Update proxy_pool.py
Browse files- proxy_pool.py +38 -79
proxy_pool.py
CHANGED
|
@@ -1,7 +1,3 @@
|
|
| 1 |
-
"""
|
| 2 |
-
Proxy Pool — avec merge intelligent
|
| 3 |
-
"""
|
| 4 |
-
|
| 5 |
import random
|
| 6 |
import threading
|
| 7 |
import time
|
|
@@ -12,7 +8,6 @@ class ProxyEntry:
|
|
| 12 |
__slots__ = (
|
| 13 |
'proxy', 'protocol', 'proxy_url', 'latency',
|
| 14 |
'proxy_ip', 'verified', 'fail_count', 'total_used',
|
| 15 |
-
'last_seen', 'test_count',
|
| 16 |
)
|
| 17 |
|
| 18 |
def __init__(self, proxy, protocol, proxy_url,
|
|
@@ -25,8 +20,6 @@ class ProxyEntry:
|
|
| 25 |
self.verified = verified
|
| 26 |
self.fail_count = 0
|
| 27 |
self.total_used = 0
|
| 28 |
-
self.last_seen = time.time()
|
| 29 |
-
self.test_count = 1
|
| 30 |
|
| 31 |
def to_dict(self):
|
| 32 |
return {
|
|
@@ -37,8 +30,6 @@ class ProxyEntry:
|
|
| 37 |
"proxy_ip": self.proxy_ip,
|
| 38 |
"verified": self.verified,
|
| 39 |
"total_used": self.total_used,
|
| 40 |
-
"age_s": round(time.time() - self.last_seen, 1),
|
| 41 |
-
"test_count": self.test_count,
|
| 42 |
}
|
| 43 |
|
| 44 |
|
|
@@ -50,69 +41,28 @@ class ProxyPool:
|
|
| 50 |
self._rr = defaultdict(int)
|
| 51 |
self._lr = 0.0
|
| 52 |
self._rc = 0
|
| 53 |
-
# config
|
| 54 |
-
self.max_age = 600 # 10 min max sans retest
|
| 55 |
-
self.max_pool_size = 500 # pas plus de 500 proxies
|
| 56 |
|
| 57 |
def refresh(self, new_proxies):
|
| 58 |
-
"""
|
| 59 |
-
MERGE les nouveaux proxies avec les existants.
|
| 60 |
-
- Nouveaux: ajoutés/mis à jour
|
| 61 |
-
- Anciens pas retestés: gardés si < max_age
|
| 62 |
-
- Très vieux: supprimés
|
| 63 |
-
"""
|
| 64 |
with self._lk:
|
| 65 |
-
|
| 66 |
-
|
| 67 |
-
# 1. Update/add les nouveaux
|
| 68 |
for px in new_proxies:
|
| 69 |
-
|
| 70 |
-
|
| 71 |
-
|
| 72 |
-
|
| 73 |
-
|
| 74 |
-
|
| 75 |
-
|
| 76 |
-
|
| 77 |
-
e.
|
| 78 |
-
e.
|
| 79 |
-
|
| 80 |
-
|
| 81 |
-
|
| 82 |
-
|
| 83 |
-
|
| 84 |
-
|
| 85 |
-
px.get("proxy_ip", ""),
|
| 86 |
-
px.get("verified", False),
|
| 87 |
-
)
|
| 88 |
-
e.last_seen = now
|
| 89 |
-
self._um[url] = e
|
| 90 |
-
|
| 91 |
-
# 2. Nettoyer les vieux proxies
|
| 92 |
-
to_remove = []
|
| 93 |
-
for url, e in self._um.items():
|
| 94 |
-
age = now - e.last_seen
|
| 95 |
-
if age > self.max_age:
|
| 96 |
-
to_remove.append(url)
|
| 97 |
-
|
| 98 |
-
for url in to_remove:
|
| 99 |
-
del self._um[url]
|
| 100 |
-
|
| 101 |
-
# 3. Reconstruire la liste triée
|
| 102 |
-
self._px = sorted(
|
| 103 |
-
self._um.values(),
|
| 104 |
-
key=lambda x: x.latency
|
| 105 |
-
)
|
| 106 |
-
|
| 107 |
-
# 4. Limiter la taille
|
| 108 |
-
if len(self._px) > self.max_pool_size:
|
| 109 |
-
# garder les plus rapides
|
| 110 |
-
removed = self._px[self.max_pool_size:]
|
| 111 |
-
self._px = self._px[:self.max_pool_size]
|
| 112 |
-
for e in removed:
|
| 113 |
-
del self._um[e.proxy_url]
|
| 114 |
-
|
| 115 |
-
self._lr = now
|
| 116 |
self._rc += 1
|
| 117 |
|
| 118 |
@property
|
|
@@ -175,7 +125,6 @@ class ProxyPool:
|
|
| 175 |
p = self._um.get(url)
|
| 176 |
if p:
|
| 177 |
p.fail_count = 0
|
| 178 |
-
p.last_seen = time.time()
|
| 179 |
|
| 180 |
def get_all(self, protocol=None, verified=False, limit=500):
|
| 181 |
with self._lk:
|
|
@@ -186,26 +135,36 @@ class ProxyPool:
|
|
| 186 |
|
| 187 |
def get_stats(self):
|
| 188 |
with self._lk:
|
| 189 |
-
now = time.time()
|
| 190 |
bp = defaultdict(lambda: {"total": 0, "alive": 0})
|
| 191 |
-
ages = []
|
| 192 |
for p in self._px:
|
| 193 |
bp[p.protocol]["total"] += 1
|
| 194 |
if p.fail_count < 3:
|
| 195 |
bp[p.protocol]["alive"] += 1
