X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=stolen_chunk_of_toke.c;h=c667eaa77ba2704b3cc69ae93e6b802b7555a47a;hb=b52072dc659832e3465ecb4801024bc7c05ddad7;hp=5da609536eca054d4d1cb7fc99e1fc86c3c52e33;hpb=e807ee502b7c21edb01594764925aabdc8948f37;p=p5sagit%2FDevel-Declare.git diff --git a/stolen_chunk_of_toke.c b/stolen_chunk_of_toke.c index 5da6095..c667eaa 100644 --- a/stolen_chunk_of_toke.c +++ b/stolen_chunk_of_toke.c @@ -17,13 +17,70 @@ * up but if it does blame me (Matt S Trout), not the poor original authors */ -/* the following #defines are stolen from assorted headers, not toke.c */ +/* the following #defines are stolen from assorted headers, not toke.c (mst) */ + +#define skipspace(a) S_skipspace(aTHX_ a, 0) +#define peekspace(a) S_skipspace(aTHX_ a, 1) +#define skipspace_force(a) S_skipspace(aTHX_ a, 2) +#define incline(a) S_incline(aTHX_ a) +#define filter_gets(a,b,c) S_filter_gets(aTHX_ a,b,c) +#define scan_str(a,b,c) S_scan_str(aTHX_ a,b,c) +#define scan_word(a,b,c,d,e) S_scan_word(aTHX_ a,b,c,d,e) +#define scan_ident(a,b,c,d,e) S_scan_ident(aTHX_ a,b,c,d,e) + +STATIC void S_incline(pTHX_ char *s); +STATIC char* S_skipspace(pTHX_ char *s, int incline); +STATIC char * S_filter_gets(pTHX_ SV *sv, PerlIO *fp, STRLEN append); +STATIC char* S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims); +STATIC char* S_scan_word(pTHX_ char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp); #define DPTR2FPTR(t,p) ((t)PTR2nat(p)) /* data pointer to function pointer */ #define FPTR2DPTR(t,p) ((t)PTR2nat(p)) /* function pointer to data pointer */ -#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t), +#define PTR2nat(p) (PTRV)(p) /* pointer to integer of PTRSIZE */ + +/* conditionalise these two because as of 5.9.5 we already get them from + the headers (mst) */ +#ifndef Newx #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)safemalloc((MEM_SIZE)((n)*sizeof(t))))) +#endif +#ifndef SvPVX_const +#define SvPVX_const(sv) ((const char*) (0 + SvPVX(sv))) +#endif +#ifndef MEM_WRAP_CHECK_ +#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t), +#endif + +#define SvPV_renew(sv,n) \ + STMT_START { SvLEN_set(sv, n); \ + SvPV_set((sv), (MEM_WRAP_CHECK_(n,char) \ + (char*)saferealloc((Malloc_t)SvPVX(sv), \ + (MEM_SIZE)((n))))); \ + } STMT_END +#define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x))) + +/* On MacOS, respect nonbreaking spaces */ +#ifdef MACOS_TRADITIONAL +#define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t') +#else +#define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t') +#endif + +/* + * Normally, during compile time, PL_curcop == &PL_compiling is true. However, + * Devel::Declare makes the interpreter call back to perl during compile time, + * which temporarily enters runtime. Then perl space calls various functions + * from this file, which are designed to work during compile time. They all + * happen to operate on PL_curcop, not PL_compiling. That doesn't make a + * difference in the core, but it does for Devel::Declare, which operates at + * runtime, but still wants to mangle the things that are about to be compiled. + * That's why we define our own PL_curcop and make it point to PL_compiling + * here. + */ +#undef PL_curcop +#define PL_curcop (&PL_compiling) + +#define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline)) #define LEX_NORMAL 10 /* normal code (ie not within "...") */ #define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */ @@ -41,6 +98,111 @@ #define LEX_FORMLINE 1 /* expecting a format line */ #define LEX_KNOWNEXT 0 /* next token known; just return it */ +/* and these two are my own madness (mst) */ + +#if PERL_REVISION == 5 && PERL_VERSION == 8 && PERL_SUBVERSION >= 8 +#define PERL_5_8_8_PLUS +#endif + +#if PERL_REVISION == 5 && PERL_VERSION > 8 +#define PERL_5_9_PLUS +#endif + +#if !defined(PERL_5_9_PLUS) && defined(PERL_IMPLICIT_CONTEXT) +/* These two are not exported from the core on Windows. With 5.9+ + it's not an issue, because they're part of the PL_parser structure, + which is exported. On multiplicity/thread builds we can work + around the lack of export by this formulation, where we provide + a substitute implementation of the unexported accessor functions. + On single-interpreter builds we can't, because access is directly + via symbols that are not exported. */ +# define Perl_Ilinestart_ptr my_Ilinestart_ptr +char **my_Ilinestart_ptr(pTHX) { return &(aTHX->Ilinestart); } +# define Perl_Isublex_info_ptr my_Isublex_info_ptr +static SUBLEXINFO *my_Isublex_info_ptr(pTHX) { return &(aTHX->Isublex_info); } +#endif + +#ifdef PERL_5_9_PLUS +/* 5.9+ moves a bunch of things to a PL_parser struct so we need to + declare the backcompat macros for things to still work (mst) */ + +/* XXX temporary backwards compatibility */ +#define PL_lex_brackets (PL_parser->lex_brackets) +#define PL_lex_brackstack (PL_parser->lex_brackstack) +#define PL_lex_casemods (PL_parser->lex_casemods) +#define PL_lex_casestack (PL_parser->lex_casestack) +#define PL_lex_defer (PL_parser->lex_defer) +#define PL_lex_dojoin (PL_parser->lex_dojoin) +#define PL_lex_expect (PL_parser->lex_expect) +#define PL_lex_formbrack (PL_parser->lex_formbrack) +#define PL_lex_inpat (PL_parser->lex_inpat) +#define PL_lex_inwhat (PL_parser->lex_inwhat) +#define PL_lex_op (PL_parser->lex_op) +#define PL_lex_repl (PL_parser->lex_repl) +#define PL_lex_starts (PL_parser->lex_starts) +#define PL_lex_stuff (PL_parser->lex_stuff) +#define PL_multi_start (PL_parser->multi_start) +#define PL_multi_open (PL_parser->multi_open) +#define PL_multi_close (PL_parser->multi_close) +#define PL_pending_ident (PL_parser->pending_ident) +#define PL_preambled (PL_parser->preambled) +#define PL_sublex_info (PL_parser->sublex_info) +#define PL_linestr (PL_parser->linestr) +#define PL_sublex_info (PL_parser->sublex_info) +#define PL_linestr (PL_parser->linestr) +#define PL_expect (PL_parser->expect) +#define PL_copline (PL_parser->copline) +#define PL_bufptr (PL_parser->bufptr) +#define PL_oldbufptr (PL_parser->oldbufptr) +#define PL_oldoldbufptr (PL_parser->oldoldbufptr) +#define PL_linestart (PL_parser->linestart) +#define PL_bufend (PL_parser->bufend) +#define PL_last_uni (PL_parser->last_uni) +#define PL_last_lop (PL_parser->last_lop) +#define PL_last_lop_op (PL_parser->last_lop_op) +#define PL_lex_state (PL_parser->lex_state) +#define PL_rsfp (PL_parser->rsfp) +#define PL_rsfp_filters (PL_parser->rsfp_filters) +#define PL_in_my (PL_parser->in_my) +#define PL_in_my_stash (PL_parser->in_my_stash) +#define PL_tokenbuf (PL_parser->tokenbuf) +#define PL_multi_end (PL_parser->multi_end) +#define PL_error_count (PL_parser->error_count) +#define PL_nexttoke (PL_parser->nexttoke) +/* these are from the non-PERL_MAD path but I don't -think- I need + the PERL_MAD stuff since my code isn't really populating things (mst) */ +# ifdef PERL_MAD +# define PL_curforce (PL_parser->curforce) +# define PL_lasttoke (PL_parser->lasttoke) +# else +# define PL_nexttype (PL_parser->nexttype) +# define PL_nextval (PL_parser->nextval) +# endif +/* end of backcompat macros from 5.9 toke.c (mst) */ +#endif + +/* when ccflags include -DDEBUGGING we need this for earlier 5.8 perls */ +#ifndef SvPV_nolen_const +#define SvPV_nolen_const SvPV_nolen +#endif + +/* Name changed in 5.17; use new name in our code. Apparently we're meant + to use something else instead, but no non-underscored way to achieve + this is apparent. */ + +#ifndef _is_utf8_mark +#define _is_utf8_mark is_utf8_mark +#endif + +/* utf8_to_uvchr_buf() not defined in earlier perls, but less-capable + * substitute is available */ + +#ifndef utf8_to_uvchr_buf +#define utf8_to_uvchr_buf(s, e, lp) ((e), utf8_to_uvchr(s, lp)) +#endif + +/* and now we're back to the toke.c stuff again (mst) */ + static const char ident_too_long[] = "Identifier too long"; static const char c_without_g[] = @@ -139,7 +301,7 @@ S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append) */ STATIC char * -S_skipspace(pTHX_ register char *s) +S_skipspace(pTHX_ register char *s, int incline) { if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) { while (s < PL_bufend && SPACE_OR_TAB(*s)) @@ -151,7 +313,7 @@ S_skipspace(pTHX_ register char *s) SSize_t oldprevlen, oldoldprevlen; SSize_t oldloplen = 0, oldunilen = 0; while (s < PL_bufend && isSPACE(*s)) { - if (*s++ == '\n' && PL_in_eval && !PL_rsfp) + if (*s++ == '\n' && ((incline == 2) || (PL_in_eval && !PL_rsfp && !incline))) incline(s); } @@ -161,13 +323,21 @@ S_skipspace(pTHX_ register char *s) s++; if (s < PL_bufend) { s++; - if (PL_in_eval && !PL_rsfp) { + if (PL_in_eval && !PL_rsfp && !incline) { incline(s); continue; } } } + /* also skip leading whitespace on the beginning of a line before deciding + * whether or not to recharge the linestr. --rafl + */ + while (s < PL_bufend && isSPACE(*s)) { + if (*s++ == '\n' && PL_in_eval && !PL_rsfp && !incline) + incline(s); + } + /* only continue to recharge the buffer if we're at the end * of the buffer, we're not reading from a source filter, and * we're in normal lexing mode @@ -199,16 +369,20 @@ S_skipspace(pTHX_ register char *s) PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); PL_last_lop = PL_last_uni = Nullch; - /* Close the filehandle. Could be from -P preprocessor, + /* In perl versions previous to p4-rawid: //depot/perl@32954 -P + * preprocessors were supported here. We don't support -P at all, even + * on perls that support it, and use the following chunk from blead + * perl. (rafl) + */ + + /* Close the filehandle. Could be from * STDIN, or a regular file. If we were reading code from * STDIN (because the commandline held no -e or filename) * then we don't close it, we reset it so the code can * read from STDIN too. */ - if (PL_preprocess && !PL_in_eval) - (void)PerlProc_pclose(PL_rsfp); - else if ((PerlIO*)PL_rsfp == PerlIO_stdin()) + if ((PerlIO*)PL_rsfp == PerlIO_stdin()) PerlIO_clearerr(PL_rsfp); else (void)PerlIO_close(PL_rsfp); @@ -233,19 +407,22 @@ S_skipspace(pTHX_ register char *s) PL_last_uni = s + oldunilen; if (PL_last_lop) PL_last_lop = s + oldloplen; - incline(s); + if (!incline) + incline(s); /* debugger active and we're not compiling the debugger code, * so store the line into the debugger's array of lines */ if (PERLDB_LINE && PL_curstash != PL_debstash) { - SV * const sv = NEWSV(85,0); - - sv_upgrade(sv, SVt_PVMG); - sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr); - (void)SvIOK_on(sv); - SvIV_set(sv, 0); - av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv); + AV *fileav = CopFILEAV(PL_curcop); + if (fileav) { + SV * const sv = NEWSV(85,0); + sv_upgrade(sv, SVt_PVMG); + sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr); + (void)SvIOK_on(sv); + SvIV_set(sv, 