Formal Languages and Compilation

408 Pages·2013·6.148 MB·

by  Stefano Crespi Reghizzi, Luca Breveglieri, Angelo Morzenti

Texts in Computer Science
Stefano Crespi Reghizzi
Luca Breveglieri
Angelo Morzenti
Formal
Languages and
Compilation
Second Edition
Texts in Computer Science
Editors
DavidGries
FredB.Schneider
Forfurthervolumes:
www.springer.com/series/3191

Stefano Crespi Reghizzi (cid:2) Luca Breveglieri (cid:2)
Angelo Morzenti
Formal Languages
and Compilation
Second Edition
StefanoCrespiReghizzi AngeloMorzenti
DipartimentoElettronica DipartimentoElettronica
InformazioneeBioingegneria InformazioneeBioingegneria
PolitecnicodiMilano PolitecnicodiMilano
Milan,Italy Milan,Italy
LucaBreveglieri
DipartimentoElettronica
InformazioneeBioingegneria
PolitecnicodiMilano
Milan,Italy
SeriesEditors
DavidGries FredB.Schneider
DepartmentofComputerScience DepartmentofComputerScience
CornellUniversity CornellUniversity
Ithaca,NY,USA Ithaca,NY,USA
ISSN1868-0941 ISSN1868-095X(electronic)
TextsinComputerScience
ISBN978-1-4471-5513-3 ISBN978-1-4471-5514-0(eBook)
DOI10.1007/978-1-4471-5514-0
SpringerLondonHeidelbergNewYorkDordrecht
©Springer-VerlagLondon2009,2013
Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpartof
thematerialisconcerned,specificallytherightsoftranslation,reprinting,reuseofillustrations,recitation,
broadcasting,reproductiononmicrofilmsorinanyotherphysicalway,andtransmissionorinformation
storageandretrieval,electronicadaptation,computersoftware,orbysimilarordissimilarmethodology
nowknownorhereafterdeveloped.Exemptedfromthislegalreservationarebriefexcerptsinconnection
with reviews or scholarly analysis or material supplied specifically for the purpose of being entered
and executed on a computer system, for exclusive use by the purchaser of the work. Duplication of
this publication or parts thereof is permitted only under the provisions of the Copyright Law of the
Publisher’slocation,initscurrentversion,andpermissionforusemustalwaysbeobtainedfromSpringer.
PermissionsforusemaybeobtainedthroughRightsLinkattheCopyrightClearanceCenter.Violations
areliabletoprosecutionundertherespectiveCopyrightLaw.
Theuseofgeneraldescriptivenames,registerednames,trademarks,servicemarks,etc.inthispublication
doesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfromtherelevant
protectivelawsandregulationsandthereforefreeforgeneraluse.
Whiletheadviceandinformationinthisbookarebelievedtobetrueandaccurateatthedateofpub-
lication,neithertheauthorsnortheeditorsnorthepublishercanacceptanylegalresponsibilityforany
errorsoromissionsthatmaybemade.Thepublishermakesnowarranty,expressorimplied,withrespect
tothematerialcontainedherein.
Printedonacid-freepaper
SpringerispartofSpringerScience+BusinessMedia(www.springer.com)
Preface
The textbook derives from the well-known identically titled volume1 published in
2009:twoyoungerco-authorshavebroughttheirexperiencetoenrichandstream-
linethebook,withoutdullingitsoriginalstructure.Theselectionofmaterialsand
thepresentationstylehavenotsubstantiallychanged.Thebookreflectsmanyyears
ofteachingcompilercoursesandofdoingresearchonformallanguagetheoryand
formal methods, on compiler and language technology, and to a lesser extent on
naturallanguageprocessing.Themoreimportantchangeconcernsthecentraltopic
oflanguageparsing.Itisacompletelynew,systematic,andunifiedpresentationof
themostimportantparsingalgorithms,includingalsoparallelparsing.
