def populate_csv_file():
backup(csv_path)
f_csv = CsvFile(csv_path, 'w')
seen_courses = set()
course_tags_soup = BeautifulSoup(open(html_path), 'html.parser')
for course_tag in course_tags_soup.findAll('a'):
displayed_course_information = course_tag.contents[0]
department_and_level_regex = '[A-Zx0-9 \.\-]+'
if re.match('%s \- ' % department_and_level_regex, displayed_course_information):
department_and_level, title = displayed_course_information.split(' - ', 1)
elif re.search(' \- %s($|\s)' % department_and_level_regex, displayed_course_information):
title, department_and_level = displayed_course_information.rsplit(' - ', 1)
else:
title, department_and_level = displayed_course_information, ''
url = course_tag.get('href')
if (title, department_and_level) not in seen_courses:
f_csv.add_row(title, department_and_level, url)
seen_courses.add((title, department_and_level))
f_csv.close()
