def ocrd_cli_wrap_processor(processorClass,
ocrd_tool=None,
mets=None,
working_dir=None,
dump_json=False,
version=False,
**kwargs):
if dump_json:
processorClass(workspace=None, dump_json=True)
elif version:
p = processorClass(workspace=None)
print("Version %s, ocrd/core %s" % (p.version, OCRD_VERSION))
elif mets is None:
raise Exception('Error: Missing option "-m" / "--mets".')
else:
if mets.find('://') == -1:
mets = 'file://' + os.path.abspath(mets)
if mets.startswith('file://') and not os.path.exists(
mets[len('file://'):]):
raise Exception("File does not exist: %s" % mets)
resolver = Resolver()
workspace = resolver.workspace_from_url(mets, working_dir)
run_processor(processorClass,
ocrd_tool,
mets,
workspace=workspace,
**kwargs)
def test_run_agent(self):
no_agents_before = len(self.workspace.mets.agents)
run_processor(DummyProcessor,
ocrd_tool=DUMMY_TOOL,
workspace=self.workspace)
self.assertEqual(len(self.workspace.mets.agents), no_agents_before + 1,
'one more agent')
def test_copies_ok(self):
with copy_of_directory(assets.url_of('SBB0000F29300010000/data')) as wsdir:
workspace = Workspace(Resolver(), wsdir)
input_files = workspace.mets.find_files(fileGrp='OCR-D-IMG')
self.assertEqual(len(input_files), 3)
output_files = workspace.mets.find_files(fileGrp='OUTPUT')
self.assertEqual(len(output_files), 0)
run_processor(
DummyProcessor,
input_file_grp='OCR-D-IMG',
output_file_grp='OUTPUT',
workspace=workspace
)
output_files = workspace.mets.find_files(fileGrp='OUTPUT')
output_files.sort(key=lambda x: x.url)
print([str(s) for s in output_files])
self.assertEqual(output_files[0].url, 'OUTPUT/OUTPUT_0001.tif')
self.assertEqual(output_files[1].url, 'OUTPUT/OUTPUT_0001.xml')
self.assertEqual(page_from_file(output_files[1]).pcGtsId, output_files[1].ID)
self.assertEqual(page_from_file(output_files[1]).get_Page().imageFilename, output_files[0].url)
self.assertEqual(len(output_files), 6)
self.assertEqual(len(workspace.mets.find_files(ID='//OUTPUT.*')), 6)
self.assertEqual(len(workspace.mets.find_files(ID='//OUTPUT.*_PAGE')), 3)
self.assertEqual(len(workspace.mets.find_files(fileGrp='OUTPUT', mimetype=MIMETYPE_PAGE)), 3)
run_processor(
DummyProcessor,
input_file_grp='OUTPUT',
output_file_grp='OUTPUT2',
workspace=workspace
)
output2_files = workspace.mets.find_files(fileGrp='OUTPUT2')
output2_files.sort(key=lambda x: x.url)
self.assertEqual(len(output2_files), 3)
def test_crop(self):
with copy_of_directory(assets.path_to('dfki-testdata/data')) as wsdir:
ws = Workspace(self.resolver, wsdir)
pagexml_before = len(ws.mets.find_files(mimetype=MIMETYPE_PAGE))
run_processor(
OcrdAnybaseocrCropper,
resolver=self.resolver,
mets_url=str(Path(wsdir, 'mets.xml')),
input_file_grp='BIN',
output_file_grp='CROP-TEST',
parameter={},
)
ws.reload_mets()
pagexml_after = len(ws.mets.find_files(mimetype=MIMETYPE_PAGE))
self.assertEqual(pagexml_after, pagexml_before + 1)
def test_crop(self):
if not torch.cuda.is_available():
pytest.skip('CUDA is not available, cannot test dewarping')
with copy_of_directory(assets.path_to('dfki-testdata/data')) as wsdir:
ws = Workspace(self.resolver, wsdir)
pagexml_before = len(ws.mets.find_files(mimetype=MIMETYPE_PAGE))
