def set_product_data(self , page_url, soup, product_ctx ) :
#
#
try :
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
#self.set_product_category_first(product_data, soup)
self.set_product_category_second(page_url, product_data, soup)
###########################
# 상품 이미지 확인
#
###########################
self.set_product_image_second( product_data, product_ctx )
# 품절여부 확인
#
self.set_product_soldout_first(product_data, product_ctx )
###########################
# 상품명/URL
###########################
crw_post_url = self.set_product_name_url_fourth( product_data, product_ctx , 'p', 'name')
if(crw_post_url == '') : crw_post_url = self.set_product_name_url_fourth( product_data, product_ctx , 'strong', 'name')
##############################
# 가격
# <ul class="xans-element- xans-product xans-product-listitem"><li class=" xans-record-">
# <strong class="title displaynone"><span style="font-size:12px;color:#555555;font-weight:bold;"></span> :</strong><span style="font-size:12px;color:#555555;font-weight:bold;">₩15,000</span><span id="span_product_tax_type_text" style=""> </span></li>
# </ul>
##############################
li_list = product_ctx.find_all('li')
li_num = 0
for li_ctx in li_list :
li_num += 1
value_str = li_ctx.get_text().strip()
split_list = value_str.split(':')
sub_split_list = split_list[1].split('(')
price_str = sub_split_list[0].strip()
if( li_num == 1) : product_data.crw_price = int( __UTIL__.get_only_digit( price_str ) )
elif( li_num == 2) : product_data.crw_price_sale = int( __UTIL__.get_only_digit( price_str ))
if( crw_post_url != '' ) :
self.set_product_url_hash( product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self, page_url, soup, product_ctx):
#
#
try:
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
#self.set_product_category_first(product_data, soup)
self.set_product_category_second(page_url, product_data, soup)
###########################
# 상품 이미지 확인
#
###########################
self.set_product_image_second(product_data, product_ctx)
# 품절여부 확인
#
#<div class="status">
#<div class="icon"> <img src="/web/upload/custom_6.gif" alt=""></div>
#</div>
#self.set_product_soldout_first(product_data, product_ctx )
soldout_div_list = product_ctx.find_all(
self.C_PRODUCT_SOLDOUT_SELECTOR,
class_=self.C_PRODUCT_SOLDOUT_SELECTOR_CLASSNAME)
for soldout_div_ctx in soldout_div_list:
img_list = soldout_div_ctx.find_all('img')
for img_ctx in img_list:
if ('src' in img_ctx.attrs):
if (0 <= img_ctx.attrs['src'].find('custom_6.gif')):
product_data.crw_is_soldout = 1
###########################
# 상품명/URL
###########################
crw_post_url = self.set_product_name_url_first(
product_data, product_ctx, 'p', 'name')
if (crw_post_url == ''):
crw_post_url = self.set_product_name_url_first(
product_data, product_ctx, 'strong', 'name')
##############################
# 가격
##############################
self.set_product_price_brand_first(product_data, product_ctx)
if (crw_post_url != ''):
self.set_product_url_hash(product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self , page_url, soup, product_ctx ) :
#
#
try :
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
self.set_product_category_first(product_data, soup)
###########################
# 상품 이미지 확인
#
# <img src="//ai-doggi.com/web/product/medium/20191220/a8ebb002293a954628763cf4a9ab6c38.jpg" alt="" class="thumb">
###########################
self.set_product_image_second( product_data, product_ctx )
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx )
###########################
#
# <p class="name">
# <a href="/product/detail.html?product_no=286&cate_no=43&display_group=1"><strong class="title displaynone"><span style="font-size:12px;color:#555555;">상품명</span> :</strong> <span style="font-size:12px;color:#555555;">Frill Neck Sleeve Blouse Lavender [20%SALE]</span></a>
# </p>
###########################
crw_post_url = self.set_product_name_url_second( product_data, product_ctx , 'div', '-name')
############################
# 가격 및 브랜드
#
#<div class="xans-element- xans-product xans-product-listitem -description"><div rel="판매가" class=" xans-record-">
#<span class="title displaynone"><span style="font-size:12px;color:#333333;font-weight:bold;">판매가</span> :</span> <span style="font-size:12px;color:#333333;font-weight:bold;">39,000원</span><span id="span_product_tax_type_text" style=""> </span></div>
#</div>
#
############################
price_div_list = product_ctx.find_all('div', {'rel':'판매가'})
for price_div_ctx in price_div_list :
span_list = price_div_ctx.find_all('span')
for span_ctx in span_list :
value_str = span_ctx.get_text().strip()
if(value_str != '') and (value_str.find('판매가') < 0) and (value_str.find(':') < 0) :
product_data.crw_price_sale = int( __UTIL__.get_only_digit( value_str ))
if( crw_post_url != '' ) :
self.set_product_url_hash( product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self, page_url, soup, product_ctx):
#
#
try:
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
#self.set_product_category_first(product_data, soup)
self.set_product_category_second(page_url, product_data, soup)
###########################
# 상품 이미지 확인
#
###########################
self.set_product_image_second(product_data, product_ctx)
# 품절여부 확인
#
# 장바구니 이미지가 없으면 품절
# <img src="/web/upload/icon_201905151555185500.png" onclick="CAPP_SHOP_NEW_PRODUCT_OPTIONSELECT.selectOptionCommon(856, 163, 'basket', '')" alt="장바구니 담기" class="ec-admin-icon cart">
soldout_img_ctx = product_ctx.find('img',
class_='ec-admin-icon cart')
if (soldout_img_ctx == None): product_data.crw_is_soldout = 1
###########################
# 상품명/URL
###########################
crw_post_url = self.set_product_name_url_fourth(
product_data, product_ctx, 'p', 'name')
if (crw_post_url == ''):
crw_post_url = self.set_product_name_url_fourth(
product_data, product_ctx, 'strong', 'name')
##############################
# 가격
##############################
price_ctx = product_ctx.find('p', class_='prices')
if (price_ctx != None):
span_ctx = price_ctx.find('span', class_='price normal')
if (span_ctx != None):
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(span_ctx.get_text().strip()))
if (crw_post_url != ''):
self.set_product_url_hash(product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self, page_url, soup, product_ctx):
#
#
try:
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
#self.set_product_category_third(product_data, soup)
#self.set_product_category_second(page_url, product_data, soup)
title_list = soup.select(
'#-common > div > div > div.xans-element-.xans-product.xans-product-menupackage > div.xans-element-.xans-product.xans-product-headcategory.titleArea.sub'
)
for title_ctx in title_list:
h2_ctx = title_ctx.find('h2')
if (h2_ctx != None):
product_data.crw_category1 = h2_ctx.get_text().strip()
break
###########################
# 상품 이미지 확인
#
###########################
self.set_product_image_fourth(product_data, product_ctx)
# 품절여부 확인
#
#self.set_product_soldout_first(product_data, product_ctx )
sold_out_ctx = product_ctx.find('img', class_='ec-admin-icon cart')
if (sold_out_ctx == None): product_data.crw_is_soldout = 1
###########################
# 상품명/URL
###########################
crw_post_url = self.set_product_name_url_first(
product_data, product_ctx, 'div', 'name')
##############################
# 가격
#
##############################
self.set_product_price_brand_first(product_data, product_ctx)
if (crw_post_url != ''):
self.set_product_url_hash(product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self, page_url, soup, product_ctx):
#
try:
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
#self.set_product_category_first(product_data, soup)
self.set_product_category_second(page_url, product_data, soup)
# 상품 이미지 확인
self.set_product_image_second(product_data, product_ctx)
product_link_list = product_ctx.find_all('a')
for product_link_ctx in product_link_list:
if ('href' in product_link_ctx.attrs):
tmp_product_link = product_link_ctx.attrs['href'].strip()
if (0 != tmp_product_link.find('http')):
tmp_product_link = '%s%s' % (
self.BASIC_PRODUCT_URL,
product_link_ctx.attrs['href'].strip())
crw_post_url = tmp_product_link
if (self.C_PRODUCT_STRIP_STR != ''):
crw_post_url = tmp_product_link.replace(
self.C_PRODUCT_STRIP_STR, '')
split_list = crw_post_url.split('?product_no=')
crw_goods_code_list = split_list[1].strip().split('&')
product_data.crw_goods_code = crw_goods_code_list[0].strip(
)
name_div_list = product_ctx.find_all('div', class_='name')
for name_div_ctx in name_div_list:
#
# 상품 링크 정보 및 상품명 / 상품코드
product_data.crw_name = name_div_ctx.get_text().strip()
self.set_product_price_brand_first(product_data, product_ctx)
if (crw_post_url != ''):
self.set_product_url_hash(product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self, page_url, soup, product_ctx):
#
#
try:
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
self.set_product_category_first(product_data, soup)
#product_data.crw_category1 = self.PAGE_URL_HASH[page_url]
###########################
# 상품 이미지 확인
#
###########################
self.set_product_image_third(product_data, product_ctx)
# 품절여부 확인
#
self.set_product_soldout_first(product_data, product_ctx)
###########################
# 상품명/URL
###########################
crw_post_url = self.set_product_name_url_fourth(
product_data, product_ctx, 'div', 'name')
##############################
# 가격
#
# <div>
# <p><span class="info displaynone"> / </span><span class="price normal">19,900 won</span><span class="price normal displaynone"></span></p>
# <p class="icon"></p>
# </div>
##############################
#self.set_product_price_brand_first(product_data, product_ctx)
price_ctx = product_ctx.find('span', class_='price normal')
if (price_ctx != None):
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(price_ctx.get_text().strip()))
if (crw_post_url != ''):
