def setUp(self):
"""Setup connection."""
self.session = db.connect(database=db.DEFAULT_DB)
def __init__(self, *args):
"""Setup connection."""
unittest.TestCase.__init__(self, *args)
self.session = db.connect(database=db.DEFAULT_DB, debug=True)
root = os.path.join('test', 'texts')
TanachParser.parse(self.session, root, 'Index.xml')
def setUp(self):
"""Establish a database connection."""
self.session = db.connect(database=db.DEFAULT_DB, debug=True)
root_file (str): name of the "Tanach.xml" file
Returns:
dict. Dictionary of words parsed.
"""
logger.info(MSG_PARSING, 'tanach')
logger.info(MSG_PARSING, 'index')
path = os.path.join(root, root_file)
xml_index = BeautifulSoup(open(path), 'xml')
logger.info(MSG_PARSED, 'index')
xml_books = xml_index.Tanach.find_all('index') # pylint: disable=E1101
xml_books = xml_books[0].books
for xml_book in xml_books.find_all('names'):
book_path = os.path.join(root, xml_book.filename.string + '.xml')
with open(book_path) as stream:
xml_soup = BeautifulSoup(stream, 'xml')
book_name, book_counts = count(xml_soup)
save(session, book_name, book_counts)
logger.info(MSG_PARSED, 'tanach')
if '__main__' == __name__: # pragma: no cover
logger = logging.getLogger('hebphonics.parsers.TanachParser')
logger.info('connecting...')
parse(db.connect(database=os.path.join('db', 'hebphonics.db')),
os.path.join('texts', 'tanach.us'))
