| 12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087208820892090209120922093209420952096209720982099210021012102210321042105210621072108210921102111211221132114211521162117211821192120212121222123212421252126212721282129213021312132213321342135213621372138213921402141214221432144214521462147214821492150215121522153215421552156215721582159216021612162216321642165216621672168216921702171217221732174217521762177217821792180218121822183218421852186218721882189219021912192219321942195219621972198219922002201220222032204220522062207220822092210221122122213221422152216221722182219222022212222222322242225222622272228222922302231223222332234223522362237223822392240224122422243224422452246224722482249225022512252225322542255225622572258225922602261226222632264226522662267226822692270227122722273227422752276227722782279228022812282228322842285228622872288228922902291229222932294229522962297229822992300230123022303230423052306230723082309231023112312231323142315231623172318231923202321232223232324232523262327232823292330233123322333233423352336233723382339234023412342234323442345234623472348234923502351235223532354235523562357235823592360236123622363236423652366236723682369237023712372237323742375237623772378237923802381238223832384238523862387238823892390239123922393239423952396239723982399240024012402240324042405240624072408240924102411241224132414241524162417241824192420242124222423242424252426242724282429243024312432243324342435243624372438243924402441244224432444244524462447244824492450245124522453245424552456245724582459246024612462246324642465246624672468246924702471247224732474247524762477247824792480248124822483248424852486248724882489249024912492249324942495249624972498249925002501250225032504250525062507250825092510251125122513251425152516251725182519252025212522252325242525252625272528252925302531253225332534253525362537253825392540254125422543254425452546254725482549255025512552255325542555255625572558255925602561256225632564256525662567256825692570257125722573257425752576257725782579258025812582258325842585258625872588258925902591259225932594259525962597259825992600260126022603260426052606260726082609261026112612261326142615261626172618261926202621262226232624262526262627262826292630263126322633263426352636263726382639264026412642264326442645264626472648264926502651265226532654265526562657265826592660266126622663266426652666266726682669267026712672267326742675267626772678267926802681268226832684268526862687268826892690269126922693269426952696269726982699270027012702270327042705270627072708270927102711271227132714271527162717271827192720272127222723272427252726272727282729273027312732273327342735273627372738273927402741274227432744274527462747274827492750275127522753275427552756275727582759276027612762276327642765276627672768276927702771277227732774277527762777277827792780278127822783278427852786278727882789279027912792279327942795279627972798279928002801280228032804280528062807280828092810281128122813281428152816281728182819282028212822282328242825282628272828282928302831283228332834283528362837283828392840284128422843284428452846284728482849285028512852285328542855285628572858285928602861286228632864286528662867286828692870287128722873287428752876287728782879288028812882288328842885288628872888288928902891289228932894289528962897289828992900290129022903290429052906290729082909291029112912291329142915291629172918291929202921292229232924292529262927292829292930293129322933293429352936293729382939294029412942294329442945294629472948294929502951295229532954295529562957295829592960296129622963296429652966296729682969297029712972297329742975297629772978297929802981298229832984298529862987298829892990299129922993299429952996299729982999300030013002300330043005300630073008300930103011301230133014301530163017301830193020302130223023302430253026302730283029303030313032303330343035303630373038303930403041304230433044304530463047304830493050305130523053305430553056305730583059306030613062306330643065306630673068306930703071307230733074307530763077307830793080308130823083308430853086308730883089309030913092309330943095309630973098309931003101310231033104310531063107310831093110311131123113311431153116311731183119312031213122312331243125312631273128312931303131313231333134313531363137313831393140314131423143314431453146314731483149315031513152315331543155315631573158315931603161316231633164316531663167316831693170317131723173317431753176317731783179318031813182318331843185318631873188318931903191319231933194319531963197319831993200320132023203320432053206320732083209321032113212321332143215321632173218321932203221322232233224322532263227322832293230323132323233323432353236323732383239324032413242324332443245324632473248324932503251325232533254325532563257325832593260326132623263326432653266326732683269327032713272327332743275327632773278327932803281328232833284328532863287328832893290329132923293329432953296329732983299330033013302330333043305330633073308330933103311331233133314331533163317331833193320332133223323332433253326332733283329333033313332333333343335333633373338333933403341334233433344334533463347334833493350335133523353335433553356335733583359336033613362336333643365336633673368336933703371337233733374337533763377337833793380338133823383338433853386338733883389339033913392339333943395339633973398339934003401340234033404340534063407340834093410341134123413341434153416341734183419342034213422342334243425342634273428342934303431343234333434343534363437343834393440344134423443344434453446344734483449345034513452345334543455345634573458345934603461346234633464346534663467346834693470347134723473347434753476347734783479348034813482348334843485348634873488348934903491349234933494349534963497349834993500350135023503350435053506350735083509351035113512351335143515351635173518351935203521352235233524352535263527352835293530353135323533353435353536353735383539354035413542354335443545354635473548354935503551355235533554355535563557355835593560356135623563356435653566356735683569357035713572357335743575357635773578357935803581358235833584358535863587358835893590359135923593359435953596359735983599360036013602360336043605360636073608360936103611361236133614361536163617361836193620362136223623362436253626362736283629363036313632363336343635363636373638363936403641364236433644364536463647364836493650365136523653365436553656365736583659366036613662366336643665366636673668366936703671367236733674367536763677367836793680368136823683368436853686368736883689369036913692369336943695369636973698369937003701370237033704370537063707370837093710371137123713371437153716371737183719372037213722372337243725372637273728372937303731373237333734373537363737373837393740374137423743374437453746374737483749375037513752375337543755375637573758375937603761376237633764376537663767376837693770377137723773377437753776377737783779378037813782378337843785378637873788378937903791379237933794379537963797379837993800380138023803380438053806380738083809381038113812381338143815381638173818381938203821382238233824382538263827382838293830383138323833383438353836383738383839384038413842384338443845384638473848384938503851385238533854385538563857385838593860386138623863386438653866386738683869387038713872387338743875387638773878387938803881388238833884388538863887388838893890389138923893389438953896389738983899390039013902390339043905390639073908390939103911391239133914391539163917391839193920392139223923392439253926392739283929393039313932393339343935393639373938393939403941394239433944394539463947394839493950395139523953395439553956395739583959396039613962396339643965396639673968396939703971397239733974397539763977397839793980398139823983398439853986398739883989399039913992399339943995399639973998399940004001400240034004400540064007400840094010401140124013401440154016401740184019402040214022402340244025402640274028402940304031403240334034403540364037403840394040404140424043404440454046404740484049405040514052405340544055405640574058405940604061406240634064406540664067406840694070407140724073407440754076407740784079408040814082408340844085408640874088408940904091409240934094409540964097409840994100410141024103410441054106410741084109411041114112411341144115411641174118411941204121412241234124412541264127412841294130413141324133413441354136413741384139414041414142414341444145414641474148414941504151415241534154415541564157415841594160416141624163416441654166416741684169417041714172417341744175417641774178417941804181418241834184418541864187418841894190419141924193419441954196419741984199420042014202420342044205420642074208420942104211421242134214421542164217421842194220422142224223422442254226422742284229423042314232423342344235423642374238423942404241424242434244424542464247424842494250425142524253425442554256425742584259426042614262426342644265426642674268426942704271427242734274427542764277427842794280428142824283428442854286428742884289429042914292429342944295429642974298429943004301430243034304430543064307430843094310431143124313431443154316431743184319432043214322432343244325432643274328432943304331433243334334433543364337433843394340434143424343434443454346434743484349435043514352435343544355435643574358435943604361436243634364436543664367436843694370437143724373437443754376437743784379438043814382438343844385438643874388438943904391439243934394439543964397439843994400440144024403440444054406440744084409441044114412441344144415441644174418441944204421442244234424442544264427442844294430443144324433443444354436443744384439444044414442444344444445444644474448444944504451445244534454445544564457445844594460446144624463446444654466446744684469447044714472447344744475447644774478447944804481448244834484448544864487448844894490449144924493449444954496449744984499450045014502450345044505450645074508450945104511451245134514451545164517451845194520452145224523452445254526452745284529453045314532453345344535453645374538453945404541454245434544454545464547454845494550455145524553455445554556455745584559456045614562456345644565456645674568456945704571457245734574457545764577457845794580458145824583458445854586458745884589459045914592459345944595459645974598459946004601460246034604460546064607460846094610461146124613461446154616461746184619462046214622462346244625462646274628462946304631463246334634463546364637463846394640464146424643464446454646464746484649465046514652465346544655465646574658465946604661466246634664466546664667466846694670467146724673467446754676467746784679468046814682468346844685468646874688468946904691469246934694469546964697469846994700470147024703470447054706470747084709471047114712471347144715471647174718471947204721472247234724472547264727472847294730473147324733473447354736473747384739474047414742474347444745474647474748474947504751475247534754475547564757475847594760476147624763476447654766476747684769477047714772477347744775477647774778477947804781478247834784478547864787478847894790479147924793479447954796479747984799480048014802480348044805480648074808480948104811481248134814481548164817481848194820482148224823482448254826482748284829483048314832483348344835483648374838483948404841484248434844484548464847484848494850485148524853485448554856485748584859486048614862486348644865486648674868486948704871487248734874487548764877487848794880488148824883488448854886488748884889489048914892489348944895489648974898489949004901490249034904490549064907490849094910491149124913491449154916491749184919492049214922492349244925492649274928492949304931493249334934493549364937493849394940494149424943494449454946494749484949495049514952495349544955495649574958495949604961496249634964496549664967496849694970497149724973497449754976497749784979498049814982498349844985498649874988498949904991499249934994499549964997499849995000500150025003500450055006500750085009501050115012501350145015501650175018501950205021502250235024502550265027502850295030503150325033503450355036503750385039504050415042504350445045504650475048504950505051505250535054505550565057505850595060506150625063506450655066506750685069507050715072507350745075507650775078507950805081508250835084508550865087508850895090509150925093509450955096509750985099510051015102510351045105510651075108510951105111511251135114511551165117511851195120512151225123512451255126512751285129513051315132513351345135513651375138513951405141514251435144514551465147514851495150515151525153515451555156515751585159516051615162516351645165516651675168516951705171517251735174517551765177517851795180518151825183518451855186518751885189519051915192519351945195519651975198519952005201520252035204520552065207520852095210521152125213521452155216521752185219522052215222522352245225522652275228522952305231523252335234523552365237523852395240524152425243524452455246524752485249525052515252525352545255525652575258525952605261526252635264526552665267526852695270527152725273527452755276527752785279528052815282528352845285528652875288528952905291529252935294529552965297529852995300530153025303530453055306530753085309531053115312531353145315531653175318531953205321532253235324532553265327532853295330533153325333533453355336533753385339534053415342534353445345534653475348534953505351535253535354535553565357535853595360536153625363536453655366536753685369537053715372537353745375537653775378537953805381538253835384538553865387538853895390539153925393539453955396539753985399540054015402540354045405540654075408540954105411541254135414541554165417541854195420542154225423542454255426542754285429543054315432543354345435543654375438543954405441544254435444544554465447544854495450545154525453545454555456545754585459546054615462546354645465546654675468546954705471547254735474547554765477547854795480548154825483548454855486548754885489549054915492549354945495549654975498549955005501550255035504550555065507550855095510551155125513551455155516551755185519552055215522552355245525552655275528552955305531553255335534553555365537553855395540554155425543554455455546554755485549555055515552555355545555555655575558555955605561556255635564556555665567556855695570557155725573557455755576557755785579558055815582558355845585558655875588558955905591559255935594559555965597559855995600560156025603560456055606560756085609561056115612561356145615561656175618561956205621562256235624562556265627562856295630563156325633563456355636563756385639564056415642564356445645564656475648564956505651565256535654565556565657565856595660566156625663566456655666566756685669567056715672567356745675567656775678567956805681568256835684568556865687568856895690569156925693569456955696569756985699570057015702570357045705570657075708570957105711571257135714571557165717571857195720572157225723572457255726572757285729573057315732573357345735573657375738573957405741574257435744574557465747574857495750575157525753575457555756575757585759576057615762576357645765576657675768576957705771577257735774577557765777577857795780578157825783578457855786578757885789579057915792579357945795579657975798579958005801580258035804580558065807580858095810581158125813581458155816581758185819582058215822582358245825582658275828582958305831583258335834583558365837583858395840 |
- //-------------------------------------------------------------------------------------------------------
- // Copyright (C) Microsoft. All rights reserved.
- // Licensed under the MIT license. See LICENSE.txt file in the project root for full license information.
- //-------------------------------------------------------------------------------------------------------
- #include "ParserPch.h"
- namespace UnifiedRegex
- {
- // ----------------------------------------------------------------------
- // CountDomain
- // ----------------------------------------------------------------------
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void CountDomain::Print(DebugWriter* w) const
- {
- if (upper != CharCountFlag && lower == (CharCount)upper)
- {
- w->Print(_u("[%u]"), lower);
- }
- else
- {
- w->Print(_u("[%u-"), lower);
- if (upper == CharCountFlag)
- w->Print(_u("inf]"));
- else
- w->Print(_u("%u]"), (CharCount)upper);
- }
- }
- #endif
- // ----------------------------------------------------------------------
- // Matcher (inlined, called from instruction Exec methods)
- // ----------------------------------------------------------------------
- #define PUSH(contStack, T, ...) (new (contStack.Push<T>()) T(__VA_ARGS__))
- #define PUSHA(assertionStack, T, ...) (new (assertionStack.Push()) T(__VA_ARGS__))
- #define L2I(O, label) LabelToInstPointer<O##Inst>(Inst::InstTag::O, label)
- #define FAIL_PARAMETERS input, inputOffset, instPointer, contStack, assertionStack, qcTicks
- #define HARDFAIL_PARAMETERS(mode) input, inputLength, matchStart, inputOffset, instPointer, contStack, assertionStack, qcTicks, mode
- // Regex QC heuristics:
- // - TicksPerQC
- // - Number of ticks from a previous QC needed to cause another QC. The value affects how often QC will be triggered, so
- // on slower machines or debug builds, the value needs to be smaller to maintain a reasonable frequency of QCs.
- // - TicksPerQcTimeCheck
- // - Number of ticks from a previous QC needed to trigger a time check. Elapsed time from the previous QC is checked to
- // see if a QC needs to be triggered. The value must be less than TicksPerQc and small enough to reasonably guarantee
- // a QC every TimePerQc milliseconds without affecting perf.
- // - TimePerQc
- // - The target time between QCs
- #if defined(_M_ARM)
- const uint Matcher::TicksPerQc = 1u << 19
- #else
- const uint Matcher::TicksPerQc = 1u << (AutoSystemInfo::ShouldQCMoreFrequently() ? 17 : 21)
- #endif
- #if DBG
- >> 2
- #endif
- ;
- const uint Matcher::TicksPerQcTimeCheck = Matcher::TicksPerQc >> 2;
- const uint Matcher::TimePerQc = AutoSystemInfo::ShouldQCMoreFrequently() ? 50 : 100; // milliseconds
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void Matcher::PushStats(ContStack& contStack, const Char* const input) const
- {
- if (stats != 0)
- {
- stats->numPushes++;
- if (contStack.Position() > stats->stackHWM)
- stats->stackHWM = contStack.Position();
- }
- if (w != 0)
- {
- w->Print(_u("PUSH "));
- contStack.Top()->Print(w, input);
- }
- }
- void Matcher::PopStats(ContStack& contStack, const Char* const input) const
- {
- if (stats != 0)
- {
- stats->numPops++;
- }
- if (w != 0)
- {
- const Cont* top = contStack.Top();
- if (top == 0)
- w->PrintEOL(_u("<empty stack>"));
- else
- {
- w->Print(_u("POP "));
- top->Print(w, input);
- }
- }
- }
- void Matcher::UnPopStats(ContStack& contStack, const Char* const input) const
- {
- if (stats != 0)
- {
- stats->numPops--;
- }
- if (w != 0)
- {
- const Cont* top = contStack.Top();
- if (top == 0)
- w->PrintEOL(_u("<empty stack>"));
- else
- {
- w->Print(_u("UNPOP "));
- top->Print(w, input);
- }
- }
- }
- void Matcher::CompStats() const
- {
- if (stats != 0)
- {
- stats->numCompares++;
- }
- }
- void Matcher::InstStats() const
- {
- if (stats != 0)
- {
- stats->numInsts++;
- }
- }
- #endif
- inline void Matcher::QueryContinue(uint &qcTicks)
- {
- // See definition of TimePerQc for description of regex QC heuristics
- Assert(!(TicksPerQc & TicksPerQc - 1)); // must be a power of 2
- Assert(!(TicksPerQcTimeCheck & TicksPerQcTimeCheck - 1)); // must be a power of 2
- Assert(TicksPerQcTimeCheck < TicksPerQc);
- if (PHASE_OFF1(Js::RegexQcPhase))
- {
- return;
- }
- if (++qcTicks & TicksPerQcTimeCheck - 1)
- {
- return;
- }
- DoQueryContinue(qcTicks);
- }
- inline bool Matcher::HardFail(
- const Char* const input
- , const CharCount inputLength
- , CharCount &matchStart
- , CharCount &inputOffset
- , const uint8 *&instPointer
- , ContStack &contStack
- , AssertionStack &assertionStack
- , uint &qcTicks
- , HardFailMode mode)
- {
- switch (mode)
- {
- case HardFailMode::BacktrackAndLater:
- return Fail(FAIL_PARAMETERS);
- case HardFailMode::BacktrackOnly:
- if (Fail(FAIL_PARAMETERS))
- {
- // No use trying any more start positions
- matchStart = inputLength;
- return true; // STOP EXECUTING
- }
- else
- {
- return false;
- }
- case HardFailMode::LaterOnly:
- #if ENABLE_REGEX_CONFIG_OPTIONS
- if (w != 0)
- {
- w->PrintEOL(_u("CLEAR"));
- }
- #endif
- contStack.Clear();
- assertionStack.Clear();
- return true; // STOP EXECUTING
- case HardFailMode::ImmediateFail:
- // No use trying any more start positions
- matchStart = inputLength;
- return true; // STOP EXECUTING
- default:
- Assume(false);
- }
- return true;
- }
- inline bool Matcher::PopAssertion(CharCount &inputOffset, const uint8 *&instPointer, ContStack &contStack, AssertionStack &assertionStack, bool succeeded)
- {
- AssertionInfo* info = assertionStack.Top();
- Assert(info != 0);
- assertionStack.Pop();
- BeginAssertionInst* begin = L2I(BeginAssertion, info->beginLabel);
- // Cut the existing continuations (we never backtrack into an assertion)
- // NOTE: We don't include the effective pops in the stats
- #if ENABLE_REGEX_CONFIG_OPTIONS
- if (w != 0)
- {
- w->PrintEOL(_u("POP TO %llu"), (unsigned long long)info->contStackPosition);
- }
- #endif
- contStack.PopTo(info->contStackPosition);
- // succeeded isNegation action
- // --------- ---------- ----------------------------------------------------------------------------------
- // false false Fail into outer continuations (inner group bindings will have been undone)
- // true false Jump to next label (inner group bindings are now frozen)
- // false true Jump to next label (inner group bindings will have been undone and are now frozen)
- // true true Fail into outer continuations (inner group binding MUST BE CLEARED)
- if (succeeded && begin->isNegation)
- {
- ResetInnerGroups(begin->minBodyGroupId, begin->maxBodyGroupId);
- }
- if (succeeded == begin->isNegation)
- {
- // Assertion failed
- return false;
- }
- else
- {
- // Continue with next label but from original input position
- inputOffset = info->startInputOffset;
- instPointer = LabelToInstPointer(begin->nextLabel);
- return true;
- }
- }
- inline void Matcher::SaveInnerGroups(
- const int fromGroupId,
- const int toGroupId,
- const bool reset,
- const Char *const input,
- ContStack &contStack)
- {
- if (toGroupId >= 0)
- {
- DoSaveInnerGroups(fromGroupId, toGroupId, reset, input, contStack);
- }
- }
- void Matcher::DoSaveInnerGroups(
- const int fromGroupId,
- const int toGroupId,
- const bool reset,
- const Char *const input,
- ContStack &contStack)
- {
- Assert(fromGroupId >= 0);
- Assert(toGroupId >= 0);
- Assert(fromGroupId <= toGroupId);
- int undefinedRangeFromId = -1;
- int groupId = fromGroupId;
- do
- {
- GroupInfo *const groupInfo = GroupIdToGroupInfo(groupId);
- if (groupInfo->IsUndefined())
- {
- if (undefinedRangeFromId < 0)
- {
- undefinedRangeFromId = groupId;
- }
- continue;
- }
- if (undefinedRangeFromId >= 0)
- {
- Assert(groupId > 0);
- DoSaveInnerGroups_AllUndefined(undefinedRangeFromId, groupId - 1, input, contStack);
- undefinedRangeFromId = -1;
- }
- PUSH(contStack, RestoreGroupCont, groupId, *groupInfo);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- PushStats(contStack, input);
