def parse_entry(self, driver):
driver.implicitly_wait(15)
product_count = 0
while True:
elements = of_utils.find_elements_by_css_selector(
driver, 'div.productItemContainer > a')
if not elements:
elements = of_utils.find_elements_by_css_selector(
driver, 'li.productItemContainer > a')
if not elements:
elements = of_utils.find_elements_by_css_selector(
driver, 'li.productItem > a')
if not elements:
elements = of_utils.find_elements_by_css_selector(
driver, 'li.lookItem > a')
if not elements:
elements = of_utils.find_elements_by_css_selector(
driver, 'ul.lv-list-container a')
if len(elements) > product_count:
product_count = len(elements)
driver.execute_script(
'window.scrollBy(0, document.body.scrollHeight);')
of_utils.sleep(4)
else:
break
return [element.get_attribute('href').strip() for element in elements]
def parse_product(self, driver):
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(
driver, 'div.goods_about > p.about_tit')
if element:
product['title'] = element.text.strip().replace('\n', ' ')
else:
raise Exception('Title not found')
# code
element = of_utils.find_element_by_css_selector(
driver, 'p.bianhao > span')
if element:
product['code'] = element.text.strip()
# price_cny
element = of_utils.find_element_by_css_selector(
driver, 'p.about_cost > span')
if element:
price_text = element.text.strip()[1:].strip()
product['price_cny'] = int(float(price_text))
# images
elements = of_utils.find_elements_by_css_selector(
driver, 'ul.swiper-wrapper > li > img')
images = [
element.get_attribute('src').strip().replace('80X80', '540X540')
for element in elements
]
product['images'] = ';'.join(images)
# detail
elements = of_utils.find_elements_by_css_selector(
driver, 'div.word > div > p > span')
texts = [element.text.strip() for element in elements]
product['detail'] = '\n'.join(texts)
return product
def parse_product(self, driver):
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(
driver, 'p.tdr-watch-details__header-watch-name > span')
if element:
product['title'] = element.text.strip().replace('\n', '')
else:
raise Exception('Title not found')
# code
element = of_utils.find_element_by_css_selector(
driver, 'p.tdr-watch-details__header-watch-reference')
if element:
product['code'] = element.text.strip().split('：')[-1]
# price_cny N/A
# images
elements = of_utils.find_elements_by_css_selector(
driver, 'div.tdr-variations__main-image-canvas-wrapper > img')
images = [element.get_attribute('src').strip() for element in elements]
product['images'] = ';'.join(images)
# detail
texts = []
elements = of_utils.find_elements_by_css_selector(
driver,
'ul.tdr-watch-details__column > li > div.tdr-watch-details__text')
for element in elements:
k_element = of_utils.find_element_by_css_selector(
element, 'p.tdr-watch-details__title')
v_element = of_utils.find_element_by_css_selector(
element, 'p.tdr-watch-details__spectext')
texts.append(k_element.text.strip() + '：' + v_element.text.strip())
product['detail'] = '\n'.join(texts)
return product
def parse_product(self, driver):
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(driver, 'span.product.attribute.name')
if element:
product['title'] = element.text.strip()
else:
raise Exception('Title not found')
# code
element = of_utils.find_element_by_css_selector(driver, 'span.value[itemprop=sku]')
if element:
product['code'] = element.text.strip()
# price_cny
element = of_utils.find_element_by_css_selector(driver, 'span.product-price-reveal__action__show')
if element:
driver.execute_script('arguments[0].click();', element)
of_utils.sleep(2)
element = of_utils.find_element_by_css_selector(driver, 'span.price')
if element:
price_text = element.text.strip()[1:].strip().replace(',', '') # 去掉开头的¥
product['price_cny'] = int(float(price_text))
# images
elements = of_utils.find_elements_by_css_selector(driver, 'div.fotorama__stage__shaft > div > img')
images = [element.get_attribute('src').strip() for element in elements]
product['images'] = ';'.join(images)
# detail
elements = of_utils.find_elements_by_css_selector(driver, 'div.product-info-details-content > div.value > p')
texts = [element.get_attribute('innerHTML').strip() for element in elements]
