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_third(product_data, soup)
#self.set_product_category_second(page_url, product_data, soup)
###########################
# 상품 이미지 확인
#
# <a href="/product/detail.html?product_no=348&cate_no=105&display_group=1" name="anchorBoxName_348"><img class="lazy thumb" src="//www.urbanbeast.co.kr/web/product/big/201901/270edfd2fcebe218147a9cdf9904bc06.jpg" data-original="//www.urbanbeast.co.kr/web/product/big/201901/270edfd2fcebe218147a9cdf9904bc06.jpg" alt="" style="display: inline;"></a>
###########################
img_link_list = product_ctx.find_all('a')
for img_link_ctx in img_link_list:
if ('name' in img_link_ctx.attrs):
if (0 <=
img_link_ctx.attrs['name'].find('anchorBoxName_')):
img_ctx = img_link_ctx.find('img')
if (img_ctx != None):
if ('data-original' in img_ctx.attrs):
img_src = img_ctx.attrs['data-original'].strip(
)
if (img_src != ''):
img_link = self.set_img_url(
self.BASIC_IMAGE_URL, img_src)
if (product_data.product_img == ''):
product_data.product_img = self.get_hangul_url_convert(
img_link)
# 품절여부 확인
#
self.set_product_soldout_first(product_data, product_ctx)
###########################
# 상품명/URL
#
# <a href="/product/detail.html?product_no=146&cate_no=24&display_group=1" class="text-wrap ease">
# <span class="text ease">
# <p class="name"><span style="font-size:12px;color:#1e1e1e;">Beast Den Bumper Bed Dynamite Red<br>비스트덴 다이너마이트 레드</span> </p>
# <ul class="xans-element- xans-product xans-product-listitem"><li class=" xans-record-">
# <strong class="title displaynone"><span style="font-size:12px;color:#1e1e1e;font-weight:bold;">판매가</span> :</strong> <span style="font-size:12px;color:#1e1e1e;font-weight:bold;">98,000원</span><span id="span_product_tax_type_text" style=""> </span></li>
# </ul>
# </span>
# </a>
###########################
product_link_ctx = product_ctx.find('a', class_='text-wrap ease')
if (product_link_ctx != None):
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()
product_name_ctx = product_ctx.find('p', class_='name')
if (product_name_ctx != None):
product_data.crw_name = product_name_ctx.get_text().strip()
##############################
# 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_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 = ''
self.reset_product_category(product_data)
####################################
# 상품 카테고리 추출
####################################
__LOG__.Trace(self.PAGE_URL_HASH[page_url])
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
'''
div_list = soup.find_all( 'div' , class_='cntbody' )
for div_ctx in div_list :
category_list = div_ctx.find_all( 'h2', class_='subtitle' )
for category_ctx in category_list :
product_data.crw_category1 = category_ctx.get_text().strip()
'''
'''
####################################
# 브랜드 추출
####################################
brand_div_list = product_ctx.find_all('span', class_='item_brand')
for brand_ctx in brand_div_list :
brand_name = brand_ctx.get_text().strip()
if( brand_name != '') : product_data.crw_brand1 = brand_name.replace('[','').replace(']','').strip()
'''
####################################
# 상품 이미지 확인
#
# <div class="prdimg"><a href="https://www.howlpot.com/shop/detail.php?pno=41AE36ECB9B3EEE609D05B90C14222FB&rURL=https%3A%2F%2Fwww.howlpot.com%2Fshop%2Fbig_section.php%3Fcno1%3D1037&ctype=1&cno1=1037"><img src="https://howlpotdesign.wisacdn.com/_data/product/d0dcc887757a47bd539823e77b7a3da6.jpg" width="292" height="292"></a></div>
#
####################################
img_div_list = product_ctx.find_all('div', class_='prdimg')
for img_div_ctx in img_div_list:
img_ctx = img_div_ctx.find('img')
#for img_ctx in img_list :
if (img_ctx != None):
img_src = ''
if ('src' in img_ctx.attrs):
img_src = img_ctx.attrs['src'].strip()
if (img_src != ''):
img_link = self.set_img_url(self.BASIC_IMAGE_URL,
img_src)
product_data.product_img = self.get_hangul_url_convert(
img_link)
####################################
# 품절여부 추출
# 품절시 <div class="box out"> 으로 표현됨
#
# <div class="box out">
# <div class="no">03</div>
# <div class="img">
# 생략
# </div>
# <div class="info">
# 생략
# </div>
# </div>
#
####################################
if ('class' in product_ctx.attrs):
class_name_list = product_ctx.attrs['class']
if (len(class_name_list) == 2):
if (class_name_list[1] == 'out'):
product_data.crw_is_soldout = 1
####################################
# 상품 링크 정보 및 상품명 / 상품코드
#
# <div class="name">
# <a href="https://www.howlpot.com/shop/detail.php?pno=41AE36ECB9B3EEE609D05B90C14222FB&rURL=https%3A%2F%2Fwww.howlpot.com%2Fshop%2Fbig_section.php%3Fcno1%3D1037&ctype=1&cno1=1037">메모리폼_라이트 그레이</a>
# <span class="wish"><a href="#" onclick="wishPartCartAjax("41AE36ECB9B3EEE609D05B90C14222FB", this); return false;">관심상품 담기</a></span>
# </div>
#
####################################
name_strong_list = product_ctx.find_all('div', class_='name')
for name_strong_ctx in name_strong_list:
product_link_ctx = name_strong_ctx.find('a')
if (product_link_ctx != None):
#__LOG__.Trace( product_link_ctx )
if ('href' in product_link_ctx.attrs):
product_data.crw_name = product_link_ctx.get_text(
).strip()
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('?pno=')
second_split_list = split_list[1].split('&')
product_data.crw_goods_code = second_split_list[
0].strip()
####################################
# 가격
#
# <div class="price">
# <span class="sell"><span class="font">98,000</span></span>
# </div>
#
####################################
div_list = product_ctx.find_all('div', class_='price')
for div_ctx in div_list:
sell_ctx = div_ctx.find('span', class_='sell')
consumer_ctx = div_ctx.find('span', class_='consumer')
if (consumer_ctx != None):
product_data.crw_price = int(
__UTIL__.get_only_digit(
consumer_ctx.get_text().strip()))
if (sell_ctx != None):
# 타임세일일때 뒷부분의 별도의 값이 붙어서, 값 이상 문제 해결법,
crw_price_sale = sell_ctx.get_text().strip().split('\n')
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(crw_price_sale[0].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.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_second(self, page_url, soup):
#
#
try:
product_data = ProductData()
crw_post_url = page_url
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()
# 상품 카테고리
#
product_data.crw_category1 = 'PRODUCT'
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
img_list = soup.find_all('img', class_='BigImage')
for img_ctx in img_list:
if ('src' in img_ctx.attrs):
img_src = img_ctx.attrs['src'].strip()
if (img_src != ''):
img_link = self.set_img_url(self.BASIC_IMAGE_URL,
img_src)
if (product_data.product_img == ''):
product_data.product_img = self.get_hangul_url_convert(
img_link)
p_list = soup.find_all('p', class_='prd_model')
for p_ctx in p_list:
if (product_data.crw_name == ''):
product_data.crw_name = p_ctx.get_text().strip()
# 품절여부 확인
sold_out_ctx = soup.find('span', {'id': 'btnReserve'})
if (sold_out_ctx != None):
if ('class' in sold_out_ctx.attrs):
if ('displaynone' != sold_out_ctx.attrs['class'][0]):
product_data.crw_is_soldout = 1
else:
product_data.crw_is_soldout = 1
# 가격
price_list = soup.find_all('div', class_='info_price')
for price_ctx in price_list:
sell_ctx = price_ctx.find('span', class_='sell')
customer_ctx = price_ctx.find('span', class_='customer')
if (sell_ctx != None):
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(sell_ctx.get_text().strip()))
if (customer_ctx != None):
product_data.crw_price = int(
__UTIL__.get_only_digit(
customer_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.strip()
elif(idx == 2 ) : product_data.crw_category2 = split_data.strip()
elif(idx == 3 ) : product_data.crw_category3 = split_data.strip()
# 상품 이미지 확인
div_list = product_ctx.find_all('div', class_='thumbnail')
for div_ctx in div_list :
a_link_list = product_ctx.find_all('a')
for a_link_ctx in a_link_list :
img_list = a_link_ctx.find_all('img')
for img_ctx in img_list :
if('src' in img_ctx.attrs ) :
img_src = img_ctx.attrs['src'].strip()
if( img_src != '' ) :
img_link = self.set_img_url( self.BASIC_IMAGE_URL, img_src )
if(product_data.product_img == '') : product_data.product_img = self.get_hangul_url_convert( img_link )
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx )
name_div_list = product_ctx.find_all('strong', class_='name')
for name_div_ctx in name_div_list :
#
# 상품명 / 상품코드
#
product_link_list = name_div_ctx.find_all('a')
for product_link_ctx in product_link_list :
if('href' in product_link_ctx.attrs ) :
product_data.crw_name = product_link_ctx.get_text().strip()
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()
#
# 가격 / 브랜드
#
div_list = product_ctx.find_all('div', class_='description')
for div_ctx in div_list :
span_list = div_ctx.find_all('span')
for span_ctx in span_list :
if('class' in span_ctx.attrs ) :
class_name_list = span_ctx.attrs['class']
if(len(class_name_list) == 1) and ( class_name_list[0].strip() == 'summary') : product_data.crw_brand1 = span_ctx.get_text().strip()
li_list = div_ctx.find_all('li')
for li_ctx in li_list :
if('class' in li_ctx.attrs ) :
class_name_list = li_ctx.attrs['class']
if(len(class_name_list) == 1) and ( class_name_list[0].strip() == 'price') : product_data.crw_price = int( __UTIL__.get_only_digit( li_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.CRW_CATEGORY_1 = ''
self.CRW_CATEGORY_2 = ''
self.CRW_CATEGORY_3 = ''
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_second(page_url, product_data, soup)
self.CRW_CATEGORY_1 = product_data.crw_category1
self.CRW_CATEGORY_2 = product_data.crw_category2
self.CRW_CATEGORY_3 = product_data.crw_category3
#product_data.crw_category1 = self.PAGE_URL_HASH[ page_url ]
# 상품 이미지 확인
self.set_product_image_fourth(product_data, product_ctx)
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx)
img_div_list = product_ctx.find_all(
'div', class_=self.C_PRODUCT_IMG_SELECTOR_CLASSNAME)
for img_div_ctx in img_div_list:
product_link_ctx = img_div_ctx.find('a')
if (product_link_ctx != None):
if ('href' in product_link_ctx.attrs):
