def extract_coaching_start_date(page):
start_date = ''
start_date_text = page.find('div',attrs={"class":"flip-container"}).find("div",attrs={"class":"front"}).find("p", attrs={"class":"info"}).text
month = find_month(start_date_text)
year = find_year(start_date_text)
date = find_coaching_start_date(start_date_text)
start_date = f'{year}-{month}-{date}'
return start_date
def extract_coaching_end_date(page):
end_date = ''
end_date_text = page.find_all('div',attrs={"class":"flip-container"})[2].find("div",attrs={"class":"front"}).find(
"p",
attrs={"class":"info"}).text
month = find_month(end_date_text)
year = find_year(end_date_text)
date = find_coaching_end_date(end_date_text)
end_date = f'{year}-{month}-{date}'
return end_date
def deal_with_globle_ceo_start_date(start_date):
month = find_month(start_date)
year = find_year(start_date)
date = extract_en_globle_ceo_date_digit(start_date)
start_date = f'{year}-{month}-{date}'
return start_date
def deal_with_trans_start_date(start_date):
month = find_month(start_date)
year = find_year(start_date)
date = extract_es_trans_date_digit(start_date)
start_date = f'{year}-{month}-{date}'
return start_date
def online_version_detail(page):
location = ''
version = 2
effective_date_start = ''
effective_date_end = ''
tuition_number = ''
currency = ''
tuition_note = ''
credential = ''
type = 'Online-Virtual'
duration_type = ''
duration_num = ''
schedule = [["", "", "", "formal"]]
languages = 'English'
try:
location = page.find('div', attrs={
"class": "fourcol last"
}).find('div', attrs={
"class": "segment clearfix cajita_datos"
}).find_all('p')[1].strong.text
location = format_location(location, '')
except:
pass
try:
page.find('div', attrs={
"class": "fourcol last"
}).find('div', attrs={
"class": "segment clearfix cajita_datos"
}).find_all('p')[1].strong.extract()
start_date = page.find('div', attrs={
"class": "fourcol last"
}).find('div', attrs={
"class": "segment clearfix cajita_datos"
}).find_all('p')[1].text
year = re.findall('\d{4}', start_date)[0]
date = re.findall('\d{1,2}', start_date)[0]
if len(date) < 2:
date = '0' + date
month = find_month(start_date)
effective_date_start = f'{year}-{month}-{date}'
except Exception as e:
print(e)
try:
date_info = page.find('div', attrs={
"class": "fourcol last"
}).find('div', attrs={
"class": "segment clearfix cajita_datos"
}).find_all('p')[1].text
duration_type = get_duration_type(date_info)
duration_type = 'duration_' + duration_type
duration_num = re.findall('\d{1}', date_info)[-1]
except Exception as e:
print(e)
try:
price_related_session = page.find('div',
attrs={"class": "toggle-label"},
text="Live Online Edition Admission "
"Process & Fees")
price_session = price_related_session.find_next('div').find(
'strong', text='General fee').find_parent('li').text
tuition_number = get_tuition_number(price_session)
currency = get_currency_type(price_session)
tuition_note = price_related_session.find_next('div').find('ul').text
if 'Certificate' in tuition_note:
credential = 'Certificate'
except:
pass
try:
schedule[0][0] = effective_date_start
schedule[0][1] = effective_date_end
schedule[0][2] = str(duration_num)
except:
pass
info = {
'location': location,
'version': version,
'tuition_number': tuition_number,
'currency': currency,
'tuition_note': tuition_note,
'credential': credential,
'effective_date_start': effective_date_start,
'effective_date_end': effective_date_end,
'type': type,
'schedule': schedule,
'languages': languages
}
if 'weeks' in date_info:
duration_info = {duration_type: duration_num}
info.update(duration_info)
return info