def test_shemot_filter(self):
"""Expected to filter out shemot."""
query = search.search(self.session, shemot=False)
for test in query:
self.assertFalse(hebrew.isshemot(test.hebrew))
query = search.search(self.session, shemot=True)
expected = [hebrew.isshemot(test.hebrew) for test in query]
self.assertTrue(sum(expected) > 0, 'expected at least one name of G-d')
def save(session, name, counts):
"""Saves the contents of a book.
Args:
session (session): database session
name (str): the name of the book
counts (dict): unparsed words mapped to frequency counts
"""
logger.info(MSG_SAVING, name)
db_book = db.Book(name=name)
progress = Progress(len(counts)).start()
for i, (stripped, (frequency, uni_word)) in enumerate(counts.iteritems()):
progress.update(i)
# check if word appeared in different book
db_word = session.query(db.Word)\
.filter_by(hebrew=stripped)\
.first()
if not db_word: # first occurrence of word ever
groups = hebrew.clusters(uni_word)
syllables = hebrew.syllabify(groups=groups, strict=True)
syllables_hatafs = hebrew.syllabify(groups=groups, strict=False)
db_word = db.Word(hebrew=stripped,
shemot=hebrew.isshemot(uni_word),
gematria=hebrew.gematria(stripped),
syllables=str(syllables),
syllen=len(syllables),
syllen_hatafs=len(syllables_hatafs))
db_rel = db.Occurrence(frequency=frequency)
db_rel.word = db_word
db_book.words.append(db_rel) # pylint: disable=E1101
session.add(db_book)
session.commit()
progress.end()
logger.info(MSG_SAVED, name)
def test_isshemot(self):
"""Expected to correctly identify names of G-d."""
test = u'אֱלֹהִים'
self.assertTrue(hebrew.isshemot(test))
test = u'בֵאלֹהִים'
self.assertTrue(hebrew.isshemot(test))
test = u'אֱלוֹהֵי'
self.assertTrue(hebrew.isshemot(test))
test = u'אֱלוֹהַי'
self.assertTrue(hebrew.isshemot(test))
test = u'אֵל'
self.assertTrue(hebrew.isshemot(test))
test = u'אֵלַי'
self.assertFalse(hebrew.isshemot(test), 'should not match')