3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
4 * 2000, 2001, 2002, 2003, 2004, 2005, by Larry Wall and others
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
12 * "It all comes from here, the stench and the peril." --Frodo
16 * This file is the lexer for Perl. It's closely linked to the
19 * The main routine is yylex(), which returns the next token.
23 #define PERL_IN_TOKE_C
26 #define yychar (*PL_yycharp)
27 #define yylval (*PL_yylvalp)
29 static const char ident_too_long[] = "Identifier too long";
31 static void restore_rsfp(pTHX_ void *f);
32 #ifndef PERL_NO_UTF16_FILTER
33 static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen);
34 static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen);
37 #define XFAKEBRACK 128
40 #ifdef USE_UTF8_SCRIPTS
41 # define UTF (!IN_BYTES)
43 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
46 /* In variables named $^X, these are the legal values for X.
47 * 1999-02-27 mjd-perl-patch@plover.com */
48 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
50 /* On MacOS, respect nonbreaking spaces */
51 #ifdef MACOS_TRADITIONAL
52 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
54 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
57 /* LEX_* are values for PL_lex_state, the state of the lexer.
58 * They are arranged oddly so that the guard on the switch statement
59 * can get by with a single comparison (if the compiler is smart enough).
62 /* #define LEX_NOTPARSING 11 is done in perl.h. */
64 #define LEX_NORMAL 10 /* normal code (ie not within "...") */
65 #define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
66 #define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
67 #define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
68 #define LEX_INTERPSTART 6 /* expecting the start of a $var */
70 /* at end of code, eg "$x" followed by: */
71 #define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
72 #define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
74 #define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
75 string or after \E, $foo, etc */
76 #define LEX_INTERPCONST 2 /* NOT USED */
77 #define LEX_FORMLINE 1 /* expecting a format line */
78 #define LEX_KNOWNEXT 0 /* next token known; just return it */
82 static const char* const lex_state_names[] = {
101 #include "keywords.h"
103 /* CLINE is a macro that ensures PL_copline has a sane value */
108 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
111 * Convenience functions to return different tokens and prime the
112 * lexer for the next token. They all take an argument.
114 * TOKEN : generic token (used for '(', DOLSHARP, etc)
115 * OPERATOR : generic operator
116 * AOPERATOR : assignment operator
117 * PREBLOCK : beginning the block after an if, while, foreach, ...
118 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
119 * PREREF : *EXPR where EXPR is not a simple identifier
120 * TERM : expression term
121 * LOOPX : loop exiting command (goto, last, dump, etc)
122 * FTST : file test operator
123 * FUN0 : zero-argument function
124 * FUN1 : not used, except for not, which isn't a UNIOP
125 * BOop : bitwise or or xor
127 * SHop : shift operator
128 * PWop : power operator
129 * PMop : pattern-matching operator
130 * Aop : addition-level operator
131 * Mop : multiplication-level operator
132 * Eop : equality-testing operator
133 * Rop : relational operator <= != gt
135 * Also see LOP and lop() below.
138 #ifdef DEBUGGING /* Serve -DT. */
139 # define REPORT(retval) tokereport((I32)retval)
141 # define REPORT(retval) (retval)
144 #define TOKEN(retval) return ( PL_bufptr = s, REPORT(retval))
145 #define OPERATOR(retval) return (PL_expect = XTERM, PL_bufptr = s, REPORT(retval))
146 #define AOPERATOR(retval) return ao((PL_expect = XTERM, PL_bufptr = s, REPORT(retval)))
147 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s, REPORT(retval))
148 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s, REPORT(retval))
149 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s, REPORT(retval))
150 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR, PL_bufptr = s, REPORT(retval))
151 #define LOOPX(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)LOOPEX))
152 #define FTST(f) return (yylval.ival=f, PL_expect=XTERMORDORDOR, PL_bufptr=s, REPORT((int)UNIOP))
153 #define FUN0(f) return (yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0))
154 #define FUN1(f) return (yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC1))
155 #define BOop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITOROP)))
156 #define BAop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITANDOP)))
157 #define SHop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)SHIFTOP)))
158 #define PWop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)POWOP)))
159 #define PMop(f) return(yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MATCHOP))
160 #define Aop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)ADDOP)))
161 #define Mop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MULOP)))
162 #define Eop(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)EQOP))
163 #define Rop(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)RELOP))
165 /* This bit of chicanery makes a unary function followed by
166 * a parenthesis into a function with one argument, highest precedence.
167 * The UNIDOR macro is for unary functions that can be followed by the //
168 * operator (such as C<shift // 0>).
170 #define UNI2(f,x) { \
174 PL_last_uni = PL_oldbufptr; \
175 PL_last_lop_op = f; \
177 return REPORT( (int)FUNC1 ); \
179 return REPORT( *s=='(' ? (int)FUNC1 : (int)UNIOP ); \
181 #define UNI(f) UNI2(f,XTERM)
182 #define UNIDOR(f) UNI2(f,XTERMORDORDOR)
184 #define UNIBRACK(f) { \
187 PL_last_uni = PL_oldbufptr; \
189 return REPORT( (int)FUNC1 ); \
191 return REPORT( (*s == '(') ? (int)FUNC1 : (int)UNIOP ); \
194 /* grandfather return to old style */
195 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
199 /* how to interpret the yylval associated with the token */
203 TOKENTYPE_OPNUM, /* yylval.ival contains an opcode number */
209 static struct debug_tokens { const int token, type; const char *name; }
210 const debug_tokens[] =
212 { ADDOP, TOKENTYPE_OPNUM, "ADDOP" },
213 { ANDAND, TOKENTYPE_NONE, "ANDAND" },
214 { ANDOP, TOKENTYPE_NONE, "ANDOP" },
215 { ANONSUB, TOKENTYPE_IVAL, "ANONSUB" },
216 { ARROW, TOKENTYPE_NONE, "ARROW" },
217 { ASSIGNOP, TOKENTYPE_OPNUM, "ASSIGNOP" },
218 { BITANDOP, TOKENTYPE_OPNUM, "BITANDOP" },
219 { BITOROP, TOKENTYPE_OPNUM, "BITOROP" },
220 { COLONATTR, TOKENTYPE_NONE, "COLONATTR" },
221 { CONTINUE, TOKENTYPE_NONE, "CONTINUE" },
222 { DEFAULT, TOKENTYPE_NONE, "DEFAULT" },
223 { DO, TOKENTYPE_NONE, "DO" },
224 { DOLSHARP, TOKENTYPE_NONE, "DOLSHARP" },
225 { DORDOR, TOKENTYPE_NONE, "DORDOR" },
226 { DOROP, TOKENTYPE_OPNUM, "DOROP" },
227 { DOTDOT, TOKENTYPE_IVAL, "DOTDOT" },
228 { ELSE, TOKENTYPE_NONE, "ELSE" },
229 { ELSIF, TOKENTYPE_IVAL, "ELSIF" },
230 { EQOP, TOKENTYPE_OPNUM, "EQOP" },
231 { FOR, TOKENTYPE_IVAL, "FOR" },
232 { FORMAT, TOKENTYPE_NONE, "FORMAT" },
233 { FUNC, TOKENTYPE_OPNUM, "FUNC" },
234 { FUNC0, TOKENTYPE_OPNUM, "FUNC0" },
235 { FUNC0SUB, TOKENTYPE_OPVAL, "FUNC0SUB" },
236 { FUNC1, TOKENTYPE_OPNUM, "FUNC1" },
237 { FUNCMETH, TOKENTYPE_OPVAL, "FUNCMETH" },
238 { GIVEN, TOKENTYPE_IVAL, "GIVEN" },
239 { HASHBRACK, TOKENTYPE_NONE, "HASHBRACK" },
240 { IF, TOKENTYPE_IVAL, "IF" },
241 { LABEL, TOKENTYPE_PVAL, "LABEL" },
242 { LOCAL, TOKENTYPE_IVAL, "LOCAL" },
243 { LOOPEX, TOKENTYPE_OPNUM, "LOOPEX" },
244 { LSTOP, TOKENTYPE_OPNUM, "LSTOP" },
245 { LSTOPSUB, TOKENTYPE_OPVAL, "LSTOPSUB" },
246 { MATCHOP, TOKENTYPE_OPNUM, "MATCHOP" },
247 { METHOD, TOKENTYPE_OPVAL, "METHOD" },
248 { MULOP, TOKENTYPE_OPNUM, "MULOP" },
249 { MY, TOKENTYPE_IVAL, "MY" },
250 { MYSUB, TOKENTYPE_NONE, "MYSUB" },
251 { NOAMP, TOKENTYPE_NONE, "NOAMP" },
252 { NOTOP, TOKENTYPE_NONE, "NOTOP" },
253 { OROP, TOKENTYPE_IVAL, "OROP" },
254 { OROR, TOKENTYPE_NONE, "OROR" },
255 { PACKAGE, TOKENTYPE_NONE, "PACKAGE" },
256 { PMFUNC, TOKENTYPE_OPVAL, "PMFUNC" },
257 { POSTDEC, TOKENTYPE_NONE, "POSTDEC" },
258 { POSTINC, TOKENTYPE_NONE, "POSTINC" },
259 { POWOP, TOKENTYPE_OPNUM, "POWOP" },
260 { PREDEC, TOKENTYPE_NONE, "PREDEC" },
261 { PREINC, TOKENTYPE_NONE, "PREINC" },
262 { PRIVATEREF, TOKENTYPE_OPVAL, "PRIVATEREF" },
263 { REFGEN, TOKENTYPE_NONE, "REFGEN" },
264 { RELOP, TOKENTYPE_OPNUM, "RELOP" },
265 { SHIFTOP, TOKENTYPE_OPNUM, "SHIFTOP" },
266 { SUB, TOKENTYPE_NONE, "SUB" },
267 { THING, TOKENTYPE_OPVAL, "THING" },
268 { UMINUS, TOKENTYPE_NONE, "UMINUS" },
269 { UNIOP, TOKENTYPE_OPNUM, "UNIOP" },
270 { UNIOPSUB, TOKENTYPE_OPVAL, "UNIOPSUB" },
271 { UNLESS, TOKENTYPE_IVAL, "UNLESS" },
272 { UNTIL, TOKENTYPE_IVAL, "UNTIL" },
273 { USE, TOKENTYPE_IVAL, "USE" },
274 { WHEN, TOKENTYPE_IVAL, "WHEN" },
275 { WHILE, TOKENTYPE_IVAL, "WHILE" },
276 { WORD, TOKENTYPE_OPVAL, "WORD" },
277 { 0, TOKENTYPE_NONE, 0 }
280 /* dump the returned token in rv, plus any optional arg in yylval */
283 S_tokereport(pTHX_ I32 rv)
286 const char *name = Nullch;
287 enum token_type type = TOKENTYPE_NONE;
288 const struct debug_tokens *p;
289 SV* const report = newSVpvn("<== ", 4);
291 for (p = debug_tokens; p->token; p++) {
292 if (p->token == (int)rv) {
299 Perl_sv_catpv(aTHX_ report, name);
300 else if ((char)rv > ' ' && (char)rv < '~')
301 Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
303 Perl_sv_catpv(aTHX_ report, "EOF");
305 Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
308 case TOKENTYPE_GVVAL: /* doesn't appear to be used */
311 Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", (IV)yylval.ival);
313 case TOKENTYPE_OPNUM:
314 Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
315 PL_op_name[yylval.ival]);
318 Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", yylval.pval);
320 case TOKENTYPE_OPVAL:
322 Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
323 PL_op_name[yylval.opval->op_type]);
324 if (yylval.opval->op_type == OP_CONST) {
325 Perl_sv_catpvf(aTHX_ report, " %s",
326 SvPEEK(cSVOPx_sv(yylval.opval)));
331 Perl_sv_catpv(aTHX_ report, "(opval=null)");
334 PerlIO_printf(Perl_debug_log, "### %s\n\n", SvPV_nolen_const(report));
340 /* print the buffer with suitable escapes */
343 S_printbuf(pTHX_ const char* fmt, const char* s)
345 SV* const tmp = newSVpvn("", 0);
346 PerlIO_printf(Perl_debug_log, fmt, pv_display(tmp, s, strlen(s), 0, 60));
355 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
356 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
360 S_ao(pTHX_ int toketype)
362 if (*PL_bufptr == '=') {
364 if (toketype == ANDAND)
365 yylval.ival = OP_ANDASSIGN;
366 else if (toketype == OROR)
367 yylval.ival = OP_ORASSIGN;
368 else if (toketype == DORDOR)
369 yylval.ival = OP_DORASSIGN;
377 * When Perl expects an operator and finds something else, no_op
378 * prints the warning. It always prints "<something> found where
379 * operator expected. It prints "Missing semicolon on previous line?"
380 * if the surprise occurs at the start of the line. "do you need to
381 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
382 * where the compiler doesn't know if foo is a method call or a function.
383 * It prints "Missing operator before end of line" if there's nothing
384 * after the missing operator, or "... before <...>" if there is something
385 * after the missing operator.
389 S_no_op(pTHX_ const char *what, char *s)
391 char * const oldbp = PL_bufptr;
392 const bool is_first = (PL_oldbufptr == PL_linestart);
398 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
399 if (ckWARN_d(WARN_SYNTAX)) {
401 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
402 "\t(Missing semicolon on previous line?)\n");
403 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
405 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
406 if (t < PL_bufptr && isSPACE(*t))
407 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
408 "\t(Do you need to predeclare %.*s?)\n",
409 (int)(t - PL_oldoldbufptr), PL_oldoldbufptr);
413 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
414 "\t(Missing operator before %.*s?)\n", (int)(s - oldbp), oldbp);
422 * Complain about missing quote/regexp/heredoc terminator.
423 * If it's called with (char *)NULL then it cauterizes the line buffer.
424 * If we're in a delimited string and the delimiter is a control
425 * character, it's reformatted into a two-char sequence like ^C.
430 S_missingterm(pTHX_ char *s)
435 char * const nl = strrchr(s,'\n');
441 iscntrl(PL_multi_close)
443 PL_multi_close < 32 || PL_multi_close == 127
447 tmpbuf[1] = (char)toCTRL(PL_multi_close);
452 *tmpbuf = (char)PL_multi_close;
456 q = strchr(s,'"') ? '\'' : '"';
457 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
460 #define FEATURE_IS_ENABLED(name, namelen) \
461 ((0 != (PL_hints & HINT_LOCALIZE_HH)) \
462 && feature_is_enabled(name, namelen))
464 * S_feature_is_enabled
465 * Check whether the named feature is enabled.
468 S_feature_is_enabled(pTHX_ char *name, STRLEN namelen)
470 HV * const hinthv = GvHV(PL_hintgv);
471 char he_name[32] = "feature_";
472 (void) strncpy(&he_name[8], name, 24);
474 return (hinthv && hv_exists(hinthv, he_name, 8 + namelen));
482 Perl_deprecate(pTHX_ const char *s)
484 if (ckWARN(WARN_DEPRECATED))
485 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
489 Perl_deprecate_old(pTHX_ const char *s)
491 /* This function should NOT be called for any new deprecated warnings */
492 /* Use Perl_deprecate instead */
494 /* It is here to maintain backward compatibility with the pre-5.8 */
495 /* warnings category hierarchy. The "deprecated" category used to */
496 /* live under the "syntax" category. It is now a top-level category */
497 /* in its own right. */
499 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
500 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
501 "Use of %s is deprecated", s);
506 * Deprecate a comma-less variable list.
512 deprecate_old("comma-less variable list");
516 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
517 * utf16-to-utf8-reversed.
520 #ifdef PERL_CR_FILTER
524 register const char *s = SvPVX_const(sv);
525 register const char * const e = s + SvCUR(sv);
526 /* outer loop optimized to do nothing if there are no CR-LFs */
528 if (*s++ == '\r' && *s == '\n') {
529 /* hit a CR-LF, need to copy the rest */
530 register char *d = s - 1;
533 if (*s == '\r' && s[1] == '\n')
544 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
546 const I32 count = FILTER_READ(idx+1, sv, maxlen);
547 if (count > 0 && !maxlen)
555 * Initialize variables. Uses the Perl save_stack to save its state (for
556 * recursive calls to the parser).
560 Perl_lex_start(pTHX_ SV *line)
565 SAVEI32(PL_lex_dojoin);
566 SAVEI32(PL_lex_brackets);
567 SAVEI32(PL_lex_casemods);
568 SAVEI32(PL_lex_starts);
569 SAVEI32(PL_lex_state);
570 SAVEVPTR(PL_lex_inpat);
571 SAVEI32(PL_lex_inwhat);
572 if (PL_lex_state == LEX_KNOWNEXT) {
573 I32 toke = PL_nexttoke;
574 while (--toke >= 0) {
575 SAVEI32(PL_nexttype[toke]);
576 SAVEVPTR(PL_nextval[toke]);
578 SAVEI32(PL_nexttoke);
580 SAVECOPLINE(PL_curcop);
583 SAVEPPTR(PL_oldbufptr);
584 SAVEPPTR(PL_oldoldbufptr);
585 SAVEPPTR(PL_last_lop);
586 SAVEPPTR(PL_last_uni);
587 SAVEPPTR(PL_linestart);
588 SAVESPTR(PL_linestr);
589 SAVEGENERICPV(PL_lex_brackstack);
590 SAVEGENERICPV(PL_lex_casestack);
591 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
592 SAVESPTR(PL_lex_stuff);
593 SAVEI32(PL_lex_defer);
594 SAVEI32(PL_sublex_info.sub_inwhat);
595 SAVESPTR(PL_lex_repl);
597 SAVEINT(PL_lex_expect);
599 PL_lex_state = LEX_NORMAL;
603 Newx(PL_lex_brackstack, 120, char);
604 Newx(PL_lex_casestack, 12, char);
606 *PL_lex_casestack = '\0';
609 PL_lex_stuff = Nullsv;
610 PL_lex_repl = Nullsv;
614 PL_sublex_info.sub_inwhat = 0;
616 if (SvREADONLY(PL_linestr))
617 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
618 s = SvPV_const(PL_linestr, len);
619 if (!len || s[len-1] != ';') {
620 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
621 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
622 sv_catpvn(PL_linestr, "\n;", 2);
624 SvTEMP_off(PL_linestr);
625 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
626 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
627 PL_last_lop = PL_last_uni = Nullch;
633 * Finalizer for lexing operations. Must be called when the parser is
634 * done with the lexer.
640 PL_doextract = FALSE;
645 * This subroutine has nothing to do with tilting, whether at windmills
646 * or pinball tables. Its name is short for "increment line". It
647 * increments the current line number in CopLINE(PL_curcop) and checks
648 * to see whether the line starts with a comment of the form
649 * # line 500 "foo.pm"
650 * If so, it sets the current line number and file to the values in the comment.
654 S_incline(pTHX_ char *s)
661 CopLINE_inc(PL_curcop);
664 while (SPACE_OR_TAB(*s)) s++;
665 if (strnEQ(s, "line", 4))
669 if (SPACE_OR_TAB(*s))
673 while (SPACE_OR_TAB(*s)) s++;
679 while (SPACE_OR_TAB(*s))
681 if (*s == '"' && (t = strchr(s+1, '"'))) {
686 for (t = s; !isSPACE(*t); t++) ;
689 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
691 if (*e != '\n' && *e != '\0')
692 return; /* false alarm */
698 const char * const cf = CopFILE(PL_curcop);
699 STRLEN tmplen = cf ? strlen(cf) : 0;
700 if (tmplen > 7 && strnEQ(cf, "(eval ", 6)) {
701 /* must copy *{"::_<(eval N)[oldfilename:L]"}
702 * to *{"::_<newfilename"} */
703 char smallbuf[256], smallbuf2[256];
704 char *tmpbuf, *tmpbuf2;
706 STRLEN tmplen2 = strlen(s);
707 if (tmplen + 3 < sizeof smallbuf)
710 Newx(tmpbuf, tmplen + 3, char);
711 if (tmplen2 + 3 < sizeof smallbuf2)
714 Newx(tmpbuf2, tmplen2 + 3, char);
715 tmpbuf[0] = tmpbuf2[0] = '_';
716 tmpbuf[1] = tmpbuf2[1] = '<';
717 memcpy(tmpbuf + 2, cf, ++tmplen);
718 memcpy(tmpbuf2 + 2, s, ++tmplen2);
720 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
722 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
724 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
725 /* adjust ${"::_<newfilename"} to store the new file name */
726 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
727 GvHV(gv2) = (HV*)SvREFCNT_inc(GvHV(*gvp));
728 GvAV(gv2) = (AV*)SvREFCNT_inc(GvAV(*gvp));
730 if (tmpbuf != smallbuf) Safefree(tmpbuf);
731 if (tmpbuf2 != smallbuf2) Safefree(tmpbuf2);
734 CopFILE_free(PL_curcop);
735 CopFILE_set(PL_curcop, s);
738 CopLINE_set(PL_curcop, atoi(n)-1);
743 * Called to gobble the appropriate amount and type of whitespace.
744 * Skips comments as well.
