir, 'tmp.djvu')
with interim(sys, stdout=stdout, stderr=stderr):
with interim(ocrodjvu, system_encoding='ASCII'):
rc = try_run(ocrodjvu.main, [
'', '--engine', '_dummy', '--save-bundled', out_path, path
])
assert_equal(stderr.getvalue(), '')
assert_equal(rc, 0)
assert_equal(stdout.getvalue(), '')
def _test_ocr(engine, layers):
if not distutils.spawn.find_executable(engine):
raise SkipTest('{cmd} not found'.format(cmd=engine))
remove_logging_handlers('ocrodjvu.')
here = os.path.dirname(__file__)
here = os.path.abspath(here)
path = os.path.join(here, '..', 'data', 'alice.djvu')
stdout = io.BytesIO()
stderr = io.BytesIO()
with temporary.directory() as tmpdir:
tmp_path = os.path.join(tmpdir, 'tmp.djvu')
with interim(sys, stdout=stdout, stderr=stderr):
rc = try_run(ocrodjvu.main, [
'', '--engine', engine, '--render', layers, '--save-bundled',
tmp_path, path
])
assert_multi_line_equal(stderr.getvalue(), '')
assert_equal(rc, 0)
assert_multi_line_equal(stdout.getvalue(), '')
def _test_from_file(base_filename, index, extra_args):
base_filename = os.path.join(here, base_filename)
test_filename = '{base}.test{i}'.format(base=base_filename, i=index)
html_filename = '{base}.html'.format(base=base_filename)
with open(test_filename, 'rb') as file:
commandline = file.readline()
expected_output = file.read()
args = shlex.split(commandline) + shlex.split(extra_args)
assert_equal(args[0], '#')
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_multi_line_equal(
normalize_djvused(expected_output),
normalize_djvused(output)
)
def _test_from_file(base_filename, index):
base_filename = os.path.join(here, base_filename)
test_filename = '{base}.test{i}'.format(base=base_filename, i=index)
djvused_filename = base_filename + '.djvused'
with open(test_filename, 'rb') as file:
commandline = file.readline()
expected_output = file.read()
args = shlex.split(commandline)
assert_equal(args[0], '#')
with temporary.directory() as tmpdir:
djvu_filename = os.path.join(tmpdir, 'empty.djvu')
args += [djvu_filename]
shutil.copy(
os.path.join(os.path.dirname(__file__), '..', 'data',
'empty.djvu'), djvu_filename)
ipc.Subprocess(
['djvused', '-f', djvused_filename, '-s', djvu_filename]).wait()
xml_filename = os.path.join(tmpdir, 'output.html')
with open(xml_filename, 'w+b') as xml_file:
xmllint = ipc.Subprocess(['xmllint', '--format', '-'],
stdin=ipc.PIPE,
stdout=xml_file)
try:
with open(os.devnull, 'w') as null:
with interim(sys, stdout=xmllint.stdin, stderr=null):
with interim(djvu2hocr.logger, handlers=[]):
rc = try_run(djvu2hocr.main, args)
finally:
xmllint.stdin.close()
try:
xmllint.wait()
except ipc.CalledProcessError:
# Raising the exception here is likely to hide the real
# reason of the failure.
pass
assert_equal(rc, 0)
xml_file.seek(0)
output = xml_file.read()
assert_multi_line_equal(expected_output, output)
def _test_from_file(base_filename, index):
base_filename = os.path.join(here, base_filename)
test_filename = '{base}.test{i}'.format(base=base_filename, i=index)
djvused_filename = base_filename + '.djvused'
with open(test_filename, 'rb') as file:
commandline = file.readline()
expected_output = file.read()
args = shlex.split(commandline)
assert_equal(args[0], '#')
with temporary.directory() as tmpdir:
djvu_filename = os.path.join(tmpdir, 'empty.djvu')
args += [djvu_filename]
shutil.copy(
os.path.join(os.path.dirname(__file__), '..', 'data', 'empty.djvu'),
djvu_filename)
ipc.Subprocess(['djvused', '-f', djvused_filename, '-s', djvu_filename]).wait()
xml_filename = os.path.join(tmpdir, 'output.html')
with open(xml_filename, 'w+b') as xml_file:
xmllint = ipc.Subprocess(['xmllint', '--format', '-'], stdin=ipc.PIPE, stdout=xml_file)
try:
with open(os.devnull, 'w') as null:
with interim(sys, stdout=xmllint.stdin, stderr=null):
with interim(djvu2hocr.logger, handlers=[]):
rc = try_run(djvu2hocr.main, args)
finally:
xmllint.stdin.close()
try:
xmllint.wait()
except ipc.CalledProcessError:
# Raising the exception here is likely to hide the real
# reason of the failure.
pass
assert_equal(rc, 0)
xml_file.seek(0)
output = xml_file.read()
assert_multi_line_equal(expected_output, output)
