func doWork(links <-chan string, results chan<- string) {
	for link := range links {
		var doc *goquery.Document
		for i := 1; ; i++ {
			var err error
			doc, err = goquery.NewDocument(link)
			if err == nil {
				break
			}
			fmt.Fprintf(os.Stderr, "[Tentativa %d] Erro tentando processar página de servidor: %s. Erro: %q", i, link, err)
			if i == maxRetries {
				fmt.Fprintf(os.Stderr, "Página não processada: %s", link)
				return
			}
			time.Sleep(time.Duration(i) * time.Duration(rand.Intn(5)) * time.Second)
		}
		var row []string
		doc.Find("td.desc").Each(func(i int, s *goquery.Selection) {
			cell := strings.Replace(
				strings.Trim(s.Next().Text(), " \n"),
				",",
				".",
				1)
			row = append(row, cell)
		})
		if len(row) > 0 {
			results <- strings.Join(row, *sep)
		} else {
			fmt.Fprintf(os.Stderr, "Não achou td.desc: %s\n", link)
		}
	}
}

func parseTrendingRepos(doc *goquery.Document) []GithubRepo {
	var repos []GithubRepo
	var regStars = regexp.MustCompile("[0-9]+")

	doc.Find("li.repo-list-item").Each(func(i int, s *goquery.Selection) {
		title := strings.Trim(s.Find("h3.repo-list-name a").Text(), "\n\t ")
		title = strings.Replace(title, " ", "", -1)
		title = strings.Replace(title, "\n", "", -1)
		description := strings.Trim(s.Find("p.repo-list-description").Text(), "\n\t ")
		url, _ := s.Find("h3.repo-list-name a").Attr("href")
		url = "https://github.com" + url
		starsString := s.Find("p.repo-list-meta").Text()
		starsString = strings.Replace(starsString, ",", "", -1)
		starsString = regStars.FindString(starsString)
		if starsString == "" {
			starsString = "0"
		}
		stars, _ := strconv.Atoi(starsString)

		repo := GithubRepo{
			Title:       title,
			Description: description,
			Url:         url,
			Stars:       stars,
			Forks:       0,
			Date:        time.Now().UTC().Unix(),
		}

		repos = append(repos, repo)
	})

	return repos
}

func (this *cleaner) cleanCites(doc *goquery.Document) *goquery.Document {
	cites := doc.Find("cite")
	cites.Each(func(i int, s *goquery.Selection) {
		this.config.parser.removeNode(s)
	})
	return doc
}

// ogtags extracts the og:title, og:image, ... tags from a webpage
func defaultHTML(i *data.Item, sourceURL string, doc *goquery.Document) {
	fmt.Println("Running OG extract.")

	selection := doc.Find("title")
	if len(selection.Nodes) != 0 {
		i.Caption = selection.Nodes[0].FirstChild.Data
	}

	selection = doc.Find("meta[property*='og']")

	for _, e := range selection.Nodes {
		m := htmlAttributeToMap(e.Attr)

		if m["property"] == "og:title" {
			i.Caption = m["content"]
		}
		if m["property"] == "og:image" {
			if !govalidator.IsRequestURL(m["content"]) {
				log.Println("Invalid url in og:image. " + sourceURL)
				continue
			}
			i.ImageURL = m["content"]
		}
		if m["property"] == "og:url" {
			if !govalidator.IsRequestURL(m["content"]) {
				log.Println("Invalid url in og:url. " + sourceURL)
				continue
			}
			i.URL = m["content"]
		}
		if m["property"] == "og:description" {
			i.Description = m["content"]
		}
	}
}

func parseOrderListPage(s *goquery.Document) ([]Order, bool, error) {
	c := s.Find(".container").First()
	t := c.Find("div").First().Text()
	if t != ">注文情報(一覧)<" && t != ">注文情報(検索)<" {
		return nil, false, fmt.Errorf("cannot open \"注文情報(一覧)\", but %#v", t)
	}
	// タイトル行の削除
	c.Find("hr").First().Next().PrevAll().Remove()

	results := []Order{}
	c.Find("a").Each(
		func(_ int, s *goquery.Selection) {
			href, ok := s.Attr("href")
			if !ok || !strings.HasPrefix(href, "../otc/C003.html?") {
				return
			}
			u, err := url.Parse(href)
			if err != nil || u.RawQuery == "" {
				return
			}
			v, err := url.ParseQuery(u.RawQuery)
			results = append(results, Order{
				OrderId:     v.Get("order_id"),
				OrderMethod: v.Get("order_method"),
			})
		})

	return results, c.Find("a[accesskey=\"#\"]").Length() == 1, nil
}

//获取店铺名称
func GetShopName(p *goquery.Document) string {
	name := p.Find(".tb-shop-name").Text()
	if name == "" {
		name = p.Find(".slogo-shopname").Text()
	}
	return strings.TrimSpace(name)
}

func feedsFromDoc(doc *goquery.Document, text string) []string {
	sel := "link[type='application/rss+xml']"
	sel += ", link[type='application/atom+xml']"
	matches := doc.Find(sel)

	if matches.Length() > 0 {
		feeds := make([]string, matches.Length())
		matches.Each(func(i int, s *goquery.Selection) {
			url, _ := s.Attr("href")
			feeds[i] = url
		})
		return feeds
	}

	rx := regexp.MustCompile(`href=['"]([^'"]*(rss|atom|feed|xml)[^'"]*)['"]`)
	if rx.FindString(text) != "" {
		matches := rx.FindAllStringSubmatch(text, -1)
		feeds := make([]string, len(matches))
		for i, e := range matches {
			feeds[i] = e[1]
		}
		return feeds
	}

	return make([]string, 0)
}

// Parse html
func perseHTML(htmldata *goquery.Document) []string {
	var dates []string

	htmldata.Find("a.bt-open").Each(func(_ int, s *goquery.Selection) {
		if jsonData, ok := s.Attr("id"); ok {

			//decode
			htmlStringDecode(&jsonData)

			//analyze json object
			var jsonObject map[string]interface{}
			//json.JsonAnalyze(jsonData, &jsonObject)
			json.Unmarshal([]byte(jsonData), &jsonObject)

			//extract date from json object
			//e.g. 2016-02-27 03:30:00
			strDate := jsonObject["field19"].(string)
			if isTimeApplicable(strDate) {
				dates = append(dates, strDate)
			}
		}
	})

	return dates
}

func garfield(i *data.Item, sourceURL string, doc *goquery.Document) {
	if !strings.Contains(sourceURL, "www.gocomics.com/garfield") {
		return
	}

	fmt.Println("Running Garfield plugin.")

	// update title

	selection := doc.Find(".strip")
	if len(selection.Nodes) == 0 {
		fmt.Println("Garfield plugin found no .strip. " + sourceURL)
	} else {
		if len(selection.Nodes) > 1 {
			fmt.Println("Garfield plugin found >1 .strip. " + sourceURL)
		}
		m := htmlAttributeToMap(selection.Nodes[0].Attr)

		if govalidator.IsRequestURL(m["src"]) {
			i.Description = "<img src =\""
			i.Description += m["src"]
			i.Description += "\" />"
		} else {
			fmt.Println("Amazon plugin invalid url. " + m["src"])
		}
		i.ImageURL = ""
	}

}

func getItems(doc *goquery.Document) (items []item, maxWidth int) {
	doc.Find("td.title a").EachWithBreak(func(i int, s *goquery.Selection) bool {
		if i == maxItems {
			return false
		}

		if s.Text() == "More" {
			return true
		}

		href, _ := s.Attr("href")
		title := s.Text()
		points := s.Parent().Parent().Next().Find("span").Text()
		a, b := len(fmt.Sprintf("%s (%s)", title, points)), len(href)
		maxWidth = max(a, b, maxWidth)

		items = append(items, item{
			title:  title,
			url:    href,
			points: points,
		})

		return true
	})
	return
}

/*
Get the two teams in a match
*/
func getTeamsId(d *goquery.Document) ([2]int, error) {
	var ids [2]int

	url1, ok := d.Find("div.container.left h3 a").Attr("href")
	if !ok {
		return ids, errors.New("could not find team a")
	}

	idA, err := parseTeam(BASE + url1)
	if err != nil {
		return ids, err
	}

	url2, ok := d.Find("div.container.right h3 a").Attr("href")
	if !ok {
		return ids, errors.New("could not find team b")
	}

	idB, err := parseTeam(BASE + url2)
	if err != nil {
		return ids, err
	}

	ids[0] = idA
	ids[1] = idB
	return ids, nil
}

/*
** get friends' friends info
 */
func (w *SocialWorker) GetFFInfo(query *goquery.Document) {
	var user User
	// var uid string
	var usex string
	// var usersId []string
	// var usersName []string
	// uidString, _ := query.Find("div.c").Eq(1).Find("a").Attr("href")
	// var digitsRegexp = regexp.MustCompile(`(^|&|\?)uid=([^&]*)(&|$)`)
	/*
	 ** 获取粉丝的粉丝的uid（str)
	 */
	// str := digitsRegexp.FindStringSubmatch(uidString)
	// uid = crawlUrl.Id
	// usersId = append(usersId, uid)
	uStr := query.Find("div.c").Eq(2).Text()
	nameStr_1 := GetBetweenStr(uStr, ":", "性别")
	nameStr_2 := GetBetweenStr(nameStr_1, ":", "认证")
	nameStr_3 := strings.Split(nameStr_2, ":")
	uname := nameStr_3[1]
	sexStr_1 := GetBetweenStr(uStr, "性别", "地区")
	sexStr_2 := strings.Split(sexStr_1, ":")
	if sexStr_2[1] == "男" {
		usex = "male"
	} else {
		usex = "famale"
	}

	user.uid = crawlUrl.FatherId
	user.friendid = crawlUrl.Id
	user.uname = uname
	user.usex = usex
	glog.Infoln(user)
	w.putItems(user)
}

/*
**get friends url
 */
func (w *SocialWorker) GetFriendsUrl(query *goquery.Document, p *page.Page) {
	var str_1 string
	// newCrawlUrl := models.CrawlUrl{}
	query.Find("div.c").Find("table").Find("tbody").Find("tr").Find("a:last-child").Each(func(j int, s *goquery.Selection) {
		if j%2 != 0 {
			friendsUrlString, _ := s.Attr("href")
			var digitsRegexp = regexp.MustCompile(`(^|&|\?)uid=([^&]*)(&|$)`)
			str := digitsRegexp.FindStringSubmatch(friendsUrlString)
			if str == nil {
				str_1 = "1"
			} else {
				str_1 = str[2]
			}
			friendsInfoUrl := "http://weibo.cn/" + str_1 + "/info"
			// newCrawlUrl.Url = "http://weibo.cn/" + str_1 + "/fans"
			// p.AddTargetRequestWithHeaderFile(friendsInfoUrl, "html", "./header.json")
			// newCrawlUrl.Id = str_1
			// newCrawlUrl.Layer = crawlUrl.Layer + 1
			// newCrawlUrl.FatherId = crawlUrl.Id
			// w.SendMessageToSQS(newCrawlUrl)

			Urls = append(Urls, friendsInfoUrl)
			UrlsLevel = append(UrlsLevel, UrlsLevel[i]+1)
		}
	})
}

// Parse 获取url对应的资源并根据规则进行解析
func (this *RedditLogic) Parse(redditUrl string) error {
	redditUrl = strings.TrimSpace(redditUrl)
	if redditUrl == "" {
		redditUrl = this.domain + this.golang
	} else if !strings.HasPrefix(redditUrl, "https") {
		redditUrl = "https://" + redditUrl
	}

	var (
		doc *goquery.Document
		err error
	)

	// if doc, err = goquery.NewDocument(redditUrl); err != nil {
	if doc, err = this.newDocumentFromResp(redditUrl); err != nil {
		logger.Errorln("goquery reddit newdocument error:", err)
		return err
	}

	// 最后面的先入库处理
	resourcesSelection := doc.Find("#siteTable .link")

	for i := resourcesSelection.Length() - 1; i >= 0; i-- {
		err = this.dealRedditOneResource(goquery.NewDocumentFromNode(resourcesSelection.Get(i)).Selection)

		if err != nil {
			logger.Errorln(err)
		}
	}

	return err
}

// ARSOPotresi returs slice of Potres struct
func ARSOPotresi() []Potres {
	var potresi []Potres
	var doc *goquery.Document
	var e error

	if res, found := cacheArso.Get("potresi"); found {
		return res.([]Potres)
	}

	if doc, e = goquery.NewDocument("http://www.arso.gov.si/potresi/obvestila%20o%20potresih/aip/"); e != nil {
		return potresi
	}

	doc.Find("#glavna td.vsebina table tr").Each(func(i int, s *goquery.Selection) {
		magnituda, err := strconv.ParseFloat(s.Find("td:nth-child(4)").Text(), 2)
		if magnituda > 0 && err == nil {
			potres := Potres{}
			potres.Magnituda = magnituda
			potres.Lat, _ = strconv.ParseFloat(s.Find("td:nth-child(2)").Text(), 3)
			potres.Lon, _ = strconv.ParseFloat(s.Find("td:nth-child(3)").Text(), 3)
			potres.Lokacija = s.Find("td:nth-child(6)").Text()
			potres.Datum = s.Find("td:nth-child(1)").Text()
			potresi = append(potresi, potres)
		}
	})
	cacheArso.Set("potresi", potresi, cache.DefaultExpiration)
	return potresi
}

func scrapeSearch(document *goquery.Document, url string) {
	pagesStr := document.Find("a.next_page").Prev().Text()
	pages, _ := strconv.Atoi(pagesStr)
	page := 1
	for page <= pages {
		pageURL := url + "&p=" + strconv.Itoa(page)
		fmt.Println("Analyzing page: " + pageURL)
		doc := downloadURL(pageURL)
		doc.Find(".user-list-item").Each(func(i int, s *goquery.Selection) {
			email := s.Find("a.email").Text()
			profileURL, _ := s.Find("a").Eq(1).Attr("href")
			username := profileURL[1:len(profileURL)]
			profileURL = "http://github.com" + profileURL
			info := s.Find(".user-list-info")
			_ = info.Find("ul.user-list-meta").Remove()
			_ = info.Find("a").Remove()
			name := strings.TrimSpace(info.Text())
			fmt.Println("Parsed user: " + username)
			user := user{name: name, email: email, url: profileURL, username: username}
			dumpToCSV(user)
		})

		page = page + 1
	}
}

//获取淘宝属性信息
func GetAttrbuites(p *goquery.Document) string {
	attribute := make([]string, 0, 20)
	p.Find("#J_AttrUL li").Each(func(index int, element *goquery.Selection) {
		as := strings.Split(element.Text(), ":")
		if len(as) < 2 {
			as = strings.Split(element.Text(), "：")
		}

		b := ""

		if len(as) >= 2 && !utf8.ValidString(as[1]) {
			as[1] = as[1]
			b = as[1]
		}

		attribute = append(attribute, as[0]+":"+b)
	})

	if len(attribute) == 0 {
		p.Find("#attributes .attributes-list li").Each(func(index int, element *goquery.Selection) {
			attribute = append(attribute, element.Text())
		})
	}

	return strings.Join(attribute, "##")
}

func getTerms(doc *goquery.Document) ([]string, error) {
	terms := make([]string, 0)
	doc.Find("p").Each(func(i int, s *goquery.Selection) {
		// Decode any HTML-encoded characters so they can be parsed correctly.
		bdy := html.UnescapeString(s.Text())
		// TODO: condense into a regex?
		bdy = strings.Replace(bdy, "-", " ", -1)
		bdy = strings.Replace(bdy, ",", " ", -1)
		bdy = strings.Replace(bdy, ".", " ", -1)
		bdy = strings.Replace(bdy, ";", " ", -1)
		bdy = strings.Replace(bdy, "\"", " ", -1)
		terms = append(terms, strings.Fields(bdy)...)
	})

	re, err := regexp.Compile("[^A-Za-z0-9]+")
	if err != nil {
		log.Println("Unexpected regex compilation error: " + err.Error())
		return []string{}, err
	}

	for i := 0; i < len(terms); i++ {
		terms[i] = re.ReplaceAllString(terms[i], "")
	}

	return terms, nil
}

//获取店铺地址
func GetShopUrl(p *goquery.Document) string {
	href, _ := p.Find(".tb-seller-name").Attr("href")
	if href == "" {
		href, _ = p.Find(".slogo-shopname").Attr("href")
	}
	return strings.TrimSpace("https:" + href)
}

func (t *TownClient) getSValue() (sValue string) {
	log.WithField("tag", TAG).Info("getting sValue for town login")
	sValue = ""
	var doc *goquery.Document
	var e error
	log.WithField("tag", TAG).Infof("GET %v", ROOT)
	if doc, e = goquery.NewDocument(ROOT); e != nil {
		log.WithField("tag", TAG).Errorf("%s", e.Error())
		return
	}

	doc.Find("input").Each(func(i int, s *goquery.Selection) {
		attr, exists := s.Attr("name")
		if exists == true {
			if attr == "s" {
				bla, exists := s.Attr("value")
				if exists == true {
					sValue = bla
				}
			}
		}

	})
	log.WithField("tag", TAG).Infof("sValue: %v", sValue)
	return sValue
}