Java Stream（摘自java8 API文档及stackflow）

Streamoperationsaredividedintointermediate(Stream-producing)operationsand
terminal(value-orside-effect-producing)operations.Intermediateoperationsarealwayslazy.
1、Streamscanbeobtainedinanumberofways.
·     Froma Collection via
thestream()andparallelStream()methods;
·     Fromanarrayvia Arrays.stream(Object[]);
·     Fromstaticfactorymethodsonthestreamclasses,suchas Stream.of(Object[]), IntStream.range(int,
int) orStream.iterate(Object,UnaryOperator);
·     Thelinesofafilecanbeobtainedfrom BufferedReader.lines();
·     Streamsoffilepathscanbeobtainedfrommethodsin Files;
·     Streamsofrandomnumberscanbeobtainedfrom Random.ints();
·     Numerousotherstream-bearingmethodsintheJDK,including BitSet.stream(),Pattern.splitAsStream(java.lang.CharSequence),
and JarFile.stream().

2、Streamoperationsandpipelines

（1）Streamoperations

1、intermediateoperations

Intermediateoperationsreturnanewstream.Theyarealwayslazy.

Intermediateoperationsarefurth
ca43
erdividedinto
stateless(suchas

filter

and

map---

retainnostatefrompreviouslyseenelementwhenprocessinganewelement)andstatefuloperations(suchas

distinct

and

sorted---

).

Traversalofthepipelinesourcedoesnotbeginuntiltheterminaloperationofthepipelineisexecuted.

2、terminaloperations

Aftertheterminaloperationisperformed,thestreampipelineisconsideredconsumed,andcannolongerbeused.

Inalmostallcases,terminaloperationsareeager,completingtheirtraversalofthedatasourceandprocessingofthepipelinebeforereturning.Onlytheterminaloperationsiterator()andspliterator()arenot.(eg:

Stream.forEachor
IntStream.sum

)

 

Aboutshortcuiting:

 Anintermediateoperationisshort-circuitingif,whenpresentedwithinfiniteinput,itmayproduceafinitestreamasaresult.Aterminaloperationisshort-circuitingif,whenpresentedwithinfiniteinput,itmayterminate
infinitetime.Havingashort-circuitingoperationinthepipelineisanecessary,butnotsufficient,conditionfortheprocessingofaninfinitestreamtoterminatenormallyinfinitetime.

3、SideEffect解析（摘自stackflow）

链接：https://stackoverflow.com/questions/1073909/side-effect-whats-this?noredirect=1

Asideeffectisanythingamethoddoesbesidescomputingandreturningavalue.Anychangeofinstanceorclassfieldvaluesisasideeffect,asisdrawingsomethingonthescreen,writingtoafileoranetworkconnection.

Strictlyspeaking,a"function"isdefinedasnothavingsideeffects-whichiswhyJavausestheword"method"instead.Arealfunctionwithnoreturnvaluewouldbepointless.

Obviously,amethodthatdoesnothaveareturnvaluemusthavesomesortofsideeffectthatjustifiesitsexistence.Setmethodsareanexample-thesideeffectischangingtheobject'sinternalstate

ForExample：

Asideeffectiswhenamethodcallchangesaclass'sstate.So

 

publicclassSideEffectClass{

   privateintstate=0;

  publicdoSomething(intarg0){

       state+=arg0;

   }

}

Here,doSomething(intarg0)hasthesideeffectofchangingthestatevariable.

Whenyouthinkofaprogram,youcanthinkofitasinstructions+state+input.（指令+状态+输入）Soifthedomainofaprogramistherangeofallpossibleinput*state,andtheprogramhassideeffects,youcanseethatthecodomain（值域）
ofpossibleresultsfortheapplicationcangrowexplosively,asthenumberofsideeffectsincrease.Thismakesthepossiblestatesfortheprogramlarge,whichleadstocomplicatedtesting.Thefunctionalprogrammingparadigmisdesignedtoeliminatesideeffects.
Bymakingfunctionsfirstclasscitizensandbymakingalldeclarationsimmutablefunctionalprogrammingpreventssideeffects,whichmakesfunctionalprogrammingshineinparallelprocessing,assynchronizationissuesarereduced.

