def test_validate_sequence(self):
resolver = Resolver()
with TemporaryDirectory() as tempdir:
workspace = resolver.workspace_from_url(
assets.path_to('kant_aufklaerung_1784/data/mets.xml'),
dst_dir=tempdir)
params_path = Path(tempdir, 'params.json')
params_path.write_text('{"param1": true}')
with self.assertRaisesRegex(
Exception,
"Input file group not contained in METS or produced by previous steps: FOO'"
):
validate_tasks([
ProcessorTask.parse(x) for x in [
'%s -I OCR-D-IMG -O OUT1 -p %s' %
(SAMPLE_NAME_REQUIRED_PARAM, params_path),
'%s -I FOO -O OUT2 -p %s' %
(SAMPLE_NAME_REQUIRED_PARAM, params_path)
]
], workspace)
with self.assertRaisesRegex(
Exception, "Input fileGrp.@USE='IN'. not in METS!"):
validate_tasks([
ProcessorTask.parse(x) for x in [
'%s -I IN -O OUT1 -p %s' %
(SAMPLE_NAME_REQUIRED_PARAM, params_path),
]
], workspace)
def test_overwrite(self):
resolver = Resolver()
with TemporaryDirectory() as tempdir:
workspace = resolver.workspace_from_url(
assets.path_to('kant_aufklaerung_1784/data/mets.xml'),
dst_dir=tempdir)
# should fail at step 3
workspace.mets.add_file('OCR-D-SEG-WORD',
url='foo/bar',
ID='foo',
pageId='page1',
mimetype='image/tif')
with self.assertRaisesRegex(
Exception,
r"Invalid task sequence input/output file groups: \[\"Output fileGrp\[@USE='OCR-D-SEG-WORD'\] already in METS!\"\]"
):
validate_tasks([
ProcessorTask.parse(x) for x in [
"sample-processor -I OCR-D-IMG -O OCR-D-SEG-BLOCK",
"sample-processor -I OCR-D-SEG-BLOCK -O OCR-D-SEG-LINE",
"sample-processor -I OCR-D-SEG-LINE -O OCR-D-SEG-WORD",
"sample-processor -I OCR-D-SEG-WORD -O OCR-D-OCR-TESS",
]
], workspace)
# should succeed b/c overwrite
validate_tasks([
ProcessorTask.parse(x) for x in [
"sample-processor -I OCR-D-IMG -O OCR-D-SEG-BLOCK",
"sample-processor -I OCR-D-SEG-BLOCK -O OCR-D-SEG-LINE",
"sample-processor -I OCR-D-SEG-LINE -O OCR-D-SEG-WORD",
"sample-processor -I OCR-D-SEG-WORD -O OCR-D-OCR-TESS",
]
],
workspace,
overwrite=True)
def test_422(self):
"""
# OCR-D/core#422
"""
resolver = Resolver()
with TemporaryDirectory() as tempdir:
workspace = resolver.workspace_from_url(
assets.path_to('kant_aufklaerung_1784/data/mets.xml'),
dst_dir=tempdir)
validate_tasks([
ProcessorTask.parse(x) for x in [
"sample-processor -I OCR-D-IMG -O OCR-D-SEG-BLOCK",
"sample-processor -I OCR-D-SEG-BLOCK -O OCR-D-SEG-LINE",
"sample-processor -I OCR-D-SEG-LINE -O OCR-D-SEG-WORD",
"sample-processor -I OCR-D-SEG-WORD -O OCR-D-OCR-TESS",
]
], workspace)
def validate_process(tasks, workspace):
'''
Validate a sequence of tasks passable to 'ocrd process'
'''
if workspace:
_inform_of_result(
validate_tasks([ProcessorTask.parse(t) for t in tasks],
Workspace(Resolver(), directory=workspace)))
else:
for t in [ProcessorTask.parse(t) for t in tasks]:
_inform_of_result(t.validate())