def _get_category_from_menu_info(self, category_node):
item_nodes = HtmlUtils.get_elements(category_node, './/div[@class="menu-item"]')
items_info = []
for item_node in item_nodes:
item_name = HtmlUtils.get_element_value(item_node, './/h3[@class="menu-item-name"]/text()')
if item_name != '':
info = {}
info['name'] = item_name
info['prices'] = self._get_menu_item_prices_info(item_node)
items_info.append(info)
return items_info
def get_rank(self, item_node):
return HtmlUtils.get_element_value(
item_node, './div[@class="bs-product-rank"]/h2/text()')
def get_category_name(self, html_doc):
return HtmlUtils.get_element_value(
html_doc,
'//div[@class="flex-horiz-product-list"]//h2[@class="best-seller-cat-title"]/text()'
)
def is_top_deal(self, deal_node):
return HtmlUtils.get_element_value(
deal_node, './/div[@class="deal-box"]/text()') != ''
def _get_dispensary_minimum_age(self, deal_node):
return '21' if HtmlUtils.get_element_value(
deal_node,
'//ul[@class="detail-list"]/li/span[contains(@class, "icon-retail")]'
) != '' else 'unkown'
def get_product_price(self, item_node):
value = HtmlUtils.get_element_value(
item_node, './/div[@class="bs-product-price"]/h4[2]/text()')
currency = HtmlUtils.get_element_value(
item_node, './/div[@class="bs-product-price"]/h4[1]/text()')
return value + currency
def get_title(self, html_document, title_xpath):
return HtmlUtils.get_element_value(html_document, title_xpath)
def _get_dispensary_address(self, deal_node):
return HtmlUtils.get_element_value(
deal_node,
'.//div[@class="text-box"]//span[@class="txt"]/text()').strip()
def _get_dispensary_phone_number(self, deal_node):
return HtmlUtils.get_element_value(
deal_node,
'.//div[@class="text-box"]//span[@class="tel-link"]/text()')
def _get_dispensary_name(self, deal_node):
return HtmlUtils.get_element_value(
deal_node, './/div[@class="profile-link"]/text()')
def _get_image_url(self, deal_node):
return 'https://www.leafbuyer.com' + HtmlUtils.get_element_value(
deal_node, '//div[contains(@class, "img-block")]/a/img/@src')
def _get_dispensary_url(self, deal_node):
return 'https://www.leafbuyer.com' + HtmlUtils.get_element_value(
deal_node, './/div[@class="loc-name-addr"]/strong/a/@href')
def get_date(self, html_document, date_xpath):
return HtmlUtils.get_element_value(html_document, date_xpath)
def get_image_url(self, html_document, image_xpath):
return HtmlUtils.get_element_value(html_document, image_xpath)
def get_brand_image(self, item_node):
image_url = HtmlUtils.get_element_value(
item_node, './/div[@class="bs-product-brand-image"]//img/@src')
return image_url if 'http' in image_url else ''
def _get_deal_name(self, deal_node):
return HtmlUtils.get_element_value(
deal_node, './/div[@class="text-wrap"]/h1/text()')
def get_brand_name(self, item_node):
return HtmlUtils.get_element_value(
item_node, './/div[@class="bs-product-name"]/div/text()')
def _strip_quantity(self, quantity):
if quantity is str:
return quantity.strip()
return HtmlUtils.get_element_value(quantity, './text()')
def get_image_chart(self, item_node):
return HtmlUtils.get_element_value(
item_node, './/div[@class="bs-product-chart"]//img/@src')
def _search_script_info(self, deal_node, script_infos):
node_id = HtmlUtils.get_element_value(deal_node, './@id')
for i in script_infos:
if i[0] in node_id:
return i
return []