0); + av_store(fileav,(I32)CopLINE(PL_curcop),sv); + } } } } @@ -271,7 +448,7 @@ S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_packag } else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) { char *t = s + UTF8SKIP(s); - while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t)) + while (UTF8_IS_CONTINUED(*t) && _is_utf8_mark((U8*)t)) t += UTF8SKIP(t); if (d + (t - s) > e) Perl_croak(aTHX_ ident_too_long); @@ -341,6 +518,12 @@ S_incline(pTHX_ char *s) ch = *t; *t = '\0'; if (t - s > 0) { +/* this chunk was added to S_incline during 5.8.8. I don't know why but I don't + honestly care since I probably want to be bug-compatible anyway (mst) */ + +/* ... my kingdom for a perl parser in perl ... (mst) */ + +#ifdef PERL_5_8_8_PLUS #ifndef USE_ITHREADS const char *cf = CopFILE(PL_curcop); if (cf && strlen(cf) > 7 && strnEQ(cf, "(eval ", 6)) { @@ -378,9 +561,510 @@ S_incline(pTHX_ char *s) if (tmpbuf2 != smallbuf2) Safefree(tmpbuf2); } #endif +#endif +/* second endif closes out the "are we 5.8.(8+)" conditional */ CopFILE_free(PL_curcop); CopFILE_set(PL_curcop, s); } *t = ch; CopLINE_set(PL_curcop, atoi(n)-1); } + +/* scan_str + takes: start position in buffer + keep_quoted preserve \ on the embedded delimiter(s) + keep_delims preserve the delimiters around the string + returns: position to continue reading from buffer + side-effects: multi_start, multi_close, lex_repl or lex_stuff, and + updates the read buffer. + + This subroutine pulls a string out of the input. It is called for: + q single quotes q(literal text) + ' single quotes 'literal text' + qq double quotes qq(interpolate $here please) + " double quotes "interpolate $here please" + qx backticks qx(/bin/ls -l) + ` backticks `/bin/ls -l` + qw quote words @EXPORT_OK = qw( func() $spam ) + m// regexp match m/this/ + s/// regexp substitute s/this/that/ + tr/// string transliterate tr/this/that/ + y/// string transliterate y/this/that/ + ($*@) sub prototypes sub foo ($) + (stuff) sub attr parameters sub foo : attr(stuff) + <> readline or globs , <>, <$fh>, or <*.c> + + In most of these cases (all but <>, patterns and transliterate) + yylex() calls scan_str(). m// makes yylex() call scan_pat() which + calls scan_str(). s/// makes yylex() call scan_subst() which calls + scan_str(). tr/// and y/// make yylex() call scan_trans() which + calls scan_str(). + + It skips whitespace before the string starts, and treats the first + character as the delimiter. If the delimiter is one of ([{< then + the corresponding "close" character )]}> is used as the closing + delimiter. It allows quoting of delimiters, and if the string has + balanced delimiters ([{<>}]) it allows nesting. + + On success, the SV with the resulting string is put into lex_stuff or, + if that is already non-NULL, into lex_repl. The second case occurs only + when parsing the RHS of the special constructs s/// and tr/// (y///). + For convenience, the terminating delimiter character is stuffed into + SvIVX of the SV. +*/ + +STATIC char * +S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims) +{ + SV *sv; /* scalar value: string */ + char *tmps; /* temp string, used for delimiter matching */ + register char *s = start; /* current position in the buffer */ + register char term; /* terminating character */ + register char *to; /* current position in the sv's data */ + I32 brackets = 1; /* bracket nesting level */ + bool has_utf8 = FALSE; /* is there any utf8 content? */ + I32 termcode; /* terminating char. code */ + /* 5.8.7+ uses UTF8_MAXBYTES