def parse(url):
clean_url = pat.findall(url)[0]
with get_session() as s:
data = s.get(clean_url).text
try:
lds = json.loads('{"@context"' +
find_str(data, '{"@context"', '</script'))
except JSONDecodeError:
return message(code='invalid_url')
price = price_sale = int(lds['offers'][0]['price'])
try:
x = find_str(data, 'data:[[null,null,[[[[null,[', ']\n]')
price = x.split('","')[-1].split('\xa0')[-1][:-1].replace(',', '')
price = 0 if price == '' else int(price)
except IndexError:
pass
return dict(url=clean_url,
name=lds['name'],
price=price,
price_sale=price_sale,
price_card=price_sale,
image=lds['image'],
raw=lds)
def vtex_parser(content, url):
"""Para los sitios basados en el e-commerce Vtex, que exponen los datos del producto de la misma forma"""
try:
data = json.loads(find_str(content, vtex_str, ';'))
except JSONDecodeError:
return None
if not data.get('available', False):
return message(code='out_of_stock')
sku = data.get('skus', [{}])[0]
price = sku.get('spotPrice', 0)
price_sale = sku.get('bestPrice', price)
if sku.get('listPrice') > 0:
# En las pruebas, Ilko no usa este campo, pero Casa de la Cerveza sí (precio normal)
price = sku.get('listPrice', 0)
# El precio en int aparece como "1000000" en vez de "10000.00", entonces busco en el valor formateado
# cuántos decimales hay, asumiendo que el separador decimal es una coma
price_fmt = sku.get('listPriceFormated', '').split(',')
if len(price_fmt) > 1 and len(price_fmt[-1]) > 0:
price /= (10**len(price_fmt[-1]))
price_sale /= (10**len(price_fmt[-1]))
return StoreItem(url=url,
name=data.get('name', ''),
image=sku.get('image', ''),
price=price,
price_sale=price_sale,
price_card=price_sale,
raw=data)
def parse(url):
with get_session() as s:
req = s.get(url)
try:
data_json = find_str(req.text, '__NEXT_DATA__" type="application/json">', '</script>')
data = json.loads(data_json.strip())
del data['props']['features']
del data['props']['footer']
del data['props']['header']
except json.decoder.JSONDecodeError:
return None
prod = data['props']['pageProps']['productProps']['result']
prices = {i['type']: only_numbers(i['price']) for i in prod['variants'][0]['price']}
price = prices['NORMAL']
price_sale = prices.get('AB', prices.get('INTERNET', price))
price_card = prices.get('CMR', price_sale)
return dict(
url=url,
name=prod['name'],
price=price,
price_sale=price_sale,
price_card=price_card,
image=media_url + prod['id'],
raw=data
)
def parse(url):
with get_session() as s:
req = s.get(url)
req2 = s.get(url_promos,
headers={
'x-api-key': 'IuimuMneIKJd3tapno2Ag1c1WcAES97j'
}).json()
try:
data_json = find_str(req.text, '__renderData = ', ';</script>')
data = json.loads(json.loads(data_json))
except json.decoder.JSONDecodeError:
return message(code='product_not_found')
del data['menu']
prod = data['pdp']['product'][0]['items'][0]
prod_sell = prod['sellers'][0]['commertialOffer']
price_card = prod_sell['Price']
prod_id = data['pdp']['product'][0]['productId']
if prod_id in req2['products']:
for offer_id in req2['products'][prod_id]:
if offer_id in req2['promotions']:
offer = req2['promotions'][offer_id]
if offer['tcenco'] and offer['value'] < price_card:
price_card = offer['value']
# TODO: dónde está el precio con tarjeta?¿?¿¿
return dict(url=url,
name=prod['name'],
price=prod_sell['ListPrice'],
price_sale=prod_sell['Price'],
price_card=price_card,
image=prod['images'][0]['imageUrl'],
raw=data)
def parse(url):
with get_session() as s:
req = s.get(url)
if 'https://schema.org/OutOfStock' in req.text:
return message(code='out_of_stock')
try:
data_json = find_str(req.text, '"variants":', ',"layoutType"')
data_prod = json.loads(data_json)[0]
except (json.decoder.JSONDecodeError, TypeError):
return message(code='invalid_url')
prices = {
i['type']: int(i['price'][0].replace('.', ''))
for i in data_prod['prices']
}
price = prices.get('normalPrice', prices.get('internetPrice', 0))
price_sale = prices.get('internetPrice', price)
