olivia_finder.myrequests.proxy_builders.ssl_proxies

 1from typing import List
 2from typing_extensions import override
 3from bs4 import BeautifulSoup
 4import requests
 5from .proxy_builder import ProxyBuilder
 6
 7
 8class SSLProxiesBuilder(ProxyBuilder):
 9    '''
10    SSLProxies proxy builder, gets the proxies from https://www.sslproxies.org/
11
12    Implements the abstract method `ProxyBuilder._parse_request()` to parse the response 
13    '''
14    
15    def __init__(self, request_timeout: int = 60):
16        super().__init__(
17            url="https://www.sslproxies.org/", 
18            request_timeout=request_timeout
19        )
20    
21    @override
22    def _parse_request(self, response: requests.Response) -> List[str]:
23        '''
24        Parse the response and return a list of proxies
25        Expected data:
26
27        <tbody>
28            <tr>
29                <td>35.247.248.104</td>
30                <td>3129</td><td>BR</td>
31                <td class="hm">Brazil</td>
32                <td>anonymous</td>
33                <td class="hm"></td>
34                <td class="hx">yes</td>
35                <td class="hm">22 secs ago</td>
36            </tr>
37            ...
38        </tbody>
39
40        Parameters
41        ----------
42        response : requests.Response
43            The response of the request
44
45        Returns
46        -------
47        List[str]
48            A list of proxies            
49        '''
50        
51        proxies = []
52
53        # Get the table using beautiful soup
54        soup = BeautifulSoup(response.text, 'html.parser')
55        tbody = soup.find('tbody')
56
57        # iterate over the rows getting the data
58        rows = tbody.find_all('tr')
59        for row in rows:
60            cols = row.find_all('td')
61
62            # add the proxy to the list
63            ip = cols[0].text
64            port = cols[1].text
65            proxies.append(f"{ip}:{port}")
66        
67        return proxies