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) {
4089 assert (orig_keyword == 0);
4097 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4099 /* Some keywords can be followed by any delimiter, including ':' */
4100 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
4101 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
4102 (PL_tokenbuf[0] == 'q' &&
4103 strchr("qwxr", PL_tokenbuf[1])))));
4105 /* x::* is just a word, unless x is "CORE" */
4106 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
4110 while (d < PL_bufend && isSPACE(*d))
4111 d++; /* no comments skipped here, or s### is misparsed */
4113 /* Is this a label? */
4114 if (!tmp && PL_expect == XSTATE
4115 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
4117 yylval.pval = savepv(PL_tokenbuf);
4122 /* Check for keywords */
4123 tmp = keyword(PL_tokenbuf, len);
4125 /* Is this a word before a => operator? */
4126 if (*d == '=' && d[1] == '>') {
4129 = (OP*)newSVOP(OP_CONST, 0,
4130 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
4131 yylval.opval->op_private = OPpCONST_BARE;
4135 if (tmp < 0) { /* second-class keyword? */
4136 GV *ogv = Nullgv; /* override (winner) */
4137 GV *hgv = Nullgv; /* hidden (loser) */
4138 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
4140 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
4143 if (GvIMPORTED_CV(gv))
4145 else if (! CvMETHOD(cv))
4149 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
4150 (gv = *gvp) != (GV*)&PL_sv_undef &&
4151 GvCVu(gv) && GvIMPORTED_CV(gv))
4158 tmp = 0; /* overridden by import or by GLOBAL */
4161 && -tmp==KEY_lock /* XXX generalizable kludge */
4163 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
4165 tmp = 0; /* any sub overrides "weak" keyword */
4170 && PL_expect != XOPERATOR
4171 && PL_expect != XTERMORDORDOR)
4173 /* any sub overrides the "err" keyword, except when really an
4174 * operator is expected */
4177 else { /* no override */
4179 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
4180 Perl_warner(aTHX_ packWARN(WARN_MISC),
4181 "dump() better written as CORE::dump()");
4185 if (hgv && tmp != KEY_x && tmp != KEY_CORE
4186 && ckWARN(WARN_AMBIGUOUS)) /* never ambiguous */
4187 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4188 "Ambiguous call resolved as CORE::%s(), %s",
4189 GvENAME(hgv), "qualify as such or use &");
4196 default: /* not a keyword */
4200 const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
4202 /* Get the rest if it looks like a package qualifier */
4204 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
4206 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
4209 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
4210 *s == '\'' ? "'" : "::");
4215 if (PL_expect == XOPERATOR) {
4216 if (PL_bufptr == PL_linestart) {
4217 CopLINE_dec(PL_curcop);
4218 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
4219 CopLINE_inc(PL_curcop);
4222 no_op("Bareword",s);
4225 /* Look for a subroutine with this name in current package,
4226 unless name is "Foo::", in which case Foo is a bearword
4227 (and a package name). */
4230 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
4232 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
4233 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
4234 "Bareword \"%s\" refers to nonexistent package",
4237 PL_tokenbuf[len] = '\0';
4244 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
4247 /* if we saw a global override before, get the right name */
4250 sv = newSVpvn("CORE::GLOBAL::",14);
4251 sv_catpv(sv,PL_tokenbuf);
4254 /* If len is 0, newSVpv does strlen(), which is correct.
4255 If len is non-zero, then it will be the true length,
4256 and so the scalar will be created correctly. */
4257 sv = newSVpv(PL_tokenbuf,len);
4260 /* Presume this is going to be a bareword of some sort. */
4263 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
4264 yylval.opval->op_private = OPpCONST_BARE;
4265 /* UTF-8 package name? */
4266 if (UTF && !IN_BYTES &&
4267 is_utf8_string((U8*)SvPVX_const(sv), SvCUR(sv)))
4270 /* And if "Foo::", then that's what it certainly is. */
4275 /* See if it's the indirect object for a list operator. */
4277 if (PL_oldoldbufptr &&
4278 PL_oldoldbufptr < PL_bufptr &&
4279 (PL_oldoldbufptr == PL_last_lop
4280 || PL_oldoldbufptr == PL_last_uni) &&
4281 /* NO SKIPSPACE BEFORE HERE! */
4282 (PL_expect == XREF ||
4283 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
4285 bool immediate_paren = *s == '(';
4287 /* (Now we can afford to cross potential line boundary.) */
4290 /* Two barewords in a row may indicate method call. */
4292 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
4295 /* If not a declared subroutine, it's an indirect object. */
4296 /* (But it's an indir obj regardless for sort.) */
4297 /* Also, if "_" follows a filetest operator, it's a bareword */
4300 ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
4301 ((!gv || !GvCVu(gv)) &&
4302 (PL_last_lop_op != OP_MAPSTART &&
4303 PL_last_lop_op != OP_GREPSTART))))
4304 || (PL_tokenbuf[0] == '_' && PL_tokenbuf[1] == '\0'
4305 && ((PL_opargs[PL_last_lop_op] & OA_CLASS_MASK) == OA_FILESTATOP))
4308 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
4313 PL_expect = XOPERATOR;
4316 /* Is this a word before a => operator? */
4317 if (*s == '=' && s[1] == '>' && !pkgname) {
4319 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
4320 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
4321 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
4325 /* If followed by a paren, it's certainly a subroutine. */
4328 if (gv && GvCVu(gv)) {
4329 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
4330 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
4335 PL_nextval[PL_nexttoke].opval = yylval.opval;
4336 PL_expect = XOPERATOR;
4342 /* If followed by var or block, call it a method (unless sub) */
4344 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
4345 PL_last_lop = PL_oldbufptr;
4346 PL_last_lop_op = OP_METHOD;
4350 /* If followed by a bareword, see if it looks like indir obj. */
4353 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
4354 && (tmp = intuit_method(s,gv)))
4357 /* Not a method, so call it a subroutine (if defined) */
4359 if (gv && GvCVu(gv)) {
4361 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4362 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4363 "Ambiguous use of -%s resolved as -&%s()",
4364 PL_tokenbuf, PL_tokenbuf);
4365 /* Check for a constant sub */
4367 if ((sv = cv_const_sv(cv))) {
4369 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4370 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4371 yylval.opval->op_private = 0;
4375 /* Resolve to GV now. */
4376 op_free(yylval.opval);
4377 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4378 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4379 PL_last_lop = PL_oldbufptr;
4380 PL_last_lop_op = OP_ENTERSUB;
4381 /* Is there a prototype? */
4384 const char *proto = SvPV_const((SV*)cv, len);
4387 if (*proto == '$' && proto[1] == '\0')
4389 while (*proto == ';')
4391 if (*proto == '&' && *s == '{') {
4392 sv_setpv(PL_subname, PL_curstash ?
4393 "__ANON__" : "__ANON__::__ANON__");
4397 PL_nextval[PL_nexttoke].opval = yylval.opval;
4403 /* Call it a bare word */
4405 if (PL_hints & HINT_STRICT_SUBS)
4406 yylval.opval->op_private |= OPpCONST_STRICT;
4409 if (lastchar != '-') {
4410 if (ckWARN(WARN_RESERVED)) {
4411 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4412 if (!*d && !gv_stashpv(PL_tokenbuf,FALSE))
4413 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
4420 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')
4421 && ckWARN_d(WARN_AMBIGUOUS)) {
4422 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4423 "Operator or semicolon missing before %c%s",
4424 lastchar, PL_tokenbuf);
4425 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4426 "Ambiguous use of %c resolved as operator %c",
4427 lastchar, lastchar);
4433 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4434 newSVpv(CopFILE(PL_curcop),0));
4438 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4439 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4442 case KEY___PACKAGE__:
4443 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4445 ? newSVhek(HvNAME_HEK(PL_curstash))
4452 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4453 const char *pname = "main";
4454 if (PL_tokenbuf[2] == 'D')
4455 pname = HvNAME_get(PL_curstash ? PL_curstash : PL_defstash);
4456 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4459 GvIOp(gv) = newIO();
4460 IoIFP(GvIOp(gv)) = PL_rsfp;
4461 #if defined(HAS_FCNTL) && defined(F_SETFD)
4463 const int fd = PerlIO_fileno(PL_rsfp);
4464 fcntl(fd,F_SETFD,fd >= 3);
4467 /* Mark this internal pseudo-handle as clean */
4468 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4470 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4471 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4472 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4474 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4475 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4476 /* if the script was opened in binmode, we need to revert
4477 * it to text mode for compatibility; but only iff it has CRs
4478 * XXX this is a questionable hack at best. */
4479 if (PL_bufend-PL_bufptr > 2
4480 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4483 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4484 loc = PerlIO_tell(PL_rsfp);
4485 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4488 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4490 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4491 #endif /* NETWARE */
4492 #ifdef PERLIO_IS_STDIO /* really? */
4493 # if defined(__BORLANDC__)
4494 /* XXX see note in do_binmode() */
4495 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4499 PerlIO_seek(PL_rsfp, loc, 0);
4503 #ifdef PERLIO_LAYERS
4506 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4507 else if (PL_encoding) {
4514 XPUSHs(PL_encoding);
4516 call_method("name", G_SCALAR);
4520 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
4521 Perl_form(aTHX_ ":encoding(%"SVf")",
4539 if (PL_expect == XSTATE) {
4546 if (*s == ':' && s[1] == ':') {
4549 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4550 if (!(tmp = keyword(PL_tokenbuf, len)))
4551 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4554 else if (tmp == KEY_require || tmp == KEY_do)
4555 /* that's a way to remember we saw "CORE::" */
4568 LOP(OP_ACCEPT,XTERM);
4574 LOP(OP_ATAN2,XTERM);
4580 LOP(OP_BINMODE,XTERM);
4583 LOP(OP_BLESS,XTERM);
4592 /* When 'use switch' is in effect, continue has a dual
4593 life as a control operator. */
4595 if (!FEATURE_IS_ENABLED("switch", 6))
4598 /* We have to disambiguate the two senses of
4599 "continue". If the next token is a '{' then
4600 treat it as the start of a continue block;
4601 otherwise treat it as a control operator.