Goals Intheturmoilofinformationtechnologydevelopments,thesubjectofthe
book has kept the same fundamental principles since half a century, and has pre-
served its conceptual importance and practical relevance. This state of affairs in a
topicthatiscentraltocomputerscienceandisbasedonestablishedprinciples,might
lead some people to believe that the corresponding textbooks are by now consoli-
dated,muchastheclassicalbooksonmathematicsandphysics.Inreality,thisisnot
thecase:thereexistfineclassicalbooksonthemathematicalaspectsoflanguageand
automatatheory,butforwhatconcernstheapplicationtocompiling,thebestbooks
are sort of encyclopedias of algorithms, design methods, and practical tricks used
incompilerdesign.Indeed,acompilerisamicrocosm,andfeaturesmanydifferent
aspectsrangingfromalgorithmicwisdomtocomputerhardware.Asaconsequence,
thetextbookshavegrowninsize,andcompetewithrespecttotheircoverageofthe
last developments on programming languages, processor architectures and clever
mappingsfromtheformertothelatter.
Toputthingsinorder,itisbettertoseparatesuchcomplextopicsintotwoparts,
basicandadvanced,whichtoalargeextentcorrespondtothetwosubsystemsthat
make a compiler: the user-language specific front-end, and the machine-language
specificback-end.Thebasicpartisthesubjectofthisbook.Itcoverstheprinciples
andalgorithmstobeusedfordefiningthesyntaxoflanguagesandforimplementing
simpletranslators.Itdoesnotinclude:thespecializedknow-howneededforvarious
classesofprogramminglanguages(imperative,functional,objectoriented,etc.),the
computerarchitecturerelatedaspects,andtheoptimizationmethodsusedtoimprove
themachinecodeproducedbythecompiler.
1S.CrespiReghizzi,FormalLanguagesandCompilation(Springer,London,2009).
v
vi Preface
OrganizationandFeatures Inothertextbooksthebiastowardspracticalaspects
hasreducedtheattentiontofundamentalconcepts.Thishaspreventedtheirauthors
from taking advantage of the improvements and simplifications made possible by
decades of extensive use, and from avoiding the irritating variants and repetitions
that are found in the original papers. Moving from these premises, we decided to
present, in a simple minimalistway, the principles and methods used in designing
languagesyntaxandsyntax-directedtranslators.
Chapter2coversregularexpressionsandcontext-freegrammars,withemphasis
onthestructuraladequacyofsyntacticdefinitionsandacarefulanalysisofambigu-
ityandhowtoavoidit.
Chapter3presentsfinite-staterecognizersandtheirconversionbackandforthto
regularexpressionsandgrammars.
Chapter4presentspush-downmachinesandparsingalgorithms,focusingatten-
tionontheLL,LRandEarleymethods.Wehavesubstantiallyimprovedthestandard
presentationofsuchalgorithms,byunifyingtheconceptsandnotationsusedinvar-
iousapproaches,andbyextendingthemethodcoveragewithareduceddefinitional
apparatus.Anexamplethatexpertreadersandinstructorsshouldappreciate,isthe
unificationofthetop-down(LL)andbottom-up(LR)parsingalgorithms,aswellas
the tabular (Early) one, within a novel practical framework. In this way, the effort
and space spared have made room for advanced methods typically not present in
similar textbooks. First, our parsing algorithms apply to the Extended BNF gram-
mars, which are the de facto standard in the language reference manuals. Second,
weprovideaparallelparsingalgorithmthattakesadvantageofthemanyprocessing
unitsofmodernmicroprocessors,tospeed-uptheanalysisoflargefiles.
The book is not restricted to syntax: Chap. 5, the last one, studies translations,
semantic functions (attribute grammars), and the static program analysis by data
flow equations. This provides a comprehensive understanding of the compilation
process,andcoverstheessentialelementsofasyntax-directedtranslator.
Thepresentationisillustratedbymanysmallyetrealisticexamplesandpictures,
to ease the understanding of the theory and the transfer to application. Theoret-
ical models of automata, transducers and formal grammars are extensively used,
whenever practical motivations exist, without insisting too much on their formal
definition.Algorithmsaredescribedinapseudo-codetoavoidthedisturbingdetails
of a programming language, yet they are straightforward to convert to executable
procedures.
Thisbookshouldbewelcomebythosewillingtoteachortolearntheessential
concepts of syntax-directed compilation, without the need of relying on software
toolsandimplementations.Webelievethatlearningbydoingisnotalwaysthebest
approach,andthatover-commitmenttopracticalworkmaysometimesobscurethe
conceptualfoundations.Inthecaseofformallanguages,theeleganceandsimplicity
oftheunderlyingtheoryallowsthestudentstoacquirethefundamentalparadigms
of language structures, to avoid pitfalls such as ambiguity, and to adequately map
structuretomeaning.Inthisfield,therelevantalgorithmsaresimpleenoughtobe
practicedbypaperandpencil.Ofcourse,studentsshouldbeencouragedtoenrollin
ahands-onlaboratoryandtoexperimentsyntax-directedtools(likeflexandbison)
onrealisticcases.
Preface vii
IntendedAudiences Thisisprimarilyatextbooktargetedtograduate(orupper-
divisionundergraduate)studentsincomputerscienceorcomputerengineering.We
listasprerequisites:familiaritywithsomeprogramminglanguagemodelsandwith
algorithm design; and the ability to use elementary mathematical and logical no-
tation. If the reader or student has already taken a course on theoretical computer
scienceincludingafewelementsofformallanguageandautomatatheory,thetime
andeffortneededtolearnthefirstchapterscanbereduced.Butitisfairtosaythat
thisbookdoesnotcompetewiththemanyavailablebooksonthetheoryofformal
languagesandcomputation:weusuallydonotincludemathematicalproofsofprop-
erties,andwerelyinsteadonexamplesandinformalarguments.Yetmathematically
orientedreaderswillfindheremanymotivatingexamplesandapplications.
Alargecollectionofproblemsandsolutionscomplementingthenumerousexam-
plesinthebookisavailableontheauthors’coursewebsiteatPolitecnicodiMilano.
Similarly,acomprehensivesetoflectureslidesisalsoavailableonthecourseweb
site.
The Authors Thank The colleagues who have taught this book to computer
engineering students, Giampaolo Agosta and Licia Sbattella; the Italian National
Research Group on Formal Languages, in particular Antonio Restivo, Alessan-
dra Cherubini, Dino Mandrioli, Matteo Pradella and Pierluigi San Pietro; our past
andpresentPh.D.studentsandteachingassistants,AlessandroBarenghi,Marcello
Bersani, Andrea Di Biagio, Simone Campanoni, Silvia Lovergine, Michele Scan-
dale, Ettore Speziale, Martino Sykora and Michele Tartara. We also acknowledge
thesupportof STMicroelectronicsCompanyforR.&D.projectsoncompilersfor
advancedmicroprocessorarchitectures.
The first author remembers the late Antonio Grasselli, a pioneer of computer
sciencestudies,whofirstfascinatedhimwithasubjectcombininglinguistic,math-
ematical,andtechnologicalaspects.
Milan,Italy StefanoCrespiReghizzi
July2013 LucaBreveglieri
AngeloMorzenti
Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 IntendedScopeandAudience . . . . . . . . . . . . . . . . . . . 1
1.2 CompilerPartsandCorrespondingConcepts . . . . . . . . . . . 2
2 Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.1 ArtificialandFormalLanguages . . . . . . . . . . . . . 5
2.1.2 LanguageTypes . . . . . . . . . . . . . . . . . . . . . . 6
2.1.3 ChapterOutline . . . . . . . . . . . . . . . . . . . . . . 7
2.2 FormalLanguageTheory . . . . . . . . . . . . . . . . . . . . . 7
2.2.1 AlphabetandLanguage . . . . . . . . . . . . . . . . . . 7
2.2.2 LanguageOperations . . . . . . . . . . . . . . . . . . . 11
2.2.3 SetOperations . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.4 StarandCross . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.5 Quotient . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3 RegularExpressionsandLanguages . . . . . . . . . . . . . . . 17
2.3.1 DefinitionofRegularExpression . . . . . . . . . . . . . 17
2.3.2 DerivationandLanguage . . . . . . . . . . . . . . . . . 19
