go網站爬蟲
阿新 • • 發佈:2021-01-20
公司有需求爬取全國所有城市的樓盤,解析出經緯度,這種大範圍的爬取一般用scrapy框架比較合適,但是剛好最近學完了go語言,go語言的http庫用起來也很方便,因此嘗試用go語言完成
一,程式支援根據使用者輸入城市進行爬取,使用flag方法可以解析使用者資料,定義好要爬取的城市、爬取翻頁時等待時長(慢速爬取更友好)和是否列印爬取結果
func init() { flag.StringVar(&p, "p", "湖北", "省份") flag.StringVar(&c, "c", "武漢", "城市") flag.IntVar(&t, "t", 5, "翻頁間隔") flag.BoolVar(&r, "r", true, "列印記錄") flag.Parse() }
二,判斷使用者輸入的省份城市是否合法(是否在該網站中存在記錄),在進行元素定位時用到了htmlquery庫,採用xpath語法定位
func CheckCityExist() bool { rsp, err := engine.GetRsp("https://www.ke.com/city/") if err != nil { log.Fatalln("請求查詢城市失敗: ", err) } defer rsp.Body.Close() doc, _ := htmlquery.Parse(rsp.Body) retLis := htmlquery.Find(doc, `//div[@class="city_province"]`) for _, i := range retLis { pr := strings.TrimSpace(htmlquery.InnerText(htmlquery.FindOne(i, `./div[@class="city_list_tit c_b"]`))) if p == pr { ct := htmlquery.FindOne(i, fmt.Sprintf(`./ul/li[@class="CLICKDATA"]/a[text()="%s"]`, c)) if ct != nil { ctURL = htmlquery.SelectAttr(ct, "href") if ctURL != "" { ctURL = "https:" + ctURL return true } } } } return false }
三,每個城市會存在樓盤(新房)和小區(舊房)兩個爬取源,GetCityURL方法找到樓盤和小區的根路徑,並將其存入結構體pp中
func GetCityURL(u string) { rsp, err := engine.GetRsp(u) if err != nil { log.Fatalln("請求查詢城市失敗: ", err) } defer rsp.Body.Close() doc, _ := htmlquery.Parse(rsp.Body) louPan := htmlquery.FindOne(doc, `//li[@class="CLICKDATA"]/a[text()="新房"]`) // 新房 xiaoQu := htmlquery.FindOne(doc, `//li[@class="CLICKDATA"]/a[text()="小區"]`) // 舊房 lp := htmlquery.SelectAttr(louPan, "href") if lp == "" { louPan2 := htmlquery.FindOne(doc, `//li[@class="new-link-list"]/a[text()="新房"]`) // 新房 lp = htmlquery.SelectAttr(louPan2, "href") lp = strings.TrimRight(lp, "/") } if !strings.HasPrefix(lp, "http") { lp = "https:" + lp } pp.Loupan = lp pp.Xiaoqu = htmlquery.SelectAttr(xiaoQu, "href") pp.Province = p pp.City = c }
四,爬取樓盤,解析欄位,該網站請求樓盤返回的資料是json格式,每次返回10條記錄,解析很方便,注意該網站在爬到一定頁數後,每次會返回200條記錄,並且都是重複資料,因此程式對此作了判斷並結束迴圈
func ProcessLoupanData(data *utils.JData, pp *utils.CityMeta, r bool) { // 提取出資料庫需要的欄位 for _, item := range data.Data.List { item.CityName = pp.City item.Province = pp.Province db.InsertRow(&item, true) if !r { continue } fmt.Println("封面", item.CoverPic) fmt.Println("省份", item.Province) fmt.Println("城市", item.CityName) fmt.Println("區域", item.District) fmt.Println("地址", item.Address) fmt.Println("商圈", item.CircleName) fmt.Println("小區名", item.Loupan) fmt.Println("均價", item.AvePrice) fmt.Println("價格單位", item.PriceUnit) fmt.Println("銷售狀態", item.Status) fmt.Println("面積", item.Area) fmt.Println("房型", item.RoomType) fmt.Println("經度", item.LNG) fmt.Println("緯度", item.LAT) fmt.Println(strings.Repeat("#", 50)) } } func ParseCityLoupan(r bool, t int, pp *utils.CityMeta) { // 爬取樓盤資訊 if pp.Loupan == "" { log.Printf("%s-%s 無新房記錄", pp.Province, pp.City) return } url := pp.Loupan + "/pg1/?