4612 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4629 if (!PL_cryptseen) {
4630 PL_cryptseen = TRUE;
4634 LOP(OP_CRYPT,XTERM);
4637 LOP(OP_CHMOD,XTERM);
4640 LOP(OP_CHOWN,XTERM);
4643 LOP(OP_CONNECT,XTERM);
4662 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4663 if (orig_keyword == KEY_do) {
4672 PL_hints |= HINT_BLOCK_SCOPE;
4682 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4683 LOP(OP_DBMOPEN,XTERM);
4689 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4696 yylval.ival = CopLINE(PL_curcop);
4710 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4711 UNIBRACK(OP_ENTEREVAL);
4729 case KEY_endhostent:
4735 case KEY_endservent:
4738 case KEY_endprotoent:
4749 yylval.ival = CopLINE(PL_curcop);
4751 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4753 if ((PL_bufend - p) >= 3 &&
4754 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4756 else if ((PL_bufend - p) >= 4 &&
4757 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4760 if (isIDFIRST_lazy_if(p,UTF)) {
4761 p = scan_ident(p, PL_bufend,
4762 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4766 Perl_croak(aTHX_ "Missing $ on loop variable");
4771 LOP(OP_FORMLINE,XTERM);
4777 LOP(OP_FCNTL,XTERM);
4783 LOP(OP_FLOCK,XTERM);
4792 LOP(OP_GREPSTART, XREF);
4795 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4810 case KEY_getpriority:
4811 LOP(OP_GETPRIORITY,XTERM);
4813 case KEY_getprotobyname:
4816 case KEY_getprotobynumber:
4817 LOP(OP_GPBYNUMBER,XTERM);
4819 case KEY_getprotoent:
4831 case KEY_getpeername:
4832 UNI(OP_GETPEERNAME);
4834 case KEY_gethostbyname:
4837 case KEY_gethostbyaddr:
4838 LOP(OP_GHBYADDR,XTERM);
4840 case KEY_gethostent:
4843 case KEY_getnetbyname:
4846 case KEY_getnetbyaddr:
4847 LOP(OP_GNBYADDR,XTERM);
4852 case KEY_getservbyname:
4853 LOP(OP_GSBYNAME,XTERM);
4855 case KEY_getservbyport:
4856 LOP(OP_GSBYPORT,XTERM);
4858 case KEY_getservent:
4861 case KEY_getsockname:
4862 UNI(OP_GETSOCKNAME);
4864 case KEY_getsockopt:
4865 LOP(OP_GSOCKOPT,XTERM);
4880 yylval.ival = CopLINE(PL_curcop);
4891 yylval.ival = CopLINE(PL_curcop);
4895 LOP(OP_INDEX,XTERM);
4901 LOP(OP_IOCTL,XTERM);
4913 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4945 LOP(OP_LISTEN,XTERM);
4954 s = scan_pat(s,OP_MATCH);
4955 TERM(sublex_start());
4958 LOP(OP_MAPSTART, XREF);
4961 LOP(OP_MKDIR,XTERM);
4964 LOP(OP_MSGCTL,XTERM);
4967 LOP(OP_MSGGET,XTERM);
4970 LOP(OP_MSGRCV,XTERM);
4973 LOP(OP_MSGSND,XTERM);
4979 if (isIDFIRST_lazy_if(s,UTF)) {
4980 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4981 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4983 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4984 if (!PL_in_my_stash) {
4987 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4995 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5002 s = tokenize_use(0, s);
5006 if (*s == '(' || (s = skipspace(s), *s == '('))
5013 if (isIDFIRST_lazy_if(s,UTF)) {
5015 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
5016 for (t=d; *t && isSPACE(*t); t++) ;
5017 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
5019 && !(t[0] == '=' && t[1] == '>')
5021 int len = (int)(d-s);
5022 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
5023 "Precedence problem: open %.*s should be open(%.*s)",
5030 yylval.ival = OP_OR;
5040 LOP(OP_OPEN_DIR,XTERM);
5043 checkcomma(s,PL_tokenbuf,"filehandle");
5047 checkcomma(s,PL_tokenbuf,"filehandle");
5066 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5070 LOP(OP_PIPE_OP,XTERM);
5073 s = scan_str(s,FALSE,FALSE);
5075 missingterm((char*)0);
5076 yylval.ival = OP_CONST;
5077 TERM(sublex_start());
5083 s = scan_str(s,FALSE,FALSE);
5085 missingterm((char*)0);
5086 PL_expect = XOPERATOR;
5088 if (SvCUR(PL_lex_stuff)) {
5091 d = SvPV_force(PL_lex_stuff, len);
5094 for (; isSPACE(*d) && len; --len, ++d) ;
5097 if (!warned && ckWARN(WARN_QW)) {
5098 for (; !isSPACE(*d) && len; --len, ++d) {
5100 Perl_warner(aTHX_ packWARN(WARN_QW),
5101 "Possible attempt to separate words with commas");
5104 else if (*d == '#') {
5105 Perl_warner(aTHX_ packWARN(WARN_QW),
5106 "Possible attempt to put comments in qw() list");
5112 for (; !isSPACE(*d) && len; --len, ++d) ;
5114 sv = newSVpvn(b, d-b);
5115 if (DO_UTF8(PL_lex_stuff))
5117 words = append_elem(OP_LIST, words,
5118 newSVOP(OP_CONST, 0, tokeq(sv)));
5122 PL_nextval[PL_nexttoke].opval = words;
5127 SvREFCNT_dec(PL_lex_stuff);
5128 PL_lex_stuff = Nullsv;
5134 s = scan_str(s,FALSE,FALSE);
5136 missingterm((char*)0);
5137 yylval.ival = OP_STRINGIFY;
5138 if (SvIVX(PL_lex_stuff) == '\'')
5139 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
5140 TERM(sublex_start());
5143 s = scan_pat(s,OP_QR);
5144 TERM(sublex_start());
5147 s = scan_str(s,FALSE,FALSE);
5149 missingterm((char*)0);
5150 yylval.ival = OP_BACKTICK;
5152 TERM(sublex_start());
5160 s = force_version(s, FALSE);
5162 else if (*s != 'v' || !isDIGIT(s[1])
5163 || (s = force_version(s, TRUE), *s == 'v'))
5165 *PL_tokenbuf = '\0';
5166 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5167 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
5168 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
5170 yyerror("<> should be quotes");
5172 if (orig_keyword == KEY_require) {
5180 PL_last_uni = PL_oldbufptr;
5181 PL_last_lop_op = OP_REQUIRE;
5183 return REPORT( (int)REQUIRE );
5189 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5193 LOP(OP_RENAME,XTERM);
5202 LOP(OP_RINDEX,XTERM);
5212 UNIDOR(OP_READLINE);
5225 LOP(OP_REVERSE,XTERM);
5228 UNIDOR(OP_READLINK);
5236 TERM(sublex_start());
5238 TOKEN(1); /* force error */
5241 checkcomma(s,PL_tokenbuf,"filehandle");
5251 LOP(OP_SELECT,XTERM);
5257 LOP(OP_SEMCTL,XTERM);
5260 LOP(OP_SEMGET,XTERM);
5263 LOP(OP_SEMOP,XTERM);
5269 LOP(OP_SETPGRP,XTERM);
5271 case KEY_setpriority:
5272 LOP(OP_SETPRIORITY,XTERM);
5274 case KEY_sethostent:
5280 case KEY_setservent:
5283 case KEY_setprotoent:
5293 LOP(OP_SEEKDIR,XTERM);
5295 case KEY_setsockopt:
5296 LOP(OP_SSOCKOPT,XTERM);
5302 LOP(OP_SHMCTL,XTERM);
5305 LOP(OP_SHMGET,XTERM);
5308 LOP(OP_SHMREAD,XTERM);
5311 LOP(OP_SHMWRITE,XTERM);
5314 LOP(OP_SHUTDOWN,XTERM);
5323 LOP(OP_SOCKET,XTERM);
5325 case KEY_socketpair:
5326 LOP(OP_SOCKPAIR,XTERM);
5329 checkcomma(s,PL_tokenbuf,"subroutine name");
5331 if (*s == ';' || *s == ')') /* probably a close */
5332 Perl_croak(aTHX_ "sort is now a reserved word");
5334 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5338 LOP(OP_SPLIT,XTERM);
5341 LOP(OP_SPRINTF,XTERM);
5344 LOP(OP_SPLICE,XTERM);
5359 LOP(OP_SUBSTR,XTERM);
5365 char tmpbuf[sizeof PL_tokenbuf];
5366 SSize_t tboffset = 0;
5367 expectation attrful;
5368 bool have_name, have_proto, bad_proto;
5369 const int key = tmp;
5373 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
5374 (*s == ':' && s[1] == ':'))
5377 attrful = XATTRBLOCK;
5378 /* remember buffer pos'n for later force_word */
5379 tboffset = s - PL_oldbufptr;
5380 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
5381 if (strchr(tmpbuf, ':'))
5382 sv_setpv(PL_subname, tmpbuf);
5384 sv_setsv(PL_subname,PL_curstname);
5385 sv_catpvn(PL_subname,"::",2);
5386 sv_catpvn(PL_subname,tmpbuf,len);
5393 Perl_croak(aTHX_ "Missing name in \"my sub\"");
5394 PL_expect = XTERMBLOCK;
5395 attrful = XATTRTERM;
5396 sv_setpvn(PL_subname,"?",1);
5400 if (key == KEY_format) {
5402 PL_lex_formbrack = PL_lex_brackets + 1;
5404 (void) force_word(PL_oldbufptr + tboffset, WORD,
5409 /* Look for a prototype */
5413 s = scan_str(s,FALSE,FALSE);
5415 Perl_croak(aTHX_ "Prototype not terminated");
5416 /* strip spaces and check for bad characters */
5417 d = SvPVX(PL_lex_stuff);
5420 for (p = d; *p; ++p) {
5423 if (!strchr("$@%*;[]&\\", *p))
5428 if (bad_proto && ckWARN(WARN_SYNTAX))
5429 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5430 "Illegal character in prototype for %"SVf" : %s",
5432 SvCUR_set(PL_lex_stuff, tmp);
5440 if (*s == ':' && s[1] != ':')
5441 PL_expect = attrful;
5442 else if (*s != '{' && key == KEY_sub) {
5444 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
5446 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, PL_subname);
5450 PL_nextval[PL_nexttoke].opval =
5451 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5452 PL_lex_stuff = Nullsv;
5456 sv_setpv(PL_subname,
5457 PL_curstash ? "__ANON__" : "__ANON__::__ANON__");
5460 (void) force_word(PL_oldbufptr + tboffset, WORD,
5469 LOP(OP_SYSTEM,XREF);
5472 LOP(OP_SYMLINK,XTERM);
5475 LOP(OP_SYSCALL,XTERM);
5478 LOP(OP_SYSOPEN,XTERM);
5481 LOP(OP_SYSSEEK,XTERM);
5484 LOP(OP_SYSREAD,XTERM);
5487 LOP(OP_SYSWRITE,XTERM);
5491 TERM(sublex_start());
5512 LOP(OP_TRUNCATE,XTERM);
5524 yylval.ival = CopLINE(PL_curcop);
5528 yylval.ival = CopLINE(PL_curcop);
5532 LOP(OP_UNLINK,XTERM);
5538 LOP(OP_UNPACK,XTERM);
5541 LOP(OP_UTIME,XTERM);
5547 LOP(OP_UNSHIFT,XTERM);
5550 s = tokenize_use(1, s);
5560 yylval.ival = CopLINE(PL_curcop);
5564 yylval.ival = CopLINE(PL_curcop);
5568 PL_hints |= HINT_BLOCK_SCOPE;
5575 LOP(OP_WAITPID,XTERM);
5584 ctl_l[0] = toCTRL('L');
5586 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5589 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5594 if (PL_expect == XOPERATOR)
5600 yylval.ival = OP_XOR;
5605 TERM(sublex_start());
5610 #pragma segment Main
5614 S_pending_ident(pTHX)
5617 register I32 tmp = 0;
5618 /* pit holds the identifier we read and pending_ident is reset */
5619 char pit = PL_pending_ident;
5620 PL_pending_ident = 0;
5622 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5623 "### Pending identifier '%s'\n", PL_tokenbuf); });
5625 /* if we're in a my(), we can't allow dynamics here.
