前面提到了有关个人网站的实时在线人数问题,本文要讨论的是如何自己来实现一个这样的统计服务。因为网站也同时部署在Github上,海外用户访问Github镜像网站的访问日志Pi是拿不到的,这怎么办?
Google Channel Service
Google Channel Service允许应用和GAE (Google App Engine) 保持一个长连接,允许应用实时发送消息给JavaScript客户端,而不用让客户端用效率很低的定时轮询获取新消息。这个服务是允许有多个发布者和多个订阅者,也能创建多个主题来关联发布者和订阅者。
使用这个服务分两步:
客户端请求服务器端(部署在GAE上)获取一个Channel的Token:
客户端根据Channel Token和服务器建立长连接,并开始接收消息,这时其它的客户端(或服务器端)可以想这个通道发送消息
在线人数统计实现
在页面上部署beacon
通常的网站流量统计是依赖部署在页面上的beacon(Javascript或图片标签)来实现的,这样做的好处是可以直接过滤掉一些机器流量,并且可以将日志集中存储在日志收集服务器上,和网站分离开。
于是可以利用GAE实现一个简单的Beacon服务,这里采用Go语言来实现,用Java或Python也是可以的。
package counter
import (
"encoding/base64"
"fmt"
"time"
"net/http"
"appengine"
"appengine/channel"
)
var GIF []byte
const (
TOPIC = "counter"
)
func init() {
GIF, _ = base64.StdEncoding.DecodeString("R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7")
http.HandleFunc("/beacon.gif", handler)
http.HandleFunc("/new_token", handler_new_token)
}
func handler(w http.ResponseWriter, r *http.Request) {
context := appengine.NewContext(r)
now := time.Now()
expire := now.AddDate(30, 0, 0)
zcookie, _ := r.Cookie("z")
if zcookie == nil {
zcookie = &http.Cookie{}
zcookie.Name = "z"
zcookie.Value = make_hash("<your_salt>", r.RemoteAddr, now.UnixNano())
zcookie.Expires = expire
zcookie.Path = "/"
http.SetCookie(w, zcookie)
}
w.Header().Set("Content-type", "image/gif")
w.Header().Set("Cache-control", "no-cache, must-revalidate")
w.Header().Set("Expires", "Sat, 26 Jul 1997 05:00:00 GMT")
fmt.Fprintf(w, "%s", GIF)
channel.Send(context, TOPIC, zcookie.Value+"\n"+r.RemoteAddr+"\n"+r.Referer()+"\n"+r.UserAgent())
}
func handler_new_token(w http.ResponseWriter, r *http.Request) {
c := appengine.NewContext(r)
tok, err := channel.Create(c, TOPIC)
callback := r.FormValue("callback")
if err != nil {
http.Error(w, "Couldn't create Channel", http.StatusInternalServerError)
c.Errorf("channel.Create: %v", err)
return
}
if callback == "" {
w.Header().Set("Content-type", "text/javascript")
fmt.Fprintf(w, "%s", tok)
} else {
fmt.Fprintf(w, callback+"('%s')", tok)
}
}
代码最后一行是将访问日志实时通过Channel发送出去,该通道有一个指定的主题,这样订阅该主题的客户端都可以收到相应的消息。
app.yaml
application: <your_app_name>
version: 1
runtime: go
api_version: go1
handlers:
- url: /(.*).html
static_files: html/\1.html
upload: html/(.*\.html)
- url: (/.*)
script: _go_app
将该应用发布到GAE后,可以通过在页面底部放置<img src="http://<your_app_name>.appspot.com/beacon.gif/>"
标签来将网站的流量引到GAE,并通过Channel发布出去。
接收实时消息
比较遗憾的是Google只提供了Javascript库来供浏览器接收实时消息,相应的使用代码如下。
<html>
<body>
<script type="text/javascript" src="/_ah/channel/jsapi"></script>
<script>
function onOpened() {
alert("onOpened")
}
function onMessage(obj) {
console.log(obj)
alert("onMessage: " + obj.data)
}
function onError(x) {
alert("onError:"+x)
}
function onClose(x) {
alert("onClose:"+x)
}
function init(token) {
channel = new goog.appengine.Channel(token);
socket = channel.open();
socket.onopen = onOpened;
socket.onmessage = onMessage;
socket.onerror = onError;
socket.onclose = onClose;
}
</script>
<script type="text/javascript" src="/new_token?callback=init"></script>
</body>
</html>
显然Javascript是不能满足统计网站实时流量的需求,我们需要一个服务器端能接收消息的方法,经过对Javascript的逻辑分析,我实现了一个同样能接收消息的Go语言客户端
使用方法如下:
go get github.com/hugozhu/gae-channel
package main
import (
. "github.com/hugozhu/gae-channel"
"log"
)
func main() {
log.Println("started")
stop_chan := make(chan bool)
channel := NewChannel("http://<your_app_name>.appspot.com/new_token")
socket := channel.Open()
socket.OnOpened = func() {
log.Println("socket opened!")
