123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115211621172118211921202121212221232124212521262127212821292130213121322133213421352136213721382139214021412142214321442145214621472148214921502151215221532154215521562157215821592160216121622163216421652166216721682169217021712172217321742175217621772178217921802181218221832184218521862187218821892190219121922193219421952196219721982199220022012202220322042205220622072208220922102211221222132214221522162217221822192220222122222223222422252226222722282229223022312232223322342235223622372238223922402241224222432244224522462247224822492250225122522253225422552256225722582259226022612262226322642265226622672268226922702271227222732274227522762277227822792280228122822283228422852286228722882289229022912292229322942295229622972298229923002301230223032304230523062307230823092310231123122313231423152316231723182319232023212322232323242325232623272328232923302331233223332334233523362337233823392340234123422343234423452346234723482349235023512352235323542355235623572358235923602361236223632364236523662367236823692370237123722373237423752376237723782379238023812382238323842385238623872388238923902391239223932394239523962397239823992400240124022403240424052406240724082409241024112412241324142415241624172418241924202421242224232424242524262427242824292430243124322433243424352436243724382439244024412442244324442445244624472448244924502451245224532454245524562457245824592460246124622463246424652466246724682469247024712472247324742475247624772478247924802481248224832484248524862487248824892490249124922493249424952496249724982499250025012502250325042505250625072508250925102511251225132514251525162517251825192520252125222523252425252526252725282529253025312532253325342535253625372538253925402541254225432544254525462547254825492550255125522553255425552556255725582559256025612562256325642565256625672568256925702571257225732574257525762577257825792580258125822583258425852586258725882589259025912592259325942595259625972598259926002601260226032604260526062607260826092610261126122613261426152616261726182619262026212622262326242625262626272628262926302631263226332634263526362637263826392640264126422643264426452646264726482649265026512652265326542655265626572658265926602661266226632664266526662667266826692670267126722673267426752676267726782679268026812682268326842685268626872688268926902691269226932694269526962697269826992700270127022703270427052706270727082709271027112712271327142715271627172718271927202721272227232724272527262727272827292730273127322733273427352736273727382739274027412742274327442745274627472748274927502751275227532754275527562757275827592760276127622763276427652766276727682769277027712772277327742775277627772778277927802781278227832784278527862787278827892790279127922793279427952796279727982799280028012802280328042805280628072808280928102811281228132814281528162817281828192820282128222823282428252826282728282829283028312832283328342835283628372838283928402841284228432844284528462847284828492850285128522853285428552856285728582859286028612862286328642865286628672868286928702871287228732874287528762877287828792880288128822883288428852886288728882889289028912892289328942895289628972898289929002901290229032904290529062907290829092910291129122913291429152916291729182919292029212922292329242925292629272928292929302931293229332934293529362937293829392940294129422943294429452946294729482949295029512952295329542955295629572958295929602961296229632964296529662967296829692970297129722973297429752976297729782979298029812982298329842985298629872988298929902991299229932994299529962997299829993000300130023003300430053006300730083009301030113012301330143015301630173018301930203021302230233024302530263027302830293030303130323033303430353036303730383039304030413042304330443045304630473048304930503051305230533054305530563057305830593060306130623063306430653066306730683069307030713072307330743075307630773078307930803081308230833084308530863087308830893090309130923093309430953096309730983099310031013102310331043105310631073108310931103111311231133114311531163117 |
- /*
- *
- * Copyright (c) 2004
- * John Maddock
- *
- * Use, modification and distribution are subject to the
- * Boost Software License, Version 1.0. (See accompanying file
- * LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
- *
- */
- /*
- * LOCATION: see http://www.boost.org for most recent version.
- * FILE basic_regex_parser.cpp
- * VERSION see <boost/version.hpp>
- * DESCRIPTION: Declares template class basic_regex_parser.
- */
- #ifndef BOOST_REGEX_V5_BASIC_REGEX_PARSER_HPP
- #define BOOST_REGEX_V5_BASIC_REGEX_PARSER_HPP
- namespace boost{
- namespace BOOST_REGEX_DETAIL_NS{
- #ifdef BOOST_REGEX_MSVC
- #pragma warning(push)
- #pragma warning(disable:4244 4459)
- #if BOOST_REGEX_MSVC < 1910
- #pragma warning(disable:4800)
- #endif
- #endif
- inline std::intmax_t umax(std::integral_constant<bool, false> const&)
- {
- // Get out clause here, just in case numeric_limits is unspecialized:
- return std::numeric_limits<std::intmax_t>::is_specialized ? (std::numeric_limits<std::intmax_t>::max)() : INT_MAX;
- }
- inline std::intmax_t umax(std::integral_constant<bool, true> const&)
- {
- return (std::numeric_limits<std::size_t>::max)();
- }
- inline std::intmax_t umax()
- {
- return umax(std::integral_constant<bool, std::numeric_limits<std::intmax_t>::digits >= std::numeric_limits<std::size_t>::digits>());
- }
- template <class charT, class traits>
- class basic_regex_parser : public basic_regex_creator<charT, traits>
- {
- public:
- basic_regex_parser(regex_data<charT, traits>* data);
- void parse(const charT* p1, const charT* p2, unsigned flags);
- void fail(regex_constants::error_type error_code, std::ptrdiff_t position);
- void fail(regex_constants::error_type error_code, std::ptrdiff_t position, std::string message, std::ptrdiff_t start_pos);
- void fail(regex_constants::error_type error_code, std::ptrdiff_t position, const std::string& message)
- {
- fail(error_code, position, message, position);
- }
- bool parse_all();
- bool parse_basic();
- bool parse_extended();
- bool parse_literal();
- bool parse_open_paren();
- bool parse_basic_escape();
- bool parse_extended_escape();
- bool parse_match_any();
- bool parse_repeat(std::size_t low = 0, std::size_t high = (std::numeric_limits<std::size_t>::max)());
- bool parse_repeat_range(bool isbasic);
- bool parse_alt();
- bool parse_set();
- bool parse_backref();
- void parse_set_literal(basic_char_set<charT, traits>& char_set);
- bool parse_inner_set(basic_char_set<charT, traits>& char_set);
- bool parse_QE();
- bool parse_perl_extension();
- bool parse_perl_verb();
- bool match_verb(const char*);
- bool add_emacs_code(bool negate);
- bool unwind_alts(std::ptrdiff_t last_paren_start);
- digraph<charT> get_next_set_literal(basic_char_set<charT, traits>& char_set);
- charT unescape_character();
- regex_constants::syntax_option_type parse_options();
- private:
- typedef bool (basic_regex_parser::*parser_proc_type)();
- typedef typename traits::string_type string_type;
- typedef typename traits::char_class_type char_class_type;
- parser_proc_type m_parser_proc; // the main parser to use
- const charT* m_base; // the start of the string being parsed
- const charT* m_end; // the end of the string being parsed
- const charT* m_position; // our current parser position
- unsigned m_mark_count; // how many sub-expressions we have
- int m_mark_reset; // used to indicate that we're inside a (?|...) block.
- unsigned m_max_mark; // largest mark count seen inside a (?|...) block.
- std::ptrdiff_t m_paren_start; // where the last seen ')' began (where repeats are inserted).
- std::ptrdiff_t m_alt_insert_point; // where to insert the next alternative
- bool m_has_case_change; // true if somewhere in the current block the case has changed
- unsigned m_recursion_count; // How many times we've called parse_all.
- #if defined(BOOST_REGEX_MSVC) && defined(_M_IX86)
- // This is an ugly warning suppression workaround (for warnings *inside* std::vector
- // that can not otherwise be suppressed)...
- static_assert(sizeof(long) >= sizeof(void*), "Long isn't long enough!");
- std::vector<long> m_alt_jumps; // list of alternative in the current scope.
- #else
- std::vector<std::ptrdiff_t> m_alt_jumps; // list of alternative in the current scope.
- #endif
- basic_regex_parser& operator=(const basic_regex_parser&);
- basic_regex_parser(const basic_regex_parser&);
- };
- template <class charT, class traits>
- basic_regex_parser<charT, traits>::basic_regex_parser(regex_data<charT, traits>* data)
- : basic_regex_creator<charT, traits>(data), m_parser_proc(), m_base(0), m_end(0), m_position(0),
- m_mark_count(0), m_mark_reset(-1), m_max_mark(0), m_paren_start(0), m_alt_insert_point(0), m_has_case_change(false), m_recursion_count(0)
- {
- }
- template <class charT, class traits>
- void basic_regex_parser<charT, traits>::parse(const charT* p1, const charT* p2, unsigned l_flags)
- {
- // pass l_flags on to base class:
- this->init(l_flags);
- // set up pointers:
- m_position = m_base = p1;
- m_end = p2;
- // empty strings are errors:
- if((p1 == p2) &&
- (
- ((l_flags & regbase::main_option_type) != regbase::perl_syntax_group)
- || (l_flags & regbase::no_empty_expressions)
- )
- )
- {
- fail(regex_constants::error_empty, 0);
- return;
- }
- // select which parser to use:
- switch(l_flags & regbase::main_option_type)
- {
- case regbase::perl_syntax_group:
- {
- m_parser_proc = &basic_regex_parser<charT, traits>::parse_extended;
- //
- // Add a leading paren with index zero to give recursions a target:
- //
- re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_startmark, sizeof(re_brace)));
- br->index = 0;
- br->icase = this->flags() & regbase::icase;
- break;
- }
- case regbase::basic_syntax_group:
- m_parser_proc = &basic_regex_parser<charT, traits>::parse_basic;
- break;
- case regbase::literal:
- m_parser_proc = &basic_regex_parser<charT, traits>::parse_literal;
- break;
- default:
- // Oops, someone has managed to set more than one of the main option flags,
- // so this must be an error:
- fail(regex_constants::error_unknown, 0, "An invalid combination of regular expression syntax flags was used.");
- return;
- }
- // parse all our characters:
- bool result = parse_all();
- //
- // Unwind our alternatives:
- //
- unwind_alts(-1);
- // reset l_flags as a global scope (?imsx) may have altered them:
- this->flags(l_flags);
- // if we haven't gobbled up all the characters then we must
- // have had an unexpected ')' :
- if(!result)
- {
- fail(regex_constants::error_paren, std::distance(m_base, m_position), "Found a closing ) with no corresponding opening parenthesis.");
- return;
- }
- // if an error has been set then give up now:
- if(this->m_pdata->m_status)
- return;
- // fill in our sub-expression count:
- this->m_pdata->m_mark_count = 1u + (std::size_t)m_mark_count;
- this->finalize(p1, p2);
- }
- template <class charT, class traits>
- void basic_regex_parser<charT, traits>::fail(regex_constants::error_type error_code, std::ptrdiff_t position)
- {
- // get the error message:
- std::string message = this->m_pdata->m_ptraits->error_string(error_code);
- fail(error_code, position, message);
- }
- template <class charT, class traits>
- void basic_regex_parser<charT, traits>::fail(regex_constants::error_type error_code, std::ptrdiff_t position, std::string message, std::ptrdiff_t start_pos)
- {
- if(0 == this->m_pdata->m_status) // update the error code if not already set
- this->m_pdata->m_status = error_code;
- m_position = m_end; // don't bother parsing anything else
- //
- // Augment error message with the regular expression text:
- //
- if(start_pos == position)
- start_pos = (std::max)(static_cast<std::ptrdiff_t>(0), position - static_cast<std::ptrdiff_t>(10));
- std::ptrdiff_t end_pos = (std::min)(position + static_cast<std::ptrdiff_t>(10), static_cast<std::ptrdiff_t>(m_end - m_base));
- if(error_code != regex_constants::error_empty)
- {
- if((start_pos != 0) || (end_pos != (m_end - m_base)))
- message += " The error occurred while parsing the regular expression fragment: '";
- else
- message += " The error occurred while parsing the regular expression: '";
- if(start_pos != end_pos)
- {
- message += std::string(m_base + start_pos, m_base + position);
- message += ">>>HERE>>>";
- message += std::string(m_base + position, m_base + end_pos);
- }
- message += "'.";
- }
- #ifndef BOOST_NO_EXCEPTIONS
- if(0 == (this->flags() & regex_constants::no_except))
- {
- boost::regex_error e(message, error_code, position);
- e.raise();
- }
- #else
- (void)position; // suppress warnings.