748 S_skipspace(pTHX_ register char *s)
750 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
751 while (s < PL_bufend && SPACE_OR_TAB(*s))
757 SSize_t oldprevlen, oldoldprevlen;
758 SSize_t oldloplen = 0, oldunilen = 0;
759 while (s < PL_bufend && isSPACE(*s)) {
760 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
765 if (s < PL_bufend && *s == '#') {
766 while (s < PL_bufend && *s != '\n')
770 if (PL_in_eval && !PL_rsfp) {
777 /* only continue to recharge the buffer if we're at the end
778 * of the buffer, we're not reading from a source filter, and
779 * we're in normal lexing mode
781 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
782 PL_lex_state == LEX_FORMLINE)
785 /* try to recharge the buffer */
786 if ((s = filter_gets(PL_linestr, PL_rsfp,
787 (prevlen = SvCUR(PL_linestr)))) == Nullch)
789 /* end of file. Add on the -p or -n magic */
792 ";}continue{print or die qq(-p destination: $!\\n);}");
793 PL_minus_n = PL_minus_p = 0;
795 else if (PL_minus_n) {
796 sv_setpvn(PL_linestr, ";}", 2);
800 sv_setpvn(PL_linestr,";", 1);
802 /* reset variables for next time we lex */
803 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
805 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
806 PL_last_lop = PL_last_uni = Nullch;
808 /* Close the filehandle. Could be from -P preprocessor,
809 * STDIN, or a regular file. If we were reading code from
810 * STDIN (because the commandline held no -e or filename)
811 * then we don't close it, we reset it so the code can
812 * read from STDIN too.
815 if (PL_preprocess && !PL_in_eval)
816 (void)PerlProc_pclose(PL_rsfp);
817 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
818 PerlIO_clearerr(PL_rsfp);
820 (void)PerlIO_close(PL_rsfp);
825 /* not at end of file, so we only read another line */
826 /* make corresponding updates to old pointers, for yyerror() */
827 oldprevlen = PL_oldbufptr - PL_bufend;
828 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
830 oldunilen = PL_last_uni - PL_bufend;
832 oldloplen = PL_last_lop - PL_bufend;
833 PL_linestart = PL_bufptr = s + prevlen;
834 PL_bufend = s + SvCUR(PL_linestr);
836 PL_oldbufptr = s + oldprevlen;
837 PL_oldoldbufptr = s + oldoldprevlen;
839 PL_last_uni = s + oldunilen;
841 PL_last_lop = s + oldloplen;
844 /* debugger active and we're not compiling the debugger code,
845 * so store the line into the debugger's array of lines
847 if (PERLDB_LINE && PL_curstash != PL_debstash) {
848 SV * const sv = NEWSV(85,0);
850 sv_upgrade(sv, SVt_PVMG);
851 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
854 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
861 * Check the unary operators to ensure there's no ambiguity in how they're
862 * used. An ambiguous piece of code would be:
864 * This doesn't mean rand() + 5. Because rand() is a unary operator,
865 * the +5 is its argument.
874 if (PL_oldoldbufptr != PL_last_uni)
876 while (isSPACE(*PL_last_uni))
878 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
879 if ((t = strchr(s, '(')) && t < PL_bufptr)
881 if (ckWARN_d(WARN_AMBIGUOUS)){
884 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
885 "Warning: Use of \"%s\" without parentheses is ambiguous",
892 * LOP : macro to build a list operator. Its behaviour has been replaced
893 * with a subroutine, S_lop() for which LOP is just another name.
896 #define LOP(f,x) return lop(f,x,s)
900 * Build a list operator (or something that might be one). The rules:
901 * - if we have a next token, then it's a list operator [why?]
902 * - if the next thing is an opening paren, then it's a function
903 * - else it's a list operator
907 S_lop(pTHX_ I32 f, int x, char *s)
913 PL_last_lop = PL_oldbufptr;
914 PL_last_lop_op = (OPCODE)f;
916 return REPORT(LSTOP);
923 return REPORT(LSTOP);
928 * When the lexer realizes it knows the next token (for instance,
929 * it is reordering tokens for the parser) then it can call S_force_next
930 * to know what token to return the next time the lexer is called. Caller
931 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
932 * handles the token correctly.
936 S_force_next(pTHX_ I32 type)
938 PL_nexttype[PL_nexttoke] = type;
940 if (PL_lex_state != LEX_KNOWNEXT) {
941 PL_lex_defer = PL_lex_state;
942 PL_lex_expect = PL_expect;
943 PL_lex_state = LEX_KNOWNEXT;
948 S_newSV_maybe_utf8(pTHX_ const char *start, STRLEN len)
950 SV * const sv = newSVpvn(start,len);
951 if (UTF && !IN_BYTES && is_utf8_string((const U8*)start, len))
958 * When the lexer knows the next thing is a word (for instance, it has
959 * just seen -> and it knows that the next char is a word char, then
960 * it calls S_force_word to stick the next word into the PL_next lookahead.
963 * char *start : buffer position (must be within PL_linestr)
964 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
965 * int check_keyword : if true, Perl checks to make sure the word isn't
966 * a keyword (do this if the word is a label, e.g. goto FOO)
967 * int allow_pack : if true, : characters will also be allowed (require,
969 * int allow_initial_tick : used by the "sub" lexer only.
973 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
978 start = skipspace(start);
980 if (isIDFIRST_lazy_if(s,UTF) ||
981 (allow_pack && *s == ':') ||
982 (allow_initial_tick && *s == '\'') )
984 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
985 if (check_keyword && keyword(PL_tokenbuf, len))
987 if (token == METHOD) {
992 PL_expect = XOPERATOR;
995 PL_nextval[PL_nexttoke].opval
996 = (OP*)newSVOP(OP_CONST,0,
997 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
998 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
1006 * Called when the lexer wants $foo *foo &foo etc, but the program
1007 * text only contains the "foo" portion. The first argument is a pointer
1008 * to the "foo", and the second argument is the type symbol to prefix.
1009 * Forces the next token to be a "WORD".
1010 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
1014 S_force_ident(pTHX_ register const char *s, int kind)
1017 OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
1018 PL_nextval[PL_nexttoke].opval = o;
1021 o->op_private = OPpCONST_ENTERED;
1022 /* XXX see note in pp_entereval() for why we forgo typo
1023 warnings if the symbol must be introduced in an eval.
1025 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
1026 kind == '$' ? SVt_PV :
1027 kind == '@' ? SVt_PVAV :
1028 kind == '%' ? SVt_PVHV :
1036 Perl_str_to_version(pTHX_ SV *sv)
1041 const char *start = SvPV_const(sv,len);
1042 const char * const end = start + len;
1043 const bool utf = SvUTF8(sv) ? TRUE : FALSE;
1044 while (start < end) {
1048 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
1053 retval += ((NV)n)/nshift;
1062 * Forces the next token to be a version number.
1063 * If the next token appears to be an invalid version number, (e.g. "v2b"),
1064 * and if "guessing" is TRUE, then no new token is created (and the caller
1065 * must use an alternative parsing method).
1069 S_force_version(pTHX_ char *s, int guessing)
1071 OP *version = Nullop;
1080 while (isDIGIT(*d) || *d == '_' || *d == '.')
1082 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
1084 s = scan_num(s, &yylval);
1085 version = yylval.opval;
1086 ver = cSVOPx(version)->op_sv;
1087 if (SvPOK(ver) && !SvNIOK(ver)) {
1088 SvUPGRADE(ver, SVt_PVNV);
1089 SvNV_set(ver, str_to_version(ver));
1090 SvNOK_on(ver); /* hint that it is a version */
1097 /* NOTE: The parser sees the package name and the VERSION swapped */
1098 PL_nextval[PL_nexttoke].opval = version;
1106 * Tokenize a quoted string passed in as an SV. It finds the next
1107 * chunk, up to end of string or a backslash. It may make a new
1108 * SV containing that chunk (if HINT_NEW_STRING is on). It also
1113 S_tokeq(pTHX_ SV *sv)
1116 register char *send;
1124 s = SvPV_force(sv, len);
1125 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
1128 while (s < send && *s != '\\')
1133 if ( PL_hints & HINT_NEW_STRING ) {
1134 pv = sv_2mortal(newSVpvn(SvPVX_const(pv), len));
1140 if (s + 1 < send && (s[1] == '\\'))
1141 s++; /* all that, just for this */
1146 SvCUR_set(sv, d - SvPVX_const(sv));
1148 if ( PL_hints & HINT_NEW_STRING )
1149 return new_constant(NULL, 0, "q", sv, pv, "q");
1154 * Now come three functions related to double-quote context,
1155 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
1156 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
1157 * interact with PL_lex_state, and create fake ( ... ) argument lists
1158 * to handle functions and concatenation.
1159 * They assume that whoever calls them will be setting up a fake
1160 * join call, because each subthing puts a ',' after it. This lets
1163 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
1165 * (I'm not sure whether the spurious commas at the end of lcfirst's
1166 * arguments and join's arguments are created or not).
1171 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
1173 * Pattern matching will set PL_lex_op to the pattern-matching op to
1174 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
1176 * OP_CONST and OP_READLINE are easy--just make the new op and return.
1178 * Everything else becomes a FUNC.
1180 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
1181 * had an OP_CONST or OP_READLINE). This just sets us up for a
1182 * call to S_sublex_push().
1186 S_sublex_start(pTHX)
1188 register const I32 op_type = yylval.ival;
1190 if (op_type == OP_NULL) {
1191 yylval.opval = PL_lex_op;
1195 if (op_type == OP_CONST || op_type == OP_READLINE) {
1196 SV *sv = tokeq(PL_lex_stuff);
1198 if (SvTYPE(sv) == SVt_PVIV) {
1199 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
1201 const char *p = SvPV_const(sv, len);
1202 SV * const nsv = newSVpvn(p, len);
1208 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1209 PL_lex_stuff = Nullsv;
1210 /* Allow <FH> // "foo" */
1211 if (op_type == OP_READLINE)
1212 PL_expect = XTERMORDORDOR;
1216 PL_sublex_info.super_state = PL_lex_state;
1217 PL_sublex_info.sub_inwhat = op_type;
1218 PL_sublex_info.sub_op = PL_lex_op;
1219 PL_lex_state = LEX_INTERPPUSH;
1223 yylval.opval = PL_lex_op;
1233 * Create a new scope to save the lexing state. The scope will be
1234 * ended in S_sublex_done. Returns a '(', starting the function arguments
1235 * to the uc, lc, etc. found before.
1236 * Sets PL_lex_state to LEX_INTERPCONCAT.
1245 PL_lex_state = PL_sublex_info.super_state;
1246 SAVEI32(PL_lex_dojoin);
1247 SAVEI32(PL_lex_brackets);
1248 SAVEI32(PL_lex_casemods);
1249 SAVEI32(PL_lex_starts);
1250 SAVEI32(PL_lex_state);
1251 SAVEVPTR(PL_lex_inpat);
1252 SAVEI32(PL_lex_inwhat);
1253 SAVECOPLINE(PL_curcop);
1254 SAVEPPTR(PL_bufptr);
1255 SAVEPPTR(PL_bufend);
1256 SAVEPPTR(PL_oldbufptr);
1257 SAVEPPTR(PL_oldoldbufptr);
1258 SAVEPPTR(PL_last_lop);
1259 SAVEPPTR(PL_last_uni);
1260 SAVEPPTR(PL_linestart);
1261 SAVESPTR(PL_linestr);
1262 SAVEGENERICPV(PL_lex_brackstack);
1263 SAVEGENERICPV(PL_lex_casestack);
1265 PL_linestr = PL_lex_stuff;
1266 PL_lex_stuff = Nullsv;
1268 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1269 = SvPVX(PL_linestr);
1270 PL_bufend += SvCUR(PL_linestr);
1271 PL_last_lop = PL_last_uni = Nullch;
1272 SAVEFREESV(PL_linestr);
1274 PL_lex_dojoin = FALSE;
1275 PL_lex_brackets = 0;
1276 Newx(PL_lex_brackstack, 120, char);
1277 Newx(PL_lex_casestack, 12, char);
1278 PL_lex_casemods = 0;
1279 *PL_lex_casestack = '\0';
1281 PL_lex_state = LEX_INTERPCONCAT;
1282 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1284 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1285 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1286 PL_lex_inpat = PL_sublex_info.sub_op;
1288 PL_lex_inpat = Nullop;
1295 * Restores lexer state after a S_sublex_push.
1302 if (!PL_lex_starts++) {
1303 SV * const sv = newSVpvn("",0);
1304 if (SvUTF8(PL_linestr))
1306 PL_expect = XOPERATOR;
1307 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1311 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1312 PL_lex_state = LEX_INTERPCASEMOD;
1316 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1317 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1318 PL_linestr = PL_lex_repl;
1320 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1321 PL_bufend += SvCUR(PL_linestr);
1322 PL_last_lop = PL_last_uni = Nullch;
1323 SAVEFREESV(PL_linestr);
1324 PL_lex_dojoin = FALSE;
1325 PL_lex_brackets = 0;
1326 PL_lex_casemods = 0;
1327 *PL_lex_casestack = '\0';
1329 if (SvEVALED(PL_lex_repl)) {
1330 PL_lex_state = LEX_INTERPNORMAL;
1332 /* we don't clear PL_lex_repl here, so that we can check later
1333 whether this is an evalled subst; that means we rely on the
1334 logic to ensure sublex_done() is called again only via the
1335 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1338 PL_lex_state = LEX_INTERPCONCAT;
1339 PL_lex_repl = Nullsv;
1345 PL_bufend = SvPVX(PL_linestr);
1346 PL_bufend += SvCUR(PL_linestr);
1347 PL_expect = XOPERATOR;
1348 PL_sublex_info.sub_inwhat = 0;
1356 Extracts a pattern, double-quoted string, or transliteration. This
1359 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1360 processing a pattern (PL_lex_inpat is true), a transliteration
1361 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1363 Returns a pointer to the character scanned up to. Iff this is
1364 advanced from the start pointer supplied (ie if anything was
1365 successfully parsed), will leave an OP for the substring scanned
1366 in yylval. Caller must intuit reason for not parsing further
1367 by looking at the next characters herself.
1371 double-quoted style: \r and \n
1372 regexp special ones: \D \s
1374 backrefs: \1 (deprecated in substitution replacements)
1375 case and quoting: \U \Q \E
1376 stops on @ and $, but not for $ as tail anchor
1378 In transliterations:
1379 characters are VERY literal, except for - not at the start or end
1380 of the string, which indicates a range. scan_const expands the
1381 range to the full set of intermediate characters.
1383 In double-quoted strings:
1385 double-quoted style: \r and \n
1387 backrefs: \1 (deprecated)
1388 case and quoting: \U \Q \E
1391 scan_const does *not* construct ops to handle interpolated strings.
1392 It stops processing as soon as it finds an embedded $ or @ variable
1393 and leaves it to the caller to work out what's going on.
1395 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @::foo.
1397 $ in pattern could be $foo or could be tail anchor. Assumption:
1398 it's a tail anchor if $ is the last thing in the string, or if it's
1399 followed by one of ")| \n\t"
1401 \1 (backreferences) are turned into $1
1403 The structure of the code is
1404 while (there's a character to process) {
1405 handle transliteration ranges
1406 skip regexp comments
1407 skip # initiated comments in //x patterns
1408 check for embedded @foo
1409 check for embedded scalars
1411 leave intact backslashes from leave (below)
1412 deprecate \1 in strings and sub replacements
1413 handle string-changing backslashes \l \U \Q \E, etc.
1414 switch (what was escaped) {
1415 handle - in a transliteration (becomes a literal -)
1416 handle \132 octal characters
1417 handle 0x15 hex characters
1418 handle \cV (control V)
1419 handle printf backslashes (\f, \r, \n, etc)
1421 } (end if backslash)
1422 } (end while character to read)
1427 S_scan_const(pTHX_ char *start)
1429 register char *send = PL_bufend; /* end of the constant */
1430 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1431 register char *s = start; /* start of the constant */
1432 register char *d = SvPVX(sv); /* destination for copies */
1433 bool dorange = FALSE; /* are we in a translit range? */
1434 bool didrange = FALSE; /* did we just finish a range? */
1435 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1436 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1439 UV literal_endpoint = 0;
1442 const char *leaveit = /* set of acceptably-backslashed characters */
1444 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxz0123456789[{]} \t\n\r\f\v#"
1447 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1448 /* If we are doing a trans and we know we want UTF8 set expectation */
1449 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1450 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1454 while (s < send || dorange) {
1455 /* get transliterations out of the way (they're most literal) */
1456 if (PL_lex_inwhat == OP_TRANS) {
1457 /* expand a range A-Z to the full set of characters. AIE! */
1459 I32 i; /* current expanded character */
1460 I32 min; /* first character in range */
1461 I32 max; /* last character in range */
1464 char * const c = (char*)utf8_hop((U8*)d, -1);
1468 *c = (char)UTF_TO_NATIVE(0xff);
1469 /* mark the range as done, and continue */
1475 i = d - SvPVX_const(sv); /* remember current offset */
1476 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1477 d = SvPVX(sv) + i; /* refresh d after realloc */
1478 d -= 2; /* eat the first char and the - */
1480 min = (U8)*d; /* first char in range */
1481 max = (U8)d[1]; /* last char in range */
1485 "Invalid range \"%c-%c\" in transliteration operator",
1486 (char)min, (char)max);
1490 if (literal_endpoint == 2 &&
1491 ((isLOWER(min) && isLOWER(max)) ||
1492 (isUPPER(min) && isUPPER(max)))) {
1494 for (i = min; i <= max; i++)
1496 *d++ = NATIVE_TO_NEED(has_utf8,i);
1498 for (i = min; i <= max; i++)
1500 *d++ = NATIVE_TO_NEED(has_utf8,i);
1505 for (i = min; i <= max; i++)
1508 /* mark the range as done, and continue */
1512 literal_endpoint = 0;
1517 /* range begins (ignore - as first or last char) */
1518 else if (*s == '-' && s+1 < send && s != start) {
1520 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1523 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
1533 literal_endpoint = 0;
1538 /* if we get here, we're not doing a transliteration */
1540 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1541 except for the last char, which will be done separately. */
1542 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1544 while (s+1 < send && *s != ')')
1545 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1547 else if (s[2] == '{' /* This should match regcomp.c */
1548 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1551 char *regparse = s + (s[2] == '{' ? 3 : 4);
1554 while (count && (c = *regparse)) {
1555 if (c == '\\' && regparse[1])
1563 if (*regparse != ')')
1564 regparse--; /* Leave one char for continuation. */
1565 while (s < regparse)
1566 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1570 /* likewise skip #-initiated comments in //x patterns */
1571 else if (*s == '#' && PL_lex_inpat &&
1572 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1573 while (s+1 < send && *s != '\n')
1574 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1577 /* check for embedded arrays
1578 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
1580 else if (*s == '@' && s[1]
1581 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1584 /* check for embedded scalars. only stop if we're sure it's a
1587 else if (*s == '$') {
1588 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1590 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
1591 break; /* in regexp, $ might be tail anchor */
1594 /* End of else if chain - OP_TRANS rejoin rest */
1597 if (*s == '\\' && s+1 < send) {
1600 /* some backslashes we leave behind */
1601 if (*leaveit && *s && strchr(leaveit, *s)) {
1602 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
1603 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1607 /* deprecate \1 in strings and substitution replacements */
1608 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1609 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1611 if (ckWARN(WARN_SYNTAX))
1612 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
1617 /* string-change backslash escapes */
1618 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1623 /* if we get here, it's either a quoted -, or a digit */
1626 /* quoted - in transliterations */
1628 if (PL_lex_inwhat == OP_TRANS) {
1638 Perl_warner(aTHX_ packWARN(WARN_MISC),
1639 "Unrecognized escape \\%c passed through",
1641 /* default action is to copy the quoted character */
1642 goto default_action;
1645 /* \132 indicates an octal constant */
1646 case '0': case '1': case '2': case '3':
1647 case '4': case '5': case '6': case '7':
1651 uv = grok_oct(s, &len, &flags, NULL);
1654 goto NUM_ESCAPE_INSERT;
1656 /* \x24 indicates a hex constant */
1660 char* const e = strchr(s, '}');
1661 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1662 PERL_SCAN_DISALLOW_PREFIX;
1667 yyerror("Missing right brace on \\x{}");
1671 uv = grok_hex(s, &len, &flags, NULL);
1677 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
1678 uv = grok_hex(s, &len, &flags, NULL);
1684 /* Insert oct or hex escaped character.
1685 * There will always enough room in sv since such
1686 * escapes will be longer than any UTF-8 sequence
1687 * they can end up as. */
1689 /* We need to map to chars to ASCII before doing the tests
1692 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
1693 if (!has_utf8 && uv > 255) {
1694 /* Might need to recode whatever we have
1695 * accumulated so far if it contains any
1698 * (Can't we keep track of that and avoid
1699 * this rescan? --jhi)
1703 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
1704 if (!NATIVE_IS_INVARIANT(*c)) {
1709 const STRLEN offset = d - SvPVX_const(sv);
1711 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
1715 while (src >= (const U8 *)SvPVX_const(sv)) {
1716 if (!NATIVE_IS_INVARIANT(*src)) {
1717 const U8 ch = NATIVE_TO_ASCII(*src);
1718 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
1719 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
1729 if (has_utf8 || uv > 255) {
1730 d = (char*)uvchr_to_utf8((U8*)d, uv);
1732 if (PL_lex_inwhat == OP_TRANS &&
1733 PL_sublex_info.sub_op) {
1734 PL_sublex_info.sub_op->op_private |=
1735 (PL_lex_repl ? OPpTRANS_FROM_UTF
1748 /* \N{LATIN SMALL LETTER A} is a named character */
1752 char* e = strchr(s, '}');
1758 yyerror("Missing right brace on \\N{}");
1762 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
1764 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1765 PERL_SCAN_DISALLOW_PREFIX;
1768 uv = grok_hex(s, &len, &flags, NULL);
1770 goto NUM_ESCAPE_INSERT;
1772 res = newSVpvn(s + 1, e - s - 1);
1773 res = new_constant( Nullch, 0, "charnames",
1774 res, Nullsv, "\\N{...}" );
1776 sv_utf8_upgrade(res);
1777 str = SvPV_const(res,len);
1778 #ifdef EBCDIC_NEVER_MIND
1779 /* charnames uses pack U and that has been
1780 * recently changed to do the below uni->native
1781 * mapping, so this would be redundant (and wrong,
1782 * the code point would be doubly converted).