product['detail'] = '\n'.join(texts)
return product
def parse_entry(self, driver):
# 手袋
elements = of_utils.find_elements_by_css_selector(driver, 'div.fs-products-grid__product.fs-gridelement > div.fs-products-grid__product__illu > a')
if not elements:
# 手袋2
elements = of_utils.find_elements_by_css_selector(driver, 'div.fs-products-grid > div.fs-products-grid__product.fs-gridelement > div.fs-products-grid__product__wrapper > a')
if not elements:
# 彩妆
elements = of_utils.find_elements_by_css_selector(driver, 'div.fnb_col-wd6.fnb_product-img > a')
# 手表
if not elements:
load_more = of_utils.find_element_by_css_selector(driver, 'div.pd-action-btns > button[role=button]')
if not load_more: # 戒指
load_more = of_utils.find_element_by_css_selector(driver, 'div.display-all > a')
if load_more:
driver.execute_script('arguments[0].click();', load_more)
of_utils.sleep(5)
product_count = 0
while True:
elements = of_utils.find_elements_by_css_selector(driver, 'div.products > div.row > div > ul > li > div.product-item-wrapper > a')
if len(elements) > product_count:
product_count = len(elements)
driver.execute_script('window.scrollBy(0, document.body.scrollHeight);')
of_utils.sleep(4)
else:
break
return [element.get_attribute('href').strip() for element in elements]
def parse_product(self, driver):
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(
driver, 'div.frabic-detail-right > h3')
if element:
product['title'] = element.text.strip()
else:
raise Exception('Title not found')
# code + price_cny
elements = of_utils.find_elements_by_css_selector(
driver, 'div.detail-chose-box > div > strong')
if len(elements) >= 2:
product['code'] = elements[0].text.strip()
product['price_cny'] = int(float(elements[1].text.strip()))
# images
elements = of_utils.find_elements_by_css_selector(
driver, 'div.frabic-detail-left > img')
images = [element.get_attribute('src').strip() for element in elements]
product['images'] = ';'.join(images)
# detail
element = of_utils.find_element_by_css_selector(
driver, 'table.detail > tbody > tr > td > span.detail')
product['detail'] = element.text.strip()
return product
def parse_product(self, driver):
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(
driver, 'div.product-title.desktop > div.inner > h1[lang=en]')
if element:
product['title'] = element.text.strip().replace('\n', ' ')
else:
raise Exception('Title not found')
# code N/A
# price_cny N/A
# images
elements = of_utils.find_elements_by_css_selector(
driver, 'img.main-image')
images = [element.get_attribute('src').strip() for element in elements]
product['images'] = ';'.join(images)
# detail
texts = []
element = of_utils.find_element_by_css_selector(driver, 'div.readmore')
texts.append(element.text.strip())
elements = of_utils.find_elements_by_css_selector(
driver, 'div.list-details > div.row')
for element in elements:
spans = of_utils.find_elements_by_css_selector(element, 'span')
texts.append(spans[0].text.strip() + '：' + spans[1].text.strip())
product['detail'] = '\n'.join(texts)
return product
def parse_entry(self, driver):
elements = of_utils.find_elements_by_css_selector(
driver, 'div.product > div.product-image > a')
if not elements:
elements = of_utils.find_elements_by_css_selector(
driver, 'div.product > a.product-link')
return [element.get_attribute('href').strip() for element in elements]
def parse_product(self, driver):
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(
driver, 'div.product-title > h1')
if element:
product['title'] = element.text.strip()
else:
raise Exception('Title not found')
# code N/A
# price_cny
element = of_utils.find_element_by_css_selector(
driver, 'span.regular-price > span.price')
if element:
price_text = element.text.strip()[1:].strip().replace(',', '')
product['price_cny'] = int(float(price_text))
# images
elements = of_utils.find_elements_by_css_selector(
driver,
'div#product-page-top > div > div > div > div > div > div > img')
if not elements:
elements = of_utils.find_elements_by_css_selector(
driver, 'div.img-cont > img')
images = [element.get_attribute('src').strip() for element in elements]
product['images'] = ';'.join(images)
# detail
element = of_utils.find_element_by_css_selector(
