main.go 6.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288
  1. package main
  2. import (
  3. "bufio"
  4. "encoding/csv"
  5. "flag"
  6. "fmt"
  7. "io"
  8. "log"
  9. "net"
  10. "net/http"
  11. "os"
  12. "strings"
  13. "time"
  14. "github.com/google/gopacket"
  15. "github.com/google/gopacket/layers"
  16. "github.com/google/gopacket/pcap"
  17. "github.com/nats-io/nats"
  18. "github.com/nats-io/nats/encoders/protobuf"
  19. "git.scraperwall.com/scw/data"
  20. "git.scraperwall.com/scw/ip"
  21. )
  22. var (
  23. doLiveCapture = flag.Bool("live", false, "Capture data in real time from a given interface")
  24. iface = flag.String("interface", "eth0", "Interface to get packets from")
  25. snapshotLen = flag.Int("snapshot-len", 8192, "Snapshot Length in Bytes")
  26. filter = flag.String("filter", "tcp", "PCAP filter expression")
  27. promiscuous = flag.Bool("promiscuous", false, "Switch interface into promiscuous mode?")
  28. natsURL = flag.String("nats-url", "nats://127.0.0.1:4222", "The URL of the NATS server")
  29. natsQueue = flag.String("nats-queue", "mami_requests", "The NATS queue name")
  30. sleepFor = flag.Duration("sleep", 0, "Sleep this long between sending data")
  31. requestsFile = flag.String("requests", "", "CSV file containing requests (IP and URL)")
  32. useXForwardedAsSource = flag.Bool("use-x-forwarded", false, "Use the IP address in X-Forwarded-For as source")
  33. doVersion = flag.Bool("version", false, "Show version information")
  34. dataChan chan data.Request
  35. natsEC *nats.EncodedConn
  36. count uint64
  37. timeout = -1 * time.Second
  38. ipPriv *ip.IP
  39. // Version contains the program Version, e.g. 1.0.1
  40. Version string
  41. // BuildDate contains the date and time at which the program was compiled
  42. BuildDate string
  43. )
  44. func init() {
  45. flag.Parse()
  46. nats.RegisterEncoder(protobuf.PROTOBUF_ENCODER, &protobuf.ProtobufEncoder{})
  47. }
  48. func main() {
  49. if *doVersion {
  50. version()
  51. os.Exit(0)
  52. }
  53. go func(c *uint64) {
  54. for {
  55. fmt.Printf("%d items per second\n", *c)
  56. *c = 0
  57. time.Sleep(time.Second)
  58. }
  59. }(&count)
  60. // NATS
  61. //
  62. if *natsURL == "" {
  63. log.Fatal("No NATS URL specified (-nats-url)!")
  64. }
  65. natsConn, err := nats.Connect(*natsURL)
  66. if err != nil {
  67. log.Fatal(err)
  68. }
  69. natsEC, err = nats.NewEncodedConn(natsConn, protobuf.PROTOBUF_ENCODER)
  70. if err != nil {
  71. log.Fatalf("Encoded Connection: %v!\n", err)
  72. }
  73. // What should I do?
  74. if *requestsFile != "" {
  75. replayFile()
  76. } else if *doLiveCapture {
  77. liveCapture()
  78. }
  79. }
  80. func liveCapture() {
  81. // PCAP setup
  82. //
  83. handle, err := pcap.OpenLive(*iface, int32(*snapshotLen), *promiscuous, timeout)
  84. if err != nil {
  85. log.Fatal(err)
  86. }
  87. defer handle.Close()
  88. err = handle.SetBPFFilter(*filter)
  89. if err != nil {
  90. log.Fatal(err)
  91. }
  92. packetSource := gopacket.NewPacketSource(handle, handle.LinkType())
  93. for packet := range packetSource.Packets() {
  94. go processPacket(packet)
  95. }
  96. }
  97. // processPacket receives a raw packet from pcap, builds a Request item from it and sends it to the queue
  98. func processPacket(packet gopacket.Packet) {
  99. count++
  100. ipLayer := packet.Layer(layers.LayerTypeIPv4)
  101. if ipLayer == nil {
  102. log.Println("No IPv4 Layer!")
  103. return
  104. }
  105. ip, _ := ipLayer.(*layers.IPv4)
  106. if ip.Protocol != layers.IPProtocolTCP {
  107. log.Println("No TCP Protocol!")
