def _enrich_deals_from_soup(self, soup, deals):
for item in soup.find_all(
'article', attrs={'itemtype': 'http://schema.org/Product'}):
product_name = item.find('div', attrs={'class': 'summary'}).text
product_price = clean_price(
item.find('div', attrs={
'class': 'price'
}).text)
deals[product_name] = product_price
def _enrich_deals_from_soup(self, soup, deals):
products = soup.findAll('div', attrs={'class': re.compile('ppp-*')})
for product in products:
tag = product.find('div', attrs={'class': re.compile('product*')})
brand = tag.find('span', attrs={'class': 'marque'}).text
self.remove_all_span(tag)
second_part = tag.find('a', attrs={'class': 'prod_txt_left'}).text
product_name = brand + self.remove_garbage_characters(second_part)
product_price = clean_price(
product.find('div', attrs={
'class': 'price_prod_resp'
}).text)
deals[product_name] = product_price
def _enrich_deals_from_soup(self, soup, deals):
items = soup.find_all('li',
attrs={'data-ref': re.compile("[A-Za-z0-9]")})
for item in items:
product_name = item.find('div', attrs={
'class': 'description'
}).find('h2').text
for script_section in soup.findAll('script'):
if '.price-wrapper' in script_section.text:
print(script_section.prettify())
product_price = clean_price(
item.find('div', attrs={
'class': 'price_prod_resp'
}).text)
deals[product_name] = product_price
def _enrich_deals_from_soup(self, soup, deals):
for product in soup.find('table', attrs={
'id': 'listing_mode_display'
}).findAll('tr'):
try:
product_name = product.find('div',
attrs={
'class': 'product_description'
}).find('a').text
product_price = clean_price(
product.find('td', attrs={
'class': 'btn_price_wrapper'
}).find('b').text)
deals[product_name] = product_price
except Exception as exception:
pass
def _enrich_deals_from_soup(self, soup: bs4.BeautifulSoup, deals):
for item in soup.find_all("script"):
if "ecommerce" in item.text:
for gpu in [x for x in item.text.split("{") if "name" in x]:
product_name = None
product_price = None
# Look for name and price
for token in gpu.strip().split(","):
if "name" in token:
product_name = token.split(":")[1].replace("'", "").strip()
elif "price" in token:
product_price = token.split(":")[1]
if product_name and product_price:
deals[product_name] = clean_price(product_price)
else:
logger.warning(f"Product name [{product_name}] and price [{product_price}] seem not available.")
def _enrich_deals_from_soup(self, soup, deals):
for product in soup.findAll('div', attrs={'class': 'pcontent'}):
product_name = product.find('div', attrs={'class': 'pname'}).text
product_price = product.find('div', attrs={'class': 'pprice'}).text
deals[product_name] = clean_price(product_price)
def _enrich_deals_from_soup(self, soup, deals):
for item in soup.findAll('article', attrs={'class': 'grille-produit'}):
product_name = item.find('h3').text
product_price = clean_price(item.find('div', attrs={'itemprop': 'price'}).text)
deals[product_name] = product_price
def test_case_5(self):
self.assertEqual("1167.83", clean_price('€ 1.167,83*'))
def test_case_4(self):
self.assertEqual(self.reference_price, clean_price('4 22€45'))
def test_case_2(self):
self.assertEqual(self.reference_price,
clean_price(' --- 422,45 €*'))
def test_case_1(self):
self.assertEqual(self.reference_price,
clean_price(' 422.45 €*'))