self.set_product_url_hash(product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self , page_url, soup, product_ctx ) :
#
#
try :
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
#product_data.crw_category1 = self.PAGE_URL_HASH[page_url]
#self.set_product_category_third(product_data, soup)
self.set_product_category_second(page_url, product_data, soup)
sub_cate = product_data.crw_category1
if(self.TOP_CATEGORY_NAME.get(sub_cate, -1) != -1) :
product_data.crw_category3 = product_data.crw_category2
product_data.crw_category2 = product_data.crw_category1
product_data.crw_category1 = self.TOP_CATEGORY_NAME[sub_cate]
###########################
# 상품 이미지 확인
###########################
self.set_product_image_second( product_data, product_ctx )
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx )
###########################
# 상품명/URL
###########################
crw_post_url = self.set_product_name_url_second( product_data, product_ctx , 'strong', 'name')
if(crw_post_url == '') : crw_post_url = self.set_product_name_url_second( product_data, product_ctx , 'p', 'name')
##############################
# 가격
##############################
self.set_product_price_brand_second(product_data, product_ctx)
if( crw_post_url != '' ) :
self.set_product_url_hash( product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self, page_url, soup, product_ctx):
#
#
try:
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
self.set_product_category_first(product_data, soup)
#self.set_product_category_second(page_url, product_data, soup)
###########################
# 상품 이미지 확인
###########################
self.set_product_image_third(product_data, product_ctx)
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx)
###########################
# 상품명/URL
###########################
crw_post_url = self.set_product_name_url_second(
product_data, product_ctx, 'strong', 'name')
if (crw_post_url == ''):
crw_post_url = self.set_product_name_url_second(
product_data, product_ctx, 'p', 'name')
##############################
# 가격
#
##############################
li_list = product_ctx.find_all('li')
for li_ctx in li_list:
span_ctx = li_ctx.find_all('span')
if (1 < len(span_ctx)):
split_list = span_ctx[0].get_text().strip().split('(')
value_str = split_list[0].strip()
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(value_str))
if (crw_post_url != ''):
self.set_product_url_hash(product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self, page_url, soup, product_ctx):
#
#
try:
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
#self.set_product_category_first(product_data, soup)
self.set_product_category_second(page_url, product_data, soup)
###########################
# 상품 이미지 확인
#
###########################
self.set_product_image_fourth(product_data, product_ctx)
# 품절여부 확인
#
# <div class="sold"><img src="/web/upload/icon_201709051101436600.gif" class="icon_img" alt="" title=""></div>
soldout_div_ctx = product_ctx.find('div', class_='sold')
if (soldout_div_ctx != None):
soldout_img_ctx = soldout_div_ctx.find('img')
if (soldout_img_ctx != None): product_data.crw_is_soldout = 1
###########################
# 상품명/URL
###########################
crw_post_url = self.set_product_name_url_second(
product_data, product_ctx, 'strong', 'name')
if (crw_post_url == ''):
crw_post_url = self.set_product_name_url_second(
product_data, product_ctx, 'p', 'name')
##############################
# 가격
#
##############################
self.set_product_price_brand_second(product_data, product_ctx)
if (crw_post_url != ''):
self.set_product_url_hash(product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self , page_url, soup, product_ctx ) :
#
#
try :
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
self.set_product_category_first(product_data, soup)
###########################
# 상품 이미지 확인
#
###########################
self.set_product_image_fourth( product_data, product_ctx )
# 품절여부 확인
#self.set_product_soldout_first(product_data, product_ctx )
###########################
#
###########################
crw_post_url = self.set_product_name_url_first( product_data, product_ctx , 'strong', 'name')
if(crw_post_url == '') : crw_post_url = self.set_product_name_url_first( product_data, product_ctx , 'p', 'name')
# 품절여부 확인
# 상품명에 품절이라고 표시함
# <strong class="name"><a href="/product/detail.html?product_no=154&cate_no=54&display_group=1" class=""><span class="title displaynone"><span style="font-size:13px;color:#1c1c1c;font-weight:bold;">상품명</span> :</span> <span style="font-size:13px;color:#1c1c1c;font-weight:bold;">(품절) Hanbok Embroidery Cape - Navy</span></a></strong>
if(0 <= product_data.crw_name.find('(품절)') ) : product_data.crw_is_soldout = 1
##############################
#
##############################
self.set_product_price_brand_first(product_data, product_ctx)
if( crw_post_url != '' ) :
self.set_product_url_hash( product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self , page_url, soup, product_ctx ) :
