def parse_product_detail(self, response):
product = items.Ecommerce_product_items()
product['domain_name'] = 'www.othoba.com'
product['url'] = response.url
product['title'] = response.selector.xpath(
'//h2/span[@itemprop="name"]/text()').extract_first()
a = response.selector.xpath(
'//ol[@class="breadcrumb"]//li/a/text()').extract()
category_name = a[1:]
product['categories'] = category_name
product['currency'] = 'BDT'
s = response.selector.xpath(
'//span[@itemprop="price"]/text()').extract_first()
s = s.replace(' ', '')
s = s.replace(',', '')
num = float(s)
product['price'] = num
print 'price float--------------------------> '
print num
img1 = response.selector.xpath(
'//div[@style="margin-top: 10px;text-align: left;"]//div/a/@href'
).extract()
img2 = response.selector.xpath(
'//a[@itemprop="image"]/@href').extract()
product['images'] = img1 + img2
product['last_updated'] = datetime.datetime.now()
yield product
def parse_product_detail(self, response):
product = items.Ecommerce_product_items() ## this name is same as class name in items.py
product['domain_name'] = 'www.priyoshop.com'
product['url'] = response.url
title = response.selector.xpath('//div[@class="product-name"]/h1[@itemprop="name"]/text()').extract_first()
product['title'] = title[33:]
a = response.selector.xpath('//div[@class="breadcrumb"]/ul//li/span/a/span/text()').extract()
category_name = a[1:]
product['categories'] = category_name
product['currency'] ='BDT'
s = response.selector.xpath('//span[@itemprop="price"]/text()').extract_first()
s = s.replace(' ','')
s = s.replace(',','')
s = s.replace('Tk','')
num = float(s)
product['price'] = num
print num
img1 = response.selector.xpath('//div[@class="picture-thumbs"]//a/@href').extract()
img2 = response.selector.xpath('//img[@itemprop="image"]/@src').extract()
s = img1 + img2
product['images'] = s
product['last_updated'] = datetime.datetime.now()
yield product
def parse_product_detail(self, response):
product = items.Ecommerce_product_items()
product['domain_name'] = 'www.pickaboo.com'
product['url'] = response.url
a = response.selector.xpath('//div[@class="breadcrumbs"]/ul//li/a/span/text()').extract()
category_name = a[1:]
product['categories'] = category_name
product['title'] = response.selector.xpath('//h1[@itemprop="name"]/text()').extract_first()
product['currency'] ='BDT'
price = response.selector.xpath('//span[@class="regular-price"]//span[@class="price"]/text()').extract_first()
if price is not None:
s = price
else:
s = response.selector.xpath('//p[@class="special-price"]//span[@class="price"]/text()').extract_first()
s = s.replace(' ','')
s = s.replace(',','')
s = s[1:]
num = float(s)
product['price'] = num
print num
img1 = response.selector.xpath('//a[@class="magnify-zoom-gallery"]/img/@src').extract()
img2 = response.selector.xpath('//img[@id="magnify-small"]/@src').extract()
product['images'] = img1 + img2
product['last_updated'] = datetime.datetime.now()
yield product
def parse_product_detail(self, response):
product = items.Ecommerce_product_items()
product['domain_name'] = 'www.bagdoom.com'
product['url'] = response.url
product['title'] = response.selector.xpath('//div[@class="product-name"]/h1/text()').extract_first()
a = response.selector.xpath('//div[@class="breadcrumbs"]/ul//li/a/text()').extract()
category_name = a[1:]
product['categories'] = category_name
product['currency'] ='BDT'
price = response.selector.xpath('//p [@class="special-price"]/span [@class="price"]/text()').extract_first()
if price is not None:
s = price
else:
s = response.selector.xpath('//span[@class="price"]/text()').extract_first()
s = s.replace(' ','')