- #endif
- }
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::parse_all()
- {
- if (++m_recursion_count > 400)
- {
- // exceeded internal limits
- fail(boost::regex_constants::error_complexity, m_position - m_base, "Exceeded nested brace limit.");
- }
- bool result = true;
- while(result && (m_position != m_end))
- {
- result = (this->*m_parser_proc)();
- }
- --m_recursion_count;
- return result;
- }
- #ifdef BOOST_REGEX_MSVC
- #pragma warning(push)
- #pragma warning(disable:4702)
- #endif
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::parse_basic()
- {
- switch(this->m_traits.syntax_type(*m_position))
- {
- case regex_constants::syntax_escape:
- return parse_basic_escape();
- case regex_constants::syntax_dot:
- return parse_match_any();
- case regex_constants::syntax_caret:
- ++m_position;
- this->append_state(syntax_element_start_line);
- break;
- case regex_constants::syntax_dollar:
- ++m_position;
- this->append_state(syntax_element_end_line);
- break;
- case regex_constants::syntax_star:
- if(!(this->m_last_state) || (this->m_last_state->type == syntax_element_start_line))
- return parse_literal();
- else
- {
- ++m_position;
- return parse_repeat();
- }
- case regex_constants::syntax_plus:
- if(!(this->m_last_state) || (this->m_last_state->type == syntax_element_start_line) || !(this->flags() & regbase::emacs_ex))
- return parse_literal();
- else
- {
- ++m_position;
- return parse_repeat(1);
- }
- case regex_constants::syntax_question:
- if(!(this->m_last_state) || (this->m_last_state->type == syntax_element_start_line) || !(this->flags() & regbase::emacs_ex))
- return parse_literal();
- else
- {
- ++m_position;
- return parse_repeat(0, 1);
- }
- case regex_constants::syntax_open_set:
- return parse_set();
- case regex_constants::syntax_newline:
- if(this->flags() & regbase::newline_alt)
- return parse_alt();
- else
- return parse_literal();
- default:
- return parse_literal();
- }
- return true;
- }
- #ifdef BOOST_REGEX_MSVC
- # pragma warning(push)
- #if BOOST_REGEX_MSVC >= 1800
- #pragma warning(disable:26812)
- #endif
- #endif
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::parse_extended()
- {
- bool result = true;
- switch(this->m_traits.syntax_type(*m_position))
- {
- case regex_constants::syntax_open_mark:
- return parse_open_paren();
- case regex_constants::syntax_close_mark:
- return false;
- case regex_constants::syntax_escape:
- return parse_extended_escape();
- case regex_constants::syntax_dot:
- return parse_match_any();
- case regex_constants::syntax_caret:
- ++m_position;
- this->append_state(
- (this->flags() & regex_constants::no_mod_m ? syntax_element_buffer_start : syntax_element_start_line));
- break;
- case regex_constants::syntax_dollar:
- ++m_position;
- this->append_state(
- (this->flags() & regex_constants::no_mod_m ? syntax_element_buffer_end : syntax_element_end_line));
- break;
- case regex_constants::syntax_star:
- if(m_position == this->m_base)
- {
- fail(regex_constants::error_badrepeat, 0, "The repeat operator \"*\" cannot start a regular expression.");
- return false;
- }
- ++m_position;
- return parse_repeat();
- case regex_constants::syntax_question:
- if(m_position == this->m_base)
- {
- fail(regex_constants::error_badrepeat, 0, "The repeat operator \"?\" cannot start a regular expression.");
- return false;
- }
- ++m_position;
- return parse_repeat(0,1);
- case regex_constants::syntax_plus:
- if(m_position == this->m_base)
- {
- fail(regex_constants::error_badrepeat, 0, "The repeat operator \"+\" cannot start a regular expression.");
- return false;
- }
- ++m_position;
- return parse_repeat(1);
- case regex_constants::syntax_open_brace:
- ++m_position;
- return parse_repeat_range(false);
- case regex_constants::syntax_close_brace:
- if((this->flags() & regbase::no_perl_ex) == regbase::no_perl_ex)
- {
- fail(regex_constants::error_brace, this->m_position - this->m_base, "Found a closing repetition operator } with no corresponding {.");
- return false;
- }
- result = parse_literal();
- break;
- case regex_constants::syntax_or:
- return parse_alt();
- case regex_constants::syntax_open_set:
- return parse_set();
- case regex_constants::syntax_newline:
- if(this->flags() & regbase::newline_alt)
- return parse_alt();
- else
- return parse_literal();
- case regex_constants::syntax_hash:
- //
- // If we have a mod_x flag set, then skip until
- // we get to a newline character:
- //
- if((this->flags()
- & (regbase::no_perl_ex|regbase::mod_x))
- == regbase::mod_x)
- {
- while((m_position != m_end) && !is_separator(*m_position++)){}
- return true;
- }
- BOOST_REGEX_FALLTHROUGH;
- default:
- result = parse_literal();
- break;
- }
- return result;
- }
- #ifdef BOOST_REGEX_MSVC
- # pragma warning(pop)
- #endif
- #ifdef BOOST_REGEX_MSVC
- #pragma warning(pop)
- #endif
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::parse_literal()
- {
- // append this as a literal provided it's not a space character
- // or the perl option regbase::mod_x is not set:
- if(
- ((this->flags()
- & (regbase::main_option_type|regbase::mod_x|regbase::no_perl_ex))
- != regbase::mod_x)
- || !this->m_traits.isctype(*m_position, this->m_mask_space))
- this->append_literal(*m_position);
- ++m_position;
- return true;
- }
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::parse_open_paren()
- {
- //
- // skip the '(' and error check:
- //
- if(++m_position == m_end)
- {
- fail(regex_constants::error_paren, m_position - m_base);
- return false;
- }
- //
- // begin by checking for a perl-style (?...) extension:
- //
- if(
- ((this->flags() & (regbase::main_option_type | regbase::no_perl_ex)) == 0)
- || ((this->flags() & (regbase::main_option_type | regbase::emacs_ex)) == (regbase::basic_syntax_group|regbase::emacs_ex))
- )
- {
- if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_question)
- return parse_perl_extension();
- if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_star)
- return parse_perl_verb();
- }
- //
- // update our mark count, and append the required state:
- //
- unsigned markid = 0;
- if(0 == (this->flags() & regbase::nosubs))
- {
- markid = ++m_mark_count;
- if(this->flags() & regbase::save_subexpression_location)
- this->m_pdata->m_subs.push_back(std::pair<std::size_t, std::size_t>(std::distance(m_base, m_position) - 1, 0));
- }
- re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_startmark, sizeof(re_brace)));
- pb->index = markid;
- pb->icase = this->flags() & regbase::icase;
- std::ptrdiff_t last_paren_start = this->getoffset(pb);
- // back up insertion point for alternations, and set new point:
- std::ptrdiff_t last_alt_point = m_alt_insert_point;
- this->m_pdata->m_data.align();
- m_alt_insert_point = this->m_pdata->m_data.size();
- //
- // back up the current flags in case we have a nested (?imsx) group:
- //
- regex_constants::syntax_option_type opts = this->flags();
- bool old_case_change = m_has_case_change;
- m_has_case_change = false; // no changes to this scope as yet...
- //
- // Back up branch reset data in case we have a nested (?|...)
