# 數(shù)據(jù)庫(kù)地址
HOST = '127.0.0.1'
# MySql端口
MYSQL_PORT = 3306
# MySQl用戶名、密碼
MYSQL_USERNAME = '***'
MYSQL_PASSWORD = '***'
# 數(shù)據(jù)庫(kù)名
SQL_NAME = 'test'

# 代理等級(jí)
MAX_SCORE = 30
MIN_SCORE = 0
INITIAL_SCORE = 10

# 代理池?cái)?shù)量界限
POOL_UPPER_THRESHOLD = 1000

import pymysql
from error import PoolEmptyError
from setting import *
from random import choice
import re


class MySqlClient(object):
 # 初始化
 def __init__(self, host=HOST, port=MYSQL_PORT, username=MYSQL_USERNAME, password=MYSQL_PASSWORD, sqlname=SQL_NAME):
  self.db = pymysql.connect(host=host, user=username, password=password, port=port, db=sqlname)
  self.cursor = self.db.cursor()

 # 添加代理IP
 def add(self, ip, score=INITIAL_SCORE):
  sql_add = "INSERT INTO PROXY (IP,SCORE) VALUES ('%s', %s)" % (ip, score)
  if not re.match('\d+\.\d+\.\d+\.\d+\:\d+', ip):
   print('代理不符合規(guī)范', ip, '丟棄')
   return
  if not self.exists(ip):
   self.cursor.execute(sql_add)
   self.db.commit()

 # 減少代理分?jǐn)?shù)
 def decrease(self, ip):
  sql_get = "SELECT * FROM PROXY WHERE IP='%s'" % (ip)
  self.cursor.execute(sql_get)
  score = self.cursor.fetchone()[1]
  print(score)
  if score and score > MIN_SCORE:
   print('代理', ip, '當(dāng)前分?jǐn)?shù)', score, '減1')
   sql_change = "UPDATE PROXY SET SCORE = %s WHERE IP = '%s'" % (score-1, ip)
  else:
   print('代理', ip, '當(dāng)前分?jǐn)?shù)', score, '移除')
   sql_change = "DELETE FROM PROXY WHERE IP = %s" % (ip)
  self.cursor.execute(sql_change)
  self.db.commit()

 # 分?jǐn)?shù)最大化
 def max(self, ip):
  print('代理', ip, '可用，設(shè)置為', MAX_SCORE)
  sql_max = "UPDATE PROXY SET SCORE = %s WHERE IP = '%s'" % (MAX_SCORE, ip)
  self.cursor.execute(sql_max)
  self.db.commit()
  
 # 隨機(jī)獲取有效代理
 def random(self):
  # 先從滿分中隨機(jī)選一個(gè)
  sql_max = "SELECT * FROM PROXY WHERE SCORE=%s" % (MAX_SCORE)
  if self.cursor.execute(sql_max):
   results = self.cursor.fetchall()
   return choice(results)[0]
  # 沒(méi)有滿分則隨機(jī)選一個(gè)
  else:
   sql_all = "SELECT * FROM PROXY WHERE SCORE BETWEEN %s AND %s" % (MIN_SCORE, MAX_SCORE)
   if self.cursor.execute(sql_all):
    results = self.cursor.fetchall()
    return choice(results)[0]
   else:
    raise PoolEmptyError

 # 判斷是否存在
 def exists(self, ip):
  sql_exists = "SELECT 1 FROM PROXY WHERE IP='%s' limit 1" % ip
  return self.cursor.execute(sql_exists)
  
 # 獲取數(shù)量
 def count(self):
  sql_count = "SELECT * FROM PROXY"
  return self.cursor.execute(sql_count)

 # 獲取全部
 def all(self):
  self.count()
  return self.cursor.fetchall()

 # 批量獲取
 def batch(self, start, stop):
  sql_batch = "SELECT * FROM PROXY LIMIT %s, %s" % (start, stop - start)
  self.cursor.execute(sql_batch)
  return self.cursor.fetchall()

from crawler import Crawler
from db import MySqlClient
from setting import *
import sys


class Getter():
 def __init__(self):
  self.mysql = MySqlClient()
  self.crawler = Crawler()

 # 判斷數(shù)量是否足夠
 def is_over_threshold(self):
  if self.mysql.count() >= POOL_UPPER_THRESHOLD:
   return True
  else:
   return False
 
 def run(self):
  print('獲取器開(kāi)始執(zhí)行')
  if not self.is_over_threshold():
   for callback_label in range(self.crawler.__CrawlFuncCount__):
    callback = self.crawler.__CrawlFunc__[callback_label]
    # 獲取代理
    all_ip = self.crawler.get_proxies(callback)
    sys.stdout.flush()
    for ip in all_ip:
     self.mysql.add(ip)


if __name__ == '__main__':
 get = Getter()
 get.run()