crw_post_url = self.get_crw_post_url(
product_link_ctx, 'href')
if (crw_post_url != ''):
split_list = crw_post_url.split('/')
if (5 < len(split_list)):
product_data.crw_goods_code = split_list[
5].strip()
name_div_list = product_ctx.find_all('div', class_='description')
for name_div_ctx in name_div_list:
#
# 상품 링크 정보 및 상품명 / 상품코드
#
name_strong_list = name_div_ctx.find_all('strong',
class_='name')
for name_strong_ctx in name_strong_list:
product_link_list = name_strong_ctx.find_all('a')
for product_link_ctx in product_link_list:
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(':')) and (
name_value.strip() != ''):
product_data.crw_name = name_value
# 가격
li_list = name_div_ctx.find_all('li')
for li_ctx in li_list:
span_list = li_ctx.find_all('span')
for span_ctx in span_list:
price_value = span_ctx.get_text().strip()
if (price_value != '') and (price_value != ':'):
if ('style' in span_ctx.attrs):
if (0 < span_ctx.attrs['style'].find(
'text-decoration')):
product_data.crw_price = int(
__UTIL__.get_only_digit(price_value))
else:
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(price_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_third(product_data, soup)
# 상품 이미지 확인
# <a href="/product/detail.html?product_no=417&cate_no=29&display_group=1" name="anchorBoxName_417"><img src="//www.wefam.co.kr/web/product/medium/201704/417_shop1_670038.jpg?cmd=thumb&width=300&height=300" data-original="//www.wefam.co.kr/web/product/medium/201704/417_shop1_670038.jpg?cmd=thumb&width=300&height=300" id="eListPrdImage417_1" alt="" class="thumb" style="display: inline; opacity: 1;"></a>
###########################
img_link_list = product_ctx.find_all('a')
for img_link_ctx in img_link_list:
if ('name' in img_link_ctx.attrs):
if (0 <=
img_link_ctx.attrs['name'].find('anchorBoxName_')):
tmp_product_link = img_link_ctx.attrs['href'].strip()
if (0 != tmp_product_link.find('http')):
tmp_product_link = '%s%s' % (
self.BASIC_PRODUCT_URL,
img_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()
img_ctx = img_link_ctx.find('img')
if (img_ctx != None):
if ('data-original' in img_ctx.attrs):
img_src = img_ctx.attrs['data-original'].strip(
)
if (img_src != ''):
img_link = self.set_img_url(
self.BASIC_IMAGE_URL, img_src)
if (product_data.product_img == ''):
product_data.product_img = self.get_hangul_url_convert(
img_link)
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx)
#
# 상품명 / 가격
#
# <ul class="xans-product-listitem">
# <li class="manu">we.fam (위팸)</li>
# <li class="second"><span><span style="font-size:13px;color:#2e2e2e;">Zigzag Pattern Sleeveless_Navy</span></span></li>
# <li class="custom">25,000 WON</li>
# <li class="price">15,800 WON</li>
# </ul>
#
li_list = product_ctx.find_all('li')
for li_ctx in li_list:
if ('class' in li_ctx.attrs):
class_name_list = li_ctx.attrs['class']
value_str = li_ctx.get_text().strip()
if (class_name_list[0] == 'manu'):
product_data.crw_brand1 = value_str
elif (class_name_list[0] == 'second'):
product_data.crw_name = value_str
elif (class_name_list[0] == 'custom'):
int_value_str = int(__UTIL__.get_only_digit(value_str))
if (int_value_str != 0):
product_data.crw_price = int(
__UTIL__.get_only_digit(value_str))
elif (class_name_list[0] == 'price'):
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 = ''
self.reset_product_category(product_data)
#__LOG__.Trace( page_url )
#__LOG__.Trace( self.PAGE_URL_HASH[page_url])
if (self.C_DETAIL_CATEGORY_VALUE.strip() != ''):
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
else:
self.set_godo_category_data(soup, product_data)
####################################
# 브랜드 추출
#
# <span class="item_brand">
# <strong>[지그니쳐]</strong>
# </span>
#
####################################
brand_div_list = product_ctx.find_all('span', class_='item_brand')
for brand_ctx in brand_div_list:
brand_name = brand_ctx.get_text().strip()
if (brand_name != ''):
product_data.crw_brand1 = brand_name.replace(
'[', '').replace(']', '').strip()
####################################
# 상품 이미지 확인
#
# <div class="item_photo_box">
# <a href="../goods/goods_view.php?goodsNo=1000000896&mtn=%5E%7C%5E%5E%7C%5E">
# <img data-original="/data/goods/19/10/43/1000000896/1000000896_add2_085.jpg" width="250" alt="바잇밀 - 치킨/오리/말고기 샘플러 100g (3종)" title="바잇밀 - 치킨/오리/말고기 샘플러 100g (3종)" class="middle gd_image_lazy" src="/data/goods/19/10/43/1000000896/1000000896_add2_085.jpg" style="display: inline;">
# </a>
# </div>
####################################
img_div_list = product_ctx.find_all('div', class_='item_photo_box')
for img_div_ctx in img_div_list:
img_ctx = img_div_ctx.find('img')
#for img_ctx in img_list :
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 != ''):
img_link = self.set_img_url(self.BASIC_IMAGE_URL,
img_src)
if (product_data.product_img == ''):
product_data.product_img = self.get_hangul_url_convert(
img_link)
####################################
# 품절여부 추출 (2가지 방법)
#
# <div class="item_icon_box">
# <img src="/data/icon/goods_icon/당일출고아이콘.gif" alt="벌써배송상품" title="벌써배송상품" class="middle">
# <img src="/data/icon/goods_icon/icon_soldout.gif">
# </div>
#
#------------------------------------
#
# <div class="item_photo_box">
# <a href="../goods/goods_view.php?goodsNo=1000001614">
# <img data-original="/data/goods/20/05/20/1000001614/1000001614_add2_099.jpg" width="250" alt="빅독 리틀 바이트 사료 - 악어고기 100g" title="빅독 리틀 바이트 사료 - 악어고기 100g" class="middle gd_image_lazy" src="/data/goods/20/05/20/1000001614/1000001614_add2_099.jpg" style="display: inline;">
# <strong class="item_soldout_bg" style="background-image:url(/data/icon/goods_icon/soldout-1.png);">SOLD OUT</strong>
# </a>
# </div>
#
####################################
soldout_div_list = product_ctx.find_all('div',
class_='item_icon_box')
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('soldout')):
product_data.crw_is_soldout = 1
if (product_data.crw_is_soldout != 1):
soldout_div_list = product_ctx.find_all(
'div', class_='item_photo_box')
for soldout_div_ctx in soldout_div_list:
img_list = soldout_div_ctx.find_all(
'strong', class_='item_soldout_bg')
for img_ctx in img_list:
product_data.crw_is_soldout = 1
####################################
# 상품 링크 정보 및 상품명 / 상품코드
#
# <div class="item_tit_box">
# <a href="../goods/goods_view.php?goodsNo=1000001614">
# <strong class="item_name">빅독 리틀 바이트 사료 - 악어고기 100g</strong>
# </a>
# </div>
#
####################################
name_strong_list = product_ctx.find_all('div',
class_='item_tit_box')
for name_strong_ctx in name_strong_list:
product_link_list = name_strong_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('strong')
for span_ctx in span_list:
product_data.crw_name = span_ctx.get_text().strip()
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('?goodsNo=')
product_data.crw_goods_code = split_list[1].strip()
####################################
# 가격
#
# <div class="item_money_box">
# <strong class="item_price">
# <span>23,000원 </span>
# </strong>
# </div>
#
####################################
div_list = product_ctx.find_all('div', class_='item_money_box')
for div_ctx in div_list:
del_ctx = div_ctx.find('del')
strong_ctx = div_ctx.find('strong', class_='item_price')
if (del_ctx != None):
product_data.crw_price = int(
__UTIL__.get_only_digit(del_ctx.get_text().strip()))
if (strong_ctx != None):
# 타임세일일때 뒷부분의 별도의 값이 붙어서, 값 이상 문제 해결법,
crw_price_sale = strong_ctx.get_text().strip().split('\n')
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(crw_price_sale[0].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.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_second(self, page_url, soup, product_ctx):
#
#
try:
product_data = ProductData()
crw_post_url = ''
self.reset_product_category(product_data)
####################################
# 상품 카테고리 추출
####################################
self.set_godo_category_data_second(soup, product_data)
'''
div_list = soup.find_all("div" , class_='cg-main')
for div_ctx in div_list :
category_list = div_ctx.find_all('h2')
for category_ctx in category_list :
product_data.crw_category1 = category_ctx.get_text().strip()
'''
'''
# 브랜드 확인
brand_div_list = product_ctx.find_all('span', class_='item_brand')
for brand_ctx in brand_div_list :
brand_name = brand_ctx.get_text().strip()
if( brand_name != '') : product_data.crw_brand1 = brand_name.replace('[','').replace(']','').strip()
'''
####################################
# 상품 이미지 확인
#
# <div class="thumbnail">
# <a href="../goods/goods_view.php?goodsNo=1000000030"><img src="/data/goods/16/10/43/1000000030/1000000030_main_072.jpg" width="184" alt="Pet Esthé Spa Mud Conditioner (스파 머드 컨디셔너) (3L)" title="Pet Esthé Spa Mud Conditioner (스파 머드 컨디셔너) (3L)" class="middle">
# </a>
# </div>
####################################
img_div_list = product_ctx.find_all('div', class_='thumbnail')
for img_div_ctx in img_div_list:
img_ctx = img_div_ctx.find('img')
#for img_ctx in img_list :
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 != ''):
img_link = self.set_img_url(self.BASIC_IMAGE_URL,
img_src)
if (product_data.product_img == ''):
product_data.product_img = self.get_hangul_url_convert(
img_link)
####################################
# 상품 링크 정보 및 상품명 / 상품코드
#
# <div class="txt">
# <a href="../goods/goods_view.php?goodsNo=1000000030">
# <strong>Pet Esthé Spa Mud Conditioner (스파 머드 컨디셔너) (3L)</strong> </a>
# </div>
#
####################################
name_div_list = product_ctx.find_all('div', class_='txt')
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('strong')
for span_ctx in span_list:
product_data.crw_name = span_ctx.get_text().strip()
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('?goodsNo=')
sub_split_list = split_list[1].split('&')
product_data.crw_goods_code = sub_split_list[0].strip()
####################################
# 가격 / 품절여부
#
# <div class="price gd-default">
# <span class="cost">
# <strong>180,000</strong>원
# </span>
# <br>
# </div>
#
# -------------품절시 -----------------
# <div class="price gd-default">
# <span class="cost">
# <strong>일시품절</strong>
# </span>
# <br>
# </div>
#
####################################