|
| 196 |
-
ages.append(now - p.last_seen)
|
| 197 |
-
|
| 198 |
alive = sum(d["alive"] for d in bp.values())
|
| 199 |
-
avg_age = round(sum(ages) / len(ages), 1) if ages else 0
|
| 200 |
-
|
| 201 |
return {
|
| 202 |
"total": len(self._px),
|
| 203 |
"alive": alive,
|
| 204 |
"by_protocol": dict(bp),
|
| 205 |
-
"avg_age_s": avg_age,
|
| 206 |
-
"max_age_config": self.max_age,
|
| 207 |
"refresh_count": self._rc,
|
| 208 |
"last_refresh_ago": round(
|
| 209 |
-
|
| 210 |
) if self._lr else None,
|
| 211 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
import random
|
| 2 |
import threading
|
| 3 |
import time
|
|
|
|
| 8 |
__slots__ = (
|
| 9 |
'proxy', 'protocol', 'proxy_url', 'latency',
|
| 10 |
'proxy_ip', 'verified', 'fail_count', 'total_used',
|
|
|
|
| 11 |
)
|
| 12 |
|
| 13 |
def __init__(self, proxy, protocol, proxy_url,
|
|
|
|
| 20 |
self.verified = verified
|
| 21 |
self.fail_count = 0
|
| 22 |
self.total_used = 0
|
|
|
|
|
|
|
| 23 |
|
| 24 |
def to_dict(self):
|
| 25 |
return {
|
|
|
|
| 30 |
"proxy_ip": self.proxy_ip,
|
| 31 |
"verified": self.verified,
|
| 32 |
"total_used": self.total_used,
|
|
|
|
|
|
|
| 33 |
}
|
| 34 |
|
| 35 |
|
|
|
|
| 41 |
self._rr = defaultdict(int)
|
| 42 |
self._lr = 0.0
|
| 43 |
self._rc = 0
|
|
|
|
|
|
|
|
|
|
| 44 |
|
| 45 |
def refresh(self, new_proxies):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 46 |
with self._lk:
|
| 47 |
+
old = dict(self._um)
|
| 48 |
+
fresh, um = [], {}
|
|
|
|
| 49 |
for px in new_proxies:
|
| 50 |
+
e = ProxyEntry(
|
| 51 |
+
px["proxy"], px["protocol"],
|
| 52 |
+
px["proxy_url"], px["latency"],
|
| 53 |
+
px.get("proxy_ip", ""),
|
| 54 |
+
px.get("verified", False),
|
| 55 |
+
)
|
| 56 |
+
o = old.get(e.proxy_url)
|
| 57 |
+
if o:
|
| 58 |
+
e.total_used = o.total_used
|
| 59 |
+
e.fail_count = max(0, o.fail_count - 1)
|
| 60 |
+
fresh.append(e)
|
| 61 |
+
um[e.proxy_url] = e
|
| 62 |
+
fresh.sort(key=lambda x: x.latency)
|
| 63 |
+
self._px = fresh
|
| 64 |
+
self._um = um
|
| 65 |
+
self._lr = time.time()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 66 |
self._rc += 1
|
| 67 |
|
| 68 |
@property
|
|
|
|
| 125 |
p = self._um.get(url)
|
| 126 |
if p:
|
| 127 |
p.fail_count = 0
|
|
|
|
| 128 |
|
| 129 |
def get_all(self, protocol=None, verified=False, limit=500):
|
| 130 |
with self._lk:
|
|
|
|
| 135 |
|
| 136 |
def get_stats(self):
|
| 137 |
with self._lk:
|
|
|
|
| 138 |
bp = defaultdict(lambda: {"total": 0, "alive": 0})
|
|
|
|
| 139 |
for p in self._px:
|
| 140 |
bp[p.protocol]["total"] += 1
|
| 141 |
if p.fail_count < 3:
|
| 142 |
bp[p.protocol]["alive"] += 1
|
|
|
|
|
|
|
| 143 |
alive = sum(d["alive"] for d in bp.values())
|
|
|
|
|
|
|
| 144 |
return {
|
| 145 |
"total": len(self._px),
|
| 146 |
"alive": alive,
|
| 147 |
"by_protocol": dict(bp),
|
|
|
|
|
|
|
| 148 |
"refresh_count": self._rc,
|
| 149 |
"last_refresh_ago": round(
|
| 150 |
+
time.time() - self._lr, 1
|
| 151 |
) if self._lr else None,
|
| 152 |
+
}
|
| 153 |
+
|
| 154 |
+
def export_json(self):
|
| 155 |
+
with self._lk:
|
| 156 |
+
return [p.to_dict() for p in self._px]
|
| 157 |
+
|
| 158 |
+
def import_json(self, data):
|
| 159 |
+
proxies = []
|
| 160 |
+
for p in data:
|
| 161 |
+
proxies.append({
|
| 162 |
+
"proxy": p["proxy"],
|
| 163 |
+
"protocol": p["protocol"],
|
| 164 |
+
"proxy_url": p["proxy_url"],
|
| 165 |
+
"latency": p["latency"],
|
| 166 |
+
"proxy_ip": p.get("proxy_ip", ""),
|
| 167 |
+
"verified": p.get("verified", False),
|
| 168 |
+
})
|
| 169 |
+
if proxies:
|
| 170 |
+
self.refresh(proxies)
|