def get_data(context):
html = get_html(
'https://www.colins.com.tr/urun/893-julia-normal-kesim-orta-bel-kisa-paca-mavi-jean-pantolon-16434'
)
data = {}
soup = BeautifulSoup(html, 'lxml')
base = soup.find('div', id='page').find(
'div', class_='fullContent top-space').find('div').find('div').find(
'div').find(
'div',
class_='row product-details-page').find('div').find('form')
images_soup = base.find(
'div', class_='productDetailLeftPanel col-lg-6 col-sm-6'
).find('div').find('div').find(
'div',
class_='module-frameproduct-picture-wrapper').find('div').find_all('a')
description = ''
try:
tr = (soup.find('div', id='page').find(
'div', class_='fullContent top-space').find('div').find(
'div').find('div').find(
'div',
class_='product-details-mobile row').find('div').find(
'div',
class_='PDTABSCONTENT clearfix tab-content').find(
'div', class_='tab-pane active').find(
'table').find('tbody').find_all('tr')[1])
description = tr.find('td').text.strip()
except:
pass
images = []
for i in images_soup:
images.append(i['data-image'])
right = base.find('div', class_='productDetailRightPanel')
name_soup = right.find('h1')
name = name_soup.text.strip()
others = right.find('div', class_='product-collateral').find('div').find(
'div', class_='variant-wrapper').find('div')
color = others['data-variant-name'].lower().capitalize()
price = others.find('div', class_='productPriceContent clearfix').find(
'span').text.strip().split(' ')[0].replace(',', '.')
variants = others.find(
'div', class_='product-options').find('div').find_all('div')
sizes = []
for i in variants:
classes = i['class']
size = {'value': i.find('label').text.strip()}
if len(classes) == 2:
size['stock'] = False
else:
size['stock'] = True
sizes.append(size)
print(sizes)
return []
def main():
url = 'https://magicbox.izishop.kg/api/v1/project/brands/?brand=bershka'
# url = 'http://127.0.0.1:8000/api/v1/project/brands/zara/'
request_url = 'https://www.bershka.com/itxrest/2/catalog/store/44109521/40259537/category?languageId=-43&typeCatalog=1&appId=2'
html = get_html(request_url)
massimo = json.loads(html)
categories = [massimo['categories'][0], massimo['categories'][3]]
departments = []
for i in categories:
department = {
'department': i['name'],
'link': 'https://www.bershka.com/tr/',
}
collection = None
for j in i['subcategories']:
if j['name'] == 'KOLEKSİYON':
collection = j
break
sub_categories = []
for category in collection['subcategories']:
if len(category['subcategories']
) == 0 and category['viewCategoryId'] > 0:
n_category = {
'category':
category['name'],
'link':
'https://www.bershka.com/tr/%s/koleksi̇yon/%s-c%s.html' %
(str(i['name']).lower(), str(category['name']).replace(
' ', '-').lower(), category['id'])
}
sub_categories.append(n_category)
elif len(category['subcategories']) > 0:
for sub in category['subcategories'][1:]:
n_category = {
'category':
'%s (%s)' % (category['name'], sub['name']),
'link':
'https://www.bershka.com/tr/%s/koleksi̇yon/%s/%s-c%s.html'
%
(str(i['name']).lower(), str(category['name']).replace(
' ', '-').lower(), str(sub['name']).replace(
' ', '-').lower(), sub['id'])
}
sub_categories.append(n_category)
department['categories'] = sub_categories
departments.append(department)
pprint(departments)
headers = {
'Content-type': 'application/json',
'Accept': 'application/json'
}
r = requests.put(url, data=json.dumps(departments), headers=headers)
print(r.status_code)
departments = []
def main():
url = 'https://magicbox.izishop.kg/api/v1/project/brands/?brand=stradivarius'
# url = 'http://127.0.0.1:8000/api/v1/project/brands/?brand=stradivarius'
request_url = 'https://www.stradivarius.com/itxrest/2/catalog/store/54009571/50331081/category?languageId=-43&typeCatalog=1&appId=2'
html = get_html(request_url)
stradivarius = json.loads(html)
categories = stradivarius['categories'][0]['subcategories']
new_categories = [categories[1], categories[3]]
departments = []
department = {
'department': 'Kadin',
'link': 'https://www.stradivarius.com/tr/',
}
sub_categories = []
for i in new_categories:
parent = i['subcategories'][0]
for category in parent['subcategories'][1:]:
if len(category['subcategories']) == 0:
print(category['name'])
n_category = {
'category':
category['name'],
'link':
'https://www.stradivarius.com/tr/yeni-koleksiyon/%s/ürüne-göre-alışveriş/%s-c%s.html'
% (str(i['name']).lower(), str(category['name']).replace(
' ', '-').lower(), category['id'])
}
sub_categories.append(n_category)
elif len(category['subcategories']) > 0:
for sub in category['subcategories'][1:]:
n_category = {
'category':
'%s (%s)' % (category['name'], sub['name']),
'link':
'https://www.stradivarius.com/tr/yeni-koleksi̇yon/%s/ürüne-göre-alışveriş/%s/%s-c%s.html'
%
(str(i['name']).lower(), str(category['name']).replace(
' ', '-').lower(), str(sub['name']).replace(
' ', '-').lower(), sub['id'])
}
sub_categories.append(n_category)
department['categories'] = sub_categories
departments.append(department)
pprint(departments)
headers = {
'Content-type': 'application/json',
'Accept': 'application/json'
}