- //
- int mark_reset = m_mark_reset;
- m_mark_reset = -1;
- //
- // now recursively add more states, this will terminate when we get to a
- // matching ')' :
- //
- parse_all();
- //
- // Unwind pushed alternatives:
- //
- if(0 == unwind_alts(last_paren_start))
- return false;
- //
- // restore flags:
- //
- if(m_has_case_change)
- {
- // the case has changed in one or more of the alternatives
- // within the scoped (...) block: we have to add a state
- // to reset the case sensitivity:
- static_cast<re_case*>(
- this->append_state(syntax_element_toggle_case, sizeof(re_case))
- )->icase = opts & regbase::icase;
- }
- this->flags(opts);
- m_has_case_change = old_case_change;
- //
- // restore branch reset:
- //
- m_mark_reset = mark_reset;
- //
- // we either have a ')' or we have run out of characters prematurely:
- //
- if(m_position == m_end)
- {
- this->fail(regex_constants::error_paren, std::distance(m_base, m_end));
- return false;
- }
- if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)
- return false;
- if(markid && (this->flags() & regbase::save_subexpression_location))
- this->m_pdata->m_subs.at(markid - 1).second = std::distance(m_base, m_position);
- ++m_position;
- //
- // append closing parenthesis state:
- //
- pb = static_cast<re_brace*>(this->append_state(syntax_element_endmark, sizeof(re_brace)));
- pb->index = markid;
- pb->icase = this->flags() & regbase::icase;
- this->m_paren_start = last_paren_start;
- //
- // restore the alternate insertion point:
- //
- this->m_alt_insert_point = last_alt_point;
- //
- // allow backrefs to this mark:
- //
- if(markid > 0)
- this->m_backrefs.set(markid);
- return true;
- }
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::parse_basic_escape()
- {
- if(++m_position == m_end)
- {
- fail(regex_constants::error_paren, m_position - m_base);
- return false;
- }
- bool result = true;
- switch(this->m_traits.escape_syntax_type(*m_position))
- {
- case regex_constants::syntax_open_mark:
- return parse_open_paren();
- case regex_constants::syntax_close_mark:
- return false;
- case regex_constants::syntax_plus:
- if(this->flags() & regex_constants::bk_plus_qm)
- {
- ++m_position;
- return parse_repeat(1);
- }
- else
- return parse_literal();
- case regex_constants::syntax_question:
- if(this->flags() & regex_constants::bk_plus_qm)
- {
- ++m_position;
- return parse_repeat(0, 1);
- }
- else
- return parse_literal();
- case regex_constants::syntax_open_brace:
- if(this->flags() & regbase::no_intervals)
- return parse_literal();
- ++m_position;
- return parse_repeat_range(true);
- case regex_constants::syntax_close_brace:
- if(this->flags() & regbase::no_intervals)
- return parse_literal();
- fail(regex_constants::error_brace, this->m_position - this->m_base, "Found a closing repetition operator } with no corresponding {.");
- return false;
- case regex_constants::syntax_or:
- if(this->flags() & regbase::bk_vbar)
- return parse_alt();
- else
- result = parse_literal();
- break;
- case regex_constants::syntax_digit:
- return parse_backref();
- case regex_constants::escape_type_start_buffer:
- if(this->flags() & regbase::emacs_ex)
- {
- ++m_position;
- this->append_state(syntax_element_buffer_start);
- }
- else
- result = parse_literal();
- break;
- case regex_constants::escape_type_end_buffer:
- if(this->flags() & regbase::emacs_ex)
- {
- ++m_position;
- this->append_state(syntax_element_buffer_end);
- }
- else
- result = parse_literal();
- break;
- case regex_constants::escape_type_word_assert:
- if(this->flags() & regbase::emacs_ex)
- {
- ++m_position;
- this->append_state(syntax_element_word_boundary);
- }
- else
- result = parse_literal();
- break;
- case regex_constants::escape_type_not_word_assert:
- if(this->flags() & regbase::emacs_ex)
- {
- ++m_position;
- this->append_state(syntax_element_within_word);
- }
- else
- result = parse_literal();
- break;
- case regex_constants::escape_type_left_word:
- if(this->flags() & regbase::emacs_ex)
- {
- ++m_position;
- this->append_state(syntax_element_word_start);
- }
- else
- result = parse_literal();
- break;
- case regex_constants::escape_type_right_word:
- if(this->flags() & regbase::emacs_ex)
- {
- ++m_position;
- this->append_state(syntax_element_word_end);
- }
- else
- result = parse_literal();
- break;
- default:
- if(this->flags() & regbase::emacs_ex)
- {
- bool negate = true;
- switch(*m_position)
- {
- case 'w':
- negate = false;
- BOOST_REGEX_FALLTHROUGH;
- case 'W':
- {
- basic_char_set<charT, traits> char_set;
- if(negate)
- char_set.negate();
- char_set.add_class(this->m_word_mask);
- if(0 == this->append_set(char_set))
- {
- fail(regex_constants::error_ctype, m_position - m_base);
- return false;
- }
- ++m_position;
- return true;
- }
- case 's':
- negate = false;
- BOOST_REGEX_FALLTHROUGH;
- case 'S':
- return add_emacs_code(negate);
- case 'c':
- case 'C':
- // not supported yet:
- fail(regex_constants::error_escape, m_position - m_base, "The \\c and \\C escape sequences are not supported by POSIX basic regular expressions: try the Perl syntax instead.");
- return false;
- default:
- break;
- }
- }
- result = parse_literal();
- break;
- }
- return result;
- }
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::parse_extended_escape()
- {
- ++m_position;
- if(m_position == m_end)
- {
- fail(regex_constants::error_escape, m_position - m_base, "Incomplete escape sequence found.");
- return false;
- }
- bool negate = false; // in case this is a character class escape: \w \d etc
- switch(this->m_traits.escape_syntax_type(*m_position))
- {
- case regex_constants::escape_type_not_class:
- negate = true;
- BOOST_REGEX_FALLTHROUGH;
- case regex_constants::escape_type_class:
- {
- escape_type_class_jump:
- typedef typename traits::char_class_type m_type;
- m_type m = this->m_traits.lookup_classname(m_position, m_position+1);
- if(m != 0)
- {
- basic_char_set<charT, traits> char_set;
- if(negate)
- char_set.negate();
- char_set.add_class(m);
- if(0 == this->append_set(char_set))
- {
- fail(regex_constants::error_ctype, m_position - m_base);
- return false;
- }
- ++m_position;
- return true;
- }
- //
- // not a class, just a regular unknown escape:
- //
- this->append_literal(unescape_character());
- break;
- }
- case regex_constants::syntax_digit:
- return parse_backref();
- case regex_constants::escape_type_left_word:
- ++m_position;
- this->append_state(syntax_element_word_start);
- break;
- case regex_constants::escape_type_right_word:
- ++m_position;
- this->append_state(syntax_element_word_end);
- break;
- case regex_constants::escape_type_start_buffer:
- ++m_position;
- this->append_state(syntax_element_buffer_start);
- break;
- case regex_constants::escape_type_end_buffer:
- ++m_position;
- this->append_state(syntax_element_buffer_end);
- break;
- case regex_constants::escape_type_word_assert:
- ++m_position;
- this->append_state(syntax_element_word_boundary);
- break;
- case regex_constants::escape_type_not_word_assert:
- ++m_position;
- this->append_state(syntax_element_within_word);
- break;
- case regex_constants::escape_type_Z:
- ++m_position;
- this->append_state(syntax_element_soft_buffer_end);
- break;
- case regex_constants::escape_type_Q:
- return parse_QE();
- case regex_constants::escape_type_C:
- return parse_match_any();
- case regex_constants::escape_type_X:
- ++m_position;
- this->append_state(syntax_element_combining);
- break;
- case regex_constants::escape_type_G:
- ++m_position;
- this->append_state(syntax_element_restart_continue);
- break;
- case regex_constants::escape_type_not_property:
- negate = true;
- BOOST_REGEX_FALLTHROUGH;
- case regex_constants::escape_type_property:
- {
- ++m_position;
- char_class_type m;
- if(m_position == m_end)
- {
- fail(regex_constants::error_escape, m_position - m_base, "Incomplete property escape found.");
- return false;
- }
- // maybe have \p{ddd}
- if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_open_brace)
- {
- const charT* base = m_position;
- // skip forward until we find enclosing brace:
- while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_brace))
- ++m_position;
- if(m_position == m_end)
- {
- fail(regex_constants::error_escape, m_position - m_base, "Closing } missing from property escape sequence.");
- return false;
- }
- m = this->m_traits.lookup_classname(++base, m_position++);
- }
- else
- {
- m = this->m_traits.lookup_classname(m_position, m_position+1);
- ++m_position;
- }
- if(m != 0)
- {
- basic_char_set<charT, traits> char_set;
- if(negate)
- char_set.negate();
- char_set.add_class(m);
- if(0 == this->append_set(char_set))
- {
- fail(regex_constants::error_ctype, m_position - m_base);
- return false;
- }
- return true;
- }
- fail(regex_constants::error_ctype, m_position - m_base, "Escape sequence was neither a valid property nor a valid character class name.");
- return false;
- }
- case regex_constants::escape_type_reset_start_mark:
- if(0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex)))
- {
- re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_startmark, sizeof(re_brace)));
- pb->index = -5;
- pb->icase = this->flags() & regbase::icase;
- this->m_pdata->m_data.align();
- ++m_position;
- return true;
- }
- goto escape_type_class_jump;
- case regex_constants::escape_type_line_ending:
- if(0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex)))
- {
- const charT* e = get_escape_R_string<charT>();
- const charT* old_position = m_position;
- const charT* old_end = m_end;
- const charT* old_base = m_base;
- m_position = e;
- m_base = e;
- m_end = e + traits::length(e);
- bool r = parse_all();
- m_position = ++old_position;
- m_end = old_end;
- m_base = old_base;
- return r;
- }
- goto escape_type_class_jump;
- case regex_constants::escape_type_extended_backref:
- if(0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex)))
- {
- bool have_brace = false;
- bool negative = false;
- static const char incomplete_message[] = "Incomplete \\g escape found.";
- if(++m_position == m_end)
- {
- fail(regex_constants::error_escape, m_position - m_base, incomplete_message);
- return false;
- }
- // maybe have \g{ddd}
- regex_constants::syntax_type syn = this->m_traits.syntax_type(*m_position);
- regex_constants::syntax_type syn_end = 0;
- if((syn == regex_constants::syntax_open_brace)
- || (syn == regex_constants::escape_type_left_word)
- || (syn == regex_constants::escape_type_end_buffer))
- {
- if(++m_position == m_end)
- {
- fail(regex_constants::error_escape, m_position - m_base, incomplete_message);
- return false;
- }
- have_brace = true;
- switch(syn)
- {
- case regex_constants::syntax_open_brace:
- syn_end = regex_constants::syntax_close_brace;
- break;
- case regex_constants::escape_type_left_word:
- syn_end = regex_constants::escape_type_right_word;
- break;
- default:
- syn_end = regex_constants::escape_type_end_buffer;
- break;
- }
- }
- negative = (*m_position == static_cast<charT>('-'));
- if((negative) && (++m_position == m_end))
- {
- fail(regex_constants::error_escape, m_position - m_base, incomplete_message);
- return false;
- }
- const charT* pc = m_position;
- std::intmax_t i = this->m_traits.toi(pc, m_end, 10);
- if((i < 0) && syn_end)
- {
- // Check for a named capture, get the leftmost one if there is more than one:
- const charT* base = m_position;
- while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != syn_end))
- {
- ++m_position;
- }
- i = hash_value_from_capture_name(base, m_position);
- pc = m_position;
- }
- if(negative)
- i = 1 + (static_cast<std::intmax_t>(m_mark_count) - i);
- if(((i < hash_value_mask) && (i > 0) && (this->m_backrefs.test((std::size_t)i))) || ((i >= hash_value_mask) && (this->m_pdata->get_id((int)i) > 0) && (this->m_backrefs.test(this->m_pdata->get_id((int)i)))))
- {
- m_position = pc;
- re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_backref, sizeof(re_brace)));
- pb->index = (int)i;
- pb->icase = this->flags() & regbase::icase;
- }
- else
- {
- fail(regex_constants::error_backref, m_position - m_base);
- return false;
- }
- m_position = pc;
- if(have_brace)
- {
- if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != syn_end))
- {
- fail(regex_constants::error_escape, m_position - m_base, incomplete_message);
- return false;
- }
- ++m_position;
- }
- return true;
- }
- goto escape_type_class_jump;
- case regex_constants::escape_type_control_v:
- if(0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex)))
- goto escape_type_class_jump;
- BOOST_REGEX_FALLTHROUGH;
- default:
- this->append_literal(unescape_character());
- break;
- }
- return true;
- }
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::parse_match_any()
- {
- //
- // we have a '.' that can match any character:
- //
- ++m_position;
- static_cast<re_dot*>(
- this->append_state(syntax_element_wild, sizeof(re_dot))
- )->mask = static_cast<unsigned char>(this->flags() & regbase::no_mod_s
- ? BOOST_REGEX_DETAIL_NS::force_not_newline
- : this->flags() & regbase::mod_s ?
