|
| 1 | +package api |
| 2 | + |
| 3 | +import ( |
| 4 | + "context" |
| 5 | + _ "embed" |
| 6 | + "fmt" |
| 7 | + "log" |
| 8 | + "net/http" |
| 9 | + "os" |
| 10 | + "os/signal" |
| 11 | + "strconv" |
| 12 | + "sync" |
| 13 | + "syscall" |
| 14 | + |
| 15 | + // |
| 16 | + |
| 17 | + "github.com/twiny/spidy/v2/internal/pkg/spider/v1" |
| 18 | + "github.com/twiny/spidy/v2/internal/service/cache" |
| 19 | + "github.com/twiny/spidy/v2/internal/service/writer" |
| 20 | + |
| 21 | + // |
| 22 | + "github.com/twiny/domaincheck" |
| 23 | + "github.com/twiny/flog" |
| 24 | + "github.com/twiny/wbot" |
| 25 | +) |
| 26 | + |
| 27 | +//go:embed version |
| 28 | +var Version string |
| 29 | + |
| 30 | +// Spider |
| 31 | +type Spider struct { |
| 32 | + wg *sync.WaitGroup |
| 33 | + setting *spider.Setting |
| 34 | + bot *wbot.WBot |
| 35 | + pages chan *spider.Page |
| 36 | + check *domaincheck.Checker |
| 37 | + store spider.Storage |
| 38 | + write spider.Writer |
| 39 | + log *flog.Logger |
| 40 | +} |
| 41 | + |
| 42 | +// NewSpider |
| 43 | +func NewSpider(fp string) (*Spider, error) { |
| 44 | + // get settings |
| 45 | + setting := spider.ParseSetting(fp) |
| 46 | + |
| 47 | + // crawler opts |
| 48 | + opts := []wbot.Option{ |
| 49 | + wbot.SetParallel(setting.Parralle), |
| 50 | + wbot.SetMaxDepth(setting.Crawler.MaxDepth), |
| 51 | + wbot.SetRateLimit(setting.Crawler.Limit.Rate, setting.Crawler.Limit.Interval), |
| 52 | + wbot.SetMaxBodySize(setting.Crawler.MaxBodySize), |
| 53 | + wbot.SetUserAgents(setting.Crawler.UserAgents), |
| 54 | + wbot.SetProxies(setting.Crawler.Proxies), |
| 55 | + } |
| 56 | + |
| 57 | + bot := wbot.NewWBot(opts...) |
| 58 | + |
| 59 | + check, err := domaincheck.NewChecker() |
| 60 | + if err != nil { |
| 61 | + return nil, err |
| 62 | + } |
| 63 | + |
| 64 | + // store |
| 65 | + store, err := cache.NewCache(setting.Store.TTL, setting.Store.Path) |
| 66 | + if err != nil { |
| 67 | + return nil, err |
| 68 | + } |
| 69 | + |
| 70 | + // logger |
| 71 | + log, err := flog.NewLogger(setting.Log.Path, "spidy", setting.Log.Rotate) |
| 72 | + if err != nil { |
| 73 | + return nil, err |
| 74 | + } |
| 75 | + |
| 76 | + write, err := writer.NewCSVWriter(setting.Result.Path) |
| 77 | + if err != nil { |
| 78 | + return nil, err |
| 79 | + } |
| 80 | + |
| 81 | + return &Spider{ |
| 82 | + wg: &sync.WaitGroup{}, |
| 83 | + setting: setting, |
| 84 | + bot: bot, |
| 85 | + pages: make(chan *spider.Page, setting.Parralle), |
| 86 | + check: check, |
| 87 | + store: store, |
| 88 | + write: write, |
| 89 | + log: log, |
| 90 | + }, nil |
| 91 | +} |
| 92 | + |
| 93 | +// Start |
| 94 | +func (s *Spider) Start(links []string) error { |
| 95 | + // go crawl |
| 96 | + s.wg.Add(len(links)) |
| 97 | + for _, link := range links { |
| 98 | + go func(l string) { |
| 99 | + defer s.wg.Done() |
| 100 | + // |
| 101 | + if err := s.bot.Crawl(l); err != nil { |
| 102 | + s.log.Error(err.Error(), map[string]string{"url": l}) |
| 103 | + } |
| 104 | + }(link) |
| 105 | + } |
| 106 | + |
| 107 | + // check domains |
| 108 | + s.wg.Add(s.setting.Parralle) |
| 109 | + for i := 0; i < s.setting.Parralle; i++ { |
| 110 | + go func() { |