5626 $foo'bar has already been turned into $foo::bar, so
5627 just check for colons.
5629 if it's a legal name, the OP is a PADANY.
5632 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5633 if (strchr(PL_tokenbuf,':'))
5634 yyerror(Perl_form(aTHX_ "No package name allowed for "
5635 "variable %s in \"our\"",
5637 tmp = allocmy(PL_tokenbuf);
5640 if (strchr(PL_tokenbuf,':'))
5641 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5643 yylval.opval = newOP(OP_PADANY, 0);
5644 yylval.opval->op_targ = allocmy(PL_tokenbuf);
5650 build the ops for accesses to a my() variable.
5652 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5653 then used in a comparison. This catches most, but not
5654 all cases. For instance, it catches
5655 sort { my($a); $a <=> $b }
5657 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5658 (although why you'd do that is anyone's guess).
5661 if (!strchr(PL_tokenbuf,':')) {
5663 tmp = pad_findmy(PL_tokenbuf);
5664 if (tmp != NOT_IN_PAD) {
5665 /* might be an "our" variable" */
5666 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
5667 /* build ops for a bareword */
5668 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
5669 HEK * const stashname = HvNAME_HEK(stash);
5670 SV * const sym = newSVhek(stashname);
5671 sv_catpvn(sym, "::", 2);
5672 sv_catpv(sym, PL_tokenbuf+1);
5673 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5674 yylval.opval->op_private = OPpCONST_ENTERED;
5677 ? (GV_ADDMULTI | GV_ADDINEVAL)
5680 ((PL_tokenbuf[0] == '$') ? SVt_PV
5681 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5686 /* if it's a sort block and they're naming $a or $b */
5687 if (PL_last_lop_op == OP_SORT &&
5688 PL_tokenbuf[0] == '$' &&
5689 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5692 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5693 d < PL_bufend && *d != '\n';
5696 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5697 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5703 yylval.opval = newOP(OP_PADANY, 0);
5704 yylval.opval->op_targ = tmp;
5710 Whine if they've said @foo in a doublequoted string,
5711 and @foo isn't a variable we can find in the symbol
5714 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5715 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5716 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5717 && ckWARN(WARN_AMBIGUOUS))
5719 /* Downgraded from fatal to warning 20000522 mjd */
5720 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5721 "Possible unintended interpolation of %s in string",
5726 /* build ops for a bareword */
5727 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5728 yylval.opval->op_private = OPpCONST_ENTERED;
5732 ? (GV_ADDMULTI | GV_ADDINEVAL)
5733 /* If the identifier refers to a stash, don't autovivify it.
5734 * Change 24660 had the side effect of causing symbol table
5735 * hashes to always be defined, even if they were freshly
5736 * created and the only reference in the entire program was
5737 * the single statement with the defined %foo::bar:: test.
5738 * It appears that all code in the wild doing this actually
5739 * wants to know whether sub-packages have been loaded, so
5740 * by avoiding auto-vivifying symbol tables, we ensure that
5741 * defined %foo::bar:: continues to be false, and the existing
5742 * tests still give the expected answers, even though what
5743 * they're actually testing has now changed subtly.
5745 : !(*PL_tokenbuf == '%' && *(d = PL_tokenbuf + strlen(PL_tokenbuf) - 1) == ':' && d[-1] == ':'),
5746 ((PL_tokenbuf[0] == '$') ? SVt_PV
5747 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5753 * The following code was generated by perl_keyword.pl.
5757 Perl_keyword (pTHX_ const char *name, I32 len)
5761 case 1: /* 5 tokens of length 1 */
5793 case 2: /* 18 tokens of length 2 */
5939 case 3: /* 29 tokens of length 3 */
5943 if (name[1] == 'N' &&
6006 if (name[1] == 'i' &&
6046 if (name[1] == 'o' &&
6055 if (name[1] == 'e' &&
6064 if (name[1] == 'n' &&
6073 if (name[1] == 'o' &&
6082 if (name[1] == 'a' &&
6091 if (name[1] == 'o' &&
6153 if (name[1] == 'e' &&
6167 return (FEATURE_IS_ENABLED("say", 3) ? -KEY_say : 0);
6193 if (name[1] == 'i' &&
6202 if (name[1] == 's' &&
6211 if (name[1] == 'e' &&
6220 if (name[1] == 'o' &&
6232 case 4: /* 41 tokens of length 4 */
6236 if (name[1] == 'O' &&
6246 if (name[1] == 'N' &&
6256 if (name[1] == 'i' &&
6266 if (name[1] == 'h' &&
6276 if (name[1] == 'u' &&
6289 if (name[2] == 'c' &&
6298 if (name[2] == 's' &&
6307 if (name[2] == 'a' &&
6343 if (name[1] == 'o' &&
6356 if (name[2] == 't' &&
6365 if (name[2] == 'o' &&
6374 if (name[2] == 't' &&
6383 if (name[2] == 'e' &&
6396 if (name[1] == 'o' &&
6409 if (name[2] == 'y' &&
6418 if (name[2] == 'l' &&
6434 if (name[2] == 's' &&
6443 if (name[2] == 'n' &&
6452 if (name[2] == 'c' &&
6465 if (name[1] == 'e' &&
6475 if (name[1] == 'p' &&
6488 if (name[2] == 'c' &&
6497 if (name[2] == 'p' &&
6506 if (name[2] == 's' &&
6522 if (name[2] == 'n' &&
6592 if (name[2] == 'r' &&
6601 if (name[2] == 'r' &&
6610 if (name[2] == 'a' &&
6626 if (name[2] == 'l' &&
6688 if (name[2] == 'e' &&
6691 return (FEATURE_IS_ENABLED("switch", 6) ? KEY_when : 0);
6704 case 5: /* 38 tokens of length 5 */
6708 if (name[1] == 'E' &&
6719 if (name[1] == 'H' &&
6733 if (name[2] == 'a' &&
6743 if (name[2] == 'a' &&
6760 if (name[2] == 'e' &&
6770 if (name[2] == 'e' &&
6774 return (FEATURE_IS_ENABLED("switch", 6) ? -KEY_break : 0);
6790 if (name[3] == 'i' &&
6799 if (name[3] == 'o' &&
6835 if (name[2] == 'o' &&
6845 if (name[2] == 'y' &&
6859 if (name[1] == 'l' &&
6873 if (name[2] == 'n' &&
6883 if (name[2] == 'o' &&
6897 if (name[1] == 'i' &&
6902 return (FEATURE_IS_ENABLED("switch", 6) ? KEY_given : 0);
6911 if (name[2] == 'd' &&
6921 if (name[2] == 'c' &&
6938 if (name[2] == 'c' &&
6948 if (name[2] == 't' &&
6962 if (name[1] == 'k' &&
6973 if (name[1] == 'r' &&
6987 if (name[2] == 's' &&
6997 if (name[2] == 'd' &&
7014 if (name[2] == 'm' &&
7024 if (name[2] == 'i' &&
7034 if (name[2] == 'e' &&
7044 if (name[2] == 'l' &&
7054 if (name[2] == 'a' &&
7064 if (name[2] == 'u' &&
7078 if (name[1] == 'i' &&
7092 if (name[2] == 'a' &&
7105 if (name[3] == 'e' &&
7140 if (name[2] == 'i' &&
7157 if (name[2] == 'i' &&
7167 if (name[2] == 'i' &&
7184 case 6: /* 33 tokens of length 6 */
7188 if (name[1] == 'c' &&
7203 if (name[2] == 'l' &&
7214 if (name[2] == 'r' &&
7229 if (name[1] == 'e' &&
7244 if (name[2] == 's' &&
7249 if(ckWARN_d(WARN_SYNTAX))
7250 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
7256 if (name[2] == 'i' &&
7274 if (name[2] == 'l' &&
7285 if (name[2] == 'r' &&
7300 if (name[1] == 'm' &&
7315 if (name[2] == 'n' &&
7326 if (name[2] == 's' &&
7341 if (name[1] == 's' &&
7347 if (name[4] == 't' &&
7356 if (name[4] == 'e' &&
7365 if (name[4] == 'c' &&
7374 if (name[4] == 'n' &&
7390 if (name[1] == 'r' &&
7408 if (name[3] == 'a' &&
