sink_node.go 9.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361
  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. //static
  17. input chan interface{}
  18. name string
  19. sinkType string
  20. mutex sync.RWMutex
  21. //configs (also static for sinks)
  22. concurrency int
  23. options map[string]interface{}
  24. isMock bool
  25. //states varies after restart
  26. ctx api.StreamContext
  27. statManagers []StatManager
  28. sinks []api.Sink
  29. }
  30. func NewSinkNode(name string, sinkType string, props map[string]interface{}) *SinkNode {
  31. bufferLength := 1024
  32. if c, ok := props["bufferLength"]; ok {
  33. if t, err := common.ToInt(c); err != nil || t <= 0 {
  34. //invalid property bufferLength
  35. } else {
  36. bufferLength = t
  37. }
  38. }
  39. return &SinkNode{
  40. input: make(chan interface{}, bufferLength),
  41. name: name,
  42. sinkType: sinkType,
  43. options: props,
  44. concurrency: 1,
  45. ctx: nil,
  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. input: make(chan interface{}, 1024),
  52. name: name,
  53. sinks: []api.Sink{sink},
  54. options: props,
  55. concurrency: 1,
  56. ctx: nil,
  57. isMock: true,
  58. }
  59. }
  60. func (m *SinkNode) Open(ctx api.StreamContext, result chan<- error) {
  61. m.ctx = ctx
  62. logger := ctx.GetLogger()
  63. logger.Debugf("open sink node %s", m.name)
  64. go func() {
  65. if c, ok := m.options["concurrency"]; ok {
  66. if t, err := common.ToInt(c); err != nil || t <= 0 {
  67. logger.Warnf("invalid type for concurrency property, should be positive integer but found %t", c)
  68. } else {
  69. m.concurrency = t
  70. }
  71. }
  72. runAsync := false
  73. if c, ok := m.options["runAsync"]; ok {
  74. if t, ok := c.(bool); !ok {
  75. logger.Warnf("invalid type for runAsync property, should be bool but found %t", c)
  76. } else {
  77. runAsync = t
  78. }
  79. }
  80. retryInterval := 1000
  81. if c, ok := m.options["retryInterval"]; ok {
  82. if t, err := common.ToInt(c); err != nil || t < 0 {
  83. logger.Warnf("invalid type for retryInterval property, should be positive integer but found %t", c)
  84. } else {
  85. retryInterval = t
  86. }
  87. }
  88. cacheLength := 1024
  89. if c, ok := m.options["cacheLength"]; ok {
  90. if t, err := common.ToInt(c); err != nil || t < 0 {
  91. logger.Warnf("invalid type for cacheLength property, should be positive integer but found %t", c)
  92. } else {
  93. cacheLength = t
  94. }
  95. }
  96. cacheSaveInterval := 1000
  97. if c, ok := m.options["cacheSaveInterval"]; ok {
  98. if t, err := common.ToInt(c); err != nil || t < 0 {
  99. logger.Warnf("invalid type for cacheSaveInterval property, should be positive integer but found %t", c)
  100. } else {
  101. cacheSaveInterval = t
  102. }
  103. }
  104. omitIfEmpty := false
  105. if c, ok := m.options["omitIfEmpty"]; ok {
  106. if t, ok := c.(bool); !ok {
  107. logger.Warnf("invalid type for omitIfEmpty property, should be a bool value 'true/false'.", c)
  108. } else {
  109. omitIfEmpty = t
  110. }
  111. }
  112. sendSingle := false
  113. if c, ok := m.options["sendSingle"]; ok {
  114. if t, ok := c.(bool); !ok {
  115. logger.Warnf("invalid type for sendSingle property, should be a bool value 'true/false'.", c)
  116. } else {
  117. sendSingle = t
  118. }
  119. }
  120. var tp *template.Template = nil
  121. if c, ok := m.options["dataTemplate"]; ok {
  122. if t, ok := c.(string); !ok {
  123. logger.Warnf("invalid type for dateTemplate property, should be a string value.", c)
  124. } else {
  125. funcMap := template.FuncMap{
  126. "json": templates.JsonMarshal,
  127. "base64": templates.Base64Encode,
  128. "add": templates.Add,
  129. }
  130. temp, err := template.New("sink").Funcs(funcMap).Parse(t)
  131. if err != nil {
  132. msg := fmt.Sprintf("property dataTemplate %v is invalid: %v", t, err)
  133. result <- fmt.Errorf(msg)
  134. logger.Warnf(msg)
  135. return
  136. } else {
  137. tp = temp
  138. }
  139. }
  140. }
  141. m.reset()
  142. logger.Infof("open sink node %d instances", m.concurrency)
  143. for i := 0; i < m.concurrency; i++ { // workers
  144. go func(instance int) {
  145. var sink api.Sink
  146. var err error
  147. if !m.isMock {
  148. sink, err = getSink(m.sinkType, m.options)
  149. if err != nil {
  150. m.drainError(result, err, ctx, logger)
  151. return
  152. }
  153. m.mutex.Lock()
  154. m.sinks = append(m.sinks, sink)
  155. m.mutex.Unlock()
  156. if err := sink.Open(ctx); err != nil {
  157. m.drainError(result, err, ctx, logger)
  158. return
  159. }
  160. } else {
  161. sink = m.sinks[instance]
  162. }
  163. stats, err := NewStatManager("sink", ctx)
  164. if err != nil {
  165. m.drainError(result, err, ctx, logger)
  166. return
  167. }
  168. m.mutex.Lock()
  169. m.statManagers = append(m.statManagers, stats)
  170. m.mutex.Unlock()
  171. cache := NewCache(m.input, cacheLength, cacheSaveInterval, result, ctx)
  172. for {
  173. select {
  174. case data := <-cache.Out:
  175. stats.SetBufferLength(int64(cache.Length()))
  176. if runAsync {
  177. go doCollect(sink, data, stats, retryInterval, omitIfEmpty, sendSingle, tp, cache.Complete, ctx)
  178. } else {
  179. doCollect(sink, data, stats, retryInterval, omitIfEmpty, sendSingle, tp, cache.Complete, ctx)
  180. }
  181. case <-ctx.Done():
  182. logger.Infof("sink node %s instance %d done", m.name, instance)
  183. if err := sink.Close(ctx); err != nil {
  184. logger.Warnf("close sink node %s instance %d fails: %v", m.name, instance, err)
  185. }
  186. return
  187. }
  188. }
  189. }(i)
  190. }
  191. }()
  192. }
  193. func (m *SinkNode) reset() {
  194. if !m.isMock {
  195. m.sinks = nil
  196. }
  197. m.statManagers = nil
  198. }
  199. func extractInput(v []byte) ([]map[string]interface{}, error) {
  200. var j []map[string]interface{}
  201. if err := json.Unmarshal(v, &j); err != nil {
  202. return nil, fmt.Errorf("fail to decode the input %s as json: %v", v, err)
  203. }
  204. return j, nil
  205. }
  206. func doCollect(sink api.Sink, item *CacheTuple, stats StatManager, retryInterval int, omitIfEmpty bool, sendSingle bool, tp *template.Template, signalCh chan<- int, ctx api.StreamContext) {
  207. stats.IncTotalRecordsIn()
  208. stats.ProcessTimeStart()
  209. defer stats.ProcessTimeEnd()
  210. logger := ctx.GetLogger()
  211. var outdatas [][]byte
  212. switch val := item.data.(type) {
  213. case []byte:
  214. if omitIfEmpty && string(val) == "[{}]" {
  215. return
  216. }
  217. var (
  218. err error
  219. j []map[string]interface{}
  220. )
  221. if sendSingle || tp != nil {
  222. j, err = extractInput(val)
  223. if err != nil {
  224. stats.IncTotalExceptions()
  225. logger.Warnf("sink node %s instance %d publish %s error: %v", ctx.GetOpId(), ctx.GetInstanceId(), val, err)
  226. return
  227. }
  228. logger.Debugf("receive %d records", len(j))
  229. }
  230. if !sendSingle {
  231. if tp != nil {
  232. var output bytes.Buffer
  233. err := tp.Execute(&output, j)
  234. if err != nil {
  235. logger.Warnf("sink node %s instance %d publish %s decode template error: %v", ctx.GetOpId(), ctx.GetInstanceId(), val, err)
  236. stats.IncTotalExceptions()
  237. return
  238. }
  239. outdatas = append(outdatas, output.Bytes())
  240. } else {
  241. outdatas = [][]byte{val}
  242. }
  243. } else {
  244. for _, r := range j {
  245. var output bytes.Buffer
  246. err := tp.Execute(&output, r)
  247. if err != nil {
  248. logger.Warnf("sink node %s instance %d publish %s decode template error: %v", ctx.GetOpId(), ctx.GetInstanceId(), val, err)
  249. stats.IncTotalExceptions()
  250. return
  251. }
  252. outdatas = append(outdatas, output.Bytes())
  253. }
  254. }
  255. case error:
  256. outdatas = [][]byte{[]byte(fmt.Sprintf(`[{"error":"%s"}]`, val.Error()))}
  257. default:
  258. outdatas = [][]byte{[]byte(fmt.Sprintf(`[{"error":"result is not a string but found %#v"}]`, val))}
  259. }
  260. for _, outdata := range outdatas {
  261. outerloop:
  262. for {
  263. select {
  264. case <-ctx.Done():
  265. logger.Infof("sink node %s instance %d stops data resending", ctx.GetOpId(), ctx.GetInstanceId())
  266. return
  267. default:
  268. if err := sink.Collect(ctx, outdata); err != nil {
  269. stats.IncTotalExceptions()
  270. logger.Warnf("sink node %s instance %d publish %s error: %v", ctx.GetOpId(), ctx.GetInstanceId(), outdata, err)
  271. if retryInterval > 0 {
  272. time.Sleep(time.Duration(retryInterval) * time.Millisecond)
  273. logger.Debugf("try again")
  274. } else {
  275. break outerloop
  276. }
  277. } else {
  278. logger.Debugf("success")
  279. stats.IncTotalRecordsOut()
  280. signalCh <- item.index
  281. break outerloop
  282. }
  283. }
  284. }
  285. }
  286. }
  287. func doGetSink(name string, action map[string]interface{}) (api.Sink, error) {
  288. var (
  289. s api.Sink
  290. err error
  291. )
  292. switch name {
  293. case "log":
  294. s = sinks.NewLogSink()
  295. case "logToMemory":
  296. s = sinks.NewLogSinkToMemory()
  297. case "mqtt":
  298. s = &sinks.MQTTSink{}
  299. case "rest":
  300. s = &sinks.RestSink{}
  301. case "nop":
  302. s = &sinks.NopSink{}
  303. default:
  304. s, err = plugins.GetSink(name)
  305. if err != nil {
  306. return nil, err
  307. }
  308. }
  309. err = s.Configure(action)
  310. if err != nil {
  311. return nil, err
  312. }
  313. return s, nil
  314. }
  315. func (m *SinkNode) GetName() string {
  316. return m.name
  317. }
  318. func (m *SinkNode) GetInput() (chan<- interface{}, string) {
  319. return m.input, m.name
  320. }
  321. func (m *SinkNode) GetMetrics() (result [][]interface{}) {
  322. for _, stats := range m.statManagers {
  323. result = append(result, stats.GetMetrics())
  324. }
  325. return result
  326. }
  327. func (m *SinkNode) drainError(errCh chan<- error, err error, ctx api.StreamContext, logger api.Logger) {
  328. go func() {
  329. select {
  330. case errCh <- err:
  331. case <-ctx.Done():
  332. m.close(ctx, logger)
  333. }
  334. }()
  335. }
  336. func (m *SinkNode) close(ctx api.StreamContext, logger api.Logger) {
  337. for _, s := range m.sinks {
  338. if err := s.Close(ctx); err != nil {
  339. logger.Warnf("close sink fails: %v", err)
  340. }
  341. }
  342. }