2.3.3 OtherOperators . . . . . . . . . . . . . . . . . . . . . . 22
2.3.4 ClosurePropertiesofREGFamily . . . . . . . . . . . . 23
2.4 LinguisticAbstraction . . . . . . . . . . . . . . . . . . . . . . . 24
2.4.1 AbstractandConcreteLists . . . . . . . . . . . . . . . . 25
2.5 Context-FreeGenerativeGrammars . . . . . . . . . . . . . . . . 28
2.5.1 LimitsofRegularLanguages . . . . . . . . . . . . . . . 29
2.5.2 IntroductiontoContext-FreeGrammars . . . . . . . . . 29
2.5.3 ConventionalGrammarRepresentations . . . . . . . . . 32
2.5.4 DerivationandLanguageGeneration . . . . . . . . . . . 33
2.5.5 ErroneousGrammarsandUselessRules . . . . . . . . . 35
2.5.6 RecursionandLanguageInfinity . . . . . . . . . . . . . 37
2.5.7 SyntaxTreesandCanonicalDerivations . . . . . . . . . 38
2.5.8 ParenthesisLanguages . . . . . . . . . . . . . . . . . . . 41
2.5.9 RegularCompositionofContext-FreeLanguages . . . . 43
2.5.10 Ambiguity . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.5.11 CatalogueofAmbiguousFormsandRemedies . . . . . . 47
ix
x Contents
2.5.12 WeakandStructuralEquivalence . . . . . . . . . . . . . 54
2.5.13 GrammarTransformationsandNormalForms . . . . . . 56
2.6 GrammarsofRegularLanguages . . . . . . . . . . . . . . . . . 67
2.6.1 FromRegularExpressionstoContext-FreeGrammars . . 67
2.6.2 LinearGrammars . . . . . . . . . . . . . . . . . . . . . 68
2.6.3 LinearLanguageEquations . . . . . . . . . . . . . . . . 71
2.7 ComparisonofRegularandContext-FreeLanguages . . . . . . . 73
2.7.1 LimitsofContext-FreeLanguages . . . . . . . . . . . . 76
2.7.2 ClosurePropertiesofREGandCF . . . . . . . . . . . . 78
2.7.3 AlphabeticTransformations . . . . . . . . . . . . . . . . 79
2.7.4 GrammarswithRegularExpressions . . . . . . . . . . . 82
2.8 MoreGeneralGrammarsandLanguageFamilies . . . . . . . . . 85
2.8.1 ChomskyClassification . . . . . . . . . . . . . . . . . . 86
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
3 FiniteAutomataasRegularLanguageRecognizers . . . . . . . . . 91
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
3.2 RecognitionAlgorithmsandAutomata . . . . . . . . . . . . . . 92
3.2.1 AGeneralAutomaton . . . . . . . . . . . . . . . . . . . 93
3.3 IntroductiontoFiniteAutomata . . . . . . . . . . . . . . . . . . 96
3.4 DeterministicFiniteAutomata. . . . . . . . . . . . . . . . . . . 97
3.4.1 ErrorStateandTotalAutomata . . . . . . . . . . . . . . 98
3.4.2 CleanAutomata . . . . . . . . . . . . . . . . . . . . . . 99
3.4.3 MinimalAutomata . . . . . . . . . . . . . . . . . . . . . 100
3.4.4 FromAutomatatoGrammars . . . . . . . . . . . . . . . 103
3.5 NondeterministicAutomata . . . . . . . . . . . . . . . . . . . . 104
3.5.1 MotivationofNondeterminism . . . . . . . . . . . . . . 105
3.5.2 NondeterministicRecognizers . . . . . . . . . . . . . . . 107
3.5.3 AutomatawithSpontaneousMoves . . . . . . . . . . . . 109
3.5.4 CorrespondenceBetweenAutomataandGrammars . . . 110
3.5.5 AmbiguityofAutomata . . . . . . . . . . . . . . . . . . 111
3.5.6 Left-LinearGrammarsandAutomata . . . . . . . . . . . 112
3.6 DirectlyfromAutomatatoRegularExpressions:BMCMethod . 112
3.7 EliminationofNondeterminism . . . . . . . . . . . . . . . . . . 114
3.7.1 ConstructionofAccessibleSubsets . . . . . . . . . . . . 115
3.8 FromRegularExpressiontoRecognizer . . . . . . . . . . . . . 119
3.8.1 ThompsonStructuralMethod . . . . . . . . . . . . . . . 119
3.8.2 Berry–SethiMethod . . . . . . . . . . . . . . . . . . . . 121
3.9 RegularExpressionswithComplementandIntersection . . . . . 132
3.9.1 ProductofAutomata . . . . . . . . . . . . . . . . . . . . 134
3.10 SummaryofRelationsBetweenRegularLanguages,Grammars,
andAutomata . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139