7418 if (name[3] == 'u' &&
7432 if (name[2] == 'n' &&
7450 if (name[2] == 'a' &&
7464 if (name[3] == 'e' &&
7477 if (name[4] == 't' &&
7486 if (name[4] == 'e' &&
7508 if (name[4] == 't' &&
7517 if (name[4] == 'e' &&
7533 if (name[2] == 'c' &&
7544 if (name[2] == 'l' &&
7555 if (name[2] == 'b' &&
7566 if (name[2] == 's' &&
7589 if (name[4] == 's' &&
7598 if (name[4] == 'n' &&
7611 if (name[3] == 'a' &&
7628 if (name[1] == 'a' &&
7643 case 7: /* 29 tokens of length 7 */
7647 if (name[1] == 'E' &&
7660 if (name[1] == '_' &&
7673 if (name[1] == 'i' &&
7680 return -KEY_binmode;
7686 if (name[1] == 'o' &&
7693 return -KEY_connect;
7702 if (name[2] == 'm' &&
7708 return -KEY_dbmopen;
7719 if (name[4] == 'u' &&
7723 return (FEATURE_IS_ENABLED("switch", 6) ? KEY_default : 0);
7729 if (name[4] == 'n' &&
7750 if (name[1] == 'o' &&
7763 if (name[1] == 'e' &&
7770 if (name[5] == 'r' &&
7773 return -KEY_getpgrp;
7779 if (name[5] == 'i' &&
7782 return -KEY_getppid;
7795 if (name[1] == 'c' &&
7802 return -KEY_lcfirst;
7808 if (name[1] == 'p' &&
7815 return -KEY_opendir;
7821 if (name[1] == 'a' &&
7839 if (name[3] == 'd' &&
7844 return -KEY_readdir;
7850 if (name[3] == 'u' &&
7861 if (name[3] == 'e' &&
7866 return -KEY_reverse;
7885 if (name[3] == 'k' &&
7890 return -KEY_seekdir;
7896 if (name[3] == 'p' &&
7901 return -KEY_setpgrp;
7911 if (name[2] == 'm' &&
7917 return -KEY_shmread;
7923 if (name[2] == 'r' &&
7929 return -KEY_sprintf;
7938 if (name[3] == 'l' &&
7943 return -KEY_symlink;
7952 if (name[4] == 'a' &&
7956 return -KEY_syscall;
7962 if (name[4] == 'p' &&
7966 return -KEY_sysopen;
7972 if (name[4] == 'e' &&
7976 return -KEY_sysread;
7982 if (name[4] == 'e' &&
7986 return -KEY_sysseek;
8004 if (name[1] == 'e' &&
8011 return -KEY_telldir;
8020 if (name[2] == 'f' &&
8026 return -KEY_ucfirst;
8032 if (name[2] == 's' &&
8038 return -KEY_unshift;
8048 if (name[1] == 'a' &&
8055 return -KEY_waitpid;
8064 case 8: /* 26 tokens of length 8 */
8068 if (name[1] == 'U' &&
8076 return KEY_AUTOLOAD;
8087 if (name[3] == 'A' &&
8093 return KEY___DATA__;
8099 if (name[3] == 'I' &&
8105 return -KEY___FILE__;
8111 if (name[3] == 'I' &&
8117 return -KEY___LINE__;
8133 if (name[2] == 'o' &&
8140 return -KEY_closedir;
8146 if (name[2] == 'n' &&
8153 return -KEY_continue;
8163 if (name[1] == 'b' &&
8171 return -KEY_dbmclose;
8177 if (name[1] == 'n' &&
8183 if (name[4] == 'r' &&
8188 return -KEY_endgrent;
8194 if (name[4] == 'w' &&
8199 return -KEY_endpwent;
8212 if (name[1] == 'o' &&
8220 return -KEY_formline;
8226 if (name[1] == 'e' &&
8237 if (name[6] == 'n' &&
8240 return -KEY_getgrent;
8246 if (name[6] == 'i' &&
8249 return -KEY_getgrgid;
8255 if (name[6] == 'a' &&
8258 return -KEY_getgrnam;
8271 if (name[4] == 'o' &&
8276 return -KEY_getlogin;
8287 if (name[6] == 'n' &&
8290 return -KEY_getpwent;
8296 if (name[6] == 'a' &&
8299 return -KEY_getpwnam;
8305 if (name[6] == 'i' &&
8308 return -KEY_getpwuid;
8328 if (name[1] == 'e' &&
8335 if (name[5] == 'i' &&
8342 return -KEY_readline;
8347 return -KEY_readlink;
8358 if (name[5] == 'i' &&
8362 return -KEY_readpipe;
8383 if (name[4] == 'r' &&
8388 return -KEY_setgrent;
8394 if (name[4] == 'w' &&
8399 return -KEY_setpwent;
8415 if (name[3] == 'w' &&
8421 return -KEY_shmwrite;
8427 if (name[3] == 't' &&
8433 return -KEY_shutdown;
8443 if (name[2] == 's' &&
8450 return -KEY_syswrite;
8460 if (name[1] == 'r' &&
8468 return -KEY_truncate;
8477 case 9: /* 8 tokens of length 9 */
8481 if (name[1] == 'n' &&
8490 return -KEY_endnetent;
8496 if (name[1] == 'e' &&
8505 return -KEY_getnetent;
8511 if (name[1] == 'o' &&
8520 return -KEY_localtime;
8526 if (name[1] == 'r' &&
8535 return KEY_prototype;
8541 if (name[1] == 'u' &&
8550 return -KEY_quotemeta;
8556 if (name[1] == 'e' &&
8565 return -KEY_rewinddir;
8571 if (name[1] == 'e' &&
8580 return -KEY_setnetent;
8586 if (name[1] == 'a' &&
8595 return -KEY_wantarray;
8604 case 10: /* 9 tokens of length 10 */
8608 if (name[1] == 'n' &&
8614 if (name[4] == 'o' &&
8621 return -KEY_endhostent;
8627 if (name[4] == 'e' &&
8634 return -KEY_endservent;
8647 if (name[1] == 'e' &&
8653 if (name[4] == 'o' &&
8660 return -KEY_gethostent;
8669 if (name[5] == 'r' &&
8675 return -KEY_getservent;
8681 if (name[5] == 'c' &&
8687 return -KEY_getsockopt;
8712 if (name[4] == 'o' &&
8719 return -KEY_sethostent;
8728 if (name[5] == 'r' &&
8734 return -KEY_setservent;
8740 if (name[5] == 'c' &&
8746 return -KEY_setsockopt;
8763 if (name[2] == 'c' &&
8772 return -KEY_socketpair;
8785 case 11: /* 8 tokens of length 11 */
8789 if (name[1] == '_' &&
8800 return -KEY___PACKAGE__;
8806 if (name[1] == 'n' &&
8817 return -KEY_endprotoent;
8823 if (name[1] == 'e' &&
8832 if (name[5] == 'e' &&
8839 return -KEY_getpeername;
8848 if (name[6] == 'o' &&
8854 return -KEY_getpriority;
8860 if (name[6] == 't' &&
8866 return -KEY_getprotoent;
8880 if (name[4] == 'o' &&
8888 return -KEY_getsockname;
8901 if (name[1] == 'e' &&
8909 if (name[6] == 'o' &&
8915 return -KEY_setpriority;
8921 if (name[6] == 't' &&
8927 return -KEY_setprotoent;
8943 case 12: /* 2 tokens of length 12 */
8944 if (name[0] == 'g' &&
8956 if (name[9] == 'd' &&
8959 { /* getnetbyaddr */
8960 return -KEY_getnetbyaddr;
8966 if (name[9] == 'a' &&
8969 { /* getnetbyname */
8970 return -KEY_getnetbyname;
8982 case 13: /* 4 tokens of length 13 */
8983 if (name[0] == 'g' &&
8990 if (name[4] == 'o' &&
8999 if (name[10] == 'd' &&
9002 { /* gethostbyaddr */
9003 return -KEY_gethostbyaddr;
9009 if (name[10] == 'a' &&
9012 { /* gethostbyname */
9013 return -KEY_gethostbyname;
9026 if (name[4] == 'e' &&
9035 if (name[10] == 'a' &&
9038 { /* getservbyname */
9039 return -KEY_getservbyname;
9045 if (name[10] == 'o' &&
9048 { /* getservbyport */
9049 return -KEY_getservbyport;
9068 case 14: /* 1 tokens of length 14 */
9069 if (name[0] == 'g' &&
9083 { /* getprotobyname */
9084 return -KEY_getprotobyname;
9089 case 16: /* 1 tokens of length 16 */
9090 if (name[0] == 'g' &&
9106 { /* getprotobynumber */
9107 return -KEY_getprotobynumber;
9121 S_checkcomma(pTHX_ register char *s, const char *name, const char *what)
9125 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
9126 if (ckWARN(WARN_SYNTAX)) {
9128 for (w = s+2; *w && level; w++) {
9135 for (; *w && isSPACE(*w); w++) ;
9136 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
9137 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
9138 "%s (...) interpreted as function",name);
9141 while (s < PL_bufend && isSPACE(*s))
9145 while (s < PL_bufend && isSPACE(*s))
9147 if (isIDFIRST_lazy_if(s,UTF)) {
9149 while (isALNUM_lazy_if(s,UTF))
9151 while (s < PL_bufend && isSPACE(*s))
9155 *s = '\0'; /* XXX If we didn't do this, we could const a lot of toke.c */
9156 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
9160 Perl_croak(aTHX_ "No comma allowed after %s", what);
9165 /* Either returns sv, or mortalizes sv and returns a new SV*.
