def make_test_course(): course = CourseraCourse() course.cid = 'introstats2' course.n_seq = 1 # '001' course.title = 'Introduction to Statistics' course.description = 'Introduction to Statistics is nice course!' course.start_date = datetime.date(2013, 4, 5) course.duration_in_weeks = 8 # course.workload_in_hours_per_day = 3 course.workload_in_hours_per_week = 3 institution = Institution() institution.id = 10 institution.name = 'Harvard Univ.' course.institutions = [institution] professor = Instructor() professor.id = 10 professor.name = 'John Joey' professor.institution = institution course.instructors = [professor] category = Category() category.id = 10 category.name = 'Mathematics & Statistics' course.categories = [category] print 'course', course print 'Instructors', course.instructors.values() return course
def process_element(element):
elements_l1 = element.getchildren()
# first element of incoming element <div> is expected to be <h3>
print 'len(elements_l1)', len(elements_l1)
h3 = elements_l1[0]
# first element is expected to be <h3>, if not, return
if h3.tag != 'h3':
return None
print 'Got <h3>', h3.getchildren(), 'l1',elements_l1[1].tag, elements_l1[2].tag, elements_l1[3].tag
elements_l2 = h3.getchildren()
a = elements_l2[0]
# first element of h3 is expected to be <a>, if not, return
if a.tag != 'a':
return None
url = a.get('href')
course_id = url.split('/')[-1]
print 'course_id', course_id
if not is_course_id_good(course_id):
return None
# "coursera-course-listing-more coursera-course-my-listing-more"
university = Institution()
try:
div_that_has_university_info = elements_l1[3]
inner_a = div_that_has_university_info.getchildren()[0]
print 'university_class_node.text', inner_a.text
university.name = inner_a.text
except IndexError:
pass
course = Course()
course.course_id = course_id
course.title = a.text
course.university = university
return course
def save_courses_subset_to_db(self):
for i, course_subset in enumerate(self.courses_subset):
#if course_subset.university == None:
#continue
try:
print str(i+1).zfill(3), 'Saving to db', course_subset
except UnicodeEncodeError:
print str(i+1).zfill(3), 'Saving to db'
course = CourseraCourse()
course.cid = course_subset.cid
course.n_seq = course_subset.get_n_seq()
course.title = course_subset.title
if course_subset.start_date != None:
course.start_date = course_subset.start_date
if course_subset.duration_in_weeks != None:
course.duration_in_weeks = course_subset.duration_in_weeks
#course.save()
if course_subset.university != None:
university_name = course_subset.university
try:
institution = Institution.objects.get(name=university_name)
except Institution.DoesNotExist:
institution = Institution()
institution.name = university_name
#institution.courseracourse_id = course.cid
institution.save()
print 'institution id', institution.id
#CourseraCourse.objects.create(name=university_name)
except AttributeError, e:
print 'university_name', university_name
print e
#sys.exit(0)
pass
#course.in
#course.institutions.add(institution.id)
course.save()