- BOOST_REGEX_DETAIL_NS::force_newline : BOOST_REGEX_DETAIL_NS::dont_care);
- return true;
- }
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::parse_repeat(std::size_t low, std::size_t high)
- {
- bool greedy = true;
- bool possessive = false;
- std::size_t insert_point;
- //
- // when we get to here we may have a non-greedy ? mark still to come:
- //
- if((m_position != m_end)
- && (
- (0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex)))
- || ((regbase::basic_syntax_group|regbase::emacs_ex) == (this->flags() & (regbase::main_option_type | regbase::emacs_ex)))
- )
- )
- {
- // OK we have a perl or emacs regex, check for a '?':
- if ((this->flags() & (regbase::main_option_type | regbase::mod_x | regbase::no_perl_ex)) == regbase::mod_x)
- {
- // whitespace skip:
- while ((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space))
- ++m_position;
- }
- if((m_position != m_end) && (this->m_traits.syntax_type(*m_position) == regex_constants::syntax_question))
- {
- greedy = false;
- ++m_position;
- }
- // for perl regexes only check for possessive ++ repeats.
- if((m_position != m_end)
- && (0 == (this->flags() & regbase::main_option_type))
- && (this->m_traits.syntax_type(*m_position) == regex_constants::syntax_plus))
- {
- possessive = true;
- ++m_position;
- }
- }
- if(0 == this->m_last_state)
- {
- fail(regex_constants::error_badrepeat, std::distance(m_base, m_position), "Nothing to repeat.");
- return false;
- }
- if(this->m_last_state->type == syntax_element_endmark)
- {
- // insert a repeat before the '(' matching the last ')':
- insert_point = this->m_paren_start;
- }
- else if((this->m_last_state->type == syntax_element_literal) && (static_cast<re_literal*>(this->m_last_state)->length > 1))
- {
- // the last state was a literal with more than one character, split it in two:
- re_literal* lit = static_cast<re_literal*>(this->m_last_state);
- charT c = (static_cast<charT*>(static_cast<void*>(lit+1)))[lit->length - 1];
- lit->length -= 1;
- // now append new state:
- lit = static_cast<re_literal*>(this->append_state(syntax_element_literal, sizeof(re_literal) + sizeof(charT)));
- lit->length = 1;
- (static_cast<charT*>(static_cast<void*>(lit+1)))[0] = c;
- insert_point = this->getoffset(this->m_last_state);
- }
- else
- {
- // repeat the last state whatever it was, need to add some error checking here:
- switch(this->m_last_state->type)
- {
- case syntax_element_start_line:
- case syntax_element_end_line:
- case syntax_element_word_boundary:
- case syntax_element_within_word:
- case syntax_element_word_start:
- case syntax_element_word_end:
- case syntax_element_buffer_start:
- case syntax_element_buffer_end:
- case syntax_element_alt:
- case syntax_element_soft_buffer_end:
- case syntax_element_restart_continue:
- case syntax_element_jump:
- case syntax_element_startmark:
- case syntax_element_backstep:
- // can't legally repeat any of the above:
- fail(regex_constants::error_badrepeat, m_position - m_base);
- return false;
- default:
- // do nothing...
- break;
- }
- insert_point = this->getoffset(this->m_last_state);
- }
- //
- // OK we now know what to repeat, so insert the repeat around it:
- //
- re_repeat* rep = static_cast<re_repeat*>(this->insert_state(insert_point, syntax_element_rep, re_repeater_size));
- rep->min = low;
- rep->max = high;
- rep->greedy = greedy;
- rep->leading = false;
- // store our repeater position for later:
- std::ptrdiff_t rep_off = this->getoffset(rep);
- // and append a back jump to the repeat:
- re_jump* jmp = static_cast<re_jump*>(this->append_state(syntax_element_jump, sizeof(re_jump)));
- jmp->alt.i = rep_off - this->getoffset(jmp);
- this->m_pdata->m_data.align();
- // now fill in the alt jump for the repeat:
- rep = static_cast<re_repeat*>(this->getaddress(rep_off));
- rep->alt.i = this->m_pdata->m_data.size() - rep_off;
- //
- // If the repeat is possessive then bracket the repeat with a (?>...)
- // independent sub-expression construct:
- //
- if(possessive)
- {
- if(m_position != m_end)
- {
- //
- // Check for illegal following quantifier, we have to do this here, because
- // the extra states we insert below circumvents our usual error checking :-(
- //
- bool contin = false;
- do
- {
- if ((this->flags() & (regbase::main_option_type | regbase::mod_x | regbase::no_perl_ex)) == regbase::mod_x)
- {
- // whitespace skip:
- while ((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space))
- ++m_position;
- }
- if (m_position != m_end)
- {
- switch (this->m_traits.syntax_type(*m_position))
- {
- case regex_constants::syntax_star:
- case regex_constants::syntax_plus:
- case regex_constants::syntax_question:
- case regex_constants::syntax_open_brace:
- fail(regex_constants::error_badrepeat, m_position - m_base);
- return false;
- case regex_constants::syntax_open_mark:
- // Do we have a comment? If so we need to skip it here...
- if ((m_position + 2 < m_end) && this->m_traits.syntax_type(*(m_position + 1)) == regex_constants::syntax_question
- && this->m_traits.syntax_type(*(m_position + 2)) == regex_constants::syntax_hash)
- {
- while ((m_position != m_end)
- && (this->m_traits.syntax_type(*m_position++) != regex_constants::syntax_close_mark)) {
- }
- contin = true;
- }
- else
- contin = false;
- }
- }
- else
- contin = false;
- } while (contin);
- }
- re_brace* pb = static_cast<re_brace*>(this->insert_state(insert_point, syntax_element_startmark, sizeof(re_brace)));
- pb->index = -3;
- pb->icase = this->flags() & regbase::icase;
- jmp = static_cast<re_jump*>(this->insert_state(insert_point + sizeof(re_brace), syntax_element_jump, sizeof(re_jump)));
- this->m_pdata->m_data.align();
- jmp->alt.i = this->m_pdata->m_data.size() - this->getoffset(jmp);
- pb = static_cast<re_brace*>(this->append_state(syntax_element_endmark, sizeof(re_brace)));
- pb->index = -3;
- pb->icase = this->flags() & regbase::icase;
- }
- return true;
- }
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::parse_repeat_range(bool isbasic)
- {
- static const char incomplete_message[] = "Missing } in quantified repetition.";
- //
- // parse a repeat-range:
- //
- std::size_t min, max;
- std::intmax_t v;
- // skip whitespace:
- while((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space))
- ++m_position;
- if(this->m_position == this->m_end)
- {
- if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex))
- {
- fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message);
- return false;
- }
- // Treat the opening '{' as a literal character, rewind to start of error:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position;
- return parse_literal();
- }
- // get min:
- v = this->m_traits.toi(m_position, m_end, 10);
- // skip whitespace:
- if((v < 0) || (v > umax()))
- {
- if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex))
- {
- fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message);
- return false;
- }
- // Treat the opening '{' as a literal character, rewind to start of error:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position;
- return parse_literal();
- }
- while((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space))
- ++m_position;
- if(this->m_position == this->m_end)
- {
- if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex))
- {
- fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message);
- return false;
- }
- // Treat the opening '{' as a literal character, rewind to start of error:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position;
- return parse_literal();
- }
- min = static_cast<std::size_t>(v);
- // see if we have a comma:
- if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_comma)
- {
- // move on and error check:
- ++m_position;
- // skip whitespace:
- while((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space))
- ++m_position;
- if(this->m_position == this->m_end)
- {
- if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex))
- {
- fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message);
- return false;
- }
- // Treat the opening '{' as a literal character, rewind to start of error:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position;
- return parse_literal();
- }
- // get the value if any:
- v = this->m_traits.toi(m_position, m_end, 10);
- max = ((v >= 0) && (v < umax())) ? (std::size_t)v : (std::numeric_limits<std::size_t>::max)();
- }
- else
- {
- // no comma, max = min:
- max = min;
- }
- // skip whitespace:
- while((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space))
- ++m_position;
- // OK now check trailing }:
- if(this->m_position == this->m_end)
- {
- if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex))
- {
- fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message);
- return false;
- }
- // Treat the opening '{' as a literal character, rewind to start of error:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position;
- return parse_literal();
- }
- if(isbasic)
- {
- if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_escape)
- {
- ++m_position;
- if(this->m_position == this->m_end)
- {
- fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message);
- return false;
- }
- }
- else
- {
- fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message);
- return false;
- }
- }
- if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_brace)
- ++m_position;
- else
- {
- // Treat the opening '{' as a literal character, rewind to start of error:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position;
- return parse_literal();
- }
- //
- // finally go and add the repeat, unless error:
- //
- if(min > max)
- {
- // Backtrack to error location:
- m_position -= 2;
- while(this->m_traits.isctype(*m_position, this->m_word_mask)) --m_position;
- ++m_position;
- fail(regex_constants::error_badbrace, m_position - m_base);
- return false;
- }
- return parse_repeat(min, max);
- }
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::parse_alt()
- {
- //
- // error check: if there have been no previous states,
- // or if the last state was a '(' then error:
- //
- if(
- ((this->m_last_state == 0) || (this->m_last_state->type == syntax_element_startmark))
- &&
- !(
- ((this->flags() & regbase::main_option_type) == regbase::perl_syntax_group)
- &&
- ((this->flags() & regbase::no_empty_expressions) == 0)
- )
- )
- {
- fail(regex_constants::error_empty, this->m_position - this->m_base, "A regular expression cannot start with the alternation operator |.");
- return false;
- }
- //
- // Reset mark count if required:
- //
- if(m_max_mark < m_mark_count)
- m_max_mark = m_mark_count;
- if(m_mark_reset >= 0)
- m_mark_count = m_mark_reset;
- ++m_position;
- //
- // we need to append a trailing jump:
- //
- re_syntax_base* pj = this->append_state(BOOST_REGEX_DETAIL_NS::syntax_element_jump, sizeof(re_jump));
- std::ptrdiff_t jump_offset = this->getoffset(pj);
- //
- // now insert the alternative:
- //
- re_alt* palt = static_cast<re_alt*>(this->insert_state(this->m_alt_insert_point, syntax_element_alt, re_alt_size));
- jump_offset += re_alt_size;
- this->m_pdata->m_data.align();
- palt->alt.i = this->m_pdata->m_data.size() - this->getoffset(palt);
- //
- // update m_alt_insert_point so that the next alternate gets
- // inserted at the start of the second of the two we've just created:
- //
- this->m_alt_insert_point = this->m_pdata->m_data.size();
- //
- // the start of this alternative must have a case changes state
- // if the current block has messed around with case changes:
- //
- if(m_has_case_change)
- {
- static_cast<re_case*>(
- this->append_state(syntax_element_toggle_case, sizeof(re_case))
- )->icase = this->m_icase;
- }
- //
- // push the alternative onto our stack, a recursive
- // implementation here is easier to understand (and faster
- // as it happens), but causes all kinds of stack overflow problems
- // on programs with small stacks (COM+).
