SHOW:
|
|
- or go back to the newest paste.
| 1 | import requests | |
| 2 | ||
| 3 | ||
| 4 | def get_page(category: str, page_id: int) -> str: | |
| 5 | if page_id: | |
| 6 | url = 'https://www.ozon.ru/brand/{0}/?page={1}'.format(category, page_id)
| |
| 7 | else: | |
| 8 | url = 'https://www.ozon.ru/brand/{0}/'.format(category)
| |
| 9 | print('get url: {0}'.format(url))
| |
| 10 | response = requests.get(url) | |
| 11 | return response.text | |
| 12 | ||
| 13 | ||
| 14 | def load_data(): | |
| 15 | category_list = ['adidas-144082850', 'puma-87235756'] | |
| 16 | for category in category_list: | |
| 17 | for page_id in range(50): | |
| 18 | text = get_page(category, page_id) | |
| 19 | # обрабатываем полученный текст, сохраняем в файл/базу | |
| 20 | ||
| 21 | ||
| 22 | if __name__ == '__main__': | |
| 23 | load_data() | |
| 24 |
