def test_update_anahashes(dbsession, datafiles):
wfs = pd.DataFrame()
wfs['wordform'] = ['wf-a', 'wf-b', 'wf-c']
alphabet_file = datafiles.listdir()[0]
bulk_add_wordforms(dbsession, wfs)
anahashes = dbsession.query(Anahash).order_by(Anahash.anahash_id).all()
assert len(anahashes) == 0
wrdfrms = dbsession.query(Wordform).order_by(Wordform.wordform_id).all()
for w in wrdfrms:
assert (w.anahash) is None
update_anahashes(dbsession, alphabet_file)
# If we don't commit here, the anahashes won't be updated when we do the
# tests.
dbsession.commit()
wrdfrms = dbsession.query(Wordform).order_by(Wordform.wordform_id).all()
anahashes = dbsession.query(Anahash).order_by(Anahash.anahash_id).all()
# Three anahashes were added
assert len(anahashes) == 3
# The anahashes are connected to the correct wordforms
for wf, a in zip(wrdfrms, (3, 2, 1)):
assert wf.anahash_id == a
def test_bulk_add_wordforms_not_unique(dbsession):
wfs = pd.DataFrame()
wfs['wordform'] = ['wf1', 'wf1', 'wf2']
print(dbsession)
bulk_add_wordforms(dbsession, wfs)
wrdfrms = dbsession.query(Wordform).order_by(Wordform.wordform_id).all()
assert len(wrdfrms) == 2
def test_update_anahashes_empty_wf(dbsession, datafiles):
wfs = pd.DataFrame()
wfs['wordform'] = ['wf-a', 'wf-b', 'wf-c', ' ']
alphabet_file = datafiles.listdir()[0]
bulk_add_wordforms(dbsession, wfs)
# make sure ticcl doesn't choke on the empty wordform (it must not be added
# to the database)
update_anahashes(dbsession, alphabet_file)
def test_bulk_add_wordforms_all_new(dbsession):
wfs = pd.DataFrame()
wfs['wordform'] = ['wf1', 'wf2', 'wf3']
print(dbsession)
bulk_add_wordforms(dbsession, wfs)
wordforms = dbsession.query(Wordform).order_by(Wordform.wordform_id).all()
assert len(wordforms) == len(wfs['wordform'])
assert [wf.wordform for wf in wordforms] == list(wfs['wordform'])
def test_bulk_add_wordforms_drop_empty_and_nan(dbsession):
wfs = pd.DataFrame()
wfs["wordform"] = ["wf1", "", "wf2", np.NaN]
print(dbsession)
bulk_add_wordforms(dbsession, wfs)
wrdfrms = dbsession.query(Wordform).order_by(Wordform.wordform_id).all()
assert len(wrdfrms) == 2
assert wrdfrms[0].wordform == "wf1"
assert wrdfrms[1].wordform == "wf2"
def test_bulk_add_wordforms_whitespace(dbsession):
wfs = pd.DataFrame()
wfs['wordform'] = ['wf1 ', ' wf2', ' ', ' \t']
print(dbsession)
bulk_add_wordforms(dbsession, wfs)
wrdfrms = dbsession.query(Wordform).order_by(Wordform.wordform_id).all()
assert len(wrdfrms) == 2
assert wrdfrms[0].wordform == 'wf1'
assert wrdfrms[1].wordform == 'wf2'
def test_bulk_add_wordforms_replace_underscores(dbsession):
wfs = pd.DataFrame()
wfs["wordform"] = ["wf_1", "wf 2"]
print(dbsession)
bulk_add_wordforms(dbsession, wfs)
wrdfrms = dbsession.query(Wordform).order_by(Wordform.wordform_id).all()
assert len(wrdfrms) == 2
assert wrdfrms[0].wordform == "wf*1"
assert wrdfrms[1].wordform == "wf_2"
def test_get_word_frequency_df(dbsession):
wfs = pd.DataFrame()
wfs['wordform'] = ['wf1', 'wf2', 'wf3']
bulk_add_wordforms(dbsession, wfs)
freq_df = get_word_frequency_df(dbsession)
expected = pd.DataFrame({
'wordform': ['wf1', 'wf2', 'wf3'],
'frequency': [1, 1, 1]
}).set_index('wordform')
assert freq_df.equals(expected)
def test_bulk_add_anahashes(dbsession):
wfs = pd.DataFrame()
wfs['wordform'] = ['wf1', 'wf2', 'wf3']
bulk_add_wordforms(dbsession, wfs)
a = pd.DataFrame({
'wordform': ['wf1', 'wf2', 'wf3'],
'anahash': [1, 2, 3]
}).set_index('wordform')
bulk_add_anahashes(dbsession, a)
ahs = dbsession.query(Anahash).order_by(Anahash.anahash_id).all()
print(ahs[0])
assert [a.anahash for a in ahs] == list(a['anahash'])
def test_bulk_add_wordforms_some_new(dbsession):
wfs = pd.DataFrame()
wfs['wordform'] = ['wf1', 'wf2', 'wf3']
print(dbsession)
bulk_add_wordforms(dbsession, wfs)
wfs['wordform'] = ['wf3', 'wf4', 'wf5']
wfs['wordform_lowercase'] = ['wf3', 'wf4', 'wf4']
n = bulk_add_wordforms(dbsession, wfs)
assert n == 2
wrdfrms = dbsession.query(Wordform).order_by(Wordform.wordform_id).all()
assert len(wrdfrms) == 5
assert [w.wordform for w in wrdfrms] == ['wf1', 'wf2', 'wf3', 'wf4', 'wf5']
def test_update_anahashes_nothing_to_update(dbsession, datafiles):
wfs = pd.DataFrame()
wfs['wordform'] = ['wf1', 'wf2', 'wf3']
bulk_add_wordforms(dbsession, wfs)
a = pd.DataFrame({
'wordform': ['wf1', 'wf2', 'wf3'],
'anahash': [1, 2, 3]
}).set_index('wordform')
bulk_add_anahashes(dbsession, a)
connect_anahashes_to_wordforms(dbsession, a, a['anahash'].to_dict())
alphabet_file = datafiles.listdir()[0]
update_anahashes(dbsession, alphabet_file)
wrdfrms = dbsession.query(Wordform).order_by(Wordform.wordform_id).all()
assert [wf.anahash.anahash for wf in wrdfrms] == list(a['anahash'])
def test_connect_anahashes_to_wordforms(dbsession):
wfs = pd.DataFrame()
wfs['wordform'] = ['wf1', 'wf2', 'wf3']
bulk_add_wordforms(dbsession, wfs)
wfs = get_word_frequency_df(dbsession, add_ids=True)
wf_mapping = wfs['wordform_id'].to_dict()
a = pd.DataFrame({
'wordform': ['wf1', 'wf2', 'wf3'],
'anahash': [1, 2, 3]
}).set_index('wordform')
bulk_add_anahashes(dbsession, a)
connect_anahashes_to_wordforms(dbsession, a, wf_mapping)
wrdfrms = dbsession.query(Wordform).order_by(Wordform.wordform_id).all()
assert [wf.anahash.anahash for wf in wrdfrms] == list(a['anahash'])
def test_connect_anahashes_to_wordforms_empty(dbsession):
wfs = pd.DataFrame()
wfs['wordform'] = ['wf1', 'wf2', 'wf3']
bulk_add_wordforms(dbsession, wfs)
a = pd.DataFrame({
'wordform': ['wf1', 'wf2', 'wf3'],
'anahash': [1, 2, 3]
}).set_index('wordform')
bulk_add_anahashes(dbsession, a)
connect_anahashes_to_wordforms(dbsession, a, a['anahash'].to_dict())
# nothing was updated the second time around (the values didn't change)
# (and there is no error when running this)
connect_anahashes_to_wordforms(dbsession, a, a['anahash'].to_dict())
wrdfrms = dbsession.query(Wordform).order_by(Wordform.wordform_id).all()
assert [wf.anahash.anahash for wf in wrdfrms] == list(a['anahash'])
