sink_node.go 9.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366
  1. package nodes
  2. import (
  3. "bytes"
  4. "encoding/json"
  5. "fmt"
  6. "github.com/emqx/kuiper/common"
  7. "github.com/emqx/kuiper/common/templates"
  8. "github.com/emqx/kuiper/plugins"
  9. "github.com/emqx/kuiper/xstream/api"
  10. "github.com/emqx/kuiper/xstream/sinks"
  11. "sync"
  12. "text/template"
  13. "time"
  14. )
  15. type SinkNode struct {
  16. *defaultSinkNode
  17. //static
  18. sinkType string
  19. mutex sync.RWMutex
  20. //configs (also static for sinks)
  21. options map[string]interface{}
  22. isMock bool
  23. //states varies after restart
  24. sinks []api.Sink
  25. }
  26. func NewSinkNode(name string, sinkType string, props map[string]interface{}) *SinkNode {
  27. bufferLength := 1024
  28. if c, ok := props["bufferLength"]; ok {
  29. if t, err := common.ToInt(c); err != nil || t <= 0 {
  30. //invalid property bufferLength
  31. } else {
  32. bufferLength = t
  33. }
  34. }
  35. return &SinkNode{
  36. defaultSinkNode: &defaultSinkNode{
  37. input: make(chan interface{}, bufferLength),
  38. defaultNode: &defaultNode{
  39. name: name,
  40. concurrency: 1,
  41. ctx: nil,
  42. },
  43. },
  44. sinkType: sinkType,
  45. options: props,
  46. }
  47. }
  48. //Only for mock source, do not use it in production
  49. func NewSinkNodeWithSink(name string, sink api.Sink, props map[string]interface{}) *SinkNode {
  50. return &SinkNode{
  51. defaultSinkNode: &defaultSinkNode{
  52. input: make(chan interface{}, 1024),
  53. defaultNode: &defaultNode{
  54. name: name,
  55. concurrency: 1,
  56. ctx: nil,
  57. },
  58. },
  59. sinks: []api.Sink{sink},
  60. options: props,
  61. isMock: true,
  62. }
  63. }
  64. func (m *SinkNode) Open(ctx api.StreamContext, result chan<- error) {
  65. m.ctx = ctx
  66. logger := ctx.GetLogger()
  67. logger.Debugf("open sink node %s", m.name)
  68. go func() {
  69. if c, ok := m.options["concurrency"]; ok {
  70. if t, err := common.ToInt(c); err != nil || t <= 0 {
  71. logger.Warnf("invalid type for concurrency property, should be positive integer but found %t", c)
  72. } else {
  73. m.concurrency = t
  74. }
  75. }
  76. runAsync := false
  77. if c, ok := m.options["runAsync"]; ok {
  78. if t, ok := c.(bool); !ok {
  79. logger.Warnf("invalid type for runAsync property, should be bool but found %t", c)
  80. } else {
  81. runAsync = t
  82. }
  83. }
  84. retryInterval := 1000
  85. if c, ok := m.options["retryInterval"]; ok {
  86. if t, err := common.ToInt(c); err != nil || t < 0 {
  87. logger.Warnf("invalid type for retryInterval property, should be positive integer but found %t", c)
  88. } else {
  89. retryInterval = t
  90. }
  91. }
  92. cacheLength := 1024
  93. if c, ok := m.options["cacheLength"]; ok {
  94. if t, err := common.ToInt(c); err != nil || t < 0 {
  95. logger.Warnf("invalid type for cacheLength property, should be positive integer but found %t", c)
  96. } else {
  97. cacheLength = t
  98. }
  99. }
  100. cacheSaveInterval := 1000
  101. if c, ok := m.options["cacheSaveInterval"]; ok {
  102. if t, err := common.ToInt(c); err != nil || t < 0 {
  103. logger.Warnf("invalid type for cacheSaveInterval property, should be positive integer but found %t", c)
  104. } else {
  105. cacheSaveInterval = t
  106. }
  107. }
  108. omitIfEmpty := false
  109. if c, ok := m.options["omitIfEmpty"]; ok {
  110. if t, ok := c.(bool); !ok {
  111. logger.Warnf("invalid type for omitIfEmpty property, should be a bool value 'true/false'.", c)
  112. } else {
  113. omitIfEmpty = t
  114. }
  115. }
  116. sendSingle := false
  117. if c, ok := m.options["sendSingle"]; ok {
  118. if t, ok := c.(bool); !ok {
  119. logger.Warnf("invalid type for sendSingle property, should be a bool value 'true/false'.", c)
  120. } else {
  121. sendSingle = t
  122. }
  123. }
  124. var tp *template.Template = nil
  125. if c, ok := m.options["dataTemplate"]; ok {
  126. if t, ok := c.(string); !ok {
  127. logger.Warnf("invalid type for dateTemplate property, should be a string value.", c)
  128. } else {
  129. funcMap := template.FuncMap{
  130. "json": templates.JsonMarshal,
  131. "base64": templates.Base64Encode,
  132. "add": templates.Add,
  133. }
  134. temp, err := template.New("sink").Funcs(funcMap).Parse(t)
  135. if err != nil {
  136. msg := fmt.Sprintf("property dataTemplate %v is invalid: %v", t, err)
  137. result <- fmt.Errorf(msg)
  138. logger.Warnf(msg)
  139. return
  140. } else {
  141. tp = temp
  142. }
  143. }
  144. }
  145. m.reset()
  146. logger.Infof("open sink node %d instances", m.concurrency)
  147. for i := 0; i < m.concurrency; i++ { // workers
  148. go func(instance int) {
  149. var sink api.Sink
  150. var err error
  151. if !m.isMock {
  152. sink, err = getSink(m.sinkType, m.options)
  153. if err != nil {
  154. m.drainError(result, err, ctx, logger)
  155. return
  156. }
  157. m.mutex.Lock()
  158. m.sinks = append(m.sinks, sink)
  159. m.mutex.Unlock()
  160. if err := sink.Open(ctx); err != nil {
  161. m.drainError(result, err, ctx, logger)
  162. return
  163. }
  164. } else {
  165. sink = m.sinks[instance]
  166. }
  167. stats, err := NewStatManager("sink", ctx)
  168. if err != nil {
  169. m.drainError(result, err, ctx, logger)
  170. return
  171. }
  172. m.mutex.Lock()
  173. m.statManagers = append(m.statManagers, stats)
  174. m.mutex.Unlock()
  175. cache := NewCache(m.input, cacheLength, cacheSaveInterval, result, ctx)
  176. for {
  177. select {
  178. case data := <-cache.Out:
  179. if newdata, processed := m.preprocess(data.data); processed {
  180. break
  181. } else {
  182. data.data = newdata
  183. }
  184. stats.SetBufferLength(int64(cache.Length()))
  185. if runAsync {
  186. go doCollect(sink, data, stats, retryInterval, omitIfEmpty, sendSingle, tp, cache.Complete, ctx)
  187. } else {
  188. doCollect(sink, data, stats, retryInterval, omitIfEmpty, sendSingle, tp, cache.Complete, ctx)
  189. }
  190. case <-ctx.Done():
  191. logger.Infof("sink node %s instance %d done", m.name, instance)
  192. if err := sink.Close(ctx); err != nil {
  193. logger.Warnf("close sink node %s instance %d fails: %v", m.name, instance, err)
  194. }
  195. return
  196. }
  197. }
  198. }(i)
  199. }
  200. }()
  201. }
  202. func (m *SinkNode) reset() {
  203. if !m.isMock {
  204. m.sinks = nil
  205. }
  206. m.statManagers = nil
  207. }
  208. func extractInput(v []byte) ([]map[string]interface{}, error) {
  209. var j []map[string]interface{}
  210. if err := json.Unmarshal(v, &j); err != nil {
  211. return nil, fmt.Errorf("fail to decode the input %s as json: %v", v, err)
  212. }
  213. return j, nil
  214. }
  215. func doCollect(sink api.Sink, item *CacheTuple, stats StatManager, retryInterval int, omitIfEmpty bool, sendSingle bool, tp *template.Template, signalCh chan<- int, ctx api.StreamContext) {
  216. stats.IncTotalRecordsIn()
  217. stats.ProcessTimeStart()
  218. defer stats.ProcessTimeEnd()
  219. logger := ctx.GetLogger()
  220. var outdatas [][]byte
  221. switch val := item.data.(type) {
  222. case []byte:
  223. if omitIfEmpty && string(val) == "[{}]" {
  224. return
  225. }
  226. var (
  227. err error
  228. j []map[string]interface{}
  229. )
  230. if sendSingle || tp != nil {
  231. j, err = extractInput(val)