1783 * But leave this in just in case the pack U change
1784 * gets revoked, but the semantics is still
1785 * desireable for charnames. --jhi */
1787 UV uv = utf8_to_uvchr((const U8*)str, 0);
1790 U8 tmpbuf[UTF8_MAXBYTES+1], *d;
1792 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
1793 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
1794 str = SvPV_const(res, len);
1798 if (!has_utf8 && SvUTF8(res)) {
1799 const char * const ostart = SvPVX_const(sv);
1800 SvCUR_set(sv, d - ostart);
1803 sv_utf8_upgrade(sv);
1804 /* this just broke our allocation above... */
1805 SvGROW(sv, (STRLEN)(send - start));
1806 d = SvPVX(sv) + SvCUR(sv);
1809 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
1810 const char * const odest = SvPVX_const(sv);
1812 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
1813 d = SvPVX(sv) + (d - odest);
1815 Copy(str, d, len, char);
1822 yyerror("Missing braces on \\N{}");
1825 /* \c is a control character */
1834 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
1837 yyerror("Missing control char name in \\c");
1841 /* printf-style backslashes, formfeeds, newlines, etc */
1843 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
1846 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
1849 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
1852 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
1855 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
1858 *d++ = ASCII_TO_NEED(has_utf8,'\033');
1861 *d++ = ASCII_TO_NEED(has_utf8,'\007');
1867 } /* end if (backslash) */
1874 /* If we started with encoded form, or already know we want it
1875 and then encode the next character */
1876 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1878 const UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1879 const STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1882 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1883 const STRLEN off = d - SvPVX_const(sv);
1884 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1886 d = (char*)uvchr_to_utf8((U8*)d, uv);
1890 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1892 } /* while loop to process each character */
1894 /* terminate the string and set up the sv */
1896 SvCUR_set(sv, d - SvPVX_const(sv));
1897 if (SvCUR(sv) >= SvLEN(sv))
1898 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
1901 if (PL_encoding && !has_utf8) {
1902 sv_recode_to_utf8(sv, PL_encoding);
1908 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1909 PL_sublex_info.sub_op->op_private |=
1910 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1914 /* shrink the sv if we allocated more than we used */
1915 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1916 SvPV_shrink_to_cur(sv);
1919 /* return the substring (via yylval) only if we parsed anything */
1920 if (s > PL_bufptr) {
1921 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1922 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1924 ( PL_lex_inwhat == OP_TRANS
1926 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1929 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1936 * Returns TRUE if there's more to the expression (e.g., a subscript),
1939 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1941 * ->[ and ->{ return TRUE
1942 * { and [ outside a pattern are always subscripts, so return TRUE
1943 * if we're outside a pattern and it's not { or [, then return FALSE
1944 * if we're in a pattern and the first char is a {
1945 * {4,5} (any digits around the comma) returns FALSE
1946 * if we're in a pattern and the first char is a [
1948 * [SOMETHING] has a funky algorithm to decide whether it's a
1949 * character class or not. It has to deal with things like
1950 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1951 * anything else returns TRUE
1954 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1957 S_intuit_more(pTHX_ register char *s)
1959 if (PL_lex_brackets)
1961 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1963 if (*s != '{' && *s != '[')
1968 /* In a pattern, so maybe we have {n,m}. */
1985 /* On the other hand, maybe we have a character class */
1988 if (*s == ']' || *s == '^')
1991 /* this is terrifying, and it works */
1992 int weight = 2; /* let's weigh the evidence */
1994 unsigned char un_char = 255, last_un_char;
1995 const char * const send = strchr(s,']');
1996 char tmpbuf[sizeof PL_tokenbuf * 4];
1998 if (!send) /* has to be an expression */
2001 Zero(seen,256,char);
2004 else if (isDIGIT(*s)) {
2006 if (isDIGIT(s[1]) && s[2] == ']')
2012 for (; s < send; s++) {
2013 last_un_char = un_char;
2014 un_char = (unsigned char)*s;
2019 weight -= seen[un_char] * 10;
2020 if (isALNUM_lazy_if(s+1,UTF)) {
2021 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
2022 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
2027 else if (*s == '$' && s[1] &&
2028 strchr("[#!%*<>()-=",s[1])) {
2029 if (/*{*/ strchr("])} =",s[2]))
2038 if (strchr("wds]",s[1]))
2040 else if (seen['\''] || seen['"'])
2042 else if (strchr("rnftbxcav",s[1]))
2044 else if (isDIGIT(s[1])) {
2046 while (s[1] && isDIGIT(s[1]))
2056 if (strchr("aA01! ",last_un_char))
2058 if (strchr("zZ79~",s[1]))
2060 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
2061 weight -= 5; /* cope with negative subscript */
2064 if (!isALNUM(last_un_char)
2065 && !(last_un_char == '$' || last_un_char == '@'
2066 || last_un_char == '&')
2067 && isALPHA(*s) && s[1] && isALPHA(s[1])) {
2072 if (keyword(tmpbuf, d - tmpbuf))
2075 if (un_char == last_un_char + 1)
2077 weight -= seen[un_char];
2082 if (weight >= 0) /* probably a character class */
2092 * Does all the checking to disambiguate
2094 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
2095 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
2097 * First argument is the stuff after the first token, e.g. "bar".
2099 * Not a method if bar is a filehandle.
2100 * Not a method if foo is a subroutine prototyped to take a filehandle.
2101 * Not a method if it's really "Foo $bar"
2102 * Method if it's "foo $bar"
2103 * Not a method if it's really "print foo $bar"
2104 * Method if it's really "foo package::" (interpreted as package->foo)
2105 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
2106 * Not a method if bar is a filehandle or package, but is quoted with
2111 S_intuit_method(pTHX_ char *start, GV *gv)
2113 char *s = start + (*start == '$');
2114 char tmpbuf[sizeof PL_tokenbuf];
2122 if ((cv = GvCVu(gv))) {
2123 const char *proto = SvPVX_const(cv);
2133 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
2134 /* start is the beginning of the possible filehandle/object,
2135 * and s is the end of it
2136 * tmpbuf is a copy of it
2139 if (*start == '$') {
2140 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
2145 return *s == '(' ? FUNCMETH : METHOD;
2147 if (!keyword(tmpbuf, len)) {
2148 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
2153 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
2154 if (indirgv && GvCVu(indirgv))
2156 /* filehandle or package name makes it a method */
2157 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
2159 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
2160 return 0; /* no assumptions -- "=>" quotes bearword */
2162 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
2163 newSVpvn(tmpbuf,len));
2164 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
2168 return *s == '(' ? FUNCMETH : METHOD;
2176 * Return a string of Perl code to load the debugger. If PERL5DB
2177 * is set, it will return the contents of that, otherwise a
2178 * compile-time require of perl5db.pl.
2185 const char * const pdb = PerlEnv_getenv("PERL5DB");
2189 SETERRNO(0,SS_NORMAL);
2190 return "BEGIN { require 'perl5db.pl' }";
2196 /* Encoded script support. filter_add() effectively inserts a
2197 * 'pre-processing' function into the current source input stream.
2198 * Note that the filter function only applies to the current source file
2199 * (e.g., it will not affect files 'require'd or 'use'd by this one).
2201 * The datasv parameter (which may be NULL) can be used to pass
2202 * private data to this instance of the filter. The filter function
2203 * can recover the SV using the FILTER_DATA macro and use it to
2204 * store private buffers and state information.
2206 * The supplied datasv parameter is upgraded to a PVIO type
2207 * and the IoDIRP/IoANY field is used to store the function pointer,
2208 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
2209 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
2210 * private use must be set using malloc'd pointers.
2214 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
2219 if (!PL_rsfp_filters)
2220 PL_rsfp_filters = newAV();
2222 datasv = NEWSV(255,0);
2223 SvUPGRADE(datasv, SVt_PVIO);
2224 IoANY(datasv) = FPTR2DPTR(void *, funcp); /* stash funcp into spare field */
2225 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2226 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2227 IoANY(datasv), SvPV_nolen(datasv)));
2228 av_unshift(PL_rsfp_filters, 1);
2229 av_store(PL_rsfp_filters, 0, datasv) ;
2234 /* Delete most recently added instance of this filter function. */
2236 Perl_filter_del(pTHX_ filter_t funcp)
2241 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", FPTR2DPTR(XPVIO *, funcp)));
2243 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2245 /* if filter is on top of stack (usual case) just pop it off */
2246 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2247 if (IoANY(datasv) == FPTR2DPTR(void *, funcp)) {
2248 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2249 IoANY(datasv) = (void *)NULL;
2250 sv_free(av_pop(PL_rsfp_filters));
2254 /* we need to search for the correct entry and clear it */
2255 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2259 /* Invoke the idxth filter function for the current rsfp. */
2260 /* maxlen 0 = read one text line */
2262 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2267 if (!PL_rsfp_filters)
2269 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
2270 /* Provide a default input filter to make life easy. */
2271 /* Note that we append to the line. This is handy. */
2272 DEBUG_P(PerlIO_printf(Perl_debug_log,
2273 "filter_read %d: from rsfp\n", idx));
2277 const int old_len = SvCUR(buf_sv);
2279 /* ensure buf_sv is large enough */
2280 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
2281 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2282 if (PerlIO_error(PL_rsfp))
2283 return -1; /* error */
2285 return 0 ; /* end of file */
2287 SvCUR_set(buf_sv, old_len + len) ;
2290 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2291 if (PerlIO_error(PL_rsfp))
2292 return -1; /* error */
2294 return 0 ; /* end of file */
2297 return SvCUR(buf_sv);
2299 /* Skip this filter slot if filter has been deleted */
2300 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
2301 DEBUG_P(PerlIO_printf(Perl_debug_log,
2302 "filter_read %d: skipped (filter deleted)\n",
2304 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2306 /* Get function pointer hidden within datasv */
2307 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
2308 DEBUG_P(PerlIO_printf(Perl_debug_log,
2309 "filter_read %d: via function %p (%s)\n",
2310 idx, datasv, SvPV_nolen_const(datasv)));
2311 /* Call function. The function is expected to */
2312 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2313 /* Return: <0:error, =0:eof, >0:not eof */
2314 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2318 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2320 #ifdef PERL_CR_FILTER
2321 if (!PL_rsfp_filters) {
2322 filter_add(S_cr_textfilter,NULL);
2325 if (PL_rsfp_filters) {
2327 SvCUR_set(sv, 0); /* start with empty line */
2328 if (FILTER_READ(0, sv, 0) > 0)
2329 return ( SvPVX(sv) ) ;
2334 return (sv_gets(sv, fp, append));
2338 S_find_in_my_stash(pTHX_ const char *pkgname, I32 len)
2342 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2346 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2347 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2349 return GvHV(gv); /* Foo:: */
2352 /* use constant CLASS => 'MyClass' */
2353 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2355 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2356 pkgname = SvPV_nolen_const(sv);
2360 return gv_stashpv(pkgname, FALSE);
2364 S_tokenize_use(pTHX_ int is_use, char *s) {
2365 if (PL_expect != XSTATE)
2366 yyerror(Perl_form(aTHX_ "\"%s\" not allowed in expression",
2367 is_use ? "use" : "no"));
2369 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
2370 s = force_version(s, TRUE);
2371 if (*s == ';' || (s = skipspace(s), *s == ';')) {
2372 PL_nextval[PL_nexttoke].opval = Nullop;
2375 else if (*s == 'v') {
2376 s = force_word(s,WORD,FALSE,TRUE,FALSE);
2377 s = force_version(s, FALSE);
2381 s = force_word(s,WORD,FALSE,TRUE,FALSE);
2382 s = force_version(s, FALSE);
2384 yylval.ival = is_use;
2388 static const char* const exp_name[] =
2389 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2390 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
2397 Works out what to call the token just pulled out of the input
2398 stream. The yacc parser takes care of taking the ops we return and
2399 stitching them into a tree.
2405 if read an identifier
2406 if we're in a my declaration
2407 croak if they tried to say my($foo::bar)
2408 build the ops for a my() declaration
2409 if it's an access to a my() variable
2410 are we in a sort block?
2411 croak if my($a); $a <=> $b
2412 build ops for access to a my() variable
2413 if in a dq string, and they've said @foo and we can't find @foo
2415 build ops for a bareword
2416 if we already built the token before, use it.
2421 #pragma segment Perl_yylex
2426 register char *s = PL_bufptr;
2433 I32 orig_keyword = 0;
2436 SV* tmp = newSVpvn("", 0);
2437 PerlIO_printf(Perl_debug_log, "### %"IVdf":LEX_%s/X%s %s\n",
2438 (IV)CopLINE(PL_curcop),
2439 lex_state_names[PL_lex_state],
2440 exp_name[PL_expect],
2441 pv_display(tmp, s, strlen(s), 0, 60));
2444 /* check if there's an identifier for us to look at */
2445 if (PL_pending_ident)
2446 return REPORT(S_pending_ident(aTHX));
2448 /* no identifier pending identification */
2450 switch (PL_lex_state) {
2452 case LEX_NORMAL: /* Some compilers will produce faster */
2453 case LEX_INTERPNORMAL: /* code if we comment these out. */
2457 /* when we've already built the next token, just pull it out of the queue */
2460 yylval = PL_nextval[PL_nexttoke];
2462 PL_lex_state = PL_lex_defer;
2463 PL_expect = PL_lex_expect;
2464 PL_lex_defer = LEX_NORMAL;
2466 return REPORT(PL_nexttype[PL_nexttoke]);
2468 /* interpolated case modifiers like \L \U, including \Q and \E.
2469 when we get here, PL_bufptr is at the \
2471 case LEX_INTERPCASEMOD:
2473 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2474 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2476 /* handle \E or end of string */
2477 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2479 if (PL_lex_casemods) {
2480 const char oldmod = PL_lex_casestack[--PL_lex_casemods];
2481 PL_lex_casestack[PL_lex_casemods] = '\0';
2483 if (PL_bufptr != PL_bufend
2484 && (oldmod == 'L' || oldmod == 'U' || oldmod == 'Q')) {
2486 PL_lex_state = LEX_INTERPCONCAT;
2490 if (PL_bufptr != PL_bufend)
2492 PL_lex_state = LEX_INTERPCONCAT;
2496 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2497 "### Saw case modifier\n"); });
2499 if (s[1] == '\\' && s[2] == 'E') {
2501 PL_lex_state = LEX_INTERPCONCAT;
2505 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2506 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
2507 if ((*s == 'L' || *s == 'U') &&
2508 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
2509 PL_lex_casestack[--PL_lex_casemods] = '\0';
2512 if (PL_lex_casemods > 10)
2513 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2514 PL_lex_casestack[PL_lex_casemods++] = *s;
2515 PL_lex_casestack[PL_lex_casemods] = '\0';
2516 PL_lex_state = LEX_INTERPCONCAT;
2517 PL_nextval[PL_nexttoke].ival = 0;
2520 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2522 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2524 PL_nextval[PL_nexttoke].ival = OP_LC;
2526 PL_nextval[PL_nexttoke].ival = OP_UC;
2528 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2530 Perl_croak(aTHX_ "panic: yylex");
2534 if (PL_lex_starts) {
2537 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2538 if (PL_lex_casemods == 1 && PL_lex_inpat)
2547 case LEX_INTERPPUSH:
2548 return REPORT(sublex_push());
2550 case LEX_INTERPSTART:
2551 if (PL_bufptr == PL_bufend)
2552 return REPORT(sublex_done());
2553 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2554 "### Interpolated variable\n"); });
2556 PL_lex_dojoin = (*PL_bufptr == '@');
2557 PL_lex_state = LEX_INTERPNORMAL;
2558 if (PL_lex_dojoin) {
2559 PL_nextval[PL_nexttoke].ival = 0;
2561 force_ident("\"", '$');
2562 PL_nextval[PL_nexttoke].ival = 0;
2564 PL_nextval[PL_nexttoke].ival = 0;
2566 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2569 if (PL_lex_starts++) {
2571 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2572 if (!PL_lex_casemods && PL_lex_inpat)
2579 case LEX_INTERPENDMAYBE:
2580 if (intuit_more(PL_bufptr)) {
2581 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2587 if (PL_lex_dojoin) {
2588 PL_lex_dojoin = FALSE;
2589 PL_lex_state = LEX_INTERPCONCAT;
2592 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2593 && SvEVALED(PL_lex_repl))
2595 if (PL_bufptr != PL_bufend)
2596 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2597 PL_lex_repl = Nullsv;
2600 case LEX_INTERPCONCAT:
2602 if (PL_lex_brackets)
2603 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2605 if (PL_bufptr == PL_bufend)
2606 return REPORT(sublex_done());
2608 if (SvIVX(PL_linestr) == '\'') {
2609 SV *sv = newSVsv(PL_linestr);
2612 else if ( PL_hints & HINT_NEW_RE )
2613 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2614 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2618 s = scan_const(PL_bufptr);
2620 PL_lex_state = LEX_INTERPCASEMOD;
2622 PL_lex_state = LEX_INTERPSTART;
2625 if (s != PL_bufptr) {
2626 PL_nextval[PL_nexttoke] = yylval;
2629 if (PL_lex_starts++) {
2630 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2631 if (!PL_lex_casemods && PL_lex_inpat)
2644 PL_lex_state = LEX_NORMAL;
2645 s = scan_formline(PL_bufptr);
2646 if (!PL_lex_formbrack)
2652 PL_oldoldbufptr = PL_oldbufptr;
2658 if (isIDFIRST_lazy_if(s,UTF))
2660 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2663 goto fake_eof; /* emulate EOF on ^D or ^Z */
2668 if (PL_lex_brackets) {
2669 yyerror(PL_lex_formbrack
2670 ? "Format not terminated"
2671 : "Missing right curly or square bracket");
2673 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2674 "### Tokener got EOF\n");
2678 if (s++ < PL_bufend)
2679 goto retry; /* ignore stray nulls */
2682 if (!PL_in_eval && !PL_preambled) {
2683 PL_preambled = TRUE;
2684 sv_setpv(PL_linestr,incl_perldb());
2685 if (SvCUR(PL_linestr))
2686 sv_catpvn(PL_linestr,";", 1);
2688 while(AvFILLp(PL_preambleav) >= 0) {
2689 SV *tmpsv = av_shift(PL_preambleav);
2690 sv_catsv(PL_linestr, tmpsv);
2691 sv_catpvn(PL_linestr, ";", 1);
2694 sv_free((SV*)PL_preambleav);
2695 PL_preambleav = NULL;
2697 if (PL_minus_n || PL_minus_p) {
2698 sv_catpv(PL_linestr, "LINE: while (<>) {");
2700 sv_catpv(PL_linestr,"chomp;");
2703 if ((*PL_splitstr == '/' || *PL_splitstr == '\''
2704 || *PL_splitstr == '"')
2705 && strchr(PL_splitstr + 1, *PL_splitstr))
2706 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
2708 /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
2709 bytes can be used as quoting characters. :-) */
2710 /* The count here deliberately includes the NUL
2711 that terminates the C string constant. This
2712 embeds the opening NUL into the string. */
2713 const char *splits = PL_splitstr;
2714 sv_catpvn(PL_linestr, "our @F=split(q", 15);
2717 if (*splits == '\\')
2718 sv_catpvn(PL_linestr, splits, 1);
2719 sv_catpvn(PL_linestr, splits, 1);
2720 } while (*splits++);
2721 /* This loop will embed the trailing NUL of
2722 PL_linestr as the last thing it does before
2724 sv_catpvn(PL_linestr, ");", 2);
2728 sv_catpv(PL_linestr,"our @F=split(' ');");
2731 sv_catpvn(PL_linestr, "\n", 1);
2732 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2733 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2734 PL_last_lop = PL_last_uni = Nullch;
2735 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2736 SV * const sv = NEWSV(85,0);
2738 sv_upgrade(sv, SVt_PVMG);
2739 sv_setsv(sv,PL_linestr);
2742 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2747 bof = PL_rsfp ? TRUE : FALSE;
2748 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2751 if (PL_preprocess && !PL_in_eval)
2752 (void)PerlProc_pclose(PL_rsfp);
2753 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2754 PerlIO_clearerr(PL_rsfp);
2756 (void)PerlIO_close(PL_rsfp);
2758 PL_doextract = FALSE;
2760 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2761 sv_setpv(PL_linestr,PL_minus_p
2762 ? ";}continue{print;}" : ";}");
2763 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2764 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2765 PL_last_lop = PL_last_uni = Nullch;
2766 PL_minus_n = PL_minus_p = 0;
2769 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2770 PL_last_lop = PL_last_uni = Nullch;
2771 sv_setpvn(PL_linestr,"",0);
2772 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2774 /* If it looks like the start of a BOM or raw UTF-16,
2775 * check if it in fact is. */
2781 #ifdef PERLIO_IS_STDIO
2782 # ifdef __GNU_LIBRARY__
2783 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2784 # define FTELL_FOR_PIPE_IS_BROKEN
2788 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2789 # define FTELL_FOR_PIPE_IS_BROKEN
2794 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2795 /* This loses the possibility to detect the bof
2796 * situation on perl -P when the libc5 is being used.
