def get_products(cat, subcat, url):
print('Downloading from %s' % url)
rez = requests.get(url, verify=False)
soup = BeautifulSoup(rez.text, 'html.parser')
for item in soup.findAll('div', {'class': 'company_pic'}):
img = item.find('img')
in_stop = False
# отсеиваем ненужное
for w in STOP_WORDS:
if img.get('title').find(w) > -1:
in_stop = True
if img.get('src').find('no_image') > -1:
in_stop = True
if not in_stop:
print(img.get('title'))
pr = Product()
pr.category = cat
pr.name = img.get('title')
pr.subcategory = subcat
img_url = 'https://gastronoma.net/%s' % img.get('src')
img_response = requests.get(img_url, stream=True, verify=False)
# сохраняем файл
with open('tmp.png', 'wb') as out_file:
shutil.copyfileobj(img_response.raw, out_file)
# читаем временный файл и загружаем его программно в модель
with open('%s/tmp.png' % BASE_DIR, 'rb') as img_file:
pr.image.save('cat.png', File(img_file), save=True)
pr.save()
def handle(self, *args, **options):
print('Clearing DB')
Category.objects.all().delete()
Product.objects.all().delete()
print('Start importing from excel %s' % DATA_DIR)
wb = load_workbook(DATA_DIR + '/price.xlsx')
sheet = wb.get_sheet_by_name(wb.get_sheet_names()[0])
cat = None
for cnt in range(1, sheet.max_row, 1):
item = sheet.cell(row=cnt, column=3).value
id = sheet.cell(row=cnt, column=2).value
if id == None:
print('Create a new category')
cat = Category()
cat.name = item
cat.save()
else:
print('Create a new good')
if cat:
p = Product()
p.name = item
p.category = cat
p.save()
def post(self, request, format=None):
cat = Category.objects.get(pk=request.data.get('cat'))
subcat = SubCategory.objects.get(pk=request.data.get('subcat'))
p = Product()
p.category = cat
p.user = request.user.userprofile
p.subcategory = subcat
p.name = request.data.get('name')
if "image" in request.data:
p.image = (request.data['image'])
if "image_base64" in request.data:
try:
format, imgstr = request.data.get('image_base64').split(
';base64,')
ext = format.split('/')[-1]
data = ContentFile(base64.b64decode(imgstr))
file_name = '%s_user.%s' % (p.id, ext)
p.image.save(file_name, data, save=True)
except:
pass
p.save()
return Response(ProductSerializer(p).data)
def get_product_description(link, category):
URL = 'https://tainabox.com.ua'
print('Start importing from %s' % URL + link)
# rez = requests.get(URL + link, verify=False)
rez = requests.get(URL + link)
soup = BeautifulSoup(rez.text, 'html.parser')
for desc in soup.findAll('div', {'class': 'product__big-item'}):
image = desc.find('div', {
'class': 'product__big-item_right'
}).find('img')
consist = desc.find('div',
{'class': 'product__item__composition__value'})
price = desc.find('div', {'class': 'to-order__value'})
name = desc.find('div', {'product__big-item__name'})
in_stop = False
for w in STOP_WORDS:
if name.text.find(w) > -1:
in_stop = True
if consist.text.find(w) > -1:
in_stop = True
if not in_stop:
p = Product()
p.name = re.sub('\n', '', name.text)
p.price = re.split(r'\n', price.text)[0]
p.consist = consist.text
p.category = category
img_url = URL + image['src']
img_temp = NamedTemporaryFile(delete=True)
req = urllib.request.Request(
img_url,
data=None,
headers={
'User-Agent':
'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_3) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/35.0.1916.47 Safari/537.36'
})
img_temp.write(urllib.request.urlopen(req).read())
img_temp.flush()
img_res = re.split(r'\.', image['src'])
p.image.save("image_.{}".format(img_res[-1]), File(img_temp))
p.save()
def get_products(cat, subcat, url):
try:
shutil.rmtree('%s/media/category/subcategory' %
BASE_DIR) # очищает католог от картинок
except:
pass
print('Downloading fm %s' % url)
requests.packages.urllib3.disable_warnings(category=InsecureRequestWarning)
rez = requests.get(url, verify=False)
soup = BeautifulSoup(rez.text, 'html.parser')
for item in soup.findAll('div', {'class': 'company_pic'}):
img = item.find('img')
in_stop = False
# отсеиваем ненужное
for w in STOP_WORDS:
if img.get('title').find(w) > -1:
in_stop = True
if img.get('src').find('no_image') > -1:
in_stop = True
if not in_stop:
print(img.get('title'))
pr = Product()
pr.category = cat
pr.name = img.get('title')
pr.subcategory = subcat
img_url = 'https://gastronoma.net/%s' % img.get('src')
requests.packages.urllib3.disable_warnings(
category=InsecureRequestWarning)
img_response = requests.get(img_url, stream=True, verify=False)
# сохраняем временный файл
with open(
'tmp.png', 'wb'
) as out_file: # создается временный файл 'b' - binary, для записи 'w'-write
shutil.copyfileobj(img_response.raw, out_file)
# читаем временный файл и загружаем его программно в модель
with open('%s/tmp.png' % BASE_DIR, 'rb') as img_file:
pr.image.save('product.png', File(img_file), save=True)
pr.save()
def handle(self, *args, **options):
print('Clear DB')
Category.objects.all().delete()
Product.objects.all().delete()
print('Start import from excel %s' % DATA_DIR)
wb = load_workbook((DATA_DIR + '\\price.xlsx'))
worksheet = wb.get_sheet_by_name(wb.get_sheet_names()[0])
category = None
for cnt in range(1, worksheet.max_row + 1):
item = worksheet.cell(row=cnt, column=5).value
cat = worksheet.cell(row=cnt, column=1).value
if item == None:
category = Category()
category.name = cat
category.save()
print('Create category')
else:
if category:
product = Product()
product.name = item
product.category = category
product.save()
print('Create item')