_t=1" b, err := engine.Download(url) if err != nil { log.Printf("首頁新房請求失敗", err) return } var data utils.JData err = json.Unmarshal(*b, &data) if err != nil { log.Printf("解析新房第1頁失敗:%v\n", err) return } total, err := strconv.Atoi(data.Data.Total) if err != nil { log.Fatalln("無法獲取新房總頁數", data.Data.Total, err) } ProcessLoupanData(&data, pp, r) pages := int(math.Ceil(float64(total)/ 10)) + 1 count := 0 for page := 2; page < pages; page++ { var data utils.JData time.Sleep(time.Second * time.Duration(t)) url := fmt.Sprintf("%s/pg%d/?_t=1", pp.Loupan, page) b, err := engine.Download(url) if err != nil { log.Printf("請求第%d頁失敗\n", page) continue } err = json.Unmarshal(*b, &data) if err != nil { log.Printf("解析第%d頁失敗\n", page) continue } if len(data.Data.List) > 10 { count ++ } if count == 2 { fmt.Println("結束重複值") break } ProcessLoupanData(&data, pp, r) } }
五,爬取小區,解析欄位,該網站請求小區返回的是html格式內容,需要用htmlquery庫xpath,方便解析欄位
func ParsePageRegionXiaoqu(r bool, url string, areaName string, p string, c string) { rsp, err := engine.GetRsp(url) if err != nil { log.Printf("爬取連結%s失敗\n", url) return } defer rsp.Body.Close() doc, _ := htmlquery.Parse(rsp.Body) items := htmlquery.Find(doc, `//li[contains(@class, "xiaoquListItem")]`) for _, item := range items { var data utils.MData titleNode := htmlquery.FindOne(item, `.//a[@class="maidian-detail"]`) title := htmlquery.SelectAttr(titleNode, "title") if title == "" { continue } circleNode := htmlquery.FindOne(item, `//a[@class="bizcircle"]`) circle := strings.TrimSpace(htmlquery.InnerText(circleNode)) if circle == "" { continue } lng, lat := engine.BaiduAPI(c, areaName+" "+title) if lng == 0 || lat == 0 { continue } data.Province = p data.CityName = c data.Loupan = title data.District = areaName data.CircleName = circle data.Address = circle + title data.LNG = strconv.FormatFloat(lng, 'f', 6, 64) data.LAT = strconv.FormatFloat(lat, 'f', 6, 64) if r { fmt.Println("封面", data.CoverPic) fmt.Println("省份", data.Province) fmt.Println("城市", data.CityName) fmt.Println("區域", data.District) fmt.Println("地址", data.Address) fmt.Println("商圈", data.CircleName) fmt.Println("小區名", data.Loupan) fmt.Println("均價", data.AvePrice) fmt.Println("價格單位", data.PriceUnit) fmt.Println("銷售狀態", data.Status) fmt.Println("面積", data.Area) fmt.Println("房型", data.RoomType) fmt.Println("經度", data.LNG) fmt.Println("緯度", data.LAT) fmt.Println(strings.Repeat("#", 50)) } db.InsertRow(&data, false) } } func ParseRegionXiaoqu(r bool, t int, baseURL string, areaName string, p string, c string) { page := 1 url := baseURL + strconv.Itoa(page) rsp, err := engine.GetRsp(url) if err != nil { log.Printf("爬取連結%s失敗\n", url) return } defer rsp.Body.Close() doc, _ := htmlquery.Parse(rsp.Body) totalNode := htmlquery.FindOne(doc, `//h2[contains(@class, "total")]/span`) total, err := strconv.Atoi(strings.TrimSpace(htmlquery.InnerText(totalNode))) if err != nil { log.Printf("非有效數字", total) return } pages := int(math.Ceil(float64(total) / 30)) for page:=1;page<=pages;page++{ time.Sleep(time.Second * time.Duration(t)) url := baseURL + strconv.Itoa(page) ParsePageRegionXiaoqu(r, url, areaName, p, c) } } // 爬取舊房資訊 func ParseCityXiaoqu(r bool, t int, pp *utils.CityMeta) { if !strings.HasSuffix(pp.Xiaoqu, "xiaoqu/") { log.Printf("%s-%s 無小區記錄", pp.Province, pp.City) return } url := pp.Xiaoqu + "pg1/" b, err := engine.Download(url) if err != nil { log.Printf("首頁新房請求失敗:%v", err) return } pt, _ := regexp.Compile(`href="/xiaoqu/([a-z0-9]{2,20}/pg)1".*?>(.*?)</a>`) retLis := pt.FindAllStringSubmatch(string(*b), -1) mp := make(map[string]string) // 去重後的區域url for _, v := range retLis { mp[pp.Xiaoqu+v[1]] = v[2] } for aURL, aName := range mp { fmt.Println("列印當前", aURL, aName) ParseRegionXiaoqu(r, t, aURL, aName, pp.Province, pp.City) } }
六,無論是爬取樓盤還是小區都有可能返回資料中沒有經緯度,這裡需要呼叫百度api獲取地址對應的經緯度(需要申請百度開發者賬號)
func BaiduAPI(city string, addr string) (lng, lat float64) { // 呼叫百度api,獲取地理位置 if useIdx == len(utils.AK) { log.Fatalln("無可用ak") } params := url.Values{} params.Set("address", addr) params.Set("output", "json") params.Set("city", city) params.Set("ak", utils.AK[useIdx]) toURL := apiURL + "?" + params.Encode() req, err = http.NewRequest("GET", toURL, nil) if err != nil { return 0, 0 } req.Header.Set("User-Agent", "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/84.0.4147.135 Safari/537.36") rsp, err = client.Do(req) if err != nil { return 0, 0 } defer rsp.Body.Close() b, _ := ioutil.ReadAll(rsp.Body) var ret apiRet err = json.Unmarshal(b, &ret) if err != nil { return 0, 0 } if ret.Msg == "天配額超限,限制訪問" { log.Println(utils.AK[useIdx], ret.Msg) useIdx += 1 return BaiduAPI(city, addr) } return ret.Result.Location.LNG, ret.Result.Location.LAT }
七,將資料存入mysql,使用經緯度md5值去重
func generateID(data *utils.MData) string { md := md5.New() md.Write([]byte(strings.Join([]string{data.LNG, data.LAT}, ""))) ret := md.Sum([]byte("")) return fmt.Sprintf("%x", ret) } // InsertRow ... func InsertRow(data *utils.MData, checkGeo bool) { // 若經緯度任意為空,則根據地址,呼叫百度api獲取經緯度 if data.LNG == "" || data.LAT == "" { if !checkGeo { return } if data.CityName == "" || data.Address == "" || data.Loupan == "" { return } lng, lat := engine.BaiduAPI(data.CityName, fmt.Sprintf("%s %s", data.Address, data.Loupan)) if lng == 0 || lat == 0 { return } data.LNG = strconv.FormatFloat(lng, 'f', 6, 64) data.LAT = strconv.FormatFloat(lat, 'f', 6, 64) } iid = generateID(data) if _, ok := mp[iid]; ok { // 過濾重複座標 return } mp[iid] = "" sqlStr := "insert into lbs(id, province, city, region, avenue, community_name, community_addr, longitude, dimension, status) values (?,?,?,?,?,?,?,?,?,?)" _, err := sdb.Exec(sqlStr, iid, data.Province, data.CityName, data.District, data.CircleName, data.Loupan, data.Address, data.LNG, data.LAT, 0) if err != nil { fmt.Printf("insert failed, err:%v\n", err) return } }
編譯後執行後效果如下:
完整程式碼參考github專案地址:https://github.com/Tarantiner/golang_beike