  108. return
  109. }
  110. ipSrc := ip.SrcIP.String()
  111. ipDst := ip.DstIP.String()
  112. tcpLayer := packet.Layer(layers.LayerTypeTCP)
  113. if tcpLayer == nil {
  114. return
  115. }
  116. tcp, _ := tcpLayer.(*layers.TCP)
  117. portSrc := tcp.SrcPort
  118. portDst := tcp.DstPort
  119. sequence := tcp.Seq
  120. applicationLayer := packet.ApplicationLayer()
  121. if applicationLayer == nil {
  122. return
  123. }
  124. data := data.Request{}
  125. reader := bufio.NewReader(strings.NewReader(string(applicationLayer.Payload())))
  126. req, err := http.ReadRequest(reader)
  127. if err != nil {
  128. return
  129. }
  130. data.IpSrc = ipSrc
  131. data.IpDst = ipDst
  132. data.PortSrc = uint32(portSrc)
  133. data.PortDst = uint32(portDst)
  134. data.TcpSeq = uint32(sequence)
  135. data.CreatedAt = time.Now().UnixNano()
  136. data.Url = req.URL.String()
  137. data.Method = req.Method
  138. data.Referer = req.Referer()
  139. data.Host = req.Host
  140. data.Protocol = req.Proto
  141. data.Origin = data.Host
  142. if _, ok := req.Header["Connection"]; ok {
  143. data.Connection = req.Header["Connection"][0]
  144. }
  145. if _, ok := req.Header["X-Forwarded-For"]; ok {
  146. data.XForwardedFor = req.Header["X-Forwarded-For"][0]
  147. }
  148. if _, ok := req.Header["X-Real-IP"]; ok {
  149. data.XRealIP = req.Header["X-Real-IP"][0]
  150. }
  151. if _, ok := req.Header["X-Requested-With"]; ok {
  152. data.XRequestedWith = req.Header["X-Requested-With"][0]
  153. }
  154. if _, ok := req.Header["Accept-Encoding"]; ok {
  155. data.AcceptEncoding = req.Header["Accept-Encoding"][0]
  156. }
  157. if _, ok := req.Header["Accept-Language"]; ok {
  158. data.AcceptLanguage = req.Header["Accept-Language"][0]
  159. }
  160. if _, ok := req.Header["User-Agent"]; ok {
  161. data.UserAgent = req.Header["User-Agent"][0]
  162. }
  163. if _, ok := req.Header["Accept"]; ok {
  164. data.Accept = req.Header["Accept"][0]
  165. }
  166. if _, ok := req.Header["Cookie"]; ok {
  167. data.Cookie = req.Header["Cookie"][0]
  168. }
  169. data.Source = data.IpSrc
  170. if *useXForwardedAsSource && data.XForwardedFor != "" {
  171. if strings.Contains(data.XForwardedFor, ",") {
  172. ips := strings.Split(data.XForwardedFor, ",")
  173. for i := len(ips) - 1; i >= 0; i-- {
  174. ipRaw := strings.TrimSpace(ips[i])
  175. ipAddr := net.ParseIP(ipRaw)
  176. if ipAddr != nil && !ipPriv.IsPrivate(ipAddr) {
  177. data.Source = ipRaw
  178. break
  179. }
  180. }
  181. } else {
  182. ipAddr := net.ParseIP(strings.TrimSpace(data.XForwardedFor))
  183. if !ipPriv.IsPrivate(ipAddr) {
  184. data.Source = data.XForwardedFor
  185. }
  186. }
  187. }
  188. if data.Source == data.IpSrc && data.XRealIP != "" {
  189. data.Source = data.XRealIP
  190. }
  191. dataChan <- data
  192. }
  193. // replayFile takes a file containing a list of requests (SourceIP Url) and queues the requests
  194. // e.g.
  195. // 157.55.39.229 /gross-gerau/12012260-beate-anstatt
  196. // 103.232.100.98 /weinsheim-eifel/13729444-plus-warenhandelsges-mbh
  197. func replayFile() {
  198. var req data.Request
  199. var startTs time.Time
  200. var endTs time.Time
  201. for {
  202. fh, err := os.Open(*requestsFile)
  203. if err != nil {
  204. log.Fatalf("Failed to open request file '%s': %s", *requestsFile, err)
  205. }
  206. c := csv.NewReader(fh)
  207. c.Comma = ' '
  208. for {
  209. if *sleepFor > time.Nanosecond {
  210. startTs = time.Now()
  211. }
  212. r, err := c.Read()
  213. if err == io.EOF {
  214. break
  215. }
  216. if err != nil {
  217. log.Println(err)
  218. continue
  219. }
  220. req.IpSrc = r[0]
  221. req.Source = r[0]
  222. req.Url = r[1]
  223. req.UserAgent = "Munch/1.0"
  224. req.Host = "www.scraperwall.com"
  225. req.CreatedAt = time.Now().UnixNano()
  226. natsEC.Publish(*natsQueue, &req)
  227. count++
  228. if *sleepFor >= time.Nanosecond {
  229. endTs = time.Now()
  230. if endTs.Before(startTs.Add(*sleepFor)) {
  231. time.Sleep(*sleepFor - endTs.Sub(startTs))
  232. }
  233. }
  234. }
  235. }
  236. }
  237. // version outputs build information
  238. func version() {
  239. fmt.Printf("munchclient %s, built on %s\n", Version, BuildDate)
  240. }