- //
- m_alt_jumps.push_back(jump_offset);
- return true;
- }
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::parse_set()
- {
- static const char incomplete_message[] = "Character set declaration starting with [ terminated prematurely - either no ] was found or the set had no content.";
- ++m_position;
- if(m_position == m_end)
- {
- fail(regex_constants::error_brack, m_position - m_base, incomplete_message);
- return false;
- }
- basic_char_set<charT, traits> char_set;
- const charT* base = m_position; // where the '[' was
- const charT* item_base = m_position; // where the '[' or '^' was
- while(m_position != m_end)
- {
- switch(this->m_traits.syntax_type(*m_position))
- {
- case regex_constants::syntax_caret:
- if(m_position == base)
- {
- char_set.negate();
- ++m_position;
- item_base = m_position;
- }
- else
- parse_set_literal(char_set);
- break;
- case regex_constants::syntax_close_set:
- if(m_position == item_base)
- {
- parse_set_literal(char_set);
- break;
- }
- else
- {
- ++m_position;
- if(0 == this->append_set(char_set))
- {
- fail(regex_constants::error_ctype, m_position - m_base);
- return false;
- }
- }
- return true;
- case regex_constants::syntax_open_set:
- if(parse_inner_set(char_set))
- break;
- return true;
- case regex_constants::syntax_escape:
- {
- //
- // look ahead and see if this is a character class shortcut
- // \d \w \s etc...
- //
- ++m_position;
- if(this->m_traits.escape_syntax_type(*m_position)
- == regex_constants::escape_type_class)
- {
- char_class_type m = this->m_traits.lookup_classname(m_position, m_position+1);
- if(m != 0)
- {
- char_set.add_class(m);
- ++m_position;
- break;
- }
- }
- else if(this->m_traits.escape_syntax_type(*m_position)
- == regex_constants::escape_type_not_class)
- {
- // negated character class:
- char_class_type m = this->m_traits.lookup_classname(m_position, m_position+1);
- if(m != 0)
- {
- char_set.add_negated_class(m);
- ++m_position;
- break;
- }
- }
- // not a character class, just a regular escape:
- --m_position;
- parse_set_literal(char_set);
- break;
- }
- default:
- parse_set_literal(char_set);
- break;
- }
- }
- return m_position != m_end;
- }
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::parse_inner_set(basic_char_set<charT, traits>& char_set)
- {
- static const char incomplete_message[] = "Character class declaration starting with [ terminated prematurely - either no ] was found or the set had no content.";
- //
- // we have either a character class [:name:]
- // a collating element [.name.]
- // or an equivalence class [=name=]
- //
- if(m_end == ++m_position)
- {
- fail(regex_constants::error_brack, m_position - m_base, incomplete_message);
- return false;
- }
- switch(this->m_traits.syntax_type(*m_position))
- {
- case regex_constants::syntax_dot:
- //
- // a collating element is treated as a literal:
- //
- --m_position;
- parse_set_literal(char_set);
- return true;
- case regex_constants::syntax_colon:
- {
- // check that character classes are actually enabled:
- if((this->flags() & (regbase::main_option_type | regbase::no_char_classes))
- == (regbase::basic_syntax_group | regbase::no_char_classes))
- {
- --m_position;
- parse_set_literal(char_set);
- return true;
- }
- // skip the ':'
- if(m_end == ++m_position)
- {
- fail(regex_constants::error_brack, m_position - m_base, incomplete_message);
- return false;
- }
- const charT* name_first = m_position;
- // skip at least one character, then find the matching ':]'
- if(m_end == ++m_position)
- {
- fail(regex_constants::error_brack, m_position - m_base, incomplete_message);
- return false;
- }
- while((m_position != m_end)
- && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_colon))
- ++m_position;
- const charT* name_last = m_position;
- if(m_end == m_position)
- {
- fail(regex_constants::error_brack, m_position - m_base, incomplete_message);
- return false;
- }
- if((m_end == ++m_position)
- || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set))
- {
- fail(regex_constants::error_brack, m_position - m_base, incomplete_message);
- return false;
- }
- //
- // check for negated class:
- //
- bool negated = false;
- if(this->m_traits.syntax_type(*name_first) == regex_constants::syntax_caret)
- {
- ++name_first;
- negated = true;
- }
- typedef typename traits::char_class_type m_type;
- m_type m = this->m_traits.lookup_classname(name_first, name_last);
- if(m == 0)
- {
- if(char_set.empty() && (name_last - name_first == 1))
- {
- // maybe a special case:
- ++m_position;
- if( (m_position != m_end)
- && (this->m_traits.syntax_type(*m_position)
- == regex_constants::syntax_close_set))
- {
- if(this->m_traits.escape_syntax_type(*name_first)
- == regex_constants::escape_type_left_word)
- {
- ++m_position;
- this->append_state(syntax_element_word_start);
- return false;
- }
- if(this->m_traits.escape_syntax_type(*name_first)
- == regex_constants::escape_type_right_word)
- {
- ++m_position;
- this->append_state(syntax_element_word_end);
- return false;
- }
- }
- }
- fail(regex_constants::error_ctype, name_first - m_base);
- return false;
- }
- if(!negated)
- char_set.add_class(m);
- else
- char_set.add_negated_class(m);
- ++m_position;
- break;
- }
- case regex_constants::syntax_equal:
- {
- // skip the '='
- if(m_end == ++m_position)
- {
- fail(regex_constants::error_brack, m_position - m_base, incomplete_message);
- return false;
- }
- const charT* name_first = m_position;
- // skip at least one character, then find the matching '=]'
- if(m_end == ++m_position)
- {
- fail(regex_constants::error_brack, m_position - m_base, incomplete_message);
- return false;
- }
- while((m_position != m_end)
- && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_equal))
- ++m_position;
- const charT* name_last = m_position;
- if(m_end == m_position)
- {
- fail(regex_constants::error_brack, m_position - m_base, incomplete_message);
- return false;
- }
- if((m_end == ++m_position)
- || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set))
- {
- fail(regex_constants::error_brack, m_position - m_base, incomplete_message);
- return false;
- }
- string_type m = this->m_traits.lookup_collatename(name_first, name_last);
- if(m.empty() || (m.size() > 2))
- {
- fail(regex_constants::error_collate, name_first - m_base);
- return false;
- }
- digraph<charT> d;
- d.first = m[0];
- if(m.size() > 1)
- d.second = m[1];
- else
- d.second = 0;
- char_set.add_equivalent(d);
- ++m_position;
- break;
- }
- default:
- --m_position;
- parse_set_literal(char_set);
- break;
- }
- return true;
- }
- template <class charT, class traits>
- void basic_regex_parser<charT, traits>::parse_set_literal(basic_char_set<charT, traits>& char_set)
- {
- digraph<charT> start_range(get_next_set_literal(char_set));
- if(m_end == m_position)
- {
- fail(regex_constants::error_brack, m_position - m_base);
- return;
- }
- if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_dash)
- {
- // we have a range:
- if(m_end == ++m_position)
- {
- fail(regex_constants::error_brack, m_position - m_base);
- return;
- }
- if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set)
- {
- digraph<charT> end_range = get_next_set_literal(char_set);
- char_set.add_range(start_range, end_range);
- if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_dash)
- {
- if(m_end == ++m_position)
- {
- fail(regex_constants::error_brack, m_position - m_base);
- return;
- }
- if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_set)
- {
- // trailing - :
- --m_position;
- return;
- }
- fail(regex_constants::error_range, m_position - m_base);
- return;
- }
- return;
- }
- --m_position;
- }
- char_set.add_single(start_range);
- }
- template <class charT, class traits>
- digraph<charT> basic_regex_parser<charT, traits>::get_next_set_literal(basic_char_set<charT, traits>& char_set)
- {
- digraph<charT> result;
- switch(this->m_traits.syntax_type(*m_position))
- {
- case regex_constants::syntax_dash:
- if(!char_set.empty())
- {
- // see if we are at the end of the set:
- if((++m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set))
- {
- fail(regex_constants::error_range, m_position - m_base);
- return result;
- }
- --m_position;
- }
- result.first = *m_position++;
- return result;
- case regex_constants::syntax_escape:
- // check to see if escapes are supported first:
- if(this->flags() & regex_constants::no_escape_in_lists)
- {
- result = *m_position++;
- break;
- }
- ++m_position;
- result = unescape_character();
- break;
- case regex_constants::syntax_open_set:
- {
- if(m_end == ++m_position)
- {
- fail(regex_constants::error_collate, m_position - m_base);
- return result;
- }
- if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_dot)
- {
- --m_position;
- result.first = *m_position;
- ++m_position;
- return result;
- }
- if(m_end == ++m_position)
- {
- fail(regex_constants::error_collate, m_position - m_base);
- return result;
- }
- const charT* name_first = m_position;
- // skip at least one character, then find the matching ':]'
- if(m_end == ++m_position)
- {
- fail(regex_constants::error_collate, name_first - m_base);
- return result;
- }
- while((m_position != m_end)
- && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_dot))
- ++m_position;
- const charT* name_last = m_position;
- if(m_end == m_position)
- {
- fail(regex_constants::error_collate, name_first - m_base);
- return result;
- }
- if((m_end == ++m_position)
- || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set))
- {
- fail(regex_constants::error_collate, name_first - m_base);
- return result;
- }
- ++m_position;
- string_type s = this->m_traits.lookup_collatename(name_first, name_last);
- if(s.empty() || (s.size() > 2))
- {
- fail(regex_constants::error_collate, name_first - m_base);
- return result;
- }
- result.first = s[0];
- if(s.size() > 1)
- result.second = s[1];
- else
- result.second = 0;
- return result;
- }
- default:
- result = *m_position++;
- }
- return result;
- }
- //
- // does a value fit in the specified charT type?