driver, 'div#product-description')
product['detail'] = element.text.strip()
return product
def parse_product(self, driver):
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(
driver, 'div.right[align=left] > div')
if element:
texts = element.text.split('\n')
product['title'] = ' '.join(texts[:-1])
price_text = texts[-1].split('/')[0].split('$')[-1].strip()
product['price_hkd'] = int(float(price_text))
else:
raise Exception('Title not found')
# code N/A
# price_hkd Processed in title
# images
elements = of_utils.find_elements_by_css_selector(
driver, 'div.left[align=center] > div >img')
images = [element.get_attribute('src').strip() for element in elements]
product['images'] = ';'.join(images)
# detail
texts = []
elements = of_utils.find_elements_by_css_selector(
driver, 'table.table-productdetails > tbody > tr > td')
for element in elements:
text = element.text.strip()
if text:
texts.append(text)
product['detail'] = '\n'.join(texts)
return product
def parse_product(self, driver):
flag = int(driver.current_url.split('?')[-1])
product = of_spider.empty_product.copy()
#title
element = of_utils.find_elements_by_css_selector(
driver, '.product-list .name')[flag]
if element:
product['title'] = element.text.strip()
else:
raise Exception('Title not found')
# code N/A
# price_cny
element = of_utils.find_elements_by_css_selector(
driver, '.product-list .value')[flag]
if element:
product['price_cny'] = of_utils.convert_price(element.text.strip())
# images
images = []
element = of_utils.find_elements_by_css_selector(
driver, '.product-list .product-img img')[flag]
if element:
images.append(element.get_attribute('src'))
product['images'] = ';'.join(images)
# detail N/A
return product
def parse_product(self, driver):
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(
driver, 'div.line1 > h1')
if element:
product['title'] = element.text.strip()
else:
raise Exception('Title not found')
# code
element = of_utils.find_element_by_css_selector(
driver, 'div.reference > span')
if element:
product['code'] = element.text.strip().split(' ')[-1]
# price_cny
element = of_utils.find_element_by_css_selector(
driver, 'div#priceDisplayValue')
if element:
try:
price_text = element.text.strip()[1:].replace(",", '')
product['price_cny'] = int(float(price_text))
except:
pass
# images
elements = of_utils.find_elements_by_css_selector(
driver, 'div.callage > img.pngfix')
images = [element.get_attribute('src').strip() for element in elements]
product['images'] = ';'.join(images)
# detail
elements = of_utils.find_elements_by_css_selector(
driver, 'ul.listInfos.specifications > li')
texts = [element.text.strip() for element in elements]
product['detail'] = '\n'.join(texts).strip()
return product
def parse_product(self, driver):
driver.implicitly_wait(10)
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(driver, '.iwc-buying-options-title')
if element:
product['title'] = element.text.strip().replace('\n添加至我的愿望清单','')
else:
raise Exception('Title not found')
# code
element = of_utils.find_element_by_css_selector(driver, '.iwc-buying-options-reference')
if element:
product['code'] = element.text.strip()
product['title'] = product['title'] + ' ' + element.text.strip()
# price_cny
element = of_utils.find_element_by_css_selector(driver, '.iwc-buying-options-price')
if element:
product['price_cny'] = of_utils.convert_price(element.text.strip())
# images
elements = of_utils.find_elements_by_css_selector(driver, '.iwc-buying-option-thumbnails .iwc-watch-thumbnail-container:not(.slick-cloned) .iwc-watch-thumbnail')
images = ['https://www.iwc.cn'+ element.get_attribute('data-srcset').strip() for element in elements]
product['images'] = ';'.join(images)
# detail
elements = of_utils.find_elements_by_css_selector(driver, 'ul.detailList > li')
texts = [element.text.strip() for element in elements]
product['detail'] = '\n'.join(texts).strip()
return product
def parse_product(self, driver):
driver.implicitly_wait(10)
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(driver, 'h2.pageTitle.watch-title')
if element:
product['title'] = element.text.strip()
else:
raise Exception('Title not found')