#
#
try :
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
#self.set_product_category_first(product_data, soup)
self.set_product_category_second(page_url, product_data, soup)
###########################
# 상품 이미지 확인
#
###########################
self.set_product_image_second( product_data, product_ctx )
# 품절여부 확인
#
#self.set_product_soldout_first(product_data, product_ctx )
###########################
# 상품명/URL
###########################
crw_post_url = self.set_product_name_url_second( product_data, product_ctx , 'strong', 'name')
if(crw_post_url == '') : crw_post_url = self.set_product_name_url_second( product_data, product_ctx , 'p', 'name')
if( 0 < product_data.crw_name.find('[품절]') ) or ( 0 < product_data.crw_name.find('-품절-') ) : product_data.crw_is_soldout = 1
##############################
# 가격
#
##############################
self.set_product_price_brand_second(product_data, product_ctx)
if( crw_post_url != '' ) :
self.set_product_url_hash( product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self, page_url, soup, product_ctx):
#
#
try:
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
self.set_product_category_first(product_data, soup)
###########################
# 상품 이미지 확인
#
###########################
self.set_product_image_second(product_data, product_ctx)
# 품절여부 확인
#
self.set_product_soldout_first(product_data, product_ctx)
###########################
# 상품명/URL
###########################
crw_post_url = self.set_product_name_url_fourth(
product_data, product_ctx, 'p', 'name')
if (crw_post_url == ''):
crw_post_url = self.set_product_name_url_fourth(
product_data, product_ctx, 'strong', 'name')
##############################
# 가격
#
# <p class="price">KRW 46,000</p>
##############################
price_ctx = product_ctx.find('p', class_='price')
if (price_ctx != None):
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(price_ctx.get_text().strip()))
if (crw_post_url != ''):
self.set_product_url_hash(product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self, page_url, soup, product_ctx):
#
#
try:
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
split_list = self.PAGE_URL_HASH[page_url].split('|')
idx = 0
for split_data in split_list:
idx += 1
if (idx == 1): product_data.crw_category1 = split_data
elif (idx == 2): product_data.crw_category2 = split_data
elif (idx == 3): product_data.crw_category3 = split_data
# self.set_product_category_third(product_data, soup)
# self.set_product_category_second(page_url, product_data, soup)
###########################
# 상품 이미지 확인
###########################
self.set_product_image_fourth(product_data, product_ctx)
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx)
###########################
# 상품명/URL
###########################
crw_post_url = self.set_product_name_url_second(
product_data, product_ctx, 'strong', 'name')
##############################
# 가격
##############################
self.set_product_price_brand_first(product_data, product_ctx)
if (crw_post_url != ''):
self.set_product_url_hash(product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self , page_url, soup, product_ctx ) :
#
#
try :
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
self.set_product_category_third(product_data, soup)
###########################
# 상품 이미지 확인
#
###########################
self.set_product_image_fourth( product_data, product_ctx )
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx )
###########################
#
# <p class="name">
# <a href="/product/detail.html?product_no=286&cate_no=43&display_group=1"><strong class="title displaynone"><span style="font-size:12px;color:#555555;">상품명</span> :</strong> <span style="font-size:12px;color:#555555;">Frill Neck Sleeve Blouse Lavender [20%SALE]</span></a>
# </p>
###########################
crw_post_url = self.set_product_name_url_second( product_data, product_ctx , 'strong', 'name')
if(crw_post_url == '') : crw_post_url = self.set_product_name_url_second( product_data, product_ctx , 'p', 'name')
##############################
#
#<p><strike class="displaynone"></strike><strong class="price">25,000원</strong></p>
##############################
strong_ctx = product_ctx.find('strong', class_='price')
if(strong_ctx != None) : product_data.crw_price_sale = int( __UTIL__.get_only_digit( strong_ctx.get_text().strip() ))
if( crw_post_url != '' ) :
self.set_product_url_hash( product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self, page_url, soup, product_ctx):
#
#
try:
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
#self.set_product_category_first(product_data, soup)
self.set_product_category_third(product_data, soup)
###########################
# 상품 이미지 확인
#
# <a name="anchorBoxName_741" href="/product/detail.html?product_no=741&cate_no=42&display_group=1" class="prdImg"><img src="//babiana.co.kr/web/product/medium/201907/23d619d612a7e377f9f6eb3a8ffd193a.jpg" style="border: 1px solid transparent;" class="borderEffect" onmouseover="this.style.border='1px solid #8bc1c6'" onmouseout="this.style.border='1px solid transparent'"></a>
###########################
self.set_product_image_first(product_data, product_ctx)
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx)
###########################
#
# <p class="name">
# <a href="/product/detail.html?product_no=286&cate_no=43&display_group=1"><strong class="title displaynone"><span style="font-size:12px;color:#555555;">상품명</span> :</strong> <span style="font-size:12px;color:#555555;">Frill Neck Sleeve Blouse Lavender [20%SALE]</span></a>
# </p>
###########################
crw_post_url = self.set_product_name_url_first(
product_data, product_ctx, 'p', 'name')
price_ctx = product_ctx.find('p', class_='price')
if (price_ctx != None):
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(price_ctx.get_text().strip()))
if (crw_post_url != ''):
self.set_product_url_hash(product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self, page_url, soup, product_ctx):
#
#
try:
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
# self.set_product_category_first(product_data, soup)
product_data.crw_category1 = self.PAGE_URL_HASH[page_url]
###########################
# 상품 이미지 확인
#
###########################
self.set_product_image_second(product_data, product_ctx)
# 품절여부 확인
#
self.set_product_soldout_first(product_data, product_ctx)
###########################
# 상품명/URL
###########################
crw_post_url = self.set_product_name_url_first(
product_data, product_ctx, 'p', 'name')
if (crw_post_url == ''):
crw_post_url = self.set_product_name_url_first(
product_data, product_ctx, 'strong', 'name')
##############################
# 가격
##############################
self.set_product_price_brand_first(product_data, product_ctx)
if (crw_post_url != ''):
self.set_product_url_hash(product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self , page_url, soup, product_ctx ) :
#
#
try :
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
self.set_product_category_first(product_data, soup)
# 상품 이미지 확인
self.set_product_image_fourth(product_data, product_ctx )
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx )
crw_post_url = self.set_product_name_url_second( product_data, product_ctx , 'div', 'description')
self.set_product_price_brand_first(product_data, product_ctx )
if( crw_post_url != '' ) :
self.set_product_url_hash( product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self, page_url, soup, product_ctx):
#
#
try:
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
#self.set_product_category_first(product_data, soup)
#__LOG__.Trace( self.PAGE_URL_HASH[page_url] )
self.set_product_category_third(product_data, soup)
split_list = self.PAGE_URL_HASH[page_url].split('|')
idx = 0
for split_data in split_list:
idx += 1
if (idx == 1): product_data.crw_category2 = split_data.strip()
elif (idx == 2):
product_data.crw_category3 = split_data.strip()
# 상품 이미지 확인
self.set_product_image_fourth(product_data, product_ctx)
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx)
crw_post_url = self.set_product_name_url_second(
product_data, product_ctx, 'div', 'description')
self.set_product_price_brand_second(product_data, product_ctx)
if (crw_post_url != ''):
self.set_product_url_hash(product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self , page_url, soup, product_ctx ) :
#
#
try :
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
#self.set_product_category_first(product_data, soup)
self.set_product_category_second(page_url, product_data, soup)
###########################
# 상품 이미지 확인
#
# <div class="thumbnail">
# <a href="/product/detail.html?product_no=42&cate_no=70&display_group=1" name="anchorBoxName_42"><img src="//cocotail.co.kr/web/product/medium/201909/cc8ffbd9cd198a80d979f47bcb24cc6e.jpg" onmouseover="this.src='//cocotail.co.kr/web/product/big/201909/027c792045bf9ece1e6c77312f20ff3d.jpg'" onmouseout="this.src='//cocotail.co.kr/web/product/medium/201909/cc8ffbd9cd198a80d979f47bcb24cc6e.jpg'" id="eListPrdImage42_1" alt=""></a>
# <span class="wish"><img src="//img.echosting.cafe24.com/design/skin/admin/ko_KR/btn_wish_before.png" class="icon_img ec-product-listwishicon" alt="관심상품 등록 전" productno="42" categoryno="70" icon_status="off" login_status="T" individual-set="F"></span>
# </div>
###########################
self.set_product_image_fourth( product_data, product_ctx )
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx )
###########################
#
# <p class="name">
# <a href="/product/detail.html?product_no=81&cate_no=56&display_group=1"><span style="font-size:12px;color:#333333;"><b><font color="ff0000"></font><b>LUXURY SOFA</b><br><span style="color: rgb(0,0,0);">블랙</span></b></span></a><b><br><span class="zoom"><img src="//img.echosting.cafe24.com/design/skin/admin/ko_KR/btn_prd_zoom.gif" onclick="zoom('81', '56', '1','', '');" style="cursor:pointer" alt="상품 큰 이미지 보기"></span>
# </b></p>
###########################
