Spaces:
Paused
Paused
Delete proxy_pool.py
Browse files- proxy_pool.py +0 -175
proxy_pool.py
DELETED
|
@@ -1,175 +0,0 @@
|
|
| 1 |
-
import requests
|
| 2 |
-
import time
|
| 3 |
-
import logging
|
| 4 |
-
import random
|
| 5 |
-
from threading import Lock, RLock
|
| 6 |
-
from collections import defaultdict
|
| 7 |
-
from datetime import datetime, timedelta
|
| 8 |
-
|
| 9 |
-
logger = logging.getLogger(__name__)
|
| 10 |
-
|
| 11 |
-
class ProxyPool:
|
| 12 |
-
def __init__(self):
|
| 13 |
-
self.proxies = []
|
| 14 |
-
self.proxy_usage = defaultdict(list)
|
| 15 |
-
self.lock = RLock()
|
| 16 |
-
self.last_fetch_time = datetime.min
|
| 17 |
-
self.fetch_interval = 300
|
| 18 |
-
self.last_cleanup_time = time.time()
|
| 19 |
-
self.cleanup_interval = 60
|
| 20 |
-
|
| 21 |
-
def _cleanup_usage_records(self):
|
| 22 |
-
"""清理过期的使用记录"""
|
| 23 |
-
current_time = time.time()
|
| 24 |
-
if current_time - self.last_cleanup_time < self.cleanup_interval:
|
| 25 |
-
return
|
| 26 |
-
|
| 27 |
-
with self.lock:
|
| 28 |
-
for ip in list(self.proxy_usage.keys()):
|
| 29 |
-
self.proxy_usage[ip] = [t for t in self.proxy_usage[ip]
|
| 30 |
-
if current_time - t < 1.0]
|
| 31 |
-
if not self.proxy_usage[ip]:
|
| 32 |
-
del self.proxy_usage[ip]
|
| 33 |
-
self.last_cleanup_time = current_time
|
| 34 |
-
|
| 35 |
-
def _parse_proxy_list(self, text):
|
| 36 |
-
"""解析代理列表文本"""
|
| 37 |
-
proxies = []
|
| 38 |
-
try:
|
| 39 |
-
lines = [line.strip() for line in text.strip().split(',') if line.strip()]
|
| 40 |
-
|
| 41 |
-
for line in lines:
|
| 42 |
-
try:
|
| 43 |
-
if line.startswith('http://'):
|
| 44 |
-
proxy_type = 'http'
|
| 45 |
-
proxy = line[7:]
|
| 46 |
-
elif line.startswith('socks4://'):
|
| 47 |
-
proxy_type = 'socks4'
|
| 48 |
-
proxy = line[9:]
|
| 49 |
-
elif line.startswith('socks5://'):
|
| 50 |
-
proxy_type = 'socks5'
|
| 51 |
-
proxy = line[9:]
|
| 52 |
-
else:
|
| 53 |
-
continue
|
| 54 |
-
|
| 55 |
-
if ':' in proxy:
|
| 56 |
-
ip, port = proxy.split(':')
|
| 57 |
-
proxies.append({
|
| 58 |
-
'type': proxy_type,
|
| 59 |
-
'ip': ip.strip(),
|
| 60 |
-
'port': int(port.strip())
|
| 61 |
-
})
|
| 62 |
-
except Exception as e:
|
| 63 |
-
logger.error(f"解析代理失败: {line}, 错误: {e}")
|
| 64 |
-
continue
|
| 65 |
-
|
| 66 |
-
except Exception as e:
|
| 67 |
-
logger.error(f"解析代理列表失败: {e}")
|
| 68 |
-
return proxies
|
| 69 |
-
|
| 70 |
-
def _fetch_proxies(self):
|
| 71 |
-
"""从API获取代理列表"""
|
| 72 |
-
try:
|
| 73 |
-
urls = [
|
| 74 |
-
"https://269900.xyz/fetch_http?count=5",
|
| 75 |
-
"https://269900.xyz/fetch_https?count=5",
|
| 76 |
-
"https://269900.xyz/fetch_socks5?count=5"
|
| 77 |
-
]
|
| 78 |
-
|
| 79 |
-
new_proxies = []
|
| 80 |
-
for url in urls:
|
| 81 |
-
try:
|
| 82 |
-
response = requests.get(url, timeout=5)
|
| 83 |
-
if response.status_code == 200:
|
| 84 |
-
proxies = self._parse_proxy_list(response.text)
|
| 85 |
-
new_proxies.extend(proxies)
|
| 86 |
-
logger.info(f"成功获取 {len(proxies)} 个代理从 {url}")
|
| 87 |
-