9166 Best used as sv=new_constant(..., sv, ...).
9167 If s, pv are NULL, calls subroutine with one argument,
9168 and type is used with error messages only. */
9171 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, SV *sv, SV *pv,
9175 HV * const table = GvHV(PL_hintgv); /* ^H */
9179 const char *why1 = "", *why2 = "", *why3 = "";
9181 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
9184 why2 = strEQ(key,"charnames")
9185 ? "(possibly a missing \"use charnames ...\")"
9187 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
9188 (type ? type: "undef"), why2);
9190 /* This is convoluted and evil ("goto considered harmful")
9191 * but I do not understand the intricacies of all the different
9192 * failure modes of %^H in here. The goal here is to make
9193 * the most probable error message user-friendly. --jhi */
9198 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
9199 (type ? type: "undef"), why1, why2, why3);
9201 yyerror(SvPVX_const(msg));
9205 cvp = hv_fetch(table, key, strlen(key), FALSE);
9206 if (!cvp || !SvOK(*cvp)) {
9209 why3 = "} is not defined";
9212 sv_2mortal(sv); /* Parent created it permanently */
9215 pv = sv_2mortal(newSVpvn(s, len));
9217 typesv = sv_2mortal(newSVpv(type, 0));
9219 typesv = &PL_sv_undef;
9221 PUSHSTACKi(PERLSI_OVERLOAD);
9233 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
9237 /* Check the eval first */
9238 if (!PL_in_eval && SvTRUE(ERRSV)) {
9239 sv_catpv(ERRSV, "Propagated");
9240 yyerror(SvPV_nolen_const(ERRSV)); /* Duplicates the message inside eval */
9242 res = SvREFCNT_inc(sv);
9246 (void)SvREFCNT_inc(res);
9255 why1 = "Call to &{$^H{";
9257 why3 = "}} did not return a defined value";
9265 /* Returns a NUL terminated string, with the length of the string written to
9269 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
9271 register char *d = dest;
9272 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
9275 Perl_croak(aTHX_ ident_too_long);
9276 if (isALNUM(*s)) /* UTF handled below */
9278 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
9283 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
9287 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
9288 char *t = s + UTF8SKIP(s);
9289 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
9291 if (d + (t - s) > e)
9292 Perl_croak(aTHX_ ident_too_long);
9293 Copy(s, d, t - s, char);
9306 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
9310 char *bracket = Nullch;
9316 e = d + destlen - 3; /* two-character token, ending NUL */
9318 while (isDIGIT(*s)) {
9320 Perl_croak(aTHX_ ident_too_long);
9327 Perl_croak(aTHX_ ident_too_long);
9328 if (isALNUM(*s)) /* UTF handled below */
9330 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
9335 else if (*s == ':' && s[1] == ':') {
9339 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
9340 char *t = s + UTF8SKIP(s);
9341 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
9343 if (d + (t - s) > e)
9344 Perl_croak(aTHX_ ident_too_long);
9345 Copy(s, d, t - s, char);
9356 if (PL_lex_state != LEX_NORMAL)
9357 PL_lex_state = LEX_INTERPENDMAYBE;
9360 if (*s == '$' && s[1] &&
9361 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
9374 if (*d == '^' && *s && isCONTROLVAR(*s)) {
9379 if (isSPACE(s[-1])) {
9381 const char ch = *s++;
9382 if (!SPACE_OR_TAB(ch)) {
9388 if (isIDFIRST_lazy_if(d,UTF)) {
9392 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
9394 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
9397 Copy(s, d, e - s, char);
9402 while ((isALNUM(*s) || *s == ':') && d < e)
9405 Perl_croak(aTHX_ ident_too_long);
9408 while (s < send && SPACE_OR_TAB(*s)) s++;
9409 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
9410 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
9411 const char *brack = *s == '[' ? "[...]" : "{...}";
9412 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
9413 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
9414 funny, dest, brack, funny, dest, brack);
9417 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
9421 /* Handle extended ${^Foo} variables
9422 * 1999-02-27 mjd-perl-patch@plover.com */
9423 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
9427 while (isALNUM(*s) && d < e) {
9431 Perl_croak(aTHX_ ident_too_long);
9436 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
9437 PL_lex_state = LEX_INTERPEND;
9442 if (PL_lex_state == LEX_NORMAL) {
9443 if (ckWARN(WARN_AMBIGUOUS) &&
9444 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
9446 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
9447 "Ambiguous use of %c{%s} resolved to %c%s",
9448 funny, dest, funny, dest);
9453 s = bracket; /* let the parser handle it */
9457 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
9458 PL_lex_state = LEX_INTERPEND;
9463 Perl_pmflag(pTHX_ U32* pmfl, int ch)
9468 *pmfl |= PMf_GLOBAL;
9470 *pmfl |= PMf_CONTINUE;
9474 *pmfl |= PMf_MULTILINE;
9476 *pmfl |= PMf_SINGLELINE;
9478 *pmfl |= PMf_EXTENDED;
9482 S_scan_pat(pTHX_ char *start, I32 type)
9485 char *s = scan_str(start,FALSE,FALSE);
9488 char * const delimiter = skipspace(start);
9489 Perl_croak(aTHX_ *delimiter == '?'
9490 ? "Search pattern not terminated or ternary operator parsed as search pattern"
9491 : "Search pattern not terminated" );
9494 pm = (PMOP*)newPMOP(type, 0);
9495 if (PL_multi_open == '?')
9496 pm->op_pmflags |= PMf_ONCE;
9498 while (*s && strchr("iomsx", *s))
9499 pmflag(&pm->op_pmflags,*s++);
9502 while (*s && strchr("iogcmsx", *s))
9503 pmflag(&pm->op_pmflags,*s++);
9505 /* issue a warning if /c is specified,but /g is not */
9506 if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL)
9507 && ckWARN(WARN_REGEXP))
9509 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless without /g" );
9512 pm->op_pmpermflags = pm->op_pmflags;
9514 PL_lex_op = (OP*)pm;
9515 yylval.ival = OP_MATCH;
9520 S_scan_subst(pTHX_ char *start)
9528 yylval.ival = OP_NULL;
9530 s = scan_str(start,FALSE,FALSE);
9533 Perl_croak(aTHX_ "Substitution pattern not terminated");
9535 if (s[-1] == PL_multi_open)
9538 first_start = PL_multi_start;
9539 s = scan_str(s,FALSE,FALSE);
9542 SvREFCNT_dec(PL_lex_stuff);
9543 PL_lex_stuff = Nullsv;
9545 Perl_croak(aTHX_ "Substitution replacement not terminated");
9547 PL_multi_start = first_start; /* so whole substitution is taken together */
9549 pm = (PMOP*)newPMOP(OP_SUBST, 0);
9555 else if (strchr("iogcmsx", *s))
9556 pmflag(&pm->op_pmflags,*s++);
9561 if ((pm->op_pmflags & PMf_CONTINUE) && ckWARN(WARN_REGEXP)) {
9562 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless in s///" );
9567 PL_sublex_info.super_bufptr = s;
9568 PL_sublex_info.super_bufend = PL_bufend;
9570 pm->op_pmflags |= PMf_EVAL;
9571 repl = newSVpvn("",0);
9573 sv_catpv(repl, es ? "eval " : "do ");
9574 sv_catpvn(repl, "{ ", 2);
9575 sv_catsv(repl, PL_lex_repl);
9576 sv_catpvn(repl, " };", 2);
9578 SvREFCNT_dec(PL_lex_repl);
9582 pm->op_pmpermflags = pm->op_pmflags;
9583 PL_lex_op = (OP*)pm;
9584 yylval.ival = OP_SUBST;
9589 S_scan_trans(pTHX_ char *start)
9598 yylval.ival = OP_NULL;
9600 s = scan_str(start,FALSE,FALSE);
9602 Perl_croak(aTHX_ "Transliteration pattern not terminated");
9603 if (s[-1] == PL_multi_open)
9606 s = scan_str(s,FALSE,FALSE);
9609 SvREFCNT_dec(PL_lex_stuff);
9610 PL_lex_stuff = Nullsv;
9612 Perl_croak(aTHX_ "Transliteration replacement not terminated");
9615 complement = del = squash = 0;
9619 complement = OPpTRANS_COMPLEMENT;
9622 del = OPpTRANS_DELETE;
9625 squash = OPpTRANS_SQUASH;
9634 Newx(tbl, complement&&!del?258:256, short);
9635 o = newPVOP(OP_TRANS, 0, (char*)tbl);
9636 o->op_private &= ~OPpTRANS_ALL;
9637 o->op_private |= del|squash|complement|
9638 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
9639 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
9642 yylval.ival = OP_TRANS;
9647 S_scan_heredoc(pTHX_ register char *s)
9650 I32 op_type = OP_SCALAR;
9654 const char newline[] = "\n";
9655 const char *found_newline;
9659 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
9663 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
9666 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
9667 if (*peek == '`' || *peek == '\'' || *peek =='"') {
9670 s = delimcpy(d, e, s, PL_bufend, term, &len);
9680 if (!isALNUM_lazy_if(s,UTF))
9681 deprecate_old("bare << to mean <<\"\"");
9682 for (; isALNUM_lazy_if(s,UTF); s++) {
9687 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
9688 Perl_croak(aTHX_ "Delimiter for here document is too long");
9691 len = d - PL_tokenbuf;
9692 #ifndef PERL_STRICT_CR
9693 d = strchr(s, '\r');
9695 char * const olds = s;
9697 while (s < PL_bufend) {
9703 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
9712 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
9716 if ( outer || !(found_newline = ninstr(s,PL_bufend,newline,newline+1)) ) {
9717 herewas = newSVpvn(s,PL_bufend-s);
9721 herewas = newSVpvn(s,found_newline-s);
9723 s += SvCUR(herewas);
9725 tmpstr = NEWSV(87,79);
9726 sv_upgrade(tmpstr, SVt_PVIV);
9729 SvIV_set(tmpstr, -1);
9731 else if (term == '`') {
9732 op_type = OP_BACKTICK;
9733 SvIV_set(tmpstr, '\\');
9737 PL_multi_start = CopLINE(PL_curcop);
9738 PL_multi_open = PL_multi_close = '<';
9739 term = *PL_tokenbuf;
9740 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
9741 char *bufptr = PL_sublex_info.super_bufptr;
9742 char *bufend = PL_sublex_info.super_bufend;
9743 char * const olds = s - SvCUR(herewas);
9744 s = strchr(bufptr, '\n');
9748 while (s < bufend &&
9749 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
9751 CopLINE_inc(PL_curcop);
9754 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9755 missingterm(PL_tokenbuf);
9757 sv_setpvn(herewas,bufptr,d-bufptr+1);
9758 sv_setpvn(tmpstr,d+1,s-d);
9760 sv_catpvn(herewas,s,bufend-s);
9761 Copy(SvPVX_const(herewas),bufptr,SvCUR(herewas) + 1,char);
9768 while (s < PL_bufend &&
9769 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
9771 CopLINE_inc(PL_curcop);
9773 if (s >= PL_bufend) {
9774 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9775 missingterm(PL_tokenbuf);
9777 sv_setpvn(tmpstr,d+1,s-d);
9779 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
9781 sv_catpvn(herewas,s,PL_bufend-s);
9782 sv_setsv(PL_linestr,herewas);
9783 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
9784 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9785 PL_last_lop = PL_last_uni = Nullch;
9788 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
9789 while (s >= PL_bufend) { /* multiple line string? */
9791 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
9792 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9793 missingterm(PL_tokenbuf);
9795 CopLINE_inc(PL_curcop);
9796 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9797 PL_last_lop = PL_last_uni = Nullch;
9798 #ifndef PERL_STRICT_CR
9799 if (PL_bufend - PL_linestart >= 2) {
9800 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
9801 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
9803 PL_bufend[-2] = '\n';
9805 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
9807 else if (PL_bufend[-1] == '\r')
9808 PL_bufend[-1] = '\n';
9810 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
9811 PL_bufend[-1] = '\n';
9813 if (PERLDB_LINE && PL_curstash != PL_debstash) {
9814 SV *sv = NEWSV(88,0);
9816 sv_upgrade(sv, SVt_PVMG);
9817 sv_setsv(sv,PL_linestr);
9820 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
9822 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
9823 STRLEN off = PL_bufend - 1 - SvPVX_const(PL_linestr);
9824 *(SvPVX(PL_linestr) + off ) = ' ';
9825 sv_catsv(PL_linestr,herewas);
9826 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9827 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
9831 sv_catsv(tmpstr,PL_linestr);
9836 PL_multi_end = CopLINE(PL_curcop);
9837 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
9838 SvPV_shrink_to_cur(tmpstr);
9840 SvREFCNT_dec(herewas);
9842 if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
9844 else if (PL_encoding)
9845 sv_recode_to_utf8(tmpstr, PL_encoding);
9847 PL_lex_stuff = tmpstr;
9848 yylval.ival = op_type;
9853 takes: current position in input buffer
9854 returns: new position in input buffer
9855 side-effects: yylval and lex_op are set.