price_card = prices.get('cmrPrice', price_sale)
return dict(url=url,
name=data_prod['name'],
price=price,
price_sale=price_sale,
price_card=price_card,
image=media_url.format(data_prod['id']),
raw=data_prod)
def parse(url):
clean_url = pat.findall(url)[0]
with get_session() as s:
data = s.get(clean_url, cookies=conf).text
try:
page_data = json.loads(find_str(data, 'data: ', ',\n'))
except JSONDecodeError:
return message(code='product_not_found')
if not page_data or 'priceModule' not in page_data:
return message(code='product_not_found')
prices = page_data['priceModule']
price_offer = price = prices['formatedPrice']
if 'formatedActivityPrice' in prices:
price_offer = prices['formatedActivityPrice']
return dict(url=clean_url,
name=page_data['pageModule']['title'],
price=price,
price_sale=price_offer,
price_card=price_offer,
image=page_data['pageModule']['imagePath'],
raw=page_data)
def parse(url):
with get_session() as s:
req = s.get(url)
try:
data_json = find_str(req.text,
'__NEXT_DATA__" type="application/json">',
'</script>')
data = json.loads(data_json.strip())
del data['props']['categories']
del data['props']['settings']
except json.decoder.JSONDecodeError:
return None
prod = data['props']['pageProps']['data']
price = prod['prices']['regularPrice']
price_sale = prod['prices'].get('discountPrice', price) or price
price_card = prod['prices'].get('cmrPrice', price_sale) or price_sale
return dict(url=url,
name=prod['name'],
price=price,
price_sale=price_sale,
price_card=price_card,
image=prod['images'][0]['url'],
raw=data)
def parse(url):
with get_session() as s:
cont = s.get(url).text
try:
cont_json = find_str(cont, '<script type="application/ld+json">',
'</script>')
if not cont_json:
raise ValueError()
data = json.loads(cont_json.strip())
except (JSONDecodeError, ValueError):
return message(code='invalid_url')
try:
price_normal = int(
find_str(cont, '"price"><s>$', '<').replace('.', ''))
except AttributeError:
price_normal = 0
prices = sorted([int(i['price']) for i in data['offers']])
price_sale = price_card = price_normal
if price_normal != 0:
if len(prices) > 1:
price_card, price_sale = prices[0], prices[1]
elif price_normal != 0 and prices[0] != price_normal:
price_card = prices[0]
if 'class="img-tc"' not in cont:
price_sale = prices[0]
else:
price_sale = price_card = price_normal = prices[0]
return dict(url=url,
name=html.unescape(data['name']),
price=price_normal,
price_sale=price_sale,
price_card=price_card,
image=data['image'],
raw=data)
def retailrocket_parser(content, url):
"""Para tiendas basadas en RetailRocket, como AudioMúsica al parecer"""
try:
data = json.loads(find_str(content, rr_str, ');'))
except JSONDecodeError:
return None
price = data.get('price', 0)
return StoreItem(url=url,
name=data.get('name', ''),
image=data.get('pictureUrl', ''),
price=price,
price_sale=price,
price_card=price,
raw=data)
def parse(url):
with get_session() as s:
req = s.get(url)
data = json.loads(find_str(req.text, '__APP_INITIAL_STATE__ = ', '</script>'))
price = only_numbers(data['deal']['priceSummary']['value']['formattedAmount'])
price_offer = only_numbers(data['deal']['priceSummary']['price']['formattedAmount'])
return dict(
url=url,
name=data['deal']['title'],
price=price,
price_sale=price_offer,
price_card=price_offer,
image=data['deal']['largeImageUrl'],
raw=data
)
def parse(url):
with get_session() as s:
req = s.get(url)
ind = 'alternate" href="android-app://cl.baytex.ripley/product/'
start = req.text.index(ind)
if start == -1:
return None
part = find_str(req.text, ind, '"')
with get_session() as s:
data = s.get(api_url + part).json()[0]
if 'prices' not in data or 'listPrice' not in data['prices']:
return message(code='out_of_stock')
return dict(url=url,
name=data['name'],
price=int(data['prices']['listPrice']),
price_sale=int(data['prices']['offerPrice']),
price_card=int(data['prices'].get(
'cardPrice', data['prices']['offerPrice'])),
image='https:' + data['fullImage'],
raw=data)