s = s.replace(',','')
s = s.replace('Tk.','') # where 'Tk.' is substring
num = float(s)
product['price'] = num
product['images'] = response.selector.xpath('//div[@class="more-views"]/ul//li/a/@href').extract()
product['last_updated'] = datetime.datetime.now()
yield product
def parse_product_detail(self, response):
product = items.Ecommerce_product_items()
product['domain_name'] = 'easyshoppingbd.com'
product['url'] = response.url
product['title'] = response.selector.xpath('//h1[@itemprop="name"]/text()').extract_first()
a = response.selector.xpath('//span[@itemprop="title"]/text()').extract()
product['categories'] = a
product['currency'] ='BDT'
s = response.selector.xpath('//span[@id="our_price_display"]/text()').extract_first()
s = s.replace(' ','')
s = s.replace(',','')
s = s.replace('Tk.','')
s = s[1:]
num = float(s)
product['price'] = num
print num
img1 = response.selector.xpath('//ul[@id="thumbs_list_frame"]//li/a/@href').extract()
img2 = response.selector.xpath('//img[@id="bigpic" and @itemprop="image"]/@src').extract()
product['images'] = img1 + img2
product['last_updated'] = datetime.datetime.now()
yield product
def parse_product_detail(self, response):
product = items.Ecommerce_product_items()
product['url'] = response.url
product['domain_name'] = "www.daraz.com.bd"
product['title'] = response.selector.xpath(
'//div[@class="details -validate-size"]/span/h1[@class="title"]/ text()'
).extract_first()
a = response.selector.xpath(
'//nav[@class="osh-breadcrumb"]/ul//li/a/text()').extract()
l = len(a) - 1
a = a[1:l]
product['categories'] = a
product['currency'] = 'BDT'
p = response.selector.xpath(
'//span[@dir="ltr"]/text()').extract_first()
p = p.replace(' ', '')
p = p.replace(',', '')
num = float(p)
product['price'] = num
print num
#product['images'] = response.selector.xpath(u'//div[@class="product-preview"]/img/@data-zoom').extract() #one image
product['images'] = response.selector.xpath(
'//div[@class="thumbs-wrapper"]/div[@id="thumbs-slide"]//a/@href'
).extract() #image list
product['last_updated'] = datetime.datetime.now()
#'https://bd.daraz.io/EtGrdRLf6i5cz3R7boLgMrApIF4=/fit-in/680x680/filters:fill(white)/product/27/1733/1.jpg?6043',
#u'https://bd.daraz.io/zGz4XrWyA2iB8ZpGhYLM7Wilt0I=/fit-in/680x680/filters:fill(white)/product/27/1733/2.jpg?6043'
yield product
def parse_product_detail(self, response):
product = items.Ecommerce_product_items()
product['domain_name'] = 'dam.com.bd'
product['url'] = response.url
product['title'] = response.selector.xpath(
'//h2[@itemprop="name"]/text()').extract_first()
a = response.selector.xpath(
'//div[@id="breadcrumbs"]/ul//li/a/span/text()').extract()
l = len(a) - 1
product['categories'] = a[1:l]
product['currency'] = 'BDT'
s = response.selector.xpath(
'//span[@itemprop="lowPrice"]/text()').extract_first()
s = s.replace(' ', '')
s = s.replace(',', '')
s = s.replace('Tk.', '')
num = float(s)
product['price'] = num
img1 = response.selector.xpath(
'//div[@class="clearfix padder-v-sm"]//img/@src').extract()
img2 = response.selector.xpath(
'//img[@itemprop="image"]/@src').extract()
s = img1 + img2
product['images'] = s
product['last_updated'] = datetime.datetime.now()
yield product
def parse_product_detail(self, response):
product = items.Ecommerce_product_items()
product['domain_name'] = 'www.nanarokom.com'
product['url'] = response.url
product['title'] = response.selector.xpath('//h1[@class="product_title entry-title"]/text()').extract_first()
a = response.selector.xpath('//nav[@class="woocommerce-breadcrumb"]//a/text()').extract()
l = len(a)-1
category_name = a[1:l]
product['categories'] = category_name
product['currency'] ='BDT'
