def scrape(cookie):
html = get_html(cookie)
tree = get_tree(html)
containers = tree.find_all('div', {'class': 'student_container'})
# Clear all students
Student.query.delete()
for container in containers:
student = Student()
student.surname, student.forename = clean_name(
container.find('h5', {
'class': 'yalehead'
}).text)
student.year = clean_year(
container.find('div', {
'class': 'student_year'
}).text)
student.pronoun = container.find('div', {
'class': 'student_info_pronoun'
}).text
info = container.find_all('div', {'class': 'student_info'})
student.college = info[0].text.replace(' College', '')
try:
student.email = info[1].find('a').text
except AttributeError:
student.email = guess_email(student)
trivia = info[1].find_all(text=True, recursive=False)
try:
room = trivia.pop(0) if RE_ROOM.match(trivia[0]) else None
if room:
result = RE_ROOM.search(room)
student.building_code, student.entryway, student.floor, student.suite, student.room = result.groups(
)
student.birthday = trivia.pop() if RE_BIRTHDAY.match(
trivia[-1]) else None
student.major = trivia.pop() if trivia[-1] in MAJORS else None
student.address = ', '.join(trivia)
student.state = parse_address(trivia)
except IndexError:
pass
db.session.add(student)
db.session.commit()
print('Done.')
def handle(self, *args, **options):
faker = Faker()
for _ in range(options['len']):
book = Book()
book.title = uuid.uuid4()
book.save()
subject, _ = Subject.objects.get_or_create(title='HTML')
student = Student()
student.name = faker.first_name()
student.surname = faker.last_name()
student.age = faker.random_number(digits=None)
student.address = faker.address()
student.birthday = faker.date()
student.description = faker.text()
student.email = faker.email()
student.book = book
student.subject = subject
student.save()
teacher, _ = Teacher.objects.get_or_create(name=faker.name())
teacher.students.add(student)
teacher.save()