# code
element = of_utils.find_element_by_css_selector(driver, 'li.variation-reference > span.val')
if element:
product['code'] = element.text.strip()
# price_cny
element = of_utils.find_element_by_css_selector(driver, 'li.price > span.val')
if element:
try:
price_text = element.text.strip()[1:].strip()
product['price_cny'] = int(float(price_text))
except:
pass
# images
elements = of_utils.find_elements_by_css_selector(driver, 'ul.frontBack > li > img')
images = [element.get_attribute('src').strip() for element in elements]
product['images'] = ';'.join(images)
# detail
elements = of_utils.find_elements_by_css_selector(driver, 'ul.detailList > li')
texts = [element.text.strip() for element in elements]
product['detail'] = '\n'.join(texts).strip()
return product
def parse_product(self, driver):
of_utils.sleep(2)
product = of_spider.empty_product.copy()
ele = of_utils.find_element_by_css_selector(driver, 'div.product-name>h1')
product['title'] = ele.text.strip() if ele else ''
ele = of_utils.find_element_by_css_selector(driver, 'div[itemprop=sku]')
product['code'] = ele.text.strip() if ele else ''
ele = of_utils.find_element_by_css_selector(driver, 'meta[itemprop=price]')
product['price_cny'] = of_utils.convert_price(ele.get_attribute('content').strip()) if ele else 0
# images
imgs=[]
eles = of_utils.find_elements_by_css_selector(driver, '.fotorama__thumb>img')
if not eles:
eles = of_utils.find_elements_by_css_selector(driver, '.fotorama__stage__frame>img')
for ele in eles:
img = ele.get_attribute('src').strip().replace('cache', '')
for a in img.split('/'):
if len(a) == 32:
img = img.replace(a,'')
imgs.append(img)
product['images'] = ';'.join(imgs)
return product
def parse_product(self, driver):
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(
driver, 'h1.prd-ProductContent_Title')
if element:
product['title'] = element.text.strip()
else:
raise Exception('Title not found')
# code N/A
# price_cny
element = of_utils.find_element_by_css_selector(
driver, '.prd-ProductPrice_Price>span')
if element:
product['price_gbp'] = element.text.strip()[1:]
# images
elements = of_utils.find_elements_by_css_selector(
driver, '.prd-ProductImage_Thumbs a.prd-ProductImage_Link')
if elements:
images = [
element.get_attribute('href').strip() for element in elements
]
product['images'] = ';'.join(images)
else:
elements = of_utils.find_elements_by_css_selector(
driver, '.prd-ProductImage img')
images = [
element.get_attribute('src').strip() for element in elements
]
product['images'] = ';'.join(images)
# detail N/A
return product
def parse_product(self, driver):
product = of_spider.empty_product.copy()
# title
elements = of_utils.find_elements_by_css_selector(
driver, '.item-name-line')
txts = [element.text.strip() for element in elements]
if txts:
product['title'] = ' '.join(txts)
else:
raise Exception('Title not found')
# code N/A
# price_cny
element = of_utils.find_element_by_css_selector(
driver, '.item-price .itemPrice .price .value')
if element:
product['price_hkd'] = element.text.strip()
# images
elements = of_utils.find_elements_by_css_selector(
driver, '.item-zoom-images>ul.alternativeImages>li>img')
images = [element.get_attribute('src').strip() for element in elements]
product['images'] = ';'.join(images)
# detail
element = of_utils.find_element_by_css_selector(
driver, '.editorialdescription>span.value')
if element:
product['detail'] = element.text.strip()
return product
def parse_product(self, driver):
driver.implicitly_wait(10)
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(
driver, 'h1.hidden-xs[itemprop=name]')
if element:
product['title'] = element.text.strip()
else:
raise Exception('Title not found')
# code
element = of_utils.find_element_by_css_selector(
driver, 'p.reference > span.ezstring-field')
if element:
product['code'] = element.text.strip()
# price_cny N/A
# images
elements = of_utils.find_elements_by_css_selector(
driver, 'div.watch > img[itemrprop=image]')
images = [element.get_attribute('src').strip() for element in elements]
product['images'] = ';'.join(images)
# detail
texts = []
elements = of_utils.find_elements_by_css_selector(
driver, 'div.details > div > div.ezxmltext-field > p')