| 111 | + defer s.wg.Done() |
| 112 | + // results |
| 113 | + for res := range s.bot.Stream() { |
| 114 | + // if response is ok |
| 115 | + if res.Status != http.StatusOK { |
| 116 | + s.log.Info("bad HTTP status", map[string]string{ |
| 117 | + "url": res.URL.String(), |
| 118 | + "status": strconv.Itoa(res.Status), |
| 119 | + }) |
| 120 | + continue |
| 121 | + } |
| 122 | + |
| 123 | + // extract domains |
| 124 | + domains := spider.FindDomains(res.Body) |
| 125 | + |
| 126 | + // check availability |
| 127 | + for _, domain := range domains { |
| 128 | + root := fmt.Sprintf("%s.%s", domain.Name, domain.TLD) |
| 129 | + |
| 130 | + // check if allowed extension |
| 131 | + if len(s.setting.TLDs) > 0 { |
| 132 | + if ok := s.setting.TLDs[domain.TLD]; !ok { |
| 133 | + s.log.Info("unsupported domain", map[string]string{ |
| 134 | + "domain": root, |
| 135 | + "url": res.URL.String(), |
| 136 | + }) |
| 137 | + continue |
| 138 | + } |
| 139 | + } |
| 140 | + |
| 141 | + // skip if already checked |
| 142 | + if s.store.HasChecked(root) { |
| 143 | + s.log.Info("already checked", map[string]string{ |
| 144 | + "domain": root, |
| 145 | + "url": res.URL.String(), |
| 146 | + }) |
| 147 | + continue |
| 148 | + } |
| 149 | + |
| 150 | + // |
| 151 | + ctx, cancel := context.WithTimeout(context.Background(), s.setting.Timeout) |
| 152 | + defer cancel() |
| 153 | + |
| 154 | + status, err := s.check.Check(ctx, root) |
| 155 | + if err != nil { |
| 156 | + s.log.Error(err.Error(), map[string]string{ |
| 157 | + "domain": root, |
| 158 | + "url": res.URL.String(), |
| 159 | + }) |
| 160 | + continue |
| 161 | + } |
| 162 | + |
| 163 | + // save domain |
| 164 | + if err := s.write.Write(&spider.Domain{ |
| 165 | + URL: res.URL.String(), |
| 166 | + Name: domain.Name, |
| 167 | + TLD: domain.TLD, |
| 168 | + Status: status.String(), |
| 169 | + }); err != nil { |
| 170 | + s.log.Error(err.Error(), map[string]string{ |
| 171 | + "domain": root, |
| 172 | + "url": res.URL.String(), |
| 173 | + }) |
| 174 | + continue |
| 175 | + } |
| 176 | + |
| 177 | + // terminal print |
| 178 | + fmt.Printf("[Spidy] == domain: %s - status %s\n", root, status.String()) |
| 179 | + } |
| 180 | + } |
| 181 | + }() |
| 182 | + } |
| 183 | + |
| 184 | + s.wg.Wait() |
| 185 | + return nil |
| 186 | +} |
| 187 | + |
| 188 | +// Shutdown |
| 189 | +func (s *Spider) Shutdown() error { |
| 190 | + // attempt graceful shutdown |
| 191 | + sigs := make(chan os.Signal, 1) |
| 192 | + signal.Notify(sigs, syscall.SIGHUP, syscall.SIGINT, syscall.SIGTERM, syscall.SIGQUIT) |
| 193 | + |
| 194 | + <-sigs |
| 195 | + log.Println("shutting down ...") |
| 196 | + |
| 197 | + // 2nd ctrl+c kills program |
| 198 | + go func() { |
| 199 | + sigs := make(chan os.Signal, 1) |
| 200 | + signal.Notify(sigs, syscall.SIGHUP, syscall.SIGINT, syscall.SIGTERM, syscall.SIGQUIT) |
| 201 | + <-sigs |
| 202 | + log.Println("killing program ...") |
| 203 | + os.Exit(0) |
| 204 | + }() |
| 205 | + |
| 206 | + s.bot.Close() |
| 207 | + s.log.Close() |
| 208 | + if err := s.store.Close(); err != nil { |
| 209 | + return err |
| 210 | + } |
| 211 | + os.Exit(0) |
| 212 | + return nil |
| 213 | +} |
0 commit comments