mapreduce.go 5.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264
  1. package mr
  2. import (
  3. "errors"
  4. "fmt"
  5. "sync"
  6. "zero/core/errorx"
  7. "zero/core/lang"
  8. "zero/core/syncx"
  9. "zero/core/threading"
  10. )
  11. const (
  12. defaultWorkers = 16
  13. minWorkers = 1
  14. )
  15. var ErrCancelWithNil = errors.New("mr cancelled with nil")
  16. type (
  17. GenerateFunc func(source chan<- interface{})
  18. MapFunc func(item interface{}, writer Writer)
  19. VoidMapFunc func(item interface{})
  20. MapperFunc func(item interface{}, writer Writer, cancel func(error))
  21. ReducerFunc func(pipe <-chan interface{}, writer Writer, cancel func(error))
  22. VoidReducerFunc func(pipe <-chan interface{}, cancel func(error))
  23. Option func(opts *mapReduceOptions)
  24. mapReduceOptions struct {
  25. workers int
  26. }
  27. Writer interface {
  28. Write(v interface{})
  29. }
  30. )
  31. func Finish(fns ...func() error) error {
  32. if len(fns) == 0 {
  33. return nil
  34. }
  35. return MapReduceVoid(func(source chan<- interface{}) {
  36. for _, fn := range fns {
  37. source <- fn
  38. }
  39. }, func(item interface{}, writer Writer, cancel func(error)) {
  40. fn := item.(func() error)
  41. if err := fn(); err != nil {
  42. cancel(err)
  43. }
  44. }, func(pipe <-chan interface{}, cancel func(error)) {
  45. drain(pipe)
  46. }, WithWorkers(len(fns)))
  47. }
  48. func FinishVoid(fns ...func()) {
  49. if len(fns) == 0 {
  50. return
  51. }
  52. MapVoid(func(source chan<- interface{}) {
  53. for _, fn := range fns {
  54. source <- fn
  55. }
  56. }, func(item interface{}) {
  57. fn := item.(func())
  58. fn()
  59. }, WithWorkers(len(fns)))
  60. }
  61. func Map(generate GenerateFunc, mapper MapFunc, opts ...Option) chan interface{} {
  62. options := buildOptions(opts...)
  63. source := buildSource(generate)
  64. collector := make(chan interface{}, options.workers)
  65. done := syncx.NewDoneChan()
  66. go mapDispatcher(mapper, source, collector, done.Done(), options.workers)
  67. return collector
  68. }
  69. func MapReduce(generate GenerateFunc, mapper MapperFunc, reducer ReducerFunc, opts ...Option) (interface{}, error) {
  70. source := buildSource(generate)
  71. return MapReduceWithSource(source, mapper, reducer, opts...)
  72. }
  73. func MapReduceWithSource(source <-chan interface{}, mapper MapperFunc, reducer ReducerFunc,
  74. opts ...Option) (interface{}, error) {
  75. options := buildOptions(opts...)
  76. output := make(chan interface{})
  77. collector := make(chan interface{}, options.workers)
  78. done := syncx.NewDoneChan()
  79. writer := newGuardedWriter(output, done.Done())
  80. var retErr errorx.AtomicError
  81. cancel := once(func(err error) {
  82. if err != nil {
  83. retErr.Set(err)
  84. } else {
  85. retErr.Set(ErrCancelWithNil)
  86. }
  87. drain(source)
  88. done.Close()
  89. close(output)
  90. })
  91. go func() {
  92. defer func() {
  93. if r := recover(); r != nil {
  94. cancel(fmt.Errorf("%v", r))
  95. }
  96. }()
  97. reducer(collector, writer, cancel)
  98. }()
  99. go mapperDispatcher(mapper, source, collector, done.Done(), cancel, options.workers)
  100. value, ok := <-output
  101. if err := retErr.Load(); err != nil {
  102. return nil, err
  103. } else if ok {
  104. return value, nil
  105. } else {
  106. return nil, nil
  107. }
  108. }
  109. func MapReduceVoid(generator GenerateFunc, mapper MapperFunc, reducer VoidReducerFunc, opts ...Option) error {
  110. _, err := MapReduce(generator, mapper, func(input <-chan interface{}, writer Writer, cancel func(error)) {
  111. reducer(input, cancel)
  112. // We need to write a placeholder to let MapReduce to continue on reducer done,
  113. // otherwise, all goroutines are waiting. The placeholder will be discarded by MapReduce.