  232. if err != nil {
  233. stats.IncTotalExceptions()
  234. logger.Warnf("sink node %s instance %d publish %s error: %v", ctx.GetOpId(), ctx.GetInstanceId(), val, err)
  235. return
  236. }
  237. logger.Debugf("receive %d records", len(j))
  238. }
  239. if !sendSingle {
  240. if tp != nil {
  241. var output bytes.Buffer
  242. err := tp.Execute(&output, j)
  243. if err != nil {
  244. logger.Warnf("sink node %s instance %d publish %s decode template error: %v", ctx.GetOpId(), ctx.GetInstanceId(), val, err)
  245. stats.IncTotalExceptions()
  246. return
  247. }
  248. outdatas = append(outdatas, output.Bytes())
  249. } else {
  250. outdatas = [][]byte{val}
  251. }
  252. } else {
  253. for _, r := range j {
  254. var output bytes.Buffer
  255. err := tp.Execute(&output, r)
  256. if err != nil {
  257. logger.Warnf("sink node %s instance %d publish %s decode template error: %v", ctx.GetOpId(), ctx.GetInstanceId(), val, err)
  258. stats.IncTotalExceptions()
  259. return
  260. }
  261. outdatas = append(outdatas, output.Bytes())
  262. }
  263. }
  264. case error:
  265. outdatas = [][]byte{[]byte(fmt.Sprintf(`[{"error":"%s"}]`, val.Error()))}
  266. default:
  267. outdatas = [][]byte{[]byte(fmt.Sprintf(`[{"error":"result is not a string but found %#v"}]`, val))}
  268. }
  269. for _, outdata := range outdatas {
  270. outerloop:
  271. for {
  272. select {
  273. case <-ctx.Done():
  274. logger.Infof("sink node %s instance %d stops data resending", ctx.GetOpId(), ctx.GetInstanceId())
  275. return
  276. default:
  277. if err := sink.Collect(ctx, outdata); err != nil {
  278. stats.IncTotalExceptions()
  279. logger.Warnf("sink node %s instance %d publish %s error: %v", ctx.GetOpId(), ctx.GetInstanceId(), outdata, err)
  280. if retryInterval > 0 {
  281. time.Sleep(time.Duration(retryInterval) * time.Millisecond)
  282. logger.Debugf("try again")
  283. } else {
  284. break outerloop
  285. }
  286. } else {
  287. logger.Debugf("success")
  288. stats.IncTotalRecordsOut()
  289. signalCh <- item.index
  290. break outerloop
  291. }
  292. }
  293. }
  294. }
  295. }
  296. func doGetSink(name string, action map[string]interface{}) (api.Sink, error) {
  297. var (
  298. s api.Sink
  299. err error
  300. )
  301. switch name {
  302. case "log":
  303. s = sinks.NewLogSink()
  304. case "logToMemory":
  305. s = sinks.NewLogSinkToMemory()
  306. case "mqtt":
  307. s = &sinks.MQTTSink{}
  308. case "rest":
  309. s = &sinks.RestSink{}
  310. case "nop":
  311. s = &sinks.NopSink{}
  312. default:
  313. s, err = plugins.GetSink(name)
  314. if err != nil {
  315. return nil, err
  316. }
  317. }
  318. err = s.Configure(action)
  319. if err != nil {
  320. return nil, err
  321. }
  322. return s, nil
  323. }
  324. //Override defaultNode
  325. func (m *SinkNode) AddOutput(output chan<- interface{}, name string) error {
  326. return fmt.Errorf("fail to add output %s, sink %s cannot add output", name, m.name)
  327. }
  328. //Override defaultNode
  329. func (m *SinkNode) Broadcast(val interface{}) error {
  330. return fmt.Errorf("sink %s cannot add broadcast", m.name)
  331. }
  332. func (m *SinkNode) drainError(errCh chan<- error, err error, ctx api.StreamContext, logger api.Logger) {
  333. go func() {
  334. select {
  335. case errCh <- err:
  336. ctx.GetLogger().Errorf("error in sink %s", err)
  337. case <-ctx.Done():
  338. m.close(ctx, logger)
  339. }
  340. }()
  341. }
  342. func (m *SinkNode) close(ctx api.StreamContext, logger api.Logger) {
  343. for _, s := range m.sinks {
  344. if err := s.Close(ctx); err != nil {
  345. logger.Warnf("close sink fails: %v", err)
  346. }
  347. }
  348. }