def getCoursesCards(soup, isSetCategories=False):
selector = 'div.row.course-list.list > div'
cards = soup.select(selector)
data = []
for card in cards:
data.append(extractCourseCard(card))
if isSetCategories:
s = Scraper()
categoriesUrls = getCategoriesURLs(soup)
for category in categoriesUrls:
name = category.get('name', '')
url = category.get('url')
s.get(url)
tempSoup = s.html_soup()
categoryCards = getCoursesCards(tempSoup)
ids = [c['publicId'] for c in categoryCards]
for course in data:
if course['publicId'] in ids:
if course.get('categories', None):
course['categories'].append(name)
else:
course['categories'] = [name]
return data
def getTheGoodZoneDataAndMyPathes():
s = Scraper()
coursesUrl = f'{protocol}://{domain}/courses'
s.get(coursesUrl)
coursesSoup = s.html_soup()
blogsUrl = f'{protocol}://{domain}/blog'
s.get(blogsUrl)
blogsSoup = s.html_soup()
return [
{
'delPath': '/delete/courses/',
'setPath': '/set/course/',
'items': getCourses()
},
{
'delPath': '/delete/coaches/',
'setPath': '/set/coach/',
'items': getCoaches(coursesSoup)['coach']
},
{
'delPath': '/delete/instructors/',
'setPath': '/set/instructor/',
'items': getCoaches(coursesSoup)['course']
},
{
'delPath': '/delete/live-events/',
'setPath': '/set/live-event/',
'items': getCoaches(coursesSoup)['live event']
},
{
'setPath': '/set-course-description/',
'items': getCoaches(coursesSoup)['description']
},
{
'delPath': '/delete/blogs/',
'setPath': '/set/blog/',
'items': getBlogs(blogsSoup)
},
]
def extraBlogData(url):
s = Scraper()
s.get(url)
soup = s.html_soup()
data = {
'publicId': url.split('/')[4],
}
thumbnailElm = soup.select_one('.post-content img')
data.update({
'thumbnail': thumbnailElm['src'] if thumbnailElm else None,
})
return data
def getCoachingCards(soup):
selector = 'div.row.services-row.list > div'
cards = soup.select(selector)
data = {
'total': [],
'coach': [],
'live event': [],
'course': [],
'description': []
}
for card in cards:
data['total'].append(extractCoachingCard(card))
s = Scraper()
# print(data['total'])
for cardData in data['total']:
url = cardData.get('url')
s.get(url)
tempSoup = s.html_soup()
descriptionContentELm = tempSoup.select_one(
'.course-block.custom_html')
# print('Desc: ', descriptionContentELm.text)
types = ['description', 'course', 'coach', 'live event']
if not (descriptionContentELm
and any([t in descriptionContentELm.text for t in types])):
continue
descriptionContent = descriptionContentELm.text
cardData.update(parseCoaching(descriptionContent))
card = handleCoachingCardData(cardData)
data[card.pop('type')].append(card)
data.pop('total')
return data
def getCourses():
# print('get all courses')
allCourses = []
courses = []
first = True
s = Scraper()
pageNum = 1
while courses or first:
first = False
allCourses.extend(courses)
coursesUrl = f'{protocol}://{domain}/courses?page={pageNum}'
# print('\n\n\nurl: ', coursesUrl, '\n\n\n\n')
s.get(coursesUrl)
coursesSoup = s.html_soup()
courses = getCoursesCards(coursesSoup, isSetCategories=True)
pageNum += 1
return allCourses