def addCorpus(self, corpus, *args, **kwargs):
if corpus == "wauchier":
corpus = add_corpus("wauchier", db, *args, **kwargs)
else:
corpus = add_corpus("floovant", db, *args, **kwargs)
self.driver.get(self.get_server_url())
if self.AUTO_LOG_IN and not kwargs.get("no_corpus_user", False):
self.addCorpusUser(corpus.name,
self.app.config['ADMIN_EMAIL'],
is_owner=kwargs.get("is_owner", True))
return corpus
def db_fixtures():
""" Loads demo/tests data to the database
"""
with app.app_context():
from tests.db_fixtures import add_corpus
add_corpus(
"wauchier", db, with_token=True, tokens_up_to=None,
with_allowed_lemma=True, partial_allowed_lemma=False,
with_allowed_pos=True, partial_allowed_pos=False,
with_allowed_morph=True)
add_corpus(
"floovant", db, with_token=True, tokens_up_to=None,
with_allowed_lemma=True, partial_allowed_lemma=False,
with_allowed_pos=True, partial_allowed_pos=False,
with_allowed_morph=True)
click.echo("Loaded fixtures to the database")
def test_corpus_dump(self):
""" Test that data export works correctly """
# Ingest data this way, not the best practice, but the shortest. Will allow file comparison down the lane
with self.app.app_context():
add_corpus("floovant",
db,
tokens_up_to=3,
with_allowed_lemma=True,
partial_allowed_lemma=True,
with_allowed_morph=True,
partial_allowed_morph=True,
with_allowed_pos=True,
partial_allowed_pos=True)
with self.runner.isolated_filesystem() as f:
curr_dir = str(f)
test_dir = "some_dir"
result = self.invoke("corpus-dump", "2", "--path", test_dir)
self.assertHasContents(
os.path.join(curr_dir, test_dir,
"tokens.csv").replace("\r", ""),
"token_id form lemma POS morph\n"
"1 SOIGNORS seignor _ NOMB.=p|GENRE=m|CAS=n\n"
"2 or or4 _ DEGRE=-\n"
"3 escoutez escouter _ MODE=imp|PERS.=2|NOMB.=p\n")
self.assertIn("--- Tokens dumped", result.output)
self.assertHasContents(
os.path.join(curr_dir, test_dir, "allowed_lemma.txt"),
"escouter\nor4\nseignor")
self.assertIn("--- Allowed Lemma Values dumped", result.output)
self.assertHasContents(
os.path.join(curr_dir, test_dir, "allowed_pos.txt"),
"ADVgen,VERcjg,NOMcom")
self.assertIn("--- Allowed POS Values dumped", result.output)
self.assertHasContents(
os.path.join(curr_dir, test_dir,
"allowed_morph.csv").replace("\r", ""),
"label readable\n"
"_ pas de morphologie\n"
"DEGRE=- non applicable\n"
"MODE=imp|PERS.=2|NOMB.=p impératif 2e personne pluriel\n")
self.assertIn("--- Allowed Morphological Values dumped",
result.output)
def addCorpus(self, corpus, *args, **kwargs):
corpus = add_corpus(corpus.lower(), db, *args, **kwargs)
if self.AUTO_LOG_IN and not kwargs.get("no_corpus_user", False):
self.addCorpusUser(corpus.name,
self.app.config['ADMIN_EMAIL'],
is_owner=kwargs.get("is_owner", True))
self.driver.get(self.get_server_url())
return corpus
def test_corpus_dump_not_found(self):
""" Test that data export works correctly by not raising an issue if Corpus does not exist """
# Ingest data this way, not the best practice, but the shortest. Will allow file comparison down the lane
with self.app.app_context():
add_corpus("floovant",
db,
tokens_up_to=3,
with_allowed_lemma=True,
partial_allowed_lemma=True,
with_allowed_morph=True,
partial_allowed_morph=True,
with_allowed_pos=True,
partial_allowed_pos=True)
with self.runner.isolated_filesystem() as f:
curr_dir = str(f)
test_dir = "some_dir"
result = self.invoke("corpus-dump", "1", "--path", test_dir)
self.assertIn("Corpus not found", result.output)
self.assertEqual(len(os.listdir(os.path.join(curr_dir, test_dir))),
0, "There should be no files")
def test_corpus_list(self):
with self.app.app_context():
add_corpus("floovant",
db,
tokens_up_to=3,
with_allowed_lemma=True,
partial_allowed_lemma=True,
with_allowed_morph=True,
partial_allowed_morph=True,
with_allowed_pos=True,
partial_allowed_pos=True)
add_corpus("wauchier",
db,
tokens_up_to=3,
with_allowed_lemma=True,
partial_allowed_lemma=True,
with_allowed_morph=True,
partial_allowed_morph=True,
with_allowed_pos=True,
partial_allowed_pos=True)
result = self.invoke("corpus-list")
self.assertIn("1\t| Wauchier", result.output)
self.assertIn("2\t| Floovant", result.output)
def addCorpus(self, corpus, *args, **kwargs):
if corpus == "wauchier":
add_corpus("wauchier", db, *args, **kwargs)
else:
add_corpus("floovant", db, *args, **kwargs)