Files
clients/WebBasedCrawler/base/proxy2.py
mjjo 1fb61f0b4c 트위터 크롤러 수정
- 프록시를 porxy2 db에 넣고 사용
2017-08-09 15:32:57 +09:00

181 lines
5.1 KiB
Python

import base.proxy_crawler as proxy_crawler
import base.logger as logger
import sqlalchemy
import sqlalchemy.ext
import sqlalchemy.ext.declarative
import sqlalchemy.orm
import enum
import datetime
import threading
import random
Base = sqlalchemy.ext.declarative.declarative_base()
class Proxy2Model(Base):
__tablename__ = 'proxy2'
id = sqlalchemy.Column(sqlalchemy.Integer, primary_key=True, nullable=False, autoincrement=True)
ip = sqlalchemy.Column(sqlalchemy.String(15), primary_key=True)
port = sqlalchemy.Column(sqlalchemy.SmallInteger, primary_key=True)
create_at = sqlalchemy.Column(sqlalchemy.DateTime, default=datetime.datetime.now)
naver_block_at = sqlalchemy.Column(sqlalchemy.DateTime)
daum_block_at = sqlalchemy.Column(sqlalchemy.DateTime)
facebook_block_at = sqlalchemy.Column(sqlalchemy.DateTime)
kakao_block_at = sqlalchemy.Column(sqlalchemy.DateTime)
insta_block_at = sqlalchemy.Column(sqlalchemy.DateTime)
twitter_block_at = sqlalchemy.Column(sqlalchemy.DateTime)
youtube_block_at = sqlalchemy.Column(sqlalchemy.DateTime)
def __init__(self, ip, port):
self.ip = ip
self.port = port
def __repr__(self):
return '{}:{}'.format(self.ip, self.port)
def get_instance_for_http(self):
return {
'http': '{}:{}'.format(self.ip, self.port),
'https': '{}:{}'.format(self.ip, self.port),
'ip': self.ip,
'port': self.port,
}
class Platform(enum.Enum):
NAVER = 'naver'
DAUM = 'daum'
FACEBOOK = 'facebook'
KAKAO = 'kakao'
INSTA = 'insta'
TWITTER = 'twitter'
YOUTUBE = 'youtube'
class Proxy2Handler:
block_field_map = {
Platform.NAVER: Proxy2Model.naver_block_at,
Platform.DAUM: Proxy2Model.daum_block_at,
Platform.FACEBOOK: Proxy2Model.facebook_block_at,
Platform.KAKAO: Proxy2Model.kakao_block_at,
Platform.INSTA: Proxy2Model.insta_block_at,
Platform.TWITTER: Proxy2Model.twitter_block_at,
Platform.YOUTUBE: Proxy2Model.youtube_block_at,
}
def __init__(self):
self.lock = threading.Lock()
self.engine = sqlalchemy.create_engine('mysql+pymysql://admin:admin123@bigbird.iptime.org/concepters?charset=utf8')
session_factory = sqlalchemy.orm.sessionmaker(bind=self.engine)
self.session = sqlalchemy.orm.scoped_session(session_factory)
def lock_enter(self):
# logger.log('lock {}'.format(threading.current_thread().ident))
# self.lock.acquire()
pass
def lock_leave(self):
# self.lock.release()
# logger.log('unlock {}'.format(threading.current_thread().ident))
pass
def commit(self):
self.lock_enter()
self.session.commit()
self.lock_leave()
def get_oldest(self, platform):
self.lock_enter()
instance = self.session.query(Proxy2Model).order_by(self.block_field_map[platform].desc()).first()
self.lock_leave()
return instance
# def get(self, platform):
# proxy = self.session.query(Proxy2Model).filter(self.block_field_map[platform] == None).first()
# if not proxy:
# proxy_crawler.crawl_proxies()
#
# proxy = self.get_oldest(platform)
#
# return proxy
def get_query(self, ip, port):
return self.session.query(Proxy2Model).filter_by(ip=ip).filter_by(port=port)
def get_instance(self, ip, port):
self.lock_enter()
instance = self.get_query(ip, port).first()
self.lock_leave()
return instance
def get(self, platform, proc_id=-1):
self.lock_enter()
block_column = self.block_field_map[platform]
try:
instances = self.session.query(Proxy2Model).filter(block_column == None).all()
except Exception as e:
self.lock_leave()
try:
session_factory = sqlalchemy.orm.sessionmaker(bind=self.engine)
self.session = sqlalchemy.orm.scoped_session(session_factory)
except Exception as e2:
logger.log('{} session recreate'.format(proc_id))
return None
instance = instances[random.randint(0, len(instances)-1)] if len(instances) > 0 else None
if instance:
self.lock_leave()
return instance.get_instance_for_http()
else:
proxies = proxy_crawler.crawl_proxies()
self.insert_all(proxies)
self.unlock()
return self.get(platform, proc_id)
def insert(self, ip, port):
instance = self.get_instance(ip, port)
if not instance:
proxy = Proxy2Model(ip, port)
self.lock_enter()
self.session.add(proxy)
self.lock_leave()
def insert_all(self, proxies):
# INSERT INTO proxy2(ip, PORT)
# SELECT <ip>, <port> FROM DUAL
# WHERE NOT EXISTS (SELECT * FROM proxy2 WHERE ip=<ip> AND PORT=<port>)
instances_add = []
for proxy in proxies:
instance = self.get_instance(proxy['ip'], proxy['port'])
if not instance:
instances_add.append(Proxy2Model(proxy['ip'], proxy['port']))
self.session.bulk_save_objects(instances_add)
self.commit()
def set_proxy_blocked(self, ip, port, platform):
block_column = self.block_field_map[platform]
query = self.get_query(ip, port).filter(block_column == None)
query.update({block_column: datetime.datetime.now()})
self.commit()
if __name__ == '__main__':
proxy_handler = Proxy2Handler()
# proxy_handler.insert('127.0.0.5', 80)
# proxy_handler.commit()
# proxy_handler.set_proxy_blocked('127.0.0.3', 80, Platform.TWITTER)
# instance = proxy_handler.get(Platform.TWITTER)
# instance = proxy_handler.get_oldest(Platform.TWITTER)
# print(instance)
proxy = proxy_handler.get(Platform.TWITTER)