but also its utf8.h defs _MAXLEN to it so + I'm reasonably hopeful this won't destroy anything (mst) */ + U8 termstr[UTF8_MAXLEN]; /* terminating string */ + STRLEN termlen; /* length of terminating string */ + char *last = NULL; /* last position for nesting bracket */ + + /* skip space before the delimiter */ + if (isSPACE(*s)) + s = skipspace(s); + + /* mark where we are, in case we need to report errors */ + CLINE; + + /* after skipping whitespace, the next character is the terminator */ + term = *s; + if (!UTF) { + termcode = termstr[0] = term; + termlen = 1; + } + else { + termcode = utf8_to_uvchr_buf((U8*)s, PL_bufend, &termlen); + Copy(s, termstr, termlen, U8); + if (!UTF8_IS_INVARIANT(term)) + has_utf8 = TRUE; + } + + /* mark where we are */ + PL_multi_start = CopLINE(PL_curcop); + PL_multi_open = term; + + /* find corresponding closing delimiter */ + if (term && (tmps = strchr("([{< )]}> )]}>",term))) + termcode = termstr[0] = term = tmps[5]; + + PL_multi_close = term; + + /* create a new SV to hold the contents. 87 is leak category, I'm + assuming. 79 is the SV's initial length. What a random number. */ + sv = NEWSV(87,79); + sv_upgrade(sv, SVt_PVIV); + SvIV_set(sv, termcode); + (void)SvPOK_only(sv); /* validate pointer */ + + /* move past delimiter and try to read a complete string */ + if (keep_delims) + sv_catpvn(sv, s, termlen); + s += termlen; + for (;;) { + if (PL_encoding && !UTF) { + bool cont = TRUE; + + while (cont) { + int offset = s - SvPVX_const(PL_linestr); + const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr, + &offset, (char*)termstr, termlen); + const char *ns = SvPVX_const(PL_linestr) + offset; + char *svlast = SvEND(sv) - 1; + + for (; s < ns; s++) { + if (*s == '\n' && !PL_rsfp) + CopLINE_inc(PL_curcop); + } + if (!found) + goto read_more_line; + else { + /* handle quoted delimiters */ + if (SvCUR(sv) > 1 && *(svlast-1) == '\\') { + const char *t; + for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';) + t--; + if ((svlast-1 - t) % 2) { + if (!keep_quoted) { + *(svlast-1) = term; + *svlast = '\0'; + SvCUR_set(sv, SvCUR(sv) - 1); + } + continue; + } + } + if (PL_multi_open == PL_multi_close) { + cont = FALSE; + } + else { + const char *t; + char *w; + if (!last) + last = SvPVX(sv); + for (t = w = last; t < svlast; w++, t++) { + /* At here, all closes are "was quoted" one, + so we don't check PL_multi_close. */ + if (*t == '\\') { + if (!keep_quoted && *(t+1) == PL_multi_open) + t++; + else + *w++ = *t++; + } + else if (*t == PL_multi_open) + brackets++; + + *w = *t; + } + if (w < t) { + *w++ = term; + *w = '\0'; + SvCUR_set(sv, w - SvPVX_const(sv)); + } + last = w; + if (--brackets <= 0) + cont = FALSE; + } + } + } + if (!keep_delims) { + SvCUR_set(sv, SvCUR(sv) - 1); + *SvEND(sv) = '\0'; + } + break; + } + + /* extend sv if need be */ + SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1); + /* set 'to' to the next character in the sv's string */ + to = SvPVX(sv)+SvCUR(sv); + + /* if open delimiter is the close delimiter read unbridle */ + if (PL_multi_open == PL_multi_close) { + for (; s < PL_bufend; s++,to++) { + /* embedded newlines increment the current line number */ + if (*s == '\n' && !PL_rsfp) + CopLINE_inc(PL_curcop); + /* handle quoted delimiters */ + if (*s == '\\' && s+1 < PL_bufend && term != '\\') { + if (!keep_quoted && s[1] == term) + s++; + /* any other quotes are simply copied straight through */ + else + *to++ = *s++; + } + /* terminate when run out of buffer (the for() condition), or + have found the terminator */ + else if (*s == term) { + if (termlen == 1) + break; + if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen)) + break; + } + else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF) + has_utf8 = TRUE; + *to = *s; + } + } + + /* if the terminator isn't the same as the start character (e.g., + matched brackets), we have to allow more in the quoting, and + be prepared for nested brackets. + */ + else { + /* read until we run out of string, or we find the terminator */ + for (; s < PL_bufend; s++,to++) { + /* embedded newlines increment the line count */ + if (*s == '\n' && !PL_rsfp) + CopLINE_inc(PL_curcop); + /* backslashes can escape the open or closing characters */ + if (*s == '\\' && s+1 < PL_bufend) { + if (!keep_quoted && + ((s[1] == PL_multi_open) || (s[1] == PL_multi_close))) + s++; + else + *to++ = *s++; + } + /* allow nested opens and closes */ + else if (*s == PL_multi_close && --brackets <= 0) + break; + else if (*s == PL_multi_open) + brackets++; + else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF) + has_utf8 = TRUE; + *to = *s; + } + } + /* terminate the copied string and update the sv's end-of-string */ + *to = '\0'; + SvCUR_set(sv, to - SvPVX_const(sv)); + + /* + * this next chunk reads more into the buffer if we're not done yet + */ + + if (s < PL_bufend) + break; /* handle case where we are done yet :-) */ + +#ifndef PERL_STRICT_CR + if (to - SvPVX_const(sv) >= 2) { + if ((to[-2] == '\r' && to[-1] == '\n') || + (to[-2] == '\n' && to[-1] == '\r')) + { + to[-2] = '\n'; + to--; + SvCUR_set(sv, to - SvPVX_const(sv)); + } + else if (to[-1] == '\r') + to[-1] = '\n'; + } + else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r') + to[-1] = '\n'; +#endif + + read_more_line: + /* if we're out of file, or a read fails, bail and reset the current + line marker so we can report where the unterminated string began + */ + if (!PL_rsfp || + !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) { + sv_free(sv); + CopLINE_set(PL_curcop, (line_t)PL_multi_start); + return Nullch; + } + /* we read a line, so increment our line counter */ + CopLINE_inc(PL_curcop); + + /* update debugger info */ + if (PERLDB_LINE && PL_curstash != PL_debstash) { + AV *fileav = CopFILEAV(PL_curcop); + if (fileav) { + SV *sv = NEWSV(88,0); + sv_upgrade(sv, SVt_PVMG); + sv_setsv(sv,PL_linestr); + (void)SvIOK_on(sv); + SvIV_set(sv, 0); + av_store(fileav, (I32)CopLINE(PL_curcop), sv); + } + } + + /* having changed the buffer, we must update PL_bufend */ + PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); + PL_last_lop = PL_last_uni = Nullch; + } + + /* at this point, we have successfully read the delimited string */ + + if (!PL_encoding || UTF) { + if (keep_delims) + sv_catpvn(sv, s, termlen); + s += termlen; + } + if (has_utf8 || PL_encoding) + SvUTF8_on(sv); + + PL_multi_end = CopLINE(PL_curcop); + + /* if we allocated too much space, give some back */ + if (SvCUR(sv) + 5 < SvLEN(sv)) { + SvLEN_set(sv, SvCUR(sv) + 1); +/* 5.8.8 uses SvPV_renew, no prior version actually has the damn thing (mst) */ +#ifdef PERL_5_8_8_PLUS + SvPV_renew(sv, SvLEN(sv)); +#else + Renew(SvPVX(sv), SvLEN(sv), char); +#endif + } + + /* decide whether this is the first or second quoted string we've read + for this op + */ + + if (PL_lex_stuff) + PL_lex_repl = sv; + else + PL_lex_stuff = sv; + return s; +} + +#define XFAKEBRACK 128 + +STATIC char * +S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni) +{ + register char *d; + register char *e; + char *bracket = Nullch; + char funny = *s++; + + if (isSPACE(*s)) + s = skipspace(s); + d = dest; + e = d + destlen - 3; /* two-character token, ending NUL */ + if (isDIGIT(*s)) { + while (isDIGIT(*s)) { + if (d >= e) + Perl_croak(aTHX_ ident_too_long); + *d++ = *s++; + } + } + else { + for (;;) { + if (d >= e) + Perl_croak(aTHX_ ident_too_long); + if (isALNUM(*s)) /* UTF handled below */ + *d++ = *s++; + else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) { + *d++ = ':'; + *d++ = ':'; + s++; + } + else if (*s == ':' && s[1] == ':') { + *d++ = *s++; + *d++ = *s++; + } + else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) { + char *t = s + UTF8SKIP(s); + while (UTF8_IS_CONTINUED(*t) && _is_utf8_mark((U8*)t)) + t += UTF8SKIP(t); + if (d + (t - s) > e) + Perl_croak(aTHX_ ident_too_long); + Copy(s, d, t - s, char); + d += t - s; + s = t; + } + else + break; + } + } + *d = '\0'; + d = dest; + if (*d) { + if (PL_lex_state != LEX_NORMAL) + PL_lex_state = LEX_INTERPENDMAYBE; + return s; + } + if (*s == '$' && s[1] && + (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) ) + { + return s; + } + if (*s == '{') { + bracket = s; + s++; + } else if (ck_uni) { + /* we always call this with ck_uni == 0, so no need for check_uni() */ + /* check_uni(); */ + } + if (s < send) + *d = *s++; + d[1] = '\0'; + if (*d == '^' && *s && isCONTROLVAR(*s)) { + *d = toCTRL(*s); + s++; + } + if (bracket) { + if (isSPACE(s[-1])) { + while (s < send) { + const char ch = *s++; + if (!SPACE_OR_TAB(ch)) { + *d = ch; + break; + } + } + } + if (isIDFIRST_lazy_if(d,UTF)) { + d++; + if (UTF) { + e = s; + while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') { + e += UTF8SKIP(e); + while (e < send && UTF8_IS_CONTINUED(*e) && _is_utf8_mark((U8*)e)) + e += UTF8SKIP(e); + } + Copy(s, d, e - s, char); + d += e - s; + s = e; + } + else { + while ((isALNUM(*s) || *s == ':') && d < e) + *d++ = *s++; + if (d >= e) + Perl_croak(aTHX_ ident_too_long); + } + *d = '\0'; + while (s < send && SPACE_OR_TAB(*s)) s++; + if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) { + /* we don't want perl to guess what is meant. the keyword + * parser decides that later. (rafl) + */ + /* + if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) { + const char *brack = *s == '[' ? "[...]" : "{...}"; + Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS), + "Ambiguous use of %c{%s%s} resolved to %c%s%s", + funny, dest, brack, funny, dest, brack); + } + */ + bracket++; + PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK); + return s; + } + } + /* Handle extended ${^Foo} variables + * 1999-02-27 mjd-perl-patch@plover.com */ + else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */ + && isALNUM(*s)) + { + d++; + while (isALNUM(*s) && d < e) { + *d++ = *s++; + } + if (d >= e) + Perl_croak(aTHX_ ident_too_long); + *d = '\0'; + } + if (*s == '}') { + s++; + if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) { + PL_lex_state = LEX_INTERPEND; + PL_expect = XREF; + } + if (funny == '#') + funny = '@'; + /* we don't want perl to guess what is meant. the keyword + * parser decides that later. (rafl) + */ + /* + if (PL_lex_state == LEX_NORMAL) { + if (ckWARN(WARN_AMBIGUOUS) && + (keyword(dest, d - dest) || get_cv(dest, FALSE))) + { + Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS), + "Ambiguous use of %c{%s} resolved to %c%s", + funny, dest, funny, dest); + } + } + */ + } + else { + s = bracket; /* let the parser handle it */ + *dest = '\0'; + } + } + /* don't intuit. we really just want the string. (rafl) */ + /* + else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s)) + PL_lex_state = LEX_INTERPEND; + */ + return s; +}