9860 <FH> read from filehandle
9861 <pkg::FH> read from package qualified filehandle
9862 <pkg'FH> read from package qualified filehandle
9863 <$fh> read from filehandle in $fh
9869 S_scan_inputsymbol(pTHX_ char *start)
9871 register char *s = start; /* current position in buffer */
9877 d = PL_tokenbuf; /* start of temp holding space */
9878 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
9879 end = strchr(s, '\n');
9882 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
9884 /* die if we didn't have space for the contents of the <>,
9885 or if it didn't end, or if we see a newline
9888 if (len >= sizeof PL_tokenbuf)
9889 Perl_croak(aTHX_ "Excessively long <> operator");
9891 Perl_croak(aTHX_ "Unterminated <> operator");
9896 Remember, only scalar variables are interpreted as filehandles by
9897 this code. Anything more complex (e.g., <$fh{$num}>) will be
9898 treated as a glob() call.
9899 This code makes use of the fact that except for the $ at the front,
9900 a scalar variable and a filehandle look the same.
9902 if (*d == '$' && d[1]) d++;
9904 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
9905 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
9908 /* If we've tried to read what we allow filehandles to look like, and
9909 there's still text left, then it must be a glob() and not a getline.
9910 Use scan_str to pull out the stuff between the <> and treat it
9911 as nothing more than a string.
9914 if (d - PL_tokenbuf != len) {
9915 yylval.ival = OP_GLOB;
9917 s = scan_str(start,FALSE,FALSE);
9919 Perl_croak(aTHX_ "Glob not terminated");
9923 bool readline_overriden = FALSE;
9924 GV *gv_readline = Nullgv;
9926 /* we're in a filehandle read situation */
9929 /* turn <> into <ARGV> */
9931 Copy("ARGV",d,5,char);
9933 /* Check whether readline() is overriden */
9934 if (((gv_readline = gv_fetchpv("readline", FALSE, SVt_PVCV))
9935 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
9937 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
9938 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
9939 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
9940 readline_overriden = TRUE;
9942 /* if <$fh>, create the ops to turn the variable into a
9948 /* try to find it in the pad for this block, otherwise find
9949 add symbol table ops
9951 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
9952 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
9953 HV *stash = PAD_COMPNAME_OURSTASH(tmp);
9954 HEK *stashname = HvNAME_HEK(stash);
9955 SV *sym = sv_2mortal(newSVhek(stashname));
9956 sv_catpvn(sym, "::", 2);
9962 OP *o = newOP(OP_PADSV, 0);
9964 PL_lex_op = readline_overriden
9965 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
9966 append_elem(OP_LIST, o,
9967 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
9968 : (OP*)newUNOP(OP_READLINE, 0, o);
9977 ? (GV_ADDMULTI | GV_ADDINEVAL)
9980 PL_lex_op = readline_overriden
9981 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
9982 append_elem(OP_LIST,
9983 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
9984 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
9985 : (OP*)newUNOP(OP_READLINE, 0,
9986 newUNOP(OP_RV2SV, 0,
9987 newGVOP(OP_GV, 0, gv)));
9989 if (!readline_overriden)
9990 PL_lex_op->op_flags |= OPf_SPECIAL;
9991 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
9992 yylval.ival = OP_NULL;
9995 /* If it's none of the above, it must be a literal filehandle
9996 (<Foo::BAR> or <FOO>) so build a simple readline OP */
9998 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
9999 PL_lex_op = readline_overriden
10000 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
10001 append_elem(OP_LIST,
10002 newGVOP(OP_GV, 0, gv),
10003 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
10004 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
10005 yylval.ival = OP_NULL;
10014 takes: start position in buffer
10015 keep_quoted preserve \ on the embedded delimiter(s)
10016 keep_delims preserve the delimiters around the string
10017 returns: position to continue reading from buffer
10018 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
10019 updates the read buffer.
10021 This subroutine pulls a string out of the input. It is called for:
10022 q single quotes q(literal text)
10023 ' single quotes 'literal text'
10024 qq double quotes qq(interpolate $here please)
10025 " double quotes "interpolate $here please"
10026 qx backticks qx(/bin/ls -l)
10027 ` backticks `/bin/ls -l`
10028 qw quote words @EXPORT_OK = qw( func() $spam )
10029 m// regexp match m/this/
10030 s/// regexp substitute s/this/that/
10031 tr/// string transliterate tr/this/that/
10032 y/// string transliterate y/this/that/
10033 ($*@) sub prototypes sub foo ($)
10034 (stuff) sub attr parameters sub foo : attr(stuff)
10035 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
10037 In most of these cases (all but <>, patterns and transliterate)
10038 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
10039 calls scan_str(). s/// makes yylex() call scan_subst() which calls
10040 scan_str(). tr/// and y/// make yylex() call scan_trans() which
10043 It skips whitespace before the string starts, and treats the first
10044 character as the delimiter. If the delimiter is one of ([{< then
10045 the corresponding "close" character )]}> is used as the closing
10046 delimiter. It allows quoting of delimiters, and if the string has
10047 balanced delimiters ([{<>}]) it allows nesting.
10049 On success, the SV with the resulting string is put into lex_stuff or,
10050 if that is already non-NULL, into lex_repl. The second case occurs only
10051 when parsing the RHS of the special constructs s/// and tr/// (y///).
10052 For convenience, the terminating delimiter character is stuffed into
10057 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
10059 SV *sv; /* scalar value: string */
10060 char *tmps; /* temp string, used for delimiter matching */
10061 register char *s = start; /* current position in the buffer */
10062 register char term; /* terminating character */
10063 register char *to; /* current position in the sv's data */
10064 I32 brackets = 1; /* bracket nesting level */
10065 bool has_utf8 = FALSE; /* is there any utf8 content? */
10066 I32 termcode; /* terminating char. code */
10067 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
10068 STRLEN termlen; /* length of terminating string */
10069 char *last = NULL; /* last position for nesting bracket */
10071 /* skip space before the delimiter */
10075 /* mark where we are, in case we need to report errors */
10078 /* after skipping whitespace, the next character is the terminator */
10081 termcode = termstr[0] = term;
10085 termcode = utf8_to_uvchr((U8*)s, &termlen);
10086 Copy(s, termstr, termlen, U8);
10087 if (!UTF8_IS_INVARIANT(term))
10091 /* mark where we are */
10092 PL_multi_start = CopLINE(PL_curcop);
10093 PL_multi_open = term;
10095 /* find corresponding closing delimiter */
10096 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
10097 termcode = termstr[0] = term = tmps[5];
10099 PL_multi_close = term;
10101 /* create a new SV to hold the contents. 87 is leak category, I'm
10102 assuming. 79 is the SV's initial length. What a random number. */
10104 sv_upgrade(sv, SVt_PVIV);
10105 SvIV_set(sv, termcode);
10106 (void)SvPOK_only(sv); /* validate pointer */
10108 /* move past delimiter and try to read a complete string */
10110 sv_catpvn(sv, s, termlen);
10113 if (PL_encoding && !UTF) {
10117 int offset = s - SvPVX_const(PL_linestr);
10118 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
10119 &offset, (char*)termstr, termlen);
10120 const char *ns = SvPVX_const(PL_linestr) + offset;
10121 char *svlast = SvEND(sv) - 1;
10123 for (; s < ns; s++) {
10124 if (*s == '\n' && !PL_rsfp)
10125 CopLINE_inc(PL_curcop);
10128 goto read_more_line;
10130 /* handle quoted delimiters */
10131 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
10133 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
10135 if ((svlast-1 - t) % 2) {
10136 if (!keep_quoted) {
10137 *(svlast-1) = term;
10139 SvCUR_set(sv, SvCUR(sv) - 1);
10144 if (PL_multi_open == PL_multi_close) {
10152 for (t = w = last; t < svlast; w++, t++) {
10153 /* At here, all closes are "was quoted" one,
10154 so we don't check PL_multi_close. */
10156 if (!keep_quoted && *(t+1) == PL_multi_open)
10161 else if (*t == PL_multi_open)
10169 SvCUR_set(sv, w - SvPVX_const(sv));
10172 if (--brackets <= 0)
10177 if (!keep_delims) {
10178 SvCUR_set(sv, SvCUR(sv) - 1);
10184 /* extend sv if need be */
10185 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
10186 /* set 'to' to the next character in the sv's string */
10187 to = SvPVX(sv)+SvCUR(sv);
10189 /* if open delimiter is the close delimiter read unbridle */
10190 if (PL_multi_open == PL_multi_close) {
10191 for (; s < PL_bufend; s++,to++) {
10192 /* embedded newlines increment the current line number */
10193 if (*s == '\n' && !PL_rsfp)
10194 CopLINE_inc(PL_curcop);
10195 /* handle quoted delimiters */
10196 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
10197 if (!keep_quoted && s[1] == term)
10199 /* any other quotes are simply copied straight through */
10203 /* terminate when run out of buffer (the for() condition), or
10204 have found the terminator */
10205 else if (*s == term) {
10208 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
10211 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
10217 /* if the terminator isn't the same as the start character (e.g.,
10218 matched brackets), we have to allow more in the quoting, and
10219 be prepared for nested brackets.