div_list = product_ctx.find_all('div')
for div_ctx in div_list:
if ('class' in div_ctx.attrs):
class_name_list = div_ctx.attrs['class']
if (class_name_list[0] == 'price'):
cost_ctx = div_ctx.find('span', class_='cost')
if (cost_ctx != None):
cost_value = cost_ctx.get_text().strip()
product_data.crw_price = int(
__UTIL__.get_only_digit(cost_value))
if (0 < cost_value.find('품절')):
product_data.crw_is_soldout = 1
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 = ''
self.reset_product_category(product_data)
####################################
# 상품 카테고리 추출
####################################
self.get_category_value(product_data, page_url, soup)
####################################
# 상품 이미지 확인
#
# <img data-prodcode="s20200603cfcd165650db6" alt="" src="https://cdn.imweb.me/thumbnail/20200603/69b4e17496c01.jpg" class="_org_img org_img _lazy_img" data-original="https://cdn.imweb.me/thumbnail/20200603/69b4e17496c01.jpg" data-src="https://cdn.imweb.me/thumbnail/20200603/69b4e17496c01.jpg" style="display: inline;">
#
# class_='_org_img org_img _lazy_img'
# class_='_org_img org_img owl-lazy'
####################################
img_list = product_ctx.find_all('img')
for img_ctx in img_list:
if ('class' in img_ctx.attrs):
class_name_list = img_ctx.attrs['class']
if (2 < len(class_name_list)):
if (class_name_list[0]
== '_org_img') and (class_name_list[1]
== 'org_img'):
img_src = ''
if ('data-original' in img_ctx.attrs):
img_src = img_ctx.attrs['data-original'].strip(
)
if (img_src == ''):
if ('data-src' in img_ctx.attrs):
img_src = img_ctx.attrs['data-src'].strip()
if (img_src == ''):
if ('src' in img_ctx.attrs):
img_src = img_ctx.attrs['src'].strip()
if (img_src != ''):
img_link = self.set_img_url(
self.BASIC_IMAGE_URL, img_src)
product_data.product_img = self.get_hangul_url_convert(
img_link)
####################################
# 품절여부 추출
#
# <div class="ns-icon clearfix">
# <!--<span class="new bg-brand">NEW</span>-->
# <div class="prod_icon sale">SALE</div><div class="prod_icon sold_out">SOLDOUT</div> </div>
#
####################################
soldout_div_list = product_ctx.find_all(
'div', class_='prod_icon sold_out')
for soldout_div_ctx in soldout_div_list:
product_data.crw_is_soldout = 1
####################################
# 상품 링크 정보 및 상품명 / 상품코드
#
# <div class="item-detail" style="">
# <div class="item-pay">
# <h2 style="display: ">
# <a class="_fade_link" href="/shop/?idx=1185">어반비스트 훈련용 코만도백</a>
# </h2>
# <div class="item-pay-detail">
# <p class="sale_pay body_font_color_50" style="">78,000원</p> <p class="pay" style=";">
# 58,500원 </p>
# </div>
# <div class="ns-icon clearfix">
# <!--<span class="new bg-brand">NEW</span>-->
# <div class="prod_icon sale">SALE</div><div class="prod_icon sold_out">SOLDOUT</div> </div>
# </div>
# <div class="item-summary holder">
# <p>반려견의 산책과 훈련을 위한 코만도백</p> <a class="item-summary-link _fade_link" href="/shop/?idx=1185"><span class="sr-only">상품 요약설명</span></a>
# </div>
# <div class="item-icon">
# <span><i class="icon-bubble"></i> 0</span>
# </div>
# </div>
#
####################################
name_div_list = product_ctx.find_all('div', class_='item-detail')
for name_div_ctx in name_div_list:
h2_list = name_div_ctx.find_all('h2')
for h2_ctx in h2_list:
product_link_ctx = name_div_ctx.find('a',
class_='_fade_link')
if (product_link_ctx != None):
if ('href' in product_link_ctx.attrs):
product_data.crw_name = h2_ctx.get_text().strip()
crw_post_url = self.get_crw_post_url(
product_link_ctx, 'href')
if (crw_post_url != ''):
split_list = crw_post_url.split('?idx=')
crw_goods_code_list = split_list[1].strip(
).split('&')
product_data.crw_goods_code = crw_goods_code_list[
0].strip()
####################################
# 가격
#
# <div class="item-pay-detail">
# <p class="sale_pay body_font_color_50" style="">78,000원</p> <p class="pay" style=";">
# 58,500원 </p>
# </div>
#
####################################
price_div_list = product_ctx.find_all('div',
class_='item-pay-detail')
for price_ctx in price_div_list:
p_list = name_div_ctx.find_all('p')
for p_ctx in p_list:
if ('class' in p_ctx.attrs):
class_name_list = p_ctx.attrs['class']
if (class_name_list[0] == 'sale_pay'):
product_data.crw_price = int(
__UTIL__.get_only_digit(
p_ctx.get_text().strip()))
elif (class_name_list[0] == 'pay'):
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(
p_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.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 = ''
self.reset_product_category(product_data)
####################################
# 상품 카테고리
####################################
div_list = soup.find_all('div', class_='category_depth clearbox')
for div_ctx in div_list:
li_list = div_ctx.find_all('li', class_='item')
idx = 0
for li_ctx in li_list:
category_link = li_ctx.find('a')
if (category_link != None):
idx += 1
category_name = category_link.get_text().strip()
if (idx == 2):
product_data.crw_category1 = category_link.get_text(
).strip()
elif (idx == 3):
product_data.crw_category2 = category_link.get_text(
).strip()
elif (idx == 4):
product_data.crw_category3 = category_link.get_text(
).strip()
'''
####################################
# 브랜드 추출
####################################
'''
####################################
# 상품 이미지 확인 / 상품 링크 정보 / 상품번호
#
# <a href="/goods/view?no=792" target="">
# <img src="/data/goods/201606/792_18171332list2.jpg" width="130" onerror="this.src='/data/skin/0545blueface/images/common/noimage.gif';this.style.height='130px';">
# </a>
####################################
span_list = product_ctx.find_all('span',
class_='goodsDisplayImageWrap')
for span_ctx in span_list:
product_link_ctx = span_ctx.find('a')
if (product_link_ctx != None):
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('?no=')
product_data.crw_goods_code = split_list[1].strip()
img_list = product_link_ctx.find_all('img')
for img_ctx in img_list:
if ('onerror' in img_ctx.attrs):
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 != ''):
img_link = self.set_img_url(
self.BASIC_IMAGE_URL, img_src)
product_data.product_img = self.get_hangul_url_convert(
img_link)
####################################
# 상품명
#
# <a href="/goods/view?no=792" target=""><span style="color:#4C4C4C;font-family:dotum;font-size:10pt;font-weight:normal;text-decoration:none;">버박 칼시데리스 칼슘영양제 (30정)</span></a>
####################################
name_div_list = product_ctx.find_all('a')
for name_div_ctx in name_div_list:
span_ctx = name_div_ctx.find('span')
if (span_ctx != None):
product_data.crw_name = span_ctx.get_text().strip()
####################################
# 가격 / 품절여부
#
# <span style="color:#4C4C4C;font-family:dotum;font-size:10pt;font-weight:normal;text-decoration:line-through;">
# 21,000
# 원 </span>
#
#
# <span style="color:#4C4C4C;font-family:dotum;font-size:10pt;font-weight:bold;text-decoration:none;">
# 15,000
# 원 </span>
#
# -------- 품절시 --------------
# <td align="center">
# <img src="/data/icon/goods_status/icon_list_soldout.gif">
# </td>
####################################
div_list = product_ctx.find_all('td')
for div_ctx in div_list:
soldout_img_list = div_ctx.find_all('img')
for soldout_ctx in soldout_img_list:
if ('src' in soldout_ctx.attrs):
if (0 <= soldout_ctx.attrs['src'].find('soldout')):
product_data.crw_is_soldout = 1
span_ctx = div_ctx.find('span')
if (span_ctx != None):
span_str = span_ctx.get_text().strip()
if ('style' in span_ctx.attrs):
if (span_str != ''):
if (span_str[0].isdigit()) and (
0 < span_str.find('원')):
if (0 < span_ctx.attrs['style'].find(
'text-decoration:line-through')):
product_data.crw_price = int(
__UTIL__.get_only_digit(span_str))
if (0 < span_ctx.attrs['style'].find(
'text-decoration:none')):
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(span_str))
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.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, category_path_str,
sub_category_str, product_ctx):
#
#
try:
product_data = ProductData()
crw_post_url = ''
self.reset_product_category(product_data)
idx = 0
for category_name in category_path_str:
idx += 1
if (idx == 2): product_data.crw_category1 = category_name
elif (idx == 3): product_data.crw_category2 = category_name
elif (idx == 4): product_data.crw_category3 = category_name
if (sub_category_str != ''):
if (idx == 2): product_data.crw_category2 = sub_category_str
elif (idx == 3): product_data.crw_category3 = sub_category_str
#product_data.crw_category1 = self.PAGE_URL_HASH[ page_url ]
'''
####################################
# 브랜드 추출
####################################
brand_div_list = product_ctx.find_all('div', class_='brand')
for brand_ctx in brand_div_list :
product_data.crw_brand1 = brand_ctx.get_text().strip()
'''
####################################
# 상품 이미지 확인 / 품절여부 추출
#
# <a href="/view/product/G0OS8PKL0KAHNCUJ/YSJRQFSCH" class="itemImg" title="[맛보기초특가!] 체험팩 8종 8팩">
# <img src="http://www.dhuman.co.kr/static-root/prdct/2020/05/13/bd1f6f76032f48329e927e9f7e727fe7.jpg" alt="[맛보기초특가!] 체험팩 8종 8팩" class="" loading="lazy">
# <span class="discountThumb">
# <span class="discount"><strong>50</strong>%</span>
# </span>
# <span class="packplayWrap">
# <span class="pack_bg02"><span><strong class="pack_font">8</strong>팩</span></span>
# </span>
# </a>
#
####################################
img_div_list = product_ctx.find_all('a', class_='itemImg')
for img_div_ctx in img_div_list:
img_list = img_div_ctx.find_all('img')
for img_ctx in img_list:
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 != ''):
img_link = self.set_img_url(self.BASIC_IMAGE_URL,
img_src)
product_data.product_img = self.get_hangul_url_convert(
img_link)
span_list = img_div_ctx.find_all('span')
for span_ctx in span_list:
soldout_str = span_ctx.get_text().strip()
if (0 <= soldout_str.find('품절')):
product_data.crw_is_soldout = 1
####################################