2797 * Workaround? Maybe attach some extra state to PL_rsfp?
2800 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2802 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
2805 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2806 s = swallow_bom((U8*)s);
2810 /* Incest with pod. */
2811 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2812 sv_setpvn(PL_linestr, "", 0);
2813 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2814 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2815 PL_last_lop = PL_last_uni = Nullch;
2816 PL_doextract = FALSE;
2820 } while (PL_doextract);
2821 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2822 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2823 SV * const sv = NEWSV(85,0);
2825 sv_upgrade(sv, SVt_PVMG);
2826 sv_setsv(sv,PL_linestr);
2829 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2831 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2832 PL_last_lop = PL_last_uni = Nullch;
2833 if (CopLINE(PL_curcop) == 1) {
2834 while (s < PL_bufend && isSPACE(*s))
2836 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2840 if (*s == '#' && *(s+1) == '!')
2842 #ifdef ALTERNATE_SHEBANG
2844 static char const as[] = ALTERNATE_SHEBANG;
2845 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2846 d = s + (sizeof(as) - 1);
2848 #endif /* ALTERNATE_SHEBANG */
2857 while (*d && !isSPACE(*d))
2861 #ifdef ARG_ZERO_IS_SCRIPT
2862 if (ipathend > ipath) {
2864 * HP-UX (at least) sets argv[0] to the script name,
2865 * which makes $^X incorrect. And Digital UNIX and Linux,
2866 * at least, set argv[0] to the basename of the Perl
2867 * interpreter. So, having found "#!", we'll set it right.
2869 SV * const x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); /* $^X */
2870 assert(SvPOK(x) || SvGMAGICAL(x));
2871 if (sv_eq(x, CopFILESV(PL_curcop))) {
2872 sv_setpvn(x, ipath, ipathend - ipath);
2878 const char *bstart = SvPV_const(CopFILESV(PL_curcop),blen);
2879 const char * const lstart = SvPV_const(x,llen);
2881 bstart += blen - llen;
2882 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
2883 sv_setpvn(x, ipath, ipathend - ipath);
2888 TAINT_NOT; /* $^X is always tainted, but that's OK */
2890 #endif /* ARG_ZERO_IS_SCRIPT */
2895 d = instr(s,"perl -");
2897 d = instr(s,"perl");
2899 /* avoid getting into infinite loops when shebang
2900 * line contains "Perl" rather than "perl" */
2902 for (d = ipathend-4; d >= ipath; --d) {
2903 if ((*d == 'p' || *d == 'P')
2904 && !ibcmp(d, "perl", 4))
2914 #ifdef ALTERNATE_SHEBANG
2916 * If the ALTERNATE_SHEBANG on this system starts with a
2917 * character that can be part of a Perl expression, then if
2918 * we see it but not "perl", we're probably looking at the
2919 * start of Perl code, not a request to hand off to some
2920 * other interpreter. Similarly, if "perl" is there, but
2921 * not in the first 'word' of the line, we assume the line
2922 * contains the start of the Perl program.
2924 if (d && *s != '#') {
2925 const char *c = ipath;
2926 while (*c && !strchr("; \t\r\n\f\v#", *c))
2929 d = Nullch; /* "perl" not in first word; ignore */
2931 *s = '#'; /* Don't try to parse shebang line */
2933 #endif /* ALTERNATE_SHEBANG */
2934 #ifndef MACOS_TRADITIONAL
2939 !instr(s,"indir") &&
2940 instr(PL_origargv[0],"perl"))
2947 while (s < PL_bufend && isSPACE(*s))
2949 if (s < PL_bufend) {
2950 Newxz(newargv,PL_origargc+3,char*);
2952 while (s < PL_bufend && !isSPACE(*s))
2955 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2958 newargv = PL_origargv;
2961 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2963 Perl_croak(aTHX_ "Can't exec %s", ipath);
2967 const U32 oldpdb = PL_perldb;
2968 const bool oldn = PL_minus_n;
2969 const bool oldp = PL_minus_p;
2971 while (*d && !isSPACE(*d)) d++;
2972 while (SPACE_OR_TAB(*d)) d++;
2975 const bool switches_done = PL_doswitches;
2977 if (*d == 'M' || *d == 'm' || *d == 'C') {
2978 const char * const m = d;
2979 while (*d && !isSPACE(*d)) d++;
2980 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2983 d = moreswitches(d);
2985 if (PL_doswitches && !switches_done) {
2986 int argc = PL_origargc;
2987 char **argv = PL_origargv;
2990 } while (argc && argv[0][0] == '-' && argv[0][1]);
2991 init_argv_symbols(argc,argv);
2993 if ((PERLDB_LINE && !oldpdb) ||
2994 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2995 /* if we have already added "LINE: while (<>) {",
2996 we must not do it again */
2998 sv_setpvn(PL_linestr, "", 0);
2999 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3000 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3001 PL_last_lop = PL_last_uni = Nullch;
3002 PL_preambled = FALSE;
3004 (void)gv_fetchfile(PL_origfilename);
3007 if (PL_doswitches && !switches_done) {
3008 int argc = PL_origargc;
3009 char **argv = PL_origargv;
3012 } while (argc && argv[0][0] == '-' && argv[0][1]);
3013 init_argv_symbols(argc,argv);
3019 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
3021 PL_lex_state = LEX_FORMLINE;
3026 #ifdef PERL_STRICT_CR
3027 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
3029 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
3031 case ' ': case '\t': case '\f': case 013:
3032 #ifdef MACOS_TRADITIONAL
3039 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
3040 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
3041 /* handle eval qq[#line 1 "foo"\n ...] */
3042 CopLINE_dec(PL_curcop);
3046 while (s < d && *s != '\n')
3050 else if (s > d) /* Found by Ilya: feed random input to Perl. */
3051 Perl_croak(aTHX_ "panic: input overflow");
3053 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
3055 PL_lex_state = LEX_FORMLINE;
3065 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
3072 while (s < PL_bufend && SPACE_OR_TAB(*s))
3075 if (strnEQ(s,"=>",2)) {
3076 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
3077 DEBUG_T( { S_printbuf(aTHX_
3078 "### Saw unary minus before =>, forcing word %s\n", s);
3080 OPERATOR('-'); /* unary minus */
3082 PL_last_uni = PL_oldbufptr;
3084 case 'r': ftst = OP_FTEREAD; break;
3085 case 'w': ftst = OP_FTEWRITE; break;
3086 case 'x': ftst = OP_FTEEXEC; break;
3087 case 'o': ftst = OP_FTEOWNED; break;
3088 case 'R': ftst = OP_FTRREAD; break;
3089 case 'W': ftst = OP_FTRWRITE; break;
3090 case 'X': ftst = OP_FTREXEC; break;
3091 case 'O': ftst = OP_FTROWNED; break;
3092 case 'e': ftst = OP_FTIS; break;
3093 case 'z': ftst = OP_FTZERO; break;
3094 case 's': ftst = OP_FTSIZE; break;
3095 case 'f': ftst = OP_FTFILE; break;
3096 case 'd': ftst = OP_FTDIR; break;
3097 case 'l': ftst = OP_FTLINK; break;
3098 case 'p': ftst = OP_FTPIPE; break;
3099 case 'S': ftst = OP_FTSOCK; break;
3100 case 'u': ftst = OP_FTSUID; break;
3101 case 'g': ftst = OP_FTSGID; break;
3102 case 'k': ftst = OP_FTSVTX; break;
3103 case 'b': ftst = OP_FTBLK; break;
3104 case 'c': ftst = OP_FTCHR; break;
3105 case 't': ftst = OP_FTTTY; break;
3106 case 'T': ftst = OP_FTTEXT; break;
3107 case 'B': ftst = OP_FTBINARY; break;
3108 case 'M': case 'A': case 'C':
3109 gv_fetchpv("\024",TRUE, SVt_PV);
3111 case 'M': ftst = OP_FTMTIME; break;
3112 case 'A': ftst = OP_FTATIME; break;
3113 case 'C': ftst = OP_FTCTIME; break;
3121 PL_last_lop_op = (OPCODE)ftst;
3122 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3123 "### Saw file test %c\n", (int)tmp);
3128 /* Assume it was a minus followed by a one-letter named
3129 * subroutine call (or a -bareword), then. */
3130 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3131 "### '-%c' looked like a file test but was not\n",
3140 if (PL_expect == XOPERATOR)
3145 else if (*s == '>') {
3148 if (isIDFIRST_lazy_if(s,UTF)) {
3149 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
3157 if (PL_expect == XOPERATOR)
3160 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
3162 OPERATOR('-'); /* unary minus */
3169 if (PL_expect == XOPERATOR)
3174 if (PL_expect == XOPERATOR)
3177 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
3183 if (PL_expect != XOPERATOR) {
3184 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3185 PL_expect = XOPERATOR;
3186 force_ident(PL_tokenbuf, '*');
3199 if (PL_expect == XOPERATOR) {
3203 PL_tokenbuf[0] = '%';
3204 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
3205 if (!PL_tokenbuf[1]) {
3208 PL_pending_ident = '%';
3219 && (PL_expect == XOPERATOR || PL_expect == XTERMORDORDOR)
3220 && FEATURE_IS_ENABLED("~~", 2))
3234 switch (PL_expect) {
3237 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
3239 PL_bufptr = s; /* update in case we back off */
3245 PL_expect = XTERMBLOCK;
3249 while (isIDFIRST_lazy_if(s,UTF)) {
3250 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3251 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
3252 if (tmp < 0) tmp = -tmp;
3268 d = scan_str(d,TRUE,TRUE);
3270 /* MUST advance bufptr here to avoid bogus
3271 "at end of line" context messages from yyerror().
3273 PL_bufptr = s + len;
3274 yyerror("Unterminated attribute parameter in attribute list");
3277 return REPORT(0); /* EOF indicator */
3281 SV *sv = newSVpvn(s, len);
3282 sv_catsv(sv, PL_lex_stuff);
3283 attrs = append_elem(OP_LIST, attrs,
3284 newSVOP(OP_CONST, 0, sv));
3285 SvREFCNT_dec(PL_lex_stuff);
3286 PL_lex_stuff = Nullsv;
3289 if (len == 6 && strnEQ(s, "unique", len)) {
3290 if (PL_in_my == KEY_our)
3292 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
3294 ; /* skip to avoid loading attributes.pm */
3297 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
3300 /* NOTE: any CV attrs applied here need to be part of
3301 the CVf_BUILTIN_ATTRS define in cv.h! */
3302 else if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3303 CvLVALUE_on(PL_compcv);
3304 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3305 CvLOCKED_on(PL_compcv);
3306 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3307 CvMETHOD_on(PL_compcv);
3308 else if (!PL_in_my && len == 9 && strnEQ(s, "assertion", len))
3309 CvASSERTION_on(PL_compcv);
3310 /* After we've set the flags, it could be argued that
3311 we don't need to do the attributes.pm-based setting
3312 process, and shouldn't bother appending recognized
3313 flags. To experiment with that, uncomment the
3314 following "else". (Note that's already been
3315 uncommented. That keeps the above-applied built-in
3316 attributes from being intercepted (and possibly
3317 rejected) by a package's attribute routines, but is
3318 justified by the performance win for the common case
3319 of applying only built-in attributes.) */
3321 attrs = append_elem(OP_LIST, attrs,
3322 newSVOP(OP_CONST, 0,
3326 if (*s == ':' && s[1] != ':')
3329 break; /* require real whitespace or :'s */
3331 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3332 if (*s != ';' && *s != '}' && *s != tmp && (tmp != '=' || *s != ')')) {
3333 const char q = ((*s == '\'') ? '"' : '\'');
3334 /* If here for an expression, and parsed no attrs, back off. */
3335 if (tmp == '=' && !attrs) {
3339 /* MUST advance bufptr here to avoid bogus "at end of line"
3340 context messages from yyerror().
3344 ? Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list", q, *s, q)
3345 : "Unterminated attribute list" );
3352 PL_nextval[PL_nexttoke].opval = attrs;
3360 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3361 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3378 if (PL_lex_brackets <= 0)
3379 yyerror("Unmatched right square bracket");
3382 if (PL_lex_state == LEX_INTERPNORMAL) {
3383 if (PL_lex_brackets == 0) {
3384 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3385 PL_lex_state = LEX_INTERPEND;
3392 if (PL_lex_brackets > 100) {
3393 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
3395 switch (PL_expect) {
3397 if (PL_lex_formbrack) {
3401 if (PL_oldoldbufptr == PL_last_lop)
3402 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3404 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3405 OPERATOR(HASHBRACK);
3407 while (s < PL_bufend && SPACE_OR_TAB(*s))
3410 PL_tokenbuf[0] = '\0';
3411 if (d < PL_bufend && *d == '-') {
3412 PL_tokenbuf[0] = '-';
3414 while (d < PL_bufend && SPACE_OR_TAB(*d))
3417 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3418 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3420 while (d < PL_bufend && SPACE_OR_TAB(*d))
3423 const char minus = (PL_tokenbuf[0] == '-');
3424 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3432 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3437 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3442 if (PL_oldoldbufptr == PL_last_lop)
3443 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3445 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3448 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3450 /* This hack is to get the ${} in the message. */
3452 yyerror("syntax error");
3455 OPERATOR(HASHBRACK);
3457 /* This hack serves to disambiguate a pair of curlies
3458 * as being a block or an anon hash. Normally, expectation
3459 * determines that, but in cases where we're not in a
3460 * position to expect anything in particular (like inside
3461 * eval"") we have to resolve the ambiguity. This code
3462 * covers the case where the first term in the curlies is a
3463 * quoted string. Most other cases need to be explicitly
3464 * disambiguated by prepending a "+" before the opening
3465 * curly in order to force resolution as an anon hash.
3467 * XXX should probably propagate the outer expectation
3468 * into eval"" to rely less on this hack, but that could
3469 * potentially break current behavior of eval"".
3473 if (*s == '\'' || *s == '"' || *s == '`') {
3474 /* common case: get past first string, handling escapes */
3475 for (t++; t < PL_bufend && *t != *s;)
3476 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3480 else if (*s == 'q') {
3483 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3486 /* skip q//-like construct */
3488 char open, close, term;
3491 while (t < PL_bufend && isSPACE(*t))
3493 /* check for q => */
3494 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
3495 OPERATOR(HASHBRACK);
3499 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3503 for (t++; t < PL_bufend; t++) {
3504 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3506 else if (*t == open)
3510 for (t++; t < PL_bufend; t++) {
3511 if (*t == '\\' && t+1 < PL_bufend)
3513 else if (*t == close && --brackets <= 0)
3515 else if (*t == open)
3522 /* skip plain q word */
3523 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3526 else if (isALNUM_lazy_if(t,UTF)) {
3528 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3531 while (t < PL_bufend && isSPACE(*t))
3533 /* if comma follows first term, call it an anon hash */
3534 /* XXX it could be a comma expression with loop modifiers */
3535 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3536 || (*t == '=' && t[1] == '>')))
3537 OPERATOR(HASHBRACK);
3538 if (PL_expect == XREF)
3541 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3547 yylval.ival = CopLINE(PL_curcop);
3548 if (isSPACE(*s) || *s == '#')
3549 PL_copline = NOLINE; /* invalidate current command line number */
3554 if (PL_lex_brackets <= 0)
3555 yyerror("Unmatched right curly bracket");
3557 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3558 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3559 PL_lex_formbrack = 0;
3560 if (PL_lex_state == LEX_INTERPNORMAL) {
3561 if (PL_lex_brackets == 0) {
3562 if (PL_expect & XFAKEBRACK) {
3563 PL_expect &= XENUMMASK;
3564 PL_lex_state = LEX_INTERPEND;
3566 return yylex(); /* ignore fake brackets */
3568 if (*s == '-' && s[1] == '>')
3569 PL_lex_state = LEX_INTERPENDMAYBE;
3570 else if (*s != '[' && *s != '{')
3571 PL_lex_state = LEX_INTERPEND;
3574 if (PL_expect & XFAKEBRACK) {
3575 PL_expect &= XENUMMASK;
3577 return yylex(); /* ignore fake brackets */
3587 if (PL_expect == XOPERATOR) {
3588 if (PL_bufptr == PL_linestart && ckWARN(WARN_SEMICOLON)
3589 && isIDFIRST_lazy_if(s,UTF))
3591 CopLINE_dec(PL_curcop);
3592 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3593 CopLINE_inc(PL_curcop);
3598 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3600 PL_expect = XOPERATOR;
3601 force_ident(PL_tokenbuf, '&');
3605 yylval.ival = (OPpENTERSUB_AMPER<<8);
3624 if (tmp && isSPACE(*s) && ckWARN(WARN_SYNTAX) && strchr("+-*/%.^&|<",tmp))
3625 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Reversed %c= operator",(int)tmp);
3627 if (PL_expect == XSTATE && isALPHA(tmp) &&
3628 (s == PL_linestart+1 || s[-2] == '\n') )
3630 if (PL_in_eval && !PL_rsfp) {
3635 if (strnEQ(s,"=cut",4)) {
3649 PL_doextract = TRUE;
3652 if (PL_lex_brackets < PL_lex_formbrack) {
3654 #ifdef PERL_STRICT_CR
3655 for (t = s; SPACE_OR_TAB(*t); t++) ;
3657 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3659 if (*t == '\n' || *t == '#') {
3671 /* was this !=~ where !~ was meant?