Examplefromapi：

Asanexampleofhowtotransformastreampipelinethatinappropriatelyusesside-effectstoonethatdoesnot,thefollowingcodesearchesastreamofstringsforthosematchingagivenregularexpression,andputs
thematchesinalist.
    ArrayList<String>results=newArrayList<>();
    stream.filter(s->pattern.matcher(s).matches())
    .forEach(s->results.add(s)); //Unnecessaryuseofside-effects!
Thiscodeunnecessarilyusesside-effects.Ifexecutedinparallel,thenon-thread-safetyofArrayListwouldcauseincorrectresults,andaddingneededsynchronizationwouldcausecontention,underminingthebenefit
ofparallelism.Furthermore,usingside-effectshereiscompletelyunnecessary;theforEach()cansimplybereplacedwithareductionoperationthatissafer,moreefficient,andmoreamenabletoparallelization:
 
    List<String>results=
        stream.filter(s->pattern.matcher(s).matches())
        .collect(Collectors.toList()); //Noside-effects!

4、Ordering

Certainstreamsources(suchasListorarrays)areintrinsicallyordered,whereasothers(suchasHashSet)arenot.Someintermediateoperations,suchassorted(),mayimposeanencounterorderonanotherwiseunordered
stream,andothersmayrenderanorderedstreamunordered,suchasBaseStream.unordered().Further,someterminaloperationsmayignoreencounterorder,suchasforEach().
Ifastreamisordered,mostoperationsareconstrainedtooperateontheelementsintheirencounterorder;ifthesourceofastreamisaListcontaining[1,2,3],thentheresultofexecutingmap(x->x*2)must
be[2,4,6].However,ifthesourcehasnodefinedencounterorder,thenanypermutation（序列）ofthevalues[2,4,6]wouldbeavalidresult.

5、Reductionoperations

Areductionoperation(alsocalledafold)takesasequenceofinputelementsandcombinesthemintoasinglesummaryresultbyrepeatedapplicationofacombiningoperation,suchasfindingthesum
ormaximumofasetofnumbers,oraccumulatingelementsintoalist.Thestreamsclasseshavemultipleformsofgeneralreductionoperations,calledreduce()andcollect(),
aswellasmultiplespecializedreductionformssuchassum(),max(),orcount().

（1）未使用reduce：

intsum=0;

   for(intx:numbers){

      sum+=x;

   }

（2）使用reduce

 

   intsum=numbers.stream().reduce(0,(x,y)->x+y);

  or:

    intsum=numbers.stream().reduce(0,Integer::sum);

（3）areduceoperationonelementsoftype<T>yieldingaresultoftype<U>requiresthreeparameters:

 <U>Ureduce(Uidentity,BiFunction<U,?superT,U>accumulator,

          BinaryOperator<U>combiner);

[/code]

identityelement：bothaninitialseedvalueforthereductionandadefaultresultiftherearenoinputelements.

accumulatorfunction：takesapartialresultandthenextelement,andproducesanewpartialresult.

combinerfunction:combinestwopartialresultstoproduceanewpartialresult.

6、Mutablereduction

Amutablereductionoperationaccumulatesinputelementsintoamutableresultcontainer,suchasaCollectionorStringBuilder,asitprocessestheelementsinthestream.

ordinaryreduction:

   Stringconcatenated=strings.reduce("",String::concat)

[/code]

[/code]

mutablereductionoperation：collect()，itcollectstogetherthedesiredresultsintoaresultcontainersuchasaCollection.

[/code]

Acollectoperationrequiresthreefunctions:

asupplierfunction：toconstructnewinstancesoftheresultcontainer,anaccumulatorfunction：toincorporateaninputelementintoaresultcontainer,

acombiningfunction：tomergethecontentsofoneresultcontainerintoanother.Theformofthisisverysimilartothegeneralformofordinaryreduction:

<R>Rcollect(Supplier<R>supplier,

              BiConsumer<R,?superT>accumulator,

              BiConsumer<R,R>combiner);

Old：

ArrayList<String>strings=newArrayList<>();

[/code]

    for(Telement:stream){

[/code]

        strings.add(element.toString());

[/code]

    }

[/code]

New：

 
 ArrayList<String>strings=stream.collect(()->newArrayList<>(),
                                    (c,e)->c.add(e.toString()),
                                     (c1,c2)->c1.addAll(c2));
 
or,pullingthemappingoperationoutoftheaccumulatorfunction,wecouldexpressitmoresuccinctlyas:
 
    List<String>strings=stream.map(Object::toString)
                                 .collect(ArrayList::new,ArrayList::add,ArrayList::addAll);
Here,oursupplierisjustthe

ArrayListconstructor

,theaccumulatoraddsthestringifiedelementtoan

ArrayList

,andthecombinersimplyuses

addAll

tocopythestringsfromonecontainerintotheother.