- #endif
- if (reset)
- {
- groupInfo->Reset();
- }
- } while (++groupId <= toGroupId);
- if (undefinedRangeFromId >= 0)
- {
- Assert(toGroupId >= 0);
- DoSaveInnerGroups_AllUndefined(undefinedRangeFromId, toGroupId, input, contStack);
- }
- }
- inline void Matcher::SaveInnerGroups_AllUndefined(
- const int fromGroupId,
- const int toGroupId,
- const Char *const input,
- ContStack &contStack)
- {
- if (toGroupId >= 0)
- {
- DoSaveInnerGroups_AllUndefined(fromGroupId, toGroupId, input, contStack);
- }
- }
- void Matcher::DoSaveInnerGroups_AllUndefined(
- const int fromGroupId,
- const int toGroupId,
- const Char *const input,
- ContStack &contStack)
- {
- Assert(fromGroupId >= 0);
- Assert(toGroupId >= 0);
- Assert(fromGroupId <= toGroupId);
- #if DBG
- for (int groupId = fromGroupId; groupId <= toGroupId; ++groupId)
- {
- Assert(GroupIdToGroupInfo(groupId)->IsUndefined());
- }
- #endif
- if (fromGroupId == toGroupId)
- {
- PUSH(contStack, ResetGroupCont, fromGroupId);
- }
- else
- {
- PUSH(contStack, ResetGroupRangeCont, fromGroupId, toGroupId);
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- PushStats(contStack, input);
- #endif
- }
- inline void Matcher::ResetGroup(int groupId)
- {
- GroupInfo* info = GroupIdToGroupInfo(groupId);
- info->Reset();
- }
- inline void Matcher::ResetInnerGroups(int minGroupId, int maxGroupId)
- {
- for (int i = minGroupId; i <= maxGroupId; i++)
- {
- ResetGroup(i);
- }
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- bool Inst::IsBaselineMode()
- {
- return Js::Configuration::Global.flags.BaselineMode;
- }
- Label Inst::GetPrintLabel(Label label)
- {
- return IsBaselineMode() ? (Label)0xFFFF : label;
- }
- template <typename T>
- void Inst::PrintBytes(DebugWriter *w, Inst *inst, T *that, const char16 *annotation) const
- {
- T *start = (T*)that;
- byte *startByte = (byte *)start;
- byte *baseByte = (byte *)inst;
- ptrdiff_t offset = startByte - baseByte;
- size_t size = sizeof(*((T *)that));
- byte *endByte = startByte + size;
- byte *currentByte = startByte;
- w->Print(_u("0x%p[+0x%03x](0x%03x) [%s]:"), startByte, offset, size, annotation);
- for (; currentByte < endByte; ++currentByte)
- {
- if ((currentByte - endByte) % 4 == 0)
- {
- w->Print(_u(" "), *currentByte);
- }
- w->Print(_u("%02x"), *currentByte);
- }
- w->PrintEOL(_u(""));
- }
- template <>
- void Inst::PrintBytes(DebugWriter *w, Inst *inst, Inst *that, const char16 *annotation) const
- {
- Inst *start = (Inst *)that;
- size_t baseSize = sizeof(*(Inst *)that);
- ptrdiff_t offsetToData = (byte *)&(start->tag) - ((byte *)start);
- size_t size = baseSize - offsetToData;
- byte *startByte = (byte *)(&(start->tag));
- byte *endByte = startByte + size;
- byte *currentByte = startByte;
- w->Print(_u("0x%p[+0x%03x](0x%03x) [%s]:"), startByte, offsetToData, size, annotation);
- for (; currentByte < endByte; ++currentByte)
- {
- if ((currentByte - endByte) % 4 == 0)
- {
- w->Print(_u(" "), *currentByte);
- }
- w->Print(_u("%02x"), *currentByte);
- }
- w->PrintEOL(_u(""));
- }
- #define PRINT_BYTES(InstType) \
- Inst::PrintBytes<InstType>(w, (Inst *)this, (InstType *)this, _u(#InstType))
- #define PRINT_BYTES_ANNOTATED(InstType, Annotation) \
- Inst::PrintBytes<InstType>(w, (Inst *)this, (InstType *)this, (Annotation))
- #define PRINT_MIXIN(Mixin) \
- ((Mixin *)this)->Print(w, litbuf)
- #define PRINT_MIXIN_ARGS(Mixin, ...) \
- ((Mixin *)this)->Print(w, litbuf, __VA_ARGS__)
- #define PRINT_MIXIN_COMMA(Mixin) \
- PRINT_MIXIN(Mixin); \
- w->Print(_u(", "));
- #define PRINT_RE_BYTECODE_BEGIN(Name) \
- w->Print(_u("L%04x: "), label); \
- if (REGEX_CONFIG_FLAG(RegexBytecodeDebug)) \
- { \
- w->Print(_u("(0x%03x bytes) "), sizeof(*this)); \
- } \
- w->Print(_u(Name)); \
- w->Print(_u("("));
- #define PRINT_RE_BYTECODE_MID() \
- w->PrintEOL(_u(")")); \
- if (REGEX_CONFIG_FLAG(RegexBytecodeDebug)) \
- { \
- w->Indent(); \
- PRINT_BYTES(Inst);
- #define PRINT_RE_BYTECODE_END() \
- w->Unindent(); \
- } \
- return sizeof(*this);
- #endif
- // ----------------------------------------------------------------------
- // Mixins
- // ----------------------------------------------------------------------
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void BackupMixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("backup: "));
- backup.Print(w);
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void CharMixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("c: "));
- w->PrintQuotedChar(c);
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void Char2Mixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("c0: "));
- w->PrintQuotedChar(cs[0]);
- w->Print(_u(", c1: "));
- w->PrintQuotedChar(cs[1]);
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void Char3Mixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("c0: "));
- w->PrintQuotedChar(cs[0]);
- w->Print(_u(", c1: "));
- w->PrintQuotedChar(cs[1]);
- w->Print(_u(", c2: "));
- w->PrintQuotedChar(cs[2]);
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void Char4Mixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("c0: "));
- w->PrintQuotedChar(cs[0]);
- w->Print(_u(", c1: "));
- w->PrintQuotedChar(cs[1]);
- w->Print(_u(", c2: "));
- w->PrintQuotedChar(cs[2]);
- w->Print(_u(", c3: "));
- w->PrintQuotedChar(cs[3]);
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void LiteralMixin::Print(DebugWriter* w, const char16* litbuf, bool isEquivClass) const
- {
- if (isEquivClass)
- {
- w->Print(_u("equivLiterals: "));
- for (int i = 0; i < CaseInsensitive::EquivClassSize; i++)
- {
- if (i > 0)
- {
- w->Print(_u(", "));
- }
- w->Print(_u("\""));
- for (CharCount j = 0; j < length; j++)
- {
- w->PrintEscapedChar(litbuf[offset + j * CaseInsensitive::EquivClassSize + i]);
- }
- w->Print(_u("\""));
- }
- }
- else
- {
- w->Print(_u("literal: "));
- w->PrintQuotedString(litbuf + offset, length);
- }
- }
- #endif
- // ----------------------------------------------------------------------
- // Char2LiteralScannerMixin
- // ----------------------------------------------------------------------
- bool Char2LiteralScannerMixin::Match(Matcher& matcher, const char16* const input, const CharCount inputLength, CharCount& inputOffset) const
- {
- if (inputLength == 0)
- {
- return false;
- }
- const uint matchC0 = Chars<char16>::CTU(cs[0]);
- const uint matchC1 = Chars<char16>::CTU(cs[1]);
- const char16 * currentInput = input + inputOffset;
- const char16 * endInput = input + inputLength - 1;
- while (currentInput < endInput)
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- while (true)
- {
- const uint c1 = Chars<char16>::CTU(currentInput[1]);
- if (c1 != matchC1)
- {
- if (c1 == matchC0)
- {
- break;
- }
- currentInput += 2;
- if (currentInput >= endInput)
- {
- return false;
- }
- continue;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- // Check the first character
- const uint c0 = Chars<char16>::CTU(*currentInput);
- if (c0 == matchC0)
- {
- inputOffset = (CharCount)(currentInput - input);
- return true;
- }
- if (matchC0 == matchC1)
- {
- break;
- }
- currentInput +=2;
- if (currentInput >= endInput)
- {
- return false;
- }
- }
- // If the second character in the buffer matches the first in the pattern, continue
- // to see if the next character has the second in the pattern
- currentInput++;
- while (currentInput < endInput)
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- const uint c1 = Chars<char16>::CTU(currentInput[1]);
- if (c1 == matchC1)
- {
- inputOffset = (CharCount)(currentInput - input);
- return true;
- }
- if (c1 != matchC0)
- {
- currentInput += 2;
- break;
- }
- currentInput++;
- }
- }
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void Char2LiteralScannerMixin::Print(DebugWriter* w, const char16 * litbuf) const
- {
- Char2Mixin::Print(w, litbuf);
- w->Print(_u(" (with two character literal scanner)"));
- }
- #endif
- // ----------------------------------------------------------------------
- // ScannerMixinT
- // ----------------------------------------------------------------------
- template <typename ScannerT>
- void ScannerMixinT<ScannerT>::FreeBody(ArenaAllocator* rtAllocator)
- {
- scanner.FreeBody(rtAllocator, length);
- }
- template <typename ScannerT>
- inline bool
- ScannerMixinT<ScannerT>::Match(Matcher& matcher, const char16 * const input, const CharCount inputLength, CharCount& inputOffset) const
- {
- Assert(length <= matcher.program->rep.insts.litbufLen - offset);
- return scanner.template Match<1>(
- input
- , inputLength
- , inputOffset
- , matcher.program->rep.insts.litbuf + offset
- , length
- #if ENABLE_REGEX_CONFIG_OPTIONS
- , matcher.stats
- #endif
- );
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- template <typename ScannerT>
- void ScannerMixinT<ScannerT>::Print(DebugWriter* w, const char16* litbuf, bool isEquivClass) const
- {
- LiteralMixin::Print(w, litbuf, isEquivClass);
- w->Print(_u(" (with %s scanner)"), ScannerT::GetName());
- }
- #endif
- // explicit instantiation
- template struct ScannerMixinT<TextbookBoyerMoore<char16>>;
- template struct ScannerMixinT<TextbookBoyerMooreWithLinearMap<char16>>;
- // ----------------------------------------------------------------------
- // EquivScannerMixinT
- // ----------------------------------------------------------------------
- template <uint lastPatCharEquivClassSize>
- inline bool EquivScannerMixinT<lastPatCharEquivClassSize>::Match(Matcher& matcher, const char16* const input, const CharCount inputLength, CharCount& inputOffset) const
- {
- Assert(length * CaseInsensitive::EquivClassSize <= matcher.program->rep.insts.litbufLen - offset);
- CompileAssert(lastPatCharEquivClassSize >= 1 && lastPatCharEquivClassSize <= CaseInsensitive::EquivClassSize);
- return scanner.Match<CaseInsensitive::EquivClassSize, lastPatCharEquivClassSize>(
- input
- , inputLength
- , inputOffset
- , matcher.program->rep.insts.litbuf + offset
- , length
- #if ENABLE_REGEX_CONFIG_OPTIONS
- , matcher.stats
- #endif
- );
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- template <uint lastPatCharEquivClassSize>
- void EquivScannerMixinT<lastPatCharEquivClassSize>::Print(DebugWriter* w, const char16* litbuf) const
- {
- __super::Print(w, litbuf, true);
- w->Print(_u(" (last char equiv size:%d)"), lastPatCharEquivClassSize);
- }
- // explicit instantiation
- template struct EquivScannerMixinT<1>;
- #endif
- // ----------------------------------------------------------------------
- // ScannerInfo
- // ----------------------------------------------------------------------
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void ScannerInfo::Print(DebugWriter* w, const char16* litbuf) const
- {
- ScannerMixin::Print(w, litbuf, isEquivClass);
- }
- #endif
- ScannerInfo* ScannersMixin::Add(Recycler *recycler, Program *program, CharCount offset, CharCount length, bool isEquivClass)
- {
- Assert(numLiterals < MaxNumSyncLiterals);
- return program->AddScannerForSyncToLiterals(recycler, numLiterals++, offset, length, isEquivClass);
- }
- void ScannersMixin::FreeBody(ArenaAllocator* rtAllocator)
- {
- for (int i = 0; i < numLiterals; i++)
- {
- infos[i]->FreeBody(rtAllocator);
- #if DBG
- infos[i] = nullptr;
- #endif
- }
- #if DBG
- numLiterals = 0;
- #endif
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void ScannersMixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("literals: {"));
- for (int i = 0; i < numLiterals; i++)
- {
- if (i > 0)
- {
- w->Print(_u(", "));
- }
- infos[i]->Print(w, litbuf);
- }
- w->Print(_u("}"));
- }
- #endif
- template<bool IsNegation>
- void SetMixin<IsNegation>::FreeBody(ArenaAllocator* rtAllocator)
- {
- set.FreeBody(rtAllocator);
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- template<bool IsNegation>
- void SetMixin<IsNegation>::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("set: "));
- if (IsNegation)
- {
- w->Print(_u("not "));
- }
- set.Print(w);
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void TrieMixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- trie.Print(w);
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void GroupMixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("groupId: %d"), groupId);
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void ChompBoundedMixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("repeats: "));
- repeats.Print(w);
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void JumpMixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("targetLabel: L%04x"), Inst::GetPrintLabel(targetLabel));
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void BodyGroupsMixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("minBodyGroupId: %d, maxBodyGroupId: %d"), minBodyGroupId, maxBodyGroupId);
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void BeginLoopBasicsMixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("loopId: %d, repeats: "), loopId);
- repeats.Print(w);
- w->Print(_u(", hasOuterLoops: %s"), hasOuterLoops ? _u("true") : _u("false"));
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void BeginLoopMixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- BeginLoopBasicsMixin::Print(w, litbuf);
- w->Print(_u(", hasInnerNondet: %s, exitLabel: L%04x, "),
- hasInnerNondet ? _u("true") : _u("false"), Inst::GetPrintLabel(exitLabel));
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void GreedyMixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("greedy: %s"), isGreedy ? _u("true") : _u("false"));
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void RepeatLoopMixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("beginLabel: L%04x"), Inst::GetPrintLabel(beginLabel));
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void GreedyLoopNoBacktrackMixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("loopId: %d, exitLabel: L%04x"), loopId, exitLabel);
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void TryMixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("failLabel: L%04x"), Inst::GetPrintLabel(failLabel));
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void NegationMixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("isNegation: %s"), isNegation ? _u("true") : _u("false"));
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void NextLabelMixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("nextLabel: L%04x"), Inst::GetPrintLabel(nextLabel));
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void FixedLengthMixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("length: %u"), length);
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void FollowFirstMixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("followFirst: %c"), followFirst);
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void NoNeedToSaveMixin::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->Print(_u("noNeedToSave: %s"), noNeedToSave ? _u("true") : _u("false"));
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void SwitchCase::Print(DebugWriter* w) const
- {
- w->Print(_u("case "));
- w->PrintQuotedChar(c);
- w->PrintEOL(_u(": Jump(L%04x)"), targetLabel);
- }
- #endif
- template <uint8 n>
- void SwitchMixin<n>::AddCase(char16 c, Label targetLabel)
- {
- AnalysisAssert(numCases < MaxCases);
- uint8 i;
- for (i = 0; i < numCases; i++)
- {
- Assert(cases[i].c != c);
- if (cases[i].c > c)
- {
- break;
- }
- }
- __analysis_assume(numCases < MaxCases);
- for (uint8 j = numCases; j > i; j--)
- {
- cases[j] = cases[j - 1];
- }
- cases[i].c = c;
- cases[i].targetLabel = targetLabel;
- numCases++;
- }
- void UnifiedRegexSwitchMixinForceAllInstantiations()
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- #define SWITCH_FORCE_INSTANTIATION_PRINT x.Print(0, 0)
- #else
- #define SWITCH_FORCE_INSTANTIATION_PRINT
- #endif
- #define SWITCH_FORCE_INSTANTIATION(n) \
- { \
- SwitchMixin<n> x; \
- x.AddCase(0, 0); \
- SWITCH_FORCE_INSTANTIATION_PRINT; \
- }
- SWITCH_FORCE_INSTANTIATION(2);
- SWITCH_FORCE_INSTANTIATION(4);
- SWITCH_FORCE_INSTANTIATION(8);
- SWITCH_FORCE_INSTANTIATION(16);
- SWITCH_FORCE_INSTANTIATION(24);
- #undef SWITCH_FORCE_INSTANTIATION_PRINT
- #undef SWITCH_FORCE_INSTANTIATION
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- template <uint8 n>
- void SwitchMixin<n>::Print(DebugWriter* w, const char16* litbuf) const
- {
- w->EOL();
- w->Indent();
- for (uint8 i = 0; i < numCases; i++)
- {
- cases[i].Print(w);
- }
- w->Unindent();
- }
- #endif
- // ----------------------------------------------------------------------
- // NopInst
- // ----------------------------------------------------------------------
- inline bool NopInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- return false; // don't stop execution
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int NopInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("Nop");
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(NopInst);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // FailInst
- // ----------------------------------------------------------------------
- inline bool FailInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int FailInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("Fail");
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(NopInst);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // SuccInst
- // ----------------------------------------------------------------------
- inline bool SuccInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- GroupInfo* info = matcher.GroupIdToGroupInfo(0);
- info->offset = matchStart;
- info->length = inputOffset - matchStart;
- return true; // STOP MATCHING
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int SuccInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("Succ");
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(NopInst);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // JumpInst
- // ----------------------------------------------------------------------
- inline bool JumpInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- instPointer = matcher.LabelToInstPointer(targetLabel);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int JumpInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("Jump");
- PRINT_MIXIN(JumpMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(JumpMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // JumpIfNotCharInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool JumpIfNotCharInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset < inputLength && input[inputOffset] == c)
- {
- instPointer += sizeof(*this);
- }
- else
- {
- instPointer = matcher.LabelToInstPointer(targetLabel);
- }
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int JumpIfNotCharInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("JumpIfNotChar");
- PRINT_MIXIN_COMMA(CharMixin);
- PRINT_MIXIN(JumpMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(CharMixin);
- PRINT_BYTES(JumpMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // MatchCharOrJumpInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool MatchCharOrJumpInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset < inputLength && input[inputOffset] == c)
- {
- inputOffset++;
- instPointer += sizeof(*this);
- }
- else
- {
- instPointer = matcher.LabelToInstPointer(targetLabel);
- }
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int MatchCharOrJumpInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("MatchCharOrJump");
- PRINT_MIXIN_COMMA(CharMixin);
- PRINT_MIXIN(JumpMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(CharMixin);
- PRINT_BYTES(JumpMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // JumpIfNotSetInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool JumpIfNotSetInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset < inputLength && set.Get(input[inputOffset]))
- {
- instPointer += sizeof(*this);
- }
- else
- {
- instPointer = matcher.LabelToInstPointer(targetLabel);
- }
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int JumpIfNotSetInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("JumpIfNotSet");
- PRINT_MIXIN_COMMA(SetMixin<false>);
- PRINT_MIXIN(JumpMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(SetMixin<false>);
- PRINT_BYTES(JumpMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // MatchSetOrJumpInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool MatchSetOrJumpInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset < inputLength && set.Get(input[inputOffset]))
- {
- inputOffset++;
- instPointer += sizeof(*this);
- }
- else
- {
- instPointer = matcher.LabelToInstPointer(targetLabel);
- }
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int MatchSetOrJumpInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("MatchSetOrJump");
- PRINT_MIXIN_COMMA(SetMixin<false>);
- PRINT_MIXIN(JumpMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(SetMixin<false>);
- PRINT_BYTES(JumpMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // Switch(AndConsume)Inst (optimized instructions)
- // ----------------------------------------------------------------------
- #if ENABLE_REGEX_CONFIG_OPTIONS
- #define COMP_STATS matcher.CompStats()
- #define SwitchAndConsumeInstPrintImpl(BaseName, n) \
- int BaseName##n##Inst::Print(DebugWriter* w, Label label, const Char* litbuf) const \
- { \
- PRINT_RE_BYTECODE_BEGIN("SwitchAndConsume"#n); \
- PRINT_MIXIN(SwitchMixin<n>); \
- PRINT_RE_BYTECODE_MID(); \
- PRINT_BYTES(SwitchMixin<n>); \
- PRINT_RE_BYTECODE_END(); \
- }
- #else
- #define COMP_STATS
- #define SwitchAndConsumeInstPrintImpl(BaseName, n)
- #endif
- #define SwitchAndConsumeInstImpl(BaseName, n) \
- inline bool BaseName##n##Inst::Exec(REGEX_INST_EXEC_PARAMETERS) const \
- { \
- if (inputOffset >= inputLength) \
- { \
- return matcher.Fail(FAIL_PARAMETERS); \
- } \
- \
- const uint8 localNumCases = numCases; \
- for (int i = 0; i < localNumCases; i++) \
- { \
- COMP_STATS; \
- if (cases[i].c == input[inputOffset]) \
- { \
- CONSUME; \
- instPointer = matcher.LabelToInstPointer(cases[i].targetLabel); \
- return false; \
- } \
- else if (cases[i].c > input[inputOffset]) \
- { \
- break; \
- } \
- } \
- \
- instPointer += sizeof(*this); \
- return false; \
- } \
- SwitchAndConsumeInstPrintImpl(BaseName, n);
- #define CONSUME
- SwitchAndConsumeInstImpl(Switch, 2);
- SwitchAndConsumeInstImpl(Switch, 4);
- SwitchAndConsumeInstImpl(Switch, 8);
- SwitchAndConsumeInstImpl(Switch, 16);
- SwitchAndConsumeInstImpl(Switch, 24);
- #undef CONSUME
- #define CONSUME inputOffset++
- SwitchAndConsumeInstImpl(SwitchAndConsume, 2);
- SwitchAndConsumeInstImpl(SwitchAndConsume, 4);
- SwitchAndConsumeInstImpl(SwitchAndConsume, 8);
- SwitchAndConsumeInstImpl(SwitchAndConsume, 16);
- SwitchAndConsumeInstImpl(SwitchAndConsume, 24);
- #undef CONSUME
- #undef COMP_STATS
- #undef SwitchAndConsumeInstPrintImpl
- #undef SwitchAndConsumeInstImpl
- // ----------------------------------------------------------------------
- // BOITestInst
- // ----------------------------------------------------------------------
