博客访问量自动监测脚本

# coding=utf-8
from selenium import webdriver  # 用于实现对动态JS内容的加载
import time  # 用于获取当前时间
import os  # 用于文件操作相关
import sys  # 用于获取脚本路径相关
import smtplib  # 用于邮件的发信动作
from email.mime.text import MIMEText  # 用于构建邮件内容
from email.header import Header

# 下载网页内容
def download(http_url):
    driver = webdriver.PhantomJS()
    driver.get(http_url)
    return driver

# 解析UV
def getUV(content):
    if content.find("busuanzi_value_site_uv") == -1:
        return -1
    else:
        index_start_uv = content.find("busuanzi_value_site_uv") + len("busuanzi_value_site_uv") + 2
        index_end_uv = content.find("</span> visitors")
        num_str = content[index_start_uv:index_end_uv]
        try:
            return int(num_str)
        except:
            return -1

# 解析PV
def getPV(content):
    if content.find("busuanzi_value_site_pv") == -1:
        return -1
    else:
        index_start_pv = content.find("busuanzi_value_site_pv") + len("busuanzi_value_site_pv") + 2
        index_end_pv = content.find("</span> times")
        num_str = content[index_start_pv:index_end_pv]
        try:
            return int(num_str)
        except:
            return -1

# 读取上一次保存的数据
def readLastFlag():
    dir_path = sys.argv[0][:sys.argv[0].rfind(os.path.sep)]
    file_path = dir_path + os.path.sep + "lastflag.txt"

    if os.path.exists(file_path):
        file_lf = open(file_path, 'r')
        last_flag_uv = int(file_lf.readline().strip())
        last_flag_pv = int(file_lf.readline().strip())
        file_lf.close()
    else:
        last_flag_uv = 0
        last_flag_pv = 0
    return last_flag_uv, last_flag_pv

# 保存上一次的数据
def writeFlag(uv, pv):
    dir_path = sys.argv[0][:sys.argv[0].rfind(os.path.sep)]
    file_path = dir_path + os.path.sep + "lastflag.txt"
    file_lf = open(file_path, 'w')
    file_lf.write(str(uv) + "\n")
    file_lf.write(str(pv) + "\n")
    file_lf.close()

# 为了方便自动更新，生成并保存指定格式的数据
def writeItem(str_content):
    dir_path = sys.argv[0][:sys.argv[0].rfind(os.path.sep)]
    file_path = dir_path + os.path.sep + "items.txt"
    file_lf = open(file_path, 'a')
    file_lf.write(str_content + "\n")
    file_lf.close()

# 发送邮件
def sendEmail(type_flag, message):
    # 用于构建邮件头
    # 发信方的信息：发信邮箱，QQ邮箱授权码(不是密码)
    from_addr = 'zhaoxuhui1993@qq.com'
    password = 'xxxxxxxxxxxx'

    # 收信方邮箱
    to_addr = 'zhaoxuhui1993@qq.com'

    # 发信服务器
    smtp_server = 'smtp.qq.com'

    # 邮箱正文内容，第一个参数为内容，第二个参数为格式(plain 为纯文本)，第三个参数为编码
    msg = MIMEText(message, 'plain', 'utf-8')

    # 邮件头信息
    msg['From'] = Header(from_addr)
    msg['To'] = Header(to_addr)
    if type_flag is 'uv':
        msg['Subject'] = Header('Blog UV Update!')
    elif type_flag is 'pv':
        msg['Subject'] = Header('Blog PV Update!')
    elif type_flag is 'uv_error':
        msg['Subject'] = Header('Blog UV Update Error!')
    elif type_flag is 'pv_error':
        msg['Subject'] = Header('Blog PV Update Error!')
    elif type_flag is 'status_error':
        msg['Subject'] = Header('Blog UV&PV Update Error!')

