mirror of
https://github.com/NanmiCoder/MediaCrawler.git
synced 2026-03-04 05:00:47 +08:00
feat: cache impl done
This commit is contained in:
@@ -2,14 +2,14 @@
|
||||
# @Author : relakkes@gmail.com
|
||||
# @Time : 2023/12/2 11:18
|
||||
# @Desc : 爬虫 IP 获取实现
|
||||
# @Url : 现在实现了极速HTTP的接口,官网地址:https://www.jisuhttp.com/?pl=mAKphQ&plan=ZY&kd=Yang
|
||||
# @Url : 快代理HTTP实现,官方文档:https://www.kuaidaili.com/?ref=ldwkjqipvz6c
|
||||
import json
|
||||
from abc import ABC, abstractmethod
|
||||
from typing import Dict, List
|
||||
|
||||
import redis
|
||||
from typing import List
|
||||
|
||||
import config
|
||||
from cache.abs_cache import AbstractCache
|
||||
from cache.cache_factory import CacheFactory
|
||||
from tools import utils
|
||||
|
||||
from .types import IpInfoModel
|
||||
@@ -30,9 +30,9 @@ class ProxyProvider(ABC):
|
||||
pass
|
||||
|
||||
|
||||
class RedisDbIpCache:
|
||||
class IpCache:
|
||||
def __init__(self):
|
||||
self.redis_client = redis.Redis(host=config.REDIS_DB_HOST, password=config.REDIS_DB_PWD)
|
||||
self.cache_client: AbstractCache = CacheFactory.create_cache(cache_type=config.CACHE_TYPE_MEMORY)
|
||||
|
||||
def set_ip(self, ip_key: str, ip_value_info: str, ex: int):
|
||||
"""
|
||||
@@ -42,7 +42,7 @@ class RedisDbIpCache:
|
||||
:param ex:
|
||||
:return:
|
||||
"""
|
||||
self.redis_client.set(name=ip_key, value=ip_value_info, ex=ex)
|
||||
self.cache_client.set(key=ip_key, value=ip_value_info, expire_time=ex)
|
||||
|
||||
def load_all_ip(self, proxy_brand_name: str) -> List[IpInfoModel]:
|
||||
"""
|
||||
@@ -51,13 +51,13 @@ class RedisDbIpCache:
|
||||
:return:
|
||||
"""
|
||||
all_ip_list: List[IpInfoModel] = []
|
||||
all_ip_keys: List[bytes] = self.redis_client.keys(pattern=f"{proxy_brand_name}_*")
|
||||
all_ip_keys: List[str] = self.cache_client.keys(pattern=f"{proxy_brand_name}_*")
|
||||
try:
|
||||
for ip_key in all_ip_keys:
|
||||
ip_value = self.redis_client.get(ip_key)
|
||||
ip_value = self.cache_client.get(ip_key)
|
||||
if not ip_value:
|
||||
continue
|
||||
all_ip_list.append(IpInfoModel(**json.loads(ip_value)))
|
||||
except Exception as e:
|
||||
utils.logger.error("[RedisDbIpCache.load_all_ip] get ip err from redis db", e)
|
||||
utils.logger.error("[IpCache.load_all_ip] get ip err from redis db", e)
|
||||
return all_ip_list
|
||||
|
||||
@@ -1,14 +1,14 @@
|
||||
# -*- coding: utf-8 -*-
|
||||
# @Author : relakkes@gmail.com
|
||||
# @Time : 2024/4/5 09:32
|
||||
# @Desc : 极速HTTP代理提供类实现,官网地址:https://www.jisuhttp.com?pl=zG3Jna
|
||||
# @Desc : 已废弃!!!!!倒闭了!!!极速HTTP 代理IP实现
|
||||
import os
|
||||
from typing import Dict, List
|
||||
from urllib.parse import urlencode
|
||||
|
||||
import httpx
|
||||
|
||||
from proxy import IpGetError, ProxyProvider, RedisDbIpCache
|
||||
from proxy import IpCache, IpGetError, ProxyProvider
|
||||
from proxy.types import IpInfoModel
|
||||
from tools import utils
|
||||
|
||||
@@ -31,7 +31,7 @@ class JiSuHttpProxy(ProxyProvider):
|
||||
"pw": "1", # 是否使用账密验证, 1:是,0:否,否表示白名单验证;默认为0
|
||||
"se": "1", # 返回JSON格式时是否显示IP过期时间, 1:显示,0:不显示;默认为0
|
||||
}
|
||||
self.ip_cache = RedisDbIpCache()
|
||||
self.ip_cache = IpCache()
|
||||
|
||||
async def get_proxies(self, num: int) -> List[IpInfoModel]:
|
||||
"""
|
||||
|
||||
@@ -9,7 +9,7 @@ from typing import Dict, List
|
||||
import httpx
|
||||
from pydantic import BaseModel, Field
|
||||
|
||||
from proxy import IpGetError, IpInfoModel, ProxyProvider, RedisDbIpCache
|
||||
from proxy import IpCache, IpInfoModel, ProxyProvider
|
||||
from proxy.types import ProviderNameEnum
|
||||
from tools import utils
|
||||
|
||||
@@ -58,7 +58,7 @@ class KuaiDaiLiProxy(ProxyProvider):
|
||||
self.api_base = "https://dps.kdlapi.com/"
|
||||
self.secret_id = kdl_secret_id
|
||||
self.signature = kdl_signature
|
||||
self.ip_cache = RedisDbIpCache()
|
||||
self.ip_cache = IpCache()
|
||||
self.proxy_brand_name = ProviderNameEnum.KUAI_DAILI_PROVIDER.value
|
||||
self.params = {
|
||||
"secret_id": self.secret_id,
|
||||
|
||||
Reference in New Issue
Block a user