SHOW:
|
|
- or go back to the newest paste.
| 1 | - | fact = [] |
| 1 | + | def recursive_request(urlb, attempts): |
| 2 | data = None | |
| 3 | response = getHtml(urlb) | |
| 4 | print(urlb + " #" + str(attempts)) | |
| 5 | if response is not None: | |
| 6 | if response.status_code == 200: | |
| 7 | soup = BeautifulSoup(response.text, 'html.parser') | |
| 8 | # print(response) | |
| 9 | # print(soup) | |
| 10 | h1 = soup.find('h1')
| |
| 11 | captcha = h1.find('br')
| |
| 12 | print(h1.text + " #" + str(attempts)) | |
| 13 | ||
| 14 | if captcha is None: | |
| 15 | fact = [] | |
| 16 | div = soup.find_all('div', {'class': 'params'})
| |
| 17 | if len(div) > 0: | |
| 18 | div = div[len(div) - 1] | |
| 19 | info = div.find('table', {'class': 'info'})
| |
| 20 | ||
| 21 | if info is not None: | |
| 22 | td = (soup.find_all('td', {'class': 'card_info'}))
| |
| 23 | if td is not None: | |
| 24 | td = td[2].find('table')
| |
| 25 | if td is not None: | |
| 26 | td = td.find_all('a')
| |
| 27 | if td is not None or td is not []: | |
| 28 | for t in td: | |
| 29 | fact.append(t.text) | |
| 30 | ||
| 31 | tr = info.find_all('tr')
| |
| 32 | - | return data |
| 32 | + | |
| 33 | # print(h1.find('nobr').text)
| |
| 34 | # print((tr[0].find_all('td'))[1].text)
| |
| 35 | # print((tr[1].find_all('td'))[1].text)
| |
| 36 | ||
| 37 | data = {
| |
| 38 | 'mark': h1.find('nobr').text,
| |
| 39 | 'weight': (tr[0].find_all('td'))[1].text,
| |
| 40 | 'diam': (tr[1].find_all('td'))[1].text,
| |
| 41 | "factories": fact | |
| 42 | } | |
| 43 | print("OOOKKKKK")
| |
| 44 | print(data) | |
| 45 | print("---------------------------------------------------------------------------------->")
| |
| 46 | return attempts | |
| 47 | ||
| 48 | else: | |
| 49 | print("Cannot connect #" + str(attempts))
| |
| 50 | recursive_request(urlb, attempts) | |
| 51 | attempts -= 1 | |
| 52 | if attempts == 0: | |
| 53 | return data | |
| 54 | else: | |
| 55 | recursive_request(urlb, attempts) |
