def populate_database(search_connection, test_data):
"""Set up the database to conduct searches on the test texts.
Fixtures
--------
search_connection
TessMongoConnection for search unit tests.
test_data
Example data for unit testing.
"""
for text in test_data['texts']:
tessfile = TessFile(text['path'], metadata=Text(**text))
search_connection.insert(tessfile.metadata)
if text['language'] == 'latin':
tok = LatinTokenizer(search_connection)
unitizer = Unitizer()
tokens, tags, features = tok.tokenize(tessfile.read(),
text=tessfile.metadata)
search_connection.update(features)
lines, phrases = unitizer.unitize(tokens, tags, tessfile.metadata)
search_connection.insert(lines + phrases)
search_connection.insert(tokens)
yield
search_connection.connection['texts'].delete_many({})
search_connection.connection['tokens'].delete_many({})
search_connection.connection['features'].delete_many({})
search_connection.connection['units'].delete_many({})
search_connection.connection['matches'].delete_many({})
search_connection.connection['searches'].delete_many({})
def test_unitize_elision_file(unit_connection, tessfiles_greek_path):
tokenizer = GreekTokenizer(unit_connection)
t = Text(path=str(tessfiles_greek_path.joinpath('test.elision.tess')),
language='greek')
tessfile = TessFile(t.path, metadata=t)
unitizer = Unitizer()
tokens, tags, features = tokenizer.tokenize(tessfile.read(), text=t)
lines, phrases = unitizer.unitize(tokens, tags, tokens[0].text)
assert len(lines) == 1
def test_unitize_notag_file(unit_connection, tessfiles_latin_path):
tokenizer = LatinTokenizer(unit_connection)
t = Text(path=str(tessfiles_latin_path.joinpath('test.notag.tess')),
language='latin')
tessfile = TessFile(t.path, metadata=t)
unitizer = Unitizer()
tokens, tags, features = tokenizer.tokenize(tessfile.read(), text=t)
lines, phrases = unitizer.unitize(tokens, tags, tokens[0].text)
assert len(lines) == 1
def test_unitize_linebreak_file(unit_connection, tessfiles_latin_path):
tokenizer = LatinTokenizer(unit_connection)
t = Text(path=str(tessfiles_latin_path.joinpath('test.linebreak.tess')),
language='latin')
tessfile = TessFile(t.path, metadata=t)
unitizer = Unitizer()
tokens, tags, features = tokenizer.tokenize(tessfile.read(), text=t)
lines, phrases = unitizer.unitize(tokens, tags, tokens[0].text)
assert len(lines) == 1
first_tag = phrases[0].tags[0]
for phrase in phrases[1:]:
assert phrase.tags[0] == first_tag
def unitizer_inputs(unit_tessfiles, unit_connection):
inputs = []
tokenizer_selector = {
'latin': LatinTokenizer(unit_connection),
'greek': GreekTokenizer(unit_connection)
}
for t in unit_tessfiles:
tessfile = TessFile(t.path, metadata=t)
tokens, tags, features = tokenizer_selector[t.language].tokenize(
tessfile.read(), text=t)
features.sort(key=lambda x: x.index)
inputs.append((tokens, tags, features))
yield inputs
def test_normalize(self, latin_files, latin_tokens):
la = self.__test_class__()
for i in range(len(latin_files)):
fname = latin_files[i]
ref_tokens = [t for t in latin_tokens[i] if 'FORM' in t]
t = TessFile(fname)
tokens = la.normalize(t.read())
correct = map(
lambda x:
('FORM' in x[1] and x[0] == x[1]['FORM']) or x[0] == '',
zip(tokens, ref_tokens))
def test_unitize_diacrit_in_latin(unit_connection, tessfiles_latin_path):
tokenizer = LatinTokenizer(unit_connection)
t = Text(path=str(
tessfiles_latin_path.joinpath('test.diacrit_in_latin.tess')),
language='latin')
tessfile = TessFile(t.path, metadata=t)
unitizer = Unitizer()
tokens, tags, features = tokenizer.tokenize(tessfile.read(), text=t)
forms = {f.index: f.token for f in features if f.feature == 'form'}
lines, phrases = unitizer.unitize(tokens, tags, tokens[0].text)
for phrase in phrases:
for t in phrase.tokens:
cur_form = t['features']['form'][0]
if cur_form != -1:
normalized = tokenizer.normalize(t['display'])[0][0]
assert normalized == forms[cur_form], phrase.snippet
def test_unitize_linebreak_end(unit_connection, tessfiles_latin_path):
tokenizer = LatinTokenizer(unit_connection)
t = Text(path=str(
tessfiles_latin_path.joinpath('test.linebreak_end.tess')),
language='latin')
tessfile = TessFile(t.path, metadata=t)
unitizer = Unitizer()
tokens, tags, features = tokenizer.tokenize(tessfile.read(), text=t)
lines, phrases = unitizer.unitize(tokens, tags, tokens[0].text)
print('# lines')
for line in lines:
print(line.snippet)
print('# phrases')
for phrase in phrases:
print(phrase.snippet)
assert len(lines) == 2
def test_unitize_nopunctuation_file(unit_connection, tessfiles_latin_path):
# when there is no ending punctuation despite coming to the end of a poem
# and another poem starts after a blank line
tokenizer = LatinTokenizer(unit_connection)
t = Text(path=str(
tessfiles_latin_path.joinpath('test.nopunctuation.tess')),
language='latin')
tessfile = TessFile(t.path, metadata=t)
unitizer = Unitizer()
tokens, tags, features = tokenizer.tokenize(tessfile.read(), text=t)
lines, phrases = unitizer.unitize(tokens, tags, tokens[0].text)
assert len(lines) == 68
for prev_phrase, cur_phrase in zip(phrases[:-1], phrases[1:]):
if '2.13' in prev_phrase.tags[0] and '2.14' in cur_phrase.tags[0]:
assert prev_phrase.snippet == 'quin et Prometheus et Pelopis parens / dulci laborem decipitur sono / nec curat Orion leones / aut timidos agitare lyncas / Eheu fugaces, Postume, Postume, / labuntur anni nec pietas moram / rugis et instanti senectae / adferet indomitaeque morti, / non, si trecenis quotquot eunt dies, / amice, places inlacrimabilem / Plutona tauris, qui ter amplum / Geryonen Tityonque tristi / conpescit unda, scilicet omnibus / quicumque terrae munere vescimur / enaviganda, sive reges / sive inopes erimus coloni. / '
assert cur_phrase.snippet == 'frustra cruento Marte carebimus / fractisque rauci fluctibus Hadriae, / frustra per autumnos nocentem / corporibus metuemus Austrum: / '
break
def test_tokenize(self, greek_files, greek_tokens, greek_word_frequencies):
grc = self.__test_class__()
for k in range(len(greek_files)):
fname = greek_files[k]
ref_tokens = [t for t in greek_tokens[k] if 'FORM' in t]
ref_freqs = greek_word_frequencies[k]
t = TessFile(fname)
tokens, frequencies = grc.tokenize(t.read())
tokens = [
t for t in tokens
if re.search('[\w]', t.display, flags=re.UNICODE)
]
correct = map(lambda x: x[0].display == x[1]['DISPLAY'],
zip(tokens, ref_tokens))
if not all(correct):
print(fname)
for j in range(len(tokens)):
if tokens[j].display != ref_tokens[j]['DISPLAY']:
print(ref_tokens[j])
print('{}->{}'.format(tokens[j].display,
ref_tokens[j]['DISPLAY']))
print('{}->{}'.format(tokens[j].form,
ref_tokens[j]['FORM']))
assert all(correct)
correct = map(lambda x: x[0].form == x[1]['FORM'],
zip(tokens, ref_tokens))
if not all(correct):
print(fname)
for j in range(len(tokens)):
if tokens[j].form != ref_tokens[j]['FORM']:
print(ref_tokens[j])
print('{}->{}'.format(tokens[j].form,
ref_tokens[j]['FORM']))
assert all(correct)
def test_tokenize(self, latin_files, latin_tokens, latin_word_frequencies):
la = self.__test_class__()
for k in range(len(latin_files)):
fname = latin_files[k]
ref_tokens = [t for t in latin_tokens[k] if 'FORM' in t]
ref_freqs = latin_word_frequencies[k]
t = TessFile(fname)
tokens, frequencies = la.tokenize(t.read(), text=t.metadata)
tokens = [
t for t in tokens
if re.search(r'^[a-zA-Z]+$', t.display, flags=re.UNICODE)
]
correct = map(lambda x: x[0].display == x[1]['DISPLAY'],
zip(tokens, ref_tokens))
if not all(correct):
print(fname)
for j in range(len(tokens)):
if tokens[j].display != ref_tokens[j]['DISPLAY']:
print('{}->{}'.format(tokens[j].display,
ref_tokens[j]['DISPLAY']))
assert all(correct)
correct = map(
lambda x: ('FORM' in x[1] and x[0].form == x[1]['FORM']) or
not x[0].form, zip(tokens, ref_tokens))
if not all(correct):
print(fname)
# for j in range(len(tokens)):
# if tokens[j].form != ref_tokens[j]['FORM']:
# print('{}->{}'.format(tokens[j].form, ref_tokens[j]['FORM']))
assert all(correct)
for key in ref_freqs:
assert key in la.frequencies
assert la.frequencies[key] == ref_freqs[key]
diff = []
for word in frequencies:
if word.form not in ref_freqs and re.search(
r'[a-zA-Z]', word.form, flags=re.UNICODE):
diff.append(word.form)
print(diff)
assert len(diff) == 0
keys = sorted(list(ref_freqs.keys()))
frequencies.sort(key=lambda x: x.form)
correct = map(
lambda x: x[0].form == x[1] and x[0].frequency == ref_freqs[x[
1]], zip(frequencies, keys))
assert all(correct)
la.clear()