3672 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
3674 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
3675 const char *t = s+1;
3677 while (t < PL_bufend && isSPACE(*t))
3680 if (*t == '/' || *t == '?' ||
3681 ((*t == 'm' || *t == 's' || *t == 'y') && !isALNUM(t[1])) ||
3682 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
3683 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3684 "!=~ should be !~");
3693 if (PL_expect != XOPERATOR) {
3694 if (s[1] != '<' && !strchr(s,'>'))
3697 s = scan_heredoc(s);
3699 s = scan_inputsymbol(s);
3700 TERM(sublex_start());
3705 SHop(OP_LEFT_SHIFT);
3719 SHop(OP_RIGHT_SHIFT);
3728 if (PL_expect == XOPERATOR) {
3729 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3732 return REPORT(','); /* grandfather non-comma-format format */
3736 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3737 PL_tokenbuf[0] = '@';
3738 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3739 sizeof PL_tokenbuf - 1, FALSE);
3740 if (PL_expect == XOPERATOR)
3741 no_op("Array length", s);
3742 if (!PL_tokenbuf[1])
3744 PL_expect = XOPERATOR;
3745 PL_pending_ident = '#';
3749 PL_tokenbuf[0] = '$';
3750 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3751 sizeof PL_tokenbuf - 1, FALSE);
3752 if (PL_expect == XOPERATOR)
3754 if (!PL_tokenbuf[1]) {
3756 yyerror("Final $ should be \\$ or $name");
3760 /* This kludge not intended to be bulletproof. */
3761 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3762 yylval.opval = newSVOP(OP_CONST, 0,
3763 newSViv(PL_compiling.cop_arybase));
3764 yylval.opval->op_private = OPpCONST_ARYBASE;
3770 if (PL_lex_state == LEX_NORMAL)
3773 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3775 PL_tokenbuf[0] = '@';
3776 if (ckWARN(WARN_SYNTAX)) {
3779 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3782 PL_bufptr = skipspace(PL_bufptr);
3783 while (t < PL_bufend && *t != ']')
3785 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3786 "Multidimensional syntax %.*s not supported",
3787 (t - PL_bufptr) + 1, PL_bufptr);
3791 else if (*s == '{') {
3793 PL_tokenbuf[0] = '%';
3794 if (strEQ(PL_tokenbuf+1, "SIG") && ckWARN(WARN_SYNTAX)
3795 && (t = strchr(s, '}')) && (t = strchr(t, '=')))
3797 char tmpbuf[sizeof PL_tokenbuf];
3798 for (t++; isSPACE(*t); t++) ;
3799 if (isIDFIRST_lazy_if(t,UTF)) {
3801 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3802 for (; isSPACE(*t); t++) ;
3803 if (*t == ';' && get_cv(tmpbuf, FALSE))
3804 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3805 "You need to quote \"%s\"", tmpbuf);
3811 PL_expect = XOPERATOR;
3812 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3813 const bool islop = (PL_last_lop == PL_oldoldbufptr);
3814 if (!islop || PL_last_lop_op == OP_GREPSTART)
3815 PL_expect = XOPERATOR;
3816 else if (strchr("$@\"'`q", *s))
3817 PL_expect = XTERM; /* e.g. print $fh "foo" */
3818 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3819 PL_expect = XTERM; /* e.g. print $fh &sub */
3820 else if (isIDFIRST_lazy_if(s,UTF)) {
3821 char tmpbuf[sizeof PL_tokenbuf];
3822 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3823 if ((tmp = keyword(tmpbuf, len))) {
3824 /* binary operators exclude handle interpretations */
3836 PL_expect = XTERM; /* e.g. print $fh length() */
3841 PL_expect = XTERM; /* e.g. print $fh subr() */
3844 else if (isDIGIT(*s))
3845 PL_expect = XTERM; /* e.g. print $fh 3 */
3846 else if (*s == '.' && isDIGIT(s[1]))
3847 PL_expect = XTERM; /* e.g. print $fh .3 */
3848 else if ((*s == '?' || *s == '-' || *s == '+')
3849 && !isSPACE(s[1]) && s[1] != '=')
3850 PL_expect = XTERM; /* e.g. print $fh -1 */
3851 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '=' && s[1] != '/')
3852 PL_expect = XTERM; /* e.g. print $fh /.../
3853 XXX except DORDOR operator */
3854 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3855 PL_expect = XTERM; /* print $fh <<"EOF" */
3857 PL_pending_ident = '$';
3861 if (PL_expect == XOPERATOR)
3863 PL_tokenbuf[0] = '@';
3864 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3865 if (!PL_tokenbuf[1]) {
3868 if (PL_lex_state == LEX_NORMAL)
3870 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3872 PL_tokenbuf[0] = '%';
3874 /* Warn about @ where they meant $. */
3875 if (*s == '[' || *s == '{') {
3876 if (ckWARN(WARN_SYNTAX)) {
3877 const char *t = s + 1;
3878 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3880 if (*t == '}' || *t == ']') {
3882 PL_bufptr = skipspace(PL_bufptr);
3883 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3884 "Scalar value %.*s better written as $%.*s",
3885 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3890 PL_pending_ident = '@';
3893 case '/': /* may be division, defined-or, or pattern */
3894 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
3898 case '?': /* may either be conditional or pattern */
3899 if(PL_expect == XOPERATOR) {
3907 /* A // operator. */
3917 /* Disable warning on "study /blah/" */
3918 if (PL_oldoldbufptr == PL_last_uni
3919 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3920 || memNE(PL_last_uni, "study", 5)
3921 || isALNUM_lazy_if(PL_last_uni+5,UTF)
3924 s = scan_pat(s,OP_MATCH);
3925 TERM(sublex_start());
3929 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3930 #ifdef PERL_STRICT_CR
3933 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3935 && (s == PL_linestart || s[-1] == '\n') )
3937 PL_lex_formbrack = 0;
3941 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3947 yylval.ival = OPf_SPECIAL;
3953 if (PL_expect != XOPERATOR)
3958 case '0': case '1': case '2': case '3': case '4':
3959 case '5': case '6': case '7': case '8': case '9':
3960 s = scan_num(s, &yylval);
3961 DEBUG_T( { S_printbuf(aTHX_ "### Saw number in %s\n", s); } );
3962 if (PL_expect == XOPERATOR)
3967 s = scan_str(s,FALSE,FALSE);
3968 DEBUG_T( { S_printbuf(aTHX_ "### Saw string before %s\n", s); } );
3969 if (PL_expect == XOPERATOR) {
3970 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3973 return REPORT(','); /* grandfather non-comma-format format */
3979 missingterm((char*)0);
3980 yylval.ival = OP_CONST;
3981 TERM(sublex_start());
3984 s = scan_str(s,FALSE,FALSE);
3985 DEBUG_T( { S_printbuf(aTHX_ "### Saw string before %s\n", s); } );
3986 if (PL_expect == XOPERATOR) {
3987 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3990 return REPORT(','); /* grandfather non-comma-format format */
3996 missingterm((char*)0);
3997 yylval.ival = OP_CONST;
3998 /* FIXME. I think that this can be const if char *d is replaced by
3999 more localised variables. */
4000 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
4001 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
4002 yylval.ival = OP_STRINGIFY;
4006 TERM(sublex_start());
4009 s = scan_str(s,FALSE,FALSE);
4010 DEBUG_T( { S_printbuf(aTHX_ "### Saw backtick string before %s\n", s); } );
4011 if (PL_expect == XOPERATOR)
4012 no_op("Backticks",s);
4014 missingterm((char*)0);
4015 yylval.ival = OP_BACKTICK;
4017 TERM(sublex_start());
4021 if (PL_lex_inwhat && isDIGIT(*s) && ckWARN(WARN_SYNTAX))
4022 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
4024 if (PL_expect == XOPERATOR)
4025 no_op("Backslash",s);
4029 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
4030 char *start = s + 2;
4031 while (isDIGIT(*start) || *start == '_')
4033 if (*start == '.' && isDIGIT(start[1])) {
4034 s = scan_num(s, &yylval);
4037 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
4038 else if (!isALPHA(*start) && (PL_expect == XTERM
4039 || PL_expect == XREF || PL_expect == XSTATE
4040 || PL_expect == XTERMORDORDOR)) {
4041 const char c = *start;
4044 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
4047 s = scan_num(s, &yylval);
4054 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
4094 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4096 /* Some keywords can be followed by any delimiter, including ':' */
4097 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
4098 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
4099 (PL_tokenbuf[0] == 'q' &&
4100 strchr("qwxr", PL_tokenbuf[1])))));
4102 /* x::* is just a word, unless x is "CORE" */
4103 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
4107 while (d < PL_bufend && isSPACE(*d))
4108 d++; /* no comments skipped here, or s### is misparsed */
4110 /* Is this a label? */
4111 if (!tmp && PL_expect == XSTATE
4112 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
4114 yylval.pval = savepv(PL_tokenbuf);
4119 /* Check for keywords */
4120 tmp = keyword(PL_tokenbuf, len);
4122 /* Is this a word before a => operator? */
4123 if (*d == '=' && d[1] == '>') {
4126 = (OP*)newSVOP(OP_CONST, 0,
4127 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
4128 yylval.opval->op_private = OPpCONST_BARE;
4132 if (tmp < 0) { /* second-class keyword? */
4133 GV *ogv = Nullgv; /* override (winner) */
4134 GV *hgv = Nullgv; /* hidden (loser) */
4135 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
4137 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
4140 if (GvIMPORTED_CV(gv))
4142 else if (! CvMETHOD(cv))
4146 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
4147 (gv = *gvp) != (GV*)&PL_sv_undef &&
4148 GvCVu(gv) && GvIMPORTED_CV(gv))
4155 tmp = 0; /* overridden by import or by GLOBAL */
4158 && -tmp==KEY_lock /* XXX generalizable kludge */
4160 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
4162 tmp = 0; /* any sub overrides "weak" keyword */
4167 && PL_expect != XOPERATOR
4168 && PL_expect != XTERMORDORDOR)
4170 /* any sub overrides the "err" keyword, except when really an
4171 * operator is expected */
4174 else { /* no override */
4176 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
4177 Perl_warner(aTHX_ packWARN(WARN_MISC),
4178 "dump() better written as CORE::dump()");
4182 if (hgv && tmp != KEY_x && tmp != KEY_CORE
4183 && ckWARN(WARN_AMBIGUOUS)) /* never ambiguous */
4184 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4185 "Ambiguous call resolved as CORE::%s(), %s",
4186 GvENAME(hgv), "qualify as such or use &");
4193 default: /* not a keyword */
4197 const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
4199 /* Get the rest if it looks like a package qualifier */
4201 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
4203 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
4206 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
4207 *s == '\'' ? "'" : "::");
4212 if (PL_expect == XOPERATOR) {
4213 if (PL_bufptr == PL_linestart) {
4214 CopLINE_dec(PL_curcop);
4215 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
4216 CopLINE_inc(PL_curcop);
4219 no_op("Bareword",s);
4222 /* Look for a subroutine with this name in current package,
4223 unless name is "Foo::", in which case Foo is a bearword
4224 (and a package name). */
4227 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
4229 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
4230 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
4231 "Bareword \"%s\" refers to nonexistent package",
4234 PL_tokenbuf[len] = '\0';
4241 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
4244 /* if we saw a global override before, get the right name */
4247 sv = newSVpvn("CORE::GLOBAL::",14);
4248 sv_catpv(sv,PL_tokenbuf);
4251 /* If len is 0, newSVpv does strlen(), which is correct.
4252 If len is non-zero, then it will be the true length,
4253 and so the scalar will be created correctly. */
4254 sv = newSVpv(PL_tokenbuf,len);
4257 /* Presume this is going to be a bareword of some sort. */
4260 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
4261 yylval.opval->op_private = OPpCONST_BARE;
4262 /* UTF-8 package name? */
4263 if (UTF && !IN_BYTES &&
4264 is_utf8_string((U8*)SvPVX_const(sv), SvCUR(sv)))
4267 /* And if "Foo::", then that's what it certainly is. */
4272 /* See if it's the indirect object for a list operator. */
4274 if (PL_oldoldbufptr &&
4275 PL_oldoldbufptr < PL_bufptr &&
4276 (PL_oldoldbufptr == PL_last_lop
4277 || PL_oldoldbufptr == PL_last_uni) &&
4278 /* NO SKIPSPACE BEFORE HERE! */
4279 (PL_expect == XREF ||
4280 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
4282 bool immediate_paren = *s == '(';
4284 /* (Now we can afford to cross potential line boundary.) */
4287 /* Two barewords in a row may indicate method call. */
4289 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
4292 /* If not a declared subroutine, it's an indirect object. */
4293 /* (But it's an indir obj regardless for sort.) */
4294 /* Also, if "_" follows a filetest operator, it's a bareword */
4297 ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
4298 ((!gv || !GvCVu(gv)) &&
4299 (PL_last_lop_op != OP_MAPSTART &&
4300 PL_last_lop_op != OP_GREPSTART))))
4301 || (PL_tokenbuf[0] == '_' && PL_tokenbuf[1] == '\0'
4302 && ((PL_opargs[PL_last_lop_op] & OA_CLASS_MASK) == OA_FILESTATOP))
4305 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
4310 PL_expect = XOPERATOR;
4313 /* Is this a word before a => operator? */
4314 if (*s == '=' && s[1] == '>' && !pkgname) {
4316 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
4317 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
4318 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
4322 /* If followed by a paren, it's certainly a subroutine. */
4325 if (gv && GvCVu(gv)) {
4326 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
4327 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
4332 PL_nextval[PL_nexttoke].opval = yylval.opval;
4333 PL_expect = XOPERATOR;
4339 /* If followed by var or block, call it a method (unless sub) */
4341 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
4342 PL_last_lop = PL_oldbufptr;
4343 PL_last_lop_op = OP_METHOD;
4347 /* If followed by a bareword, see if it looks like indir obj. */
4350 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
4351 && (tmp = intuit_method(s,gv)))
4354 /* Not a method, so call it a subroutine (if defined) */
4356 if (gv && GvCVu(gv)) {
4358 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4359 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4360 "Ambiguous use of -%s resolved as -&%s()",
4361 PL_tokenbuf, PL_tokenbuf);
4362 /* Check for a constant sub */
4364 if ((sv = cv_const_sv(cv))) {
4366 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4367 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4368 yylval.opval->op_private = 0;
4372 /* Resolve to GV now. */
4373 op_free(yylval.opval);
4374 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4375 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4376 PL_last_lop = PL_oldbufptr;
4377 PL_last_lop_op = OP_ENTERSUB;
4378 /* Is there a prototype? */
4381 const char *proto = SvPV_const((SV*)cv, len);
4384 if (*proto == '$' && proto[1] == '\0')
4386 while (*proto == ';')
4388 if (*proto == '&' && *s == '{') {
4389 sv_setpv(PL_subname, PL_curstash ?
4390 "__ANON__" : "__ANON__::__ANON__");
4394 PL_nextval[PL_nexttoke].opval = yylval.opval;
4400 /* Call it a bare word */
4402 if (PL_hints & HINT_STRICT_SUBS)
4403 yylval.opval->op_private |= OPpCONST_STRICT;
4406 if (lastchar != '-') {
4407 if (ckWARN(WARN_RESERVED)) {
4408 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4409 if (!*d && !gv_stashpv(PL_tokenbuf,FALSE))
4410 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
4417 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')
4418 && ckWARN_d(WARN_AMBIGUOUS)) {
4419 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4420 "Operator or semicolon missing before %c%s",
4421 lastchar, PL_tokenbuf);
4422 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4423 "Ambiguous use of %c resolved as operator %c",
4424 lastchar, lastchar);
4430 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4431 newSVpv(CopFILE(PL_curcop),0));
4435 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4436 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4439 case KEY___PACKAGE__:
4440 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4442 ? newSVhek(HvNAME_HEK(PL_curstash))
4449 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4450 const char *pname = "main";
4451 if (PL_tokenbuf[2] == 'D')
4452 pname = HvNAME_get(PL_curstash ? PL_curstash : PL_defstash);
4453 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4456 GvIOp(gv) = newIO();
4457 IoIFP(GvIOp(gv)) = PL_rsfp;
4458 #if defined(HAS_FCNTL) && defined(F_SETFD)
4460 const int fd = PerlIO_fileno(PL_rsfp);
4461 fcntl(fd,F_SETFD,fd >= 3);
4464 /* Mark this internal pseudo-handle as clean */
4465 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4467 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4468 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4469 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4471 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4472 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4473 /* if the script was opened in binmode, we need to revert
4474 * it to text mode for compatibility; but only iff it has CRs
4475 * XXX this is a questionable hack at best. */
4476 if (PL_bufend-PL_bufptr > 2
4477 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4480 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4481 loc = PerlIO_tell(PL_rsfp);
4482 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4485 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4487 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4488 #endif /* NETWARE */
4489 #ifdef PERLIO_IS_STDIO /* really? */
4490 # if defined(__BORLANDC__)
4491 /* XXX see note in do_binmode() */
4492 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4496 PerlIO_seek(PL_rsfp, loc, 0);
4500 #ifdef PERLIO_LAYERS
4503 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4504 else if (PL_encoding) {
4511 XPUSHs(PL_encoding);
4513 call_method("name", G_SCALAR);
4517 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
4518 Perl_form(aTHX_ ":encoding(%"SVf")",
4536 if (PL_expect == XSTATE) {
4543 if (*s == ':' && s[1] == ':') {
4546 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4547 if (!(tmp = keyword(PL_tokenbuf, len)))
4548 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4551 else if (tmp == KEY_require || tmp == KEY_do)
4552 /* that's a way to remember we saw "CORE::" */
4565 LOP(OP_ACCEPT,XTERM);
4571 LOP(OP_ATAN2,XTERM);
4577 LOP(OP_BINMODE,XTERM);
4580 LOP(OP_BLESS,XTERM);
4589 /* When 'use switch' is in effect, continue has a dual
4590 life as a control operator. */
4592 if (!FEATURE_IS_ENABLED("switch", 6))
4595 /* We have to disambiguate the two senses of
4596 "continue". If the next token is a '{' then
4597 treat it as the start of a continue block;
4598 otherwise treat it as a control operator.
4609 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4626 if (!PL_cryptseen) {
4627 PL_cryptseen = TRUE;
4631 LOP(OP_CRYPT,XTERM);
4634 LOP(OP_CHMOD,XTERM);
4637 LOP(OP_CHOWN,XTERM);
4640 LOP(OP_CONNECT,XTERM);
4659 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4660 if (orig_keyword == KEY_do) {
4669 PL_hints |= HINT_BLOCK_SCOPE;
4679 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4680 LOP(OP_DBMOPEN,XTERM);
4686 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4693 yylval.ival = CopLINE(PL_curcop);
4707 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4708 UNIBRACK(OP_ENTEREVAL);
4726 case KEY_endhostent:
4732 case KEY_endservent:
4735 case KEY_endprotoent:
4746 yylval.ival = CopLINE(PL_curcop);
4748 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4750 if ((PL_bufend - p) >= 3 &&
4751 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4753 else if ((PL_bufend - p) >= 4 &&
4754 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4757 if (isIDFIRST_lazy_if(p,UTF)) {
4758 p = scan_ident(p, PL_bufend,
4759 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4763 Perl_croak(aTHX_ "Missing $ on loop variable");
4768 LOP(OP_FORMLINE,XTERM);
4774 LOP(OP_FCNTL,XTERM);
4780 LOP(OP_FLOCK,XTERM);
4789 LOP(OP_GREPSTART, XREF);
4792 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4807 case KEY_getpriority:
4808 LOP(OP_GETPRIORITY,XTERM);
4810 case KEY_getprotobyname:
4813 case KEY_getprotobynumber:
4814 LOP(OP_GPBYNUMBER,XTERM);
4816 case KEY_getprotoent:
4828 case KEY_getpeername:
4829 UNI(OP_GETPEERNAME);
4831 case KEY_gethostbyname:
4834 case KEY_gethostbyaddr:
4835 LOP(OP_GHBYADDR,XTERM);
4837 case KEY_gethostent:
4840 case KEY_getnetbyname:
4843 case KEY_getnetbyaddr:
4844 LOP(OP_GNBYADDR,XTERM);
4849 case KEY_getservbyname:
4850 LOP(OP_GSBYNAME,XTERM);
4852 case KEY_getservbyport:
4853 LOP(OP_GSBYPORT,XTERM);
4855 case KEY_getservent:
4858 case KEY_getsockname:
4859 UNI(OP_GETSOCKNAME);
4861 case KEY_getsockopt:
4862 LOP(OP_GSOCKOPT,XTERM);
4877 yylval.ival = CopLINE(PL_curcop);
4888 yylval.ival = CopLINE(PL_curcop);
4892 LOP(OP_INDEX,XTERM);
4898 LOP(OP_IOCTL,XTERM);
4910 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4942 LOP(OP_LISTEN,XTERM);
4951 s = scan_pat(s,OP_MATCH);
4952 TERM(sublex_start());
4955 LOP(OP_MAPSTART, XREF);
4958 LOP(OP_MKDIR,XTERM);
4961 LOP(OP_MSGCTL,XTERM);
4964 LOP(OP_MSGGET,XTERM);
4967 LOP(OP_MSGRCV,XTERM);
4970 LOP(OP_MSGSND,XTERM);
4976 if (isIDFIRST_lazy_if(s,UTF)) {
4977 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4978 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4980 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4981 if (!PL_in_my_stash) {
4984 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4992 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4999 s = tokenize_use(0, s);
5003 if (*s == '(' || (s = skipspace(s), *s == '('))
5010 if (isIDFIRST_lazy_if(s,UTF)) {
5012 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
5013 for (t=d; *t && isSPACE(*t); t++) ;
5014 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
5016 && !(t[0] == '=' && t[1] == '>')
5018 int len = (int)(d-s);
5019 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
5020 "Precedence problem: open %.*s should be open(%.*s)",
5027 yylval.ival = OP_OR;
5037 LOP(OP_OPEN_DIR,XTERM);
5040 checkcomma(s,PL_tokenbuf,"filehandle");
5044 checkcomma(s,PL_tokenbuf,"filehandle");
5063 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5067 LOP(OP_PIPE_OP,XTERM);
5070 s = scan_str(s,FALSE,FALSE);
5072 missingterm((char*)0);
5073 yylval.ival = OP_CONST;
5074 TERM(sublex_start());
5080 s = scan_str(s,FALSE,FALSE);
5082 missingterm((char*)0);
5083 PL_expect = XOPERATOR;
5085 if (SvCUR(PL_lex_stuff)) {
5088 d = SvPV_force(PL_lex_stuff, len);
5091 for (; isSPACE(*d) && len; --len, ++d) ;
5094 if (!warned && ckWARN(WARN_QW)) {
5095 for (; !isSPACE(*d) && len; --len, ++d) {
5097 Perl_warner(aTHX_ packWARN(WARN_QW),
5098 "Possible attempt to separate words with commas");
5101 else if (*d == '#') {
5102 Perl_warner(aTHX_ packWARN(WARN_QW),
5103 "Possible attempt to put comments in qw() list");
5109 for (; !isSPACE(*d) && len; --len, ++d) ;
5111 sv = newSVpvn(b, d-b);
5112 if (DO_UTF8(PL_lex_stuff))
5114 words = append_elem(OP_LIST, words,
5115 newSVOP(OP_CONST, 0, tokeq(sv)));
5119 PL_nextval[PL_nexttoke].opval = words;
5124 SvREFCNT_dec(PL_lex_stuff);
5125 PL_lex_stuff = Nullsv;
5131 s = scan_str(s,FALSE,FALSE);
5133 missingterm((char*)0);
5134 yylval.ival = OP_STRINGIFY;
5135 if (SvIVX(PL_lex_stuff) == '\'')
5136 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
5137 TERM(sublex_start());
5140 s = scan_pat(s,OP_QR);
5141 TERM(sublex_start());
5144 s = scan_str(s,FALSE,FALSE);
5146 missingterm((char*)0);
5147 yylval.ival = OP_BACKTICK;
5149 TERM(sublex_start());
5157 s = force_version(s, FALSE);
5159 else if (*s != 'v' || !isDIGIT(s[1])
5160 || (s = force_version(s, TRUE), *s == 'v'))
5162 *PL_tokenbuf = '\0';
5163 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5164 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
5165 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
5167 yyerror("<> should be quotes");
5169 if (orig_keyword == KEY_require) {
5177 PL_last_uni = PL_oldbufptr;
5178 PL_last_lop_op = OP_REQUIRE;
5180 return REPORT( (int)REQUIRE );
5186 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5190 LOP(OP_RENAME,XTERM);
5199 LOP(OP_RINDEX,XTERM);
5209 UNIDOR(OP_READLINE);
5222 LOP(OP_REVERSE,XTERM);
5225 UNIDOR(OP_READLINK);
5233 TERM(sublex_start());
5235 TOKEN(1); /* force error */
5238 checkcomma(s,PL_tokenbuf,"filehandle");
5248 LOP(OP_SELECT,XTERM);
5254 LOP(OP_SEMCTL,XTERM);
5257 LOP(OP_SEMGET,XTERM);
5260 LOP(OP_SEMOP,XTERM);
5266 LOP(OP_SETPGRP,XTERM);
5268 case KEY_setpriority:
5269 LOP(OP_SETPRIORITY,XTERM);
5271 case KEY_sethostent:
5277 case KEY_setservent:
5280 case KEY_setprotoent:
5290 LOP(OP_SEEKDIR,XTERM);
5292 case KEY_setsockopt:
5293 LOP(OP_SSOCKOPT,XTERM);
5299 LOP(OP_SHMCTL,XTERM);
5302 LOP(OP_SHMGET,XTERM);
5305 LOP(OP_SHMREAD,XTERM);
5308 LOP(OP_SHMWRITE,XTERM);
5311 LOP(OP_SHUTDOWN,XTERM);
5320 LOP(OP_SOCKET,XTERM);
5322 case KEY_socketpair:
5323 LOP(OP_SOCKPAIR,XTERM);
5326 checkcomma(s,PL_tokenbuf,"subroutine name");
5328 if (*s == ';' || *s == ')') /* probably a close */
5329 Perl_croak(aTHX_ "sort is now a reserved word");
5331 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5335 LOP(OP_SPLIT,XTERM);
5338 LOP(OP_SPRINTF,XTERM);
5341 LOP(OP_SPLICE,XTERM);
5356 LOP(OP_SUBSTR,XTERM);
5362 char tmpbuf[sizeof PL_tokenbuf];
5363 SSize_t tboffset = 0;
5364 expectation attrful;
5365 bool have_name, have_proto, bad_proto;
5366 const int key = tmp;
5370 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
5371 (*s == ':' && s[1] == ':'))
5374 attrful = XATTRBLOCK;
5375 /* remember buffer pos'n for later force_word */
5376 tboffset = s - PL_oldbufptr;
5377 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
5378 if (strchr(tmpbuf, ':'))
5379 sv_setpv(PL_subname, tmpbuf);
5381 sv_setsv(PL_subname,PL_curstname);
5382 sv_catpvn(PL_subname,"::",2);
5383 sv_catpvn(PL_subname,tmpbuf,len);
5390 Perl_croak(aTHX_ "Missing name in \"my sub\"");
5391 PL_expect = XTERMBLOCK;
5392 attrful = XATTRTERM;
5393 sv_setpvn(PL_subname,"?",1);
5397 if (key == KEY_format) {
5399 PL_lex_formbrack = PL_lex_brackets + 1;
5401 (void) force_word(PL_oldbufptr + tboffset, WORD,
5406 /* Look for a prototype */
5410 s = scan_str(s,FALSE,FALSE);
5412 Perl_croak(aTHX_ "Prototype not terminated");
5413 /* strip spaces and check for bad characters */
5414 d = SvPVX(PL_lex_stuff);
5417 for (p = d; *p; ++p) {
5420 if (!strchr("$@%*;[]&\\", *p))
5425 if (bad_proto && ckWARN(WARN_SYNTAX))
5426 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5427 "Illegal character in prototype for %"SVf" : %s",
5429 SvCUR_set(PL_lex_stuff, tmp);
5437 if (*s == ':' && s[1] != ':')
5438 PL_expect = attrful;
5439 else if (*s != '{' && key == KEY_sub) {
5441 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
5443 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, PL_subname);
5447 PL_nextval[PL_nexttoke].opval =
5448 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5449 PL_lex_stuff = Nullsv;
5453 sv_setpv(PL_subname,
5454 PL_curstash ? "__ANON__" : "__ANON__::__ANON__");
5457 (void) force_word(PL_oldbufptr + tboffset, WORD,
5466 LOP(OP_SYSTEM,XREF);
5469 LOP(OP_SYMLINK,XTERM);
5472 LOP(OP_SYSCALL,XTERM);
5475 LOP(OP_SYSOPEN,XTERM);
5478 LOP(OP_SYSSEEK,XTERM);
5481 LOP(OP_SYSREAD,XTERM);
5484 LOP(OP_SYSWRITE,XTERM);
5488 TERM(sublex_start());
5509 LOP(OP_TRUNCATE,XTERM);
5521 yylval.ival = CopLINE(PL_curcop);
5525 yylval.ival = CopLINE(PL_curcop);
5529 LOP(OP_UNLINK,XTERM);
5535 LOP(OP_UNPACK,XTERM);
5538 LOP(OP_UTIME,XTERM);
5544 LOP(OP_UNSHIFT,XTERM);
5547 s = tokenize_use(1, s);
5557 yylval.ival = CopLINE(PL_curcop);
5561 yylval.ival = CopLINE(PL_curcop);
5565 PL_hints |= HINT_BLOCK_SCOPE;
5572 LOP(OP_WAITPID,XTERM);
5581 ctl_l[0] = toCTRL('L');
5583 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5586 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5591 if (PL_expect == XOPERATOR)
5597 yylval.ival = OP_XOR;
5602 TERM(sublex_start());
5607 #pragma segment Main
5611 S_pending_ident(pTHX)
5614 register I32 tmp = 0;
5615 /* pit holds the identifier we read and pending_ident is reset */
5616 char pit = PL_pending_ident;
5617 PL_pending_ident = 0;
5619 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5620 "### Pending identifier '%s'\n", PL_tokenbuf); });
5622 /* if we're in a my(), we can't allow dynamics here.