# 상품 링크 정보 및 상품명 / 상품코드
#
# <div class="itemTit">
# <p class="name">
# <a href="/view/product/G0OS8PKL0KAHNCUJ/YSJRQFSCH" title="[맛보기초특가!] 체험팩 8종 8팩">
# [맛보기초특가!] 체험팩 8종 8팩
# </a>
# </p>
# <p class="cmnt">#휴먼그레이드 #신상체험팩</p>
# </div>
#
####################################
name_div_list = product_ctx.find_all('div', class_='itemTit')
for name_div_ctx in name_div_list:
name_ctx = name_div_ctx.find('p', class_='name')
if (name_ctx != None):
name_link_ctx = name_ctx.find('a')
if (name_link_ctx != None):
if ('href' in name_link_ctx.attrs):
product_data.crw_name = name_link_ctx.get_text(
).strip()
split_list = name_link_ctx.attrs['href'].strip(
).split(';')
tmp_product_link = split_list[0].strip()
if (0 != tmp_product_link.find('http')):
tmp_product_link = '%s%s' % (
self.BASIC_PRODUCT_URL,
split_list[0].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_data.crw_goods_code = split_list[6].strip()
####################################
# 가격
#
# <div class="priceWrap">
# <span class="primecost"><strong>19,900</strong></span>
# <span class="price"><strong>9,900</strong>원</span>
# </div>
####################################
div_list = product_ctx.find_all('div', class_='priceWrap')
for div_ctx in div_list:
span_list = div_ctx.find_all('span')
for span_ctx in span_list:
if ('class' in span_ctx.attrs):
class_name_list = span_ctx.attrs['class']
if (class_name_list[0] == 'primecost'):
product_data.crw_price = int(
__UTIL__.get_only_digit(
span_ctx.get_text().strip()))
elif (class_name_list[0] == 'price'):
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(
span_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.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, category_key, soup, product_ctx):
#
#
try:
product_data = ProductData()
crw_post_url = ''
self.reset_product_category(product_data)
####################################
# 상품 카테고리 추출
####################################
self.get_category_value(product_data, category_key, soup)
####################################
# 상품 이미지 확인
#
# <div class="thumb img" imgsrc="/uploadedFiles/46606/product/image_1573609552547.jpeg" style="width:100%;background-image:url(https://contents.sixshop.com/thumbnails/uploadedFiles/46606/product/image_1573609552547_1000.jpeg)"></div>
#
####################################
img_div_list = product_ctx.find_all('div', class_='thumb img')
for img_div_ctx in img_div_list:
if ('style' in img_div_ctx.attrs):
tmp_img_src = img_div_ctx.attrs['style'].strip()
split_list = tmp_img_src.split(':url(')
img_src = split_list[1].replace(')', '')
if (img_src != ''):
img_link = self.set_img_url(self.BASIC_IMAGE_URL,
img_src)
product_data.product_img = self.get_hangul_url_convert(
img_link)
####################################
# 품절여부 추출
#
# <div class="soldOutBadge badge"><span>Sold Out</span></div>
#
####################################
soldout_div_list = product_ctx.find_all(
'div', class_='soldOutBadge badge')
for soldout_div_ctx in soldout_div_list:
product_data.crw_is_soldout = 1
# 가격 부분에 sold out 문구가 있는 경우
price_div_list = product_ctx.find_all('div',
class_='shopProduct price')
for price_ctx in price_div_list:
soldout_str = price_ctx.get_text().strip()
if (0 <= soldout_str.lower().find('sold')) and (
0 < soldout_str.lower().find('out')):
product_data.crw_is_soldout = 1
####################################
# 상품 링크 정보 및 상품명 / 상품코드
#
# <div class="shopProductWrapper badgeUse" data-productno="1008345"><a href="/product/Chu"><div class="thumbDiv"><div class="thumb img" imgsrc="/uploadedFiles/46606/product/image_1573609552547.jpeg" style="width:100%;background-image:url(https://contents.sixshop.com/thumbnails/uploadedFiles/46606/product/image_1573609552547_1000.jpeg)"></div><div class="shopProductBackground"></div><div class="badgeWrapper"><div class="soldOutBadge badge"><span>Sold Out</span></div></div></div><div class="shopProductNameAndPriceDiv"><div class="shopProductNameAndPriceContent"><div class="shopProductNameAndPrice"><div class="shopProduct productName">멜로니코코 풉백</div><div class="shopProduct price"><span class="productPriceSpan">20,000원</span></div></div></div></div></a></div>
#
####################################
if ('data-productno' in product_ctx.attrs):
product_data.crw_goods_code = product_ctx.attrs[
'data-productno']
product_link_ctx = product_ctx.find('a')
if (product_link_ctx != None):
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, '')
name_div_list = product_ctx.find_all(
'div', class_='shopProduct productName')
for name_div_ctx in name_div_list:
product_data.crw_name = name_div_ctx.get_text().strip()
####################################
# 가격
#
# <span class="productPriceSpan">20,000원</span>
#
# <div class="shopProduct price"><span class="productDiscountPriceSpan">16,200원 </span><span class="productPriceWithDiscountSpan">18,000원</span></div>
####################################
price_div_list = product_ctx.find_all('div',
class_='shopProduct price')
for price_ctx in price_div_list:
span_list = price_ctx.find_all('span')
for span_ctx in span_list:
if ('class' in span_ctx.attrs):
class_name_list = span_ctx.attrs['class']
if (class_name_list[0] == 'productPriceSpan'):
product_data.crw_price = int(
__UTIL__.get_only_digit(
span_ctx.get_text().strip()))
elif (class_name_list[0] == 'productDiscountPriceSpan'
):
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(
span_ctx.get_text().strip()))
elif (class_name_list[0] ==
'productPriceWithDiscountSpan'):
product_data.crw_price = int(
__UTIL__.get_only_digit(
span_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.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 = ''
self.reset_product_category(product_data)
####################################
# 상품 카테고리 추출
####################################
self.get_category_value( product_data, page_url, soup )
####################################
# 상품 이미지 확인
#
# <div class="prdImg scroll-fade">
# <a href="/product/방수커버/57/category/61/display/1/" name="anchorBoxName_57">
# <img src="//eledog.co.kr/web/product/medium/202011/74b4cc12fc9dd4f38c49d4de2d2f6b51.jpg" id="eListPrdImage57_1" class="thumb_Img" alt="방수커버"> </a>
# </div>
#
# class_='_org_img org_img _lazy_img'
# class_='_org_img org_img owl-lazy'
####################################
img_ctx = product_ctx.find('div', class_='prdImg')
img_ = img_ctx.find('img', class_='thumb_Img')
img_src = ''
if('src' in img_.attrs ) : img_src = img_.attrs['src'].strip()
if( img_src != '' ) :
img_link = self.set_img_url( self.BASIC_IMAGE_URL, img_src )
product_data.product_img = self.get_hangul_url_convert( img_link )
crw_goods_code_ = img_ctx.find('a')
if(crw_goods_code_ != '') :
if('name' in crw_goods_code_.attrs ) :
if( 0 <= crw_goods_code_.attrs['name'].find('anchorBoxName_')) :
class_name_list = crw_goods_code_.attrs['name']
product_data.crw_goods_code = class_name_list.replace('anchorBoxName_','')
####################################
# 품절여부 추출
#
# <div class="promotion"><img src="//img.echosting.cafe24.com/design/skin/admin/ko_KR/ico_product_soldout.gif" class="icon_img" alt="품절"> </div>
#
####################################
soldout_div_list = product_ctx.find_all('img', alt='품절')
for soldout_div_ctx in soldout_div_list :
product_data.crw_is_soldout = 1
####################################
# <div class="item_list_box">
# <div class="thumbnail">
# <div class="custom_pro dj_count30"><span class="dj-mov-fade-in-out2">21%</span></div>
# <div class="button">
# <ul>
# <li class="likeButton displaynone"><button type="button">LIKE<strong></strong></button></li>
# </ul>
# <ul class="other">
# <li class="option"><span>옵션보기</span>
# </li>
# <li class="cart">
# <a href="#none"><img src="//img.echosting.cafe24.com/design/skin/admin/ko_KR/btn_list_cart.gif" onclick="CAPP_SHOP_NEW_PRODUCT_OPTIONSELECT.selectOptionCommon(43, 42, 'basket', '')" alt="장바구니 담기" class="ec-admin-icon cart"></a><span>장바구니</span>
# </li>
# <li class="zoom">
# <a href="#none"><img src="//img.echosting.cafe24.com/design/skin/admin/ko_KR/btn_prd_zoom.gif" onclick="zoom('43', '42', '1','', '');" style="cursor:pointer" alt="상품 큰 이미지 보기"></a><span>확대보기</span>
# </li>
# <li class="pop">
# <a href="/product/ver02-맞춤이가방-카키/43/category/42/display/1/" target="blank"><img src="/_dj/img/button_other_04.png"></a><span>새창보기</span>
# </li>
# </ul>
# </div>
# <div class="prdImg scroll-fade">
# <a href="/product/ver02-맞춤이가방-카키/43/category/42/display/1/" name="anchorBoxName_43">
# <img src="//eledog.co.kr/web/product/medium/202011/d05fff61e25635e2dae85c6a7dadce63.jpg" id="eListPrdImage43_1" class="thumb_Img" alt="[ver.02] 맞춤이가방 : 카키"> </a>
# </div>
# </div>
# <div class="description" onclick="window.location.href='/product/ver02-맞춤이가방-카키/43/category/42/display/1/';">
# <div class="inner">
# <div class="displaynone">
# </div>
# <div class="brand displaynone"></div>
# <strong class="name"><a href="/product/ver02-맞춤이가방-카키/43/category/42/display/1/" class=""><span style="font-size:12px;color:#555555;">[ver.02] 맞춤이가방 : 카키</span></a></strong>
# <ul class="spec">
# <li class="summary_line displaynone">
# </li>
# <li class="summary displaynone"></li>
# <li class="price_all">
# <span class="custom ">188,000원</span>
# <span class="price displaynone"><span class="strike">149,000원</span><span class="pri">149,000원</span></span>
# <span class="sale displaynone"></span>
# </li>
# </ul>
# <div class="icon">
# <div class="promotion"> <img src="//img.echosting.cafe24.com/design/skin/admin/ko_KR/ico_product_recommended.gif" class="icon_img" alt="추천"> </div>
# </div>
# </div>
# </div>
# </div>
#
####################################
name_div_list = product_ctx.find_all('div', class_='description')
for name_div_ctx in name_div_list :
h2_list = name_div_ctx.find_all('strong')
for h2_ctx in h2_list :
product_link_ctx = name_div_ctx.find('a')
if( product_link_ctx != None) :
if('href' in product_link_ctx.attrs ) :
product_data.crw_name = h2_ctx.get_text().strip()
crw_post_url = self.get_crw_post_url( product_link_ctx, 'href')
####################################
# 가격
#
# <li class="price_all">
# <span class="custom ">188,000원</span>
# <span class="price displaynone"><span class="strike">149,000원</span><span class="pri">149,000원</span></span>
# <span class="sale displaynone"></span>
# </li>
#
####################################
price_div_list = product_ctx.find_all('li', class_='price_all')
for price_ctx in price_div_list :
p_list = name_div_ctx.find_all('span')
for p_ctx in p_list :
if('class' in p_ctx.attrs ) :
class_name_list = p_ctx.attrs['class']
if(class_name_list[0] == 'custom' ) : product_data.crw_price = int( __UTIL__.get_only_digit( p_ctx.get_text().strip() ) )
elif(class_name_list[0] == 'pri' ) : product_data.crw_price_sale = int( __UTIL__.get_only_digit( p_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.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
#
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_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)
name_div_list = product_ctx.find_all('div', class_='thumb-info')
for name_div_ctx in name_div_list:
#
# 상품 링크 정보 및 상품명 / 상품코드
#
name_strong_list = name_div_ctx.find_all('p')
for name_strong_ctx in name_strong_list:
product_link_list = name_strong_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('?product_no=')
crw_goods_code_list = split_list[1].strip().split(
'&')
product_data.crw_goods_code = crw_goods_code_list[
0].strip()
# 가격
span_list = product_ctx.find_all('span', class_='price')
for span_ctx in span_list:
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)
###########################
# 상품 이미지 확인
#
###########################
self.set_product_image_first( product_data, product_ctx )
# 품절여부 확인
#
#self.set_product_soldout_first(product_data, product_ctx )
###########################
# 상품명/URL
###########################
name_div_list = product_ctx.find_all('a', class_='name')
for name_div_ctx in name_div_list :
if('href' in name_div_ctx.attrs ) :
name_value = name_div_ctx.get_text().strip()
if(product_data.crw_name == '') : product_data.crw_name = name_value
tmp_product_link = name_div_ctx.attrs['href'].strip()
if(0 != tmp_product_link.find('http')) : tmp_product_link = '%s%s' % ( self.BASIC_PRODUCT_URL, name_div_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()
##############################
# 가격
#
##############################
li_list = product_ctx.find_all('li')
for li_ctx in li_list :
main_span_ctx = li_ctx.find('span')
if( main_span_ctx != None) :
strong_ctx = main_span_ctx.find('strong')
span_ctx = main_span_ctx.find_all('span')
if(strong_ctx != None) :
if(1 < len(span_ctx) ) :
title_name = strong_ctx.get_text().strip()
split_list = span_ctx[2].get_text().strip().split('(')
value_str = split_list[0].strip()
if( 0 == title_name.find( '브랜드')) : product_data.crw_brand1 = value_str
elif( 0 == title_name.find( '제조사')) : product_data.crw_brand2 = value_str
elif( 0 == title_name.find( '소비자가')) : product_data.crw_price = int( __UTIL__.get_only_digit( value_str ) )
elif( 0 == title_name.find( '판매가')) : 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)
#
# 보정
crw_category1 = product_data.crw_category1
product_data.crw_category1 = crw_category1.replace(
'HEALTHY SNACK', '').replace('BAKERY SUSU',
'').replace('EVERYDAY MEAL', '')
# 상품 이미지 확인
self.set_product_image_second(product_data, product_ctx)
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx)
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_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('?product_no=')
crw_goods_code_list = split_list[1].strip().split('&')
product_data.crw_goods_code = crw_goods_code_list[
0].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_second(self, page_url, product_json):
#
#
try:
product_data = ProductData()
crw_post_url = ''
# 상품 카테고리
#
#self.set_product_category_second(page_url, product_data, soup)
#product_data.crw_category1 = self.PAGE_URL_HASH[ page_url ]
product_data.crw_category1 = self.CRW_CATEGORY_1
product_data.crw_category2 = self.CRW_CATEGORY_2
product_data.crw_category3 = self.CRW_CATEGORY_3
for key in product_json:
#__LOG__.Trace('%s : %s' % (key, product_json[key] ))
# 이미지
if (key == 'image_medium'):
img_src = product_json[key]
img_link = self.set_img_url(self.BASIC_IMAGE_URL, img_src)
product_data.product_img = self.get_hangul_url_convert(
img_link)
if (key == 'image_big'):
if (product_data.product_img != ''):
img_src = product_json[key]
img_link = self.set_img_url(self.BASIC_IMAGE_URL,
img_src)
product_data.product_img = self.get_hangul_url_convert(
img_link)
if (key == 'image_small'):
if (product_data.product_img != ''):
img_src = product_json[key]
img_link = self.set_img_url(self.BASIC_IMAGE_URL,
img_src)
product_data.product_img = self.get_hangul_url_convert(
img_link)
# 상품명
if (key == 'product_name_striptag'):
product_data.crw_name = product_json[key]
if (key == 'product_name_tag'):
if (product_data.crw_name != ''):
product_data.crw_name = product_json[key]
if (key == 'product_name'):
if (product_data.crw_name != ''):
product_data.crw_name = product_json[key]
# 상품명번호
if (key == 'product_no'):
product_data.crw_goods_code = str(product_json[key])
# 상품 URL
if (key == 'link_product_detail'):
tmp_product_link = product_json[key]
if (0 != tmp_product_link.find('http')):
tmp_product_link = '%s%s' % (self.BASIC_PRODUCT_URL,
product_json[key])
crw_post_url = tmp_product_link
if (self.C_PRODUCT_STRIP_STR != ''):
crw_post_url = tmp_product_link.replace(
self.C_PRODUCT_STRIP_STR, '')
# 상품 가격
if (key == 'product_custom'):
product_data.crw_price = int(product_json[key])
if (key == 'product_price'):
product_data.crw_price_sale = int(product_json[key])
# soldout
if (key == 'soldout_icon'):
if (product_json[key].strip() != ''):
product_data.crw_is_soldout = 1
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 = ''
self.reset_product_category(product_data)
location_ctx = soup.find('div', {'id': 'navigation-bar'})
if (location_ctx != None):
span_list = location_ctx.find_all('span')
idx = 0
for span_ctx in span_list:
idx += 1
if (idx == 2):
product_data.crw_category1 = span_ctx.get_text().strip(
)
elif (idx == 3):
product_data.crw_category2 = span_ctx.get_text().strip(
)
elif (idx == 4):
product_data.crw_category3 = span_ctx.get_text().strip(
)
####################################
# 브랜드 추출
####################################
product_data.crw_brand1 = product_data.crw_category1
####################################
# 상품 이미지 확인
#
# <img class="item-image" src="/_vir0001/product_img/P1449_20200421AM94623_2.jpg" alt="img1">
####################################
img_list = product_ctx.find_all('img', class_='item-image')
for img_ctx in img_list:
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 != ''):
img_link = self.set_img_url(self.BASIC_IMAGE_URL, img_src)
product_data.product_img = self.get_hangul_url_convert(
img_link)
####################################
# 상품 링크 정보 및 상품코드
#
# <div class="item" style="cursor:pointer" onclick="goHref(1449,'P1449_20200421AM94623_2.jpg')">
####################################
if ('onclick' in product_ctx.attrs):
onclick_str = product_ctx.attrs['onclick'].strip()
split_data = onclick_str.split('goHref(')
sub_split_data = split_data[1].split(',')
product_data.crw_goods_code = sub_split_data[0].strip()
crw_post_url = 'http://www.montraum.com/common/process/shopview.asp?thisCategory=22&pack_content_id=' + product_data.crw_goods_code
####################################
# 상품명 / 품절여부
#
# <p class="item-description" id="iconID1449" name="iconID1449">데일리관리 세트<br> (돈모 브러쉬+플러쉬 콤)</p>
#
# ---------- 품절시 -----------------------
# <p class="item-description" id="iconID1095" name="iconID1095">트레이닝패드 L 120매 (30매 x 4개)<img src="/_vir0001/process/partImages/icon_soldout.gif" align="absmiddle"> <img src="/_vir0001/process/partImages/icon_soldout.gif" align="absmiddle"> <img src="/_vir0001/process/partImages/icon_soldout.gif" align="absmiddle"> <img src="/_vir0001/process/partImages/icon_soldout.gif" align="absmiddle"> </p>
####################################
name_div_list = product_ctx.find_all('p',
class_='item-description')
for name_div_ctx in name_div_list:
product_data.crw_name = name_div_ctx.get_text().replace(
'\n', ' ').strip()
# 품절여부
soldout_img_list = name_div_ctx.find_all('img')
for soldout_img in soldout_img_list:
if ('src' in soldout_img.attrs):
if (0 <= soldout_img.attrs['src'].find('soldout')):
product_data.crw_is_soldout = 1
####################################
# 가격
#
# <p class="item-price">
# <span class="list-price" id="ori_count1449" name="ori_count1449">74,000</span> <span class="now-price" id="promotion_ID1449" name="promotion_ID1449">40,900</span>
# </p>
####################################
div_list = product_ctx.find_all('p', class_='item-price')
for div_ctx in div_list:
span_list = div_ctx.find_all('span')
for span_ctx in span_list:
if ('class' in span_ctx.attrs):
class_name_list = span_ctx.attrs['class']
if (class_name_list[0] == 'list-price'):
product_data.crw_price = int(
__UTIL__.get_only_digit(
span_ctx.get_text().strip()))
elif (class_name_list[0] == 'now-price'):
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(
span_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.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 = ''
self.reset_product_category(product_data)
category_ctx_list = soup.select(
'body > div.body_wrap > div.content_wrap > div.section_tit > div.close'
)
for category_ctx in category_ctx_list:
split_list = category_ctx.get_text().strip().split('>')
idx = 0
for split_data in split_list:
idx += 1
category_name = split_data.strip()
if (idx == 2): product_data.crw_category1 = category_name
elif (idx == 3): product_data.crw_category2 = category_name
elif (idx == 4): product_data.crw_category3 = category_name
#split_list = self.PAGE_URL_HASH[page_url].split('(')
#product_data.crw_category1 = split_list[0].replace('BEST','').strip()
####################################
