nginx日志分析脚本

#!/bin/bash
#Name:log.sh
#Version:V1.0
#Description:This is check nginx log script.
#用户输入的查询开始时间
#input_log_time1="2017:"$1
#input_log_time2="2017:"$2
#将用户输入的时间转换为日志记录的时间
log_time1="2017:"$1
log_time2="2017:"$2
#日志路径
log_path=$3
#生成的临时日志
tmp_path=/tmp/${log_path}.tmp
#访问最高的IP记录
top_ip_path=/tmp/top_ip.log
#IP归属地查询
ip_where_path=/tmp/ip_where.log
#响应时间排名
nginx_time_path=/tmp/nginx_time.log
#平均响应时间
avg_time_path=/tmp/avg_time.log
#pv数量
pv_num_path=/tmp/pv_num.log
#服务器IP地址
server_ip=$(hostname -I)
#服务器名为
server_name=$(cat /etc/hostname)
#搜索关键词
key_word_bbs="GET /thread-"
key_word_m="GET /bbs/thread-"
key_word_mall="GET /mall/list"

#输入帮助
usage(){
    echo "执行格式为:\nsh $0 \"18:00\" \"22:00\" bbs.vim26.com_access.log"
}

#查询服务器的信息
server_info(){
    echo "=============================="
    echo "服务器名:${server_name}"
    echo "服务器IP:${server_ip}"
    echo "查询日志为:${log_path}"
    echo "查询时间为:${input_log_time1} - ${input_log_time2}"
    echo "=============================="
}

#提取行数并生成临时日志
time_line(){
    line1=$(grep -n "${log_time1}" ${log_path}|head -1|awk -F":" '{print $1}')
    line2=$(grep -n "${log_time2}" ${log_path}|head -1|awk -F":" '{print $1}')
    sed -n "${line1},${line2}p" $log_path > ${tmp_path}
}

#PV数
pv_num(){
    awk '{print $1}' ${tmp_path}|wc -l > ${pv_num_path}
    echo "    PV数量为:  $(cat ${pv_num_path})"
    echo "=============================="
}

#最多IP
top_ip(){
    awk '{print $1}' ${tmp_path}|sort|uniq -c|sort -n|tail -10 > ${top_ip_path}
    echo "访问最多IP前10为:" 
    cat ${top_ip_path}
    echo "=============================="
}

#IP归属地查询
ip_where(){
    > ${ip_where_path}
    for i in $(cat ${top_ip_path}|awk '{print $2}')
    do
        ip_where_host=$(curl -s "http://ip138.com/ips138.asp?ip=${i}&action=2"|iconv -f gb2312 -t utf-8|grep '<ul class="ul1"><li>'|awk -F "[:<>]+" '{print $5}'|cut -c 15-)
        echo "$i $ip_where_host" >> ${ip_where_path}
    done
    echo "=============================="
    echo "IP归属地查询结果为:"
    cat -n ${ip_where_path}
}


#日志查询
log_info(){
    [ "${log_path}" = "bbs.vim26.com_access.log" ] && key_word=${key_word_bbs}
    [ "${log_path}" = "m.vim26.com_access.log" ] && key_word=${key_word_m}
    [ "${log_path}" = "mall.vim26.com_access.log" ] && key_word=${key_word_mall}
    #监控关键词
    echo "监控关键链接为：${key_word}"
    echo "=============================="
    #关键词PV次数
    key_word_pv=$(grep "${key_word}" ${tmp_path}|wc -l)
    echo "关键链接PV访问次数: ${key_word_pv}"
    echo "=============================="
    #关键词PV次数
    #关键词平均响应时间
    grep "${key_word}" ${tmp_path}|awk '{sum+=$NF} END {print  sum/NR}' > ${avg_time_path}
    echo "关键链接平均响应时间为: $(cat ${avg_time_path})"
    echo "=============================="
    #关键词访问响应时间排名
    echo "关键链接访问响应时间排名"
    grep "${key_word}" ${tmp_path}|awk '{printf ("%.1f\n",$NF)}'|sort -nk 1 -t.|uniq -c> ${nginx_time_path}
    cat ${nginx_time_path}
    echo "=============================="
}


#搜索引擎搜索情况
bot_search(){
    echo "搜索情况汇总"
    bot_all=$(egrep "bot|spider|brandwatch" ${tmp_path}|wc -l)
    bot_baidu=$(grep "Baiduspider" ${tmp_path}| wc -l)
    bot_bing=$(grep "bingbot" ${tmp_path}| wc -l)
    bot_Google=$(grep "Googlebot" ${tmp_path}| wc -l)
    bot_sougou=$(grep "Sogou web spider" ${tmp_path}| wc -l)
    bot_toutiao=$(grep "ToutiaoSpider" ${tmp_path}| wc -l)
    bot_brandwatch=$(grep "brandwatch" ${tmp_path}| wc -l)
    echo "搜索引擎总计访问次数" ${bot_all}
    echo "    Baidu访问次数：  ${bot_baidu}"
    echo "    bing访问次数：   ${bot_bing}"
    echo "    Google访问次数： ${bot_Google}"
    echo "    sougou访问次数： ${bot_sougou}"
    echo "    toutiao访问次数: ${bot_toutiao}"
    echo "    brandwatch访问次数: ${bot_brandwatch}"
    echo "=============================="
}

#bbs的执行动作
all_info(){
    time_line
    server_info
    pv_num
    time_line
    bot_search
    top_ip
    log_info
    #ip_where
}



#判断参数是否正确
[ $# != 3 ] && {
    usage
    exit 1;
} 

all_info