run_processor(OcrdAnybaseocrDewarper,
resolver=self.resolver,
mets_url=str(Path(wsdir, 'mets.xml')),
input_file_grp='BIN',
output_file_grp='DEWARP-TEST',
parameter={'model_path': str(self.model_path)})
ws.reload_mets()
pagexml_after = len(ws.mets.find_files(mimetype=MIMETYPE_PAGE))
self.assertEqual(pagexml_after, pagexml_before + 1)
def test_with_mets_url_input_files(self):
processor = run_processor(
DummyProcessor,
resolver=self.resolver,
mets_url=assets.url_of('SBB0000F29300010000/data/mets.xml'))
self.assertEqual(len(processor.input_files), 20)
self.assertTrue(
all([f.mimetype == MIMETYPE_PAGE for f in processor.input_files]))
def test_no_input_file_grp(self):
processor = run_processor(
DummyProcessor,
resolver=self.resolver,
mets_url=assets.url_of('SBB0000F29300010000/data/mets.xml'))
with self.assertRaisesRegex(Exception,
'Processor is missing input fileGrp'):
_ = processor.input_files
def ocrd_cli_wrap_processor(processorClass,
ocrd_tool=None,
mets=None,
working_dir=None,
cache_enabled=True,
*args,
**kwargs):
if mets.find('://') == -1:
mets = 'file://' + mets
if mets.startswith('file://') and not os.path.exists(
mets[len('file://'):]):
raise Exception("File does not exist: %s" % mets)
resolver = Resolver(cache_enabled=cache_enabled)
workspace = resolver.workspace_from_url(mets, working_dir)
run_processor(processorClass,
ocrd_tool,
mets,
workspace=workspace,
*args,
**kwargs)
def test_run_cli(self):
with TemporaryDirectory() as tempdir:
run_processor(DummyProcessor,
ocrd_tool=DUMMY_TOOL,
workspace=self.workspace)
run_cli(
'echo',
mets_url=assets.url_of('SBB0000F29300010000/data/mets.xml'),
resolver=Resolver(),
workspace=None,
page_id='page1',
log_level='DEBUG',
input_file_grp='INPUT',
output_file_grp='OUTPUT',
parameter='/path/to/param.json',
working_dir=tempdir)
run_cli(
'echo',
mets_url=assets.url_of('SBB0000F29300010000/data/mets.xml'),
resolver=Resolver(),
)
def test_parameter_url(self):
with TemporaryDirectory() as tempdir:
jsonpath = join(tempdir, 'params.json')
with open(jsonpath, 'w') as f:
f.write('{}')
processor = run_processor(
DummyProcessor,
parameter='file://%s' % jsonpath,
resolver=self.resolver,
mets_url=assets.url_of('SBB0000F29300010000/data/mets.xml')
)
self.assertEqual(len(processor.input_files), 35)
def test_parameter(self):
with TemporaryDirectory() as tempdir:
jsonpath = join(tempdir, 'params.json')
with open(jsonpath, 'w') as f:
f.write('{"baz": "quux"}')
with open(jsonpath, 'r') as f:
processor = run_processor(
DummyProcessor,
parameter=json.load(f),
resolver=self.resolver,
mets_url=assets.url_of(
'SBB0000F29300010000/data/mets.xml'))
self.assertEqual(len(processor.input_files), 20)
def test_no_mets_url(self):
with self.assertRaisesRegex(Exception,
'pass mets_url to create a workspace'):
run_processor(DummyProcessor, resolver=self.resolver)
def test_no_resolver(self):
with self.assertRaisesRegex(Exception,
'pass a resolver to create a workspace'):
run_processor(DummyProcessor)
def test_with_mets_url_input_files(self):
processor = run_processor(
DummyProcessor,
resolver=self.resolver,
mets_url=assets.url_of('SBB0000F29300010000/data/mets.xml'))
self.assertEqual(len(processor.input_files), 35)