# 브랜드 추출
#
# <div class="line_sub">
# 한국산 </div>
####################################
div_list = product_ctx.find_all('div', class_='line_sub')
for div_ctx in div_list:
brand_str = div_ctx.get_text().strip()
product_data.crw_brand1 = brand_str
####################################
# 상품 이미지 확인 / 상품 링크 정보 / 상품번호
#
# <div class="picture"><a href="./product.html?pd_code=A010489&event_type=%C3%CA%C6%AF%B0%A1"><img src="http://queenpuppy.co.kr/shop/pd_img/A01/489/A010489_2.jpg"></a></div>
####################################
span_list = product_ctx.find_all('div', class_='picture')
for span_ctx in span_list:
product_link_ctx = span_ctx.find('a')
if (product_link_ctx != None):
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())
if (self.C_PRODUCT_STRIP_STR != ''):
crw_post_url = tmp_product_link.replace(
self.C_PRODUCT_STRIP_STR, '')
split_list = crw_post_url.split('&event_type=')
crw_post_url = split_list[0].strip()
split_list = crw_post_url.split('?pd_code=')
sub_split_list = split_list[1].strip().split('&')
product_data.crw_goods_code = sub_split_list[0]
img_list = product_link_ctx.find_all('img')
for img_ctx in img_list:
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 != ''):
img_link = self.set_img_url(
self.BASIC_IMAGE_URL, img_src)
product_data.product_img = self.get_hangul_url_convert(
img_link)
####################################
# 상품명
#
# <div class="name">
# <div style="color:#fd705f; font-weight: bold; valign:top; height: 15px; padding-bottom: 3px;"></div>
# <a href="./product.html?pd_code=A010489&event_type=%C3%CA%C6%AF%B0%A1">
# 건국유업 프로젝트 닥터케이 펫밀크 10개입 </a>
# </div>
####################################
name_div_list = product_ctx.find_all('div', class_='name')
for name_div_ctx in name_div_list:
span_ctx = name_div_ctx.find('a')
if (span_ctx != None):
crw_name = span_ctx.get_text().strip()
product_data.crw_name = crw_name
if (0 < crw_name.find('[품절]')):
product_data.crw_is_soldout = 1
product_data.crw_name = crw_name.replace('[품절]',
'').strip()
####################################
# 가격
#
#
# <div class="line_np">20,000원</div>
# <div class="line_sp">
# 12,000원
# <span style="font-size: 0.8em; color: #666; vertical-align:bottom;">40%↓</span> </div>
####################################
div_list = product_ctx.find_all('div', class_='line_np')
for div_ctx in div_list:
price_str = div_ctx.get_text().strip()
product_data.crw_price = int(
__UTIL__.get_only_digit(price_str))
div_list = product_ctx.find_all('div', class_='line_sp')
for div_ctx in div_list:
price_str = div_ctx.get_text().strip()
span_ctx = div_ctx.find('span')
split_str = ''
if (span_ctx != None): split_str = span_ctx.get_text().strip()
if (split_str == ''):
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(price_str.strip()))
else:
split_list = price_str.split(split_str)
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(split_list[0].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.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 = ''
self.reset_product_category(product_data)
category_ctx_list = soup.select('#menu_inner')
for category_ctx in category_ctx_list:
split_list = category_ctx.get_text().strip().split('>')
idx = 0
for a_ctx in split_list:
idx += 1
category_name = a_ctx.strip()
if (idx == 2): product_data.crw_category1 = category_name
elif (idx == 3): product_data.crw_category2 = category_name
elif (idx == 4): product_data.crw_category3 = category_name
#product_data.crw_category1 = self.PAGE_URL_HASH[page_url]
############################
# 품절여부
############################
soldout_ctx = product_ctx.find('span', class_='soldOut')
if (soldout_ctx != None): product_data.crw_is_soldout = 1
####################################
# 상품 이미지 확인
#
# <div class="thumbnail">
# <div class="centered">
# <a href="prd_detail.php?idx=171&part_idx=90"><img src="/data/goodsImages/1529056838_IMAGES1.jpg" data-pin-nopin="true"></a>
# </div>
# </div>
####################################
span_list = product_ctx.find_all('div', class_='thumbnail')
for span_ctx in span_list:
product_link_ctx = span_ctx.find('a')
if (product_link_ctx != None):
img_list = product_link_ctx.find_all('img')
for img_ctx in img_list:
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 != ''):
img_link = self.set_img_url(
self.BASIC_IMAGE_URL, img_src)
product_data.product_img = self.get_hangul_url_convert(
img_link)
####################################
# 상품명 / 상품 링크 정보 / 상품번호
#
# <div class="title"><a href="prd_detail.php?idx=171&part_idx=90"><!--[보듬]--> 보듬 10mm 폴딩 리드줄 (길이 조절 가능)</a></div>
####################################
name_div_list = product_ctx.find_all('div', class_='title')
for name_div_ctx in name_div_list:
span_ctx = name_div_ctx.find('a')
if (span_ctx != None):
if ('href' in span_ctx.attrs):
tmp_product_link = span_ctx.attrs['href'].strip()
if (0 != tmp_product_link.find('http')):
tmp_product_link = '%s%s' % (
self.BASIC_PRODUCT_URL,
span_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('?idx=')
sub_split_list = split_list[1].strip().split('&')
product_data.crw_goods_code = sub_split_list[0]
split_list = span_ctx.get_text().strip().split(']')
crw_name = split_list[0].strip()
if (len(split_list) == 2):
product_data.crw_brand1 = split_list[0].replace(
'[', '').strip()
crw_name = split_list[1].strip()
product_data.crw_name = crw_name
####################################
# 가격
#
# <div class="priceWrap">
# <div class="saleprice"><span>119,000</span>원</div>
# <div class="price"><strong>101,150</strong>원</div>
# </div>
####################################
div_list = product_ctx.find_all('div', class_='priceWrap')
for div_ctx in div_list:
saleprice_ctx = div_ctx.find('div', class_='saleprice')
price_ctx = div_ctx.find('div', class_='price')
if (saleprice_ctx != None):
product_data.crw_price = int(
__UTIL__.get_only_digit(
saleprice_ctx.get_text().strip()))
if (price_ctx != None):
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(price_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.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 = ''
self.reset_product_category(product_data)
####################################
# 상품 카테고리 추출
####################################
if (self.C_DETAIL_CATEGORY_VALUE.strip() != ''):
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
#div_list = soup.find_all( 'div' , class_='sub_title_txt' )
#for div_ctx in div_list :
# category_list = div_ctx.find_all( 'h2' )
# for category_ctx in category_list :
# product_data.crw_category1 = category_ctx.get_text().strip()
'''
####################################
# 브랜드 추출
####################################
brand_div_list = product_ctx.find_all('span', class_='item_brand')
for brand_ctx in brand_div_list :
brand_name = brand_ctx.get_text().strip()
if( brand_name != '') : product_data.crw_brand1 = brand_name.replace('[','').replace(']','').strip()
'''
####################################
# 상품 이미지 확인
#
# <div class="img">
# <img src="https://img.mywisa.com/freeimg/smallstuff/_data/product/201803/31/9ee1628095bceaf0f9bb5d8dae079791.jpg" width="260" height="260">
# <!-- 상품품절 영역 -->
# 생략
# </div>
####################################
img_div_list = product_ctx.find_all('div', class_='prdimg')
for img_div_ctx in img_div_list:
img_ctx = img_div_ctx.find('img')
#for img_ctx in img_list :
if (img_ctx != None):
img_src = ''
if ('src' in img_ctx.attrs):
img_src = img_ctx.attrs['src'].strip()
if (img_src != ''):
img_link = self.set_img_url(self.BASIC_IMAGE_URL,
img_src)
product_data.product_img = self.get_hangul_url_convert(
img_link)
####################################
# 품절여부 추출
# 품절시 <div class="info out"> 으로 표현됨
#
# <!-- 상품품절 영역 -->
# <div class="soldout" onclick="location.href='https://www.smallstuff.kr/shop/detail.php?pno=A02FFD91ECE5E7EFEB46DB8F10A74059&rURL=https%3A%2F%2Fwww.smallstuff.kr%2Fshop%2Fbig_section.php%3Fcno1%3D1001&ctype=1&cno1=1001'">Sold out</div>
# <!-- //상품품절 영역 -->
####################################
if ('class' in product_ctx.attrs):
class_name_list = product_ctx.attrs['class']
# if( len(class_name_list) == 2 ) :
if (class_name_list[0] == 'soldout'):
product_data.crw_is_soldout = 1
####################################
# 상품 링크 정보 및 상품명 / 상품코드
#
# <p class="name"><a href="https://www.smallstuff.kr/shop/detail.php?pno=BEED13602B9B0E6ECB5B568FF5058F07&rURL=https%3A%2F%2Fwww.smallstuff.kr%2Fshop%2Fbig_section.php%3Fcno1%3D1001&ctype=1&cno1=1001">DINING SET 1P OLIVE</a></p>
#
####################################
name_strong_list = product_ctx.find_all('p', class_='name')
for name_strong_ctx in name_strong_list:
product_link_ctx = name_strong_ctx.find('a')
if (product_link_ctx != None):
if ('href' in product_link_ctx.attrs):
product_data.crw_name = product_link_ctx.get_text(
).strip()
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('?pno=')
second_split_list = split_list[1].split('&')
product_data.crw_goods_code = second_split_list[
0].strip()
####################################
# 가격
#
# <div class="price">
# <p class="consumer consumer">KRW 24,000</p>
# <p class="sell sell"><strong>KRW 22,800 </strong></p>
# </div>
#
####################################
div_list = product_ctx.find_all('div', class_='price')
for div_ctx in div_list:
sell_ctx = div_ctx.find('p', class_='sell')
# print('sell_ctx',sell_ctx)
if (sell_ctx != None):
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(sell_ctx.get_text().strip()))
crw_price = div_ctx.find('p', class_='consumer')
if (crw_price != None):
product_data.crw_price = int(
__UTIL__.get_only_digit(crw_price.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.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 = ''
self.reset_product_category(product_data)
category_ctx_list = soup.select('#sct_location')
for category_ctx in category_ctx_list:
a_ctx_list = category_ctx.find_all('a')
idx = 0
for a_ctx in a_ctx_list:
idx += 1
category_name = a_ctx.get_text().strip()
if (idx == 2): product_data.crw_category1 = category_name
elif (idx == 3): product_data.crw_category2 = category_name