10222 /* read until we run out of string, or we find the terminator */
10223 for (; s < PL_bufend; s++,to++) {
10224 /* embedded newlines increment the line count */
10225 if (*s == '\n' && !PL_rsfp)
10226 CopLINE_inc(PL_curcop);
10227 /* backslashes can escape the open or closing characters */
10228 if (*s == '\\' && s+1 < PL_bufend) {
10229 if (!keep_quoted &&
10230 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
10235 /* allow nested opens and closes */
10236 else if (*s == PL_multi_close && --brackets <= 0)
10238 else if (*s == PL_multi_open)
10240 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
10245 /* terminate the copied string and update the sv's end-of-string */
10247 SvCUR_set(sv, to - SvPVX_const(sv));
10250 * this next chunk reads more into the buffer if we're not done yet
10254 break; /* handle case where we are done yet :-) */
10256 #ifndef PERL_STRICT_CR
10257 if (to - SvPVX_const(sv) >= 2) {
10258 if ((to[-2] == '\r' && to[-1] == '\n') ||
10259 (to[-2] == '\n' && to[-1] == '\r'))
10263 SvCUR_set(sv, to - SvPVX_const(sv));
10265 else if (to[-1] == '\r')
10268 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
10273 /* if we're out of file, or a read fails, bail and reset the current
10274 line marker so we can report where the unterminated string began
10277 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
10279 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
10282 /* we read a line, so increment our line counter */
10283 CopLINE_inc(PL_curcop);
10285 /* update debugger info */
10286 if (PERLDB_LINE && PL_curstash != PL_debstash) {
10287 SV *sv = NEWSV(88,0);
10289 sv_upgrade(sv, SVt_PVMG);
10290 sv_setsv(sv,PL_linestr);
10291 (void)SvIOK_on(sv);
10293 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
10296 /* having changed the buffer, we must update PL_bufend */
10297 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
10298 PL_last_lop = PL_last_uni = Nullch;
10301 /* at this point, we have successfully read the delimited string */
10303 if (!PL_encoding || UTF) {
10305 sv_catpvn(sv, s, termlen);
10308 if (has_utf8 || PL_encoding)
10311 PL_multi_end = CopLINE(PL_curcop);
10313 /* if we allocated too much space, give some back */
10314 if (SvCUR(sv) + 5 < SvLEN(sv)) {
10315 SvLEN_set(sv, SvCUR(sv) + 1);
10316 SvPV_renew(sv, SvLEN(sv));
10319 /* decide whether this is the first or second quoted string we've read
10332 takes: pointer to position in buffer
10333 returns: pointer to new position in buffer
10334 side-effects: builds ops for the constant in yylval.op
10336 Read a number in any of the formats that Perl accepts:
10338 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
10339 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
10342 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
10344 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
10347 If it reads a number without a decimal point or an exponent, it will
10348 try converting the number to an integer and see if it can do so
10349 without loss of precision.
10353 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
10355 register const char *s = start; /* current position in buffer */
10356 register char *d; /* destination in temp buffer */
10357 register char *e; /* end of temp buffer */
10358 NV nv; /* number read, as a double */
10359 SV *sv = Nullsv; /* place to put the converted number */
10360 bool floatit; /* boolean: int or float? */
10361 const char *lastub = 0; /* position of last underbar */
10362 static char const number_too_long[] = "Number too long";
10364 /* We use the first character to decide what type of number this is */
10368 Perl_croak(aTHX_ "panic: scan_num");
10370 /* if it starts with a 0, it could be an octal number, a decimal in
10371 0.13 disguise, or a hexadecimal number, or a binary number. */
10375 u holds the "number so far"
10376 shift the power of 2 of the base
10377 (hex == 4, octal == 3, binary == 1)
10378 overflowed was the number more than we can hold?
10380 Shift is used when we add a digit. It also serves as an "are
10381 we in octal/hex/binary?" indicator to disallow hex characters
10382 when in octal mode.
10387 bool overflowed = FALSE;
10388 bool just_zero = TRUE; /* just plain 0 or binary number? */
10389 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
10390 static const char* const bases[5] =
10391 { "", "binary", "", "octal", "hexadecimal" };
10392 static const char* const Bases[5] =
10393 { "", "Binary", "", "Octal", "Hexadecimal" };
10394 static const char* const maxima[5] =
10396 "0b11111111111111111111111111111111",
10400 const char *base, *Base, *max;
10402 /* check for hex */
10407 } else if (s[1] == 'b') {
10412 /* check for a decimal in disguise */
10413 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
10415 /* so it must be octal */
10422 if (ckWARN(WARN_SYNTAX))
10423 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10424 "Misplaced _ in number");
10428 base = bases[shift];
10429 Base = Bases[shift];
10430 max = maxima[shift];
10432 /* read the rest of the number */
10434 /* x is used in the overflow test,
10435 b is the digit we're adding on. */
10440 /* if we don't mention it, we're done */
10444 /* _ are ignored -- but warned about if consecutive */
10446 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10447 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10448 "Misplaced _ in number");
10452 /* 8 and 9 are not octal */
10453 case '8': case '9':
10455 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
10459 case '2': case '3': case '4':
10460 case '5': case '6': case '7':
10462 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
10465 case '0': case '1':
10466 b = *s++ & 15; /* ASCII digit -> value of digit */
10470 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
10471 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
10472 /* make sure they said 0x */
10475 b = (*s++ & 7) + 9;
10477 /* Prepare to put the digit we have onto the end
10478 of the number so far. We check for overflows.
10484 x = u << shift; /* make room for the digit */
10486 if ((x >> shift) != u
10487 && !(PL_hints & HINT_NEW_BINARY)) {
10490 if (ckWARN_d(WARN_OVERFLOW))
10491 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
10492 "Integer overflow in %s number",
10495 u = x | b; /* add the digit to the end */
10498 n *= nvshift[shift];
10499 /* If an NV has not enough bits in its
10500 * mantissa to represent an UV this summing of
10501 * small low-order numbers is a waste of time
10502 * (because the NV cannot preserve the
10503 * low-order bits anyway): we could just
10504 * remember when did we overflow and in the
10505 * end just multiply n by the right
10513 /* if we get here, we had success: make a scalar value from
10518 /* final misplaced underbar check */
10519 if (s[-1] == '_') {
10520 if (ckWARN(WARN_SYNTAX))
10521 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
10526 if (n > 4294967295.0 && ckWARN(WARN_PORTABLE))
10527 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
10528 "%s number > %s non-portable",
10534 if (u > 0xffffffff && ckWARN(WARN_PORTABLE))
10535 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
10536 "%s number > %s non-portable",
10541 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
10542 sv = new_constant(start, s - start, "integer",
10544 else if (PL_hints & HINT_NEW_BINARY)
10545 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
10550 handle decimal numbers.
10551 we're also sent here when we read a 0 as the first digit
10553 case '1': case '2': case '3': case '4': case '5':
10554 case '6': case '7': case '8': case '9': case '.':
10557 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
10560 /* read next group of digits and _ and copy into d */
10561 while (isDIGIT(*s) || *s == '_') {
10562 /* skip underscores, checking for misplaced ones
10566 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10567 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10568 "Misplaced _ in number");
10572 /* check for end of fixed-length buffer */
10574 Perl_croak(aTHX_ number_too_long);
10575 /* if we're ok, copy the character */
10580 /* final misplaced underbar check */
10581 if (lastub && s == lastub + 1) {
10582 if (ckWARN(WARN_SYNTAX))
10583 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
10586 /* read a decimal portion if there is one. avoid
10587 3..5 being interpreted as the number 3. followed
10590 if (*s == '.' && s[1] != '.') {
10595 if (ckWARN(WARN_SYNTAX))
10596 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10597 "Misplaced _ in number");
10601 /* copy, ignoring underbars, until we run out of digits.
10603 for (; isDIGIT(*s) || *s == '_'; s++) {
10604 /* fixed length buffer check */
10606 Perl_croak(aTHX_ number_too_long);
10608 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10609 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10610 "Misplaced _ in number");
10616 /* fractional part ending in underbar? */
10617 if (s[-1] == '_') {
10618 if (ckWARN(WARN_SYNTAX))
10619 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10620 "Misplaced _ in number");
10622 if (*s == '.' && isDIGIT(s[1])) {
10623 /* oops, it's really a v-string, but without the "v" */
10629 /* read exponent part, if present */
10630 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
10634 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
10635 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
10637 /* stray preinitial _ */
10639 if (ckWARN(WARN_SYNTAX))
10640 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10641 "Misplaced _ in number");
10645 /* allow positive or negative exponent */
10646 if (*s == '+' || *s == '-')
10649 /* stray initial _ */
10651 if (ckWARN(WARN_SYNTAX))
10652 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10653 "Misplaced _ in number");
10657 /* read digits of exponent */
10658 while (isDIGIT(*s) || *s == '_') {
10661 Perl_croak(aTHX_ number_too_long);
10665 if (((lastub && s == lastub + 1) ||
10666 (!isDIGIT(s[1]) && s[1] != '_'))
10667 && ckWARN(WARN_SYNTAX))
10668 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10669 "Misplaced _ in number");
10676 /* make an sv from the string */
10680 We try to do an integer conversion first if no characters
10681 indicating "float" have been found.