- template <>
- BOITestInst<true>::BOITestInst() : Inst(InstTag::BOIHardFailTest) {}
- template <>
- BOITestInst<false>::BOITestInst() : Inst(InstTag::BOITest) {}
- template <bool canHardFail>
- inline bool BOITestInst<canHardFail>::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- if (inputOffset > 0)
- {
- if (canHardFail)
- {
- // Clearly trying to start from later in the input won't help, and we know backtracking can't take us earlier in the input
- return matcher.HardFail(HARDFAIL_PARAMETERS(HardFailMode::ImmediateFail));
- }
- else
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- }
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- template <bool canHardFail>
- int BOITestInst<canHardFail>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- if (canHardFail)
- {
- PRINT_RE_BYTECODE_BEGIN("BOIHardFailTest");
- }
- else
- {
- PRINT_RE_BYTECODE_BEGIN("BOITest");
- }
- w->Print(_u("<hardFail>: %s"), canHardFail ? _u("true") : _u("false"));
- PRINT_RE_BYTECODE_MID();
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // EOITestInst
- // ----------------------------------------------------------------------
- template <>
- EOITestInst<true>::EOITestInst() : Inst(InstTag::EOIHardFailTest) {}
- template <>
- EOITestInst<false>::EOITestInst() : Inst(InstTag::EOITest) {}
- template <bool canHardFail>
- inline bool EOITestInst<canHardFail>::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- if (inputOffset < inputLength)
- {
- if (canHardFail)
- {
- // We know backtracking can never take us later in the input, but starting from later in the input could help
- return matcher.HardFail(HARDFAIL_PARAMETERS(HardFailMode::LaterOnly));
- }
- else
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- }
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- template <bool canHardFail>
- int EOITestInst<canHardFail>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- if (canHardFail)
- {
- PRINT_RE_BYTECODE_BEGIN("EOIHardFailTest");
- }
- else
- {
- PRINT_RE_BYTECODE_BEGIN("EOITest");
- }
- w->Print(_u("<hardFail>: %s"), canHardFail ? _u("true") : _u("false"));
- PRINT_RE_BYTECODE_MID();
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // BOLTestInst
- // ----------------------------------------------------------------------
- inline bool BOLTestInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset > 0 && !matcher.standardChars->IsNewline(input[inputOffset - 1]))
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int BOLTestInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("BOLTest");
- PRINT_RE_BYTECODE_MID();
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // EOLTestInst
- // ----------------------------------------------------------------------
- inline bool EOLTestInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset < inputLength && !matcher.standardChars->IsNewline(input[inputOffset]))
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int EOLTestInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("EOLTest");
- PRINT_RE_BYTECODE_MID();
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // WordBoundaryTestInst
- // ----------------------------------------------------------------------
- template <>
- WordBoundaryTestInst<true>::WordBoundaryTestInst() : Inst(InstTag::NegatedWordBoundaryTest) {}
- template <>
- WordBoundaryTestInst<false>::WordBoundaryTestInst() : Inst(InstTag::WordBoundaryTest) {}
- template <bool isNegation>
- inline bool WordBoundaryTestInst<isNegation>::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- const bool prev = inputOffset > 0 && matcher.standardChars->IsWord(input[inputOffset - 1]);
- const bool curr = inputOffset < inputLength && matcher.standardChars->IsWord(input[inputOffset]);
- if (isNegation == (prev != curr))
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- template <bool isNegation>
- int WordBoundaryTestInst<isNegation>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- if (isNegation)
- {
- PRINT_RE_BYTECODE_BEGIN("NegatedWordBoundaryTest");
- }
- else
- {
- PRINT_RE_BYTECODE_BEGIN("WordBoundaryTest");
- }
- PRINT_RE_BYTECODE_MID();
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // MatchCharInst
- // ----------------------------------------------------------------------
- inline bool MatchCharInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset >= inputLength || input[inputOffset] != c)
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- inputOffset++;
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int MatchCharInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("MatchChar");
- PRINT_MIXIN(CharMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(CharMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // MatchChar2Inst
- // ----------------------------------------------------------------------
- inline bool MatchChar2Inst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset >= inputLength || (input[inputOffset] != cs[0] && input[inputOffset] != cs[1]))
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- inputOffset++;
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int MatchChar2Inst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("MatchChar2");
- PRINT_MIXIN(Char2Mixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(Char2Mixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // MatchChar3Inst
- // ----------------------------------------------------------------------
- inline bool MatchChar3Inst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset >= inputLength || (input[inputOffset] != cs[0] && input[inputOffset] != cs[1] && input[inputOffset] != cs[2]))
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- inputOffset++;
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int MatchChar3Inst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("MatchChar3");
- PRINT_MIXIN(Char3Mixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(Char3Mixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // MatchChar4Inst
- // ----------------------------------------------------------------------
- inline bool MatchChar4Inst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset >= inputLength || (input[inputOffset] != cs[0] && input[inputOffset] != cs[1] && input[inputOffset] != cs[2] && input[inputOffset] != cs[3]))
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- inputOffset++;
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int MatchChar4Inst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("MatchChar4");
- PRINT_MIXIN(Char4Mixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(Char4Mixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // MatchSetInst
- // ----------------------------------------------------------------------
- template<bool IsNegation>
- inline bool MatchSetInst<IsNegation>::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset >= inputLength || this->set.Get(input[inputOffset]) == IsNegation)
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- inputOffset++;
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- template<bool IsNegation>
- int MatchSetInst<IsNegation>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- if (IsNegation)
- {
- PRINT_RE_BYTECODE_BEGIN("MatchNegatedSet");
- PRINT_MIXIN(SetMixin<true>);
- }
- else
- {
- PRINT_RE_BYTECODE_BEGIN("MatchSet");
- PRINT_MIXIN(SetMixin<false>);
- }
- PRINT_RE_BYTECODE_MID();
- IsNegation ? PRINT_BYTES(SetMixin<true>) : PRINT_BYTES(SetMixin<false>);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // MatchLiteralInst
- // ----------------------------------------------------------------------
- inline bool MatchLiteralInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- Assert(length <= matcher.program->rep.insts.litbufLen - offset);
- if (length > inputLength - inputOffset)
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- const Char *const literalBuffer = matcher.program->rep.insts.litbuf;
- const Char * literalCurr = literalBuffer + offset;
- const Char * inputCurr = input + inputOffset;
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (*literalCurr != *inputCurr)
- {
- inputOffset++;
- return matcher.Fail(FAIL_PARAMETERS);
- }
- const Char *const literalEnd = literalCurr + length;
- literalCurr++;
- inputCurr++;
- while (literalCurr < literalEnd)
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (*literalCurr != *inputCurr++)
- {
- inputOffset = (CharCount)(inputCurr - input);
- return matcher.Fail(FAIL_PARAMETERS);
- }
- literalCurr++;
- }
- inputOffset = (CharCount)(inputCurr - input);
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int MatchLiteralInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("MatchLiteral");
- PRINT_MIXIN_ARGS(LiteralMixin, false);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(LiteralMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // MatchLiteralEquivInst
- // ----------------------------------------------------------------------
- inline bool MatchLiteralEquivInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- if (length > inputLength - inputOffset)
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- const Char *const literalBuffer = matcher.program->rep.insts.litbuf;
- CharCount literalOffset = offset;
- const CharCount literalEndOffset = offset + length * CaseInsensitive::EquivClassSize;
- Assert(literalEndOffset <= matcher.program->rep.insts.litbufLen);
- CompileAssert(CaseInsensitive::EquivClassSize == 4);
- do
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (input[inputOffset] != literalBuffer[literalOffset]
- && input[inputOffset] != literalBuffer[literalOffset + 1]
- && input[inputOffset] != literalBuffer[literalOffset + 2]
- && input[inputOffset] != literalBuffer[literalOffset + 3])
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- inputOffset++;
- literalOffset += CaseInsensitive::EquivClassSize;
- }
- while (literalOffset < literalEndOffset);
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int MatchLiteralEquivInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("MatchLiteralEquiv");
- PRINT_MIXIN_ARGS(LiteralMixin, true);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(LiteralMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // MatchTrieInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool MatchTrieInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- if (!trie.Match(
- input
- , inputLength
- , inputOffset
- #if ENABLE_REGEX_CONFIG_OPTIONS
- , matcher.stats
- #endif
- ))
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- instPointer += sizeof(*this);
- return false;
- }
- void MatchTrieInst::FreeBody(ArenaAllocator* rtAllocator)
- {
- trie.FreeBody(rtAllocator);
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int MatchTrieInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("MatchTrie");
- PRINT_MIXIN(TrieMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(TrieMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // OptMatchCharInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool OptMatchCharInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset < inputLength && input[inputOffset] == c)
- {
- inputOffset++;
- }
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int OptMatchCharInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("OptMatchChar");
- PRINT_MIXIN(CharMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(CharMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // OptMatchSetInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool OptMatchSetInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset < inputLength && set.Get(input[inputOffset]))
- {
- inputOffset++;
- }
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int OptMatchSetInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("OptMatchSet");
- PRINT_MIXIN(SetMixin<false>);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(SetMixin<false>);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // SyncToCharAndContinueInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool SyncToCharAndContinueInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- const Char matchC = c;
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- while (inputOffset < inputLength && input[inputOffset] != matchC)
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- inputOffset++;
- }
- matchStart = inputOffset;
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int SyncToCharAndContinueInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("SyncToCharAndContinue");
- PRINT_MIXIN(CharMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(CharMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // SyncToChar2SetAndContinueInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool SyncToChar2SetAndContinueInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- const Char matchC0 = cs[0];
- const Char matchC1 = cs[1];
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- while (inputOffset < inputLength && input[inputOffset] != matchC0 && input[inputOffset] != matchC1)
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- inputOffset++;
- }
- matchStart = inputOffset;
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int SyncToChar2SetAndContinueInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("SyncToChar2SetAndContinue");
- PRINT_MIXIN(Char2Mixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(Char2Mixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // SyncToSetAndContinueInst (optimized instruction)
- // ----------------------------------------------------------------------
- template<bool IsNegation>
- inline bool SyncToSetAndContinueInst<IsNegation>::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- const RuntimeCharSet<Char>& matchSet = this->set;
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- while (inputOffset < inputLength && matchSet.Get(input[inputOffset]) == IsNegation)
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- inputOffset++;
- }
- matchStart = inputOffset;
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- template<bool IsNegation>
- int SyncToSetAndContinueInst<IsNegation>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- if (IsNegation)
- {
- PRINT_RE_BYTECODE_BEGIN("SyncToNegatedSetAndContinue");
- PRINT_MIXIN(SetMixin<true>);
- }
- else
- {
- PRINT_RE_BYTECODE_BEGIN("SyncToSetAndContinue");
- PRINT_MIXIN(SetMixin<false>);
- }
- PRINT_RE_BYTECODE_MID();
- IsNegation ? PRINT_BYTES(SetMixin<true>) : PRINT_BYTES(SetMixin<false>);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // SyncToLiteralAndContinueInst (optimized instruction)
- // ----------------------------------------------------------------------
- template <typename ScannerT>
- inline bool SyncToLiteralAndContinueInstT<ScannerT>::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- if (!this->Match(matcher, input, inputLength, inputOffset))
- {
- return matcher.HardFail(HARDFAIL_PARAMETERS(HardFailMode::ImmediateFail));
- }
- matchStart = inputOffset;
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- // explicit instantiation
- template struct SyncToLiteralAndContinueInstT<Char2LiteralScannerMixin>;
- template struct SyncToLiteralAndContinueInstT<ScannerMixin>;
- template struct SyncToLiteralAndContinueInstT<ScannerMixin_WithLinearCharMap>;
- template struct SyncToLiteralAndContinueInstT<EquivScannerMixin>;
- template struct SyncToLiteralAndContinueInstT<EquivTrivialLastPatCharScannerMixin>;
- // Explicitly define each of these 5 Print functions so that the output will show the actual template param mixin and
- // actual opcode name, even though the logic is basically the same in each definition. See notes below.
- template <>
- int SyncToLiteralAndContinueInstT<Char2LiteralScannerMixin>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- // NOTE: this text is unique to this instantiation
- PRINT_RE_BYTECODE_BEGIN("SyncToLiteralAndContinueInstT<Char2LiteralScannerMixin> aka SyncToChar2LiteralAndContinue");
- PRINT_MIXIN(Char2LiteralScannerMixin); // NOTE: would work with template <typename ScannerT> ScannerT::Print
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(Char2LiteralScannerMixin); // NOTE: unique because macro expansion and _u(#InstType) happen before template is evaluated (so text would be ScannerT)
- PRINT_RE_BYTECODE_END();
- }
- template <>
- int SyncToLiteralAndContinueInstT<ScannerMixin>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- // NOTE: this text is unique to this instantiation
- PRINT_RE_BYTECODE_BEGIN("SyncToLiteralAndContinueInstT<ScannerMixin> aka SyncToLiteralAndContinue");
- PRINT_MIXIN(ScannerMixin); // NOTE: would work with template <typename ScannerT> ScannerT::Print
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(ScannerMixin); // NOTE: unique because macro expansion and _u(#InstType) happen before template is evaluated (so text would be ScannerT)
- PRINT_RE_BYTECODE_END();
- }
- template <>
- int SyncToLiteralAndContinueInstT<ScannerMixin_WithLinearCharMap>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- // NOTE: this text is unique to this instantiation
- PRINT_RE_BYTECODE_BEGIN("SyncToLiteralAndContinueInstT<ScannerMixin_WithLinearCharMap> aka SyncToLinearLiteralAndContinue");
- PRINT_MIXIN(ScannerMixin_WithLinearCharMap); // NOTE: would work with template <typename ScannerT> ScannerT::Print
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(ScannerMixin_WithLinearCharMap); // NOTE: unique because macro expansion and _u(#InstType) happen before template is evaluated (so text would be ScannerT)
- PRINT_RE_BYTECODE_END();
- }
- template <>
- int SyncToLiteralAndContinueInstT<EquivScannerMixin>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- // NOTE: this text is unique to this instantiation
- PRINT_RE_BYTECODE_BEGIN("SyncToLiteralAndContinueInstT<EquivScannerMixin> aka SyncToLiteralEquivAndContinue");
- PRINT_MIXIN(EquivScannerMixin); // NOTE: would work with template <typename ScannerT> ScannerT::Print
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(EquivScannerMixin); // NOTE: unique because macro expansion and _u(#InstType) happen before template is evaluated (so text would be ScannerT)
- PRINT_RE_BYTECODE_END();
- }
- template <>
- int SyncToLiteralAndContinueInstT<EquivTrivialLastPatCharScannerMixin>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- // NOTE: this text is unique to this instantiation
- PRINT_RE_BYTECODE_BEGIN("SyncToLiteralAndContinueInstT<EquivTrivialLastPatCharScannerMixin> aka SyncToLiteralEquivTrivialLastPatCharAndContinue");
- PRINT_MIXIN(EquivTrivialLastPatCharScannerMixin); // NOTE: would work with template <typename ScannerT> ScannerT::Print
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(EquivTrivialLastPatCharScannerMixin); // NOTE: unique because macro expansion and _u(#InstType) happen before template is evaluated (so text would be ScannerT)
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // SyncToCharAndConsumeInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool SyncToCharAndConsumeInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- const Char matchC = c;
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- while (inputOffset < inputLength && input[inputOffset] != matchC)
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- inputOffset++;
- }
- if (inputOffset >= inputLength)
- {
- return matcher.HardFail(HARDFAIL_PARAMETERS(HardFailMode::ImmediateFail));
- }
- matchStart = inputOffset++;
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int SyncToCharAndConsumeInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("SyncToCharAndConsume");
- PRINT_MIXIN(CharMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(CharMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // SyncToChar2SetAndConsumeInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool SyncToChar2SetAndConsumeInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- const Char matchC0 = cs[0];
- const Char matchC1 = cs[1];
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- while (inputOffset < inputLength && (input[inputOffset] != matchC0 && input[inputOffset] != matchC1))
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- inputOffset++;
- }
- if (inputOffset >= inputLength)
- {
- return matcher.HardFail(HARDFAIL_PARAMETERS(HardFailMode::ImmediateFail));
- }
- matchStart = inputOffset++;
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int SyncToChar2SetAndConsumeInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("SyncToChar2SetAndConsume");
- PRINT_MIXIN(Char2Mixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(Char2Mixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // SyncToSetAndConsumeInst (optimized instruction)
- // ----------------------------------------------------------------------
- template<bool IsNegation>
- inline bool SyncToSetAndConsumeInst<IsNegation>::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- const RuntimeCharSet<Char>& matchSet = this->set;
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- while (inputOffset < inputLength && matchSet.Get(input[inputOffset]) == IsNegation)
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- inputOffset++;
- }
- if (inputOffset >= inputLength)
- {
- return matcher.HardFail(HARDFAIL_PARAMETERS(HardFailMode::ImmediateFail));
- }
- matchStart = inputOffset++;
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- template<bool IsNegation>
- int SyncToSetAndConsumeInst<IsNegation>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- if (IsNegation)
- {
- PRINT_RE_BYTECODE_BEGIN("SyncToNegatedSetAndConsume");
- PRINT_MIXIN(SetMixin<true>);
- }
- else
- {
- PRINT_RE_BYTECODE_BEGIN("SyncToSetAndConsume");
- PRINT_MIXIN(SetMixin<false>);
- }
- PRINT_RE_BYTECODE_MID();
- IsNegation ? PRINT_BYTES(SetMixin<true>) : PRINT_BYTES(SetMixin<false>);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // SyncToLiteralAndConsumeInst (optimized instruction)
- // ----------------------------------------------------------------------
- template <typename ScannerT>
- inline bool SyncToLiteralAndConsumeInstT<ScannerT>::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- if (!this->Match(matcher, input, inputLength, inputOffset))
- {
- return matcher.HardFail(HARDFAIL_PARAMETERS(HardFailMode::ImmediateFail));
- }
- matchStart = inputOffset;
- inputOffset += ScannerT::GetLiteralLength();
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- // explicit instantiation
- template struct SyncToLiteralAndConsumeInstT<Char2LiteralScannerMixin>;
- template struct SyncToLiteralAndConsumeInstT<ScannerMixin>;
- template struct SyncToLiteralAndConsumeInstT<ScannerMixin_WithLinearCharMap>;
- template struct SyncToLiteralAndConsumeInstT<EquivScannerMixin>;
- template struct SyncToLiteralAndConsumeInstT<EquivTrivialLastPatCharScannerMixin>;
- // Explicitly define each of these 5 Print functions so that the output will show the actual template param mixin and
- // actual opcode name, even though the logic is basically the same in each definition. See notes below.
- template <>
- int SyncToLiteralAndConsumeInstT<Char2LiteralScannerMixin>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- // NOTE: this text is unique to this instantiation
- PRINT_RE_BYTECODE_BEGIN("SyncToLiteralAndConsumeInstT<Char2LiteralScannerMixin> aka SyncToChar2LiteralAndConsume");
- PRINT_MIXIN(Char2LiteralScannerMixin); // NOTE: would work with template <typename ScannerT> ScannerT::Print
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(Char2LiteralScannerMixin); // NOTE: unique because macro expansion and _u(#InstType) happen before template is evaluated (so text would be ScannerT)
- PRINT_RE_BYTECODE_END();
- }
- template <>
- int SyncToLiteralAndConsumeInstT<ScannerMixin>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- // NOTE: this text is unique to this instantiation
- PRINT_RE_BYTECODE_BEGIN("SyncToLiteralAndConsumeInstT<ScannerMixin> aka SyncToLiteralAndConsume");
- PRINT_MIXIN(ScannerMixin); // NOTE: would work with template <typename ScannerT> ScannerT::Print
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(ScannerMixin); // NOTE: unique because macro expansion and _u(#InstType) happen before template is evaluated (so text would be ScannerT)
- PRINT_RE_BYTECODE_END();
- }
- template <>
- int SyncToLiteralAndConsumeInstT<ScannerMixin_WithLinearCharMap>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- // NOTE: this text is unique to this instantiation
- PRINT_RE_BYTECODE_BEGIN("SyncToLiteralAndConsumeInstT<ScannerMixin_WithLinearCharMap> aka SyncToLinearLiteralAndConsume");
- PRINT_MIXIN(ScannerMixin_WithLinearCharMap); // NOTE: would work with template <typename ScannerT> ScannerT::Print
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(ScannerMixin_WithLinearCharMap); // NOTE: unique because macro expansion and _u(#InstType) happen before template is evaluated (so text would be ScannerT)
- PRINT_RE_BYTECODE_END();
- }
- template <>
- int SyncToLiteralAndConsumeInstT<EquivScannerMixin>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- // NOTE: this text is unique to this instantiation
- PRINT_RE_BYTECODE_BEGIN("SyncToLiteralAndConsumeInstT<EquivScannerMixin> aka SyncToLiteralEquivAndConsume");
- PRINT_MIXIN(EquivScannerMixin); // NOTE: would work with template <typename ScannerT> ScannerT::Print
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(EquivScannerMixin); // NOTE: unique because macro expansion and _u(#InstType) happen before template is evaluated (so text would be ScannerT)
- PRINT_RE_BYTECODE_END();
- }
- template <>
- int SyncToLiteralAndConsumeInstT<EquivTrivialLastPatCharScannerMixin>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- // NOTE: this text is unique to this instantiation
- PRINT_RE_BYTECODE_BEGIN("SyncToLiteralAndConsumeInstT<EquivTrivialLastPatCharScannerMixin> aka SyncToLiteralEquivTrivialLastPatCharAndConsume");
- PRINT_MIXIN(EquivTrivialLastPatCharScannerMixin); // NOTE: would work with template <typename ScannerT> ScannerT::Print
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(EquivTrivialLastPatCharScannerMixin); // NOTE: unique because macro expansion and _u(#InstType) happen before template is evaluated (so text would be ScannerT)
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // SyncToCharAndBackupInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool SyncToCharAndBackupInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- if (backup.lower > inputLength - matchStart)
- {
- // Even match at very end doesn't allow for minimum backup
- return matcher.HardFail(HARDFAIL_PARAMETERS(HardFailMode::ImmediateFail));
- }
- if (inputOffset < nextSyncInputOffset)
- {
- // We have not yet reached the offset in the input we last synced to before backing up, so it's unnecessary to sync
- // again since we'll sync to the same point in the input and back up to the same place we are at now
- instPointer += sizeof(*this);
- return false;
- }
- if (backup.lower > inputOffset - matchStart)
- {
- // No use looking for match until minimum backup is possible
- inputOffset = matchStart + backup.lower;
- }
- const Char matchC = c;
- while (inputOffset < inputLength && input[inputOffset] != matchC)
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- inputOffset++;
- }
- if (inputOffset >= inputLength)
- {
- return matcher.HardFail(HARDFAIL_PARAMETERS(HardFailMode::ImmediateFail));
- }
- nextSyncInputOffset = inputOffset + 1;
- if (backup.upper != CharCountFlag)
- {
- // Backup at most by backup.upper for new start
- CharCount maxBackup = inputOffset - matchStart;
- matchStart = inputOffset - min(maxBackup, (CharCount)backup.upper);
- }
- // else: leave start where it is
- // Move input to new match start
- inputOffset = matchStart;
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int SyncToCharAndBackupInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("SyncToCharAndBackup");
- PRINT_MIXIN_COMMA(CharMixin);
- PRINT_MIXIN(BackupMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(CharMixin);
- PRINT_BYTES(BackupMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // SyncToSetAndBackupInst (optimized instruction)
- // ----------------------------------------------------------------------
- template<bool IsNegation>
- inline bool SyncToSetAndBackupInst<IsNegation>::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- if (backup.lower > inputLength - matchStart)
- {
- // Even match at very end doesn't allow for minimum backup
- return matcher.HardFail(HARDFAIL_PARAMETERS(HardFailMode::ImmediateFail));
- }
- if (inputOffset < nextSyncInputOffset)
- {
- // We have not yet reached the offset in the input we last synced to before backing up, so it's unnecessary to sync
- // again since we'll sync to the same point in the input and back up to the same place we are at now
- instPointer += sizeof(*this);
- return false;
- }
- if (backup.lower > inputOffset - matchStart)
- {
- // No use looking for match until minimum backup is possible
- inputOffset = matchStart + backup.lower;
- }
- const RuntimeCharSet<Char>& matchSet = this->set;
- while (inputOffset < inputLength && matchSet.Get(input[inputOffset]) == IsNegation)
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- inputOffset++;
- }
- if (inputOffset >= inputLength)
- {
- return matcher.HardFail(HARDFAIL_PARAMETERS(HardFailMode::ImmediateFail));
- }
- nextSyncInputOffset = inputOffset + 1;
- if (backup.upper != CharCountFlag)
- {
- // Backup at most by backup.upper for new start
- CharCount maxBackup = inputOffset - matchStart;
- matchStart = inputOffset - min(maxBackup, (CharCount)backup.upper);
- }
- // else: leave start where it is
- // Move input to new match start
- inputOffset = matchStart;
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- template<bool IsNegation>
- int SyncToSetAndBackupInst<IsNegation>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- if (IsNegation)
- {
- PRINT_RE_BYTECODE_BEGIN("SyncToNegatedSetAndBackup");
- PRINT_MIXIN_COMMA(SetMixin<true>);
- }
- else
- {
- PRINT_RE_BYTECODE_BEGIN("SyncToSetAndBackup");
- PRINT_MIXIN_COMMA(SetMixin<false>);
- }
- PRINT_MIXIN(BackupMixin);
- PRINT_RE_BYTECODE_MID();
- IsNegation ? PRINT_BYTES(SetMixin<true>) : PRINT_BYTES(SetMixin<false>);
- PRINT_BYTES(BackupMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // SyncToLiteralAndBackupInst (optimized instruction)
- // ----------------------------------------------------------------------
- template <typename ScannerT>
- inline bool SyncToLiteralAndBackupInstT<ScannerT>::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- if (backup.lower > inputLength - matchStart)
- {
- // Even match at very end doesn't allow for minimum backup
- return matcher.HardFail(HARDFAIL_PARAMETERS(HardFailMode::ImmediateFail));
- }
- if(inputOffset < nextSyncInputOffset)
- {
- // We have not yet reached the offset in the input we last synced to before backing up, so it's unnecessary to sync
- // again since we'll sync to the same point in the input and back up to the same place we are at now
- instPointer += sizeof(*this);
- return false;
- }
- if (backup.lower > inputOffset - matchStart)
- {
- // No use looking for match until minimum backup is possible
- inputOffset = matchStart + backup.lower;
- }
- if (!this->Match(matcher, input, inputLength, inputOffset))
- {
- return matcher.HardFail(HARDFAIL_PARAMETERS(HardFailMode::ImmediateFail));
- }
- nextSyncInputOffset = inputOffset + 1;
- if (backup.upper != CharCountFlag)
- {
- // Set new start at most backup.upper from start of literal
- CharCount maxBackup = inputOffset - matchStart;
- matchStart = inputOffset - min(maxBackup, (CharCount)backup.upper);
- }
- // else: leave start where it is
- // Move input to new match start
- inputOffset = matchStart;
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- // explicit instantiation
- template struct SyncToLiteralAndBackupInstT<Char2LiteralScannerMixin>;
- template struct SyncToLiteralAndBackupInstT<ScannerMixin>;
- template struct SyncToLiteralAndBackupInstT<ScannerMixin_WithLinearCharMap>;
- template struct SyncToLiteralAndBackupInstT<EquivScannerMixin>;
- template struct SyncToLiteralAndBackupInstT<EquivTrivialLastPatCharScannerMixin>;
- // Explicitly define each of these 5 Print functions so that the output will show the actual template param mixin and
- // actual opcode name, even though the logic is basically the same in each definition. See notes below.
- template <>
- int SyncToLiteralAndBackupInstT<Char2LiteralScannerMixin>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- // NOTE: this text is unique to this instantiation
- PRINT_RE_BYTECODE_BEGIN("SyncToLiteralAndBackupInstT<Char2LiteralScannerMixin> aka SyncToChar2LiteralAndBackup");
- PRINT_MIXIN_COMMA(Char2LiteralScannerMixin); // NOTE: would work with template <typename ScannerT> ScannerT::Print
- PRINT_MIXIN(BackupMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(Char2LiteralScannerMixin); // NOTE: unique because macro expansion and _u(#InstType) happen before template is evaluated (so text would be ScannerT)
- PRINT_BYTES(BackupMixin);
- PRINT_RE_BYTECODE_END();
- }
- template <>
- int SyncToLiteralAndBackupInstT<ScannerMixin>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- // NOTE: this text is unique to this instantiation
- PRINT_RE_BYTECODE_BEGIN("SyncToLiteralAndBackupInstT<ScannerMixin> aka SyncToLiteralAndBackup");
- PRINT_MIXIN_COMMA(ScannerMixin); // NOTE: would work with template <typename ScannerT> ScannerT::Print
- PRINT_MIXIN(BackupMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(ScannerMixin); // NOTE: unique because macro expansion and _u(#InstType) happen before template is evaluated (so text would be ScannerT)
- PRINT_BYTES(BackupMixin);
- PRINT_RE_BYTECODE_END();
- }
- template <>
- int SyncToLiteralAndBackupInstT<ScannerMixin_WithLinearCharMap>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- // NOTE: this text is unique to this instantiation
- PRINT_RE_BYTECODE_BEGIN("SyncToLiteralAndBackupInstT<ScannerMixin_WithLinearCharMap> aka SyncToLinearLiteralAndBackup");
- PRINT_MIXIN_COMMA(ScannerMixin_WithLinearCharMap); // NOTE: would work with template <typename ScannerT> ScannerT::Print
- PRINT_MIXIN(BackupMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(ScannerMixin_WithLinearCharMap); // NOTE: unique because macro expansion and _u(#InstType) happen before template is evaluated (so text would be ScannerT)
- PRINT_BYTES(BackupMixin);
- PRINT_RE_BYTECODE_END();
- }
- template <>
- int SyncToLiteralAndBackupInstT<EquivScannerMixin>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- // NOTE: this text is unique to this instantiation
- PRINT_RE_BYTECODE_BEGIN("SyncToLiteralAndBackupInstT<EquivScannerMixin> aka SyncToLiteralEquivAndBackup");
- PRINT_MIXIN_COMMA(EquivScannerMixin); // NOTE: would work with template <typename ScannerT> ScannerT::Print
- PRINT_MIXIN(BackupMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(EquivScannerMixin); // NOTE: unique because macro expansion and _u(#InstType) happen before template is evaluated (so text would be ScannerT)
- PRINT_BYTES(BackupMixin);
- PRINT_RE_BYTECODE_END();
- }
- template <>
- int SyncToLiteralAndBackupInstT<EquivTrivialLastPatCharScannerMixin>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- // NOTE: this text is unique to this instantiation
- PRINT_RE_BYTECODE_BEGIN("SyncToLiteralAndBackupInstT<EquivTrivialLastPatCharScannerMixin> aka SyncToLiteralEquivTrivialLastPatCharAndBackup");
- PRINT_MIXIN_COMMA(EquivTrivialLastPatCharScannerMixin); // NOTE: would work with template <typename ScannerT> ScannerT::Print
- PRINT_MIXIN(BackupMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(EquivTrivialLastPatCharScannerMixin); // NOTE: unique because macro expansion and _u(#InstType) happen before template is evaluated (so text would be ScannerT)
- PRINT_BYTES(BackupMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // SyncToLiteralsAndBackupInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool SyncToLiteralsAndBackupInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- if (backup.lower > inputLength - matchStart)
- {
- // Even match at very end doesn't allow for minimum backup
- return matcher.HardFail(HARDFAIL_PARAMETERS(HardFailMode::ImmediateFail));
- }
- if (inputOffset < nextSyncInputOffset)
- {
- // We have not yet reached the offset in the input we last synced to before backing up, so it's unnecessary to sync
- // again since we'll sync to the same point in the input and back up to the same place we are at now
- instPointer += sizeof(*this);
- return false;
- }
- if (backup.lower > inputOffset - matchStart)
- {
- // No use looking for match until minimum backup is possible
- inputOffset = matchStart + backup.lower;
- }
- int besti = -1;
- CharCount bestMatchOffset = 0;
- if (matcher.literalNextSyncInputOffsets == nullptr)
- {
- Assert(numLiterals <= MaxNumSyncLiterals);
- matcher.literalNextSyncInputOffsets =
- RecyclerNewArrayLeaf(matcher.recycler, CharCount, ScannersMixin::MaxNumSyncLiterals);
- }
- CharCount* literalNextSyncInputOffsets = matcher.literalNextSyncInputOffsets;
- if (firstIteration)
- {
- for (int i = 0; i < numLiterals; i++)
- {
- literalNextSyncInputOffsets[i] = inputOffset;
- }
- }
- for (int i = 0; i < numLiterals; i++)
- {
- CharCount thisMatchOffset = literalNextSyncInputOffsets[i];
- if (inputOffset > thisMatchOffset)
- {
- thisMatchOffset = inputOffset;
- }
- if (infos[i]->isEquivClass
- ? (infos[i]->scanner.Match<CaseInsensitive::EquivClassSize>(
- input
- , inputLength
- , thisMatchOffset
- , matcher.program->rep.insts.litbuf + infos[i]->offset
- , infos[i]->length
- #if ENABLE_REGEX_CONFIG_OPTIONS
- , matcher.stats
- #endif
- ))
- : (infos[i]->scanner.Match<1>(
- input
- , inputLength
- , thisMatchOffset
- , matcher.program->rep.insts.litbuf + infos[i]->offset
- , infos[i]->length
- #if ENABLE_REGEX_CONFIG_OPTIONS
- , matcher.stats
- #endif
- )))
- {
- if (besti < 0 || thisMatchOffset < bestMatchOffset)
- {
- besti = i;
- bestMatchOffset = thisMatchOffset;
- }
- literalNextSyncInputOffsets[i] = thisMatchOffset;
- }
- else
- {
- literalNextSyncInputOffsets[i] = inputLength;
- }
- }
- if (besti < 0)
- {
- // No literals matched
- return matcher.HardFail(HARDFAIL_PARAMETERS(HardFailMode::ImmediateFail));
- }
- nextSyncInputOffset = bestMatchOffset + 1;
- if (backup.upper != CharCountFlag)
- {
- // Set new start at most backup.upper from start of literal
- CharCount maxBackup = bestMatchOffset - matchStart;
- matchStart = bestMatchOffset - min(maxBackup, (CharCount)backup.upper);
- }
- // else: leave start where it is
- // Move input to new match start
- inputOffset = matchStart;
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int SyncToLiteralsAndBackupInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("SyncToLiteralsAndBackup");
- PRINT_MIXIN_COMMA(ScannersMixin);
- PRINT_MIXIN(BackupMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(ScannersMixin);
- PRINT_BYTES(BackupMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // MatchGroupInst
- // ----------------------------------------------------------------------
- inline bool MatchGroupInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- GroupInfo* const info = matcher.GroupIdToGroupInfo(groupId);
- if (!info->IsUndefined() && info->length > 0)
- {
- if (info->length > inputLength - inputOffset)
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- CharCount groupOffset = info->offset;
- const CharCount groupEndOffset = groupOffset + info->length;
- bool isCaseInsensitiveMatch = (matcher.program->flags & IgnoreCaseRegexFlag) != 0;
- bool isCodePointList = (matcher.program->flags & UnicodeRegexFlag) != 0;
- // This is the only place in the runtime machinery we need to convert characters to their equivalence class
- if (isCaseInsensitiveMatch && isCodePointList)
- {
- auto getNextCodePoint = [=](CharCount &offset, CharCount endOffset, codepoint_t &codePoint) {
- if (endOffset <= offset)
- {
- return false;
- }
- Char lowerPart = input[offset];
- if (!Js::NumberUtilities::IsSurrogateLowerPart(lowerPart) || offset + 1 == endOffset)
- {
- codePoint = lowerPart;
- offset += 1;
- return true;
- }
- Char upperPart = input[offset + 1];
- if (!Js::NumberUtilities::IsSurrogateUpperPart(upperPart))
- {
- codePoint = lowerPart;
- offset += 1;
- }
- else
- {
- codePoint = Js::NumberUtilities::SurrogatePairAsCodePoint(lowerPart, upperPart);
- offset += 2;
- }
- return true;
- };
- codepoint_t equivs[CaseInsensitive::EquivClassSize];
- while (true)
- {
- codepoint_t groupCodePoint;
- bool hasGroupCodePoint = getNextCodePoint(groupOffset, groupEndOffset, groupCodePoint);
- if (!hasGroupCodePoint)
- {
- break;
- }
- // We don't need to verify that there is a valid input code point since at the beginning
- // of the function, we make sure that the length of the input is at least as long as the
- // length of the group.
- codepoint_t inputCodePoint;
- getNextCodePoint(inputOffset, inputLength, inputCodePoint);
- bool doesMatch = false;
- if (!Js::NumberUtilities::IsInSupplementaryPlane(groupCodePoint))
- {
- auto toCanonical = [&](codepoint_t c) {
- return matcher.standardChars->ToCanonical(
- CaseInsensitive::MappingSource::CaseFolding,
- static_cast<char16>(c));
- };
- doesMatch = (toCanonical(groupCodePoint) == toCanonical(inputCodePoint));
- }
- else
- {
- uint tblidx = 0;
- uint acth = 0;
- CaseInsensitive::RangeToEquivClass(tblidx, groupCodePoint, groupCodePoint, acth, equivs);
- CompileAssert(CaseInsensitive::EquivClassSize == 4);
- doesMatch =
- inputCodePoint == equivs[0]
- || inputCodePoint == equivs[1]
- || inputCodePoint == equivs[2]
- || inputCodePoint == equivs[3];
- }
- if (!doesMatch)
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- }
- }
- else if (isCaseInsensitiveMatch)
- {
- do
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- auto toCanonical = [&](CharCount &offset) {
- return matcher.standardChars->ToCanonical(CaseInsensitive::MappingSource::UnicodeData, input[offset++]);
- };
- if (toCanonical(groupOffset) != toCanonical(inputOffset))
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- }
- while (groupOffset < groupEndOffset);
- }
- else
- {
- do
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (input[groupOffset++] != input[inputOffset++])
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- }
- while (groupOffset < groupEndOffset);
- }
- }
- // else: trivially match empty string
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int MatchGroupInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("MatchGroup");
- PRINT_MIXIN(GroupMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(GroupMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // BeginDefineGroupInst
- // ----------------------------------------------------------------------
- inline bool BeginDefineGroupInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- GroupInfo *const groupInfo = matcher.GroupIdToGroupInfo(groupId);
- Assert(groupInfo->IsUndefined());
- groupInfo->offset = inputOffset;
- Assert(groupInfo->IsUndefined());
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int BeginDefineGroupInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("BeginDefineGroup");
- PRINT_MIXIN(GroupMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(GroupMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // EndDefineGroupInst
- // ----------------------------------------------------------------------
- inline bool EndDefineGroupInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- if (!noNeedToSave)
- {
- // UNDO ACTION: Restore group on backtrack
- PUSH(contStack, ResetGroupCont, groupId);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- GroupInfo *const groupInfo = matcher.GroupIdToGroupInfo(groupId);
- Assert(groupInfo->IsUndefined());
- Assert(inputOffset >= groupInfo->offset);
- groupInfo->length = inputOffset - groupInfo->offset;
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int EndDefineGroupInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("EndDefineGroup");
- PRINT_MIXIN_COMMA(GroupMixin);
- PRINT_MIXIN(NoNeedToSaveMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(GroupMixin);
- PRINT_BYTES(NoNeedToSaveMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // DefineGroupFixedInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool DefineGroupFixedInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- if (!noNeedToSave)
- {
- // UNDO ACTION: Restore group on backtrack
- PUSH(contStack, ResetGroupCont, groupId);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- GroupInfo *const groupInfo = matcher.GroupIdToGroupInfo(groupId);
- Assert(groupInfo->IsUndefined());
- groupInfo->offset = inputOffset - length;
- groupInfo->length = length;
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int DefineGroupFixedInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("DefineGroupFixed");
- PRINT_MIXIN_COMMA(GroupMixin);
- PRINT_MIXIN_COMMA(FixedLengthMixin);
- PRINT_MIXIN(NoNeedToSaveMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(GroupMixin);
- PRINT_BYTES(FixedLengthMixin);
- PRINT_BYTES(NoNeedToSaveMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // BeginLoopInst
- // ----------------------------------------------------------------------
- inline bool BeginLoopInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- LoopInfo* loopInfo = matcher.LoopIdToLoopInfo(loopId);
- // If loop has outer loops, the continuation stack may have choicepoints from an earlier "run" of this loop
- // which, when backtracked to, may expect the loopInfo state to be as it was at the time the choicepoint was
- // pushed.
- // - If the loop is greedy with deterministic body, there may be Resumes into the follow of the loop, but
- // they won't look at the loopInfo state so there's nothing to do.
- // - If the loop is greedy, or if it is non-greedy with lower > 0, AND it has a non-deterministic body,
- // we may have Resume entries which will resume inside the loop body, which may then run to a
- // RepeatLoop, which will then look at the loopInfo state. However, each iteration is protected by
- // a RestoreLoop by RepeatLoopInst below. (****)
- // - If the loop is non-greedy there may be a RepeatLoop on the stack, so we must restore the loopInfo
- // state before backtracking to it.
- if (!isGreedy && hasOuterLoops)
- {
- PUSH(contStack, RestoreLoopCont, loopId, *loopInfo, matcher);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- // The loop body must always begin with empty inner groups
- // - if the loop is not in an outer they will be empty due to the reset when the match began
- // - if the loop is in an outer loop, they will have been reset by the outer loop's RepeatLoop instruction
- #if DBG
- for (int i = minBodyGroupId; i <= maxBodyGroupId; i++)
- {
- Assert(matcher.GroupIdToGroupInfo(i)->IsUndefined());
- }
- #endif
- loopInfo->number = 0;
- loopInfo->startInputOffset = inputOffset;
- if (repeats.lower == 0)
- {
- if (isGreedy)
- {
- // CHOICEPOINT: Try one iteration of body, if backtrack continue from here with no iterations
- PUSH(contStack, ResumeCont, inputOffset, exitLabel);
- instPointer += sizeof(*this);
- }
- else
- {
- // CHOICEPOINT: Try no iterations of body, if backtrack do one iteration of body from here
- Assert(instPointer == (uint8*)this);
- PUSH(contStack, RepeatLoopCont, matcher.InstPointerToLabel(instPointer), inputOffset);
- instPointer = matcher.LabelToInstPointer(exitLabel);
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- else
- {
- // Must match minimum iterations, so continue to loop body
- instPointer += sizeof(*this);
- }
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int BeginLoopInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("BeginLoop");
- PRINT_MIXIN_COMMA(BeginLoopMixin);
- PRINT_MIXIN_COMMA(BodyGroupsMixin);
- PRINT_MIXIN(GreedyMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(BeginLoopMixin);
- PRINT_BYTES(BodyGroupsMixin);
- PRINT_BYTES(GreedyMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // RepeatLoopInst
- // ----------------------------------------------------------------------
- inline bool RepeatLoopInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- BeginLoopInst* begin = matcher.L2I(BeginLoop, beginLabel);
- LoopInfo* loopInfo = matcher.LoopIdToLoopInfo(begin->loopId);
- // See comment (****) above.
- if (begin->hasInnerNondet)
- {
- PUSH(contStack, RestoreLoopCont, begin->loopId, *loopInfo, matcher);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- loopInfo->number++;
- if (loopInfo->number < begin->repeats.lower)
- {
- // Must match another iteration of body.
- loopInfo->startInputOffset = inputOffset;
- if(begin->hasInnerNondet)
- {
- // If it backtracks into the loop body of an earlier iteration, it must restore inner groups for that iteration.
- // Save the inner groups and reset them for the next iteration.
- matcher.SaveInnerGroups(begin->minBodyGroupId, begin->maxBodyGroupId, true, input, contStack);
- }
- else
- {
- // If it backtracks, the entire loop will fail, so no need to restore groups. Just reset the inner groups for
- // the next iteration.
- matcher.ResetInnerGroups(begin->minBodyGroupId, begin->maxBodyGroupId);
- }
- instPointer = matcher.LabelToInstPointer(beginLabel + sizeof(BeginLoopInst));
- }
- else if (inputOffset == loopInfo->startInputOffset && loopInfo->number > begin->repeats.lower)
- {
- // The minimum number of iterations has been satisfied but the last iteration made no progress.
- // - With greedy & deterministic body, FAIL so as to undo that iteration and restore group bindings.
- // - With greedy & non-deterministic body, FAIL so as to try another body alternative
- // - With non-greedy, we're trying an additional iteration because the follow failed. But
- // since we didn't consume anything the follow will fail again, so fail
- //
- return matcher.Fail(FAIL_PARAMETERS);
- }
- else if (begin->repeats.upper != CharCountFlag && loopInfo->number >= (CharCount)begin->repeats.upper)
- {
- // Success: proceed to remainder.
- instPointer = matcher.LabelToInstPointer(begin->exitLabel);
- }
- else if (begin->isGreedy)
- {
- // CHOICEPOINT: Try one more iteration of body, if backtrack continue from here with no more iterations
- PUSH(contStack, ResumeCont, inputOffset, begin->exitLabel);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- loopInfo->startInputOffset = inputOffset;
- // If backtrack, we must continue with previous group bindings
- matcher.SaveInnerGroups(begin->minBodyGroupId, begin->maxBodyGroupId, true, input, contStack);
- instPointer = matcher.LabelToInstPointer(beginLabel + sizeof(BeginLoopInst));
- }
- else
- {
- // CHOICEPOINT: Try no more iterations of body, if backtrack do one more iteration of body from here
- PUSH(contStack, RepeatLoopCont, beginLabel, inputOffset);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- instPointer = matcher.LabelToInstPointer(begin->exitLabel);
- }
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int RepeatLoopInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("RepeatLoop");
- PRINT_MIXIN(RepeatLoopMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(RepeatLoopMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // BeginLoopIfCharInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool BeginLoopIfCharInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset < inputLength && input[inputOffset] == c)
- {
- // Commit to at least one iteration of loop
- LoopInfo* loopInfo = matcher.LoopIdToLoopInfo(loopId);
- // All inner groups must begin reset
- #if DBG
- for (int i = minBodyGroupId; i <= maxBodyGroupId; i++)
- {
- Assert(matcher.GroupIdToGroupInfo(i)->IsUndefined());
- }
- #endif
- loopInfo->number = 0;
- instPointer += sizeof(*this);
- return false;
- }
- if (repeats.lower > 0)
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- instPointer = matcher.LabelToInstPointer(exitLabel);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int BeginLoopIfCharInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("BeginLoopIfChar");
- PRINT_MIXIN_COMMA(CharMixin);
- PRINT_MIXIN_COMMA(BeginLoopMixin);
- PRINT_MIXIN(BodyGroupsMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(CharMixin);
- PRINT_BYTES(BeginLoopMixin);
- PRINT_BYTES(BodyGroupsMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // BeginLoopIfSetInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool BeginLoopIfSetInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset < inputLength && set.Get(input[inputOffset]))
- {
- // Commit to at least one iteration of loop
- LoopInfo* loopInfo = matcher.LoopIdToLoopInfo(loopId);
- // All inner groups must be begin reset
- #if DBG
- for (int i = minBodyGroupId; i <= maxBodyGroupId; i++)
- {
- Assert(matcher.GroupIdToGroupInfo(i)->IsUndefined());
- }
- #endif
- loopInfo->startInputOffset = inputOffset;
- loopInfo->number = 0;
- instPointer += sizeof(*this);
- return false;
- }
- if (repeats.lower > 0)
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- instPointer = matcher.LabelToInstPointer(exitLabel);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int BeginLoopIfSetInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("BeginLoopIfSet");
- PRINT_MIXIN_COMMA(SetMixin<false>);
- PRINT_MIXIN_COMMA(BeginLoopMixin);
- PRINT_MIXIN(BodyGroupsMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(SetMixin<false>);
- PRINT_BYTES(BeginLoopMixin);
- PRINT_BYTES(BodyGroupsMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // RepeatLoopIfCharInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool RepeatLoopIfCharInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- BeginLoopIfCharInst* begin = matcher.L2I(BeginLoopIfChar, beginLabel);
- LoopInfo* loopInfo = matcher.LoopIdToLoopInfo(begin->loopId);
- if (begin->hasInnerNondet)
- {
- // May end up backtracking into loop body for iteration just completed: see above.
- PUSH(contStack, RestoreLoopCont, begin->loopId, *loopInfo, matcher);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- loopInfo->number++;
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset < inputLength && input[inputOffset] == begin->c)
- {
- if (begin->repeats.upper != CharCountFlag && loopInfo->number >= (CharCount)begin->repeats.upper)
- {
- // If the loop body's first set and the loop's follow set are disjoint, we can just fail here since
- // we know the next character in the input is in the loop body's first set.
- return matcher.Fail(FAIL_PARAMETERS);
- }
- // Commit to one more iteration
- if(begin->hasInnerNondet)
- {
- // If it backtracks into the loop body of an earlier iteration, it must restore inner groups for that iteration.
- // Save the inner groups and reset them for the next iteration.
- matcher.SaveInnerGroups(begin->minBodyGroupId, begin->maxBodyGroupId, true, input, contStack);
- }
- else
- {
- // If it backtracks, the entire loop will fail, so no need to restore groups. Just reset the inner groups for
- // the next iteration.
- matcher.ResetInnerGroups(begin->minBodyGroupId, begin->maxBodyGroupId);
- }
- instPointer = matcher.LabelToInstPointer(beginLabel + sizeof(BeginLoopIfCharInst));
- return false;
- }
- if (loopInfo->number < begin->repeats.lower)
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- // Proceed to exit
- instPointer = matcher.LabelToInstPointer(begin->exitLabel);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int RepeatLoopIfCharInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("RepeatLoopIfChar");
- PRINT_MIXIN(RepeatLoopMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(RepeatLoopMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // RepeatLoopIfSetInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool RepeatLoopIfSetInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- BeginLoopIfSetInst* begin = matcher.L2I(BeginLoopIfSet, beginLabel);
- LoopInfo* loopInfo = matcher.LoopIdToLoopInfo(begin->loopId);
- if (begin->hasInnerNondet)
- {
- // May end up backtracking into loop body for iteration just completed: see above.
- PUSH(contStack, RestoreLoopCont, begin->loopId, *loopInfo, matcher);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- loopInfo->number++;
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset < inputLength && begin->set.Get(input[inputOffset]))
- {
- if (begin->repeats.upper != CharCountFlag && loopInfo->number >= (CharCount)begin->repeats.upper)
- {
- // If the loop body's first set and the loop's follow set are disjoint, we can just fail here since
- // we know the next character in the input is in the loop body's first set.
- return matcher.Fail(FAIL_PARAMETERS);
- }
- // Commit to one more iteration
- if (begin->hasInnerNondet)
- {
- // If it backtracks into the loop body of an earlier iteration, it must restore inner groups for that iteration.
- // Save the inner groups and reset them for the next iteration.
- matcher.SaveInnerGroups(begin->minBodyGroupId, begin->maxBodyGroupId, true, input, contStack);
- }
- else
- {
- // If it backtracks, the entire loop will fail, so no need to restore groups. Just reset the inner groups for
- // the next iteration.
- matcher.ResetInnerGroups(begin->minBodyGroupId, begin->maxBodyGroupId);
- }
- instPointer = matcher.LabelToInstPointer(beginLabel + sizeof(BeginLoopIfSetInst));
- return false;
- }
- if (loopInfo->number < begin->repeats.lower)
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- // Proceed to exit
- instPointer = matcher.LabelToInstPointer(begin->exitLabel);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int RepeatLoopIfSetInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("RepeatLoopIfSet");
- PRINT_MIXIN(RepeatLoopMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(RepeatLoopMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // BeginLoopFixedInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool BeginLoopFixedInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- LoopInfo* loopInfo = matcher.LoopIdToLoopInfo(loopId);
- // If loop is contained in an outer loop, continuation stack may already have a RewindLoopFixed entry for
- // this loop. We must make sure it's state is preserved on backtrack.
- if (hasOuterLoops)
- {
- PUSH(contStack, RestoreLoopCont, loopId, *loopInfo, matcher);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- // startInputOffset will stay here for all iterations, and we'll use number of length to figure out
- // where in the input to rewind to
- loopInfo->number = 0;
- loopInfo->startInputOffset = inputOffset;
- if (repeats.lower == 0)
- {
- // CHOICEPOINT: Try one iteration of body. Failure of body will rewind input to here and resume with follow.
- Assert(instPointer == (uint8*)this);
- PUSH(contStack, RewindLoopFixedCont, matcher.InstPointerToLabel(instPointer), true);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- // else: Must match minimum iterations, so continue to loop body. Failure of body signals failure of entire loop.
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int BeginLoopFixedInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("BeginLoopFixed");
- PRINT_MIXIN_COMMA(BeginLoopMixin);
- PRINT_MIXIN(FixedLengthMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(BeginLoopMixin);
- PRINT_BYTES(FixedLengthMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // RepeatLoopFixedInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool RepeatLoopFixedInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- BeginLoopFixedInst* begin = matcher.L2I(BeginLoopFixed, beginLabel);
- LoopInfo* loopInfo = matcher.LoopIdToLoopInfo(begin->loopId);
- loopInfo->number++;
- if (loopInfo->number < begin->repeats.lower)
- {
- // Must match another iteration of body. Failure of body signals failure of the entire loop.
- instPointer = matcher.LabelToInstPointer(beginLabel + sizeof(BeginLoopFixedInst));
- }
- else if (begin->repeats.upper != CharCountFlag && loopInfo->number >= (CharCount)begin->repeats.upper)
- {
- // Matched maximum number of iterations. Continue with follow.
- if (begin->repeats.lower < begin->repeats.upper)
- {
- // Failure of follow will try one fewer iterations (subject to repeats.lower).
- // Since loop body is non-deterministic and does not define groups the rewind continuation must be on top of the stack.
- Cont *top = contStack.Top();
- Assert(top != 0);
- Assert(top->tag == Cont::ContTag::RewindLoopFixed);
- RewindLoopFixedCont* rewind = (RewindLoopFixedCont*)top;
- rewind->tryingBody = false;
- }
- // else: we never pushed a rewind continuation
- instPointer = matcher.LabelToInstPointer(begin->exitLabel);
- }
- else
- {
- // CHOICEPOINT: Try one more iteration of body. Failure of body will rewind input to here and
- // try follow.
- if (loopInfo->number == begin->repeats.lower)
- {
- // i.e. begin->repeats.lower > 0, so continuation won't have been pushed in BeginLoopFixed
- PUSH(contStack, RewindLoopFixedCont, beginLabel, true);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- instPointer = matcher.LabelToInstPointer(beginLabel + sizeof(BeginLoopFixedInst));
- }
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int RepeatLoopFixedInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("RepeatLoopFixed");
- PRINT_MIXIN(RepeatLoopMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(RepeatLoopMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // LoopSetInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool LoopSetInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- LoopInfo* loopInfo = matcher.LoopIdToLoopInfo(loopId);
- // If loop is contained in an outer loop, continuation stack may already have a RewindLoopFixed entry for
- // this loop. We must make sure it's state is preserved on backtrack.
- if (hasOuterLoops)
- {
- PUSH(contStack, RestoreLoopCont, loopId, *loopInfo, matcher);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- // startInputOffset will stay here for all iterations, and we'll use number of length to figure out
- // where in the input to rewind to
- loopInfo->startInputOffset = inputOffset;
- // Consume as many elements of set as possible
- const RuntimeCharSet<Char>& matchSet = this->set;
- const CharCount loopMatchStart = inputOffset;
- const CharCountOrFlag repeatsUpper = repeats.upper;
- const CharCount inputEndOffset =
- static_cast<CharCount>(repeatsUpper) >= inputLength - inputOffset
- ? inputLength
- : inputOffset + static_cast<CharCount>(repeatsUpper);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- while (inputOffset < inputEndOffset && matchSet.Get(input[inputOffset]))
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- inputOffset++;
- }
- loopInfo->number = inputOffset - loopMatchStart;
- if (loopInfo->number < repeats.lower)
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- else if (loopInfo->number > repeats.lower)
- {
- // CHOICEPOINT: If follow fails, try consuming one fewer characters
- Assert(instPointer == (uint8*)this);
- PUSH(contStack, RewindLoopSetCont, matcher.InstPointerToLabel(instPointer));
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- // else: failure of follow signals failure of entire loop
- // Continue with follow
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int LoopSetInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("LoopSetInst");
- PRINT_MIXIN_COMMA(SetMixin<false>);
- PRINT_MIXIN(BeginLoopBasicsMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(SetMixin<false>);
- PRINT_BYTES(BeginLoopBasicsMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- inline bool LoopSetWithFollowFirstInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- LoopInfo* loopInfo = matcher.LoopIdToLoopInfo(loopId);
- // If loop is contained in an outer loop, continuation stack may already have a RewindLoopFixed entry for
- // this loop. We must make sure it's state is preserved on backtrack.
- if (hasOuterLoops)
- {
- PUSH(contStack, RestoreLoopCont, loopId, *loopInfo, matcher);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- if (loopInfo->offsetsOfFollowFirst)
- {
- loopInfo->offsetsOfFollowFirst->Clear();
- }
- // startInputOffset will stay here for all iterations, and we'll use number of length to figure out
- // where in the input to rewind to
- loopInfo->startInputOffset = inputOffset;
- // Consume as many elements of set as possible
- const RuntimeCharSet<Char>& matchSet = this->set;
- const CharCount loopMatchStart = inputOffset;
- const CharCountOrFlag repeatsUpper = repeats.upper;
- const CharCount inputEndOffset =
- static_cast<CharCount>(repeatsUpper) >= inputLength - inputOffset
- ? inputLength
- : inputOffset + static_cast<CharCount>(repeatsUpper);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- while (inputOffset < inputEndOffset && matchSet.Get(input[inputOffset]))
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (input[inputOffset] == this->followFirst)
- {
- loopInfo->EnsureOffsetsOfFollowFirst(matcher);
- loopInfo->offsetsOfFollowFirst->Add(inputOffset - loopInfo->startInputOffset);
- }
- inputOffset++;
- }
- loopInfo->number = inputOffset - loopMatchStart;
- if (loopInfo->number < repeats.lower)
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- else if (loopInfo->number > repeats.lower)
- {
- // CHOICEPOINT: If follow fails, try consuming one fewer characters
- Assert(instPointer == (uint8*)this);
- PUSH(contStack, RewindLoopSetWithFollowFirstCont, matcher.InstPointerToLabel(instPointer));
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- // else: failure of follow signals failure of entire loop
- // Continue with follow
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int LoopSetWithFollowFirstInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("LoopSetWithFollowFirstInst");
- PRINT_MIXIN_COMMA(SetMixin<false>);
- PRINT_MIXIN_COMMA(BeginLoopBasicsMixin);
- PRINT_MIXIN(FollowFirstMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(SetMixin<false>);
- PRINT_BYTES(BeginLoopBasicsMixin);
- PRINT_MIXIN(FollowFirstMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // BeginLoopFixedGroupLastIterationInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool BeginLoopFixedGroupLastIterationInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- Assert(matcher.GroupIdToGroupInfo(groupId)->IsUndefined());
- LoopInfo* loopInfo = matcher.LoopIdToLoopInfo(loopId);
- // If loop is contained in an outer loop, continuation stack may already have a RewindLoopFixedGroupLastIteration entry
- // for this loop. We must make sure it's state is preserved on backtrack.
- if (hasOuterLoops)
- {
- PUSH(contStack, RestoreLoopCont, loopId, *loopInfo, matcher);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- // If loop is contained in an outer loop or assertion, we must reset the group binding if we backtrack all the way out
- if (!noNeedToSave)
- {
- PUSH(contStack, ResetGroupCont, groupId);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- // startInputOffset will stay here for all iterations, and we'll use number of length to figure out
- // where in the input to rewind to
- loopInfo->number = 0;
- loopInfo->startInputOffset = inputOffset;
- if (repeats.lower == 0)
- {
- // CHOICEPOINT: Try one iteration of body. Failure of body will rewind input to here and resume with follow.
- Assert(instPointer == (uint8*)this);
- PUSH(contStack, RewindLoopFixedGroupLastIterationCont, matcher.InstPointerToLabel(instPointer), true);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- // else: Must match minimum iterations, so continue to loop body. Failure of body signals failure of entire loop.
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int BeginLoopFixedGroupLastIterationInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("BeginLoopFixedGroupLastIteration");
- PRINT_MIXIN_COMMA(BeginLoopMixin);
- PRINT_MIXIN_COMMA(FixedLengthMixin);
- PRINT_MIXIN_COMMA(GroupMixin);
- PRINT_MIXIN(NoNeedToSaveMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(BeginLoopMixin);
- PRINT_BYTES(FixedLengthMixin);
- PRINT_BYTES(GroupMixin);
- PRINT_BYTES(NoNeedToSaveMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // RepeatLoopFixedGroupLastIterationInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool RepeatLoopFixedGroupLastIterationInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- BeginLoopFixedGroupLastIterationInst* begin = matcher.L2I(BeginLoopFixedGroupLastIteration, beginLabel);
- LoopInfo* loopInfo = matcher.LoopIdToLoopInfo(begin->loopId);
- loopInfo->number++;
- if (loopInfo->number < begin->repeats.lower)
- {
- // Must match another iteration of body. Failure of body signals failure of the entire loop.
- instPointer = matcher.LabelToInstPointer(beginLabel + sizeof(BeginLoopFixedGroupLastIterationInst));
- }
- else if (begin->repeats.upper != CharCountFlag && loopInfo->number >= (CharCount)begin->repeats.upper)
- {
- // Matched maximum number of iterations. Continue with follow.
- if (begin->repeats.lower < begin->repeats.upper)
- {
- // Failure of follow will try one fewer iterations (subject to repeats.lower).
- // Since loop body is non-deterministic and does not define groups the rewind continuation must be on top of the stack.
- Cont *top = contStack.Top();
- Assert(top != 0);
- Assert(top->tag == Cont::ContTag::RewindLoopFixedGroupLastIteration);
- RewindLoopFixedGroupLastIterationCont* rewind = (RewindLoopFixedGroupLastIterationCont*)top;
- rewind->tryingBody = false;
- }
- // else: we never pushed a rewind continuation
- // Bind group
- GroupInfo* groupInfo = matcher.GroupIdToGroupInfo(begin->groupId);
- groupInfo->offset = inputOffset - begin->length;
- groupInfo->length = begin->length;
- instPointer = matcher.LabelToInstPointer(begin->exitLabel);
- }
- else
- {
- // CHOICEPOINT: Try one more iteration of body. Failure of body will rewind input to here and
- // try follow.
- if (loopInfo->number == begin->repeats.lower)
- {
- // i.e. begin->repeats.lower > 0, so continuation won't have been pushed in BeginLoopFixed
- PUSH(contStack, RewindLoopFixedGroupLastIterationCont, beginLabel, true);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- instPointer = matcher.LabelToInstPointer(beginLabel + sizeof(BeginLoopFixedGroupLastIterationInst));
- }
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int RepeatLoopFixedGroupLastIterationInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("RepeatLoopFixedGroupLastIteration");
- PRINT_MIXIN(RepeatLoopMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(RepeatLoopMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // BeginGreedyLoopNoBacktrackInst
- // ----------------------------------------------------------------------
- inline bool BeginGreedyLoopNoBacktrackInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- LoopInfo* loopInfo = matcher.LoopIdToLoopInfo(loopId);
- loopInfo->number = 0;
- loopInfo->startInputOffset = inputOffset;
- // CHOICEPOINT: Try one iteration of body, if backtrack continue from here with no iterations
- PUSH(contStack, ResumeCont, inputOffset, exitLabel);
- instPointer += sizeof(*this);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int BeginGreedyLoopNoBacktrackInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("BeginGreedyLoopNoBacktrack");
- PRINT_MIXIN(GreedyLoopNoBacktrackMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(GreedyLoopNoBacktrackMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // RepeatGreedyLoopNoBacktrackInst
- // ----------------------------------------------------------------------
- inline bool RepeatGreedyLoopNoBacktrackInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- BeginGreedyLoopNoBacktrackInst* begin = matcher.L2I(BeginGreedyLoopNoBacktrack, beginLabel);
- LoopInfo* loopInfo = matcher.LoopIdToLoopInfo(begin->loopId);
- loopInfo->number++;
- if (inputOffset == loopInfo->startInputOffset)
- {
- // No progress
- return matcher.Fail(FAIL_PARAMETERS);
- }
- else
- {
- // CHOICEPOINT: Try one more iteration of body, if backtrack, continue from here with no more iterations.
- // Since the loop body is deterministic and group free, it wouldn't have left any continuation records.
- // Therefore we can simply update the Resume continuation still on the top of the stack with the current
- // input pointer.
- Cont* top = contStack.Top();
- Assert(top != 0 && top->tag == Cont::ContTag::Resume);
- ResumeCont* resume = (ResumeCont*)top;
- resume->origInputOffset = inputOffset;
- loopInfo->startInputOffset = inputOffset;
- instPointer = matcher.LabelToInstPointer(beginLabel + sizeof(BeginGreedyLoopNoBacktrackInst));
- }
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int RepeatGreedyLoopNoBacktrackInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("RepeatGreedyLoopNoBacktrack");
- PRINT_MIXIN(RepeatLoopMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(RepeatLoopMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // ChompCharInst (optimized instruction)
- // ----------------------------------------------------------------------
- template<ChompMode Mode>
- inline bool ChompCharInst<Mode>::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- const Char matchC = c;
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (Mode == ChompMode::Star || (inputOffset < inputLength && input[inputOffset] == matchC))
- {
- while (true)
- {
- if (Mode != ChompMode::Star)
- {
- ++inputOffset;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset < inputLength && input[inputOffset] == matchC)
- {
- if (Mode == ChompMode::Star)
- {
- ++inputOffset;
- }
- continue;
- }
- break;
- }
- instPointer += sizeof(*this);
- return false;
- }
- return matcher.Fail(FAIL_PARAMETERS);
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- template<ChompMode Mode>
- int ChompCharInst<Mode>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- if (Mode == ChompMode::Star)
- {
- PRINT_RE_BYTECODE_BEGIN("ChompChar<Star>");
- }
- else
- {
- PRINT_RE_BYTECODE_BEGIN("ChompChar<Plus>");
- }
- PRINT_MIXIN(CharMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(CharMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // ChompSetInst (optimized instruction)
- // ----------------------------------------------------------------------
- template<ChompMode Mode>
- inline bool ChompSetInst<Mode>::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- const RuntimeCharSet<Char>& matchSet = this->set;
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if(Mode == ChompMode::Star || (inputOffset < inputLength && matchSet.Get(input[inputOffset])))
- {
- while(true)
- {
- if (Mode != ChompMode::Star)
- {
- ++inputOffset;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset < inputLength && matchSet.Get(input[inputOffset]))
- {
- if (Mode == ChompMode::Star)
- {
- ++inputOffset;
- }
- continue;
- }
- break;
- }
- instPointer += sizeof(*this);
- return false;
- }
- return matcher.Fail(FAIL_PARAMETERS);
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- template<ChompMode Mode>
- int ChompSetInst<Mode>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- if (Mode == ChompMode::Star)
- {
- PRINT_RE_BYTECODE_BEGIN("ChompSet<Star>");
- }
- else
- {
- PRINT_RE_BYTECODE_BEGIN("ChompSet<Plus>");
- }
- PRINT_MIXIN(SetMixin<false>);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(SetMixin<false>);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // ChompCharGroupInst (optimized instruction)
- // ----------------------------------------------------------------------
- template<ChompMode Mode>
- inline bool ChompCharGroupInst<Mode>::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- Assert(matcher.GroupIdToGroupInfo(groupId)->IsUndefined());
- const CharCount inputStartOffset = inputOffset;
- const Char matchC = c;
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if(Mode == ChompMode::Star || (inputOffset < inputLength && input[inputOffset] == matchC))
- {
- while (true)
- {
- if (Mode != ChompMode::Star)
- {
- ++inputOffset;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset < inputLength && input[inputOffset] == matchC)
- {
- if (Mode == ChompMode::Star)
- {
- ++inputOffset;
- }
- continue;
- }
- break;
- }
- if (!noNeedToSave)
- {
- // UNDO ACTION: Restore group on backtrack
- PUSH(contStack, ResetGroupCont, groupId);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- GroupInfo *const groupInfo = matcher.GroupIdToGroupInfo(groupId);
- groupInfo->offset = inputStartOffset;
- groupInfo->length = inputOffset - inputStartOffset;
- instPointer += sizeof(*this);
- return false;
- }
- return matcher.Fail(FAIL_PARAMETERS);
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- template<ChompMode Mode>
- int ChompCharGroupInst<Mode>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- if (Mode == ChompMode::Star)
- {
- PRINT_RE_BYTECODE_BEGIN("ChompCharGroup<Star>");
- }
- else
- {
- PRINT_RE_BYTECODE_BEGIN("ChompCharGroup<Plus>");
- }
- PRINT_MIXIN_COMMA(CharMixin);
- PRINT_MIXIN_COMMA(GroupMixin);
- PRINT_MIXIN(NoNeedToSaveMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(CharMixin);
- PRINT_BYTES(GroupMixin);
- PRINT_BYTES(NoNeedToSaveMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // ChompSetGroupInst (optimized instruction)
- // ----------------------------------------------------------------------
- template<ChompMode Mode>
- inline bool ChompSetGroupInst<Mode>::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- Assert(matcher.GroupIdToGroupInfo(groupId)->IsUndefined());
- const CharCount inputStartOffset = inputOffset;
- const RuntimeCharSet<Char>& matchSet = this->set;
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (Mode == ChompMode::Star || (inputOffset < inputLength && matchSet.Get(input[inputOffset])))
- {
- while (true)
- {
- if (Mode != ChompMode::Star)
- {
- ++inputOffset;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset < inputLength && matchSet.Get(input[inputOffset]))
- {
- if (Mode == ChompMode::Star)
- {
- ++inputOffset;
- }
- continue;
- }
- break;
- }
- if (!noNeedToSave)
- {
- // UNDO ACTION: Restore group on backtrack
- PUSH(contStack, ResetGroupCont, groupId);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- GroupInfo *const groupInfo = matcher.GroupIdToGroupInfo(groupId);
- groupInfo->offset = inputStartOffset;
- groupInfo->length = inputOffset - inputStartOffset;
- instPointer += sizeof(*this);
- return false;
- }
- return matcher.Fail(FAIL_PARAMETERS);
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- template<ChompMode Mode>
- int ChompSetGroupInst<Mode>::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- if (Mode == ChompMode::Star)
- {
- PRINT_RE_BYTECODE_BEGIN("ChompSetGroup<Star>");
- }
- else
- {
- PRINT_RE_BYTECODE_BEGIN("ChompSetGroup<Plus>");
- }
- PRINT_MIXIN_COMMA(SetMixin<false>);
- PRINT_MIXIN_COMMA(GroupMixin);
- PRINT_MIXIN(NoNeedToSaveMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(SetMixin<false>);
- PRINT_BYTES(GroupMixin);
- PRINT_BYTES(NoNeedToSaveMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // ChompCharBoundedInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool ChompCharBoundedInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- const Char matchC = c;
- const CharCount loopMatchStart = inputOffset;
- const CharCountOrFlag repeatsUpper = repeats.upper;
- const CharCount inputEndOffset =
- static_cast<CharCount>(repeatsUpper) >= inputLength - inputOffset
- ? inputLength
- : inputOffset + static_cast<CharCount>(repeatsUpper);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- while (inputOffset < inputEndOffset && input[inputOffset] == matchC)
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- inputOffset++;
- }
- if (inputOffset - loopMatchStart < repeats.lower)
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int ChompCharBoundedInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("ChompCharBounded");
- PRINT_MIXIN_COMMA(CharMixin);
- PRINT_MIXIN(ChompBoundedMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(CharMixin);
- PRINT_BYTES(ChompBoundedMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // ChompSetBoundedInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool ChompSetBoundedInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- const RuntimeCharSet<Char>& matchSet = this->set;
- const CharCount loopMatchStart = inputOffset;
- const CharCountOrFlag repeatsUpper = repeats.upper;
- const CharCount inputEndOffset =
- static_cast<CharCount>(repeatsUpper) >= inputLength - inputOffset
- ? inputLength
- : inputOffset + static_cast<CharCount>(repeatsUpper);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- while (inputOffset < inputEndOffset && matchSet.Get(input[inputOffset]))
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- inputOffset++;
- }
- if (inputOffset - loopMatchStart < repeats.lower)
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int ChompSetBoundedInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("ChompSetBounded");
- PRINT_MIXIN_COMMA(SetMixin<false>);
- PRINT_MIXIN(ChompBoundedMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(SetMixin<false>);
- PRINT_BYTES(ChompBoundedMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // ChompSetBoundedGroupLastCharInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool ChompSetBoundedGroupLastCharInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- Assert(matcher.GroupIdToGroupInfo(groupId)->IsUndefined());
- const RuntimeCharSet<Char>& matchSet = this->set;
- const CharCount loopMatchStart = inputOffset;
- const CharCountOrFlag repeatsUpper = repeats.upper;
- const CharCount inputEndOffset =
- static_cast<CharCount>(repeatsUpper) >= inputLength - inputOffset
- ? inputLength
- : inputOffset + static_cast<CharCount>(repeatsUpper);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- while (inputOffset < inputEndOffset && matchSet.Get(input[inputOffset]))
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- inputOffset++;
- }
- if (inputOffset - loopMatchStart < repeats.lower)
- {
- return matcher.Fail(FAIL_PARAMETERS);
- }
- if (inputOffset > loopMatchStart)
- {
- if (!noNeedToSave)
- {
- PUSH(contStack, ResetGroupCont, groupId);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- }
- GroupInfo *const groupInfo = matcher.GroupIdToGroupInfo(groupId);
- groupInfo->offset = inputOffset - 1;
- groupInfo->length = 1;
- }
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int ChompSetBoundedGroupLastCharInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("ChompSetBoundedGroupLastChar");
- PRINT_MIXIN_COMMA(SetMixin<false>);
- PRINT_MIXIN_COMMA(ChompBoundedMixin);
- PRINT_MIXIN_COMMA(GroupMixin);
- PRINT_MIXIN(NoNeedToSaveMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(SetMixin<false>);
- PRINT_BYTES(ChompBoundedMixin);
- PRINT_BYTES(GroupMixin);
- PRINT_BYTES(NoNeedToSaveMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // TryInst
- // ----------------------------------------------------------------------
- inline bool TryInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- // CHOICEPOINT: Resume at fail label on backtrack
- PUSH(contStack, ResumeCont, inputOffset, failLabel);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int TryInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("Try");
- PRINT_MIXIN(TryMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(TryMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // TryIfCharInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool TryIfCharInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset < inputLength && input[inputOffset] == c)
- {
- // CHOICEPOINT: Resume at fail label on backtrack
- PUSH(contStack, ResumeCont, inputOffset, failLabel);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- instPointer += sizeof(*this);
- return false;
- }
- // Proceed directly to exit
- instPointer = matcher.LabelToInstPointer(failLabel);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int TryIfCharInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("TryIfChar");
- PRINT_MIXIN_COMMA(CharMixin);
- PRINT_MIXIN(TryMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(CharMixin);
- PRINT_BYTES(TryMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // TryMatchCharInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool TryMatchCharInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset < inputLength && input[inputOffset] == c)
- {
- // CHOICEPOINT: Resume at fail label on backtrack
- PUSH(contStack, ResumeCont, inputOffset, failLabel);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- inputOffset++;
- instPointer += sizeof(*this);
- return false;
- }
- // Proceed directly to exit
- instPointer = matcher.LabelToInstPointer(failLabel);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int TryMatchCharInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("TryMatchChar");
- PRINT_MIXIN_COMMA(CharMixin);
- PRINT_MIXIN(TryMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(CharMixin);
- PRINT_BYTES(TryMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // TryIfSetInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool TryIfSetInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset < inputLength && set.Get(input[inputOffset]))
- {
- // CHOICEPOINT: Resume at fail label on backtrack
- PUSH(contStack, ResumeCont, inputOffset, failLabel);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- instPointer += sizeof(*this);
- return false;
- }
- // Proceed directly to exit
- instPointer = matcher.LabelToInstPointer(failLabel);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int TryIfSetInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("TryIfSet");
- PRINT_MIXIN_COMMA(SetMixin<false>);
- PRINT_MIXIN(TryMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(SetMixin<false>);
- PRINT_BYTES(TryMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // TryMatchSetInst (optimized instruction)
- // ----------------------------------------------------------------------
- inline bool TryMatchSetInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.CompStats();
- #endif
- if (inputOffset < inputLength && set.Get(input[inputOffset]))
- {
- // CHOICEPOINT: Resume at fail label on backtrack
- PUSH(contStack, ResumeCont, inputOffset, failLabel);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- inputOffset++;
- instPointer += sizeof(*this);
- return false;
- }
- // Proceed directly to exit
- instPointer = matcher.LabelToInstPointer(failLabel);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int TryMatchSetInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("TryMatchSet");
- PRINT_MIXIN_COMMA(SetMixin<false>);
- PRINT_MIXIN(TryMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(SetMixin<false>);
- PRINT_BYTES(TryMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // BeginAssertionInst
- // ----------------------------------------------------------------------
- inline bool BeginAssertionInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- Assert(instPointer == (uint8*)this);
- if (!isNegation)
- {
- // If the positive assertion binds some groups then on success any RestoreGroup continuations pushed
- // in the assertion body will be cut. Hence if the entire assertion is backtracked over we must restore
- // the current inner group bindings.
- matcher.SaveInnerGroups(minBodyGroupId, maxBodyGroupId, false, input, contStack);
- }
- PUSHA(assertionStack, AssertionInfo, matcher.InstPointerToLabel(instPointer), inputOffset, contStack.Position());
- PUSH(contStack, PopAssertionCont);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.PushStats(contStack, input);
- #endif
- instPointer += sizeof(*this);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int BeginAssertionInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("BeginAssertion");
- PRINT_MIXIN_COMMA(BodyGroupsMixin);
- PRINT_MIXIN_COMMA(NegationMixin);
- PRINT_MIXIN(NextLabelMixin);
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(BodyGroupsMixin);
- PRINT_BYTES(NegationMixin);
- PRINT_BYTES(NextLabelMixin);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // EndAssertionInst
- // ----------------------------------------------------------------------
- inline bool EndAssertionInst::Exec(REGEX_INST_EXEC_PARAMETERS) const
- {
- if (!matcher.PopAssertion(inputOffset, instPointer, contStack, assertionStack, true))
- {
- // Body of negative assertion succeeded, so backtrack
- return matcher.Fail(FAIL_PARAMETERS);
- }
- // else: body of positive assertion succeeded, instruction pointer already at next instruction
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int EndAssertionInst::Print(DebugWriter* w, Label label, const Char* litbuf) const
- {
- PRINT_RE_BYTECODE_BEGIN("EndAssertion");
- PRINT_RE_BYTECODE_MID();
- PRINT_BYTES(EndAssertionInst);
- PRINT_RE_BYTECODE_END();
- }
- #endif
- // ----------------------------------------------------------------------
- // Matcher state
- // ----------------------------------------------------------------------
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void LoopInfo::Print(DebugWriter* w) const
- {
- w->Print(_u("number: %u, startInputOffset: %u"), number, startInputOffset);
- }
- #endif
- void LoopInfo::EnsureOffsetsOfFollowFirst(Matcher& matcher)
- {
- if (this->offsetsOfFollowFirst == nullptr)
- {
- this->offsetsOfFollowFirst = JsUtil::List<CharCount, ArenaAllocator>::New(matcher.pattern->library->GetScriptContext()->RegexAllocator());
- }
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void GroupInfo::Print(DebugWriter* w, const Char* const input) const
- {
- if (IsUndefined())
- {
- w->Print(_u("<undefined> (%u)"), offset);
- }
- else
- {
- w->PrintQuotedString(input + offset, (CharCount)length);
- w->Print(_u(" (%u+%u)"), offset, (CharCount)length);
- }
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void AssertionInfo::Print(DebugWriter* w) const
- {
- w->PrintEOL(_u("beginLabel: L%04x, startInputOffset: %u, contStackPosition: $llu"), beginLabel, startInputOffset, static_cast<unsigned long long>(contStackPosition));
- }
- #endif
- // ----------------------------------------------------------------------
- // ResumeCont
- // ----------------------------------------------------------------------
- inline bool ResumeCont::Exec(REGEX_CONT_EXEC_PARAMETERS)
- {
- inputOffset = origInputOffset;
- instPointer = matcher.LabelToInstPointer(origInstLabel);
- return true; // STOP BACKTRACKING
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int ResumeCont::Print(DebugWriter* w, const Char* const input) const
- {
- w->PrintEOL(_u("Resume(origInputOffset: %u, origInstLabel: L%04x)"), origInputOffset, origInstLabel);
- return sizeof(*this);
- }
- #endif
- // ----------------------------------------------------------------------
- // RestoreLoopCont
- // ----------------------------------------------------------------------
- inline RestoreLoopCont::RestoreLoopCont(int loopId, LoopInfo& origLoopInfo, Matcher& matcher) : Cont(ContTag::RestoreLoop), loopId(loopId)
- {
- this->origLoopInfo.number = origLoopInfo.number;
- this->origLoopInfo.startInputOffset = origLoopInfo.startInputOffset;
- this->origLoopInfo.offsetsOfFollowFirst = nullptr;
- if (origLoopInfo.offsetsOfFollowFirst != nullptr)
- {
- this->origLoopInfo.offsetsOfFollowFirst = JsUtil::List<CharCount, ArenaAllocator>::New(matcher.pattern->library->GetScriptContext()->RegexAllocator());
- this->origLoopInfo.offsetsOfFollowFirst->Copy(origLoopInfo.offsetsOfFollowFirst);
- }
- }
- inline bool RestoreLoopCont::Exec(REGEX_CONT_EXEC_PARAMETERS)
- {
- matcher.QueryContinue(qcTicks);
- *matcher.LoopIdToLoopInfo(loopId) = origLoopInfo;
- return false; // KEEP BACKTRACKING
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int RestoreLoopCont::Print(DebugWriter* w, const Char* const input) const
- {
- w->Print(_u("RestoreLoop(loopId: %d, "), loopId);
- origLoopInfo.Print(w);
- w->PrintEOL(_u(")"));
- return sizeof(*this);
- }
- #endif
- // ----------------------------------------------------------------------
- // RestoreGroupCont
- // ----------------------------------------------------------------------
- inline bool RestoreGroupCont::Exec(REGEX_CONT_EXEC_PARAMETERS)
- {
- *matcher.GroupIdToGroupInfo(groupId) = origGroupInfo;
- return false; // KEEP BACKTRACKING
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int RestoreGroupCont::Print(DebugWriter* w, const Char* const input) const
- {
- w->Print(_u("RestoreGroup(groupId: %d, "), groupId);
- origGroupInfo.Print(w, input);
- w->PrintEOL(_u(")"));
- return sizeof(*this);
- }
- #endif
- // ----------------------------------------------------------------------
- // ResetGroupCont
- // ----------------------------------------------------------------------
- inline bool ResetGroupCont::Exec(REGEX_CONT_EXEC_PARAMETERS)
- {
- matcher.ResetGroup(groupId);
- return false; // KEEP BACKTRACKING
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int ResetGroupCont::Print(DebugWriter* w, const Char* const input) const
- {
- w->PrintEOL(_u("ResetGroup(groupId: %d)"), groupId);
- return sizeof(*this);
- }
- #endif
- // ----------------------------------------------------------------------
- // ResetGroupRangeCont
- // ----------------------------------------------------------------------
- inline bool ResetGroupRangeCont::Exec(REGEX_CONT_EXEC_PARAMETERS)
- {
- matcher.ResetInnerGroups(fromGroupId, toGroupId);
- return false; // KEEP BACKTRACKING
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int ResetGroupRangeCont::Print(DebugWriter* w, const Char* const input) const
- {
- w->PrintEOL(_u("ResetGroupRange(fromGroupId: %d, toGroupId: %d)"), fromGroupId, toGroupId);
- return sizeof(*this);
- }
- #endif
- // ----------------------------------------------------------------------
- // RepeatLoopCont
- // ----------------------------------------------------------------------
- inline bool RepeatLoopCont::Exec(REGEX_CONT_EXEC_PARAMETERS)
- {
- matcher.QueryContinue(qcTicks);
- // Try one more iteration of a non-greedy loop
- BeginLoopInst* begin = matcher.L2I(BeginLoop, beginLabel);
- LoopInfo* loopInfo = matcher.LoopIdToLoopInfo(begin->loopId);
- loopInfo->startInputOffset = inputOffset = origInputOffset;
- instPointer = matcher.LabelToInstPointer(beginLabel + sizeof(BeginLoopInst));
- if(begin->hasInnerNondet)
- {
- // If it backtracks into the loop body of an earlier iteration, it must restore inner groups for that iteration.