5623 $foo'bar has already been turned into $foo::bar, so
5624 just check for colons.
5626 if it's a legal name, the OP is a PADANY.
5629 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5630 if (strchr(PL_tokenbuf,':'))
5631 yyerror(Perl_form(aTHX_ "No package name allowed for "
5632 "variable %s in \"our\"",
5634 tmp = allocmy(PL_tokenbuf);
5637 if (strchr(PL_tokenbuf,':'))
5638 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5640 yylval.opval = newOP(OP_PADANY, 0);
5641 yylval.opval->op_targ = allocmy(PL_tokenbuf);
5647 build the ops for accesses to a my() variable.
5649 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5650 then used in a comparison. This catches most, but not
5651 all cases. For instance, it catches
5652 sort { my($a); $a <=> $b }
5654 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5655 (although why you'd do that is anyone's guess).
5658 if (!strchr(PL_tokenbuf,':')) {
5660 tmp = pad_findmy(PL_tokenbuf);
5661 if (tmp != NOT_IN_PAD) {
5662 /* might be an "our" variable" */
5663 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
5664 /* build ops for a bareword */
5665 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
5666 HEK * const stashname = HvNAME_HEK(stash);
5667 SV * const sym = newSVhek(stashname);
5668 sv_catpvn(sym, "::", 2);
5669 sv_catpv(sym, PL_tokenbuf+1);
5670 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5671 yylval.opval->op_private = OPpCONST_ENTERED;
5674 ? (GV_ADDMULTI | GV_ADDINEVAL)
5677 ((PL_tokenbuf[0] == '$') ? SVt_PV
5678 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5683 /* if it's a sort block and they're naming $a or $b */
5684 if (PL_last_lop_op == OP_SORT &&
5685 PL_tokenbuf[0] == '$' &&
5686 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5689 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5690 d < PL_bufend && *d != '\n';
5693 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5694 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5700 yylval.opval = newOP(OP_PADANY, 0);
5701 yylval.opval->op_targ = tmp;
5707 Whine if they've said @foo in a doublequoted string,
5708 and @foo isn't a variable we can find in the symbol
5711 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5712 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5713 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5714 && ckWARN(WARN_AMBIGUOUS))
5716 /* Downgraded from fatal to warning 20000522 mjd */
5717 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5718 "Possible unintended interpolation of %s in string",
5723 /* build ops for a bareword */
5724 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5725 yylval.opval->op_private = OPpCONST_ENTERED;
5729 ? (GV_ADDMULTI | GV_ADDINEVAL)
5730 /* If the identifier refers to a stash, don't autovivify it.
5731 * Change 24660 had the side effect of causing symbol table
5732 * hashes to always be defined, even if they were freshly
5733 * created and the only reference in the entire program was
5734 * the single statement with the defined %foo::bar:: test.
5735 * It appears that all code in the wild doing this actually
5736 * wants to know whether sub-packages have been loaded, so
5737 * by avoiding auto-vivifying symbol tables, we ensure that
5738 * defined %foo::bar:: continues to be false, and the existing
5739 * tests still give the expected answers, even though what
5740 * they're actually testing has now changed subtly.
5742 : !(*PL_tokenbuf == '%' && *(d = PL_tokenbuf + strlen(PL_tokenbuf) - 1) == ':' && d[-1] == ':'),
5743 ((PL_tokenbuf[0] == '$') ? SVt_PV
5744 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5750 * The following code was generated by perl_keyword.pl.
5754 Perl_keyword (pTHX_ const char *name, I32 len)
5758 case 1: /* 5 tokens of length 1 */
5790 case 2: /* 18 tokens of length 2 */
5936 case 3: /* 29 tokens of length 3 */
5940 if (name[1] == 'N' &&
6003 if (name[1] == 'i' &&
6043 if (name[1] == 'o' &&
6052 if (name[1] == 'e' &&
6061 if (name[1] == 'n' &&
6070 if (name[1] == 'o' &&
6079 if (name[1] == 'a' &&
6088 if (name[1] == 'o' &&
6150 if (name[1] == 'e' &&
6164 return (FEATURE_IS_ENABLED("say", 3) ? -KEY_say : 0);
6190 if (name[1] == 'i' &&
6199 if (name[1] == 's' &&
6208 if (name[1] == 'e' &&
6217 if (name[1] == 'o' &&
6229 case 4: /* 41 tokens of length 4 */
6233 if (name[1] == 'O' &&
6243 if (name[1] == 'N' &&
6253 if (name[1] == 'i' &&
6263 if (name[1] == 'h' &&
6273 if (name[1] == 'u' &&
6286 if (name[2] == 'c' &&
6295 if (name[2] == 's' &&
6304 if (name[2] == 'a' &&
6340 if (name[1] == 'o' &&
6353 if (name[2] == 't' &&
6362 if (name[2] == 'o' &&
6371 if (name[2] == 't' &&
6380 if (name[2] == 'e' &&
6393 if (name[1] == 'o' &&
6406 if (name[2] == 'y' &&
6415 if (name[2] == 'l' &&
6431 if (name[2] == 's' &&
6440 if (name[2] == 'n' &&
6449 if (name[2] == 'c' &&
6462 if (name[1] == 'e' &&
6472 if (name[1] == 'p' &&
6485 if (name[2] == 'c' &&
6494 if (name[2] == 'p' &&
6503 if (name[2] == 's' &&
6519 if (name[2] == 'n' &&
6589 if (name[2] == 'r' &&
6598 if (name[2] == 'r' &&
6607 if (name[2] == 'a' &&
6623 if (name[2] == 'l' &&
6685 if (name[2] == 'e' &&
6688 return (FEATURE_IS_ENABLED("switch", 6) ? KEY_when : 0);
6701 case 5: /* 38 tokens of length 5 */
6705 if (name[1] == 'E' &&
6716 if (name[1] == 'H' &&
6730 if (name[2] == 'a' &&
6740 if (name[2] == 'a' &&
6757 if (name[2] == 'e' &&
6767 if (name[2] == 'e' &&
6771 return (FEATURE_IS_ENABLED("switch", 6) ? -KEY_break : 0);
6787 if (name[3] == 'i' &&
6796 if (name[3] == 'o' &&
6832 if (name[2] == 'o' &&
6842 if (name[2] == 'y' &&
6856 if (name[1] == 'l' &&
6870 if (name[2] == 'n' &&
6880 if (name[2] == 'o' &&
6894 if (name[1] == 'i' &&
6899 return (FEATURE_IS_ENABLED("switch", 6) ? KEY_given : 0);
6908 if (name[2] == 'd' &&
6918 if (name[2] == 'c' &&
6935 if (name[2] == 'c' &&
6945 if (name[2] == 't' &&
6959 if (name[1] == 'k' &&
6970 if (name[1] == 'r' &&
6984 if (name[2] == 's' &&
6994 if (name[2] == 'd' &&
7011 if (name[2] == 'm' &&
7021 if (name[2] == 'i' &&
7031 if (name[2] == 'e' &&
7041 if (name[2] == 'l' &&
7051 if (name[2] == 'a' &&
7061 if (name[2] == 'u' &&
7075 if (name[1] == 'i' &&
7089 if (name[2] == 'a' &&
7102 if (name[3] == 'e' &&
7137 if (name[2] == 'i' &&
7154 if (name[2] == 'i' &&
7164 if (name[2] == 'i' &&
7181 case 6: /* 33 tokens of length 6 */
7185 if (name[1] == 'c' &&
7200 if (name[2] == 'l' &&
7211 if (name[2] == 'r' &&
7226 if (name[1] == 'e' &&
7241 if (name[2] == 's' &&
7246 if(ckWARN_d(WARN_SYNTAX))
7247 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
7253 if (name[2] == 'i' &&
7271 if (name[2] == 'l' &&
7282 if (name[2] == 'r' &&
7297 if (name[1] == 'm' &&
7312 if (name[2] == 'n' &&
7323 if (name[2] == 's' &&
7338 if (name[1] == 's' &&
7344 if (name[4] == 't' &&
7353 if (name[4] == 'e' &&
7362 if (name[4] == 'c' &&
7371 if (name[4] == 'n' &&
7387 if (name[1] == 'r' &&
7405 if (name[3] == 'a' &&
7415 if (name[3] == 'u' &&
7429 if (name[2] == 'n' &&
7447 if (name[2] == 'a' &&
7461 if (name[3] == 'e' &&
7474 if (name[4] == 't' &&
7483 if (name[4] == 'e' &&
7505 if (name[4] == 't' &&
7514 if (name[4] == 'e' &&
7530 if (name[2] == 'c' &&
7541 if (name[2] == 'l' &&
7552 if (name[2] == 'b' &&
7563 if (name[2] == 's' &&
7586 if (name[4] == 's' &&
7595 if (name[4] == 'n' &&
7608 if (name[3] == 'a' &&
7625 if (name[1] == 'a' &&
7640 case 7: /* 29 tokens of length 7 */
7644 if (name[1] == 'E' &&
7657 if (name[1] == '_' &&
7670 if (name[1] == 'i' &&
7677 return -KEY_binmode;
7683 if (name[1] == 'o' &&
7690 return -KEY_connect;
7699 if (name[2] == 'm' &&
7705 return -KEY_dbmopen;
7716 if (name[4] == 'u' &&
7720 return (FEATURE_IS_ENABLED("switch", 6) ? KEY_default : 0);
7726 if (name[4] == 'n' &&
7747 if (name[1] == 'o' &&
7760 if (name[1] == 'e' &&
7767 if (name[5] == 'r' &&
7770 return -KEY_getpgrp;
7776 if (name[5] == 'i' &&
7779 return -KEY_getppid;
7792 if (name[1] == 'c' &&
7799 return -KEY_lcfirst;
7805 if (name[1] == 'p' &&
7812 return -KEY_opendir;
7818 if (name[1] == 'a' &&
7836 if (name[3] == 'd' &&
7841 return -KEY_readdir;
7847 if (name[3] == 'u' &&
7858 if (name[3] == 'e' &&
7863 return -KEY_reverse;
7882 if (name[3] == 'k' &&
7887 return -KEY_seekdir;
7893 if (name[3] == 'p' &&
7898 return -KEY_setpgrp;
7908 if (name[2] == 'm' &&
7914 return -KEY_shmread;
7920 if (name[2] == 'r' &&
7926 return -KEY_sprintf;
7935 if (name[3] == 'l' &&
7940 return -KEY_symlink;
7949 if (name[4] == 'a' &&
7953 return -KEY_syscall;
7959 if (name[4] == 'p' &&
7963 return -KEY_sysopen;
7969 if (name[4] == 'e' &&
7973 return -KEY_sysread;
7979 if (name[4] == 'e' &&
7983 return -KEY_sysseek;
8001 if (name[1] == 'e' &&
8008 return -KEY_telldir;
8017 if (name[2] == 'f' &&
8023 return -KEY_ucfirst;
8029 if (name[2] == 's' &&
8035 return -KEY_unshift;
8045 if (name[1] == 'a' &&
8052 return -KEY_waitpid;
8061 case 8: /* 26 tokens of length 8 */
8065 if (name[1] == 'U' &&
8073 return KEY_AUTOLOAD;
8084 if (name[3] == 'A' &&
8090 return KEY___DATA__;
8096 if (name[3] == 'I' &&
8102 return -KEY___FILE__;
8108 if (name[3] == 'I' &&
8114 return -KEY___LINE__;
8130 if (name[2] == 'o' &&
8137 return -KEY_closedir;
8143 if (name[2] == 'n' &&
8150 return -KEY_continue;
8160 if (name[1] == 'b' &&
8168 return -KEY_dbmclose;
8174 if (name[1] == 'n' &&
8180 if (name[4] == 'r' &&
8185 return -KEY_endgrent;
8191 if (name[4] == 'w' &&
8196 return -KEY_endpwent;
8209 if (name[1] == 'o' &&
8217 return -KEY_formline;
8223 if (name[1] == 'e' &&
8234 if (name[6] == 'n' &&
8237 return -KEY_getgrent;
8243 if (name[6] == 'i' &&
8246 return -KEY_getgrgid;
8252 if (name[6] == 'a' &&
8255 return -KEY_getgrnam;
8268 if (name[4] == 'o' &&
8273 return -KEY_getlogin;
8284 if (name[6] == 'n' &&
8287 return -KEY_getpwent;
8293 if (name[6] == 'a' &&
8296 return -KEY_getpwnam;
8302 if (name[6] == 'i' &&
8305 return -KEY_getpwuid;
8325 if (name[1] == 'e' &&
8332 if (name[5] == 'i' &&
8339 return -KEY_readline;
8344 return -KEY_readlink;
8355 if (name[5] == 'i' &&
8359 return -KEY_readpipe;
8380 if (name[4] == 'r' &&
8385 return -KEY_setgrent;
8391 if (name[4] == 'w' &&
8396 return -KEY_setpwent;
8412 if (name[3] == 'w' &&
8418 return -KEY_shmwrite;
8424 if (name[3] == 't' &&
8430 return -KEY_shutdown;
8440 if (name[2] == 's' &&
8447 return -KEY_syswrite;
8457 if (name[1] == 'r' &&
8465 return -KEY_truncate;
8474 case 9: /* 8 tokens of length 9 */
8478 if (name[1] == 'n' &&
8487 return -KEY_endnetent;
8493 if (name[1] == 'e' &&
8502 return -KEY_getnetent;
8508 if (name[1] == 'o' &&
8517 return -KEY_localtime;
8523 if (name[1] == 'r' &&
8532 return KEY_prototype;
8538 if (name[1] == 'u' &&
8547 return -KEY_quotemeta;
8553 if (name[1] == 'e' &&
8562 return -KEY_rewinddir;
8568 if (name[1] == 'e' &&
8577 return -KEY_setnetent;
8583 if (name[1] == 'a' &&
8592 return -KEY_wantarray;
8601 case 10: /* 9 tokens of length 10 */
8605 if (name[1] == 'n' &&
8611 if (name[4] == 'o' &&
8618 return -KEY_endhostent;
8624 if (name[4] == 'e' &&
8631 return -KEY_endservent;
8644 if (name[1] == 'e' &&
8650 if (name[4] == 'o' &&
8657 return -KEY_gethostent;
8666 if (name[5] == 'r' &&
8672 return -KEY_getservent;
8678 if (name[5] == 'c' &&
8684 return -KEY_getsockopt;
8709 if (name[4] == 'o' &&
8716 return -KEY_sethostent;
8725 if (name[5] == 'r' &&
8731 return -KEY_setservent;
8737 if (name[5] == 'c' &&
8743 return -KEY_setsockopt;
8760 if (name[2] == 'c' &&
8769 return -KEY_socketpair;
8782 case 11: /* 8 tokens of length 11 */
8786 if (name[1] == '_' &&
8797 return -KEY___PACKAGE__;
8803 if (name[1] == 'n' &&
8814 return -KEY_endprotoent;
8820 if (name[1] == 'e' &&
8829 if (name[5] == 'e' &&
8836 return -KEY_getpeername;
8845 if (name[6] == 'o' &&
8851 return -KEY_getpriority;
8857 if (name[6] == 't' &&
8863 return -KEY_getprotoent;
8877 if (name[4] == 'o' &&
8885 return -KEY_getsockname;
8898 if (name[1] == 'e' &&
8906 if (name[6] == 'o' &&
8912 return -KEY_setpriority;
8918 if (name[6] == 't' &&
8924 return -KEY_setprotoent;
8940 case 12: /* 2 tokens of length 12 */
8941 if (name[0] == 'g' &&
8953 if (name[9] == 'd' &&
8956 { /* getnetbyaddr */
8957 return -KEY_getnetbyaddr;
8963 if (name[9] == 'a' &&
8966 { /* getnetbyname */
8967 return -KEY_getnetbyname;
8979 case 13: /* 4 tokens of length 13 */
8980 if (name[0] == 'g' &&
8987 if (name[4] == 'o' &&
8996 if (name[10] == 'd' &&
8999 { /* gethostbyaddr */
9000 return -KEY_gethostbyaddr;
9006 if (name[10] == 'a' &&
9009 { /* gethostbyname */
9010 return -KEY_gethostbyname;
9023 if (name[4] == 'e' &&
9032 if (name[10] == 'a' &&
9035 { /* getservbyname */
9036 return -KEY_getservbyname;
9042 if (name[10] == 'o' &&
9045 { /* getservbyport */
9046 return -KEY_getservbyport;
9065 case 14: /* 1 tokens of length 14 */
9066 if (name[0] == 'g' &&
9080 { /* getprotobyname */
9081 return -KEY_getprotobyname;
9086 case 16: /* 1 tokens of length 16 */
9087 if (name[0] == 'g' &&
9103 { /* getprotobynumber */
9104 return -KEY_getprotobynumber;
9118 S_checkcomma(pTHX_ register char *s, const char *name, const char *what)
9122 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
9123 if (ckWARN(WARN_SYNTAX)) {
9125 for (w = s+2; *w && level; w++) {
9132 for (; *w && isSPACE(*w); w++) ;
9133 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
9134 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
9135 "%s (...) interpreted as function",name);
9138 while (s < PL_bufend && isSPACE(*s))
9142 while (s < PL_bufend && isSPACE(*s))
9144 if (isIDFIRST_lazy_if(s,UTF)) {
9146 while (isALNUM_lazy_if(s,UTF))
9148 while (s < PL_bufend && isSPACE(*s))
9152 *s = '\0'; /* XXX If we didn't do this, we could const a lot of toke.c */
9153 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
9157 Perl_croak(aTHX_ "No comma allowed after %s", what);
9162 /* Either returns sv, or mortalizes sv and returns a new SV*.