  114. writer.Write(lang.Placeholder)
  115. }, opts...)
  116. return err
  117. }
  118. func MapVoid(generate GenerateFunc, mapper VoidMapFunc, opts ...Option) {
  119. drain(Map(generate, func(item interface{}, writer Writer) {
  120. mapper(item)
  121. }, opts...))
  122. }
  123. func WithWorkers(workers int) Option {
  124. return func(opts *mapReduceOptions) {
  125. if workers < minWorkers {
  126. opts.workers = minWorkers
  127. } else {
  128. opts.workers = workers
  129. }
  130. }
  131. }
  132. func buildOptions(opts ...Option) *mapReduceOptions {
  133. options := newOptions()
  134. for _, opt := range opts {
  135. opt(options)
  136. }
  137. return options
  138. }
  139. func buildSource(generate GenerateFunc) chan interface{} {
  140. source := make(chan interface{})
  141. threading.GoSafe(func() {
  142. defer close(source)
  143. generate(source)
  144. })
  145. return source
  146. }
  147. // drain drains the channel.
  148. func drain(channel <-chan interface{}) {
  149. // drain the channel
  150. for range channel {
  151. }
  152. }
  153. func executeMappers(mapper MapFunc, input <-chan interface{}, collector chan<- interface{},
  154. done <-chan lang.PlaceholderType, workers int) {
  155. var wg sync.WaitGroup
  156. defer func() {
  157. wg.Wait()
  158. close(collector)
  159. }()
  160. pool := make(chan lang.PlaceholderType, workers)
  161. writer := newGuardedWriter(collector, done)
  162. for {
  163. select {
  164. case <-done:
  165. return
  166. case pool <- lang.Placeholder:
  167. item, ok := <-input
  168. if !ok {
  169. <-pool
  170. return
  171. }
  172. wg.Add(1)
  173. // better to safely run caller defined method
  174. threading.GoSafe(func() {
  175. defer func() {
  176. wg.Done()
  177. <-pool
  178. }()
  179. mapper(item, writer)
  180. })
  181. }
  182. }
  183. }
  184. func mapDispatcher(mapper MapFunc, input <-chan interface{}, collector chan<- interface{},
  185. done <-chan lang.PlaceholderType, workers int) {
  186. executeMappers(func(item interface{}, writer Writer) {
  187. mapper(item, writer)
  188. }, input, collector, done, workers)
  189. }
  190. func mapperDispatcher(mapper MapperFunc, input <-chan interface{}, collector chan<- interface{},
  191. done <-chan lang.PlaceholderType, cancel func(error), workers int) {
  192. executeMappers(func(item interface{}, writer Writer) {
  193. mapper(item, writer, cancel)
  194. }, input, collector, done, workers)
  195. }
  196. func newOptions() *mapReduceOptions {
  197. return &mapReduceOptions{
  198. workers: defaultWorkers,
  199. }
  200. }
  201. func once(fn func(error)) func(error) {
  202. once := new(sync.Once)
  203. return func(err error) {
  204. once.Do(func() {
  205. fn(err)
  206. })
  207. }
  208. }
  209. type guardedWriter struct {
  210. channel chan<- interface{}
  211. done <-chan lang.PlaceholderType
  212. }
  213. func newGuardedWriter(channel chan<- interface{}, done <-chan lang.PlaceholderType) guardedWriter {
  214. return guardedWriter{
  215. channel: channel,
  216. done: done,
  217. }
  218. }
  219. func (gw guardedWriter) Write(v interface{}) {
  220. select {
  221. case <-gw.done:
  222. return
  223. default:
  224. gw.channel <- v
  225. }
  226. }