except requests.RequestException as e:
|
| 88 |
-
logger.error(f"获取代理失败 ({url}): {e}")
|
| 89 |
-
continue
|
| 90 |
-
|
| 91 |
-
return new_proxies
|
| 92 |
-
except Exception as e:
|
| 93 |
-
logger.error(f"获取代理失败: {e}")
|
| 94 |
-
return []
|
| 95 |
-
|
| 96 |
-
def update_proxy_pool(self):
|
| 97 |
-
"""更新代理池"""
|
| 98 |
-
current_time = datetime.now()
|
| 99 |
-
should_update = False
|
| 100 |
-
|
| 101 |
-
with self.lock:
|
| 102 |
-
if (current_time - self.last_fetch_time).total_seconds() >= self.fetch_interval:
|
| 103 |
-
should_update = True
|
| 104 |
-
self.last_fetch_time = current_time
|
| 105 |
-
|
| 106 |
-
if should_update:
|
| 107 |
-
new_proxies = self._fetch_proxies()
|
| 108 |
-
if new_proxies:
|
| 109 |
-
with self.lock:
|
| 110 |
-
self.proxies = new_proxies
|
| 111 |
-
logger.info(f"代理池更新成功,当前代理数量: {len(self.proxies)}")
|
| 112 |
-
|
| 113 |
-
def can_use_proxy(self, proxy_ip):
|
| 114 |
-
"""检查代理IP是否可以使用"""
|
| 115 |
-
current_time = time.time()
|
| 116 |
-
|
| 117 |
-
with self.lock:
|
| 118 |
-
usage_times = self.proxy_usage[proxy_ip]
|
| 119 |
-
usage_times = [t for t in usage_times if current_time - t < 1.0]
|
| 120 |
-
self.proxy_usage[proxy_ip] = usage_times
|
| 121 |
-
|
| 122 |
-
if len(usage_times) >= 5:
|
| 123 |
-
return False
|
| 124 |
-
|
| 125 |
-
self.proxy_usage[proxy_ip].append(current_time)
|
| 126 |
-
return True
|
| 127 |
-
|
| 128 |
-
def get_proxy(self):
|
| 129 |
-
"""获取一个可用的代理"""
|
| 130 |
-
self._cleanup_usage_records()
|
| 131 |
-
self.update_proxy_pool()
|
| 132 |
-
|
| 133 |
-
for _ in range(3):
|
| 134 |
-
with self.lock:
|
| 135 |
-
if not self.proxies:
|
| 136 |
-
return None
|
| 137 |
-
|
| 138 |
-
available_proxies = []
|
| 139 |
-
for proxy in self.proxies:
|
| 140 |
-
proxy_ip = proxy.get('ip')
|
| 141 |
-
if proxy_ip and self.can_use_proxy(proxy_ip):
|
| 142 |
-
available_proxies.append(proxy)
|
| 143 |
-
|
| 144 |
-
if available_proxies:
|
| 145 |
-
proxy = random.choice(available_proxies)
|
| 146 |
-
proxy_type = proxy.get('type', 'http').lower()
|
| 147 |
-
proxy_str = f"{proxy_type}://{proxy['ip']}:{proxy['port']}"
|
| 148 |
-
return {
|
| 149 |
-
'http': proxy_str,
|
| 150 |
-
'https': proxy_str if proxy_type == 'http' else None
|
| 151 |
-
}
|
| 152 |
-
|
| 153 |
-
time.sleep(0.1)
|
| 154 |
-
|
| 155 |
-
return None
|
| 156 |
-
|
| 157 |
-
def remove_proxy(self, proxy):
|
| 158 |
-
"""移除无效的代理"""
|
| 159 |
-
if not proxy:
|
| 160 |
-
return
|
| 161 |
-
|
| 162 |
-
with self.lock:
|
| 163 |
-
try:
|
| 164 |
-
proxy_str = proxy.get('http') or proxy.get('https')
|
| 165 |
-
if not proxy_str:
|
| 166 |
-
return
|
| 167 |
-
|
| 168 |
-
for p in self.proxies[:]:
|
| 169 |
-
current_proxy_str = f"{p['type']}://{p['ip']}:{p['port']}"
|
| 170 |
-
if current_proxy_str == proxy_str:
|
| 171 |
-
self.proxies.remove(p)
|
| 172 |
-
logger.info(f"移除无效代理: {proxy_str}")
|
| 173 |
-
break
|
| 174 |
-
except Exception as e:
|
| 175 |
-
logger.error(f"移除代理时出错: {e}")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|