- // Save the inner groups and reset them for the next iteration.
- matcher.SaveInnerGroups(begin->minBodyGroupId, begin->maxBodyGroupId, true, input, contStack);
- }
- else
- {
- // If it backtracks, the entire loop will fail, so no need to restore groups. Just reset the inner groups for
- // the next iteration.
- matcher.ResetInnerGroups(begin->minBodyGroupId, begin->maxBodyGroupId);
- }
- return true; // STOP BACKTRACKING
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int RepeatLoopCont::Print(DebugWriter* w, const Char* const input) const
- {
- w->PrintEOL(_u("RepeatLoop(beginLabel: L%04x, origInputOffset: %u)"), beginLabel, origInputOffset);
- return sizeof(*this);
- }
- #endif
- // ----------------------------------------------------------------------
- // PopAssertionCont
- // ----------------------------------------------------------------------
- inline bool PopAssertionCont::Exec(REGEX_CONT_EXEC_PARAMETERS)
- {
- Assert(!assertionStack.IsEmpty());
- if (matcher.PopAssertion(inputOffset, instPointer, contStack, assertionStack, false))
- {
- // Body of negative assertion failed
- return true; // STOP BACKTRACKING
- }
- else
- {
- // Body of positive assertion failed
- return false; // CONTINUE BACKTRACKING
- }
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int PopAssertionCont::Print(DebugWriter* w, const Char* const input) const
- {
- w->PrintEOL(_u("PopAssertion()"));
- return sizeof(*this);
- }
- #endif
- // ----------------------------------------------------------------------
- // RewindLoopFixedCont
- // ----------------------------------------------------------------------
- inline bool RewindLoopFixedCont::Exec(REGEX_CONT_EXEC_PARAMETERS)
- {
- matcher.QueryContinue(qcTicks);
- BeginLoopFixedInst* begin = matcher.L2I(BeginLoopFixed, beginLabel);
- LoopInfo* loopInfo = matcher.LoopIdToLoopInfo(begin->loopId);
- if (tryingBody)
- {
- tryingBody = false;
- // loopInfo->number is the number of iterations completed before trying body
- Assert(loopInfo->number >= begin->repeats.lower);
- }
- else
- {
- // loopInfo->number is the number of iterations completed before trying follow
- Assert(loopInfo->number > begin->repeats.lower);
- // Try follow with one fewer iteration
- loopInfo->number--;
- }
- // Rewind input
- inputOffset = loopInfo->startInputOffset + loopInfo->number * begin->length;
- if (loopInfo->number > begin->repeats.lower)
- {
- // Un-pop the continuation ready for next time
- contStack.UnPop<RewindLoopFixedCont>();
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.UnPopStats(contStack, input);
- #endif
- }
- // else: Can't try any fewer iterations if follow fails, so leave continuation as popped and let failure propagate
- instPointer = matcher.LabelToInstPointer(begin->exitLabel);
- return true; // STOP BACKTRACKING
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int RewindLoopFixedCont::Print(DebugWriter* w, const Char* const input) const
- {
- w->PrintEOL(_u("RewindLoopFixed(beginLabel: L%04x, tryingBody: %s)"), beginLabel, tryingBody ? _u("true") : _u("false"));
- return sizeof(*this);
- }
- #endif
- // ----------------------------------------------------------------------
- // RewindLoopSetCont
- // ----------------------------------------------------------------------
- inline bool RewindLoopSetCont::Exec(REGEX_CONT_EXEC_PARAMETERS)
- {
- matcher.QueryContinue(qcTicks);
- LoopSetInst* begin = matcher.L2I(LoopSet, beginLabel);
- LoopInfo* loopInfo = matcher.LoopIdToLoopInfo(begin->loopId);
- // loopInfo->number is the number of iterations completed before trying follow
- Assert(loopInfo->number > begin->repeats.lower);
- // Try follow with fewer iterations
- loopInfo->number--;
- // Rewind input
- inputOffset = loopInfo->startInputOffset + loopInfo->number;
- if (loopInfo->number > begin->repeats.lower)
- {
- // Un-pop the continuation ready for next time
- contStack.UnPop<RewindLoopSetCont>();
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.UnPopStats(contStack, input);
- #endif
- }
- // else: Can't try any fewer iterations if follow fails, so leave continuation as popped and let failure propagate
- instPointer = matcher.LabelToInstPointer(beginLabel + sizeof(LoopSetInst));
- return true; // STOP BACKTRACKING
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int RewindLoopSetCont::Print(DebugWriter* w, const Char* const input) const
- {
- w->PrintEOL(_u("RewindLoopSet(beginLabel: L%04x)"), beginLabel);
- return sizeof(*this);
- }
- #endif
- // ----------------------------------------------------------------------
- // RewindLoopSetWithFollowFirstCont
- // ----------------------------------------------------------------------
- inline bool RewindLoopSetWithFollowFirstCont::Exec(REGEX_CONT_EXEC_PARAMETERS)
- {
- matcher.QueryContinue(qcTicks);
- LoopSetWithFollowFirstInst* begin = matcher.L2I(LoopSetWithFollowFirst, beginLabel);
- LoopInfo* loopInfo = matcher.LoopIdToLoopInfo(begin->loopId);
- // loopInfo->number is the number of iterations completed before trying follow
- Assert(loopInfo->number > begin->repeats.lower);
- // Try follow with fewer iterations
- if (loopInfo->offsetsOfFollowFirst == nullptr)
- {
- if (begin->followFirst != MaxUChar)
- {
- // We determined the first character in the follow set at compile time,
- // but didn't find a single match for it in the last iteration of the loop.
- // So, there is no benefit in backtracking.
- loopInfo->number = begin->repeats.lower; // stop backtracking
- }
- else
- {
- // We couldn't determine the first character in the follow set at compile time;
- // fall back to backtracking by one character at a time.
- loopInfo->number--;
- }
- }
- else
- {
- if (loopInfo->offsetsOfFollowFirst->Empty())
- {
- // We have already backtracked to the first offset where we matched the LoopSet's followFirst;
- // no point in backtracking more.
- loopInfo->number = begin->repeats.lower; // stop backtracking
- }
- else
- {
- // Backtrack to the previous offset where we matched the LoopSet's followFirst
- // We will be doing one unnecessary match. But, if we wanted to avoid it, we'd have
- // to propagate to the next Inst, that the first character is already matched.
- // Seems like an overkill to avoid one match.
- loopInfo->number = loopInfo->offsetsOfFollowFirst->RemoveAtEnd();
- }
- }
- // If loopInfo->number now is less than begins->repeats.lower, the loop
- // shouldn't match anything. In that case, stop backtracking.
- loopInfo->number = max(loopInfo->number, begin->repeats.lower);
- // Rewind input
- inputOffset = loopInfo->startInputOffset + loopInfo->number;
- if (loopInfo->number > begin->repeats.lower)
- {
- // Un-pop the continuation ready for next time
- contStack.UnPop<RewindLoopSetWithFollowFirstCont>();
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.UnPopStats(contStack, input);
- #endif
- }
- // else: Can't try any fewer iterations if follow fails, so leave continuation as popped and let failure propagate
- instPointer = matcher.LabelToInstPointer(beginLabel + sizeof(LoopSetWithFollowFirstInst));
- return true; // STOP BACKTRACKING
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int RewindLoopSetWithFollowFirstCont::Print(DebugWriter* w, const Char* const input) const
- {
- w->PrintEOL(_u("RewindLoopSetWithFollowFirst(beginLabel: L%04x)"), beginLabel);
- return sizeof(*this);
- }
- #endif
- // ----------------------------------------------------------------------
- // RewindLoopFixedGroupLastIterationCont
- // ----------------------------------------------------------------------
- inline bool RewindLoopFixedGroupLastIterationCont::Exec(REGEX_CONT_EXEC_PARAMETERS)
- {
- matcher.QueryContinue(qcTicks);
- BeginLoopFixedGroupLastIterationInst* begin = matcher.L2I(BeginLoopFixedGroupLastIteration, beginLabel);
- LoopInfo* loopInfo = matcher.LoopIdToLoopInfo(begin->loopId);
- GroupInfo* groupInfo = matcher.GroupIdToGroupInfo(begin->groupId);
- if (tryingBody)
- {
- tryingBody = false;
- // loopInfo->number is the number of iterations completed before current attempt of body
- Assert(loopInfo->number >= begin->repeats.lower);
- }
- else
- {
- // loopInfo->number is the number of iterations completed before trying follow
- Assert(loopInfo->number > begin->repeats.lower);
- // Try follow with one fewer iteration
- loopInfo->number--;
- }
- // Rewind input
- inputOffset = loopInfo->startInputOffset + loopInfo->number * begin->length;
- if (loopInfo->number > 0)
- {
- // Bind previous iteration's body
- groupInfo->offset = inputOffset - begin->length;
- groupInfo->length = begin->length;
- }
- else
- {
- groupInfo->Reset();
- }
- if (loopInfo->number > begin->repeats.lower)
- {
- // Un-pop the continuation ready for next time
- contStack.UnPop<RewindLoopFixedGroupLastIterationCont>();
- #if ENABLE_REGEX_CONFIG_OPTIONS
- matcher.UnPopStats(contStack, input);
- #endif
- }
- // else: Can't try any fewer iterations if follow fails, so leave continuation as popped and let failure propagate
- instPointer = matcher.LabelToInstPointer(begin->exitLabel);
- return true; // STOP BACKTRACKING
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- int RewindLoopFixedGroupLastIterationCont::Print(DebugWriter* w, const Char* const input) const
- {
- w->PrintEOL(_u("RewindLoopFixedGroupLastIteration(beginLabel: L%04x, tryingBody: %s)"), beginLabel, tryingBody ? _u("true") : _u("false"));
- return sizeof(*this);
- }
- #endif
- // ----------------------------------------------------------------------
- // Matcher
- // ----------------------------------------------------------------------
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void ContStack::Print(DebugWriter* w, const Char* const input) const
- {
- for (Iterator it(*this); it; ++it)
- {
- w->Print(_u("%4llu: "), static_cast<unsigned long long>(it.Position()));
- it->Print(w, input);
- }
- }
- #endif
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void AssertionStack::Print(DebugWriter* w, const Matcher* matcher) const
- {
- for (Iterator it(*this); it; ++it)
- {
- it->Print(w);
- }
- }
- #endif
- Matcher::Matcher(Js::ScriptContext* scriptContext, RegexPattern* pattern)
- : pattern(pattern)
- , standardChars(scriptContext->GetThreadContext()->GetStandardChars((char16*)0))
- , program(pattern->rep.unified.program)
- , groupInfos(nullptr)
- , loopInfos(nullptr)
- , literalNextSyncInputOffsets(nullptr)
- , recycler(scriptContext->GetRecycler())
- , previousQcTime(0)
- #if ENABLE_REGEX_CONFIG_OPTIONS
- , stats(0)
- , w(0)
- #endif
- {
- // Don't need to zero out - the constructor for GroupInfo should take care of it
- groupInfos = RecyclerNewArrayLeaf(recycler, GroupInfo, program->numGroups);
- if (program->numLoops > 0)
- {
- loopInfos = RecyclerNewArrayLeafZ(recycler, LoopInfo, program->numLoops);
- }
- }
- Matcher *Matcher::New(Js::ScriptContext* scriptContext, RegexPattern* pattern)
- {
- return RecyclerNew(scriptContext->GetRecycler(), Matcher, scriptContext, pattern);
- }
- Matcher *Matcher::CloneToScriptContext(Js::ScriptContext *scriptContext, RegexPattern *pattern)
- {
- Matcher *result = New(scriptContext, pattern);
- if (groupInfos)
- {
- size_t size = program->numGroups * sizeof(GroupInfo);
- js_memcpy_s(result->groupInfos, size, groupInfos, size);
- }
- if (loopInfos)
- {
- size_t size = program->numLoops * sizeof(LoopInfo);
- js_memcpy_s(result->loopInfos, size, loopInfos, size);
- }
- return result;
- }
- #if DBG
- const Cont::ContTag contTags[] = {
- #define M(O) Cont::ContTag::O,
- #include "RegexContcodes.h"
- #undef M
- };
- const Cont::ContTag minContTag = contTags[0];
- const Cont::ContTag maxContTag = contTags[(sizeof(contTags) / sizeof(Cont::ContTag)) - 1];
- #endif
- void Matcher::DoQueryContinue(const uint qcTicks)
- {
- // See definition of TimePerQc for description of regex QC heuristics
- const uint before = previousQcTime;
- const uint now = GetTickCount();
- if ((!before || now - before < TimePerQc) && qcTicks & TicksPerQc - 1)
- {
- return;
- }
- previousQcTime = now;
- TraceQueryContinue(now);
- // Query-continue can be reentrant and run the same regex again. To prevent the matcher and other persistent objects
- // from being reused reentrantly, save and restore them around the QC call.
- class AutoCleanup
- {
- private:
- RegexPattern *const pattern;
- Matcher *const matcher;
- RegexStacks * regexStacks;
- public:
- AutoCleanup(RegexPattern *const pattern, Matcher *const matcher) : pattern(pattern), matcher(matcher)
- {
- Assert(pattern);
- Assert(matcher);
- Assert(pattern->rep.unified.matcher == matcher);
- pattern->rep.unified.matcher = nullptr;
- const auto scriptContext = pattern->GetScriptContext();
- regexStacks = scriptContext->SaveRegexStacks();
- }
- ~AutoCleanup()
- {
- pattern->rep.unified.matcher = matcher;
- const auto scriptContext = pattern->GetScriptContext();
- scriptContext->RestoreRegexStacks(regexStacks);
- }
- } autoCleanup(pattern, this);
- pattern->GetScriptContext()->GetThreadContext()->CheckScriptInterrupt();
- }
- void Matcher::TraceQueryContinue(const uint now)
- {
- if (!PHASE_TRACE1(Js::RegexQcPhase))
- {
- return;
- }
- Output::Print(_u("Regex QC"));
- static uint n = 0;
- static uint firstQcTime = 0;
- ++n;
- if (firstQcTime)
- {
- Output::Print(_u(" - frequency: %0.1f"), static_cast<double>(n * 1000) / (now - firstQcTime));
- }
- else
- {
- firstQcTime = now;
- }
- Output::Print(_u("\n"));
- Output::Flush();
- }
- bool Matcher::Fail(const Char* const input, CharCount &inputOffset, const uint8 *&instPointer, ContStack &contStack, AssertionStack &assertionStack, uint &qcTicks)
- {
- if (!contStack.IsEmpty())
- {
- if (!RunContStack(input, inputOffset, instPointer, contStack, assertionStack, qcTicks))
- {
- return false;
- }
- }
- Assert(assertionStack.IsEmpty());
- groupInfos[0].Reset();
- return true; // STOP EXECUTION
- }
- inline bool Matcher::RunContStack(const Char* const input, CharCount &inputOffset, const uint8 *&instPointer, ContStack &contStack, AssertionStack &assertionStack, uint &qcTicks)
- {
- while (true)
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- PopStats(contStack, input);
- #endif
- Cont* cont = contStack.Pop();
- if (cont == 0)
- {
- break;
- }
- Assert(cont->tag >= minContTag && cont->tag <= maxContTag);
- // All these cases RESUME EXECUTION if backtracking finds a stop point
- const Cont::ContTag tag = cont->tag;
- switch (tag)
- {
- #define M(O) case Cont::ContTag::O: if (((O##Cont*)cont)->Exec(*this, input, inputOffset, instPointer, contStack, assertionStack, qcTicks)) return false; break;
- #include "RegexContcodes.h"
- #undef M
- default:
- Assert(false); // should never be reached
- return false; // however, can't use complier optimization if we wnat to return false here
- }
- }
- return true;
- }
- #if DBG
- const Inst::InstTag instTags[] = {
- #define M(TagName) Inst::InstTag::TagName,
- #define MTemplate(TagName, ...) M(TagName)
- #include "RegexOpCodes.h"
- #undef M
- #undef MTemplate
- };
- const Inst::InstTag minInstTag = instTags[0];
- const Inst::InstTag maxInstTag = instTags[(sizeof(instTags) / sizeof(Inst::InstTag)) - 1];
- #endif
- inline void Matcher::Run(const Char* const input, const CharCount inputLength, CharCount &matchStart, CharCount &nextSyncInputOffset, ContStack &contStack, AssertionStack &assertionStack, uint &qcTicks, bool firstIteration)
- {
- CharCount inputOffset = matchStart;
- const uint8 *instPointer = program->rep.insts.insts;
- Assert(instPointer != 0);
- while (true)
- {
- Assert(inputOffset >= matchStart && inputOffset <= inputLength);
- Assert(instPointer >= program->rep.insts.insts && instPointer < program->rep.insts.insts + program->rep.insts.instsLen);
- Assert(((Inst*)instPointer)->tag >= minInstTag && ((Inst*)instPointer)->tag <= maxInstTag);
- #if ENABLE_REGEX_CONFIG_OPTIONS
- if (w != 0)
- {
- Print(w, input, inputLength, inputOffset, instPointer, contStack, assertionStack);
- }
- InstStats();
- #endif
- const Inst *inst = (const Inst*)instPointer;
- const Inst::InstTag tag = inst->tag;
- switch (tag)
- {
- #define MBase(TagName, ClassName) \
- case Inst::InstTag::TagName: \
- if (((const ClassName *)inst)->Exec(*this, input, inputLength, matchStart, inputOffset, nextSyncInputOffset, instPointer, contStack, assertionStack, qcTicks, firstIteration)) { return; } \
- break;
- #define M(TagName) MBase(TagName, TagName##Inst)
- #define MTemplate(TagName, TemplateDeclaration, GenericClassName, SpecializedClassName) MBase(TagName, SpecializedClassName)
- #include "RegexOpCodes.h"
- #undef MBase
- #undef M
- #undef MTemplate
- default:
- Assert(false);
- __assume(false);
- }
- }
- }
- #if DBG
- void Matcher::ResetLoopInfos()
- {
- for (int i = 0; i < program->numLoops; i++)
- {
- loopInfos[i].Reset();
- }
- }
- #endif
- inline bool Matcher::MatchHere(const Char* const input, const CharCount inputLength, CharCount &matchStart, CharCount &nextSyncInputOffset, ContStack &contStack, AssertionStack &assertionStack, uint &qcTicks, bool firstIteration)
- {
- // Reset the continuation and assertion stacks ready for fresh run
- // NOTE: We used to do this after the Run, but it's safer to do it here in case unusual control flow exits
- // the matcher without executing the clears.