9163 Best used as sv=new_constant(..., sv, ...).
9164 If s, pv are NULL, calls subroutine with one argument,
9165 and type is used with error messages only. */
9168 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, SV *sv, SV *pv,
9172 HV * const table = GvHV(PL_hintgv); /* ^H */
9176 const char *why1 = "", *why2 = "", *why3 = "";
9178 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
9181 why2 = strEQ(key,"charnames")
9182 ? "(possibly a missing \"use charnames ...\")"
9184 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
9185 (type ? type: "undef"), why2);
9187 /* This is convoluted and evil ("goto considered harmful")
9188 * but I do not understand the intricacies of all the different
9189 * failure modes of %^H in here. The goal here is to make
9190 * the most probable error message user-friendly. --jhi */
9195 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
9196 (type ? type: "undef"), why1, why2, why3);
9198 yyerror(SvPVX_const(msg));
9202 cvp = hv_fetch(table, key, strlen(key), FALSE);
9203 if (!cvp || !SvOK(*cvp)) {
9206 why3 = "} is not defined";
9209 sv_2mortal(sv); /* Parent created it permanently */
9212 pv = sv_2mortal(newSVpvn(s, len));
9214 typesv = sv_2mortal(newSVpv(type, 0));
9216 typesv = &PL_sv_undef;
9218 PUSHSTACKi(PERLSI_OVERLOAD);
9230 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
9234 /* Check the eval first */
9235 if (!PL_in_eval && SvTRUE(ERRSV)) {
9236 sv_catpv(ERRSV, "Propagated");
9237 yyerror(SvPV_nolen_const(ERRSV)); /* Duplicates the message inside eval */
9239 res = SvREFCNT_inc(sv);
9243 (void)SvREFCNT_inc(res);
9252 why1 = "Call to &{$^H{";
9254 why3 = "}} did not return a defined value";
9262 /* Returns a NUL terminated string, with the length of the string written to
9266 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
9268 register char *d = dest;
9269 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
9272 Perl_croak(aTHX_ ident_too_long);
9273 if (isALNUM(*s)) /* UTF handled below */
9275 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
9280 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
9284 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
9285 char *t = s + UTF8SKIP(s);
9286 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
9288 if (d + (t - s) > e)
9289 Perl_croak(aTHX_ ident_too_long);
9290 Copy(s, d, t - s, char);
9303 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
9307 char *bracket = Nullch;
9313 e = d + destlen - 3; /* two-character token, ending NUL */
9315 while (isDIGIT(*s)) {
9317 Perl_croak(aTHX_ ident_too_long);
9324 Perl_croak(aTHX_ ident_too_long);
9325 if (isALNUM(*s)) /* UTF handled below */
9327 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
9332 else if (*s == ':' && s[1] == ':') {
9336 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
9337 char *t = s + UTF8SKIP(s);
9338 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
9340 if (d + (t - s) > e)
9341 Perl_croak(aTHX_ ident_too_long);
9342 Copy(s, d, t - s, char);
9353 if (PL_lex_state != LEX_NORMAL)
9354 PL_lex_state = LEX_INTERPENDMAYBE;
9357 if (*s == '$' && s[1] &&
9358 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
9371 if (*d == '^' && *s && isCONTROLVAR(*s)) {
9376 if (isSPACE(s[-1])) {
9378 const char ch = *s++;
9379 if (!SPACE_OR_TAB(ch)) {
9385 if (isIDFIRST_lazy_if(d,UTF)) {
9389 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
9391 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
9394 Copy(s, d, e - s, char);
9399 while ((isALNUM(*s) || *s == ':') && d < e)
9402 Perl_croak(aTHX_ ident_too_long);
9405 while (s < send && SPACE_OR_TAB(*s)) s++;
9406 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
9407 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
9408 const char *brack = *s == '[' ? "[...]" : "{...}";
9409 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
9410 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
9411 funny, dest, brack, funny, dest, brack);
9414 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
9418 /* Handle extended ${^Foo} variables
9419 * 1999-02-27 mjd-perl-patch@plover.com */
9420 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
9424 while (isALNUM(*s) && d < e) {
9428 Perl_croak(aTHX_ ident_too_long);
9433 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
9434 PL_lex_state = LEX_INTERPEND;
9439 if (PL_lex_state == LEX_NORMAL) {
9440 if (ckWARN(WARN_AMBIGUOUS) &&
9441 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
9443 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
9444 "Ambiguous use of %c{%s} resolved to %c%s",
9445 funny, dest, funny, dest);
9450 s = bracket; /* let the parser handle it */
9454 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
9455 PL_lex_state = LEX_INTERPEND;
9460 Perl_pmflag(pTHX_ U32* pmfl, int ch)
9465 *pmfl |= PMf_GLOBAL;
9467 *pmfl |= PMf_CONTINUE;
9471 *pmfl |= PMf_MULTILINE;
9473 *pmfl |= PMf_SINGLELINE;
9475 *pmfl |= PMf_EXTENDED;
9479 S_scan_pat(pTHX_ char *start, I32 type)
9482 char *s = scan_str(start,FALSE,FALSE);
9485 char * const delimiter = skipspace(start);
9486 Perl_croak(aTHX_ *delimiter == '?'
9487 ? "Search pattern not terminated or ternary operator parsed as search pattern"
9488 : "Search pattern not terminated" );
9491 pm = (PMOP*)newPMOP(type, 0);
9492 if (PL_multi_open == '?')
9493 pm->op_pmflags |= PMf_ONCE;
9495 while (*s && strchr("iomsx", *s))
9496 pmflag(&pm->op_pmflags,*s++);
9499 while (*s && strchr("iogcmsx", *s))
9500 pmflag(&pm->op_pmflags,*s++);
9502 /* issue a warning if /c is specified,but /g is not */
9503 if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL)
9504 && ckWARN(WARN_REGEXP))
9506 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless without /g" );
9509 pm->op_pmpermflags = pm->op_pmflags;
9511 PL_lex_op = (OP*)pm;
9512 yylval.ival = OP_MATCH;
9517 S_scan_subst(pTHX_ char *start)
9525 yylval.ival = OP_NULL;
9527 s = scan_str(start,FALSE,FALSE);
9530 Perl_croak(aTHX_ "Substitution pattern not terminated");
9532 if (s[-1] == PL_multi_open)
9535 first_start = PL_multi_start;
9536 s = scan_str(s,FALSE,FALSE);
9539 SvREFCNT_dec(PL_lex_stuff);
9540 PL_lex_stuff = Nullsv;
9542 Perl_croak(aTHX_ "Substitution replacement not terminated");
9544 PL_multi_start = first_start; /* so whole substitution is taken together */
9546 pm = (PMOP*)newPMOP(OP_SUBST, 0);
9552 else if (strchr("iogcmsx", *s))
9553 pmflag(&pm->op_pmflags,*s++);
9558 if ((pm->op_pmflags & PMf_CONTINUE) && ckWARN(WARN_REGEXP)) {
9559 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless in s///" );
9564 PL_sublex_info.super_bufptr = s;
9565 PL_sublex_info.super_bufend = PL_bufend;
9567 pm->op_pmflags |= PMf_EVAL;
9568 repl = newSVpvn("",0);
9570 sv_catpv(repl, es ? "eval " : "do ");
9571 sv_catpvn(repl, "{ ", 2);
9572 sv_catsv(repl, PL_lex_repl);
9573 sv_catpvn(repl, " };", 2);
9575 SvREFCNT_dec(PL_lex_repl);
9579 pm->op_pmpermflags = pm->op_pmflags;
9580 PL_lex_op = (OP*)pm;
9581 yylval.ival = OP_SUBST;
9586 S_scan_trans(pTHX_ char *start)
9595 yylval.ival = OP_NULL;
9597 s = scan_str(start,FALSE,FALSE);
9599 Perl_croak(aTHX_ "Transliteration pattern not terminated");
9600 if (s[-1] == PL_multi_open)
9603 s = scan_str(s,FALSE,FALSE);
9606 SvREFCNT_dec(PL_lex_stuff);
9607 PL_lex_stuff = Nullsv;
9609 Perl_croak(aTHX_ "Transliteration replacement not terminated");
9612 complement = del = squash = 0;
9616 complement = OPpTRANS_COMPLEMENT;
9619 del = OPpTRANS_DELETE;
9622 squash = OPpTRANS_SQUASH;
9631 Newx(tbl, complement&&!del?258:256, short);
9632 o = newPVOP(OP_TRANS, 0, (char*)tbl);
9633 o->op_private &= ~OPpTRANS_ALL;
9634 o->op_private |= del|squash|complement|
9635 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
9636 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
9639 yylval.ival = OP_TRANS;
9644 S_scan_heredoc(pTHX_ register char *s)
9647 I32 op_type = OP_SCALAR;
9651 const char newline[] = "\n";
9652 const char *found_newline;
9656 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
9660 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
9663 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
9664 if (*peek == '`' || *peek == '\'' || *peek =='"') {
9667 s = delimcpy(d, e, s, PL_bufend, term, &len);
9677 if (!isALNUM_lazy_if(s,UTF))
9678 deprecate_old("bare << to mean <<\"\"");
9679 for (; isALNUM_lazy_if(s,UTF); s++) {
9684 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
9685 Perl_croak(aTHX_ "Delimiter for here document is too long");
9688 len = d - PL_tokenbuf;
9689 #ifndef PERL_STRICT_CR
9690 d = strchr(s, '\r');
9692 char * const olds = s;
9694 while (s < PL_bufend) {
9700 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
9709 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
9713 if ( outer || !(found_newline = ninstr(s,PL_bufend,newline,newline+1)) ) {
9714 herewas = newSVpvn(s,PL_bufend-s);
9718 herewas = newSVpvn(s,found_newline-s);
9720 s += SvCUR(herewas);
9722 tmpstr = NEWSV(87,79);
9723 sv_upgrade(tmpstr, SVt_PVIV);
9726 SvIV_set(tmpstr, -1);
9728 else if (term == '`') {
9729 op_type = OP_BACKTICK;
9730 SvIV_set(tmpstr, '\\');
9734 PL_multi_start = CopLINE(PL_curcop);
9735 PL_multi_open = PL_multi_close = '<';
9736 term = *PL_tokenbuf;
9737 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
9738 char *bufptr = PL_sublex_info.super_bufptr;
9739 char *bufend = PL_sublex_info.super_bufend;
9740 char * const olds = s - SvCUR(herewas);
9741 s = strchr(bufptr, '\n');
9745 while (s < bufend &&
9746 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
9748 CopLINE_inc(PL_curcop);
9751 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9752 missingterm(PL_tokenbuf);
9754 sv_setpvn(herewas,bufptr,d-bufptr+1);
9755 sv_setpvn(tmpstr,d+1,s-d);
9757 sv_catpvn(herewas,s,bufend-s);
9758 Copy(SvPVX_const(herewas),bufptr,SvCUR(herewas) + 1,char);
9765 while (s < PL_bufend &&
9766 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
9768 CopLINE_inc(PL_curcop);
9770 if (s >= PL_bufend) {
9771 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9772 missingterm(PL_tokenbuf);
9774 sv_setpvn(tmpstr,d+1,s-d);
9776 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
9778 sv_catpvn(herewas,s,PL_bufend-s);
9779 sv_setsv(PL_linestr,herewas);
9780 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
9781 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9782 PL_last_lop = PL_last_uni = Nullch;
9785 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
9786 while (s >= PL_bufend) { /* multiple line string? */
9788 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
9789 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9790 missingterm(PL_tokenbuf);
9792 CopLINE_inc(PL_curcop);
9793 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9794 PL_last_lop = PL_last_uni = Nullch;
9795 #ifndef PERL_STRICT_CR
9796 if (PL_bufend - PL_linestart >= 2) {
9797 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
9798 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
9800 PL_bufend[-2] = '\n';
9802 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
9804 else if (PL_bufend[-1] == '\r')
9805 PL_bufend[-1] = '\n';
9807 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
9808 PL_bufend[-1] = '\n';
9810 if (PERLDB_LINE && PL_curstash != PL_debstash) {
9811 SV *sv = NEWSV(88,0);
9813 sv_upgrade(sv, SVt_PVMG);
9814 sv_setsv(sv,PL_linestr);
9817 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
9819 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
9820 STRLEN off = PL_bufend - 1 - SvPVX_const(PL_linestr);
9821 *(SvPVX(PL_linestr) + off ) = ' ';
9822 sv_catsv(PL_linestr,herewas);
9823 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9824 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
9828 sv_catsv(tmpstr,PL_linestr);
9833 PL_multi_end = CopLINE(PL_curcop);
9834 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
9835 SvPV_shrink_to_cur(tmpstr);
9837 SvREFCNT_dec(herewas);
9839 if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
9841 else if (PL_encoding)
9842 sv_recode_to_utf8(tmpstr, PL_encoding);
9844 PL_lex_stuff = tmpstr;
9845 yylval.ival = op_type;
9850 takes: current position in input buffer
9851 returns: new position in input buffer
9852 side-effects: yylval and lex_op are set.
9857 <FH> read from filehandle
9858 <pkg::FH> read from package qualified filehandle
9859 <pkg'FH> read from package qualified filehandle
9860 <$fh> read from filehandle in $fh
9866 S_scan_inputsymbol(pTHX_ char *start)
9868 register char *s = start; /* current position in buffer */
9874 d = PL_tokenbuf; /* start of temp holding space */
9875 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
9876 end = strchr(s, '\n');
9879 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
9881 /* die if we didn't have space for the contents of the <>,
9882 or if it didn't end, or if we see a newline
9885 if (len >= sizeof PL_tokenbuf)
9886 Perl_croak(aTHX_ "Excessively long <> operator");
9888 Perl_croak(aTHX_ "Unterminated <> operator");
9893 Remember, only scalar variables are interpreted as filehandles by
9894 this code. Anything more complex (e.g., <$fh{$num}>) will be
9895 treated as a glob() call.
9896 This code makes use of the fact that except for the $ at the front,
9897 a scalar variable and a filehandle look the same.
9899 if (*d == '$' && d[1]) d++;
9901 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
9902 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
9905 /* If we've tried to read what we allow filehandles to look like, and
9906 there's still text left, then it must be a glob() and not a getline.
9907 Use scan_str to pull out the stuff between the <> and treat it
9908 as nothing more than a string.
9911 if (d - PL_tokenbuf != len) {
9912 yylval.ival = OP_GLOB;
9914 s = scan_str(start,FALSE,FALSE);
9916 Perl_croak(aTHX_ "Glob not terminated");
9920 bool readline_overriden = FALSE;
9921 GV *gv_readline = Nullgv;
9923 /* we're in a filehandle read situation */
9926 /* turn <> into <ARGV> */
9928 Copy("ARGV",d,5,char);
9930 /* Check whether readline() is overriden */
9931 if (((gv_readline = gv_fetchpv("readline", FALSE, SVt_PVCV))
9932 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
9934 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
9935 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
9936 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
9937 readline_overriden = TRUE;
9939 /* if <$fh>, create the ops to turn the variable into a
9945 /* try to find it in the pad for this block, otherwise find
9946 add symbol table ops
9948 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
9949 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
9950 HV *stash = PAD_COMPNAME_OURSTASH(tmp);
9951 HEK *stashname = HvNAME_HEK(stash);
9952 SV *sym = sv_2mortal(newSVhek(stashname));
9953 sv_catpvn(sym, "::", 2);
9959 OP *o = newOP(OP_PADSV, 0);
9961 PL_lex_op = readline_overriden
9962 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
9963 append_elem(OP_LIST, o,
9964 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
9965 : (OP*)newUNOP(OP_READLINE, 0, o);
9974 ? (GV_ADDMULTI | GV_ADDINEVAL)
9977 PL_lex_op = readline_overriden
9978 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
9979 append_elem(OP_LIST,
9980 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
9981 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
9982 : (OP*)newUNOP(OP_READLINE, 0,
9983 newUNOP(OP_RV2SV, 0,
9984 newGVOP(OP_GV, 0, gv)));
9986 if (!readline_overriden)
9987 PL_lex_op->op_flags |= OPf_SPECIAL;
9988 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
9989 yylval.ival = OP_NULL;
9992 /* If it's none of the above, it must be a literal filehandle
9993 (<Foo::BAR> or <FOO>) so build a simple readline OP */
9995 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
9996 PL_lex_op = readline_overriden
9997 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
9998 append_elem(OP_LIST,
9999 newGVOP(OP_GV, 0, gv),
10000 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
10001 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
10002 yylval.ival = OP_NULL;
10011 takes: start position in buffer
10012 keep_quoted preserve \ on the embedded delimiter(s)
10013 keep_delims preserve the delimiters around the string
10014 returns: position to continue reading from buffer
10015 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
10016 updates the read buffer.
10018 This subroutine pulls a string out of the input. It is called for:
10019 q single quotes q(literal text)
10020 ' single quotes 'literal text'
10021 qq double quotes qq(interpolate $here please)
10022 " double quotes "interpolate $here please"
10023 qx backticks qx(/bin/ls -l)
10024 ` backticks `/bin/ls -l`
10025 qw quote words @EXPORT_OK = qw( func() $spam )
10026 m// regexp match m/this/
10027 s/// regexp substitute s/this/that/
10028 tr/// string transliterate tr/this/that/
10029 y/// string transliterate y/this/that/
10030 ($*@) sub prototypes sub foo ($)
10031 (stuff) sub attr parameters sub foo : attr(stuff)
10032 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
10034 In most of these cases (all but <>, patterns and transliterate)
10035 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
10036 calls scan_str(). s/// makes yylex() call scan_subst() which calls
10037 scan_str(). tr/// and y/// make yylex() call scan_trans() which
10040 It skips whitespace before the string starts, and treats the first
10041 character as the delimiter. If the delimiter is one of ([{< then
10042 the corresponding "close" character )]}> is used as the closing
10043 delimiter. It allows quoting of delimiters, and if the string has
10044 balanced delimiters ([{<>}]) it allows nesting.
10046 On success, the SV with the resulting string is put into lex_stuff or,
10047 if that is already non-NULL, into lex_repl. The second case occurs only
10048 when parsing the RHS of the special constructs s/// and tr/// (y///).
10049 For convenience, the terminating delimiter character is stuffed into
10054 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
10056 SV *sv; /* scalar value: string */
10057 char *tmps; /* temp string, used for delimiter matching */
10058 register char *s = start; /* current position in the buffer */
10059 register char term; /* terminating character */
10060 register char *to; /* current position in the sv's data */
10061 I32 brackets = 1; /* bracket nesting level */
10062 bool has_utf8 = FALSE; /* is there any utf8 content? */
10063 I32 termcode; /* terminating char. code */
10064 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
10065 STRLEN termlen; /* length of terminating string */
10066 char *last = NULL; /* last position for nesting bracket */
10068 /* skip space before the delimiter */
10072 /* mark where we are, in case we need to report errors */
10075 /* after skipping whitespace, the next character is the terminator */
10078 termcode = termstr[0] = term;
10082 termcode = utf8_to_uvchr((U8*)s, &termlen);
10083 Copy(s, termstr, termlen, U8);
10084 if (!UTF8_IS_INVARIANT(term))
10088 /* mark where we are */
10089 PL_multi_start = CopLINE(PL_curcop);
10090 PL_multi_open = term;
10092 /* find corresponding closing delimiter */
10093 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
10094 termcode = termstr[0] = term = tmps[5];
10096 PL_multi_close = term;
10098 /* create a new SV to hold the contents. 87 is leak category, I'm
10099 assuming. 79 is the SV's initial length. What a random number. */
10101 sv_upgrade(sv, SVt_PVIV);
10102 SvIV_set(sv, termcode);
10103 (void)SvPOK_only(sv); /* validate pointer */
10105 /* move past delimiter and try to read a complete string */
10107 sv_catpvn(sv, s, termlen);
10110 if (PL_encoding && !UTF) {
10114 int offset = s - SvPVX_const(PL_linestr);
10115 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
10116 &offset, (char*)termstr, termlen);
10117 const char *ns = SvPVX_const(PL_linestr) + offset;
10118 char *svlast = SvEND(sv) - 1;
10120 for (; s < ns; s++) {
10121 if (*s == '\n' && !PL_rsfp)
10122 CopLINE_inc(PL_curcop);
10125 goto read_more_line;
10127 /* handle quoted delimiters */
10128 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
10130 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
10132 if ((svlast-1 - t) % 2) {
10133 if (!keep_quoted) {
10134 *(svlast-1) = term;
10136 SvCUR_set(sv, SvCUR(sv) - 1);
10141 if (PL_multi_open == PL_multi_close) {
10149 for (t = w = last; t < svlast; w++, t++) {
10150 /* At here, all closes are "was quoted" one,
10151 so we don't check PL_multi_close. */
10153 if (!keep_quoted && *(t+1) == PL_multi_open)
10158 else if (*t == PL_multi_open)
10166 SvCUR_set(sv, w - SvPVX_const(sv));
10169 if (--brackets <= 0)
10174 if (!keep_delims) {
10175 SvCUR_set(sv, SvCUR(sv) - 1);
10181 /* extend sv if need be */
10182 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
10183 /* set 'to' to the next character in the sv's string */
10184 to = SvPVX(sv)+SvCUR(sv);
10186 /* if open delimiter is the close delimiter read unbridle */
10187 if (PL_multi_open == PL_multi_close) {
10188 for (; s < PL_bufend; s++,to++) {
10189 /* embedded newlines increment the current line number */
10190 if (*s == '\n' && !PL_rsfp)
10191 CopLINE_inc(PL_curcop);
10192 /* handle quoted delimiters */
10193 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
10194 if (!keep_quoted && s[1] == term)
10196 /* any other quotes are simply copied straight through */
10200 /* terminate when run out of buffer (the for() condition), or
10201 have found the terminator */
10202 else if (*s == term) {
10205 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
10208 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
10214 /* if the terminator isn't the same as the start character (e.g.,
10215 matched brackets), we have to allow more in the quoting, and
10216 be prepared for nested brackets.