crw_post_url = self.set_product_name_url_first( product_data, product_ctx , 'strong', 'name')
if(crw_post_url == '') : crw_post_url = self.set_product_name_url_first( product_data, product_ctx , 'p', 'name')
self.set_product_price_brand_first(product_data, product_ctx )
if( crw_post_url != '' ) :
self.set_product_url_hash( product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self, page_url, soup, product_ctx):
#
#
try:
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
self.set_product_category_first(product_data, soup)
###########################
# 상품 이미지 확인
#
# <img src="//ai-doggi.com/web/product/medium/20191220/a8ebb002293a954628763cf4a9ab6c38.jpg" alt="" class="thumb">
###########################
self.set_product_image_second(product_data, product_ctx)
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx)
###########################
#
# <p class="name">
# <a href="/product/detail.html?product_no=286&cate_no=43&display_group=1"><strong class="title displaynone"><span style="font-size:12px;color:#555555;">상품명</span> :</strong> <span style="font-size:12px;color:#555555;">Frill Neck Sleeve Blouse Lavender [20%SALE]</span></a>
# </p>
###########################
crw_post_url = self.set_product_name_url_first(
product_data, product_ctx, 'p', 'name')
if (crw_post_url == ''):
crw_post_url = self.set_product_name_url_first(
product_data, product_ctx, 'strong', 'name')
##############################
#
#
##############################
self.set_product_price_brand_first(product_data, product_ctx)
if (crw_post_url != ''):
self.set_product_url_hash(product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self, page_url, soup, product_ctx):
#
#
try:
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
#self.set_product_category_first(product_data, soup)
self.set_product_category_second(page_url, product_data, soup)
###########################
# 상품 이미지 확인
#
# <a href="/product/detail.html?product_no=142&cate_no=55&display_group=1" name="anchorBoxName_142"><img src="//cokemill.com/web/product/medium/201810/be20e9b9a3981ddfea24803c128c21ad.jpg" alt="" class="thumb"></a>
###########################
self.set_product_image_fourth(product_data, product_ctx)
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx)
###########################
#
# <p class="name">
# <a href="/product/detail.html?product_no=144&cate_no=55&display_group=1"><span style="font-size:12px;color:#555555;">HOT!!!!시나몬라떼 니트조끼(카멜)</span></a><br><span class="zoom"><img src="//img.echosting.cafe24.com/design/skin/admin/ko_KR/btn_prd_zoom.gif" onclick="zoom('144', '55', '1','', '');" style="cursor:pointer" alt="상품 큰 이미지 보기"></span>
# </p>
###########################
crw_post_url = self.set_product_name_url_third(
product_data, product_ctx, 'p', 'name')
if (crw_post_url == ''):
crw_post_url = self.set_product_name_url_third(
product_data, product_ctx, 'strong', 'name')
self.set_product_price_brand_second(product_data, product_ctx)
if (crw_post_url != ''):
self.set_product_url_hash(product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self , page_url, soup, product_ctx ) :
#
#
try :
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
self.set_product_category_first(product_data, soup)
###########################
# 상품 이미지 확인
#
# <div class="thumbnail">
# <a href="/product/detail.html?product_no=335&cate_no=24&display_group=1" name="anchorBoxName_335"><img src="//cjgsfood1.cafe24.com/web/product/medium/201906/8161292e023f5f4e3d99fba4801742ab.jpg" id="eListPrdImage335_1" alt="돼지꼬리뼈 100g"></a>
# <span class="wish"></span>
# </div>
###########################
self.set_product_image_fourth( product_data, product_ctx )
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx )
###########################
#
# <strong class="name"><a href="/product/detail.html?product_no=335&cate_no=24&display_group=1" class=""><span class="title displaynone"><span style="font-size:12px;color:#555555;">상품명</span> :</span> <span style="font-size:12px;color:#555555;">돼지꼬리뼈 100g</span></a></strong>
###########################
crw_post_url = self.set_product_name_url_first( product_data, product_ctx , 'strong', 'name')
if(crw_post_url == '') : crw_post_url = self.set_product_name_url_first( product_data, product_ctx , 'p', 'name')
self.set_product_price_brand_second(product_data, product_ctx )
if( crw_post_url != '' ) :
self.set_product_url_hash( product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self, page_url, soup, product_ctx):
#
#
try:
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
self.set_product_category_first(product_data, soup)
###########################
# 상품 이미지 확인
#
###########################
self.set_product_image_third(product_data, product_ctx)
# 품절여부 확인
#
self.set_product_soldout_first(product_data, product_ctx)
###########################
# 상품명/URL
###########################