for element in elements:
text = element.text.strip()
if text:
texts.append(text)
product['detail'] = '\n'.join(texts)
return product
def parse_entry(self, driver):
# 手袋
elements = of_utils.find_elements_by_css_selector(driver, 'div.fs-products-grid__product.fs-gridelement > div.fs-products-grid__product__illu > a')
if not elements:
# 彩妆
elements = of_utils.find_elements_by_css_selector(driver, 'div.fnb_col-wd6.fnb_product-img > a')
return [element.get_attribute('href').strip() for element in elements]
def parse_product(self, driver):
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(
driver, 'h1.product-info-h1')
if element:
product['title'] = element.text.strip()
else:
raise Exception('Title not found')
# code N/A
# price_cny
element = of_utils.find_element_by_css_selector(
driver, '.product-info .price')
if element:
product['price_cny'] = of_utils.convert_price(
element.text.strip().replace('ml', ''))
# images
elements = of_utils.find_elements_by_css_selector(
driver, '.product-pic .item img')
if not elements:
elements = of_utils.find_elements_by_css_selector(
driver, '.productpage-images .productpage-image img')
images = [element.get_attribute('src').strip() for element in elements]
product['images'] = ';'.join(images)
# detail
element = of_utils.find_element_by_css_selector(
driver, '.tab-product-info')
if element:
product['detail'] = element.text.strip()
return product
def parse_product(self, driver):
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(
driver, 'h1.entry-title')
if element:
product['title'] = element.text.strip()
else:
raise Exception('Title not found')
# code
element = of_utils.find_element_by_css_selector(
driver, 'div.container > div.pdp-name > p.pdp-sku')
if element:
product['code'] = element.text.strip()
# price_cny
element = of_utils.find_element_by_css_selector(
driver, 'div.container > div.pdp-name > p.pdp-price')
if element:
price_text = element.get_attribute(
'innerHTML').strip()[1:].strip().replace(',', '') # 去掉开头的¥
product['price_cny'] = int(float(price_text))
# images
elements = of_utils.find_elements_by_css_selector(
driver, 'div.stiky-style-images > a.inventoryVariant > img')
images = [element.get_attribute('src').strip() for element in elements]
product['images'] = ';'.join(images)
# detail
texts = []
elements = of_utils.find_elements_by_css_selector(
driver, 'div.pdp-tab-longdesc > ul > li')
for element in elements:
texts.append(element.text.strip())
product['detail'] = '\n'.join(texts)
return product
def parse_product(self, driver):
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(
driver, 'p.product-title')
if element:
product['title'] = element.text.strip()
else:
raise Exception('Title not found')
# code N/A
# price_cny N/A
# images
elements = of_utils.find_elements_by_css_selector(
driver, 'ul > li.slide > a > img')
images = [element.get_attribute('src').strip() for element in elements]
product['images'] = ';'.join(images)
# detail
texts = []
element = of_utils.find_element_by_css_selector(
driver, 'p.product-information')
texts.append(element.text.strip())
elements = of_utils.find_elements_by_css_selector(
driver, 'table.gridtable > tbody > tr')
for element in elements:
_elements = of_utils.find_elements_by_css_selector(element, 'td')
texts.append(_elements[0].text.strip() + _elements[1].text.strip())
product['detail'] = '\n'.join(texts)
return product
def parse_product(self, driver):
of_utils.sleep(5)
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(driver, 'div.breadcrumbLeaf > p.attributesUpdater.Title > span.value')
if element:
product['title'] = element.text.strip()
else:
raise Exception('Title not found')
# code N/A
# price_cny N/A
# images
elements = of_utils.find_elements_by_css_selector(driver, 'div.item-alternativeImages-shots > ul > li > img')
print(elements)
images = []
for element in elements:
image = element.get_attribute('data-origin').strip().replace('_10_', '_20_')
images.append(image)
product['images'] = ';'.join(images)
# detail
element = of_utils.find_element_by_css_selector(driver, 'span.modelName')
product['code'] = element.get_attribute('innerHTML').strip() # Code here...