- contStack.Clear();
- // assertionStack may be non-empty since we can hard fail directly out of matcher without popping assertion
- assertionStack.Clear();
- Assert(contStack.IsEmpty());
- Assert(assertionStack.IsEmpty());
- ResetInnerGroups(0, program->numGroups - 1);
- #if DBG
- ResetLoopInfos();
- #endif
- Run(input, inputLength, matchStart, nextSyncInputOffset, contStack, assertionStack, qcTicks, firstIteration);
- // Leave the continuation and assertion stack memory in place so we don't have to alloc next time
- return WasLastMatchSuccessful();
- }
- inline bool Matcher::MatchSingleCharCaseInsensitive(const Char* const input, const CharCount inputLength, CharCount offset, const Char c)
- {
- CaseInsensitive::MappingSource mappingSource = program->GetCaseMappingSource();
- // If sticky flag is present, break since the 1st character didn't match the pattern character
- if ((program->flags & StickyRegexFlag) != 0)
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- CompStats();
- #endif
- if (MatchSingleCharCaseInsensitiveHere(mappingSource, input, offset, c))
- {
- GroupInfo* const info = GroupIdToGroupInfo(0);
- info->offset = offset;
- info->length = 1;
- return true;
- }
- else
- {
- ResetGroup(0);
- return false;
- }
- }
- while (offset < inputLength)
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- CompStats();
- #endif
- if (MatchSingleCharCaseInsensitiveHere(mappingSource, input, offset, c))
- {
- GroupInfo* const info = GroupIdToGroupInfo(0);
- info->offset = offset;
- info->length = 1;
- return true;
- }
- offset++;
- }
- ResetGroup(0);
- return false;
- }
- inline bool Matcher::MatchSingleCharCaseInsensitiveHere(
- CaseInsensitive::MappingSource mappingSource,
- const Char* const input,
- const CharCount offset,
- const Char c)
- {
- return (standardChars->ToCanonical(mappingSource, input[offset]) == standardChars->ToCanonical(mappingSource, c));
- }
- inline bool Matcher::MatchSingleCharCaseSensitive(const Char* const input, const CharCount inputLength, CharCount offset, const Char c)
- {
- // If sticky flag is present, break since the 1st character didn't match the pattern character
- if ((program->flags & StickyRegexFlag) != 0)
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- CompStats();
- #endif
- if (input[offset] == c)
- {
- GroupInfo* const info = GroupIdToGroupInfo(0);
- info->offset = offset;
- info->length = 1;
- return true;
- }
- else
- {
- ResetGroup(0);
- return false;
- }
- }
- while (offset < inputLength)
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- CompStats();
- #endif
- if (input[offset] == c)
- {
- GroupInfo* const info = GroupIdToGroupInfo(0);
- info->offset = offset;
- info->length = 1;
- return true;
- }
- offset++;
- }
- ResetGroup(0);
- return false;
- }
- inline bool Matcher::MatchBoundedWord(const Char* const input, const CharCount inputLength, CharCount offset)
- {
- const StandardChars<Char>& stdchrs = *standardChars;
- if (offset >= inputLength)
- {
- ResetGroup(0);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- CompStats();
- #endif
- if ((offset == 0 && stdchrs.IsWord(input[0])) ||
- (offset > 0 && (!stdchrs.IsWord(input[offset - 1]) && stdchrs.IsWord(input[offset]))))
- {
- // Already at start of word
- }
- // If sticky flag is present, return false since we are not at the beginning of the word yet
- else if ((program->flags & StickyRegexFlag) == StickyRegexFlag)
- {
- ResetGroup(0);
- return false;
- }
- else
- {
- if (stdchrs.IsWord(input[offset]))
- {
- // Scan for end of current word
- while (true)
- {
- offset++;
- if (offset >= inputLength)
- {
- ResetGroup(0);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- CompStats();
- #endif
- if (!stdchrs.IsWord(input[offset]))
- {
- break;
- }
- }
- }
- // Scan for start of next word
- while (true)
- {
- offset++;
- if (offset >= inputLength)
- {
- ResetGroup(0);
- return false;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- CompStats();
- #endif
- if (stdchrs.IsWord(input[offset]))
- {
- break;
- }
- }
- }
- GroupInfo* const info = GroupIdToGroupInfo(0);
- info->offset = offset;
- // Scan for end of word
- do
- {
- offset++;
- #if ENABLE_REGEX_CONFIG_OPTIONS
- CompStats();
- #endif
- }
- while (offset < inputLength && stdchrs.IsWord(input[offset]));
- info->length = offset - info->offset;
- return true;
- }
- inline bool Matcher::MatchLeadingTrailingSpaces(const Char* const input, const CharCount inputLength, CharCount offset)
- {
- GroupInfo* const info = GroupIdToGroupInfo(0);
- Assert(offset <= inputLength);
- Assert((program->flags & MultilineRegexFlag) == 0);
- if (offset >= inputLength)
- {
- Assert(offset == inputLength);
- if (program->rep.leadingTrailingSpaces.endMinMatch == 0 ||
- (offset == 0 && program->rep.leadingTrailingSpaces.beginMinMatch == 0))
- {
- info->offset = offset;
- info->length = 0;
- return true;
- }
- info->Reset();
- return false;
- }
- const StandardChars<Char> &stdchrs = *standardChars;
- if (offset == 0)
- {
- while (offset < inputLength && stdchrs.IsWhitespaceOrNewline(input[offset]))
- {
- offset++;
- #if ENABLE_REGEX_CONFIG_OPTIONS
- CompStats();
- #endif
- }
- if (offset >= program->rep.leadingTrailingSpaces.beginMinMatch)
- {
- info->offset = 0;
- info->length = offset;
- return true;
- }
- }
- Assert(inputLength > 0);
- const CharCount initOffset = offset;
- offset = inputLength - 1;
- while (offset >= initOffset && stdchrs.IsWhitespaceOrNewline(input[offset]))
- {
- // This can never underflow since initOffset > 0
- Assert(offset > 0);
- offset--;
- #if ENABLE_REGEX_CONFIG_OPTIONS
- CompStats();
- #endif
- }
- offset++;
- CharCount length = inputLength - offset;
- if (length >= program->rep.leadingTrailingSpaces.endMinMatch)
- {
- info->offset = offset;
- info->length = length;
- return true;
- }
- info->Reset();
- return false;
- }
- inline bool Matcher::MatchOctoquad(const Char* const input, const CharCount inputLength, CharCount offset, OctoquadMatcher* matcher)
- {
- if (matcher->Match
- ( input
- , inputLength
- , offset
- #if ENABLE_REGEX_CONFIG_OPTIONS
- , stats
- #endif
- ))
- {
- GroupInfo* const info = GroupIdToGroupInfo(0);
- info->offset = offset;
- info->length = TrigramInfo::PatternLength;
- return true;
- }
- else
- {
- ResetGroup(0);
- return false;
- }
- }
- inline bool Matcher::MatchBOILiteral2(const Char* const input, const CharCount inputLength, CharCount offset, DWORD literal2)
- {
- if (offset == 0 && inputLength >= 2)
- {
- CompileAssert(sizeof(Char) == 2);
- const Program * program = this->program;
- if (program->rep.boiLiteral2.literal == *(DWORD *)input)
- {
- GroupInfo* const info = GroupIdToGroupInfo(0);
- info->offset = 0;
- info->length = 2;
- return true;
- }
- }
- ResetGroup(0);
- return false;
- }
- bool Matcher::Match
- ( const Char* const input
- , const CharCount inputLength
- , CharCount offset
- , Js::ScriptContext * scriptContext
- #if ENABLE_REGEX_CONFIG_OPTIONS
- , RegexStats* stats
- , DebugWriter* w
- #endif
- )
- {
- #if ENABLE_REGEX_CONFIG_OPTIONS
- this->stats = stats;
- this->w = w;
- #endif
- Assert(offset <= inputLength);
- bool res;
- bool loopMatchHere = true;
- Program const *prog = this->program;
- bool isStickyPresent = this->pattern->IsSticky();
- switch (prog->tag)
- {
- case Program::ProgramTag::BOIInstructionsTag:
- if (offset != 0)
- {
- groupInfos[0].Reset();
- res = false;
- break;
- }
- // fall through
- case Program::ProgramTag::BOIInstructionsForStickyFlagTag:
- AssertMsg(prog->tag == Program::ProgramTag::BOIInstructionsTag || isStickyPresent, "prog->tag should be BOIInstructionsForStickyFlagTag if sticky = true.");
- loopMatchHere = false;
- // fall through
- case Program::ProgramTag::InstructionsTag:
- {
- previousQcTime = 0;
- uint qcTicks = 0;
- // This is the next offset in the input from where we will try to sync. For sync instructions that back up, this
- // is used to avoid trying to sync when we have not yet reached the offset in the input we last synced to before
- // backing up.
- CharCount nextSyncInputOffset = offset;
- RegexStacks * regexStacks = scriptContext->RegexStacks();
- // Need to continue matching even if matchStart == inputLim since some patterns may match an empty string at the end
- // of the input. For instance: /a*$/.exec("b")
- bool firstIteration = true;
- do
- {
- // Let there be only one call to MatchHere(), as that call expands the interpreter loop in-place. Having
- // multiple calls to MatchHere() would bloat the code.
- res = MatchHere(input, inputLength, offset, nextSyncInputOffset, regexStacks->contStack, regexStacks->assertionStack, qcTicks, firstIteration);
- firstIteration = false;
- } while(!res && loopMatchHere && ++offset <= inputLength);
- break;
- }
- case Program::ProgramTag::SingleCharTag:
- if (this->pattern->IsIgnoreCase())
- {
- res = MatchSingleCharCaseInsensitive(input, inputLength, offset, prog->rep.singleChar.c);
- }
- else
- {
- res = MatchSingleCharCaseSensitive(input, inputLength, offset, prog->rep.singleChar.c);
- }
- break;
- case Program::ProgramTag::BoundedWordTag:
- res = MatchBoundedWord(input, inputLength, offset);
- break;
- case Program::ProgramTag::LeadingTrailingSpacesTag:
- res = MatchLeadingTrailingSpaces(input, inputLength, offset);
- break;
- case Program::ProgramTag::OctoquadTag:
- res = MatchOctoquad(input, inputLength, offset, prog->rep.octoquad.matcher);
- break;
- case Program::ProgramTag::BOILiteral2Tag:
- res = MatchBOILiteral2(input, inputLength, offset, prog->rep.boiLiteral2.literal);
- break;
- default:
- Assert(false);
- __assume(false);
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- this->stats = 0;
- this->w = 0;
- #endif
- return res;
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void Matcher::Print(DebugWriter* w, const Char* const input, const CharCount inputLength, CharCount inputOffset, const uint8* instPointer, ContStack &contStack, AssertionStack &assertionStack) const
- {
- w->PrintEOL(_u("Matcher {"));
- w->Indent();
- w->Print(_u("program: "));
- w->PrintQuotedString(program->source, program->sourceLen);
- w->EOL();
- w->Print(_u("inputPointer: "));
- if (inputLength == 0)
- {
- w->PrintEOL(_u("<empty input>"));
- }
- else if (inputLength > 1024)
- {
- w->PrintEOL(_u("<string too large>"));
- }
- else
- {
- w->PrintEscapedString(input, inputOffset);
- if (inputOffset >= inputLength)
- {
- w->Print(_u("<<<>>>"));
- }
- else
- {
- w->Print(_u("<<<"));
- w->PrintEscapedChar(input[inputOffset]);
- w->Print(_u(">>>"));
- w->PrintEscapedString(input + inputOffset + 1, inputLength - inputOffset - 1);
- }
- w->EOL();
- }
- if (program->tag == Program::ProgramTag::BOIInstructionsTag || program->tag == Program::ProgramTag::InstructionsTag)
- {
- w->Print(_u("instPointer: "));
- const Inst* inst = (const Inst*)instPointer;
- switch (inst->tag)
- {
- #define MBase(TagName, ClassName) \
- case Inst::InstTag::TagName: \
- { \
- const ClassName *actualInst = static_cast<const ClassName *>(inst); \
- actualInst->Print(w, InstPointerToLabel(instPointer), program->rep.insts.litbuf); \
- break; \
- }
- #define M(TagName) MBase(TagName, TagName##Inst)
- #define MTemplate(TagName, TemplateDeclaration, GenericClassName, SpecializedClassName) MBase(TagName, SpecializedClassName)
- #include "RegexOpCodes.h"
- #undef MBase
- #undef M
- #undef MTemplate
- default:
- Assert(false);
- __assume(false);
- }
- w->PrintEOL(_u("groups:"));
- w->Indent();
- for (int i = 0; i < program->numGroups; i++)
- {
- w->Print(_u("%d: "), i);
- groupInfos[i].Print(w, input);
- w->EOL();
- }
- w->Unindent();
- w->PrintEOL(_u("loops:"));
- w->Indent();
- for (int i = 0; i < program->numLoops; i++)
- {
- w->Print(_u("%d: "), i);
- loopInfos[i].Print(w);
- w->EOL();
- }
- w->Unindent();
- w->PrintEOL(_u("contStack: (top to bottom)"));
- w->Indent();
- contStack.Print(w, input);
- w->Unindent();
- w->PrintEOL(_u("assertionStack: (top to bottom)"));
- w->Indent();
- assertionStack.Print(w, this);
- w->Unindent();
- }
- w->Unindent();
- w->PrintEOL(_u("}"));
- w->Flush();
- }
- #endif
- // ----------------------------------------------------------------------
- // Program
- // ----------------------------------------------------------------------
- Program::Program(RegexFlags flags)
- : source(nullptr)
- , sourceLen(0)
- , flags(flags)
- , numGroups(0)
- , numLoops(0)
- {
- tag = ProgramTag::InstructionsTag;
- rep.insts.insts = nullptr;
- rep.insts.instsLen = 0;
- rep.insts.litbuf = nullptr;
- rep.insts.litbufLen = 0;
- rep.insts.scannersForSyncToLiterals = nullptr;
- }
- Program *Program::New(Recycler *recycler, RegexFlags flags)
- {
- return RecyclerNew(recycler, Program, flags);
- }
- Field(ScannerInfo *)*Program::CreateScannerArrayForSyncToLiterals(Recycler *const recycler)
- {
- Assert(tag == ProgramTag::InstructionsTag);
- Assert(!rep.insts.scannersForSyncToLiterals);
- Assert(recycler);
- return
- rep.insts.scannersForSyncToLiterals =
- RecyclerNewArrayZ(recycler, Field(ScannerInfo *), ScannersMixin::MaxNumSyncLiterals);
- }
- ScannerInfo *Program::AddScannerForSyncToLiterals(
- Recycler *const recycler,
- const int scannerIndex,
- const CharCount offset,
- const CharCount length,
- const bool isEquivClass)
- {
- Assert(tag == ProgramTag::InstructionsTag);
- Assert(rep.insts.scannersForSyncToLiterals);
- Assert(recycler);
- Assert(scannerIndex >= 0);
- Assert(scannerIndex < ScannersMixin::MaxNumSyncLiterals);
- Assert(!rep.insts.scannersForSyncToLiterals[scannerIndex]);
- return
- rep.insts.scannersForSyncToLiterals[scannerIndex] =
- RecyclerNewLeaf(recycler, ScannerInfo, offset, length, isEquivClass);
- }
- void Program::FreeBody(ArenaAllocator* rtAllocator)
- {
- if (tag != ProgramTag::InstructionsTag || !rep.insts.insts)
- {
- return;
- }
- Inst *inst = reinterpret_cast<Inst *>(PointerValue(rep.insts.insts));
- const auto instEnd = reinterpret_cast<Inst *>(reinterpret_cast<uint8 *>(inst) + rep.insts.instsLen);
- Assert(inst < instEnd);
- do
- {
- switch(inst->tag)
- {
- #define MBase(TagName, ClassName) \
- case Inst::InstTag::TagName: \
- { \
- const auto actualInst = static_cast<ClassName *>(inst); \
- actualInst->FreeBody(rtAllocator); \
- inst = actualInst + 1; \
- break; \
- }
- #define M(TagName) MBase(TagName, TagName##Inst)
- #define MTemplate(TagName, TemplateDeclaration, GenericClassName, SpecializedClassName) MBase(TagName, SpecializedClassName)
- #include "RegexOpCodes.h"
- #undef MBase
- #undef M
- #undef MTemplate
- default:
- Assert(false);
- __assume(false);
- }
- } while(inst < instEnd);
- Assert(inst == instEnd);
- #if DBG
- rep.insts.insts = nullptr;
- rep.insts.instsLen = 0;
- #endif
- }
- #if ENABLE_REGEX_CONFIG_OPTIONS
- void Program::Print(DebugWriter* w)
- {
- const bool isBaselineMode = Js::Configuration::Global.flags.BaselineMode;
- w->PrintEOL(_u("Program {"));
- w->Indent();
- w->PrintEOL(_u("source: %s"), PointerValue(source));
- w->Print(_u("flags: "));
- if ((flags & GlobalRegexFlag) != 0) w->Print(_u("global "));
- if ((flags & MultilineRegexFlag) != 0) w->Print(_u("multiline "));
- if ((flags & IgnoreCaseRegexFlag) != 0) w->Print(_u("ignorecase"));
- if ((flags & UnicodeRegexFlag) != 0) w->Print(_u("unicode"));
- if ((flags & StickyRegexFlag) != 0) w->Print(_u("sticky"));
- w->EOL();
- w->PrintEOL(_u("numGroups: %d"), numGroups);
- w->PrintEOL(_u("numLoops: %d"), numLoops);
- switch (tag)
- {
- case ProgramTag::BOIInstructionsTag:
- case ProgramTag::InstructionsTag:
- {
- w->PrintEOL(_u("instructions: {"));
- w->Indent();
- if (tag == ProgramTag::BOIInstructionsTag)
- {
- w->PrintEOL(_u(" BOITest(hardFail: true)"));
- }
- uint8* instsLim = rep.insts.insts + rep.insts.instsLen;
- uint8* curr = rep.insts.insts;
- int i = 0;
- while (curr != instsLim)
- {
- const Inst *inst = (const Inst*)curr;
- switch (inst->tag)
- {
- #define MBase(TagName, ClassName) \
- case Inst::InstTag::TagName: \
- { \
- const ClassName *actualInst = static_cast<const ClassName *>(inst); \
- curr += actualInst->Print(w, (Label)(isBaselineMode ? i++ : curr - rep.insts.insts), rep.insts.litbuf); \
- break; \
- }
- #define M(TagName) MBase(TagName, TagName##Inst)
- #define MTemplate(TagName, TemplateDeclaration, GenericClassName, SpecializedClassName) MBase(TagName, SpecializedClassName)
- #include "RegexOpCodes.h"
- #undef MBase
- #undef M
- #undef MTemplate
- default:
- Assert(false);
- __assume(false);
- }
- }
- w->Unindent();
- w->PrintEOL(_u("}"));
- }
- break;
- case ProgramTag::SingleCharTag:
- w->Print(_u("special form: <match single char "));
- w->PrintQuotedChar(rep.singleChar.c);
- w->PrintEOL(_u(">"));
- break;
- case ProgramTag::BoundedWordTag:
- w->PrintEOL(_u("special form: <match bounded word>"));
- break;
- case ProgramTag::LeadingTrailingSpacesTag:
- w->PrintEOL(_u("special form: <match leading/trailing spaces: minBegin=%d minEnd=%d>"),
- rep.leadingTrailingSpaces.beginMinMatch, rep.leadingTrailingSpaces.endMinMatch);
- break;
- case ProgramTag::OctoquadTag:
- w->Print(_u("special form: <octoquad "));
- rep.octoquad.matcher->Print(w);
- w->PrintEOL(_u(">"));
- break;
- }
- w->Unindent();
- w->PrintEOL(_u("}"));
- }
- #endif
- // Template parameter here is the max number of cases
- template void UnifiedRegex::SwitchMixin<2>::AddCase(char16, Label);
- template void UnifiedRegex::SwitchMixin<4>::AddCase(char16, Label);
- template void UnifiedRegex::SwitchMixin<8>::AddCase(char16, Label);
- template void UnifiedRegex::SwitchMixin<16>::AddCase(char16, Label);
- template void UnifiedRegex::SwitchMixin<24>::AddCase(char16, Label);
- #define M(...)
- #define MTemplate(TagName, TemplateDeclaration, GenericClassName, SpecializedClassName) template struct SpecializedClassName;
- #include "RegexOpCodes.h"
- #undef M
- #undef MTemplate
- }
|