`

Golang: 利用 channel (管道) 技术多线程下载图片

阅读更多

利用 channel (管道) 技术多线程下载图片.

 

package main

import (
	"io/ioutil"
	"log"
	"net/http"
	"os"
	"regexp"
	//"strconv"
	"strings"
	"sync"
)

var urlist = [...]string{"http://stock.591hx.com/article/2014-12-03/0000850005s.shtml"}
var album chan string
var w sync.WaitGroup
var dir string

func main() {
	dir = "tmp_chenjo/"
	err := os.Mkdir(dir, 0777)
	if err != nil {
		isexist := os.IsExist(err)
		log.Println(isexist)
	}
	album = make(chan string, 200)

	for _, v := range urlist {
		w.Add(1)
		go GetAlbum(v)
		w.Wait()
	}
}

func GetAlbum(url string) {
	data := GetUrl(url)
	body := string(data)
	//<img src="http://stock.591hx.com/images/hnimg/201412/03/64/13418266510200941552.jpg" alt="" /></p>
	part := regexp.MustCompile(`<img src="(.*)" alt="" />`)
	match := part.FindAllStringSubmatch(body, -1)
	for _, v := range match {

		if m, _ := regexp.MatchString(`.*/hnimg/201412/03/.*\.jpg`, v[1]); !m {
			continue
		}
		//println(v[1])
		album <- v[1]
		w.Add(1)
		go GetItem()
	}
	w.Done()

}

func GetItem() {
	url := <-album
	println(url)
	defer func() {
		ret := recover()
		if ret != nil {
			log.Println(ret)
			w.Done()
		} else {
			w.Done()
		}
	}()

	//data := GetUrl(url)
	//if len(data) > 10 {
	//body := string(data)
	//part := regexp.MustCompile(`bigimgsrc="(.*)"`)
	//match := part.FindAllStringSubmatch(body, -1)
	//for _, v := range match {
	str := strings.Split(url, "/")
	length := len(str)
	source := GetUrl(url)
	name := str[length-1]
	file, err := os.Create(dir + name)
	if err != nil {
		panic(err)
	}
	size, err := file.Write(source)
	defer file.Close()
	if err != nil {
		panic(err)
	}
	log.Println(size)
	//}
	//}
}

func GetUrl(url string) []byte {
	ret, err := http.Get(url)
	if err != nil {
		log.Println(url)
		status := map[string]string{}
		status["status"] = "400"
		status["url"] = url
		panic(status)
	}
	body := ret.Body
	data, _ := ioutil.ReadAll(body)
	return data
}

 

 console output:

G:/share/Golang/Golang.exe [G:/share/Golang]

2015/01/18 21:54:29 true

http://stock.591hx.com/images/hnimg/201412/03/64/13418266510200941552.jpg

http://stock.591hx.com/images/hnimg/201412/03/7/10886141709285175583.jpg

http://stock.591hx.com/images/hnimg/201412/03/0/5391574706574741364.jpg

http://stock.591hx.com/images/hnimg/201412/03/1/5405780767459854941.jpg

http://stock.591hx.com/images/hnimg/201412/03/16/13722698761317688276.jpg

http://stock.591hx.com/images/hnimg/201412/03/7/16853951343108680551.jpg

http://stock.591hx.com/images/hnimg/201412/03/50/17680852843413447062.jpg

http://stock.591hx.com/images/hnimg/201412/03/34/14366548421421579970.jpg

http://stock.591hx.com/images/hnimg/201412/03/4/17141924098089490820.jpg

2015/01/18 21:54:30 29968

2015/01/18 21:54:30 28697

2015/01/18 21:54:30 51586

2015/01/18 21:54:30 53031

2015/01/18 21:54:30 31234

2015/01/18 21:54:31 47618

2015/01/18 21:54:31 54649

2015/01/18 21:54:31 40422

2015/01/18 21:54:31 155527

成功: 进程退出代码 0.

 

 

分享到:
评论

相关推荐

Global site tag (gtag.js) - Google Analytics