elif (idx == 4): product_data.crw_category3 = category_name
#product_data.crw_category1 = self.PAGE_URL_HASH[page_url]
####################################
# 상품 이미지 확인
#
# <div class="sct_img">
# <a href="http://shop.i-avec.com/shop/item.php?it_id=1585815848">
# <img src="http://shop.i-avec.com/data/item/1585815848/thumb-7KCE7ZW07IiY6riw_front_310x310.png" width="310" height="310" alt="아베크 전해수기 뿌조" title="">
# </a>
# <div class="sct_sns"><a href="https://www.facebook.com/sharer/sharer.php?u=http%3A%2F%2Fshop.i-avec.com%2Fshop%2Fitem.php%3Fit_id%3D1585815848&p=%EC%95%84%EB%B2%A0%ED%81%AC+%EC%A0%84%ED%95%B4%EC%88%98%EA%B8%B0+%EB%BF%8C%EC%A1%B0+%7C+%EC%95%84%EB%B2%A0%ED%81%AC+%ED%8E%AB%EB%93%9C%EB%9D%BC%EC%9D%B4%EB%A3%B8-%ED%8E%AB+%EC%82%B4%EA%B7%A0+%ED%86%A0%ED%83%88+%EC%BC%80%EC%96%B4%EB%A3%B8" class="share-facebook" target="_blank"><img src="http://shop.i-avec.com/theme/basic/skin/shop/basic/img/facebook.png" alt="페이스북에 공유"></a><a href="https://twitter.com/share?url=http%3A%2F%2Fshop.i-avec.com%2Fshop%2Fitem.php%3Fit_id%3D1585815848&text=%EC%95%84%EB%B2%A0%ED%81%AC+%EC%A0%84%ED%95%B4%EC%88%98%EA%B8%B0+%EB%BF%8C%EC%A1%B0+%7C+%EC%95%84%EB%B2%A0%ED%81%AC+%ED%8E%AB%EB%93%9C%EB%9D%BC%EC%9D%B4%EB%A3%B8-%ED%8E%AB+%EC%82%B4%EA%B7%A0+%ED%86%A0%ED%83%88+%EC%BC%80%EC%96%B4%EB%A3%B8" class="share-twitter" target="_blank"><img src="http://shop.i-avec.com/theme/basic/skin/shop/basic/img/twitter.png" alt="트위터에 공유"></a><a href="https://plus.google.com/share?url=http%3A%2F%2Fshop.i-avec.com%2Fshop%2Fitem.php%3Fit_id%3D1585815848" class="share-googleplus" target="_blank"><img src="http://shop.i-avec.com/theme/basic/skin/shop/basic/img/gplus.png" alt="구글플러스에 공유"></a></div>
# </div>
#
####################################
span_list = product_ctx.find_all('div', class_='sct_img')
for span_ctx in span_list:
product_link_ctx = span_ctx.find('a')
if (product_link_ctx != None):
img_list = product_link_ctx.find_all('img')
for img_ctx in img_list:
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 != ''):
img_link = self.set_img_url(
self.BASIC_IMAGE_URL, img_src)
product_data.product_img = self.get_hangul_url_convert(
img_link)
####################################
# 품절여부
#
# <div class="sct_icon"><span class="sit_icon"><br><span class="shop_icon_soldout">Sold Out</span></span></div>
####################################
name_div_list = product_ctx.find_all('div', class_='sct_icon')
for name_div_ctx in name_div_list:
soldout_ctx = name_div_ctx.find('span',
class_='shop_icon_soldout')
if (soldout_ctx != None): product_data.crw_is_soldout = 1
####################################
# 상품명 / 상품 링크 정보 / 상품번호
#
# <div class="sct_txt"><a href="http://shop.i-avec.com/shop/item.php?it_id=1585815848">
# 아베크 전해수기 뿌조
# </a></div>
####################################
name_div_ctx = product_ctx.find('div', class_='sct_txt')
if (name_div_ctx != None):
product_link_ctx = name_div_ctx.find('a')
if (product_link_ctx != None):
if ('href' in product_link_ctx.attrs):
product_data.crw_name = product_link_ctx.get_text(
).strip()
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('?it_id=')
sub_split_list = split_list[1].strip().split('&')
product_data.crw_goods_code = sub_split_list[0]
####################################
# 가격
#
# <div class="sct_cost">
# <span class="sct_discount">70,000원</span>
# 62,900원
#</div>
####################################
price_ctx = product_ctx.find('div', class_='sct_cost')
if (price_ctx != None):
price_discount_ctx = product_ctx.find('span',
class_='sct_discount')
if (price_discount_ctx != None):
product_data.crw_price = int(
__UTIL__.get_only_digit(
price_discount_ctx.get_text().strip()))
len_price_str = len(price_discount_ctx.get_text().strip())
price_str = price_ctx.get_text().strip()
crw_price_sale = price_str[len_price_str:].strip()
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(crw_price_sale))
else:
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(price_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.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 = ''
self.reset_product_category(product_data)
self.get_category_value( product_data, page_url, soup )
####################################
# 브랜드 추출
#
# <div class="brand">SALLYS LAW</div>
####################################
brand_div_list = product_ctx.find_all('div', class_='brand')
for brand_ctx in brand_div_list :
product_data.crw_brand1 = brand_ctx.get_text().strip()
####################################
# 상품 이미지 확인
#
# <div class="img">
# <img src="//image.wconcept.co.kr/productimg/image/img1/96/300972496.jpg?RS=300" alt="">
# </div>
####################################
img_div_list = product_ctx.find_all('div', class_='img')
for img_div_ctx in img_div_list :
img_list = img_div_ctx.find_all('img')
for img_ctx in img_list :
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()
split_list = img_src.split('?')
img_src = split_list[0].strip()
if( img_src != '' ) :
img_link = self.set_img_url( self.BASIC_IMAGE_URL, img_src )
product_data.product_img = self.get_hangul_url_convert( img_link )
'''
####################################
# 품절여부 추출
####################################
soldout_div_list = product_ctx.find_all('div', class_='item_icon_box')
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('soldout') ) : product_data.crw_is_soldout = 1
'''
####################################
# 상품 링크 정보 및 상품명 / 상품코드
#
# <a href="/Product/300972496">
#
####################################
product_link_ctx = product_ctx.find('a')
if( product_link_ctx != None ) :
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_data.crw_goods_code = split_list[4].strip()
name_strong_list = product_ctx.find_all('div', class_='product ellipsis multiline')
for name_strong_ctx in name_strong_list :
product_data.crw_name = name_strong_ctx.get_text().strip()
####################################
# 가격
#
# <div class="price">
# <span class="discount_price">74,400</span>
# <span class="base_price">93,000</span>
# <span class="discount_rate">20%</span>
# </div>
#
####################################
div_list = product_ctx.find_all('div', class_='price')
for div_ctx in div_list :
span_list = div_ctx.find_all('span')
for span_ctx in span_list :
if('class' in span_ctx.attrs ) :
class_name_list = span_ctx.attrs['class']
if(class_name_list[0] == 'base_price' ) : product_data.crw_price = int( __UTIL__.get_only_digit( span_ctx.get_text().strip() ) )
elif(class_name_list[0] == 'discount_price' ) : product_data.crw_price_sale = int( __UTIL__.get_only_digit( span_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.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 추출
####################################
product_data.crw_category1 = self.PAGE_URL_HASH[page_url]
'''
# 브랜드 확인
brand_div_list = product_ctx.find_all('span', class_='item_brand')
for brand_ctx in brand_div_list :
brand_name = brand_ctx.get_text().strip()
if( brand_name != '') : product_data.crw_brand1 = brand_name.replace('[','').replace(']','').strip()
'''
####################################
# 상품 이미지 확인
####################################
img_div_list = product_ctx.find_all('div', class_='goodsimg')
for img_div_ctx in img_div_list:
img_list = img_div_ctx.find_all('img')
for img_ctx in img_list:
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.startswith('..')):
tmp_img_src = '/shop%s' % img_src[2:]
img_src = tmp_img_src
if (img_src != ''):
img_link = self.set_img_url(self.BASIC_IMAGE_URL,
img_src)
product_data.product_img = self.get_hangul_url_convert(
img_link)
'''
# 품절여부 확인
soldout_div_list = product_ctx.find_all('div', class_='item_icon_box')
for soldout_div_ctx in soldout_div_list :
if(config.__DEBUG__) : __LOG__.Trace('품절여부 확인')
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('soldout') ) :product_data.crw_is_soldout = 1
# 품절여부 확인
soldout_div_list = product_ctx.find_all('div', class_='item_photo_box')
for soldout_div_ctx in soldout_div_list :
if(config.__DEBUG__) : __LOG__.Trace('품절여부 확인')
img_list = soldout_div_ctx.find_all('strong', class_='item_soldout_bg')
for img_ctx in img_list :
product_data.crw_is_soldout = 1
'''
####################################
# 상품 링크 정보 및 상품명 / 상품코드
####################################
name_div_list = product_ctx.find_all('div', class_='goods_m_name')
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):
product_data.crw_name = product_link_ctx.get_text(
).strip()
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('?goodsno=')
sub_split_list = split_list[1].strip().split('&')
product_data.crw_goods_code = sub_split_list[0].strip()
####################################
# 가격
####################################
div_list = product_ctx.find_all('div')
for div_ctx in div_list:
cost_ctx = div_ctx.find('b')
strike_ctx = div_ctx.find('strike')
if (cost_ctx != None):
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(cost_ctx.get_text().strip()))
if (strike_ctx != None):
product_data.crw_price = int(
__UTIL__.get_only_digit(strike_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_fourth(product_data, product_ctx)
# 품절여부 확인
self.set_product_soldout_first(product_data, product_ctx)
name_div_list = product_ctx.find_all('div')
for name_div_ctx in name_div_list:
class_name_list = name_div_ctx.attrs['class']
if (class_name_list[0] == 'description'):
#
# 상품 링크 정보 및 상품명 / 상품코드
#
name_strong_list = name_div_ctx.find_all('p',
class_='name')
if (len(name_strong_list) == 0):
name_strong_list = name_div_ctx.find_all('strong',
class_='name')
for name_strong_ctx in name_strong_list:
product_link_list = name_strong_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('?product_no=')
crw_goods_code_list = split_list[1].strip(
).split('&')
product_data.crw_goods_code = crw_goods_code_list[
0].strip()
#
# 가격
#
li_list = name_div_ctx.find_all('li', class_='price_all')
for li_ctx in li_list:
if (config.__DEBUG__): __LOG__.Trace('가격 체크')
span_list = li_ctx.find_all('span')
for span_ctx in span_list:
class_name_list = span_ctx.attrs['class']
