def validate_parameters(ocrd_tool, executable, param_json):
'''
Validate PARAM_JSON against parameter definition of EXECUTABLE in OCRD_TOOL
'''
with codecs.open(ocrd_tool, encoding='utf-8') as f:
ocrd_tool = loads(f.read())
_inform_of_result(
ParameterValidator(ocrd_tool['tools'][executable]).validate(
parse_json_string_or_file(param_json)))
def test_parameter_file_comments(self):
with TemporaryDirectory() as tempdir:
jsonpath = Path(tempdir, 'test.json')
jsonpath.write_text("""\
{
# Metasyntactical variables are rarely imaginative
"foo": 42,
# case in point:
"bar": 23
}""")
self.assertEqual(parse_json_string_or_file(str(jsonpath)), {'foo': 42, 'bar': 23})
def test_parse_json_string_or_file(self):
self.assertEqual(parse_json_string_or_file(), {})
self.assertEqual(parse_json_string_or_file(''), {})
self.assertEqual(parse_json_string_or_file(' '), {})
self.assertEqual(parse_json_string_or_file('{}'), {})
self.assertEqual(parse_json_string_or_file('{"foo": 32}'), {'foo': 32})
self.assertEqual(parse_json_string_or_file(
'{"dpi": -1, "textequiv_level": "word", "overwrite_words": false, "raw_lines": false, "char_whitelist": "", "char_blacklist": "", "char_unblacklist": ""}'
), {"dpi": -1, "textequiv_level": "word", "overwrite_words": False, "raw_lines": False, "char_whitelist": "", "char_blacklist": "", "char_unblacklist": ""})
def validate(self):
if not which(self.executable):
raise Exception("Executable not found in PATH: %s" %
self.executable)
if not self.input_file_grps:
raise Exception("Task must have input file group")
parameters = {}
if self.parameter_path:
parameters = parse_json_string_or_file(self.parameter_path)
param_validator = ParameterValidator(self.ocrd_tool_json)
report = param_validator.validate(parameters)
if not report.is_valid:
raise Exception(report.errors)
if 'output_file_grp' in self.ocrd_tool_json and not self.output_file_grps:
raise Exception(
"Processor requires output_file_grp but none was provided.")
return report
def validate(self):
if not which(self.executable):
raise Exception("Executable not found in PATH: %s" %
self.executable)
if not self.input_file_grps:
raise Exception("Task must have input file group")
result = run([self.executable, '--dump-json'],
stdout=PIPE,
check=True,
universal_newlines=True)
ocrd_tool_json = json.loads(result.stdout)
parameters = {}
if self.parameter_path:
parameters = parse_json_string_or_file(self.parameter_path)
param_validator = ParameterValidator(ocrd_tool_json)
report = param_validator.validate(parameters)
if not report.is_valid:
raise Exception(report.errors)
if 'output_file_grp' in ocrd_tool_json and not self.output_file_grps:
raise Exception(
"Processor requires output_file_grp but none was provided.")
return True
def test_parameters_invalid(self):
with self.assertRaisesRegex(ValueError, 'Not a valid JSON object'):
parse_json_string_or_file('[]')
with self.assertRaisesRegex(ValueError, 'Error parsing'):
parse_json_string_or_file('[}')
def test_parse_json_string_or_file(self):
self.assertEqual(parse_json_string_or_file(), {})
self.assertEqual(parse_json_string_or_file('{}'), {})
self.assertEqual(parse_json_string_or_file('{"foo": 32}'), {'foo': 32})
def _handle_param_option(ctx, param, value):
return parse_json_string_or_file(*list(value))
loglevel_option = click.option(
'-l',
'--log-level',
help="Log level",
type=click.Choice(['OFF', 'ERROR', 'WARN', 'INFO', 'DEBUG', 'TRACE']),
default=None,
callback=_set_root_logger_version)
parameter_option = click.option(
'-p',
'--parameter',
help="Parameters, either JSON string or path to JSON file",
default='{}',
callback=lambda ctx, param, value: parse_json_string_or_file(value))
def ocrd_cli_wrap_processor(processorClass,
ocrd_tool=None,
mets=None,
working_dir=None,
dump_json=False,
version=False,
**kwargs):
LOG = getLogger('ocrd_cli_wrap_processor')
if dump_json:
processorClass(workspace=None, dump_json=True)
elif version:
try:
p = processorClass(workspace=None)
