外文翻译 - 编译原理技术和工具

外文原文文献CompilersPrinciplesTechniquesAndToolsThefirstphaseofacompileriscalledlexicalanalysisorscanning.Thelexicalanalyzerreadsthestreamofcharactersmakingupthesourceprogramandgroupsthecharactersintomeaningfulsequencescalledlexemes.Foreachlexeme,thelexicalanalyzerproducesasoutputatokenoftheformtoken-name,attribute-valuethatitpassesontothesubsequentphase,syntaxanalysis.Inthetoken,thefirstcomponenttoken-nameisanabstractsymbolthatisusedduringsyntaxanalysis,andthesecondcomponentattribute-valuepointstoanentryinthesymboltableforthistoken.Informationfromthesymbol-tableentryisneededforsemanticanalysisandcodegeneration.Thesecondphaseofthecompilerissyntaxanalysisorparsing.Theparserusesthefirstcomponentsofthetokensproducedbythelexicalanalyzertocreateatree-likeintermediaterepresentationthatdepictsthegrammaticalstructureofthetokenstream.Atypicalrepresentationisasyntaxtreeinwhicheachinteriornoderepresentsanoperationandthechildrenofthenoderepresenttheargumentsoftheoperation.Thesemanticanalyzerusesthesyntaxtreeandtheinformationinthesymboltabletocheckthesourceprogramforsemanticconsistencywiththelanguagedefinition.Italsogatherstypeinformationandsavesitineitherthesyntaxtreeorthesymboltable,forsubsequentuseduringintermediate-codegeneration.Animportantpartofsemanticanalysisistypechecking,wherethecompilerchecksthateachoperatorhasmatchingoperands.Forexample,manyprogramminglanguagedefinitionsrequireanarrayindextobeaninteger;thecompilermustreportanerrorifafloating-pointnumberisusedtoindexanarray.Thecodegeneratortakesasinputanintermediaterepresentationofthesourceprogramandmapsitintothetargetlanguage.Ifthetargetlanguageismachinecode,registersOrmemorylocationsareselectedforeachofthevariablesusedbytheprogram.Then,theintermediateinstructionsaretranslatedintosequencesofmachineinstructionsthatperformthesametask.AcrucialaspectofcodeAnessentialfunctionofacompileristorecordthevariablenamesusedinthesourceprogramandcollectinformationaboutvariousattributesofeachname.Theseattributesmayprovideinformationaboutthestorageallocatedforaname,itstype,itsscope(whereintheprogramitsvaluemaybeused),andinthecaseofprocedurenames,suchthingsasthenumberandtypesofitsarguments,themethodofpassingeachargument(forexample,byvalueorbyreference),andthetypereturned.generationisthejudiciousassignmentofregisterstoholdvariables.Thediscussionofphasesdealswiththelogicalorganizationofacompiler.Inanimplementation,activitiesfromseveralphasesmaybegroupedtogetherintoapassthatreadsaninputfileandwritesanoutputfile.Forexample,thefront-endphasesoflexicalanalysis,syntaxanalysis,semanticanalysis,andintermediatecodegenerationmightbegroupedtogetherintoonepass.Codeoptimizationmightbeanoptionalpass.Thentherecouldbeaback-endpassconsistingofcodegenerationforaparticulartargetmachine.Thecompilerwriter,likeanysoftwaredeveloper,canprofitablyusemodernsoftwaredevelopmentenvironmentscontainingtoolssuchaslanguageeditors,debuggers,versionmanagers,profilers,testharnesses,andsoon.Inadditiontothesegeneralsoftware-developmenttools,othermorespecializedtoolshavebeencreatedtohelpimplementvariousphasesofacompiler.Thesetoolsusespecializedlanguagesforspecifyingandimplementingspecificcomponents,andmanyusequitesophisticatedalgorithms.Themostsuccessfultoolsarethosethathidethedetailsofthegenerationalgorithmandproducecomponentsthatcanbeeasilyintegratedintotheremainderofthecompiler.Somecommonlyusedcompiler-constructiontoolsinclude1.Parsergeneratorsthatautomaticallyproducesyntaxanalyzersfromagrammaticaldescriptionofaprogramminglanguage.2.Scannergeneratorsthatproducelexicalanalyzersfromaregular-expressiondescriptionofthetokensofalanguage.3.Syntax-directedtranslationenginesthatproducecollectionsofroutinesforwalkingaparsetreeandgeneratingintermediatecode.4.Code-generatorgeneratorsthatproduceacodegeneratorfromacollectionofrulesfortranslatingeachoperationoftheintermediatelanguageintothemachinelanguageforatargetmachine.5.Data-flowanalysisenginesthatfacilitatethegatheringofinformationabouthowvaluesaretransmittedfromonepartofaprogramtoeachotherpart.Data-flowanalysisisakeypartofcodeoptimization.6.Compiler-constructiontoolkitsthatprovideanintegratedsetofroutinesforconstructingvariousphasesofacompiler.Compilerdesignisnotonlyaboutcompilers,andmanypeopleusethetechnologylearnedbystudyingcompilersinschool,yethavenever,strictlyspeaking,written(evenpartof)acompilerforamajorprogramminglanguage.Compilertechnologyhasotherimportantusesaswell.Additionally,compilerdesignimpactsseveralotherareasofcomputerscience.外文原文翻译编译原理技术和工具编译器的第一个阶段称为词法分析或扫描。词法分析器读取组成源程序的字符流，并且将它们组织成为有意义的词素的序列。对于每个词素，词法分析器产生如下形式的词法单元作为输出：这个词法单元被传送给下一个步骤，即语法分析。在这个词法单元中，第一个分量token-name是一个由语法分析步骤使用的抽象符号，而第二个分量attribute-value指向符号表中关于这个词法单元的条目。符号表条目的信息会被语义分析和代码生成步骤使用。编译器的第二阶段是语法分析或解析。解析器使用由词法分析器生成的的第一个分量来创建类似树的中间表示。该中间表示给出了词法分析产生的词法单元流的语法结构一个常用的表示方法是语法树，树中的每个内部节点表示一个运算，而该节点的子节点代表该运算的分量。语义分析器使用语法树和符号表中的信息来检查源程序是否和语言定义的语义一致。它同时也收集类型信息，并把这些信息存放在语法树或符号表中，以便在随后的中间代码生成过程中使用。语义分析的一个重要部分是类型检查。编译器检查每个运算符是否具有匹配的运算分量。比如，很多程序设计语言的定义中要求一个数组的下标必须是整数。如果用一个浮点数作为数组下标，编译器就必须报错。代码生成器以源程序的中间表示形式作为输入，并把它映射到目标语言。如果目标语言是机器代码，那么就必须为程序使用的每个变量选择寄存器或内存位置。然后，中间指令被翻译成为能够完成相同任务的机器指令序列。代码生成的一个至关重要的方面是合理分配及寻求以存放变量的值。编译器的重要功能之一是记录源程序中使用的变量的名字，并收集和每个名字的各种属性有关的信息。这些属性可以提供一个名字的存储分配、它的类型、作用域（即在程序的哪些地方可以使用这个名字的值）等信息。对于过程名字，这些信息还包括：它的参数数量和类型、每个参数的传递方法（比如传值或传引用）以及返回类型。前面关于步骤的讨论讲的是一个编译器的逻辑组织方式。在一个特定的实现中，多个步骤的活动可以被组合成一趟。每趟读入一个输入文件并产生一个输出文件。比如，前端步骤中的词法分析、语法分析、语义分析，以及中间代码生成可以组合在一起成为一趟。代码优化可以作为一个可选的趟。然后可以有一个为特定目标机生成代码的后端