a = response.selector.xpath('//p[@class="price"]/ins/span[@class="woocommerce-Price-amount amount"]/text()').extract_first()
if a is not None:
s = a
else:
s = response.selector.xpath('//p[@class="price"]/span[@class="woocommerce-Price-amount amount"]/text()').extract_first()
s = s.replace(' ','')
s = s.replace(',','')
print " ssssssssssssssssss "
print s
num = float(s)
product['price'] = num
print num
img1 = response.selector.xpath('//figure[@class="woocommerce-product-gallery__wrapper"]//div//a/@href').extract()
img2 = response.selector.xpath('//a[@itemprop="image"]/@href').extract()
s = img1 + img2
product['images'] = s
product['last_updated'] = datetime.datetime.now()
yield product
def parse_product_detail(self, response):
product = items.Ecommerce_product_items()
product['url'] = response.url
product['domain_name'] = "www.rokomari.com"
product['title'] = response.selector.xpath(
'//div[@class="buyArea"]/h2/text()').extract_first()
cat = response.selector.xpath(
'//ol[@class="breadcrumb"]//li/a/text()').extract()
product['categories'] = cat[1:]
product['currency'] = 'BDT'
#PRICE OUTPUT FORMAT='Tk. 284'
p = response.selector.xpath(
'//span[@class="mainPrice"]/text()').extract_first()
p = p.replace(' ', '')
p = p.replace(',', '')
p = p.replace('Tk.', '') # where 'Tk.' is substring
p = p.replace('BDT', '')
l = len(p)
num = float(p)
product['price'] = num
a = response.selector.xpath(
'//div[@class="bookImgArea"]//img/@src').extract()
a = a[1:]
img_others = response.selector.xpath(
'//ul[@class="list-unstyled"]//li/@hovermax').extract()
sum_img = a + img_others
product['images'] = sum_img
product['last_updated'] = datetime.datetime.now()
yield product
def parse_product_detail(self, response):
product = items.Ecommerce_product_items()
product['url'] = response.url
product['domain_name'] = "www.kiksha.com"
t = response.selector.xpath(
'//h1[@itemprop="name"]/text()').extract_first()
l = len(t)
t = t[65:l]
product['title'] = t
product['currency'] = 'BDT'
price = response.selector.xpath(
'//p[@class="special-price"]//span[@class="price"]/text()'
).extract_first()
if price is not None:
p = price
else:
p = response.selector.xpath(
'//span[@class="price"]/text()').extract_first()
p = p.replace(' ', '')
p = p.replace(',', '')
p = p.replace('BDT', '')
num = float(p)
product['price'] = num
img1 = response.selector.xpath(
'//a[@class="thumb-link"]/img/@src').extract()
img2 = response.selector.xpath(
'//img[@id="image-main"]/@src').extract()
product['images'] = img1 + img2
product['last_updated'] = datetime.datetime.now()
yield product
def parse_product_detail(self, response):
product = items.Ecommerce_product_items()
product['url'] = response.url
product['domain_name']="kartbd.com"
product['title'] = response.selector.xpath('//div[@class="col-md-5 col-sm-6 product-info"]/h2[@class="title"]/text()').extract_first()
a = response.selector.xpath('//nav[@class="breadcrumb"]/ul//li/a/text()').extract()
category_name = a[1:]
product['categories'] = category_name
product['currency'] ='BDT'
p = response.selector.xpath('//div[@class="col-xs-7"]/text()').extract_first()
#tk
#'\n tk1,490.00'
p = p.replace(' ','')
p = p.replace(',','')
p = p.replace('tk','') # where 'Tk.' is substring
p = p.replace('\n','')
num = float(p)
product['price'] = num
img = 'http://kartbd.com'
img1 = response.selector.xpath('//a[@class="f-box"]/@href').extract()
img1 = [img + ck for ck in img1]
img2 = response.selector.xpath('//div[@id="thumbRails"]//div/a/img/@src').extract()
img2 = [img + ck for ck in img2]
product['images'] = img1 + img2
product['last_updated'] = datetime.datetime.now()
yield product