    # 开启发信服务，这里使用的是加密传输
    server = smtplib.SMTP_SSL(smtp_server)
    server.connect(smtp_server, 465)
    # 登录发信邮箱
    server.login(from_addr, password)
    # 发送邮件
    server.sendmail(from_addr, to_addr, msg.as_string())
    # 关闭服务器
    server.quit()

# 保存每次获取到的访问数据，记录历史，方便以后统计
def recordHistory(time, uv, pv):
    dir_path = sys.argv[0][:sys.argv[0].rfind(os.path.sep)]
    hist_file = open(dir_path + os.path.sep + 'history.txt', 'a')
    hist_file.write(time + "\t" + "UV:" + str(uv) + "\tPV:" + str(pv) + "\n")
    hist_file.close()


if __name__ == '__main__':
    uv = -1 # 初始uv值
    pv = -1 # 初始pv值
    max_try_time = 15    # 最大尝试次数
    sleep_time = 30 # 每次尝试的时间间隔
    log_str = ""
    run_date = time.strftime("%Y.%m.%d", time.localtime())
    run_time = time.strftime("%Y.%m.%d %H:%M:%S", time.localtime())
    for i in range(max_try_time):
        # 下载网页
        driver = download("http://zhaoxuhui.top")
        content = driver.page_source
        driver.quit()

        # 获取访问数据并进行记录
        uv = getUV(content)
        pv = getPV(content)
        # 如果没有网络或没有下载成功会输出一个空网页<html><head></head><body></body></html>
        # 因此需要进一步判断，如果不包含指定内容再尝试，否则跳出循环
        if uv != -1 and pv != -1:
            break
        else:
            # 为了降低频率和增加成功率，每次间隔指定时间
            time.sleep(sleep_time)
            log_str += "Attempt " + (i+1).__str__() + ": "
            if uv == -1:
                log_str += " get UV failed."
            if pv == -1:
                log_str += " get PV failed."
            log_str += "\n"
    # 如果达到最大次数还不行则下载失败，直接退出脚本，并发送错误提示邮件
    if uv == -1 or pv == -1:
        err_msg = "Error Time: " + run_time + "\n"
        err_flag = ""
        if uv == -1 and pv == -1:
            err_msg += "After attemping " + max_try_time.__str__() + " times,we still cannot get the UV and PV data!"
            err_flag = "status_error"
        elif uv == -1 and pv != -1:
            err_msg += "After attemping " + max_try_time.__str__() + " times,we still cannot get the UV data!"
            err_flag = "uv_error"
        elif uv != -1 and pv == -1:
            err_msg += "After attemping " + max_try_time.__str__() + " times,we still cannot get the PV data!"
            err_flag = "pv_error"
        err_msg += "\nRemember to check the data for " + run_date + " !"
        err_msg += "\nDetailed Info:\n"
        err_msg += log_str
        # 如果有问题发邮件提醒，退出本次执行的脚本
        sendEmail(err_flag, err_msg)
        exit()
    # 历史文件中追加记录
    recordHistory(run_date, uv, pv)

    # 获取上次的数据
    last_flag_uv, last_flag_pv = readLastFlag()
    write_flag = False

    # 由于是每增加10000才添加记录，所以除以10000
    flag_uv = uv / 10000
    flag_pv = pv / 10000

    # UV的数据比对
    if flag_uv != last_flag_uv:
        # 格式化的字符串
        uv_fmt_str = "\t- **" + run_date + "**" + " 网站累计访客数(UV)突破" + str(
            flag_uv * 10000)
        # 常规字符串用于阅读
        uv_str = run_date + " 网站累计访客数(UV)突破" + str(
            flag_uv * 10000)
        # 输出格式化字符串
        writeItem(uv_fmt_str)
        # 发送邮件
        sendEmail('uv', uv_str + "\n" + uv_fmt_str)
        write_flag = True
    
    # PV的数据比对
    if flag_pv != last_flag_pv:
        pv_fmt_str = "\t- **" + run_date + "**" + " 网站累计浏览量(PV)突破" + str(
            flag_pv * 10000)
        pv_str = run_date + " 网站累计浏览量(PV)突破" + str(
            flag_pv * 10000)
        writeItem(pv_fmt_str)
        sendEmail('pv', pv_str + "\n" + pv_fmt_str)
        write_flag = True

    # flag为true，说明数据有变化，将新的数据写入文件
    if write_flag:
        writeFlag(flag_uv, flag_pv)