10686 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
10688 if (flags == IS_NUMBER_IN_UV) {
10690 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
10693 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
10694 if (uv <= (UV) IV_MIN)
10695 sv_setiv(sv, -(IV)uv);
10702 /* terminate the string */
10704 nv = Atof(PL_tokenbuf);
10708 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
10709 (PL_hints & HINT_NEW_INTEGER) )
10710 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
10711 (floatit ? "float" : "integer"),
10715 /* if it starts with a v, it could be a v-string */
10718 sv = NEWSV(92,5); /* preallocate storage space */
10719 s = scan_vstring(s,sv);
10723 /* make the op for the constant and return */
10726 lvalp->opval = newSVOP(OP_CONST, 0, sv);
10728 lvalp->opval = Nullop;
10734 S_scan_formline(pTHX_ register char *s)
10736 register char *eol;
10738 SV *stuff = newSVpvn("",0);
10739 bool needargs = FALSE;
10740 bool eofmt = FALSE;
10742 while (!needargs) {
10744 #ifdef PERL_STRICT_CR
10745 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
10747 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
10749 if (*t == '\n' || t == PL_bufend) {
10754 if (PL_in_eval && !PL_rsfp) {
10755 eol = (char *) memchr(s,'\n',PL_bufend-s);
10760 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
10762 for (t = s; t < eol; t++) {
10763 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
10765 goto enough; /* ~~ must be first line in formline */
10767 if (*t == '@' || *t == '^')
10771 sv_catpvn(stuff, s, eol-s);
10772 #ifndef PERL_STRICT_CR
10773 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
10774 char *end = SvPVX(stuff) + SvCUR(stuff);
10777 SvCUR_set(stuff, SvCUR(stuff) - 1);
10786 s = filter_gets(PL_linestr, PL_rsfp, 0);
10787 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
10788 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
10789 PL_last_lop = PL_last_uni = Nullch;
10798 if (SvCUR(stuff)) {
10801 PL_lex_state = LEX_NORMAL;
10802 PL_nextval[PL_nexttoke].ival = 0;
10806 PL_lex_state = LEX_FORMLINE;
10808 if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
10810 else if (PL_encoding)
10811 sv_recode_to_utf8(stuff, PL_encoding);
10813 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
10815 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
10819 SvREFCNT_dec(stuff);
10821 PL_lex_formbrack = 0;
10832 PL_cshlen = strlen(PL_cshname);
10837 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
10839 const I32 oldsavestack_ix = PL_savestack_ix;
10840 CV* outsidecv = PL_compcv;
10843 assert(SvTYPE(PL_compcv) == SVt_PVCV);
10845 SAVEI32(PL_subline);
10846 save_item(PL_subname);
10847 SAVESPTR(PL_compcv);
10849 PL_compcv = (CV*)NEWSV(1104,0);
10850 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
10851 CvFLAGS(PL_compcv) |= flags;
10853 PL_subline = CopLINE(PL_curcop);
10854 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
10855 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
10856 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
10858 return oldsavestack_ix;
10862 #pragma segment Perl_yylex
10865 Perl_yywarn(pTHX_ const char *s)
10867 PL_in_eval |= EVAL_WARNONLY;
10869 PL_in_eval &= ~EVAL_WARNONLY;
10874 Perl_yyerror(pTHX_ const char *s)
10876 const char *where = NULL;
10877 const char *context = NULL;
10881 if (!yychar || (yychar == ';' && !PL_rsfp))
10883 else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
10884 PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
10885 PL_oldbufptr != PL_bufptr) {
10888 The code below is removed for NetWare because it abends/crashes on NetWare
10889 when the script has error such as not having the closing quotes like:
10890 if ($var eq "value)
10891 Checking of white spaces is anyway done in NetWare code.
10894 while (isSPACE(*PL_oldoldbufptr))
10897 context = PL_oldoldbufptr;
10898 contlen = PL_bufptr - PL_oldoldbufptr;
10900 else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
10901 PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
10904 The code below is removed for NetWare because it abends/crashes on NetWare
10905 when the script has error such as not having the closing quotes like:
10906 if ($var eq "value)
10907 Checking of white spaces is anyway done in NetWare code.
10910 while (isSPACE(*PL_oldbufptr))
10913 context = PL_oldbufptr;
10914 contlen = PL_bufptr - PL_oldbufptr;
10916 else if (yychar > 255)
10917 where = "next token ???";
10918 else if (yychar == -2) { /* YYEMPTY */
10919 if (PL_lex_state == LEX_NORMAL ||
10920 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
10921 where = "at end of line";
10922 else if (PL_lex_inpat)
10923 where = "within pattern";
10925 where = "within string";
10928 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
10930 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
10931 else if (isPRINT_LC(yychar))
10932 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
10934 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
10935 where = SvPVX_const(where_sv);
10937 msg = sv_2mortal(newSVpv(s, 0));
10938 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
10939 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
10941 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
10943 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
10944 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
10945 Perl_sv_catpvf(aTHX_ msg,
10946 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
10947 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
10950 if (PL_in_eval & EVAL_WARNONLY && ckWARN_d(WARN_SYNTAX))
10951 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, msg);
10954 if (PL_error_count >= 10) {
10955 if (PL_in_eval && SvCUR(ERRSV))
10956 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
10957 ERRSV, OutCopFILE(PL_curcop));
10959 Perl_croak(aTHX_ "%s has too many errors.\n",
10960 OutCopFILE(PL_curcop));
10963 PL_in_my_stash = NULL;
10967 #pragma segment Main
10971 S_swallow_bom(pTHX_ U8 *s)
10973 const STRLEN slen = SvCUR(PL_linestr);
10976 if (s[1] == 0xFE) {
10977 /* UTF-16 little-endian? (or UTF32-LE?) */
10978 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
10979 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
10980 #ifndef PERL_NO_UTF16_FILTER
10981 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
10984 if (PL_bufend > (char*)s) {
10988 filter_add(utf16rev_textfilter, NULL);
10989 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
10990 utf16_to_utf8_reversed(s, news,
10991 PL_bufend - (char*)s - 1,
10993 sv_setpvn(PL_linestr, (const char*)news, newlen);
10995 SvUTF8_on(PL_linestr);
10996 s = (U8*)SvPVX(PL_linestr);
10997 PL_bufend = SvPVX(PL_linestr) + newlen;
11000 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
11005 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
11006 #ifndef PERL_NO_UTF16_FILTER
11007 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
11010 if (PL_bufend > (char *)s) {
11014 filter_add(utf16_textfilter, NULL);
11015 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
11016 utf16_to_utf8(s, news,
11017 PL_bufend - (char*)s,
11019 sv_setpvn(PL_linestr, (const char*)news, newlen);
11021 SvUTF8_on(PL_linestr);
11022 s = (U8*)SvPVX(PL_linestr);
11023 PL_bufend = SvPVX(PL_linestr) + newlen;
11026 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
11031 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
11032 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
11033 s += 3; /* UTF-8 */
11039 if (s[2] == 0xFE && s[3] == 0xFF) {
11040 /* UTF-32 big-endian */
11041 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
11044 else if (s[2] == 0 && s[3] != 0) {
11047 * are a good indicator of UTF-16BE. */
11048 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
11053 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
11056 * are a good indicator of UTF-16LE. */
11057 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
11066 * Restore a source filter.
11070 restore_rsfp(pTHX_ void *f)
11072 PerlIO * const fp = (PerlIO*)f;
11074 if (PL_rsfp == PerlIO_stdin())
11075 PerlIO_clearerr(PL_rsfp);
11076 else if (PL_rsfp && (PL_rsfp != fp))
11077 PerlIO_close(PL_rsfp);
11081 #ifndef PERL_NO_UTF16_FILTER
11083 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
11085 const STRLEN old = SvCUR(sv);
11086 const I32 count = FILTER_READ(idx+1, sv, maxlen);
11087 DEBUG_P(PerlIO_printf(Perl_debug_log,
11088 "utf16_textfilter(%p): %d %d (%d)\n",
11089 utf16_textfilter, idx, maxlen, (int) count));
11093 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
11094 Copy(SvPVX_const(sv), tmps, old, char);
11095 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
11096 SvCUR(sv) - old, &newlen);
11097 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
11099 DEBUG_P({sv_dump(sv);});
11104 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
11106 const STRLEN old = SvCUR(sv);
11107 const I32 count = FILTER_READ(idx+1, sv, maxlen);
11108 DEBUG_P(PerlIO_printf(Perl_debug_log,
11109 "utf16rev_textfilter(%p): %d %d (%d)\n",
11110 utf16rev_textfilter, idx, maxlen, (int) count));
11114 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
11115 Copy(SvPVX_const(sv), tmps, old, char);
11116 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
11117 SvCUR(sv) - old, &newlen);
11118 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
11120 DEBUG_P({ sv_dump(sv); });
11126 Returns a pointer to the next character after the parsed
11127 vstring, as well as updating the passed in sv.
11129 Function must be called like
11132 s = scan_vstring(s,sv);
11134 The sv should already be large enough to store the vstring
11135 passed in, for performance reasons.
11140 Perl_scan_vstring(pTHX_ const char *s, SV *sv)
11142 const char *pos = s;
11143 const char *start = s;
11144 if (*pos == 'v') pos++; /* get past 'v' */
11145 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
11147 if ( *pos != '.') {
11148 /* this may not be a v-string if followed by => */
11149 const char *next = pos;
11150 while (next < PL_bufend && isSPACE(*next))
11152 if ((PL_bufend - next) >= 2 && *next == '=' && next[1] == '>' ) {
11153 /* return string not v-string */
11154 sv_setpvn(sv,(char *)s,pos-s);
11155 return (char *)pos;
11159 if (!isALPHA(*pos)) {
11160 U8 tmpbuf[UTF8_MAXBYTES+1];
11162 if (*s == 'v') s++; /* get past 'v' */
11164 sv_setpvn(sv, "", 0);
11170 /* this is atoi() that tolerates underscores */
11171 const char *end = pos;
11173 while (--end >= s) {
11178 rev += (*end - '0') * mult;
11180 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
11181 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
11182 "Integer overflow in decimal number");
11186 if (rev > 0x7FFFFFFF)
11187 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
11189 /* Append native character for the rev point */
11190 tmpend = uvchr_to_utf8(tmpbuf, rev);
11191 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
11192 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
11194 if (pos + 1 < PL_bufend && *pos == '.' && isDIGIT(pos[1]))
11200 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
11204 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
11212 * c-indentation-style: bsd
11213 * c-basic-offset: 4
11214 * indent-tabs-mode: t
11217 * ex: set ts=8 sts=4 sw=4 noet: