mapreduce.go 6.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275
  1. package mr
  2. import (
  3. "errors"
  4. "fmt"
  5. "sync"
  6. "github.com/tal-tech/go-zero/core/errorx"
  7. "github.com/tal-tech/go-zero/core/lang"
  8. "github.com/tal-tech/go-zero/core/syncx"
  9. "github.com/tal-tech/go-zero/core/threading"
  10. )
  11. const (
  12. defaultWorkers = 16
  13. minWorkers = 1
  14. )
  15. var (
  16. ErrCancelWithNil = errors.New("mapreduce cancelled with nil")
  17. ErrReduceNoOutput = errors.New("reduce not writing value")
  18. )
  19. type (
  20. GenerateFunc func(source chan<- interface{})
  21. MapFunc func(item interface{}, writer Writer)
  22. VoidMapFunc func(item interface{})
  23. MapperFunc func(item interface{}, writer Writer, cancel func(error))
  24. ReducerFunc func(pipe <-chan interface{}, writer Writer, cancel func(error))
  25. VoidReducerFunc func(pipe <-chan interface{}, cancel func(error))
  26. Option func(opts *mapReduceOptions)
  27. mapReduceOptions struct {
  28. workers int
  29. }
  30. Writer interface {
  31. Write(v interface{})
  32. }
  33. )
  34. func Finish(fns ...func() error) error {
  35. if len(fns) == 0 {
  36. return nil
  37. }
  38. return MapReduceVoid(func(source chan<- interface{}) {
  39. for _, fn := range fns {
  40. source <- fn
  41. }
  42. }, func(item interface{}, writer Writer, cancel func(error)) {
  43. fn := item.(func() error)
  44. if err := fn(); err != nil {
  45. cancel(err)
  46. }
  47. }, func(pipe <-chan interface{}, cancel func(error)) {
  48. drain(pipe)
  49. }, WithWorkers(len(fns)))
  50. }
  51. func FinishVoid(fns ...func()) {
  52. if len(fns) == 0 {
  53. return
  54. }
  55. MapVoid(func(source chan<- interface{}) {
  56. for _, fn := range fns {
  57. source <- fn
  58. }
  59. }, func(item interface{}) {
  60. fn := item.(func())
  61. fn()
  62. }, WithWorkers(len(fns)))
  63. }
  64. func Map(generate GenerateFunc, mapper MapFunc, opts ...Option) chan interface{} {
  65. options := buildOptions(opts...)
  66. source := buildSource(generate)
  67. collector := make(chan interface{}, options.workers)
  68. done := syncx.NewDoneChan()
  69. go mapDispatcher(mapper, source, collector, done.Done(), options.workers)
  70. return collector
  71. }
  72. func MapReduce(generate GenerateFunc, mapper MapperFunc, reducer ReducerFunc, opts ...Option) (interface{}, error) {
  73. source := buildSource(generate)
  74. return MapReduceWithSource(source, mapper, reducer, opts...)
  75. }
  76. func MapReduceWithSource(source <-chan interface{}, mapper MapperFunc, reducer ReducerFunc,
  77. opts ...Option) (interface{}, error) {
  78. options := buildOptions(opts...)
  79. output := make(chan interface{})
  80. collector := make(chan interface{}, options.workers)
  81. done := syncx.NewDoneChan()
  82. writer := newGuardedWriter(output, done.Done())
  83. var closeOnce sync.Once
  84. var retErr errorx.AtomicError
  85. finish := func() {
  86. closeOnce.Do(func() {
  87. done.Close()
  88. close(output)
  89. })
  90. }
  91. cancel := once(func(err error) {
  92. if err != nil {
  93. retErr.Set(err)
  94. } else {
  95. retErr.Set(ErrCancelWithNil)
  96. }
  97. drain(source)
  98. finish()
  99. })
  100. go func() {
  101. defer func() {
  102. if r := recover(); r != nil {
  103. cancel(fmt.Errorf("%v", r))
  104. } else {
  105. finish()
  106. }
  107. }()
  108. reducer(collector, writer, cancel)
  109. }()
  110. go mapperDispatcher(mapper, source, collector, done.Done(), cancel, options.workers)
  111. value, ok := <-output
  112. if err := retErr.Load(); err != nil {
  113. return nil, err
  114. } else if ok {
  115. return value, nil
  116. } else {
  117. return nil, ErrReduceNoOutput
  118. }
  119. }
  120. func MapReduceVoid(generator GenerateFunc, mapper MapperFunc, reducer VoidReducerFunc, opts ...Option) error {
  121. _, err := MapReduce(generator, mapper, func(input <-chan interface{}, writer Writer, cancel func(error)) {
  122. reducer(input, cancel)
  123. // We need to write a placeholder to let MapReduce to continue on reducer done,
  124. // otherwise, all goroutines are waiting. The placeholder will be discarded by MapReduce.