10219 /* read until we run out of string, or we find the terminator */
10220 for (; s < PL_bufend; s++,to++) {
10221 /* embedded newlines increment the line count */
10222 if (*s == '\n' && !PL_rsfp)
10223 CopLINE_inc(PL_curcop);
10224 /* backslashes can escape the open or closing characters */
10225 if (*s == '\\' && s+1 < PL_bufend) {
10226 if (!keep_quoted &&
10227 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
10232 /* allow nested opens and closes */
10233 else if (*s == PL_multi_close && --brackets <= 0)
10235 else if (*s == PL_multi_open)
10237 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
10242 /* terminate the copied string and update the sv's end-of-string */
10244 SvCUR_set(sv, to - SvPVX_const(sv));
10247 * this next chunk reads more into the buffer if we're not done yet
10251 break; /* handle case where we are done yet :-) */
10253 #ifndef PERL_STRICT_CR
10254 if (to - SvPVX_const(sv) >= 2) {
10255 if ((to[-2] == '\r' && to[-1] == '\n') ||
10256 (to[-2] == '\n' && to[-1] == '\r'))
10260 SvCUR_set(sv, to - SvPVX_const(sv));
10262 else if (to[-1] == '\r')
10265 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
10270 /* if we're out of file, or a read fails, bail and reset the current
10271 line marker so we can report where the unterminated string began
10274 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
10276 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
10279 /* we read a line, so increment our line counter */
10280 CopLINE_inc(PL_curcop);
10282 /* update debugger info */
10283 if (PERLDB_LINE && PL_curstash != PL_debstash) {
10284 SV *sv = NEWSV(88,0);
10286 sv_upgrade(sv, SVt_PVMG);
10287 sv_setsv(sv,PL_linestr);
10288 (void)SvIOK_on(sv);
10290 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
10293 /* having changed the buffer, we must update PL_bufend */
10294 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
10295 PL_last_lop = PL_last_uni = Nullch;
10298 /* at this point, we have successfully read the delimited string */
10300 if (!PL_encoding || UTF) {
10302 sv_catpvn(sv, s, termlen);
10305 if (has_utf8 || PL_encoding)
10308 PL_multi_end = CopLINE(PL_curcop);
10310 /* if we allocated too much space, give some back */
10311 if (SvCUR(sv) + 5 < SvLEN(sv)) {
10312 SvLEN_set(sv, SvCUR(sv) + 1);
10313 SvPV_renew(sv, SvLEN(sv));
10316 /* decide whether this is the first or second quoted string we've read
10329 takes: pointer to position in buffer
10330 returns: pointer to new position in buffer
10331 side-effects: builds ops for the constant in yylval.op
10333 Read a number in any of the formats that Perl accepts:
10335 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
10336 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
10339 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
10341 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
10344 If it reads a number without a decimal point or an exponent, it will
10345 try converting the number to an integer and see if it can do so
10346 without loss of precision.
10350 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
10352 register const char *s = start; /* current position in buffer */
10353 register char *d; /* destination in temp buffer */
10354 register char *e; /* end of temp buffer */
10355 NV nv; /* number read, as a double */
10356 SV *sv = Nullsv; /* place to put the converted number */
10357 bool floatit; /* boolean: int or float? */
10358 const char *lastub = 0; /* position of last underbar */
10359 static char const number_too_long[] = "Number too long";
10361 /* We use the first character to decide what type of number this is */
10365 Perl_croak(aTHX_ "panic: scan_num");
10367 /* if it starts with a 0, it could be an octal number, a decimal in
10368 0.13 disguise, or a hexadecimal number, or a binary number. */
10372 u holds the "number so far"
10373 shift the power of 2 of the base
10374 (hex == 4, octal == 3, binary == 1)
10375 overflowed was the number more than we can hold?
10377 Shift is used when we add a digit. It also serves as an "are
10378 we in octal/hex/binary?" indicator to disallow hex characters
10379 when in octal mode.
10384 bool overflowed = FALSE;
10385 bool just_zero = TRUE; /* just plain 0 or binary number? */
10386 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
10387 static const char* const bases[5] =
10388 { "", "binary", "", "octal", "hexadecimal" };
10389 static const char* const Bases[5] =
10390 { "", "Binary", "", "Octal", "Hexadecimal" };
10391 static const char* const maxima[5] =
10393 "0b11111111111111111111111111111111",
10397 const char *base, *Base, *max;
10399 /* check for hex */
10404 } else if (s[1] == 'b') {
10409 /* check for a decimal in disguise */
10410 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
10412 /* so it must be octal */
10419 if (ckWARN(WARN_SYNTAX))
10420 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10421 "Misplaced _ in number");
10425 base = bases[shift];
10426 Base = Bases[shift];
10427 max = maxima[shift];
10429 /* read the rest of the number */
10431 /* x is used in the overflow test,
10432 b is the digit we're adding on. */
10437 /* if we don't mention it, we're done */
10441 /* _ are ignored -- but warned about if consecutive */
10443 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10444 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10445 "Misplaced _ in number");
10449 /* 8 and 9 are not octal */
10450 case '8': case '9':
10452 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
10456 case '2': case '3': case '4':
10457 case '5': case '6': case '7':
10459 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
10462 case '0': case '1':
10463 b = *s++ & 15; /* ASCII digit -> value of digit */
10467 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
10468 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
10469 /* make sure they said 0x */
10472 b = (*s++ & 7) + 9;
10474 /* Prepare to put the digit we have onto the end
10475 of the number so far. We check for overflows.
10481 x = u << shift; /* make room for the digit */
10483 if ((x >> shift) != u
10484 && !(PL_hints & HINT_NEW_BINARY)) {
10487 if (ckWARN_d(WARN_OVERFLOW))
10488 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
10489 "Integer overflow in %s number",
10492 u = x | b; /* add the digit to the end */
10495 n *= nvshift[shift];
10496 /* If an NV has not enough bits in its
10497 * mantissa to represent an UV this summing of
10498 * small low-order numbers is a waste of time
10499 * (because the NV cannot preserve the
10500 * low-order bits anyway): we could just
10501 * remember when did we overflow and in the
10502 * end just multiply n by the right
10510 /* if we get here, we had success: make a scalar value from
10515 /* final misplaced underbar check */
10516 if (s[-1] == '_') {
10517 if (ckWARN(WARN_SYNTAX))
10518 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
10523 if (n > 4294967295.0 && ckWARN(WARN_PORTABLE))
10524 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
10525 "%s number > %s non-portable",
10531 if (u > 0xffffffff && ckWARN(WARN_PORTABLE))
10532 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
10533 "%s number > %s non-portable",
10538 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
10539 sv = new_constant(start, s - start, "integer",
10541 else if (PL_hints & HINT_NEW_BINARY)
10542 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
10547 handle decimal numbers.
10548 we're also sent here when we read a 0 as the first digit
10550 case '1': case '2': case '3': case '4': case '5':
10551 case '6': case '7': case '8': case '9': case '.':
10554 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
10557 /* read next group of digits and _ and copy into d */
10558 while (isDIGIT(*s) || *s == '_') {
10559 /* skip underscores, checking for misplaced ones
10563 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10564 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10565 "Misplaced _ in number");
10569 /* check for end of fixed-length buffer */
10571 Perl_croak(aTHX_ number_too_long);
10572 /* if we're ok, copy the character */
10577 /* final misplaced underbar check */
10578 if (lastub && s == lastub + 1) {
10579 if (ckWARN(WARN_SYNTAX))
10580 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
10583 /* read a decimal portion if there is one. avoid
10584 3..5 being interpreted as the number 3. followed
10587 if (*s == '.' && s[1] != '.') {
10592 if (ckWARN(WARN_SYNTAX))
10593 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10594 "Misplaced _ in number");
10598 /* copy, ignoring underbars, until we run out of digits.
10600 for (; isDIGIT(*s) || *s == '_'; s++) {
10601 /* fixed length buffer check */
10603 Perl_croak(aTHX_ number_too_long);
10605 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10606 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10607 "Misplaced _ in number");
10613 /* fractional part ending in underbar? */
10614 if (s[-1] == '_') {
10615 if (ckWARN(WARN_SYNTAX))
10616 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10617 "Misplaced _ in number");
10619 if (*s == '.' && isDIGIT(s[1])) {
10620 /* oops, it's really a v-string, but without the "v" */
10626 /* read exponent part, if present */
10627 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
10631 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
10632 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
10634 /* stray preinitial _ */
10636 if (ckWARN(WARN_SYNTAX))
10637 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10638 "Misplaced _ in number");
10642 /* allow positive or negative exponent */
10643 if (*s == '+' || *s == '-')
10646 /* stray initial _ */
10648 if (ckWARN(WARN_SYNTAX))
10649 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10650 "Misplaced _ in number");
10654 /* read digits of exponent */
10655 while (isDIGIT(*s) || *s == '_') {
10658 Perl_croak(aTHX_ number_too_long);
10662 if (((lastub && s == lastub + 1) ||
10663 (!isDIGIT(s[1]) && s[1] != '_'))
10664 && ckWARN(WARN_SYNTAX))
10665 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10666 "Misplaced _ in number");
10673 /* make an sv from the string */
10677 We try to do an integer conversion first if no characters
10678 indicating "float" have been found.
10683 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
10685 if (flags == IS_NUMBER_IN_UV) {
10687 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
10690 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
10691 if (uv <= (UV) IV_MIN)
10692 sv_setiv(sv, -(IV)uv);
10699 /* terminate the string */
10701 nv = Atof(PL_tokenbuf);
10705 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
10706 (PL_hints & HINT_NEW_INTEGER) )
10707 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
10708 (floatit ? "float" : "integer"),
10712 /* if it starts with a v, it could be a v-string */
10715 sv = NEWSV(92,5); /* preallocate storage space */
10716 s = scan_vstring(s,sv);
10720 /* make the op for the constant and return */
10723 lvalp->opval = newSVOP(OP_CONST, 0, sv);
10725 lvalp->opval = Nullop;
10731 S_scan_formline(pTHX_ register char *s)
10733 register char *eol;
10735 SV *stuff = newSVpvn("",0);
10736 bool needargs = FALSE;
10737 bool eofmt = FALSE;
10739 while (!needargs) {
10741 #ifdef PERL_STRICT_CR
10742 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
10744 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
10746 if (*t == '\n' || t == PL_bufend) {
10751 if (PL_in_eval && !PL_rsfp) {
10752 eol = (char *) memchr(s,'\n',PL_bufend-s);
10757 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
10759 for (t = s; t < eol; t++) {
10760 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
10762 goto enough; /* ~~ must be first line in formline */
10764 if (*t == '@' || *t == '^')
10768 sv_catpvn(stuff, s, eol-s);
10769 #ifndef PERL_STRICT_CR
10770 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
10771 char *end = SvPVX(stuff) + SvCUR(stuff);
10774 SvCUR_set(stuff, SvCUR(stuff) - 1);
10783 s = filter_gets(PL_linestr, PL_rsfp, 0);
10784 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
10785 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
10786 PL_last_lop = PL_last_uni = Nullch;
10795 if (SvCUR(stuff)) {
10798 PL_lex_state = LEX_NORMAL;
10799 PL_nextval[PL_nexttoke].ival = 0;
10803 PL_lex_state = LEX_FORMLINE;
10805 if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
10807 else if (PL_encoding)
10808 sv_recode_to_utf8(stuff, PL_encoding);
10810 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
10812 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
10816 SvREFCNT_dec(stuff);
10818 PL_lex_formbrack = 0;
10829 PL_cshlen = strlen(PL_cshname);
10834 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
10836 const I32 oldsavestack_ix = PL_savestack_ix;
10837 CV* outsidecv = PL_compcv;
10840 assert(SvTYPE(PL_compcv) == SVt_PVCV);
10842 SAVEI32(PL_subline);
10843 save_item(PL_subname);
10844 SAVESPTR(PL_compcv);
10846 PL_compcv = (CV*)NEWSV(1104,0);
10847 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
10848 CvFLAGS(PL_compcv) |= flags;
10850 PL_subline = CopLINE(PL_curcop);
10851 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
10852 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
10853 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
10855 return oldsavestack_ix;
10859 #pragma segment Perl_yylex
10862 Perl_yywarn(pTHX_ const char *s)
10864 PL_in_eval |= EVAL_WARNONLY;
10866 PL_in_eval &= ~EVAL_WARNONLY;
10871 Perl_yyerror(pTHX_ const char *s)
10873 const char *where = NULL;
10874 const char *context = NULL;
10878 if (!yychar || (yychar == ';' && !PL_rsfp))
10880 else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
10881 PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
10882 PL_oldbufptr != PL_bufptr) {
10885 The code below is removed for NetWare because it abends/crashes on NetWare
10886 when the script has error such as not having the closing quotes like:
10887 if ($var eq "value)
10888 Checking of white spaces is anyway done in NetWare code.
10891 while (isSPACE(*PL_oldoldbufptr))
10894 context = PL_oldoldbufptr;
10895 contlen = PL_bufptr - PL_oldoldbufptr;
10897 else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
10898 PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
10901 The code below is removed for NetWare because it abends/crashes on NetWare
10902 when the script has error such as not having the closing quotes like:
10903 if ($var eq "value)
10904 Checking of white spaces is anyway done in NetWare code.
10907 while (isSPACE(*PL_oldbufptr))
10910 context = PL_oldbufptr;
10911 contlen = PL_bufptr - PL_oldbufptr;
10913 else if (yychar > 255)
10914 where = "next token ???";
10915 else if (yychar == -2) { /* YYEMPTY */
10916 if (PL_lex_state == LEX_NORMAL ||
10917 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
10918 where = "at end of line";
10919 else if (PL_lex_inpat)
10920 where = "within pattern";
10922 where = "within string";
10925 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
10927 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
10928 else if (isPRINT_LC(yychar))
10929 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
10931 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
10932 where = SvPVX_const(where_sv);
10934 msg = sv_2mortal(newSVpv(s, 0));
10935 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
10936 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
10938 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
10940 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
10941 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
10942 Perl_sv_catpvf(aTHX_ msg,
10943 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
10944 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
10947 if (PL_in_eval & EVAL_WARNONLY && ckWARN_d(WARN_SYNTAX))
10948 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, msg);
10951 if (PL_error_count >= 10) {
10952 if (PL_in_eval && SvCUR(ERRSV))
10953 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
10954 ERRSV, OutCopFILE(PL_curcop));
10956 Perl_croak(aTHX_ "%s has too many errors.\n",
10957 OutCopFILE(PL_curcop));
10960 PL_in_my_stash = NULL;
10964 #pragma segment Main
10968 S_swallow_bom(pTHX_ U8 *s)
10970 const STRLEN slen = SvCUR(PL_linestr);
10973 if (s[1] == 0xFE) {
10974 /* UTF-16 little-endian? (or UTF32-LE?) */
10975 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
10976 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
10977 #ifndef PERL_NO_UTF16_FILTER
10978 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
10981 if (PL_bufend > (char*)s) {
10985 filter_add(utf16rev_textfilter, NULL);
10986 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
10987 utf16_to_utf8_reversed(s, news,
10988 PL_bufend - (char*)s - 1,
10990 sv_setpvn(PL_linestr, (const char*)news, newlen);
10992 SvUTF8_on(PL_linestr);
10993 s = (U8*)SvPVX(PL_linestr);
10994 PL_bufend = SvPVX(PL_linestr) + newlen;
10997 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
11002 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
11003 #ifndef PERL_NO_UTF16_FILTER
11004 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
11007 if (PL_bufend > (char *)s) {
11011 filter_add(utf16_textfilter, NULL);
11012 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
11013 utf16_to_utf8(s, news,
11014 PL_bufend - (char*)s,
11016 sv_setpvn(PL_linestr, (const char*)news, newlen);
11018 SvUTF8_on(PL_linestr);
11019 s = (U8*)SvPVX(PL_linestr);
11020 PL_bufend = SvPVX(PL_linestr) + newlen;
11023 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
11028 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
11029 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
11030 s += 3; /* UTF-8 */
11036 if (s[2] == 0xFE && s[3] == 0xFF) {
11037 /* UTF-32 big-endian */
11038 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
11041 else if (s[2] == 0 && s[3] != 0) {
11044 * are a good indicator of UTF-16BE. */
11045 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
11050 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
11053 * are a good indicator of UTF-16LE. */
11054 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
11063 * Restore a source filter.
11067 restore_rsfp(pTHX_ void *f)
11069 PerlIO * const fp = (PerlIO*)f;
11071 if (PL_rsfp == PerlIO_stdin())
11072 PerlIO_clearerr(PL_rsfp);
11073 else if (PL_rsfp && (PL_rsfp != fp))
11074 PerlIO_close(PL_rsfp);
11078 #ifndef PERL_NO_UTF16_FILTER
11080 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
11082 const STRLEN old = SvCUR(sv);
11083 const I32 count = FILTER_READ(idx+1, sv, maxlen);
11084 DEBUG_P(PerlIO_printf(Perl_debug_log,
11085 "utf16_textfilter(%p): %d %d (%d)\n",
11086 utf16_textfilter, idx, maxlen, (int) count));
11090 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
11091 Copy(SvPVX_const(sv), tmps, old, char);
11092 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
11093 SvCUR(sv) - old, &newlen);
11094 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
11096 DEBUG_P({sv_dump(sv);});
11101 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
11103 const STRLEN old = SvCUR(sv);
11104 const I32 count = FILTER_READ(idx+1, sv, maxlen);
11105 DEBUG_P(PerlIO_printf(Perl_debug_log,
11106 "utf16rev_textfilter(%p): %d %d (%d)\n",
11107 utf16rev_textfilter, idx, maxlen, (int) count));
11111 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
11112 Copy(SvPVX_const(sv), tmps, old, char);
11113 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
11114 SvCUR(sv) - old, &newlen);
11115 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
11117 DEBUG_P({ sv_dump(sv); });
11123 Returns a pointer to the next character after the parsed
11124 vstring, as well as updating the passed in sv.
11126 Function must be called like
11129 s = scan_vstring(s,sv);
11131 The sv should already be large enough to store the vstring
11132 passed in, for performance reasons.
11137 Perl_scan_vstring(pTHX_ const char *s, SV *sv)
11139 const char *pos = s;
11140 const char *start = s;
11141 if (*pos == 'v') pos++; /* get past 'v' */
11142 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
11144 if ( *pos != '.') {
11145 /* this may not be a v-string if followed by => */
11146 const char *next = pos;
11147 while (next < PL_bufend && isSPACE(*next))
11149 if ((PL_bufend - next) >= 2 && *next == '=' && next[1] == '>' ) {
11150 /* return string not v-string */
11151 sv_setpvn(sv,(char *)s,pos-s);
11152 return (char *)pos;
11156 if (!isALPHA(*pos)) {
11157 U8 tmpbuf[UTF8_MAXBYTES+1];
11159 if (*s == 'v') s++; /* get past 'v' */
11161 sv_setpvn(sv, "", 0);
11167 /* this is atoi() that tolerates underscores */
11168 const char *end = pos;
11170 while (--end >= s) {
11175 rev += (*end - '0') * mult;
11177 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
11178 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
11179 "Integer overflow in decimal number");
11183 if (rev > 0x7FFFFFFF)
11184 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
11186 /* Append native character for the rev point */
11187 tmpend = uvchr_to_utf8(tmpbuf, rev);
11188 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
11189 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
11191 if (pos + 1 < PL_bufend && *pos == '.' && isDIGIT(pos[1]))
11197 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
11201 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
11209 * c-indentation-style: bsd
11210 * c-basic-offset: 4
11211 * indent-tabs-mode: t
11214 * ex: set ts=8 sts=4 sw=4 noet: