蜘蛛抓取页面内容
(u string, num int)
| 15 | |
| 16 | // 蜘蛛抓取页面内容 |
| 17 | func Spider(u string, num int) { |
| 18 | is := true |
| 19 | defer func() { |
| 20 | config.Wg.Done() |
| 21 | if is { |
| 22 | <-config.Ch |
| 23 | } |
| 24 | |
| 25 | }() |
| 26 | config.Mux.Lock() |
| 27 | fmt.Printf("\rStart %d Spider...", config.Progress) |
| 28 | config.Progress++ |
| 29 | config.Mux.Unlock() |
| 30 | //标记完成 |
| 31 | |
| 32 | u, _ = url.QueryUnescape(u) |
| 33 | if num > 1 && cmd.D != "" && !util.RegexpMatch(cmd.D, u) { |
| 34 | return |
| 35 | } |
| 36 | if GetEndUrl(u) { |
| 37 | return |
| 38 | } |
| 39 | if cmd.M == 3 { |
| 40 | for _, v := range config.Risks { |
| 41 | if strings.Contains(u, v) { |
| 42 | return |
| 43 | } |
| 44 | } |
| 45 | } |
| 46 | AppendEndUrl(u) |
| 47 | request, err := http.NewRequest("GET", u, nil) |
| 48 | if err != nil { |
| 49 | return |
| 50 | } |
| 51 | |
| 52 | request.Header.Set("Accept-Encoding", "gzip") //使用gzip压缩传输数据让访问更快 |
| 53 | request.Header.Set("User-Agent", util.GetUserAgent()) |
| 54 | request.Header.Set("Accept", "*/*") |
| 55 | //增加header选项 |
| 56 | if cmd.C != "" { |
| 57 | request.Header.Set("Cookie", cmd.C) |
| 58 | } |
| 59 | //加载yaml配置(headers) |
| 60 | if cmd.I { |
| 61 | util.SetHeadersConfig(&request.Header) |
| 62 | } |
| 63 | |
| 64 | //处理返回结果 |
| 65 | //tr := &http.Transport{ |
| 66 | // TLSClientConfig: &tls.Config{InsecureSkipVerify: true}, |
| 67 | //} |
| 68 | //client = &http.Client{Timeout: time.Duration(cmd.TI) * time.Second, |
| 69 | // Transport: tr, |
| 70 | // CheckRedirect: func(req *http.Request, via []*http.Request) error { |
| 71 | // if len(via) >= 10 { |
| 72 | // return fmt.Errorf("Too many redirects") |
| 73 | // } |
| 74 | // if len(via) > 0 { |
no test coverage detected