def test_should_call_get_pipeline_for_configuration_with_config_and_args(
self, get_pipeline_for_configuration_and_args, args):
config = dict_to_config(DEFAULT_CONFIG)
create_simple_pipeline_runner_from_config(config, args)
get_pipeline_for_configuration_and_args.assert_called_with(config,
args=args)
def _dict_to_cfg_content(d):
config = dict_to_config(d)
fp = StringIO()
config.write(fp)
content = fp.getvalue()
LOGGER.debug('content: %s', content)
return content
def test_should_default_pipeline_to_point_to_another_pipeline_configuration(
self, import_module):
config = dict_to_config({
u'pipelines': {
u'default': u'custom',
u'custom': DEFAULT_PIPELINE_MODULE
}
})
get_pipeline_for_configuration(config)
import_module.assert_called_with(DEFAULT_PIPELINE_MODULE)
def test_should_load_steps_and_run_them_on_convert(self, pipeline, args,
step):
config = dict_to_config(DEFAULT_CONFIG)
pipeline.get_steps.return_value = [step]
pipeline_runner = create_simple_pipeline_runner_from_config(
config, args)
step.get_supported_types.return_value = {MimeTypes.PDF}
assert (pipeline_runner.convert(
content=PDF_CONTENT,
filename=PDF_FILENAME,
data_type=MimeTypes.PDF) == step.return_value)
def test_should_skip_first_step_if_content_type_does_not_match(
self, pipeline, args, step):
config = dict_to_config(DEFAULT_CONFIG)
pipeline.get_steps.return_value = [
FunctionPipelineStep(lambda _: None, set(), 'dummy'), step
]
pipeline_runner = create_simple_pipeline_runner_from_config(
config, args)
step.get_supported_types.return_value = {MimeTypes.PDF}
assert (pipeline_runner.convert(
content=PDF_CONTENT,
filename=PDF_FILENAME,
data_type=MimeTypes.PDF) == step.return_value)
def test_should_pass_includes_to_first_step(self, pipeline, args, step):
config = dict_to_config(DEFAULT_CONFIG)
pipeline.get_steps.return_value = [step]
pipeline_runner = create_simple_pipeline_runner_from_config(config, args)
step.get_supported_types.return_value = {MimeTypes.PDF}
pipeline_runner.convert(
content=PDF_CONTENT, filename=PDF_FILENAME, data_type=MimeTypes.PDF,
includes={FieldNames.TITLE}
)
step.assert_called_with({
StepDataProps.FILENAME: PDF_FILENAME,
StepDataProps.CONTENT: PDF_CONTENT,
StepDataProps.TYPE: MimeTypes.PDF,
StepDataProps.INCLUDES: {FieldNames.TITLE}
})
def get_default_config():
return dict_to_config({})
def test_should_return_pipeline_steps(self, pipeline, args, step):
config = dict_to_config(DEFAULT_CONFIG)
pipeline.get_steps.return_value = [step]
pipeline = get_pipeline_for_configuration(config)
step.get_supported_types.return_value = {MimeTypes.PDF}
assert pipeline.get_steps(config, args) == [step]
def test_should_pass_args_and_config_to_get_steps(self, pipeline, args,
step):
config = dict_to_config(DEFAULT_CONFIG)
pipeline.get_steps.return_value = [step]
get_pipeline_for_configuration(config).get_steps(config, args)
pipeline.get_steps.assert_called_with(config, args)
def test_should_call_import_module_with_configured_default_pipeline(
self, import_module):
config = dict_to_config(DEFAULT_CONFIG)
get_pipeline_for_configuration(config)
import_module.assert_called_with(DEFAULT_PIPELINE_MODULE)
def test_should_return_concatenated_pipeline(self):
config = dict_to_config(DEFAULT_CONFIG)
assert get_pipeline_expression_for_configuration(
config, '%s, %s' %
(PIPELINE_1, PIPELINE_2)) == '%s, %s' % (PIPELINE_MODULE_1,
PIPELINE_MODULE_2)
def test_should_return_specified_pipeline(self):
config = dict_to_config(DEFAULT_CONFIG)
assert get_pipeline_expression_for_configuration(
config, PIPELINE_1) == PIPELINE_MODULE_1
def test_should_return_default_pipeline(self):
config = dict_to_config(DEFAULT_CONFIG)
assert get_pipeline_expression_for_configuration(
config, '') == DEFAULT_PIPELINE_MODULE
def _config():
return dict_to_config({})
def test_should_pass_args_and_config_to_get_steps(self, pipeline, args,
step):
config = dict_to_config(DEFAULT_CONFIG)
pipeline.get_steps.return_value = [step]
create_simple_pipeline_runner_from_config(config, args)
pipeline.get_steps.assert_called_with(config, args)