- //
- template <class charT>
- bool valid_value(charT, std::intmax_t v, const std::integral_constant<bool, true>&)
- {
- return (v >> (sizeof(charT) * CHAR_BIT)) == 0;
- }
- template <class charT>
- bool valid_value(charT, std::intmax_t, const std::integral_constant<bool, false>&)
- {
- return true; // v will alsways fit in a charT
- }
- template <class charT>
- bool valid_value(charT c, std::intmax_t v)
- {
- return valid_value(c, v, std::integral_constant<bool, (sizeof(charT) < sizeof(std::intmax_t))>());
- }
- template <class charT, class traits>
- charT basic_regex_parser<charT, traits>::unescape_character()
- {
- #ifdef BOOST_REGEX_MSVC
- #pragma warning(push)
- #pragma warning(disable:4127)
- #endif
- charT result(0);
- if(m_position == m_end)
- {
- fail(regex_constants::error_escape, m_position - m_base, "Escape sequence terminated prematurely.");
- return false;
- }
- switch(this->m_traits.escape_syntax_type(*m_position))
- {
- case regex_constants::escape_type_control_a:
- result = charT('\a');
- break;
- case regex_constants::escape_type_e:
- result = charT(27);
- break;
- case regex_constants::escape_type_control_f:
- result = charT('\f');
- break;
- case regex_constants::escape_type_control_n:
- result = charT('\n');
- break;
- case regex_constants::escape_type_control_r:
- result = charT('\r');
- break;
- case regex_constants::escape_type_control_t:
- result = charT('\t');
- break;
- case regex_constants::escape_type_control_v:
- result = charT('\v');
- break;
- case regex_constants::escape_type_word_assert:
- result = charT('\b');
- break;
- case regex_constants::escape_type_ascii_control:
- ++m_position;
- if(m_position == m_end)
- {
- // Rewind to start of escape:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position;
- fail(regex_constants::error_escape, m_position - m_base, "ASCII escape sequence terminated prematurely.");
- return result;
- }
- result = static_cast<charT>(*m_position % 32);
- break;
- case regex_constants::escape_type_hex:
- ++m_position;
- if(m_position == m_end)
- {
- // Rewind to start of escape:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position;
- fail(regex_constants::error_escape, m_position - m_base, "Hexadecimal escape sequence terminated prematurely.");
- return result;
- }
- // maybe have \x{ddd}
- if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_open_brace)
- {
- ++m_position;
- if(m_position == m_end)
- {
- // Rewind to start of escape:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position;
- fail(regex_constants::error_escape, m_position - m_base, "Missing } in hexadecimal escape sequence.");
- return result;
- }
- std::intmax_t i = this->m_traits.toi(m_position, m_end, 16);
- if((m_position == m_end)
- || (i < 0)
- || ((std::numeric_limits<charT>::is_specialized) && (i > (std::intmax_t)(std::numeric_limits<charT>::max)()))
- || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_brace))
- {
- // Rewind to start of escape:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position;
- fail(regex_constants::error_badbrace, m_position - m_base, "Hexadecimal escape sequence was invalid.");
- return result;
- }
- ++m_position;
- result = charT(i);
- }
- else
- {
- std::ptrdiff_t len = (std::min)(static_cast<std::ptrdiff_t>(2), static_cast<std::ptrdiff_t>(m_end - m_position));
- std::intmax_t i = this->m_traits.toi(m_position, m_position + len, 16);
- if((i < 0)
- || !valid_value(charT(0), i))
- {
- // Rewind to start of escape:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position;
- fail(regex_constants::error_escape, m_position - m_base, "Escape sequence did not encode a valid character.");
- return result;
- }
- result = charT(i);
- }
- return result;
- case regex_constants::syntax_digit:
- {
- // an octal escape sequence, the first character must be a zero
- // followed by up to 3 octal digits:
- std::ptrdiff_t len = (std::min)(std::distance(m_position, m_end), static_cast<std::ptrdiff_t>(4));
- const charT* bp = m_position;
- std::intmax_t val = this->m_traits.toi(bp, bp + 1, 8);
- if(val != 0)
- {
- // Rewind to start of escape:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position;
- // Oops not an octal escape after all:
- fail(regex_constants::error_escape, m_position - m_base, "Invalid octal escape sequence.");
- return result;
- }
- val = this->m_traits.toi(m_position, m_position + len, 8);
- if((val < 0) || (val > (std::intmax_t)(std::numeric_limits<charT>::max)()))
- {
- // Rewind to start of escape:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position;
- fail(regex_constants::error_escape, m_position - m_base, "Octal escape sequence is invalid.");
- return result;
- }
- return static_cast<charT>(val);
- }
- case regex_constants::escape_type_named_char:
- {
- ++m_position;
- if(m_position == m_end)
- {
- // Rewind to start of escape:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position;
- fail(regex_constants::error_escape, m_position - m_base);
- return false;
- }
- // maybe have \N{name}
- if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_open_brace)
- {
- const charT* base = m_position;
- // skip forward until we find enclosing brace:
- while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_brace))
- ++m_position;
- if(m_position == m_end)
- {
- // Rewind to start of escape:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position;
- fail(regex_constants::error_escape, m_position - m_base);
- return false;
- }
- string_type s = this->m_traits.lookup_collatename(++base, m_position++);
- if(s.empty())
- {
- // Rewind to start of escape:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position;
- fail(regex_constants::error_collate, m_position - m_base);
- return false;
- }
- if(s.size() == 1)
- {
- return s[0];
- }
- }
- // fall through is a failure:
- // Rewind to start of escape:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position;
- fail(regex_constants::error_escape, m_position - m_base);
- return false;
- }
- default:
- result = *m_position;
- break;
- }
- ++m_position;
- return result;
- #ifdef BOOST_REGEX_MSVC
- #pragma warning(pop)
- #endif
- }
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::parse_backref()
- {
- BOOST_REGEX_ASSERT(m_position != m_end);
- const charT* pc = m_position;
- std::intmax_t i = this->m_traits.toi(pc, pc + 1, 10);
- if((i == 0) || (((this->flags() & regbase::main_option_type) == regbase::perl_syntax_group) && (this->flags() & regbase::no_bk_refs)))
- {
- // not a backref at all but an octal escape sequence:
- charT c = unescape_character();
- this->append_literal(c);
- }
- else if((i > 0) && (this->m_backrefs.test((std::size_t)i)))
- {
- m_position = pc;
- re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_backref, sizeof(re_brace)));
- pb->index = (int)i;
- pb->icase = this->flags() & regbase::icase;
- }
- else
- {
- // Rewind to start of escape:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position;
- fail(regex_constants::error_backref, m_position - m_base);
- return false;
- }
- return true;
- }
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::parse_QE()
- {
- #ifdef BOOST_REGEX_MSVC
- #pragma warning(push)
- #pragma warning(disable:4127)
- #endif
- //
- // parse a \Q...\E sequence:
- //
- ++m_position; // skip the Q
- const charT* start = m_position;
- const charT* end;
- do
- {
- while((m_position != m_end)
- && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape))
- ++m_position;
- if(m_position == m_end)
- {
- // a \Q...\E sequence may terminate with the end of the expression:
- end = m_position;
- break;
- }
- if(++m_position == m_end) // skip the escape
- {
- fail(regex_constants::error_escape, m_position - m_base, "Unterminated \\Q...\\E sequence.");
- return false;
- }
- // check to see if it's a \E:
- if(this->m_traits.escape_syntax_type(*m_position) == regex_constants::escape_type_E)
- {
- ++m_position;
- end = m_position - 2;
- break;
- }
- // otherwise go round again:
- }while(true);
- //
- // now add all the character between the two escapes as literals:
- //
- while(start != end)
- {
- this->append_literal(*start);
- ++start;
- }
- return true;
- #ifdef BOOST_REGEX_MSVC
- #pragma warning(pop)
- #endif
- }
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::parse_perl_extension()
- {
- if(++m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- //
- // treat comments as a special case, as these
- // are the only ones that don't start with a leading
- // startmark state:
- //
- if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_hash)
- {
- while((m_position != m_end)
- && (this->m_traits.syntax_type(*m_position++) != regex_constants::syntax_close_mark))
- {}
- return true;
- }
- //
- // backup some state, and prepare the way:
- //
- int markid = 0;
- std::ptrdiff_t jump_offset = 0;
- re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_startmark, sizeof(re_brace)));
- pb->icase = this->flags() & regbase::icase;
- std::ptrdiff_t last_paren_start = this->getoffset(pb);
- // back up insertion point for alternations, and set new point:
- std::ptrdiff_t last_alt_point = m_alt_insert_point;
- this->m_pdata->m_data.align();
- m_alt_insert_point = this->m_pdata->m_data.size();
- std::ptrdiff_t expected_alt_point = m_alt_insert_point;
- bool restore_flags = true;
- regex_constants::syntax_option_type old_flags = this->flags();
- bool old_case_change = m_has_case_change;
- m_has_case_change = false;
- charT name_delim;
- int mark_reset = m_mark_reset;
- int max_mark = m_max_mark;
- m_mark_reset = -1;
- m_max_mark = m_mark_count;
- std::intmax_t v;
- //
- // select the actual extension used:
- //
- switch(this->m_traits.syntax_type(*m_position))
- {
- case regex_constants::syntax_or:
- m_mark_reset = m_mark_count;
- BOOST_REGEX_FALLTHROUGH;
- case regex_constants::syntax_colon:
- //
- // a non-capturing mark:
- //
- pb->index = markid = 0;
- ++m_position;
- break;
- case regex_constants::syntax_digit:
- {
- //
- // a recursive subexpression:
- //
- v = this->m_traits.toi(m_position, m_end, 10);
- if((v < 0) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark))
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base, "The recursive sub-expression refers to an invalid marking group, or is unterminated.");
- return false;
- }
- insert_recursion:
- pb->index = markid = 0;
- re_recurse* pr = static_cast<re_recurse*>(this->append_state(syntax_element_recurse, sizeof(re_recurse)));
- pr->alt.i = (std::ptrdiff_t)v;
- pr->state_id = 0;
- static_cast<re_case*>(
