def test_register_unregister_synonym_key(self):
TaxonomyManager.register_synonym_key("synonym_key_1")
self.assertRaises(RecordAlreadyExistsError,
TaxonomyManager.register_synonym_key,
"synonym_key_1")
df = TaxonomyManager.get_synonym_keys()
self.assertEqual(len(df), 1)
TaxonomyManager.unregister_synonym_key("synonym_key_1")
self.assertRaises(NoRecordFoundError,
TaxonomyManager.unregister_synonym_key,
"synonym_key")
df = TaxonomyManager.get_synonym_keys()
self.assertEqual(len(df), 0)
# Test with bind
with Connector.get_connection() as connection:
TaxonomyManager.register_synonym_key("test", bind=connection)
TaxonomyManager.assert_synonym_key_exists("test", bind=connection)
TaxonomyManager.unregister_synonym_key("test", bind=connection)
TaxonomyManager.assert_synonym_key_does_not_exists("test",
bind=connection)
def _process(self,
*args,
include_mptt=False,
include_synonyms=False,
flatten=False,
**kwargs):
"""
Return the taxon dataframe.
:param include_mptt: If True, include the mptt columns.
:param include_synonyms: If True, include the stored synonyms for each
taxon.
:param flatten: If True, flattens the taxonomy hierarchy and include
it in the resulting dataframe.
"""
with Connector.get_connection() as connection:
keys = TaxonomyManager.get_synonym_keys()['name']
synonyms = []
if include_synonyms:
synonyms = [
meta.taxon.c.synonyms[k].label(k) for k in keys
if k != 'niamoto'
]
mptt = []
if include_mptt:
mptt = [
meta.taxon.c.mptt_left.label('mptt_left'),
meta.taxon.c.mptt_right.label('mptt_right'),
meta.taxon.c.mptt_tree_id.label('mptt_tree_id'),
meta.taxon.c.mptt_depth.label('mptt_depth'),
]
sel = select([
meta.taxon.c.id.label('id'),
meta.taxon.c.full_name.label('full_name'),
meta.taxon.c.rank_name.label('rank_name'),
cast(meta.taxon.c.rank, String).label('rank'),
meta.taxon.c.parent_id.label('parent_id'),
] + synonyms + mptt)
df = pd.read_sql(sel, connection, index_col='id')
# Replace None values with nan
df.fillna(value=pd.np.NAN, inplace=True)
if flatten:
df = _flatten(df)
return df, [], {'index_label': 'id'}
def test_get_synonym_keys(self):
df = TaxonomyManager.get_synonym_keys()
self.assertEqual(len(df), 0)
def test_set_taxonomy(self):
result, synonyms = TaxonomyManager.set_taxonomy(
pd.DataFrame(
columns=['full_name', 'rank_name', 'rank', 'parent_id'], ))
self.assertEqual(result, 0)
data = pd.DataFrame.from_records([
{
'id': 0,
'full_name': 'Family One',
'rank_name': 'One',
'rank': niamoto_db_meta.TaxonRankEnum.FAMILIA,
'parent_id': None,
},
],
index='id')
result, synonyms = TaxonomyManager.set_taxonomy(data)
self.assertEqual(result, 1)
data = pd.DataFrame.from_records([
{
'id': 0,
'full_name': 'Family One',
'rank_name': 'One',
'rank': niamoto_db_meta.TaxonRankEnum.FAMILIA,
'parent_id': None,
'gbif': 5,
'taxref': 1,
},
{
'id': 1,
'full_name': 'Genus Two',
'rank_name': 'Two',
'rank': niamoto_db_meta.TaxonRankEnum.GENUS,
'parent_id': 0,
'gbif': 10,
'taxref': 2,
},
{
'id': 2,
'full_name': 'Species Three',
'rank_name': 'Three',
'rank': niamoto_db_meta.TaxonRankEnum.SPECIES,
'parent_id': None,
'gbif': 7,
'taxref': 3,
},
],
index='id')
result, synonyms = TaxonomyManager.set_taxonomy(data)
self.assertEqual(result, 3)
df = TaxonomyManager.get_raw_taxon_dataframe()
self.assertEqual(len(df), 3)
synonym_keys = TaxonomyManager.get_synonym_keys()
self.assertEqual(len(synonym_keys), 2)
identity_synonyms = TaxonomyManager.get_synonyms_for_key(
TaxonomyManager.IDENTITY_SYNONYM_KEY)
self.assertEqual(len(identity_synonyms), 3)
null_synonyms = TaxonomyManager.get_synonyms_for_key(None)
self.assertEqual(len(null_synonyms), 0)
gbif_synonyms = TaxonomyManager.get_synonyms_for_key("gbif")
self.assertEqual(len(gbif_synonyms), 3)
taxref_synonyms = TaxonomyManager.get_synonyms_for_key('taxref')
self.assertEqual(len(taxref_synonyms), 3)