def reset_pipeline(self, name):
"""
Resets a pipeline given its name. (i.e. reloads it from config)
Useful if you want to re-assemble it.
:param name: Name of the pipeline to reset.
:return: Newly created pipeline.
"""
self.pipelines[name] = from_config(self.pipeline_configs[name])
return self.pipelines[name]
def test_successful_parse_raw_text(self):
p = pipeline.from_config(self.tp)
p.assemble(ending=None)
docs = p.load("tests/resources/test.txt")
nt.assert_equal(len(docs), 1)
doc = docs[0]
nt.assert_equal(doc.language.name, "English")
nt.assert_equal(doc.marco, "polo")
nt.assert_equal(len(doc.sentences), 3)
nt.assert_equal(doc.sentences[0].words[1].normalized_text,
"corporation")
def __init__(self, cfg_or_path: Union[str, ConfigTree] = None):
super().__init__()
if isinstance(cfg_or_path, Mapping):
self.cfg = cfg_or_path
else:
try:
self.cfg = util.read_config(cfg_or_path)['main']
except Exception:
raise ValueError("Could not construct main class. Param "
"cfg_or_path should be string or dict-like config! (Was {})".format(type(cfg_or_path)))
self._docs = []
self.dist_service: EmbeddingComparator = util.safe_construct(self.cfg['embedding_comparator'],
restrict_to=EmbeddingComparator,
relative_import="estrella.operate.latent")
self.languages = language.from_config(self.cfg['languages'])
self.pipeline_configs: Dict[str, ConfigTree] = {
k: v for k, v in self.cfg.get_config("pipelines").items()
} # from name to config
self.pipelines: Dict[str, Pipeline] = {
k: from_config(v) for k, v in self.pipeline_configs.items()
} # from name to actual pipeline
def test_successful_setup_from_config_flat_kwargs(self):
p = pipeline.from_config(self.tp)
p.assemble(ending=".jpg")
nt.assert_equal(p.source_reader.ending, ".jpg")
def test_successful_setup_from_config(self):
pipeline.from_config(self.tp).assemble()