- this->append_state(syntax_element_toggle_case, sizeof(re_case))
- )->icase = this->flags() & regbase::icase;
- break;
- }
- case regex_constants::syntax_plus:
- //
- // A forward-relative recursive subexpression:
- //
- ++m_position;
- v = this->m_traits.toi(m_position, m_end, 10);
- if((v <= 0) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark))
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base, "An invalid or unterminated recursive sub-expression.");
- return false;
- }
- if ((std::numeric_limits<std::intmax_t>::max)() - m_mark_count < v)
- {
- fail(regex_constants::error_perl_extension, m_position - m_base, "An invalid or unterminated recursive sub-expression.");
- return false;
- }
- v += m_mark_count;
- goto insert_recursion;
- case regex_constants::syntax_dash:
- //
- // Possibly a backward-relative recursive subexpression:
- //
- ++m_position;
- v = this->m_traits.toi(m_position, m_end, 10);
- if(v <= 0)
- {
- --m_position;
- // Oops not a relative recursion at all, but a (?-imsx) group:
- goto option_group_jump;
- }
- v = static_cast<std::intmax_t>(m_mark_count) + 1 - v;
- if(v <= 0)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base, "An invalid or unterminated recursive sub-expression.");
- return false;
- }
- goto insert_recursion;
- case regex_constants::syntax_equal:
- pb->index = markid = -1;
- ++m_position;
- jump_offset = this->getoffset(this->append_state(syntax_element_jump, sizeof(re_jump)));
- this->m_pdata->m_data.align();
- m_alt_insert_point = this->m_pdata->m_data.size();
- break;
- case regex_constants::syntax_not:
- pb->index = markid = -2;
- ++m_position;
- jump_offset = this->getoffset(this->append_state(syntax_element_jump, sizeof(re_jump)));
- this->m_pdata->m_data.align();
- m_alt_insert_point = this->m_pdata->m_data.size();
- break;
- case regex_constants::escape_type_left_word:
- {
- // a lookbehind assertion:
- if(++m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- regex_constants::syntax_type t = this->m_traits.syntax_type(*m_position);
- if(t == regex_constants::syntax_not)
- pb->index = markid = -2;
- else if(t == regex_constants::syntax_equal)
- pb->index = markid = -1;
- else
- {
- // Probably a named capture which also starts (?< :
- name_delim = '>';
- --m_position;
- goto named_capture_jump;
- }
- ++m_position;
- jump_offset = this->getoffset(this->append_state(syntax_element_jump, sizeof(re_jump)));
- this->append_state(syntax_element_backstep, sizeof(re_brace));
- this->m_pdata->m_data.align();
- m_alt_insert_point = this->m_pdata->m_data.size();
- break;
- }
- case regex_constants::escape_type_right_word:
- //
- // an independent sub-expression:
- //
- pb->index = markid = -3;
- ++m_position;
- jump_offset = this->getoffset(this->append_state(syntax_element_jump, sizeof(re_jump)));
- this->m_pdata->m_data.align();
- m_alt_insert_point = this->m_pdata->m_data.size();
- break;
- case regex_constants::syntax_open_mark:
- {
- // a conditional expression:
- pb->index = markid = -4;
- if(++m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- v = this->m_traits.toi(m_position, m_end, 10);
- if(m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- if(*m_position == charT('R'))
- {
- if(++m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- if(*m_position == charT('&'))
- {
- const charT* base = ++m_position;
- while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark))
- ++m_position;
- if(m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- v = -static_cast<int>(hash_value_from_capture_name(base, m_position));
- }
- else
- {
- v = -this->m_traits.toi(m_position, m_end, 10);
- }
- re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_assert_backref, sizeof(re_brace)));
- br->index = v < 0 ? (int)(v - 1) : 0;
- if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- if(++m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- }
- else if((*m_position == charT('\'')) || (*m_position == charT('<')))
- {
- const charT* base = ++m_position;
- while((m_position != m_end) && (*m_position != charT('>')) && (*m_position != charT('\'')))
- ++m_position;
- if(m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- v = static_cast<int>(hash_value_from_capture_name(base, m_position));
- re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_assert_backref, sizeof(re_brace)));
- br->index = (int)v;
- if(((*m_position != charT('>')) && (*m_position != charT('\''))) || (++m_position == m_end))
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base, "Unterminated named capture.");
- return false;
- }
- if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- if(++m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- }
- else if(*m_position == charT('D'))
- {
- const char* def = "DEFINE";
- while(*def && (m_position != m_end) && (*m_position == charT(*def)))
- ++m_position, ++def;
- if((m_position == m_end) || *def)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_assert_backref, sizeof(re_brace)));
- br->index = 9999; // special magic value!
- if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- if(++m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- }
- else if(v > 0)
- {
- re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_assert_backref, sizeof(re_brace)));
- br->index = (int)v;
- if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- if(++m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- }
- else
- {
- // verify that we have a lookahead or lookbehind assert:
- if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_question)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- if(++m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- if(this->m_traits.syntax_type(*m_position) == regex_constants::escape_type_left_word)
- {
- if(++m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- if((this->m_traits.syntax_type(*m_position) != regex_constants::syntax_equal)
- && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_not))
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- m_position -= 3;
- }
- else
- {
- if((this->m_traits.syntax_type(*m_position) != regex_constants::syntax_equal)
- && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_not))
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- m_position -= 2;
- }
- }
- break;
- }
- case regex_constants::syntax_close_mark:
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- case regex_constants::escape_type_end_buffer:
- {
- name_delim = *m_position;
- named_capture_jump:
- markid = 0;
- if(0 == (this->flags() & regbase::nosubs))
- {
- markid = ++m_mark_count;
- if(this->flags() & regbase::save_subexpression_location)
- this->m_pdata->m_subs.push_back(std::pair<std::size_t, std::size_t>(std::distance(m_base, m_position) - 2, 0));
- }
- pb->index = markid;
- const charT* base = ++m_position;
- if(m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- while((m_position != m_end) && (*m_position != name_delim))
- ++m_position;
- if(m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- this->m_pdata->set_name(base, m_position, markid);
- ++m_position;
- break;
- }
- default:
- if(*m_position == charT('R'))
- {
- ++m_position;
- v = 0;
- if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- goto insert_recursion;
- }
- if(*m_position == charT('&'))
- {
- ++m_position;
- const charT* base = m_position;
- while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark))
- ++m_position;
- if(m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- v = static_cast<int>(hash_value_from_capture_name(base, m_position));
- goto insert_recursion;
- }
- if(*m_position == charT('P'))
- {
- ++m_position;
- if(m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- if(*m_position == charT('>'))
- {
- ++m_position;
- const charT* base = m_position;
- while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark))
- ++m_position;
- if(m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- v = static_cast<int>(hash_value_from_capture_name(base, m_position));
- goto insert_recursion;
- }
- }
- //
- // lets assume that we have a (?imsx) group and try and parse it:
- //
- option_group_jump:
- regex_constants::syntax_option_type opts = parse_options();
- if(m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- // make a note of whether we have a case change:
- m_has_case_change = ((opts & regbase::icase) != (this->flags() & regbase::icase));
- pb->index = markid = 0;
- if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_mark)
- {
- // update flags and carry on as normal:
- this->flags(opts);
- restore_flags = false;
- old_case_change |= m_has_case_change; // defer end of scope by one ')'
- }
- else if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_colon)
- {
- // update flags and carry on until the matching ')' is found:
- this->flags(opts);
- ++m_position;
- }
- else
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- // finally append a case change state if we need it:
- if(m_has_case_change)
- {
- static_cast<re_case*>(
- this->append_state(syntax_element_toggle_case, sizeof(re_case))
- )->icase = opts & regbase::icase;
- }
- }
- //
- // now recursively add more states, this will terminate when we get to a
- // matching ')' :
- //
- parse_all();
- //
- // Unwind alternatives:
- //
- if(0 == unwind_alts(last_paren_start))
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base, "Invalid alternation operators within (?...) block.");
- return false;
- }
- //
- // we either have a ')' or we have run out of characters prematurely:
- //
- if(m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- this->fail(regex_constants::error_paren, std::distance(m_base, m_end));
- return false;
- }
- BOOST_REGEX_ASSERT(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_mark);
- ++m_position;
- //
- // restore the flags:
- //
- if(restore_flags)
- {
- // append a case change state if we need it:
- if(m_has_case_change)
- {
- static_cast<re_case*>(
- this->append_state(syntax_element_toggle_case, sizeof(re_case))
- )->icase = old_flags & regbase::icase;
- }
- this->flags(old_flags);
- }
- //
- // set up the jump pointer if we have one:
- //
- if(jump_offset)
- {
- this->m_pdata->m_data.align();
- re_jump* jmp = static_cast<re_jump*>(this->getaddress(jump_offset));
- jmp->alt.i = this->m_pdata->m_data.size() - this->getoffset(jmp);
- if((this->m_last_state == jmp) && (markid != -2))
- {
- // Oops... we didn't have anything inside the assertion.
- // Note we don't get here for negated forward lookahead as (?!)
- // does have some uses.