  125. writer.Write(lang.Placeholder)
  126. }, opts...)
  127. return err
  128. }
  129. func MapVoid(generate GenerateFunc, mapper VoidMapFunc, opts ...Option) {
  130. drain(Map(generate, func(item interface{}, writer Writer) {
  131. mapper(item)
  132. }, opts...))
  133. }
  134. func WithWorkers(workers int) Option {
  135. return func(opts *mapReduceOptions) {
  136. if workers < minWorkers {
  137. opts.workers = minWorkers
  138. } else {
  139. opts.workers = workers
  140. }
  141. }
  142. }
  143. func buildOptions(opts ...Option) *mapReduceOptions {
  144. options := newOptions()
  145. for _, opt := range opts {
  146. opt(options)
  147. }
  148. return options
  149. }
  150. func buildSource(generate GenerateFunc) chan interface{} {
  151. source := make(chan interface{})
  152. threading.GoSafe(func() {
  153. defer close(source)
  154. generate(source)
  155. })
  156. return source
  157. }
  158. // drain drains the channel.
  159. func drain(channel <-chan interface{}) {
  160. // drain the channel
  161. for range channel {
  162. }
  163. }
  164. func executeMappers(mapper MapFunc, input <-chan interface{}, collector chan<- interface{},
  165. done <-chan lang.PlaceholderType, workers int) {
  166. var wg sync.WaitGroup
  167. defer func() {
  168. wg.Wait()
  169. close(collector)
  170. }()
  171. pool := make(chan lang.PlaceholderType, workers)
  172. writer := newGuardedWriter(collector, done)
  173. for {
  174. select {
  175. case <-done:
  176. return
  177. case pool <- lang.Placeholder:
  178. item, ok := <-input
  179. if !ok {
  180. <-pool
  181. return
  182. }
  183. wg.Add(1)
  184. // better to safely run caller defined method
  185. threading.GoSafe(func() {
  186. defer func() {
  187. wg.Done()
  188. <-pool
  189. }()
  190. mapper(item, writer)
  191. })
  192. }
  193. }
  194. }
  195. func mapDispatcher(mapper MapFunc, input <-chan interface{}, collector chan<- interface{},
  196. done <-chan lang.PlaceholderType, workers int) {
  197. executeMappers(func(item interface{}, writer Writer) {
  198. mapper(item, writer)
  199. }, input, collector, done, workers)
  200. }
  201. func mapperDispatcher(mapper MapperFunc, input <-chan interface{}, collector chan<- interface{},
  202. done <-chan lang.PlaceholderType, cancel func(error), workers int) {
  203. executeMappers(func(item interface{}, writer Writer) {
  204. mapper(item, writer, cancel)
  205. }, input, collector, done, workers)
  206. }
  207. func newOptions() *mapReduceOptions {
  208. return &mapReduceOptions{
  209. workers: defaultWorkers,
  210. }
  211. }
  212. func once(fn func(error)) func(error) {
  213. once := new(sync.Once)
  214. return func(err error) {
  215. once.Do(func() {
  216. fn(err)
  217. })
  218. }
  219. }
  220. type guardedWriter struct {
  221. channel chan<- interface{}
  222. done <-chan lang.PlaceholderType
  223. }
  224. func newGuardedWriter(channel chan<- interface{}, done <-chan lang.PlaceholderType) guardedWriter {
  225. return guardedWriter{
  226. channel: channel,
  227. done: done,
  228. }
  229. }
  230. func (gw guardedWriter) Write(v interface{}) {
  231. select {
  232. case <-gw.done:
  233. return
  234. default:
  235. gw.channel <- v
  236. }
  237. }