crw_post_url = self.set_product_name_url_fifth(
product_data, product_ctx, 'div', 'prd_name')
##############################
# 가격
#
##############################
self.set_product_price_brand_first(product_data, product_ctx)
if (crw_post_url != ''):
self.set_product_url_hash(product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self, page_url, soup, product_ctx):
#
# 각 사이트별로 변경
# product_ctx -- > li
# <a href="/product/detail.html?product_no=471&cate_no=69&display_group=1" class="">
# <span class="title displaynone"><span style="font-size:12px;color:#555555;">상품명</span> :</span> <span style="font-size:12px;color:#555555;">러버스 보호자용 반팔 티셔츠</span></a>
# http://pet-paradise.kr/category/%EC%99%B8%EC%B6%9C/6040/?page=5
#
# <meta property="og:description" content="SIDE BY SIDE" />
try:
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
# <meta property="og:description" content="SIDE BY SIDE" />
# self.set_product_category_first(product_data, soup)
self.set_product_category_second(page_url, product_data, soup)
# 상품 이미지 확인
self.set_product_image_third(product_data, product_ctx)
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx)
#
# 상품 링크 정보 및 상품명
#
crw_post_url = self.set_product_name_url_first(
product_data, product_ctx, 'div', 'df-prl-desc')
self.set_product_price_brand_first(product_data, product_ctx)
if (crw_post_url != ''):
self.set_product_url_hash(product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self , page_url, soup, product_ctx ) :
#
# 각 사이트별로 변경
# product_ctx -- > li
# <a href="/product/스누피-3way-유모차-6kg/1696/category/6040/display/1/" class=""><span class="title displaynone"><span style="font-size:12px;color:#555555;font-weight:bold;">상품명</span> :</span> <span style="font-size:12px;color:#555555;font-weight:bold;">스누피 3WAY 유모차 (6kg)</span></a>
# http://pet-paradise.kr/category/%EC%99%B8%EC%B6%9C/6040/?page=5
try :
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#split_list = page_url.split('/')
#if(4 < len(split_list) ) : product_data.crw_category1 = split_list[4].strip()
#self.set_product_category_first(product_data, soup)
self.set_product_category_second(page_url, product_data, soup)
# 상품 이미지 확인
self.set_product_image_fourth(product_data, product_ctx )
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx )
#
# 상품 링크 정보 및 상품명
#
crw_post_url = self.set_product_name_url_second( product_data, product_ctx , 'div', 'description')
self.set_product_price_brand_first(product_data, product_ctx )
if( crw_post_url != '' ) :
self.set_product_url_hash( product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self, page_url, soup, product_ctx):
#
#
# <meta property="og:description" content="SIDE BY SIDE" />
try:
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
# <meta property="og:description" content="생활 • 식품">
#self.set_product_category_first(product_data, soup)
self.set_product_category_second(page_url, product_data, soup)
# 상품 이미지 확인
self.set_product_image_fourth(product_data, product_ctx)
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx)
#
# 상품 링크 정보 및 상품명
#
crw_post_url = self.set_product_name_url_second(
product_data, product_ctx, 'div', 'description')
self.set_product_price_brand_first(product_data, product_ctx)
if (crw_post_url != ''):
self.set_product_url_hash(product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self, page_url, soup, product_ctx):
#
#
try:
crw_post_url = ''
####################################
# 상품 카테고리 추출
####################################
category_list = soup.select(
'body > table > tr > td > table > tr > td.outline_side > div.indiv > form > table> tr > td > b > a'
)
for category_ctx in category_list:
crw_category = category_ctx.get_text().strip()
break
# 유효한 카테고리 체크
if (self.check_ignore_category_text(crw_category)):
product_data = ProductData()
product_data.crw_category1 = crw_category
####################################
# 상품 이미지 확인
####################################
img_ctx = product_ctx.find('img')
if (img_ctx != None):
img_src = ''
if ('data-original' in img_ctx.attrs):
img_src = img_ctx.attrs['data-original'].strip()
elif ('src' in img_ctx.attrs):
img_src = img_ctx.attrs['src'].strip()
if (img_src != ''):
tmp_img_link = self.BASIC_IMAGE_URL + '/shop' + img_src
img_link = tmp_img_link.replace('..', '')
product_data.product_img = self.get_hangul_url_convert(
img_link)
####################################
# 품절여부 추출
# <img src="/shop/data/skin/freemart/img/icon/good_icon_soldout.gif">
####################################
img_list = product_ctx.find('img')
for img_ctx in img_list:
img_src = ''
if ('src' in img_ctx.attrs):
img_src = img_ctx.attrs['src'].strip()
if (0 <= img_src.find('soldout')):
product_data.crw_is_soldout = 1
####################################
# 상품 링크 정보 및 상품명 / 상품코드
####################################
#
# 상품 링크 정보 및 상품명 / 상품코드
is_product_name = True
is_product_link = True
product_link_list = product_ctx.find_all('a')
for product_link_ctx in product_link_list:
product_name = product_link_ctx.get_text().strip()
# 첫번때 A link에 있는 Text
if (is_product_name) and (product_name != ''):
product_data.crw_name = product_name
is_product_name = False
if (is_product_link):
if ('href' in product_link_ctx.attrs):
tmp_product_link = product_link_ctx.attrs[
'href'].strip()
if (tmp_product_link.find('javascript') < 0):
if (0 != tmp_product_link.find('http')):
tmp_product_link = '%s%s' % (
self.BASIC_PRODUCT_URL,
product_link_ctx.attrs['href'].strip())
crw_post_url = tmp_product_link
if (self.C_PRODUCT_STRIP_STR != ''):
crw_post_url = tmp_product_link.replace(
self.C_PRODUCT_STRIP_STR, '')
split_list = crw_post_url.split('?goodsno=')
sub_split_list = split_list[1].strip().split(
'&')
product_data.crw_goods_code = sub_split_list[
0].strip()
is_product_link = False
####################################
# 가격
####################################
div_list = product_ctx.find_all('div')
for div_ctx in div_list:
cost_ctx = div_ctx.find('b')
if (cost_ctx != None):
product_data.crw_price = int(
__UTIL__.get_only_digit(
cost_ctx.get_text().strip()))
if (crw_post_url != ''):
if (self.PRODUCT_URL_HASH.get(crw_post_url, -1) == -1):
self.set_product_data_sub(product_data, crw_post_url)
#self.print_product_page_info( product_data )
self.process_product_api(product_data)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self , page_url, soup, product_ctx ) :
#
#
try :
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
#self.set_product_category_first(product_data, soup)
self.set_product_category_second(page_url, product_data, soup)
# 상품 이미지 확인
self.set_product_image_first(product_data, product_ctx )
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx )
name_div_list = product_ctx.find_all('div', class_='sp-product__title')
for name_div_ctx in name_div_list :
product_link_list = name_div_ctx.find_all('a')
for product_link_ctx in product_link_list :
if('href' in product_link_ctx.attrs ) :
span_list = product_link_ctx.find_all('span')
for span_ctx in span_list :
name_value = span_ctx.get_text().strip()
if(0 != name_value.find('상품명') ) and (0 != name_value.find(':') ) : product_data.crw_name = name_value
tmp_product_link = product_link_ctx.attrs['href'].strip()
if(0 != tmp_product_link.find('http')) : tmp_product_link = '%s%s' % ( self.BASIC_PRODUCT_URL, product_link_ctx.attrs['href'].strip() )
crw_post_url = tmp_product_link
if(self.C_PRODUCT_STRIP_STR != '') : crw_post_url = tmp_product_link.replace( self.C_PRODUCT_STRIP_STR,'')
split_list = crw_post_url.split('/')
if( product_data.crw_name == '') : product_data.crw_name = split_list[4].strip()
product_data.crw_goods_code = split_list[5].strip()
div_list = product_ctx.find_all('div')
for div_ctx in div_list :
if('rel' in div_ctx.attrs) :
title_name = div_ctx.attrs['rel']
span_list = div_ctx.find_all('span')
for span_ctx in span_list :
span_value = span_ctx.get_text().strip()
if(span_value != '브랜드' ) and (span_value != '판매가' ) and (span_value != '할인가' ) and (span_value != '' ) :
if(title_name == '판매가') : product_data.crw_price = int( __UTIL__.get_only_digit( span_value ) )
if(title_name == '할인가') : product_data.crw_price_sale = int( __UTIL__.get_only_digit( span_value ) )
if(title_name == '브랜드') : product_data.crw_brand1 = span_value
if( crw_post_url != '' ) :
self.set_product_url_hash( product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True
def set_product_data(self, page_url, soup, product_ctx):
#
#
try:
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
self.set_product_category_first(product_data, soup)
#self.set_product_category_second(page_url, product_data, soup)
###########################
# 상품 이미지 확인
###########################
self.set_product_image_second(product_data, product_ctx)
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx)
###########################
# 상품명/URL
###########################
crw_post_url = self.set_product_name_url_second(
product_data, product_ctx, 'strong', 'name')
if (crw_post_url == ''):
crw_post_url = self.set_product_name_url_second(
product_data, product_ctx, 'p', 'name')
##############################
# 가격
#
##############################
price_all_ctx = product_ctx.find('li', class_='price_all')
if (price_all_ctx != None):
custom_list = price_all_ctx.find_all('span', class_='custom')
for custom_ctx in custom_list:
product_data.crw_price = int(
__UTIL__.get_only_digit(custom_ctx.get_text().strip()))
custom_list = price_all_ctx.find_all('span', class_='pri')
for custom_ctx in custom_list:
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(custom_ctx.get_text().strip()))
custom_list = price_all_ctx.find_all('span', class_='strike')
for custom_ctx in custom_list:
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(custom_ctx.get_text().strip()))
if (crw_post_url != ''):
self.set_product_url_hash(product_data, crw_post_url)
rtn = True
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return True