def parse(self, response):
self.log('Scrapping: {}({})'.format(self.brand, response.url))
images = response.css('.models > div.col-4 > a > img::attr(src)').extract()
print('parsing')
for img_url in images:
# skip empty image
if img_url.endswith('no-image-170x113.jpg'):
self.log('Skipping empty image: {}'.format(img_url))
continue
# skip image if it already in DB
try:
Image.objects.get(pk=img_url)
self.log('Image already exists: {}'.format(img_url))
continue
except Image.DoesNotExist:
pass
img = Image(url=img_url,
is_car=True,
test_set=random.randint(1, 100) < 10,
brand=self.brand)
img.save()
print('saved: ', img_url)
self.log('New image saved: {}'.format(img_url))
time.sleep(5)