r = requests.put(url, data=json.dumps(departments), headers=headers)
print(r.status_code)
def parse_brand_data(brand):
data = [
{
'name': 'Kadın',
'link': 'https://www2.hm.com/tr_tr/kadin.html'
},
{
'name': 'Erkek',
'link': 'https://www2.hm.com/tr_tr/erkek.html'
},
{
'name': 'Divided',
'link': 'https://www2.hm.com/tr_tr/divided.html'
},
{
'name': 'Çocuk',
'link': 'https://www2.hm.com/tr_tr/cocuk.html'
},
]
l = []
for i in data:
context = {'department': i['name'], 'link': i['link']}
html = get_html(context['link'])
soup = BeautifulSoup(html, 'lxml')
menu = soup.find('div', class_='sidebar-plus-content').find('aside', class_='page-sidebar').find('nav',
class_='secondary-nav') \
.find('ul', id='menu-links').find_all('li', class_='list-group')
all_categories = []
for i in menu:
menu_item = None
try:
menu_item = i.find('strong',
class_='list-group-title is-opened')
except:
pass
if menu_item is not None:
ul = i.find('ul', class_='menu').find_all('li')
for j in ul:
category = {}
a = j.find('a')
if a['href'] != '/tr_tr/kadin/urune-gore-satin-al/view-all.html' \
or a['href'] != '/tr_tr/erkek/urune-gore-satin-al/view-all.html'\
or a['href'] != '/tr_tr/cocuk/urune-gore-satin-al/view-all.html':
category['link'] = 'https://www2.hm.com' + a['href']
category['name'] = a.text.strip()
all_categories.append(category)
context['categories'] = all_categories
l.append(context)
return l
def get_data():
html = get_html(
'https://www2.hm.com/tr_tr/erkek/urune-gore-satin-al/ayakkabi.html')
soup = BeautifulSoup(html, 'lxml')
menu = soup.find('div', class_='sidebar-plus-content').find('aside').find('nav',
class_='secondary-nav') \
.find('ul', id='menu-links').find_all('li', class_='list-group')
for i in menu:
menu_item = None
try:
menu_item = i.find('strong', class_='list-group-title is-opened')
except:
pass
if menu_item is not None:
ul = (i.find('ul', class_='menu').find(
'li', class_='list-group').find('ul').find_all('li'))
categories = [{
'category': j.find('a').text.strip(),
'link': 'https://www2.hm.com' + j.find('a')['href']
} for j in ul]
print(categories)
return []
def parse_brand_data(brand):
data = [
{
'name': 'Kadın',
'link': 'https://www.colins.com.tr/c/kadin-giyim-57'
},
{
'name': 'Erkek',
'link': 'https://www.colins.com.tr/c/erkek-giyim-2'
},
{
'name': 'Aksesuar',
'link': 'https://www.colins.com.tr/c/aksesuar-465'
},
]
l = []
for i in data:
context = {'department': i['name'], 'link': i['link']}
html = get_html(context['link'])
soup = BeautifulSoup(html, 'lxml')
menu = soup.find('div', id='page').find('div', class_='fullContent top-space').find('div',
class_='container catalog-category') \
.find('div', class_='productFilterContent').find('div',
class_='panel-heading clearfix group-container-Kategori') \
.find('div').find('div').find('ul').find_all('li')
pprint(menu)
categories = []
for i in menu:
category = {
'category': i.find('a').text.strip(),
'link': i.find('a')['href'],
}
categories.append(category)
context['categories'] = categories
l.append(context)
return l
def parse_brand_data(brand):
url = brand['link']
html = get_html(url)
soup = BeautifulSoup(html, 'lxml')
data = soup.find('div', id='nav-menu-group').find('nav', id='menu').find('ul', class_='category-menu') \
.find_all('li', class_='_category-link-wrapper menu-item menu-item--level-1')
data = data[1:5]
l = []
for i in data:
context = {
'department':
i['data-name'],
'link':
i.find('a',
class_='_category-link menu-item__category-link')['href']
}
html = get_html(context['link'])
soup = BeautifulSoup(html, 'lxml')
departments = soup.find('div', id='nav-menu-group').find('nav', id='menu').find('ul', class_='category-menu') \
.find('li',
class_='_category-link-wrapper menu-item menu-item--level-1 menu-item--current menu-item--selected')
ul = departments.find(
'ul',
class_=
'_subcategories subcategory-menu subcategory-menu--level-1 subcategory-menu--current'
)
classes = (ul.find_all(
'li',
class_='_category-link-wrapper menu-item menu-item--level-2'))
categories = []
for classi in classes:
name = classi['data-name']
cats = classi.find('ul', class_='_subcategories subcategory-menu subcategory-menu--level-2') \
.find_all('li', class_='_category-link-wrapper menu-item menu-item--level-3 menu-item--is-leaf')
cats = cats[1:]
for category in cats:
category_data = {
'category': '%s (%s)' % (category['data-name'], name)
}
try:
link = category.find(
'a', class_='_category-link menu-item__category-link'
)['href']
except:
link = category.find(
'a', class_='_category-link menu-item__category-link'
)['data-href']
category_data['link'] = link
categories.append(category_data)
leafs = (ul.find_all(
'li',
class_=
'_category-link-wrapper menu-item menu-item--level-2 menu-item--is-leaf'
))
for leaf in leafs:
class_data = {'category': leaf['data-name']}
try:
link = leaf.find(
'a',
class_='_category-link menu-item__category-link')['href']
except:
link = leaf.find(
'a', class_='_category-link menu-item__category-link'
)['data-href']
class_data['link'] = link
categories.append(class_data)
context['categories'] = categories
l.append(context)
return l