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base, "Invalid or empty zero width assertion.");
- return false;
- }
- }
- //
- // verify that if this is conditional expression, that we do have
- // an alternative, if not add one:
- //
- if(markid == -4)
- {
- re_syntax_base* b = this->getaddress(expected_alt_point);
- // Make sure we have exactly one alternative following this state:
- if(b->type != syntax_element_alt)
- {
- re_alt* alt = static_cast<re_alt*>(this->insert_state(expected_alt_point, syntax_element_alt, sizeof(re_alt)));
- alt->alt.i = this->m_pdata->m_data.size() - this->getoffset(alt);
- }
- else if(((std::ptrdiff_t)this->m_pdata->m_data.size() > (static_cast<re_alt*>(b)->alt.i + this->getoffset(b))) && (static_cast<re_alt*>(b)->alt.i > 0) && this->getaddress(static_cast<re_alt*>(b)->alt.i, b)->type == syntax_element_alt)
- {
- // Can't have seen more than one alternative:
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_bad_pattern, m_position - m_base, "More than one alternation operator | was encountered inside a conditional expression.");
- return false;
- }
- else
- {
- // We must *not* have seen an alternative inside a (DEFINE) block:
- b = this->getaddress(b->next.i, b);
- if((b->type == syntax_element_assert_backref) && (static_cast<re_brace*>(b)->index == 9999))
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_bad_pattern, m_position - m_base, "Alternation operators are not allowed inside a DEFINE block.");
- return false;
- }
- }
- // check for invalid repetition of next state:
- b = this->getaddress(expected_alt_point);
- b = this->getaddress(static_cast<re_alt*>(b)->next.i, b);
- if((b->type != syntax_element_assert_backref)
- && (b->type != syntax_element_startmark))
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_badrepeat, m_position - m_base, "A repetition operator cannot be applied to a zero-width assertion.");
- return false;
- }
- }
- //
- // append closing parenthesis state:
- //
- pb = static_cast<re_brace*>(this->append_state(syntax_element_endmark, sizeof(re_brace)));
- pb->index = markid;
- pb->icase = this->flags() & regbase::icase;
- this->m_paren_start = last_paren_start;
- //
- // restore the alternate insertion point:
- //
- this->m_alt_insert_point = last_alt_point;
- //
- // and the case change data:
- //
- m_has_case_change = old_case_change;
- //
- // And the mark_reset data:
- //
- if(m_max_mark > m_mark_count)
- {
- m_mark_count = m_max_mark;
- }
- m_mark_reset = mark_reset;
- m_max_mark = max_mark;
- if(markid > 0)
- {
- if(this->flags() & regbase::save_subexpression_location)
- this->m_pdata->m_subs.at((std::size_t)markid - 1).second = std::distance(m_base, m_position) - 1;
- //
- // allow backrefs to this mark:
- //
- this->m_backrefs.set(markid);
- }
- return true;
- }
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::match_verb(const char* verb)
- {
- while(*verb)
- {
- if(static_cast<charT>(*verb) != *m_position)
- {
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- if(++m_position == m_end)
- {
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- ++verb;
- }
- return true;
- }
- #ifdef BOOST_REGEX_MSVC
- # pragma warning(push)
- #if BOOST_REGEX_MSVC >= 1800
- #pragma warning(disable:26812)
- #endif
- #endif
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::parse_perl_verb()
- {
- if(++m_position == m_end)
- {
- // Rewind to start of (* sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- switch(*m_position)
- {
- case 'F':
- if(++m_position == m_end)
- {
- // Rewind to start of (* sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- if((this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_mark) || match_verb("AIL"))
- {
- if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark))
- {
- // Rewind to start of (* sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- ++m_position;
- this->append_state(syntax_element_fail);
- return true;
- }
- break;
- case 'A':
- if(++m_position == m_end)
- {
- // Rewind to start of (* sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- if(match_verb("CCEPT"))
- {
- if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark))
- {
- // Rewind to start of (* sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- ++m_position;
- this->append_state(syntax_element_accept);
- return true;
- }
- break;
- case 'C':
- if(++m_position == m_end)
- {
- // Rewind to start of (* sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- if(match_verb("OMMIT"))
- {
- if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark))
- {
- // Rewind to start of (* sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- ++m_position;
- static_cast<re_commit*>(this->append_state(syntax_element_commit, sizeof(re_commit)))->action = commit_commit;
- this->m_pdata->m_disable_match_any = true;
- return true;
- }
- break;
- case 'P':
- if(++m_position == m_end)
- {
- // Rewind to start of (* sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- if(match_verb("RUNE"))
- {
- if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark))
- {
- // Rewind to start of (* sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- ++m_position;
- static_cast<re_commit*>(this->append_state(syntax_element_commit, sizeof(re_commit)))->action = commit_prune;
- this->m_pdata->m_disable_match_any = true;
- return true;
- }
- break;
- case 'S':
- if(++m_position == m_end)
- {
- // Rewind to start of (* sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- if(match_verb("KIP"))
- {
- if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark))
- {
- // Rewind to start of (* sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- ++m_position;
- static_cast<re_commit*>(this->append_state(syntax_element_commit, sizeof(re_commit)))->action = commit_skip;
- this->m_pdata->m_disable_match_any = true;
- return true;
- }
- break;
- case 'T':
- if(++m_position == m_end)
- {
- // Rewind to start of (* sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- if(match_verb("HEN"))
- {
- if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark))
- {
- // Rewind to start of (* sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- ++m_position;
- this->append_state(syntax_element_then);
- this->m_pdata->m_disable_match_any = true;
- return true;
- }
- break;
- }
- // Rewind to start of (* sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_perl_extension, m_position - m_base);
- return false;
- }
- #ifdef BOOST_REGEX_MSVC
- # pragma warning(pop)
- #endif
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::add_emacs_code(bool negate)
- {
- //
- // parses an emacs style \sx or \Sx construct.
- //
- if(++m_position == m_end)
- {
- // Rewind to start of sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position;
- fail(regex_constants::error_escape, m_position - m_base);
- return false;
- }
- basic_char_set<charT, traits> char_set;
- if(negate)
- char_set.negate();
- static const charT s_punct[5] = { 'p', 'u', 'n', 'c', 't', };
- switch(*m_position)
- {
- case 's':
- case ' ':
- char_set.add_class(this->m_mask_space);
- break;
- case 'w':
- char_set.add_class(this->m_word_mask);
- break;
- case '_':
- char_set.add_single(digraph<charT>(charT('$')));
- char_set.add_single(digraph<charT>(charT('&')));
- char_set.add_single(digraph<charT>(charT('*')));
- char_set.add_single(digraph<charT>(charT('+')));
- char_set.add_single(digraph<charT>(charT('-')));
- char_set.add_single(digraph<charT>(charT('_')));
- char_set.add_single(digraph<charT>(charT('<')));
- char_set.add_single(digraph<charT>(charT('>')));
- break;
- case '.':
- char_set.add_class(this->m_traits.lookup_classname(s_punct, s_punct+5));
- break;
- case '(':
- char_set.add_single(digraph<charT>(charT('(')));
- char_set.add_single(digraph<charT>(charT('[')));
- char_set.add_single(digraph<charT>(charT('{')));
- break;
- case ')':
- char_set.add_single(digraph<charT>(charT(')')));
- char_set.add_single(digraph<charT>(charT(']')));
- char_set.add_single(digraph<charT>(charT('}')));
- break;
- case '"':
- char_set.add_single(digraph<charT>(charT('"')));
- char_set.add_single(digraph<charT>(charT('\'')));
- char_set.add_single(digraph<charT>(charT('`')));
- break;
- case '\'':
- char_set.add_single(digraph<charT>(charT('\'')));
- char_set.add_single(digraph<charT>(charT(',')));
- char_set.add_single(digraph<charT>(charT('#')));
- break;
- case '<':
- char_set.add_single(digraph<charT>(charT(';')));
- break;
- case '>':
- char_set.add_single(digraph<charT>(charT('\n')));
- char_set.add_single(digraph<charT>(charT('\f')));
- break;
- default:
- fail(regex_constants::error_ctype, m_position - m_base);
- return false;
- }
- if(0 == this->append_set(char_set))
- {
- fail(regex_constants::error_ctype, m_position - m_base);
- return false;
- }
- ++m_position;
- return true;
- }
- template <class charT, class traits>
- regex_constants::syntax_option_type basic_regex_parser<charT, traits>::parse_options()
- {
- // we have a (?imsx-imsx) group, convert it into a set of flags:
- regex_constants::syntax_option_type f = this->flags();
- bool breakout = false;
- do
- {
- switch(*m_position)
- {
- case 's':
- f |= regex_constants::mod_s;
- f &= ~regex_constants::no_mod_s;
- break;
- case 'm':
- f &= ~regex_constants::no_mod_m;
- break;
- case 'i':
- f |= regex_constants::icase;
- break;
- case 'x':
- f |= regex_constants::mod_x;
- break;
- default:
- breakout = true;
- continue;
- }
- if(++m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_paren, m_position - m_base);
- return false;
- }
- }
- while(!breakout);
-
- breakout = false;
- if(*m_position == static_cast<charT>('-'))
- {
- if(++m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_paren, m_position - m_base);
- return false;
- }
- do
- {
- switch(*m_position)
- {
- case 's':
- f &= ~regex_constants::mod_s;
- f |= regex_constants::no_mod_s;
- break;
- case 'm':
- f |= regex_constants::no_mod_m;
- break;
- case 'i':
- f &= ~regex_constants::icase;
- break;
- case 'x':
- f &= ~regex_constants::mod_x;
- break;
- default:
- breakout = true;
- continue;
- }
- if(++m_position == m_end)
- {
- // Rewind to start of (? sequence:
- --m_position;
- while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
- fail(regex_constants::error_paren, m_position - m_base);
- return false;
- }
- }
- while(!breakout);
- }
- return f;
- }
- template <class charT, class traits>
- bool basic_regex_parser<charT, traits>::unwind_alts(std::ptrdiff_t last_paren_start)
- {
- //
- // If we didn't actually add any states after the last
- // alternative then that's an error:
- //
- if((this->m_alt_insert_point == static_cast<std::ptrdiff_t>(this->m_pdata->m_data.size()))
- && (!m_alt_jumps.empty()) && (m_alt_jumps.back() > last_paren_start)
- &&
- !(
- ((this->flags() & regbase::main_option_type) == regbase::perl_syntax_group)
- &&
- ((this->flags() & regbase::no_empty_expressions) == 0)
- )
- )
- {
- fail(regex_constants::error_empty, this->m_position - this->m_base, "Can't terminate a sub-expression with an alternation operator |.");
- return false;
- }
- //
- // Fix up our alternatives:
- //
- while((!m_alt_jumps.empty()) && (m_alt_jumps.back() > last_paren_start))
- {
- //
- // fix up the jump to point to the end of the states
- // that we've just added:
- //
- std::ptrdiff_t jump_offset = m_alt_jumps.back();
- m_alt_jumps.pop_back();
- this->m_pdata->m_data.align();
- re_jump* jmp = static_cast<re_jump*>(this->getaddress(jump_offset));
- BOOST_REGEX_ASSERT(jmp->type == syntax_element_jump);
- jmp->alt.i = this->m_pdata->m_data.size() - jump_offset;
- }
- return true;
- }
- #ifdef BOOST_REGEX_MSVC
- #pragma warning(pop)
- #endif
- } // namespace BOOST_REGEX_DETAIL_NS
- } // namespace boost
- #endif
|