def t(*dummy):
image_path = os.path.join(os.path.dirname(__file__), 'data',
'nonexistent.png')
meta = xmp.metadata(backend=backend)
meta.import_(image_path)
with temporary.directory() as tmpdir:
os.chmod(tmpdir, 0o000)
try:
image_path = os.path.join(tmpdir, 'example.png')
meta = xmp.metadata(backend=backend)
with assert_raises(IOError):
meta.import_(image_path)
finally:
os.chmod(tmpdir, 0o700)
def test_incl(self):
path = os.path.join(datadir, 'onebit.bmp')
in_image = pil.open(path)
[width, height] = in_image.size
sjbz_path = os.path.join(datadir, 'onebit.djvu')
incl_path = os.path.join(datadir, 'shared_anno.iff')
multichunk = djvu.Multichunk(width, height, 100, sjbz=sjbz_path, incl=incl_path)
djvu_file = multichunk.save()
with temporary.directory() as tmpdir:
tmp_djvu_path = os.path.join(tmpdir, 'index.djvu')
tmp_incl_path = os.path.join(tmpdir, 'shared_anno.iff')
os.link(djvu_file.name, tmp_djvu_path)
shutil.copyfile(incl_path, tmp_incl_path)
out_image = ddjvu(tmp_djvu_path, fmt='pbm')
assert_images_equal(in_image, out_image)
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_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_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_path(self):
path = os.getenv('PATH').split(':')
with temporary.directory() as tmpdir:
command_name = temporary.name(dir=tmpdir)
command_path = os.path.join(tmpdir, command_name)
with open(command_path, 'wt') as file:
print('#!/bin/sh', file=file)
print('printf 42', file=file)
os.chmod(command_path, stat.S_IRWXU)
path[:0] = [tmpdir]
path = ':'.join(path)
with interim_environ(PATH=path):
child = ipc.Subprocess([command_name],
stdout=ipc.PIPE, stderr=ipc.PIPE,
)
stdout, stderr = child.communicate()
assert_equal(stdout, '42')
assert_equal(stderr, '')
def test_incl(self):
path = os.path.join(datadir, 'onebit.bmp')
in_image = pil.open(path)
[width, height] = in_image.size
sjbz_path = os.path.join(datadir, 'onebit.djvu')
incl_path = os.path.join(datadir, 'shared_anno.iff')
multichunk = djvu.Multichunk(width,
height,
100,
sjbz=sjbz_path,
incl=incl_path)
djvu_file = multichunk.save()
with temporary.directory() as tmpdir:
tmp_djvu_path = os.path.join(tmpdir, 'index.djvu')
tmp_incl_path = os.path.join(tmpdir, 'shared_anno.iff')
os.link(djvu_file.name, tmp_djvu_path)
shutil.copyfile(incl_path, tmp_incl_path)
out_image = ddjvu(tmp_djvu_path, fmt='pbm')
assert_images_equal(in_image, out_image)
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):
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)