#
# 구분되는 물품이 없음..
#if(class_name_list[0] == 'strike' ) : product_data.crw_price = int( __UTIL__.get_only_digit( span_ctx.get_text().strip() ) )
#elif(class_name_list[0] == 'pri' ) : product_data.crw_price_sale = int( __UTIL__.get_only_digit( span_ctx.get_text().strip() ))
if (class_name_list[0] == 'pri'):
product_data.crw_price = 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 = ''
self.reset_product_category(product_data)
category_ctx_list = soup.select('#layout_config_full > div.category_depth.clearbox > ul')
for category_ctx in category_ctx_list :
a_ctx_list = category_ctx.get_text().strip().split('>')
idx = 0
for a_ctx in a_ctx_list :
idx += 1
category_name = a_ctx.strip()
if(idx == 2 ) : product_data.crw_category1 = category_name
elif(idx == 3 ) : product_data.crw_category2 = category_name
elif(idx == 4 ) : product_data.crw_category3 = category_name
#product_data.crw_category1 = self.PAGE_URL_HASH[page_url]
####################################
# 상품 이미지 확인 / 상품 링크 정보 / 상품번호
#
# <a href="javascript:void(0)" onclick="display_goods_view('196','',this,'goods_view')"><span style="color:#000000;font-weight:normal;text-decoration:none;" class="goods_name">강아지 목줄/3M리드줄세트 <br>콤비네이션_베이지브라운</span></a>
####################################
span_list = product_ctx.find_all('div', class_='goodsDisplayImageWrap')
for span_ctx in span_list :
product_link_ctx = span_ctx.find('a')
if( product_link_ctx != None ) :
if('onclick' in product_link_ctx.attrs ) :
split_list = product_link_ctx.attrs['onclick'].split('display_goods_view(')
sub_split_list = split_list[1].split(',')
product_data.crw_goods_code = sub_split_list[0].replace("'","").strip()
tmp_product_link = self.SITE_HOME + '/goods/view?no=' + product_data.crw_goods_code
if(0 != tmp_product_link.find('http')) : tmp_product_link = '%s%s' % ( self.BASIC_PRODUCT_URL, span_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,'')
img_list = product_link_ctx.find_all('img')
for img_ctx in img_list :
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 != '' ) :
img_link = self.set_img_url( self.BASIC_IMAGE_URL, img_src )
product_data.product_img = self.get_hangul_url_convert( img_link )
####################################
# 상품명
#
# <span style="color:#000000;font-weight:normal;text-decoration:none;" class="goods_name">강아지 목줄 <br>콤비네이션_옐로우레드</span>
####################################
name_div_ctx = product_ctx.find('span', class_='goods_name')
if( name_div_ctx != None) :
product_data.crw_name = name_div_ctx.get_text().replace('\n',' ').strip()
####################################
# 가격
#
# <li>
# <span class="price_txt">판매가</span>
# <span style="color:#777777;font-weight:normal;text-decoration:line-through;" class="sale_price">
# 15,000 </span>
# </li>
#
####################################
li_list = product_ctx.find_all('li')
for li_ctx in li_list :
title_ctx = li_ctx.find('span', class_='price_txt')
value_ctx = li_ctx.find('span', class_='sale_price')
if(title_ctx != None) and (value_ctx != None) :
title_name = title_ctx.get_text().strip()
title_value = value_ctx.get_text().strip()
if( title_name == '판매가' ) : product_data.crw_price = int( __UTIL__.get_only_digit( title_value ) )
elif( title_name == '이벤트가' ) : product_data.crw_price_sale = int( __UTIL__.get_only_digit( title_value ) )
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.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:
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, category_num, soup, product_ctx):
#
#
is_loop = True
try:
product_data = ProductData()
crw_post_url = ''
self.reset_product_category(product_data)
split_list = self.CATEGORY_URL_HASH[category_num].split('|')
idx = 0
for category_name in split_list:
idx += 1
if (idx == 1): product_data.crw_category1 = category_name
elif (idx == 2): product_data.crw_category2 = category_name
elif (idx == 3): product_data.crw_category3 = category_name
#product_data.crw_category1 = self.CATEGORY_URL_HASH[category_num]
####################################
# 상품 이미지 확인
#
# <div class="saleCard__area-img"><i class="saleCard__badge--new"><svg height="33" viewBox="0 0 33 33" width="33" xmlns="http://www.w3.org/2000/svg"><g fill="none" fill-rule="evenodd"><circle cx="16.5" cy="16.5" fill="#3E0A7D" r="16" stroke="#FFF"></circle><path d="M15.38 19.37v-6.71h-1.33v6.71h1.33zm-7.01-1.23c.54-.187.99-.458 1.35-.815.36-.357.647-.768.86-1.235.2.433.475.813.825 1.14.35.327.782.577 1.295.75l.67-1.05a3.308 3.308 0 0 1-.965-.48c-.27-.2-.492-.427-.665-.68a2.796 2.796 0 0 1-.385-.82 3.242 3.242 0 0 1-.125-.89v-.9H9.89v.9c0 .307-.042.612-.125.915-.083.303-.215.59-.395.86A3.183 3.183 0 0 1 7.7 17.09l.67 1.05zm7.24 3.56v-1.05h-5.04v-1.92H9.25v2.97h6.36zm8.57-3.6v-2.22h1.22v-1.09h-1.22v-2.13h-1.32v5.44h1.32zm-6.7-.09c.527-.18.968-.44 1.325-.78s.635-.74.835-1.2c.193.393.458.738.795 1.035a3.49 3.49 0 0 0 1.235.685l.69-1.02a3.426 3.426 0 0 1-.935-.445 2.538 2.538 0 0 1-1.015-1.375c-.08-.267-.12-.54-.12-.82v-.94h-1.34v.83c0 .313-.04.622-.12.925-.08.303-.207.587-.38.85-.173.263-.397.5-.67.71-.273.21-.603.378-.99.505l.69 1.04zm3.69 3.85c.48 0 .912-.04 1.295-.12.383-.08.71-.197.98-.35.27-.153.477-.34.62-.56.143-.22.215-.47.215-.75s-.072-.528-.215-.745a1.774 1.774 0 0 0-.62-.555 3.376 3.376 0 0 0-.98-.35 6.365 6.365 0 0 0-1.295-.12c-.48 0-.913.04-1.3.12-.387.08-.715.197-.985.35-.27.153-.477.338-.62.555a1.32 1.32 0 0 0-.215.745c0 .28.072.53.215.75.143.22.35.407.62.56.27.153.598.27.985.35.387.08.82.12 1.3.12zm0-1.02c-.593 0-1.043-.062-1.35-.185-.307-.123-.46-.315-.46-.575 0-.253.153-.442.46-.565.307-.123.757-.185 1.35-.185.587 0 1.033.062 1.34.185.307.123.46.312.46.565 0 .26-.153.452-.46.575-.307.123-.753.185-1.34.185z" fill="#FFF" fill-rule="nonzero"></path></g></svg></i><div class="saleCard__area-badge"></div><img alt=" 주식캔 no.6 캥거루 200g" class="saleCard__img--no-bg" src="https://cdn.purplesto.re/media/store/sale/main_image/dogzfinefood_dog_B052DF21_thumb01.png"></div>
####################################
span_list = product_ctx.find_all('div',
class_='saleCard__area-img')
for span_ctx in span_list:
img_list = span_ctx.find_all('img',
class_='saleCard__img--no-bg')
if (len(img_list) == 0):
img_list = span_ctx.find_all(
'img', class_='saleCard__img--concept')
for img_ctx in img_list:
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 != ''):
img_link = self.set_img_url(self.BASIC_IMAGE_URL,
img_src)
product_data.product_img = self.get_hangul_url_convert(
img_link)
####################################
# 품절여부
#
# <div class="saleCard__area-price"><p class="saleCard__price--new">Sold Out</p></div>
####################################
name_div_list = product_ctx.find_all('div',
class_='saleCard__area-price')
for name_div_ctx in name_div_list:
soldout_ctx = name_div_ctx.find('p',
class_='saleCard__price--new')
if (soldout_ctx != None):
if (0 <= soldout_ctx.get_text().find('Out')):
product_data.crw_is_soldout = 1
####################################
# 상품명 / 상품 링크 정보 / 상품번호 / 브랜드 / 가격
#
# <div class="saleCard__area-content"><a class="saleCard__link" href="/products/brands/65/"><p class="saleCard__brand">로투스</p></a> <a class="saleCard__link" href="/products/sales/1500/"><p class="saleCard__title">CAT 오븐베이크 그레인프리 닭고기 4종</p></a><p class="saleCard__area-tag _saleCardTagArea"><a class="saleCard__tag" href="/products/sales/list/?type=hashtag&value=글루텐프리">#글루텐프리</a> <a class="saleCard__tag" href="/products/sales/list/?type=hashtag&value=그레인프리">#그레인프리</a> <a class="saleCard__tag" href="/products/sales/list/?type=hashtag&value=LID사료">#LID사료</a></p><a class="saleCard__link" href="/products/sales/1500/"><div class="saleCard__area-price"><ins class="saleCard__price--new"><span class="saleCard__price--discount">16%</span> 33,000원</ins> <del class="saleCard__price--old">37,000원</del></div><div class="saleCard__area-review"><div class="review-info__wrap"><i class="review-info__icon"><svg height="16" viewBox="0 0 16 16" width="16" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink"><defs><path d="M6 9.95L2.292 12 3 7.658 0 4.584 4.146 3.95 6 0 7.854 3.95 12 4.584 9 7.658 9.708 12z" id="iconStarYellowA"></path></defs><g fill="none" fill-rule="evenodd" transform="translate(2 2)"><mask fill="#fff" id="iconStarYellowB"><use xlink:href="#iconStarYellowA"></use></mask><use fill="#000" fill-rule="nonzero" xlink:href="#iconStarYellowA"></use><path d="M-2 -2H14V14H-2z" fill="#FA0" mask="url(#iconStarYellowB)"></path></g></svg></i> <em class="review-info__score">5.0</em> <span class="review-info__count">상품평 4</span></div></div></a></div>
####################################
name_div_ctx = product_ctx.find('div',
class_='saleCard__area-content')
if (name_div_ctx != None):
link_list = name_div_ctx.find_all('a', class_='saleCard__link')
for link_ctx in link_list:
brand_ctx = link_ctx.find('p', class_='saleCard__brand')
name_ctx = link_ctx.find('p', class_='saleCard__title')
if (brand_ctx != None):
product_data.crw_brand1 = brand_ctx.get_text().strip()
if (name_ctx != None):
product_data.crw_name = name_ctx.get_text().strip()
if ('href' in link_ctx.attrs):
tmp_product_link = link_ctx.attrs['href'].strip()
if (0 != tmp_product_link.find('http')):
tmp_product_link = '%s%s' % (
self.BASIC_PRODUCT_URL,
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_data.crw_goods_code = split_list[5].strip()
####################################
# 가격
#
# <div class="saleCard__area-price"><ins class="saleCard__price--new"><span class="saleCard__price--discount">16%</span> 33,000원</ins> <del class="saleCard__price--old">37,000원</del></div>
####################################
price_list = product_ctx.find_all('div',
class_='saleCard__area-price')
for price_ctx in price_list:
ins_ctx = price_ctx.find('ins', class_='saleCard__price--new')
del_ctx = price_ctx.find('del', class_='saleCard__price--old')
if (del_ctx != None):
product_data.crw_price = int(
__UTIL__.get_only_digit(del_ctx.get_text().strip()))
if (ins_ctx != None):
crw_price_sale = ins_ctx.get_text().strip()
strip_str = ''
strip_ctx = ins_ctx.find(
'span', class_='saleCard__price--discount')
if (strip_ctx != None):
len_strip_str = len(strip_ctx.get_text().strip())
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(
crw_price_sale[len_strip_str:].strip()))
else:
product_data.crw_price_sale = int(
__UTIL__.get_only_digit(crw_price_sale))
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.process_product_api(product_data)
except Exception as ex:
__LOG__.Error('에러 : set_product_data')
__LOG__.Error(ex)
pass
return is_loop