btn = of_utils.find_element_by_css_selector(driver, 'ul.itemDetails-info-accordion > li > h2 > div.plusIcon')
driver.execute_script('arguments[0].click();', btn)
elements = of_utils.find_elements_by_css_selector(driver, 'div.itemdescription > ul > li')
texts = [element.get_attribute('innerHTML').strip() for element in elements]
product['detail'] = '\n'.join(texts)
return product
def parse_product(self, driver):
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(
driver, 'div.product-card > span')
if element:
product['title'] = element.text.strip()
else:
raise Exception('Title not found')
# code
element = of_utils.find_element_by_css_selector(
driver, 'span.reference-jewelry')
if element:
product['code'] = element.text.strip()
# price_cny N/A
# images
elements = of_utils.find_elements_by_css_selector(
driver, 'div.content > img.carousel-slide__media')
images = [element.get_attribute('src').strip() for element in elements]
product['images'] = ';'.join(images)
# detail
texts = []
elements = of_utils.find_elements_by_css_selector(
driver, 'ul.fiche-details__left > li')
for element in elements:
k_element = of_utils.find_element_by_css_selector(element, 'span')
v_element = of_utils.find_element_by_css_selector(element, 'p')
txt = k_element.text.strip() + '：' + v_element.text.strip(
) if k_element else v_element.text.strip()
texts.append(txt)
product['detail'] = '\n'.join(texts)
return product
def parse_product(self, driver):
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(driver, 'div.product-detail h1.product-name')
if element:
product['title'] = element.text.strip()
else:
raise Exception('Title not found')
# code
element = of_utils.find_element_by_css_selector(driver, '.giv-ProductContent-productNumber>span')
if element:
product['code'] = element.text.strip()
# price_cny
element = of_utils.find_element_by_css_selector(driver, 'div.product-price > span.price-sales')
if element:
price_text = element.text.strip()[1:].strip().replace(',', '') # 去掉开头的¥
product['price_cny'] = int(float(price_text))
# images
elements = of_utils.find_elements_by_css_selector(driver, 'div.swiper-slide > a.fullscreen > picture > img')
if not elements:
elements = of_utils.find_elements_by_css_selector(driver, '.giv-ProductImageContainer .slick-list .slick-track a picture source')
images = [element.get_attribute('srcset').strip() for element in elements]
product['images'] = ';'.join(images)
# detail N/A
return product
def parse_product(self, driver):
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(
driver, 'div.p-name > h3.subtitle')
if element:
product['title'] = element.text.strip()
else:
raise Exception('Title not found')
# code N/A
# price_cny
element = of_utils.find_element_by_css_selector(
driver, 'div.p-name > div.price')
if element:
price_text = element.text.strip().split(' ')[1].split(
'/')[0].strip()
product['price_cny'] = int(price_text)
# images
elements = of_utils.find_elements_by_css_selector(
driver, 'ul.swiper-wrapper > li.swiper-slide > img')
images = [element.get_attribute('src').strip() for element in elements]
product['images'] = ';'.join(images)
# detail
element = of_utils.find_element_by_css_selector(
driver, 'div.product-point > div#p-intro > div.des > p')
if element:
product['detail'] = element.text.strip()
else:
elements = of_utils.find_elements_by_css_selector(
driver, 'div.product-point > div#p-intro > div.des > div')
texts = [element.text.strip() for element in elements]
product['detail'] = '\n'.join(texts)
return product
def parse_product(self, driver):
elements = of_utils.find_elements_by_css_selector(
driver, '.product_list .product_1 a')
flag = int(driver.current_url.split('?')[-1])
element = elements[flag]
driver.execute_script('arguments[0].click();', element)
of_utils.sleep(2)
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(
driver,
'.popup_product_%s .product_detail_content .jspPane h2' % flag)
if element:
product['title'] = element.text.strip()
else:
raise Exception('Title not found')
# code N/A
# price_cny
element = of_utils.find_element_by_css_selector(
driver, '.popup_product_%s .product_photo h3' % flag)
if element:
product['price_hkd'] = element.text.strip().split(
'/')[0].strip().replace('$', '').replace(',', '')
# images
elements = of_utils.find_elements_by_css_selector(
driver, '.popup_product_%s .product_photo img' % flag)
images = [element.get_attribute('src').strip() for element in elements]
product['images'] = ';'.join(images)
# detail N/A
return product
def parse_product(self, driver):
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(
driver, '#pdpMain h1.product_subtitle')
if element:
product['title'] = element.text.strip()
else:
raise Exception('Title not found')
# code N/A
# price_cny
element = of_utils.find_element_by_css_selector(
driver, '.product_add_to_cart .product_price')
if element:
product['price_cny'] = of_utils.convert_price(
element.get_attribute('data-pricevalue'))
# images
elements = of_utils.find_elements_by_css_selector(
driver, '.product_thumbnails ul li.thumb a')
if not elements:
elements = of_utils.find_elements_by_css_selector(
driver, '.product_primary_image a')
images = [
element.get_attribute('href').strip() for element in elements
]
product['images'] = ';'.join(images)
# detail
element = of_utils.find_element_by_css_selector(
driver, '#pdpMain .product_description_box>div>span')
if element:
product['detail'] = element.text.strip()
return product
def parse_entry(self, driver):
btn = of_utils.find_element_by_css_selector(driver, 'div.spice-looks-grid-button > a')
if btn:
driver.execute_script('arguments[0].click();', btn) # 点击“浏览所有”
elements = of_utils.find_elements_by_css_selector(driver, 'ul.spice-float-clearfix > li > div > div > a.spice-item-grid')
if not elements:
elements = of_utils.find_elements_by_css_selector(driver, 'div#pdlist > div.grid-cell > div.product-tiles-box > a.spice-item-grid')
return [element.get_attribute('href').strip() for element in elements]
def parse_product(self, driver):
product = of_spider.empty_product.copy()
# title
element = of_utils.find_element_by_css_selector(
driver, '.product-name h1.title')
if element:
product['title'] = element.text.strip()
else:
raise Exception('Title not found')
# code
element = of_utils.find_element_by_css_selector(driver, 'footer > p')
if element:
product['code'] = element.text.split(' ')[1].strip()
if '中国官网' in product['title']:
title = ''
element = of_utils.find_element_by_css_selector(driver, 'h1.title')
if not element:
raise Exception('Title not found')
title = element.text.strip() + '-'
element = of_utils.find_element_by_css_selector(
driver, 'div.product-sku')
if element:
title += element.text.strip()
product['title'] = title
# price_cny
element = of_utils.find_element_by_css_selector(
driver, 'div.prices > span.price > span > span.price')
if not element:
element = of_utils.find_element_by_css_selector(
driver, 'div.prices > span.price > p.old-price > span')
if not element:
element = of_utils.find_element_by_css_selector(
driver, 'div.product-price > span.price > span > span.price')
if element:
price_text = element.get_attribute('innerHTML')
price_text = price_text.split(';')[1].strip().replace(',', '')
product['price_cny'] = int(float(price_text))
# images
elements = of_utils.find_elements_by_css_selector(
driver, 'ul.images > li > a')
images = [
element.get_attribute('data-zoom-image').strip()
for element in elements
]
if not images:
elements = of_utils.find_elements_by_css_selector(
driver,
'div.product-gallery > div.gallery > div.gallery-item > a')
images = [
element.get_attribute('href').strip() for element in elements
]
product['images'] = ';'.join(images)
# detail
element = of_utils.find_element_by_css_selector(
driver, 'div#product-info')
if element:
product['detail'] = element.text.strip()
return product
