def run(self, registry: Registry, random: Generator, language: str,
document_plan: DocumentPlanNode) -> Tuple[DocumentPlanNode]:
if log.isEnabledFor(logging.DEBUG):
document_plan.print_tree()
log.debug("Aggregating")
self._aggregate(registry, language, document_plan)
if log.isEnabledFor(logging.DEBUG):
document_plan.print_tree()
return (document_plan, )
def run(self, registry: Registry, random: Generator, language: str,
document_plan: DocumentPlanNode) -> Tuple[DocumentPlanNode]:
"""
Run this pipeline component.
"""
log.info("Realizing dates")
self._recurse(registry, random, language, document_plan, None)
if log.isEnabledFor(logging.DEBUG):
document_plan.print_tree()
return (document_plan, )
def run(self, registry: Registry, random: Generator, language: str,
document_plan: DocumentPlanNode) -> Tuple[DocumentPlanNode]:
"""
Run this pipeline component.
"""
log.info("Running NER")
if language.endswith("-head"):
language = language[:-5]
log.debug(
"Language had suffix '-head', removing. Result: {}".format(
language))
previous_entities = defaultdict(lambda: None)
self._recurse(registry, random, language, document_plan,
previous_entities, set())
if log.isEnabledFor(logging.DEBUG):
document_plan.print_tree()
return (document_plan, )
def run(
self,
registry: Registry,
random: Generator,
language: str,
document_plan: DocumentPlanNode,
all_messages: List[Message],
) -> Tuple[DocumentPlanNode]:
"""
Run this pipeline component.
"""
if log.isEnabledFor(logging.DEBUG):
document_plan.print_tree()
templates = registry.get("templates")[language]
template_checker = TemplateMessageChecker(templates, all_messages)
log.info("Selecting templates from {} templates".format(len(templates)))
self._recurse(random, language, document_plan, all_messages, template_checker)
return (document_plan,)
class TestDocumentPlanNode(TestCase):
def setUp(self):
self.fact1 = Fact(
"corpus1",
"corpus_type",
"timestamp_from",
"timestamp_to",
"timestamp_type",
"analysis_type",
"result_key",
"result_value",
"outlierness",
)
self.message1 = Message(self.fact1, 0.1, 0.2, 0.3)
self.fact2 = Fact(
"corpus2",
"corpus_type",
"timestamp_from",
"timestamp_to",
"timestamp_type",
"analysis_type",
"result_key",
"result_value",
"outlierness",
)
self.message2 = Message(self.fact2, 0.1, 0.2, 0.3)
self.document_plan_node = DocumentPlanNode(
[self.message1, self.message2], Relation.ELABORATION)
def test_document_plan_node_creation_sets_values(self):
self.assertListEqual(self.document_plan_node.children,
[self.message1, self.message2])
self.assertEqual(self.document_plan_node.relation,
Relation.ELABORATION)
self.assertEqual(str(self.document_plan_node), "ELABORATION")
def test_document_plan_node_print_tree_does_not_crash(self):
self.document_plan_node.print_tree()
def run(self, registry: Registry, random: Generator, language: str,
document_plan: DocumentPlanNode) -> Tuple[DocumentPlanNode]:
"""
Run this pipeline component.
"""
log.info("Running Morphological Realizer")
if language.endswith("-head"):
language = language[:-5]
log.debug(
"Language had suffix '-head', removing. Result: {}".format(
language))
if language not in self.language_realizers:
log.warning(
"No morphological realizer for language {}".format(language))
return (document_plan, )
self._recurse(language, document_plan)
if log.isEnabledFor(logging.DEBUG):
document_plan.print_tree()
return (document_plan, )