}
socket.OnClose = func() {
log.Println("socket closed!")
stop_chan <- true
}
socket.OnMessage = func(msg *Message) {
log.Println(msg.ToString())
}
socket.OnError = func(err error) {
log.Println("error:", err)
}
<-stop_chan
}
实时分析日志
在Pi上利用Channel客户端库,脚本可以实时获取访问日志,可以通过最近15分钟内的日志计算出每分钟独立访客数和PV数,最后可以在Pi的液晶屏幕用柱状图显示出来。
package main
import (
"analytics"
"encoding/json"
"fmt"
"log"
"strings"
"time"
. "github.com/hugozhu/gae-channel"
. "github.com/hugozhu/rpi"
"github.com/hugozhu/rpi/pcd8544"
)
// pin setup
const (
_din = 3
_sclk = 5
_dc = 2
_rst = 0
_cs = 1
contrast = 43
)
var pv = analytics.NewPV(5, 60)
var uv = analytics.NewUV(5 * 60)
func init() {
//must be called before everything else!
WiringPiSetup()
pcd8544.LCDInit(_sclk, _din, _dc, _cs, _rst, contrast)
}
func main() {
log.Println("started")
stop_chan := make(chan bool)
channel := NewChannel("http://<your_app_name>.appspot.com/new_token")
socket := channel.Open()
socket.OnOpened = func() {
log.Println("socket opened!")
}
socket.OnClose = func() {
log.Println("socket closed!")
stop_chan <- true
}
socket.OnMessage = func(msg *Message) {
if msg.Level() >= 3 && msg.Child.Key == "c" {
v1 := *msg.Child.Child.Val
if len(v1) > 0 {
s := "[" + v1[0].Key + "]"
var v []string
json.Unmarshal([]byte(s), &v)
if len(v) == 2 && v[0] == "ae" {
s = v[1]
v = strings.Split(s, "\n")
zcookie := v[0]
t := time.Now()
pv.AddOne(t)
uv.AddOne(zcookie, t)
}
}
}
}
socket.OnError = func(err error) {
log.Println("error:", err)
}
go func() {
for {
pcd8544.LCDclear()
pcd8544.LCDdrawstring(0, 0, time.Now().Format("06-01-02 15:04"))
// pcd8544.LCDdrawline(0, 10, 83, 10, pcd8544.BLACK)
pcd8544.LCDdrawstring(0, 14, fmt.Sprintf("UV: %d", uv.Sum()))
pcd8544.LCDdrawstring(0, 24, fmt.Sprintf("PV: %d", pv.Sum()))
pcd8544.LCDdisplay()
time.Sleep(5 * time.Second)
}
}()
<-stop_chan
}
实际效果
![](http://ww4.sinaimg.cn/bmiddle/6bc40342jw1e3dzagtsl6j.jpg)