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;
2432 SV* tmp = newSVpvn("", 0);
2433 PerlIO_printf(Perl_debug_log, "### %"IVdf":LEX_%s/X%s %s\n",
2434 (IV)CopLINE(PL_curcop),
2435 lex_state_names[PL_lex_state],
2436 exp_name[PL_expect],
2437 pv_display(tmp, s, strlen(s), 0, 60));
2440 /* check if there's an identifier for us to look at */
2441 if (PL_pending_ident)
2442 return REPORT(S_pending_ident(aTHX));
2444 /* no identifier pending identification */
2446 switch (PL_lex_state) {
2448 case LEX_NORMAL: /* Some compilers will produce faster */
2449 case LEX_INTERPNORMAL: /* code if we comment these out. */
2453 /* when we've already built the next token, just pull it out of the queue */
2456 yylval = PL_nextval[PL_nexttoke];
2458 PL_lex_state = PL_lex_defer;
2459 PL_expect = PL_lex_expect;
2460 PL_lex_defer = LEX_NORMAL;
2462 return REPORT(PL_nexttype[PL_nexttoke]);
2464 /* interpolated case modifiers like \L \U, including \Q and \E.
2465 when we get here, PL_bufptr is at the \
2467 case LEX_INTERPCASEMOD:
2469 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2470 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2472 /* handle \E or end of string */
2473 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2475 if (PL_lex_casemods) {
2476 const char oldmod = PL_lex_casestack[--PL_lex_casemods];
2477 PL_lex_casestack[PL_lex_casemods] = '\0';
2479 if (PL_bufptr != PL_bufend
2480 && (oldmod == 'L' || oldmod == 'U' || oldmod == 'Q')) {
2482 PL_lex_state = LEX_INTERPCONCAT;
2486 if (PL_bufptr != PL_bufend)
2488 PL_lex_state = LEX_INTERPCONCAT;
2492 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2493 "### Saw case modifier\n"); });
2495 if (s[1] == '\\' && s[2] == 'E') {
2497 PL_lex_state = LEX_INTERPCONCAT;
2502 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2503 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
2504 if ((*s == 'L' || *s == 'U') &&
2505 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
2506 PL_lex_casestack[--PL_lex_casemods] = '\0';
2509 if (PL_lex_casemods > 10)
2510 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2511 PL_lex_casestack[PL_lex_casemods++] = *s;
2512 PL_lex_casestack[PL_lex_casemods] = '\0';
2513 PL_lex_state = LEX_INTERPCONCAT;
2514 PL_nextval[PL_nexttoke].ival = 0;
2517 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2519 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2521 PL_nextval[PL_nexttoke].ival = OP_LC;
2523 PL_nextval[PL_nexttoke].ival = OP_UC;
2525 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2527 Perl_croak(aTHX_ "panic: yylex");
2531 if (PL_lex_starts) {
2534 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2535 if (PL_lex_casemods == 1 && PL_lex_inpat)
2544 case LEX_INTERPPUSH:
2545 return REPORT(sublex_push());
2547 case LEX_INTERPSTART:
2548 if (PL_bufptr == PL_bufend)
2549 return REPORT(sublex_done());
2550 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2551 "### Interpolated variable\n"); });
2553 PL_lex_dojoin = (*PL_bufptr == '@');
2554 PL_lex_state = LEX_INTERPNORMAL;
2555 if (PL_lex_dojoin) {
2556 PL_nextval[PL_nexttoke].ival = 0;
2558 force_ident("\"", '$');
2559 PL_nextval[PL_nexttoke].ival = 0;
2561 PL_nextval[PL_nexttoke].ival = 0;
2563 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2566 if (PL_lex_starts++) {
2568 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2569 if (!PL_lex_casemods && PL_lex_inpat)
2576 case LEX_INTERPENDMAYBE:
2577 if (intuit_more(PL_bufptr)) {
2578 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2584 if (PL_lex_dojoin) {
2585 PL_lex_dojoin = FALSE;
2586 PL_lex_state = LEX_INTERPCONCAT;
2589 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2590 && SvEVALED(PL_lex_repl))
2592 if (PL_bufptr != PL_bufend)
2593 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2594 PL_lex_repl = Nullsv;
2597 case LEX_INTERPCONCAT:
2599 if (PL_lex_brackets)
2600 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2602 if (PL_bufptr == PL_bufend)
2603 return REPORT(sublex_done());
2605 if (SvIVX(PL_linestr) == '\'') {
2606 SV *sv = newSVsv(PL_linestr);
2609 else if ( PL_hints & HINT_NEW_RE )
2610 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2611 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2615 s = scan_const(PL_bufptr);
2617 PL_lex_state = LEX_INTERPCASEMOD;
2619 PL_lex_state = LEX_INTERPSTART;
2622 if (s != PL_bufptr) {
2623 PL_nextval[PL_nexttoke] = yylval;
2626 if (PL_lex_starts++) {
2627 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2628 if (!PL_lex_casemods && PL_lex_inpat)
2641 PL_lex_state = LEX_NORMAL;
2642 s = scan_formline(PL_bufptr);
2643 if (!PL_lex_formbrack)
2649 PL_oldoldbufptr = PL_oldbufptr;
2655 if (isIDFIRST_lazy_if(s,UTF))
2657 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2660 goto fake_eof; /* emulate EOF on ^D or ^Z */
2665 if (PL_lex_brackets) {
2666 yyerror(PL_lex_formbrack
2667 ? "Format not terminated"
2668 : "Missing right curly or square bracket");
2670 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2671 "### Tokener got EOF\n");
2675 if (s++ < PL_bufend)
2676 goto retry; /* ignore stray nulls */
2679 if (!PL_in_eval && !PL_preambled) {
2680 PL_preambled = TRUE;
2681 sv_setpv(PL_linestr,incl_perldb());
2682 if (SvCUR(PL_linestr))
2683 sv_catpvn(PL_linestr,";", 1);
2685 while(AvFILLp(PL_preambleav) >= 0) {
2686 SV *tmpsv = av_shift(PL_preambleav);
2687 sv_catsv(PL_linestr, tmpsv);
2688 sv_catpvn(PL_linestr, ";", 1);
2691 sv_free((SV*)PL_preambleav);
2692 PL_preambleav = NULL;
2694 if (PL_minus_n || PL_minus_p) {
2695 sv_catpv(PL_linestr, "LINE: while (<>) {");
2697 sv_catpv(PL_linestr,"chomp;");
2700 if ((*PL_splitstr == '/' || *PL_splitstr == '\''
2701 || *PL_splitstr == '"')
2702 && strchr(PL_splitstr + 1, *PL_splitstr))
2703 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
2705 /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
2706 bytes can be used as quoting characters. :-) */
2707 /* The count here deliberately includes the NUL
2708 that terminates the C string constant. This
2709 embeds the opening NUL into the string. */
2710 const char *splits = PL_splitstr;
2711 sv_catpvn(PL_linestr, "our @F=split(q", 15);
2714 if (*splits == '\\')
2715 sv_catpvn(PL_linestr, splits, 1);
2716 sv_catpvn(PL_linestr, splits, 1);
2717 } while (*splits++);
2718 /* This loop will embed the trailing NUL of
2719 PL_linestr as the last thing it does before
2721 sv_catpvn(PL_linestr, ");", 2);
2725 sv_catpv(PL_linestr,"our @F=split(' ');");
2728 sv_catpvn(PL_linestr, "\n", 1);
2729 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2730 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2731 PL_last_lop = PL_last_uni = Nullch;
2732 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2733 SV * const sv = NEWSV(85,0);
2735 sv_upgrade(sv, SVt_PVMG);
2736 sv_setsv(sv,PL_linestr);
2739 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2744 bof = PL_rsfp ? TRUE : FALSE;
2745 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2748 if (PL_preprocess && !PL_in_eval)
2749 (void)PerlProc_pclose(PL_rsfp);
2750 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2751 PerlIO_clearerr(PL_rsfp);
2753 (void)PerlIO_close(PL_rsfp);
2755 PL_doextract = FALSE;
2757 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2758 sv_setpv(PL_linestr,PL_minus_p
2759 ? ";}continue{print;}" : ";}");
2760 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2761 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2762 PL_last_lop = PL_last_uni = Nullch;
2763 PL_minus_n = PL_minus_p = 0;
2766 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2767 PL_last_lop = PL_last_uni = Nullch;
2768 sv_setpvn(PL_linestr,"",0);
2769 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2771 /* If it looks like the start of a BOM or raw UTF-16,
2772 * check if it in fact is. */
2778 #ifdef PERLIO_IS_STDIO
2779 # ifdef __GNU_LIBRARY__
2780 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2781 # define FTELL_FOR_PIPE_IS_BROKEN
2785 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2786 # define FTELL_FOR_PIPE_IS_BROKEN
2791 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2792 /* This loses the possibility to detect the bof
2793 * situation on perl -P when the libc5 is being used.
2794 * Workaround? Maybe attach some extra state to PL_rsfp?
2797 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2799 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
2802 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2803 s = swallow_bom((U8*)s);
2807 /* Incest with pod. */
2808 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2809 sv_setpvn(PL_linestr, "", 0);
2810 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2811 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2812 PL_last_lop = PL_last_uni = Nullch;
2813 PL_doextract = FALSE;
2817 } while (PL_doextract);
2818 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2819 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2820 SV * const sv = NEWSV(85,0);
2822 sv_upgrade(sv, SVt_PVMG);
2823 sv_setsv(sv,PL_linestr);
2826 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2828 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2829 PL_last_lop = PL_last_uni = Nullch;
2830 if (CopLINE(PL_curcop) == 1) {
2831 while (s < PL_bufend && isSPACE(*s))
2833 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2837 if (*s == '#' && *(s+1) == '!')
2839 #ifdef ALTERNATE_SHEBANG
2841 static char const as[] = ALTERNATE_SHEBANG;
2842 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2843 d = s + (sizeof(as) - 1);
2845 #endif /* ALTERNATE_SHEBANG */
2854 while (*d && !isSPACE(*d))
2858 #ifdef ARG_ZERO_IS_SCRIPT
2859 if (ipathend > ipath) {
2861 * HP-UX (at least) sets argv[0] to the script name,
2862 * which makes $^X incorrect. And Digital UNIX and Linux,
2863 * at least, set argv[0] to the basename of the Perl
2864 * interpreter. So, having found "#!", we'll set it right.
2866 SV * const x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); /* $^X */
2867 assert(SvPOK(x) || SvGMAGICAL(x));
2868 if (sv_eq(x, CopFILESV(PL_curcop))) {
2869 sv_setpvn(x, ipath, ipathend - ipath);
2875 const char *bstart = SvPV_const(CopFILESV(PL_curcop),blen);
2876 const char * const lstart = SvPV_const(x,llen);
2878 bstart += blen - llen;
2879 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
2880 sv_setpvn(x, ipath, ipathend - ipath);
2885 TAINT_NOT; /* $^X is always tainted, but that's OK */
2887 #endif /* ARG_ZERO_IS_SCRIPT */
2892 d = instr(s,"perl -");
2894 d = instr(s,"perl");
2896 /* avoid getting into infinite loops when shebang
2897 * line contains "Perl" rather than "perl" */
2899 for (d = ipathend-4; d >= ipath; --d) {
2900 if ((*d == 'p' || *d == 'P')
2901 && !ibcmp(d, "perl", 4))
2911 #ifdef ALTERNATE_SHEBANG
2913 * If the ALTERNATE_SHEBANG on this system starts with a
2914 * character that can be part of a Perl expression, then if
2915 * we see it but not "perl", we're probably looking at the
2916 * start of Perl code, not a request to hand off to some
2917 * other interpreter. Similarly, if "perl" is there, but
2918 * not in the first 'word' of the line, we assume the line
2919 * contains the start of the Perl program.
2921 if (d && *s != '#') {
2922 const char *c = ipath;
2923 while (*c && !strchr("; \t\r\n\f\v#", *c))
2926 d = Nullch; /* "perl" not in first word; ignore */
2928 *s = '#'; /* Don't try to parse shebang line */
2930 #endif /* ALTERNATE_SHEBANG */
2931 #ifndef MACOS_TRADITIONAL
2936 !instr(s,"indir") &&
2937 instr(PL_origargv[0],"perl"))
2944 while (s < PL_bufend && isSPACE(*s))
2946 if (s < PL_bufend) {
2947 Newxz(newargv,PL_origargc+3,char*);
2949 while (s < PL_bufend && !isSPACE(*s))
2952 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2955 newargv = PL_origargv;
2958 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2960 Perl_croak(aTHX_ "Can't exec %s", ipath);
2964 const U32 oldpdb = PL_perldb;
2965 const bool oldn = PL_minus_n;
2966 const bool oldp = PL_minus_p;
2968 while (*d && !isSPACE(*d)) d++;
2969 while (SPACE_OR_TAB(*d)) d++;
2972 const bool switches_done = PL_doswitches;
2974 if (*d == 'M' || *d == 'm' || *d == 'C') {
2975 const char * const m = d;
2976 while (*d && !isSPACE(*d)) d++;
2977 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2980 d = moreswitches(d);
2982 if (PL_doswitches && !switches_done) {
2983 int argc = PL_origargc;
2984 char **argv = PL_origargv;
2987 } while (argc && argv[0][0] == '-' && argv[0][1]);
2988 init_argv_symbols(argc,argv);
2990 if ((PERLDB_LINE && !oldpdb) ||
2991 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2992 /* if we have already added "LINE: while (<>) {",
2993 we must not do it again */
2995 sv_setpvn(PL_linestr, "", 0);
2996 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2997 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2998 PL_last_lop = PL_last_uni = Nullch;
2999 PL_preambled = FALSE;
3001 (void)gv_fetchfile(PL_origfilename);
3004 if (PL_doswitches && !switches_done) {
3005 int argc = PL_origargc;
3006 char **argv = PL_origargv;
3009 } while (argc && argv[0][0] == '-' && argv[0][1]);
3010 init_argv_symbols(argc,argv);
3016 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
3018 PL_lex_state = LEX_FORMLINE;
3023 #ifdef PERL_STRICT_CR
3024 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
3026 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
3028 case ' ': case '\t': case '\f': case 013:
3029 #ifdef MACOS_TRADITIONAL
3036 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
3037 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
3038 /* handle eval qq[#line 1 "foo"\n ...] */
3039 CopLINE_dec(PL_curcop);
3043 while (s < d && *s != '\n')
3047 else if (s > d) /* Found by Ilya: feed random input to Perl. */
3048 Perl_croak(aTHX_ "panic: input overflow");
3050 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
3052 PL_lex_state = LEX_FORMLINE;
3062 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
3070 while (s < PL_bufend && SPACE_OR_TAB(*s))
3073 if (strnEQ(s,"=>",2)) {
3074 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
3075 DEBUG_T( { S_printbuf(aTHX_
3076 "### Saw unary minus before =>, forcing word %s\n", s);
3078 OPERATOR('-'); /* unary minus */
3080 PL_last_uni = PL_oldbufptr;
3082 case 'r': ftst = OP_FTEREAD; break;
3083 case 'w': ftst = OP_FTEWRITE; break;
3084 case 'x': ftst = OP_FTEEXEC; break;
3085 case 'o': ftst = OP_FTEOWNED; break;
3086 case 'R': ftst = OP_FTRREAD; break;
3087 case 'W': ftst = OP_FTRWRITE; break;
3088 case 'X': ftst = OP_FTREXEC; break;
3089 case 'O': ftst = OP_FTROWNED; break;
3090 case 'e': ftst = OP_FTIS; break;
3091 case 'z': ftst = OP_FTZERO; break;
3092 case 's': ftst = OP_FTSIZE; break;
3093 case 'f': ftst = OP_FTFILE; break;
3094 case 'd': ftst = OP_FTDIR; break;
3095 case 'l': ftst = OP_FTLINK; break;
3096 case 'p': ftst = OP_FTPIPE; break;
3097 case 'S': ftst = OP_FTSOCK; break;
3098 case 'u': ftst = OP_FTSUID; break;
3099 case 'g': ftst = OP_FTSGID; break;
3100 case 'k': ftst = OP_FTSVTX; break;
3101 case 'b': ftst = OP_FTBLK; break;
3102 case 'c': ftst = OP_FTCHR; break;
3103 case 't': ftst = OP_FTTTY; break;
3104 case 'T': ftst = OP_FTTEXT; break;
3105 case 'B': ftst = OP_FTBINARY; break;
3106 case 'M': case 'A': case 'C':
3107 gv_fetchpv("\024",TRUE, SVt_PV);
3109 case 'M': ftst = OP_FTMTIME; break;
3110 case 'A': ftst = OP_FTATIME; break;
3111 case 'C': ftst = OP_FTCTIME; break;
3119 PL_last_lop_op = (OPCODE)ftst;
3120 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3121 "### Saw file test %c\n", (int)tmp);
3126 /* Assume it was a minus followed by a one-letter named
3127 * subroutine call (or a -bareword), then. */
3128 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3129 "### '-%c' looked like a file test but was not\n",
3136 const char tmp = *s++;
3139 if (PL_expect == XOPERATOR)
3144 else if (*s == '>') {
3147 if (isIDFIRST_lazy_if(s,UTF)) {
3148 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
3156 if (PL_expect == XOPERATOR)
3159 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
3161 OPERATOR('-'); /* unary minus */
3167 const char tmp = *s++;
3170 if (PL_expect == XOPERATOR)
3175 if (PL_expect == XOPERATOR)
3178 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
3185 if (PL_expect != XOPERATOR) {
3186 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3187 PL_expect = XOPERATOR;
3188 force_ident(PL_tokenbuf, '*');
3201 if (PL_expect == XOPERATOR) {
3205 PL_tokenbuf[0] = '%';
3206 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
3207 if (!PL_tokenbuf[1]) {
3210 PL_pending_ident = '%';
3221 && (PL_expect == XOPERATOR || PL_expect == XTERMORDORDOR)
3222 && FEATURE_IS_ENABLED("~~", 2))
3229 const char tmp = *s++;
3235 goto just_a_word_zero_gv;
3238 switch (PL_expect) {
3241 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
3243 PL_bufptr = s; /* update in case we back off */
3249 PL_expect = XTERMBLOCK;
3253 while (isIDFIRST_lazy_if(s,UTF)) {
3255 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3256 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
3257 if (tmp < 0) tmp = -tmp;
3273 d = scan_str(d,TRUE,TRUE);
3275 /* MUST advance bufptr here to avoid bogus
3276 "at end of line" context messages from yyerror().
3278 PL_bufptr = s + len;
3279 yyerror("Unterminated attribute parameter in attribute list");
3282 return REPORT(0); /* EOF indicator */
3286 SV *sv = newSVpvn(s, len);
3287 sv_catsv(sv, PL_lex_stuff);
3288 attrs = append_elem(OP_LIST, attrs,
3289 newSVOP(OP_CONST, 0, sv));
3290 SvREFCNT_dec(PL_lex_stuff);
3291 PL_lex_stuff = Nullsv;
3294 if (len == 6 && strnEQ(s, "unique", len)) {
3295 if (PL_in_my == KEY_our)
3297 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
3299 ; /* skip to avoid loading attributes.pm */
3302 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
3305 /* NOTE: any CV attrs applied here need to be part of
3306 the CVf_BUILTIN_ATTRS define in cv.h! */
3307 else if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3308 CvLVALUE_on(PL_compcv);
3309 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3310 CvLOCKED_on(PL_compcv);
3311 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3312 CvMETHOD_on(PL_compcv);
3313 else if (!PL_in_my && len == 9 && strnEQ(s, "assertion", len))
3314 CvASSERTION_on(PL_compcv);
3315 /* After we've set the flags, it could be argued that
3316 we don't need to do the attributes.pm-based setting
3317 process, and shouldn't bother appending recognized
3318 flags. To experiment with that, uncomment the
3319 following "else". (Note that's already been
3320 uncommented. That keeps the above-applied built-in
3321 attributes from being intercepted (and possibly
3322 rejected) by a package's attribute routines, but is
3323 justified by the performance win for the common case
3324 of applying only built-in attributes.) */
3326 attrs = append_elem(OP_LIST, attrs,
3327 newSVOP(OP_CONST, 0,
3331 if (*s == ':' && s[1] != ':')
3334 break; /* require real whitespace or :'s */
3338 = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3339 if (*s != ';' && *s != '}' && *s != tmp
3340 && (tmp != '=' || *s != ')')) {
3341 const char q = ((*s == '\'') ? '"' : '\'');
3342 /* If here for an expression, and parsed no attrs, back
3344 if (tmp == '=' && !attrs) {
3348 /* MUST advance bufptr here to avoid bogus "at end of line"
3349 context messages from yyerror().
3353 ? Perl_form(aTHX_ "Invalid separator character "
3354 "%c%c%c in attribute list", q, *s, q)
3355 : "Unterminated attribute list" );
3363 PL_nextval[PL_nexttoke].opval = attrs;
3371 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3372 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3380 const char tmp = *s++;
3385 const char tmp = *s++;
3393 if (PL_lex_brackets <= 0)
3394 yyerror("Unmatched right square bracket");
3397 if (PL_lex_state == LEX_INTERPNORMAL) {
3398 if (PL_lex_brackets == 0) {
3399 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3400 PL_lex_state = LEX_INTERPEND;
3407 if (PL_lex_brackets > 100) {
3408 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
3410 switch (PL_expect) {
3412 if (PL_lex_formbrack) {
3416 if (PL_oldoldbufptr == PL_last_lop)
3417 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3419 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3420 OPERATOR(HASHBRACK);
3422 while (s < PL_bufend && SPACE_OR_TAB(*s))
3425 PL_tokenbuf[0] = '\0';
3426 if (d < PL_bufend && *d == '-') {
3427 PL_tokenbuf[0] = '-';
3429 while (d < PL_bufend && SPACE_OR_TAB(*d))
3432 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3433 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3435 while (d < PL_bufend && SPACE_OR_TAB(*d))
3438 const char minus = (PL_tokenbuf[0] == '-');
3439 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3447 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3452 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3457 if (PL_oldoldbufptr == PL_last_lop)
3458 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3460 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3463 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3465 /* This hack is to get the ${} in the message. */
3467 yyerror("syntax error");
3470 OPERATOR(HASHBRACK);
3472 /* This hack serves to disambiguate a pair of curlies
3473 * as being a block or an anon hash. Normally, expectation
3474 * determines that, but in cases where we're not in a
3475 * position to expect anything in particular (like inside
3476 * eval"") we have to resolve the ambiguity. This code
3477 * covers the case where the first term in the curlies is a
3478 * quoted string. Most other cases need to be explicitly
3479 * disambiguated by prepending a "+" before the opening
3480 * curly in order to force resolution as an anon hash.
3482 * XXX should probably propagate the outer expectation
3483 * into eval"" to rely less on this hack, but that could
3484 * potentially break current behavior of eval"".
3488 if (*s == '\'' || *s == '"' || *s == '`') {
3489 /* common case: get past first string, handling escapes */
3490 for (t++; t < PL_bufend && *t != *s;)
3491 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3495 else if (*s == 'q') {
3498 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3501 /* skip q//-like construct */
3503 char open, close, term;
3506 while (t < PL_bufend && isSPACE(*t))
3508 /* check for q => */
3509 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
3510 OPERATOR(HASHBRACK);
3514 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3518 for (t++; t < PL_bufend; t++) {
3519 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3521 else if (*t == open)
3525 for (t++; t < PL_bufend; t++) {
3526 if (*t == '\\' && t+1 < PL_bufend)
3528 else if (*t == close && --brackets <= 0)
3530 else if (*t == open)
3537 /* skip plain q word */
3538 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3541 else if (isALNUM_lazy_if(t,UTF)) {
3543 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3546 while (t < PL_bufend && isSPACE(*t))
3548 /* if comma follows first term, call it an anon hash */
3549 /* XXX it could be a comma expression with loop modifiers */
3550 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3551 || (*t == '=' && t[1] == '>')))
3552 OPERATOR(HASHBRACK);
3553 if (PL_expect == XREF)
3556 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3562 yylval.ival = CopLINE(PL_curcop);
3563 if (isSPACE(*s) || *s == '#')
3564 PL_copline = NOLINE; /* invalidate current command line number */
3569 if (PL_lex_brackets <= 0)
3570 yyerror("Unmatched right curly bracket");
3572 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3573 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3574 PL_lex_formbrack = 0;
3575 if (PL_lex_state == LEX_INTERPNORMAL) {
3576 if (PL_lex_brackets == 0) {
3577 if (PL_expect & XFAKEBRACK) {
3578 PL_expect &= XENUMMASK;
3579 PL_lex_state = LEX_INTERPEND;
3581 return yylex(); /* ignore fake brackets */
3583 if (*s == '-' && s[1] == '>')
3584 PL_lex_state = LEX_INTERPENDMAYBE;
3585 else if (*s != '[' && *s != '{')
3586 PL_lex_state = LEX_INTERPEND;
3589 if (PL_expect & XFAKEBRACK) {
3590 PL_expect &= XENUMMASK;
3592 return yylex(); /* ignore fake brackets */
3601 if (PL_expect == XOPERATOR) {
3602 if (PL_bufptr == PL_linestart && ckWARN(WARN_SEMICOLON)
3603 && isIDFIRST_lazy_if(s,UTF))
3605 CopLINE_dec(PL_curcop);
3606 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3607 CopLINE_inc(PL_curcop);
3612 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3614 PL_expect = XOPERATOR;
3615 force_ident(PL_tokenbuf, '&');
3619 yylval.ival = (OPpENTERSUB_AMPER<<8);
3631 const char tmp = *s++;
3638 if (tmp && isSPACE(*s) && ckWARN(WARN_SYNTAX)
3639 && strchr("+-*/%.^&|<",tmp))
3640 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3641 "Reversed %c= operator",(int)tmp);
3643 if (PL_expect == XSTATE && isALPHA(tmp) &&
3644 (s == PL_linestart+1 || s[-2] == '\n') )
3646 if (PL_in_eval && !PL_rsfp) {
3651 if (strnEQ(s,"=cut",4)) {
3665 PL_doextract = TRUE;
3669 if (PL_lex_brackets < PL_lex_formbrack) {
3671 #ifdef PERL_STRICT_CR
3672 for (t = s; SPACE_OR_TAB(*t); t++) ;
3674 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3676 if (*t == '\n' || *t == '#') {
3687 const char tmp = *s++;
3689 /* was this !=~ where !~ was meant?
3690 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
3692 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
3693 const char *t = s+1;
3695 while (t < PL_bufend && isSPACE(*t))
3698 if (*t == '/' || *t == '?' ||
3699 ((*t == 'm' || *t == 's' || *t == 'y')
3700 && !isALNUM(t[1])) ||
3701 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
3702 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3703 "!=~ should be !~");
3713 if (PL_expect != XOPERATOR) {
3714 if (s[1] != '<' && !strchr(s,'>'))
3717 s = scan_heredoc(s);
3719 s = scan_inputsymbol(s);
3720 TERM(sublex_start());
3726 SHop(OP_LEFT_SHIFT);
3740 const char tmp = *s++;
3742 SHop(OP_RIGHT_SHIFT);
3752 if (PL_expect == XOPERATOR) {
3753 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3756 return REPORT(','); /* grandfather non-comma-format format */
3760 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3761 PL_tokenbuf[0] = '@';
3762 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3763 sizeof PL_tokenbuf - 1, FALSE);
3764 if (PL_expect == XOPERATOR)
3765 no_op("Array length", s);
3766 if (!PL_tokenbuf[1])
3768 PL_expect = XOPERATOR;
3769 PL_pending_ident = '#';
3773 PL_tokenbuf[0] = '$';
3774 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3775 sizeof PL_tokenbuf - 1, FALSE);
3776 if (PL_expect == XOPERATOR)
3778 if (!PL_tokenbuf[1]) {
3780 yyerror("Final $ should be \\$ or $name");
3784 /* This kludge not intended to be bulletproof. */
3785 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3786 yylval.opval = newSVOP(OP_CONST, 0,
3787 newSViv(PL_compiling.cop_arybase));
3788 yylval.opval->op_private = OPpCONST_ARYBASE;
3794 const char tmp = *s;
3795 if (PL_lex_state == LEX_NORMAL)
3798 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop)
3799 && intuit_more(s)) {
3801 PL_tokenbuf[0] = '@';
3802 if (ckWARN(WARN_SYNTAX)) {
3805 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3808 PL_bufptr = skipspace(PL_bufptr);
3809 while (t < PL_bufend && *t != ']')
3811 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3812 "Multidimensional syntax %.*s not supported",
3813 (t - PL_bufptr) + 1, PL_bufptr);
3817 else if (*s == '{') {
3819 PL_tokenbuf[0] = '%';
3820 if (strEQ(PL_tokenbuf+1, "SIG") && ckWARN(WARN_SYNTAX)
3821 && (t = strchr(s, '}')) && (t = strchr(t, '=')))
3823 char tmpbuf[sizeof PL_tokenbuf];
3824 for (t++; isSPACE(*t); t++) ;
3825 if (isIDFIRST_lazy_if(t,UTF)) {
3827 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE,
3829 for (; isSPACE(*t); t++) ;
3830 if (*t == ';' && get_cv(tmpbuf, FALSE))
3831 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3832 "You need to quote \"%s\"",
3839 PL_expect = XOPERATOR;
3840 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3841 const bool islop = (PL_last_lop == PL_oldoldbufptr);
3842 if (!islop || PL_last_lop_op == OP_GREPSTART)
3843 PL_expect = XOPERATOR;
3844 else if (strchr("$@\"'`q", *s))
3845 PL_expect = XTERM; /* e.g. print $fh "foo" */
3846 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3847 PL_expect = XTERM; /* e.g. print $fh &sub */
3848 else if (isIDFIRST_lazy_if(s,UTF)) {
3849 char tmpbuf[sizeof PL_tokenbuf];
3851 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3852 if ((t2 = keyword(tmpbuf, len))) {
3853 /* binary operators exclude handle interpretations */
3865 PL_expect = XTERM; /* e.g. print $fh length() */
3870 PL_expect = XTERM; /* e.g. print $fh subr() */
3873 else if (isDIGIT(*s))
3874 PL_expect = XTERM; /* e.g. print $fh 3 */
3875 else if (*s == '.' && isDIGIT(s[1]))
3876 PL_expect = XTERM; /* e.g. print $fh .3 */
3877 else if ((*s == '?' || *s == '-' || *s == '+')
3878 && !isSPACE(s[1]) && s[1] != '=')
3879 PL_expect = XTERM; /* e.g. print $fh -1 */
3880 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '='
3882 PL_expect = XTERM; /* e.g. print $fh /.../
3883 XXX except DORDOR operator
3885 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2])
3887 PL_expect = XTERM; /* print $fh <<"EOF" */
3890 PL_pending_ident = '$';
3894 if (PL_expect == XOPERATOR)
3896 PL_tokenbuf[0] = '@';
3897 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3898 if (!PL_tokenbuf[1]) {
3901 if (PL_lex_state == LEX_NORMAL)
3903 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3905 PL_tokenbuf[0] = '%';
3907 /* Warn about @ where they meant $. */
3908 if (*s == '[' || *s == '{') {
3909 if (ckWARN(WARN_SYNTAX)) {
3910 const char *t = s + 1;
3911 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3913 if (*t == '}' || *t == ']') {
3915 PL_bufptr = skipspace(PL_bufptr);
3916 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3917 "Scalar value %.*s better written as $%.*s",
3918 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3923 PL_pending_ident = '@';
3926 case '/': /* may be division, defined-or, or pattern */
3927 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
3931 case '?': /* may either be conditional or pattern */
3932 if(PL_expect == XOPERATOR) {
3940 /* A // operator. */
3950 /* Disable warning on "study /blah/" */
3951 if (PL_oldoldbufptr == PL_last_uni
3952 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3953 || memNE(PL_last_uni, "study", 5)
3954 || isALNUM_lazy_if(PL_last_uni+5,UTF)
3957 s = scan_pat(s,OP_MATCH);
3958 TERM(sublex_start());
3962 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3963 #ifdef PERL_STRICT_CR
3966 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3968 && (s == PL_linestart || s[-1] == '\n') )
3970 PL_lex_formbrack = 0;
3974 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3980 yylval.ival = OPf_SPECIAL;
3986 if (PL_expect != XOPERATOR)
3991 case '0': case '1': case '2': case '3': case '4':
3992 case '5': case '6': case '7': case '8': case '9':
3993 s = scan_num(s, &yylval);
3994 DEBUG_T( { S_printbuf(aTHX_ "### Saw number in %s\n", s); } );
3995 if (PL_expect == XOPERATOR)
4000 s = scan_str(s,FALSE,FALSE);
4001 DEBUG_T( { S_printbuf(aTHX_ "### Saw string before %s\n", s); } );
4002 if (PL_expect == XOPERATOR) {
4003 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
4006 return REPORT(','); /* grandfather non-comma-format format */
4012 missingterm((char*)0);
4013 yylval.ival = OP_CONST;
4014 TERM(sublex_start());
4017 s = scan_str(s,FALSE,FALSE);
4018 DEBUG_T( { S_printbuf(aTHX_ "### Saw string before %s\n", s); } );
4019 if (PL_expect == XOPERATOR) {
4020 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
4023 return REPORT(','); /* grandfather non-comma-format format */
4029 missingterm((char*)0);
4030 yylval.ival = OP_CONST;
4031 /* FIXME. I think that this can be const if char *d is replaced by
4032 more localised variables. */
4033 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
4034 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
4035 yylval.ival = OP_STRINGIFY;
4039 TERM(sublex_start());
4042 s = scan_str(s,FALSE,FALSE);
4043 DEBUG_T( { S_printbuf(aTHX_ "### Saw backtick string before %s\n", s); } );
4044 if (PL_expect == XOPERATOR)
4045 no_op("Backticks",s);
4047 missingterm((char*)0);
4048 yylval.ival = OP_BACKTICK;
4050 TERM(sublex_start());
4054 if (PL_lex_inwhat && isDIGIT(*s) && ckWARN(WARN_SYNTAX))
4055 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
4057 if (PL_expect == XOPERATOR)
4058 no_op("Backslash",s);
4062 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
4063 char *start = s + 2;
4064 while (isDIGIT(*start) || *start == '_')
4066 if (*start == '.' && isDIGIT(start[1])) {
4067 s = scan_num(s, &yylval);
4070 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
4071 else if (!isALPHA(*start) && (PL_expect == XTERM
4072 || PL_expect == XREF || PL_expect == XSTATE
4073 || PL_expect == XTERMORDORDOR)) {
4074 const char c = *start;
4077 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
4080 s = scan_num(s, &yylval);
4087 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
4123 I32 orig_keyword = 0;
4128 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4130 /* Some keywords can be followed by any delimiter, including ':' */
4131 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
4132 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
4133 (PL_tokenbuf[0] == 'q' &&
4134 strchr("qwxr", PL_tokenbuf[1])))));
4136 /* x::* is just a word, unless x is "CORE" */
4137 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
4141 while (d < PL_bufend && isSPACE(*d))
4142 d++; /* no comments skipped here, or s### is misparsed */
4144 /* Is this a label? */
4145 if (!tmp && PL_expect == XSTATE
4146 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
4148 yylval.pval = savepv(PL_tokenbuf);
4153 /* Check for keywords */
4154 tmp = keyword(PL_tokenbuf, len);
4156 /* Is this a word before a => operator? */
4157 if (*d == '=' && d[1] == '>') {
4160 = (OP*)newSVOP(OP_CONST, 0,
4161 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
4162 yylval.opval->op_private = OPpCONST_BARE;
4166 if (tmp < 0) { /* second-class keyword? */
4167 GV *ogv = Nullgv; /* override (winner) */
4168 GV *hgv = Nullgv; /* hidden (loser) */
4169 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
4171 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
4174 if (GvIMPORTED_CV(gv))
4176 else if (! CvMETHOD(cv))
4180 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
4181 (gv = *gvp) != (GV*)&PL_sv_undef &&
4182 GvCVu(gv) && GvIMPORTED_CV(gv))
4189 tmp = 0; /* overridden by import or by GLOBAL */
4192 && -tmp==KEY_lock /* XXX generalizable kludge */
4194 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
4196 tmp = 0; /* any sub overrides "weak" keyword */
4201 && PL_expect != XOPERATOR
4202 && PL_expect != XTERMORDORDOR)
4204 /* any sub overrides the "err" keyword, except when really an
4205 * operator is expected */
4208 else { /* no override */
4210 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
4211 Perl_warner(aTHX_ packWARN(WARN_MISC),
4212 "dump() better written as CORE::dump()");
4216 if (hgv && tmp != KEY_x && tmp != KEY_CORE
4217 && ckWARN(WARN_AMBIGUOUS)) /* never ambiguous */
4218 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4219 "Ambiguous call resolved as CORE::%s(), %s",
4220 GvENAME(hgv), "qualify as such or use &");
4227 default: /* not a keyword */
4228 /* Trade off - by using this evil construction we can pull the
4229 variable gv into the block labelled keylookup. If not, then
4230 we have to give it function scope so that the goto from the
4231 earlier ':' case doesn't bypass the initialisation. */
4233 just_a_word_zero_gv:
4240 const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
4242 /* Get the rest if it looks like a package qualifier */
4244 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
4246 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
4249 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
4250 *s == '\'' ? "'" : "::");
4255 if (PL_expect == XOPERATOR) {
4256 if (PL_bufptr == PL_linestart) {
4257 CopLINE_dec(PL_curcop);
4258 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
4259 CopLINE_inc(PL_curcop);
4262 no_op("Bareword",s);
4265 /* Look for a subroutine with this name in current package,
4266 unless name is "Foo::", in which case Foo is a bearword
4267 (and a package name). */
4270 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
4272 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
4273 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
4274 "Bareword \"%s\" refers to nonexistent package",
4277 PL_tokenbuf[len] = '\0';
4284 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
4287 /* if we saw a global override before, get the right name */
4290 sv = newSVpvn("CORE::GLOBAL::",14);
4291 sv_catpv(sv,PL_tokenbuf);
4294 /* If len is 0, newSVpv does strlen(), which is correct.
4295 If len is non-zero, then it will be the true length,
4296 and so the scalar will be created correctly. */
4297 sv = newSVpv(PL_tokenbuf,len);
4300 /* Presume this is going to be a bareword of some sort. */
4303 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
4304 yylval.opval->op_private = OPpCONST_BARE;
4305 /* UTF-8 package name? */
4306 if (UTF && !IN_BYTES &&
4307 is_utf8_string((U8*)SvPVX_const(sv), SvCUR(sv)))
4310 /* And if "Foo::", then that's what it certainly is. */
4315 /* See if it's the indirect object for a list operator. */
4317 if (PL_oldoldbufptr &&
4318 PL_oldoldbufptr < PL_bufptr &&
4319 (PL_oldoldbufptr == PL_last_lop
4320 || PL_oldoldbufptr == PL_last_uni) &&
4321 /* NO SKIPSPACE BEFORE HERE! */
4322 (PL_expect == XREF ||
4323 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
4325 bool immediate_paren = *s == '(';
4327 /* (Now we can afford to cross potential line boundary.) */
4330 /* Two barewords in a row may indicate method call. */
4332 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
4335 /* If not a declared subroutine, it's an indirect object. */
4336 /* (But it's an indir obj regardless for sort.) */
4337 /* Also, if "_" follows a filetest operator, it's a bareword */
4340 ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
4341 ((!gv || !GvCVu(gv)) &&
4342 (PL_last_lop_op != OP_MAPSTART &&
4343 PL_last_lop_op != OP_GREPSTART))))
4344 || (PL_tokenbuf[0] == '_' && PL_tokenbuf[1] == '\0'
4345 && ((PL_opargs[PL_last_lop_op] & OA_CLASS_MASK) == OA_FILESTATOP))
4348 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
4353 PL_expect = XOPERATOR;
4356 /* Is this a word before a => operator? */
4357 if (*s == '=' && s[1] == '>' && !pkgname) {
4359 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
4360 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
4361 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
4365 /* If followed by a paren, it's certainly a subroutine. */
4368 if (gv && GvCVu(gv)) {
4369 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
4370 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
4375 PL_nextval[PL_nexttoke].opval = yylval.opval;
4376 PL_expect = XOPERATOR;
4382 /* If followed by var or block, call it a method (unless sub) */
4384 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
4385 PL_last_lop = PL_oldbufptr;
4386 PL_last_lop_op = OP_METHOD;
4390 /* If followed by a bareword, see if it looks like indir obj. */
4393 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
4394 && (tmp = intuit_method(s,gv)))
4397 /* Not a method, so call it a subroutine (if defined) */
4399 if (gv && GvCVu(gv)) {
4401 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4402 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4403 "Ambiguous use of -%s resolved as -&%s()",
4404 PL_tokenbuf, PL_tokenbuf);
4405 /* Check for a constant sub */
4407 if ((sv = cv_const_sv(cv))) {
4409 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4410 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4411 yylval.opval->op_private = 0;
4415 /* Resolve to GV now. */
4416 op_free(yylval.opval);
4417 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4418 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4419 PL_last_lop = PL_oldbufptr;
4420 PL_last_lop_op = OP_ENTERSUB;
4421 /* Is there a prototype? */
4424 const char *proto = SvPV_const((SV*)cv, len);
4427 if (*proto == '$' && proto[1] == '\0')
4429 while (*proto == ';')
4431 if (*proto == '&' && *s == '{') {
4432 sv_setpv(PL_subname, PL_curstash ?
4433 "__ANON__" : "__ANON__::__ANON__");
4437 PL_nextval[PL_nexttoke].opval = yylval.opval;
4443 /* Call it a bare word */
4445 if (PL_hints & HINT_STRICT_SUBS)
4446 yylval.opval->op_private |= OPpCONST_STRICT;
4449 if (lastchar != '-') {
4450 if (ckWARN(WARN_RESERVED)) {
4451 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4452 if (!*d && !gv_stashpv(PL_tokenbuf,FALSE))
4453 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
4460 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')
4461 && ckWARN_d(WARN_AMBIGUOUS)) {
4462 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4463 "Operator or semicolon missing before %c%s",
4464 lastchar, PL_tokenbuf);
4465 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4466 "Ambiguous use of %c resolved as operator %c",
4467 lastchar, lastchar);
4473 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4474 newSVpv(CopFILE(PL_curcop),0));
4478 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4479 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4482 case KEY___PACKAGE__:
4483 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4485 ? newSVhek(HvNAME_HEK(PL_curstash))
4492 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4493 const char *pname = "main";
4494 if (PL_tokenbuf[2] == 'D')
4495 pname = HvNAME_get(PL_curstash ? PL_curstash : PL_defstash);
4496 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4499 GvIOp(gv) = newIO();
4500 IoIFP(GvIOp(gv)) = PL_rsfp;
4501 #if defined(HAS_FCNTL) && defined(F_SETFD)
4503 const int fd = PerlIO_fileno(PL_rsfp);
4504 fcntl(fd,F_SETFD,fd >= 3);
4507 /* Mark this internal pseudo-handle as clean */
4508 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4510 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4511 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4512 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4514 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4515 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4516 /* if the script was opened in binmode, we need to revert
4517 * it to text mode for compatibility; but only iff it has CRs
4518 * XXX this is a questionable hack at best. */
4519 if (PL_bufend-PL_bufptr > 2
4520 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4523 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4524 loc = PerlIO_tell(PL_rsfp);
4525 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4528 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4530 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4531 #endif /* NETWARE */
4532 #ifdef PERLIO_IS_STDIO /* really? */
4533 # if defined(__BORLANDC__)
4534 /* XXX see note in do_binmode() */
4535 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4539 PerlIO_seek(PL_rsfp, loc, 0);
4543 #ifdef PERLIO_LAYERS
4546 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4547 else if (PL_encoding) {
4554 XPUSHs(PL_encoding);
4556 call_method("name", G_SCALAR);
4560 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
4561 Perl_form(aTHX_ ":encoding(%"SVf")",
4579 if (PL_expect == XSTATE) {
4586 if (*s == ':' && s[1] == ':') {
4589 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4590 if (!(tmp = keyword(PL_tokenbuf, len)))
4591 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4594 else if (tmp == KEY_require || tmp == KEY_do)
4595 /* that's a way to remember we saw "CORE::" */
4608 LOP(OP_ACCEPT,XTERM);
4614 LOP(OP_ATAN2,XTERM);
4620 LOP(OP_BINMODE,XTERM);
4623 LOP(OP_BLESS,XTERM);
4632 /* When 'use switch' is in effect, continue has a dual
4633 life as a control operator. */
4635 if (!FEATURE_IS_ENABLED("switch", 6))
4638 /* We have to disambiguate the two senses of
4639 "continue". If the next token is a '{' then
4640 treat it as the start of a continue block;
4641 otherwise treat it as a control operator.
4652 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4669 if (!PL_cryptseen) {
4670 PL_cryptseen = TRUE;
4674 LOP(OP_CRYPT,XTERM);
4677 LOP(OP_CHMOD,XTERM);
4680 LOP(OP_CHOWN,XTERM);
4683 LOP(OP_CONNECT,XTERM);
4702 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4703 if (orig_keyword == KEY_do) {
4712 PL_hints |= HINT_BLOCK_SCOPE;
4722 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4723 LOP(OP_DBMOPEN,XTERM);
4729 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4736 yylval.ival = CopLINE(PL_curcop);
4750 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4751 UNIBRACK(OP_ENTEREVAL);
4769 case KEY_endhostent:
4775 case KEY_endservent:
4778 case KEY_endprotoent:
4789 yylval.ival = CopLINE(PL_curcop);
4791 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4793 if ((PL_bufend - p) >= 3 &&
4794 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4796 else if ((PL_bufend - p) >= 4 &&
4797 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4800 if (isIDFIRST_lazy_if(p,UTF)) {
4801 p = scan_ident(p, PL_bufend,
4802 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4806 Perl_croak(aTHX_ "Missing $ on loop variable");
4811 LOP(OP_FORMLINE,XTERM);
4817 LOP(OP_FCNTL,XTERM);
4823 LOP(OP_FLOCK,XTERM);
4832 LOP(OP_GREPSTART, XREF);
4835 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4850 case KEY_getpriority:
4851 LOP(OP_GETPRIORITY,XTERM);
4853 case KEY_getprotobyname:
4856 case KEY_getprotobynumber:
4857 LOP(OP_GPBYNUMBER,XTERM);
4859 case KEY_getprotoent:
4871 case KEY_getpeername:
4872 UNI(OP_GETPEERNAME);
4874 case KEY_gethostbyname:
4877 case KEY_gethostbyaddr:
4878 LOP(OP_GHBYADDR,XTERM);
4880 case KEY_gethostent:
4883 case KEY_getnetbyname:
4886 case KEY_getnetbyaddr:
4887 LOP(OP_GNBYADDR,XTERM);
4892 case KEY_getservbyname:
4893 LOP(OP_GSBYNAME,XTERM);
4895 case KEY_getservbyport:
4896 LOP(OP_GSBYPORT,XTERM);
4898 case KEY_getservent:
4901 case KEY_getsockname:
4902 UNI(OP_GETSOCKNAME);
4904 case KEY_getsockopt:
4905 LOP(OP_GSOCKOPT,XTERM);
4920 yylval.ival = CopLINE(PL_curcop);
4931 yylval.ival = CopLINE(PL_curcop);
4935 LOP(OP_INDEX,XTERM);
4941 LOP(OP_IOCTL,XTERM);
4953 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4985 LOP(OP_LISTEN,XTERM);
4994 s = scan_pat(s,OP_MATCH);
4995 TERM(sublex_start());
4998 LOP(OP_MAPSTART, XREF);
5001 LOP(OP_MKDIR,XTERM);
5004 LOP(OP_MSGCTL,XTERM);
5007 LOP(OP_MSGGET,XTERM);
5010 LOP(OP_MSGRCV,XTERM);
5013 LOP(OP_MSGSND,XTERM);
5019 if (isIDFIRST_lazy_if(s,UTF)) {
5020 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
5021 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
5023 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
5024 if (!PL_in_my_stash) {
5027 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
5035 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5042 s = tokenize_use(0, s);
5046 if (*s == '(' || (s = skipspace(s), *s == '('))
5053 if (isIDFIRST_lazy_if(s,UTF)) {
5055 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
5056 for (t=d; *t && isSPACE(*t); t++) ;
5057 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
5059 && !(t[0] == '=' && t[1] == '>')
5061 int len = (int)(d-s);
5062 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
5063 "Precedence problem: open %.*s should be open(%.*s)",
5070 yylval.ival = OP_OR;
5080 LOP(OP_OPEN_DIR,XTERM);
5083 checkcomma(s,PL_tokenbuf,"filehandle");
5087 checkcomma(s,PL_tokenbuf,"filehandle");
5106 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5110 LOP(OP_PIPE_OP,XTERM);
5113 s = scan_str(s,FALSE,FALSE);
5115 missingterm((char*)0);
5116 yylval.ival = OP_CONST;
5117 TERM(sublex_start());
5123 s = scan_str(s,FALSE,FALSE);
5125 missingterm((char*)0);
5126 PL_expect = XOPERATOR;
5128 if (SvCUR(PL_lex_stuff)) {
5131 d = SvPV_force(PL_lex_stuff, len);
5134 for (; isSPACE(*d) && len; --len, ++d) ;
5137 if (!warned && ckWARN(WARN_QW)) {
5138 for (; !isSPACE(*d) && len; --len, ++d) {
5140 Perl_warner(aTHX_ packWARN(WARN_QW),
5141 "Possible attempt to separate words with commas");
5144 else if (*d == '#') {
5145 Perl_warner(aTHX_ packWARN(WARN_QW),
5146 "Possible attempt to put comments in qw() list");
5152 for (; !isSPACE(*d) && len; --len, ++d) ;
5154 sv = newSVpvn(b, d-b);
5155 if (DO_UTF8(PL_lex_stuff))
5157 words = append_elem(OP_LIST, words,
5158 newSVOP(OP_CONST, 0, tokeq(sv)));
5162 PL_nextval[PL_nexttoke].opval = words;
5167 SvREFCNT_dec(PL_lex_stuff);
5168 PL_lex_stuff = Nullsv;
5174 s = scan_str(s,FALSE,FALSE);
5176 missingterm((char*)0);
5177 yylval.ival = OP_STRINGIFY;
5178 if (SvIVX(PL_lex_stuff) == '\'')
5179 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
5180 TERM(sublex_start());
5183 s = scan_pat(s,OP_QR);
5184 TERM(sublex_start());
5187 s = scan_str(s,FALSE,FALSE);
5189 missingterm((char*)0);
5190 yylval.ival = OP_BACKTICK;
5192 TERM(sublex_start());
5200 s = force_version(s, FALSE);
5202 else if (*s != 'v' || !isDIGIT(s[1])
5203 || (s = force_version(s, TRUE), *s == 'v'))
5205 *PL_tokenbuf = '\0';
5206 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5207 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
5208 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
5210 yyerror("<> should be quotes");
5212 if (orig_keyword == KEY_require) {
5220 PL_last_uni = PL_oldbufptr;
5221 PL_last_lop_op = OP_REQUIRE;
5223 return REPORT( (int)REQUIRE );
5229 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5233 LOP(OP_RENAME,XTERM);
5242 LOP(OP_RINDEX,XTERM);
5252 UNIDOR(OP_READLINE);
5265 LOP(OP_REVERSE,XTERM);
5268 UNIDOR(OP_READLINK);
5276 TERM(sublex_start());
5278 TOKEN(1); /* force error */
5281 checkcomma(s,PL_tokenbuf,"filehandle");
5291 LOP(OP_SELECT,XTERM);
5297 LOP(OP_SEMCTL,XTERM);
5300 LOP(OP_SEMGET,XTERM);
5303 LOP(OP_SEMOP,XTERM);
5309 LOP(OP_SETPGRP,XTERM);
5311 case KEY_setpriority:
5312 LOP(OP_SETPRIORITY,XTERM);
5314 case KEY_sethostent:
5320 case KEY_setservent:
5323 case KEY_setprotoent:
5333 LOP(OP_SEEKDIR,XTERM);
5335 case KEY_setsockopt:
5336 LOP(OP_SSOCKOPT,XTERM);
5342 LOP(OP_SHMCTL,XTERM);
5345 LOP(OP_SHMGET,XTERM);
5348 LOP(OP_SHMREAD,XTERM);
5351 LOP(OP_SHMWRITE,XTERM);
5354 LOP(OP_SHUTDOWN,XTERM);
5363 LOP(OP_SOCKET,XTERM);
5365 case KEY_socketpair:
5366 LOP(OP_SOCKPAIR,XTERM);
5369 checkcomma(s,PL_tokenbuf,"subroutine name");
5371 if (*s == ';' || *s == ')') /* probably a close */
5372 Perl_croak(aTHX_ "sort is now a reserved word");
5374 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5378 LOP(OP_SPLIT,XTERM);
5381 LOP(OP_SPRINTF,XTERM);
5384 LOP(OP_SPLICE,XTERM);
5399 LOP(OP_SUBSTR,XTERM);
5405 char tmpbuf[sizeof PL_tokenbuf];
5406 SSize_t tboffset = 0;
5407 expectation attrful;
5408 bool have_name, have_proto, bad_proto;
5409 const int key = tmp;
5413 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
5414 (*s == ':' && s[1] == ':'))
5417 attrful = XATTRBLOCK;
5418 /* remember buffer pos'n for later force_word */
5419 tboffset = s - PL_oldbufptr;
5420 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
5421 if (strchr(tmpbuf, ':'))
5422 sv_setpv(PL_subname, tmpbuf);
5424 sv_setsv(PL_subname,PL_curstname);
5425 sv_catpvn(PL_subname,"::",2);
5426 sv_catpvn(PL_subname,tmpbuf,len);
5433 Perl_croak(aTHX_ "Missing name in \"my sub\"");
5434 PL_expect = XTERMBLOCK;
5435 attrful = XATTRTERM;
5436 sv_setpvn(PL_subname,"?",1);
5440 if (key == KEY_format) {
5442 PL_lex_formbrack = PL_lex_brackets + 1;
5444 (void) force_word(PL_oldbufptr + tboffset, WORD,
5449 /* Look for a prototype */
5453 s = scan_str(s,FALSE,FALSE);
5455 Perl_croak(aTHX_ "Prototype not terminated");
5456 /* strip spaces and check for bad characters */
5457 d = SvPVX(PL_lex_stuff);
5460 for (p = d; *p; ++p) {
5463 if (!strchr("$@%*;[]&\\", *p))
5468 if (bad_proto && ckWARN(WARN_SYNTAX))
5469 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5470 "Illegal character in prototype for %"SVf" : %s",
5472 SvCUR_set(PL_lex_stuff, tmp);
5480 if (*s == ':' && s[1] != ':')
5481 PL_expect = attrful;
5482 else if (*s != '{' && key == KEY_sub) {
5484 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
5486 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, PL_subname);
5490 PL_nextval[PL_nexttoke].opval =
5491 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5492 PL_lex_stuff = Nullsv;
5496 sv_setpv(PL_subname,
5497 PL_curstash ? "__ANON__" : "__ANON__::__ANON__");
5500 (void) force_word(PL_oldbufptr + tboffset, WORD,
5509 LOP(OP_SYSTEM,XREF);
5512 LOP(OP_SYMLINK,XTERM);
5515 LOP(OP_SYSCALL,XTERM);
5518 LOP(OP_SYSOPEN,XTERM);
5521 LOP(OP_SYSSEEK,XTERM);
5524 LOP(OP_SYSREAD,XTERM);
5527 LOP(OP_SYSWRITE,XTERM);
5531 TERM(sublex_start());
5552 LOP(OP_TRUNCATE,XTERM);
5564 yylval.ival = CopLINE(PL_curcop);
5568 yylval.ival = CopLINE(PL_curcop);
5572 LOP(OP_UNLINK,XTERM);
5578 LOP(OP_UNPACK,XTERM);
5581 LOP(OP_UTIME,XTERM);
5587 LOP(OP_UNSHIFT,XTERM);
5590 s = tokenize_use(1, s);
5600 yylval.ival = CopLINE(PL_curcop);
5604 yylval.ival = CopLINE(PL_curcop);
5608 PL_hints |= HINT_BLOCK_SCOPE;
5615 LOP(OP_WAITPID,XTERM);
5624 ctl_l[0] = toCTRL('L');
5626 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5629 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5634 if (PL_expect == XOPERATOR)
5640 yylval.ival = OP_XOR;
5645 TERM(sublex_start());
5650 #pragma segment Main
5654 S_pending_ident(pTHX)
5657 register I32 tmp = 0;
5658 /* pit holds the identifier we read and pending_ident is reset */
5659 char pit = PL_pending_ident;
5660 PL_pending_ident = 0;
5662 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5663 "### Pending identifier '%s'\n", PL_tokenbuf); });
5665 /* if we're in a my(), we can't allow dynamics here.
5666 $foo'bar has already been turned into $foo::bar, so
5667 just check for colons.
5669 if it's a legal name, the OP is a PADANY.
5672 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5673 if (strchr(PL_tokenbuf,':'))
5674 yyerror(Perl_form(aTHX_ "No package name allowed for "
5675 "variable %s in \"our\"",
5677 tmp = allocmy(PL_tokenbuf);
5680 if (strchr(PL_tokenbuf,':'))
5681 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5683 yylval.opval = newOP(OP_PADANY, 0);
5684 yylval.opval->op_targ = allocmy(PL_tokenbuf);
5690 build the ops for accesses to a my() variable.
5692 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5693 then used in a comparison. This catches most, but not
5694 all cases. For instance, it catches
5695 sort { my($a); $a <=> $b }
5697 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5698 (although why you'd do that is anyone's guess).
5701 if (!strchr(PL_tokenbuf,':')) {
5703 tmp = pad_findmy(PL_tokenbuf);
5704 if (tmp != NOT_IN_PAD) {
5705 /* might be an "our" variable" */
5706 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
5707 /* build ops for a bareword */
5708 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
5709 HEK * const stashname = HvNAME_HEK(stash);
5710 SV * const sym = newSVhek(stashname);
5711 sv_catpvn(sym, "::", 2);
5712 sv_catpv(sym, PL_tokenbuf+1);
5713 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5714 yylval.opval->op_private = OPpCONST_ENTERED;
5717 ? (GV_ADDMULTI | GV_ADDINEVAL)
5720 ((PL_tokenbuf[0] == '$') ? SVt_PV
5721 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5726 /* if it's a sort block and they're naming $a or $b */
5727 if (PL_last_lop_op == OP_SORT &&
5728 PL_tokenbuf[0] == '$' &&
5729 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5732 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5733 d < PL_bufend && *d != '\n';
5736 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5737 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5743 yylval.opval = newOP(OP_PADANY, 0);
5744 yylval.opval->op_targ = tmp;
5750 Whine if they've said @foo in a doublequoted string,
5751 and @foo isn't a variable we can find in the symbol
5754 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5755 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5756 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5757 && ckWARN(WARN_AMBIGUOUS))
5759 /* Downgraded from fatal to warning 20000522 mjd */
5760 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5761 "Possible unintended interpolation of %s in string",
5766 /* build ops for a bareword */
5767 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5768 yylval.opval->op_private = OPpCONST_ENTERED;
5772 ? (GV_ADDMULTI | GV_ADDINEVAL)
5773 /* If the identifier refers to a stash, don't autovivify it.
5774 * Change 24660 had the side effect of causing symbol table
5775 * hashes to always be defined, even if they were freshly
5776 * created and the only reference in the entire program was
5777 * the single statement with the defined %foo::bar:: test.
5778 * It appears that all code in the wild doing this actually
5779 * wants to know whether sub-packages have been loaded, so
5780 * by avoiding auto-vivifying symbol tables, we ensure that
5781 * defined %foo::bar:: continues to be false, and the existing
5782 * tests still give the expected answers, even though what
5783 * they're actually testing has now changed subtly.
5785 : !(*PL_tokenbuf == '%' && *(d = PL_tokenbuf + strlen(PL_tokenbuf) - 1) == ':' && d[-1] == ':'),
5786 ((PL_tokenbuf[0] == '$') ? SVt_PV
5787 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5793 * The following code was generated by perl_keyword.pl.
5797 Perl_keyword (pTHX_ const char *name, I32 len)
5801 case 1: /* 5 tokens of length 1 */
5833 case 2: /* 18 tokens of length 2 */
5979 case 3: /* 29 tokens of length 3 */
5983 if (name[1] == 'N' &&
6046 if (name[1] == 'i' &&
6086 if (name[1] == 'o' &&
6095 if (name[1] == 'e' &&
6104 if (name[1] == 'n' &&
6113 if (name[1] == 'o' &&
6122 if (name[1] == 'a' &&
6131 if (name[1] == 'o' &&
6193 if (name[1] == 'e' &&
6207 return (FEATURE_IS_ENABLED("say", 3) ? -KEY_say : 0);
6233 if (name[1] == 'i' &&
6242 if (name[1] == 's' &&
6251 if (name[1] == 'e' &&
6260 if (name[1] == 'o' &&
6272 case 4: /* 41 tokens of length 4 */
6276 if (name[1] == 'O' &&
6286 if (name[1] == 'N' &&
6296 if (name[1] == 'i' &&
6306 if (name[1] == 'h' &&
6316 if (name[1] == 'u' &&
6329 if (name[2] == 'c' &&
6338 if (name[2] == 's' &&
6347 if (name[2] == 'a' &&
6383 if (name[1] == 'o' &&
6396 if (name[2] == 't' &&
6405 if (name[2] == 'o' &&
6414 if (name[2] == 't' &&
6423 if (name[2] == 'e' &&
6436 if (name[1] == 'o' &&
6449 if (name[2] == 'y' &&
6458 if (name[2] == 'l' &&
6474 if (name[2] == 's' &&
6483 if (name[2] == 'n' &&
6492 if (name[2] == 'c' &&
6505 if (name[1] == 'e' &&
6515 if (name[1] == 'p' &&
6528 if (name[2] == 'c' &&
6537 if (name[2] == 'p' &&
6546 if (name[2] == 's' &&
6562 if (name[2] == 'n' &&
6632 if (name[2] == 'r' &&
6641 if (name[2] == 'r' &&
6650 if (name[2] == 'a' &&
6666 if (name[2] == 'l' &&
6728 if (name[2] == 'e' &&
6731 return (FEATURE_IS_ENABLED("switch", 6) ? KEY_when : 0);
6744 case 5: /* 38 tokens of length 5 */
6748 if (name[1] == 'E' &&
6759 if (name[1] == 'H' &&
6773 if (name[2] == 'a' &&
6783 if (name[2] == 'a' &&
6800 if (name[2] == 'e' &&
6810 if (name[2] == 'e' &&
6814 return (FEATURE_IS_ENABLED("switch", 6) ? -KEY_break : 0);
6830 if (name[3] == 'i' &&
6839 if (name[3] == 'o' &&
6875 if (name[2] == 'o' &&
6885 if (name[2] == 'y' &&
6899 if (name[1] == 'l' &&
6913 if (name[2] == 'n' &&
6923 if (name[2] == 'o' &&
6937 if (name[1] == 'i' &&
6942 return (FEATURE_IS_ENABLED("switch", 6) ? KEY_given : 0);
6951 if (name[2] == 'd' &&
6961 if (name[2] == 'c' &&
6978 if (name[2] == 'c' &&
6988 if (name[2] == 't' &&
7002 if (name[1] == 'k' &&
7013 if (name[1] == 'r' &&
7027 if (name[2] == 's' &&
7037 if (name[2] == 'd' &&
7054 if (name[2] == 'm' &&
7064 if (name[2] == 'i' &&
7074 if (name[2] == 'e' &&
7084 if (name[2] == 'l' &&
7094 if (name[2] == 'a' &&
7104 if (name[2] == 'u' &&
7118 if (name[1] == 'i' &&
7132 if (name[2] == 'a' &&
7145 if (name[3] == 'e' &&
7180 if (name[2] == 'i' &&
7197 if (name[2] == 'i' &&
7207 if (name[2] == 'i' &&
7224 case 6: /* 33 tokens of length 6 */
7228 if (name[1] == 'c' &&
7243 if (name[2] == 'l' &&
7254 if (name[2] == 'r' &&
7269 if (name[1] == 'e' &&
7284 if (name[2] == 's' &&
7289 if(ckWARN_d(WARN_SYNTAX))
7290 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
7296 if (name[2] == 'i' &&
7314 if (name[2] == 'l' &&
7325 if (name[2] == 'r' &&
7340 if (name[1] == 'm' &&
7355 if (name[2] == 'n' &&
7366 if (name[2] == 's' &&
7381 if (name[1] == 's' &&
7387 if (name[4] == 't' &&
7396 if (name[4] == 'e' &&
7405 if (name[4] == 'c' &&
7414 if (name[4] == 'n' &&
7430 if (name[1] == 'r' &&
7448 if (name[3] == 'a' &&
7458 if (name[3] == 'u' &&
7472 if (name[2] == 'n' &&
7490 if (name[2] == 'a' &&
7504 if (name[3] == 'e' &&
7517 if (name[4] == 't' &&
7526 if (name[4] == 'e' &&
7548 if (name[4] == 't' &&
7557 if (name[4] == 'e' &&
7573 if (name[2] == 'c' &&
7584 if (name[2] == 'l' &&
7595 if (name[2] == 'b' &&
7606 if (name[2] == 's' &&
7629 if (name[4] == 's' &&
7638 if (name[4] == 'n' &&
7651 if (name[3] == 'a' &&
7668 if (name[1] == 'a' &&
7683 case 7: /* 29 tokens of length 7 */
7687 if (name[1] == 'E' &&
7700 if (name[1] == '_' &&
7713 if (name[1] == 'i' &&
7720 return -KEY_binmode;
7726 if (name[1] == 'o' &&
7733 return -KEY_connect;
7742 if (name[2] == 'm' &&
7748 return -KEY_dbmopen;
7759 if (name[4] == 'u' &&
7763 return (FEATURE_IS_ENABLED("switch", 6) ? KEY_default : 0);
7769 if (name[4] == 'n' &&
7790 if (name[1] == 'o' &&
7803 if (name[1] == 'e' &&
7810 if (name[5] == 'r' &&
7813 return -KEY_getpgrp;
7819 if (name[5] == 'i' &&
7822 return -KEY_getppid;
7835 if (name[1] == 'c' &&
7842 return -KEY_lcfirst;
7848 if (name[1] == 'p' &&
7855 return -KEY_opendir;
7861 if (name[1] == 'a' &&
7879 if (name[3] == 'd' &&
7884 return -KEY_readdir;
7890 if (name[3] == 'u' &&
7901 if (name[3] == 'e' &&
7906 return -KEY_reverse;
7925 if (name[3] == 'k' &&
7930 return -KEY_seekdir;
7936 if (name[3] == 'p' &&
7941 return -KEY_setpgrp;
7951 if (name[2] == 'm' &&
7957 return -KEY_shmread;
7963 if (name[2] == 'r' &&
7969 return -KEY_sprintf;
7978 if (name[3] == 'l' &&
7983 return -KEY_symlink;
7992 if (name[4] == 'a' &&
7996 return -KEY_syscall;
8002 if (name[4] == 'p' &&
8006 return -KEY_sysopen;
8012 if (name[4] == 'e' &&
8016 return -KEY_sysread;
8022 if (name[4] == 'e' &&
8026 return -KEY_sysseek;
8044 if (name[1] == 'e' &&
8051 return -KEY_telldir;
8060 if (name[2] == 'f' &&
8066 return -KEY_ucfirst;
8072 if (name[2] == 's' &&
8078 return -KEY_unshift;
8088 if (name[1] == 'a' &&
8095 return -KEY_waitpid;
8104 case 8: /* 26 tokens of length 8 */
8108 if (name[1] == 'U' &&
8116 return KEY_AUTOLOAD;
8127 if (name[3] == 'A' &&
8133 return KEY___DATA__;
8139 if (name[3] == 'I' &&
8145 return -KEY___FILE__;
8151 if (name[3] == 'I' &&
8157 return -KEY___LINE__;
8173 if (name[2] == 'o' &&
8180 return -KEY_closedir;
8186 if (name[2] == 'n' &&
8193 return -KEY_continue;
8203 if (name[1] == 'b' &&
8211 return -KEY_dbmclose;
8217 if (name[1] == 'n' &&
8223 if (name[4] == 'r' &&
8228 return -KEY_endgrent;
8234 if (name[4] == 'w' &&
8239 return -KEY_endpwent;
8252 if (name[1] == 'o' &&
8260 return -KEY_formline;
8266 if (name[1] == 'e' &&
8277 if (name[6] == 'n' &&
8280 return -KEY_getgrent;
8286 if (name[6] == 'i' &&
8289 return -KEY_getgrgid;
8295 if (name[6] == 'a' &&
8298 return -KEY_getgrnam;
8311 if (name[4] == 'o' &&
8316 return -KEY_getlogin;
8327 if (name[6] == 'n' &&
8330 return -KEY_getpwent;
8336 if (name[6] == 'a' &&
8339 return -KEY_getpwnam;
8345 if (name[6] == 'i' &&
8348 return -KEY_getpwuid;
8368 if (name[1] == 'e' &&
8375 if (name[5] == 'i' &&
8382 return -KEY_readline;
8387 return -KEY_readlink;
8398 if (name[5] == 'i' &&
8402 return -KEY_readpipe;
8423 if (name[4] == 'r' &&
8428 return -KEY_setgrent;
8434 if (name[4] == 'w' &&
8439 return -KEY_setpwent;
8455 if (name[3] == 'w' &&
8461 return -KEY_shmwrite;
8467 if (name[3] == 't' &&
8473 return -KEY_shutdown;
8483 if (name[2] == 's' &&
8490 return -KEY_syswrite;
8500 if (name[1] == 'r' &&
8508 return -KEY_truncate;
8517 case 9: /* 8 tokens of length 9 */
8521 if (name[1] == 'n' &&
8530 return -KEY_endnetent;
8536 if (name[1] == 'e' &&
8545 return -KEY_getnetent;
8551 if (name[1] == 'o' &&
8560 return -KEY_localtime;
8566 if (name[1] == 'r' &&
8575 return KEY_prototype;
8581 if (name[1] == 'u' &&
8590 return -KEY_quotemeta;
8596 if (name[1] == 'e' &&
8605 return -KEY_rewinddir;
8611 if (name[1] == 'e' &&
8620 return -KEY_setnetent;
8626 if (name[1] == 'a' &&
8635 return -KEY_wantarray;
8644 case 10: /* 9 tokens of length 10 */
8648 if (name[1] == 'n' &&
8654 if (name[4] == 'o' &&
8661 return -KEY_endhostent;
8667 if (name[4] == 'e' &&
8674 return -KEY_endservent;
8687 if (name[1] == 'e' &&
8693 if (name[4] == 'o' &&
8700 return -KEY_gethostent;
8709 if (name[5] == 'r' &&
8715 return -KEY_getservent;
8721 if (name[5] == 'c' &&
8727 return -KEY_getsockopt;
8752 if (name[4] == 'o' &&
8759 return -KEY_sethostent;
8768 if (name[5] == 'r' &&
8774 return -KEY_setservent;
8780 if (name[5] == 'c' &&
8786 return -KEY_setsockopt;
8803 if (name[2] == 'c' &&
8812 return -KEY_socketpair;
8825 case 11: /* 8 tokens of length 11 */
8829 if (name[1] == '_' &&
8840 return -KEY___PACKAGE__;
8846 if (name[1] == 'n' &&
8857 return -KEY_endprotoent;
8863 if (name[1] == 'e' &&
8872 if (name[5] == 'e' &&
8879 return -KEY_getpeername;
8888 if (name[6] == 'o' &&
8894 return -KEY_getpriority;
8900 if (name[6] == 't' &&
8906 return -KEY_getprotoent;
8920 if (name[4] == 'o' &&
8928 return -KEY_getsockname;
8941 if (name[1] == 'e' &&
8949 if (name[6] == 'o' &&
8955 return -KEY_setpriority;
8961 if (name[6] == 't' &&
8967 return -KEY_setprotoent;
8983 case 12: /* 2 tokens of length 12 */
8984 if (name[0] == 'g' &&
8996 if (name[9] == 'd' &&
8999 { /* getnetbyaddr */
9000 return -KEY_getnetbyaddr;
9006 if (name[9] == 'a' &&
9009 { /* getnetbyname */
9010 return -KEY_getnetbyname;
9022 case 13: /* 4 tokens of length 13 */
9023 if (name[0] == 'g' &&
9030 if (name[4] == 'o' &&
9039 if (name[10] == 'd' &&
9042 { /* gethostbyaddr */
9043 return -KEY_gethostbyaddr;
9049 if (name[10] == 'a' &&
9052 { /* gethostbyname */
9053 return -KEY_gethostbyname;
9066 if (name[4] == 'e' &&
9075 if (name[10] == 'a' &&
9078 { /* getservbyname */
9079 return -KEY_getservbyname;
9085 if (name[10] == 'o' &&
9088 { /* getservbyport */
9089 return -KEY_getservbyport;
9108 case 14: /* 1 tokens of length 14 */
9109 if (name[0] == 'g' &&
9123 { /* getprotobyname */
9124 return -KEY_getprotobyname;
9129 case 16: /* 1 tokens of length 16 */
9130 if (name[0] == 'g' &&
9146 { /* getprotobynumber */
9147 return -KEY_getprotobynumber;
9161 S_checkcomma(pTHX_ register char *s, const char *name, const char *what)
9165 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
9166 if (ckWARN(WARN_SYNTAX)) {
9168 for (w = s+2; *w && level; w++) {
9175 for (; *w && isSPACE(*w); w++) ;
9176 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
9177 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
9178 "%s (...) interpreted as function",name);
9181 while (s < PL_bufend && isSPACE(*s))
9185 while (s < PL_bufend && isSPACE(*s))
9187 if (isIDFIRST_lazy_if(s,UTF)) {
9189 while (isALNUM_lazy_if(s,UTF))
9191 while (s < PL_bufend && isSPACE(*s))
9195 *s = '\0'; /* XXX If we didn't do this, we could const a lot of toke.c */
9196 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
9200 Perl_croak(aTHX_ "No comma allowed after %s", what);
9205 /* Either returns sv, or mortalizes sv and returns a new SV*.
9206 Best used as sv=new_constant(..., sv, ...).
9207 If s, pv are NULL, calls subroutine with one argument,
9208 and type is used with error messages only. */
9211 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, SV *sv, SV *pv,
9215 HV * const table = GvHV(PL_hintgv); /* ^H */
9219 const char *why1 = "", *why2 = "", *why3 = "";
9221 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
9224 why2 = strEQ(key,"charnames")
9225 ? "(possibly a missing \"use charnames ...\")"
9227 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
9228 (type ? type: "undef"), why2);
9230 /* This is convoluted and evil ("goto considered harmful")
9231 * but I do not understand the intricacies of all the different
9232 * failure modes of %^H in here. The goal here is to make
9233 * the most probable error message user-friendly. --jhi */
9238 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
9239 (type ? type: "undef"), why1, why2, why3);
9241 yyerror(SvPVX_const(msg));
9245 cvp = hv_fetch(table, key, strlen(key), FALSE);
9246 if (!cvp || !SvOK(*cvp)) {
9249 why3 = "} is not defined";
9252 sv_2mortal(sv); /* Parent created it permanently */
9255 pv = sv_2mortal(newSVpvn(s, len));
9257 typesv = sv_2mortal(newSVpv(type, 0));
9259 typesv = &PL_sv_undef;
9261 PUSHSTACKi(PERLSI_OVERLOAD);
9273 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
9277 /* Check the eval first */
9278 if (!PL_in_eval && SvTRUE(ERRSV)) {
9279 sv_catpv(ERRSV, "Propagated");
9280 yyerror(SvPV_nolen_const(ERRSV)); /* Duplicates the message inside eval */
9282 res = SvREFCNT_inc(sv);
9286 (void)SvREFCNT_inc(res);
9295 why1 = "Call to &{$^H{";
9297 why3 = "}} did not return a defined value";
9305 /* Returns a NUL terminated string, with the length of the string written to
9309 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
9311 register char *d = dest;
9312 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
9315 Perl_croak(aTHX_ ident_too_long);
9316 if (isALNUM(*s)) /* UTF handled below */
9318 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
9323 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
9327 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
9328 char *t = s + UTF8SKIP(s);
9329 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
9331 if (d + (t - s) > e)
9332 Perl_croak(aTHX_ ident_too_long);
9333 Copy(s, d, t - s, char);
9346 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
9350 char *bracket = Nullch;
9356 e = d + destlen - 3; /* two-character token, ending NUL */
9358 while (isDIGIT(*s)) {
9360 Perl_croak(aTHX_ ident_too_long);
9367 Perl_croak(aTHX_ ident_too_long);
9368 if (isALNUM(*s)) /* UTF handled below */
9370 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
9375 else if (*s == ':' && s[1] == ':') {
9379 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
9380 char *t = s + UTF8SKIP(s);
9381 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
9383 if (d + (t - s) > e)
9384 Perl_croak(aTHX_ ident_too_long);
9385 Copy(s, d, t - s, char);
9396 if (PL_lex_state != LEX_NORMAL)
9397 PL_lex_state = LEX_INTERPENDMAYBE;
9400 if (*s == '$' && s[1] &&
9401 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
9414 if (*d == '^' && *s && isCONTROLVAR(*s)) {
9419 if (isSPACE(s[-1])) {
9421 const char ch = *s++;
9422 if (!SPACE_OR_TAB(ch)) {
9428 if (isIDFIRST_lazy_if(d,UTF)) {
9432 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
9434 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
9437 Copy(s, d, e - s, char);
9442 while ((isALNUM(*s) || *s == ':') && d < e)
9445 Perl_croak(aTHX_ ident_too_long);
9448 while (s < send && SPACE_OR_TAB(*s)) s++;
9449 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
9450 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
9451 const char *brack = *s == '[' ? "[...]" : "{...}";
9452 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
9453 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
9454 funny, dest, brack, funny, dest, brack);
9457 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
9461 /* Handle extended ${^Foo} variables
9462 * 1999-02-27 mjd-perl-patch@plover.com */
9463 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
9467 while (isALNUM(*s) && d < e) {
9471 Perl_croak(aTHX_ ident_too_long);
9476 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
9477 PL_lex_state = LEX_INTERPEND;
9482 if (PL_lex_state == LEX_NORMAL) {
9483 if (ckWARN(WARN_AMBIGUOUS) &&
9484 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
9486 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
9487 "Ambiguous use of %c{%s} resolved to %c%s",
9488 funny, dest, funny, dest);
9493 s = bracket; /* let the parser handle it */
9497 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
9498 PL_lex_state = LEX_INTERPEND;
9503 Perl_pmflag(pTHX_ U32* pmfl, int ch)
9508 *pmfl |= PMf_GLOBAL;
9510 *pmfl |= PMf_CONTINUE;
9514 *pmfl |= PMf_MULTILINE;
9516 *pmfl |= PMf_SINGLELINE;
9518 *pmfl |= PMf_EXTENDED;
9522 S_scan_pat(pTHX_ char *start, I32 type)
9525 char *s = scan_str(start,FALSE,FALSE);
9528 char * const delimiter = skipspace(start);
9529 Perl_croak(aTHX_ *delimiter == '?'
9530 ? "Search pattern not terminated or ternary operator parsed as search pattern"
9531 : "Search pattern not terminated" );
9534 pm = (PMOP*)newPMOP(type, 0);
9535 if (PL_multi_open == '?')
9536 pm->op_pmflags |= PMf_ONCE;
9538 while (*s && strchr("iomsx", *s))
9539 pmflag(&pm->op_pmflags,*s++);
9542 while (*s && strchr("iogcmsx", *s))
9543 pmflag(&pm->op_pmflags,*s++);
9545 /* issue a warning if /c is specified,but /g is not */
9546 if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL)
9547 && ckWARN(WARN_REGEXP))
9549 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless without /g" );
9552 pm->op_pmpermflags = pm->op_pmflags;
9554 PL_lex_op = (OP*)pm;
9555 yylval.ival = OP_MATCH;
9560 S_scan_subst(pTHX_ char *start)
9568 yylval.ival = OP_NULL;
9570 s = scan_str(start,FALSE,FALSE);
9573 Perl_croak(aTHX_ "Substitution pattern not terminated");
9575 if (s[-1] == PL_multi_open)
9578 first_start = PL_multi_start;
9579 s = scan_str(s,FALSE,FALSE);
9582 SvREFCNT_dec(PL_lex_stuff);
9583 PL_lex_stuff = Nullsv;
9585 Perl_croak(aTHX_ "Substitution replacement not terminated");
9587 PL_multi_start = first_start; /* so whole substitution is taken together */
9589 pm = (PMOP*)newPMOP(OP_SUBST, 0);
9595 else if (strchr("iogcmsx", *s))
9596 pmflag(&pm->op_pmflags,*s++);
9601 if ((pm->op_pmflags & PMf_CONTINUE) && ckWARN(WARN_REGEXP)) {
9602 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless in s///" );
9607 PL_sublex_info.super_bufptr = s;
9608 PL_sublex_info.super_bufend = PL_bufend;
9610 pm->op_pmflags |= PMf_EVAL;
9611 repl = newSVpvn("",0);
9613 sv_catpv(repl, es ? "eval " : "do ");
9614 sv_catpvn(repl, "{ ", 2);
9615 sv_catsv(repl, PL_lex_repl);
9616 sv_catpvn(repl, " };", 2);
9618 SvREFCNT_dec(PL_lex_repl);
9622 pm->op_pmpermflags = pm->op_pmflags;
9623 PL_lex_op = (OP*)pm;
9624 yylval.ival = OP_SUBST;
9629 S_scan_trans(pTHX_ char *start)
9638 yylval.ival = OP_NULL;
9640 s = scan_str(start,FALSE,FALSE);
9642 Perl_croak(aTHX_ "Transliteration pattern not terminated");
9643 if (s[-1] == PL_multi_open)
9646 s = scan_str(s,FALSE,FALSE);
9649 SvREFCNT_dec(PL_lex_stuff);
9650 PL_lex_stuff = Nullsv;
9652 Perl_croak(aTHX_ "Transliteration replacement not terminated");
9655 complement = del = squash = 0;
9659 complement = OPpTRANS_COMPLEMENT;
9662 del = OPpTRANS_DELETE;
9665 squash = OPpTRANS_SQUASH;
9674 Newx(tbl, complement&&!del?258:256, short);
9675 o = newPVOP(OP_TRANS, 0, (char*)tbl);
9676 o->op_private &= ~OPpTRANS_ALL;
9677 o->op_private |= del|squash|complement|
9678 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
9679 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
9682 yylval.ival = OP_TRANS;
9687 S_scan_heredoc(pTHX_ register char *s)
9690 I32 op_type = OP_SCALAR;
9694 const char newline[] = "\n";
9695 const char *found_newline;
9699 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
9703 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
9706 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
9707 if (*peek == '`' || *peek == '\'' || *peek =='"') {
9710 s = delimcpy(d, e, s, PL_bufend, term, &len);
9720 if (!isALNUM_lazy_if(s,UTF))
9721 deprecate_old("bare << to mean <<\"\"");
9722 for (; isALNUM_lazy_if(s,UTF); s++) {
9727 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
9728 Perl_croak(aTHX_ "Delimiter for here document is too long");
9731 len = d - PL_tokenbuf;
9732 #ifndef PERL_STRICT_CR
9733 d = strchr(s, '\r');
9735 char * const olds = s;
9737 while (s < PL_bufend) {
9743 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
9752 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
9756 if ( outer || !(found_newline = ninstr(s,PL_bufend,newline,newline+1)) ) {
9757 herewas = newSVpvn(s,PL_bufend-s);
9761 herewas = newSVpvn(s,found_newline-s);
9763 s += SvCUR(herewas);
9765 tmpstr = NEWSV(87,79);
9766 sv_upgrade(tmpstr, SVt_PVIV);
9769 SvIV_set(tmpstr, -1);
9771 else if (term == '`') {
9772 op_type = OP_BACKTICK;
9773 SvIV_set(tmpstr, '\\');
9777 PL_multi_start = CopLINE(PL_curcop);
9778 PL_multi_open = PL_multi_close = '<';
9779 term = *PL_tokenbuf;
9780 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
9781 char *bufptr = PL_sublex_info.super_bufptr;
9782 char *bufend = PL_sublex_info.super_bufend;
9783 char * const olds = s - SvCUR(herewas);
9784 s = strchr(bufptr, '\n');
9788 while (s < bufend &&
9789 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
9791 CopLINE_inc(PL_curcop);
9794 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9795 missingterm(PL_tokenbuf);
9797 sv_setpvn(herewas,bufptr,d-bufptr+1);
9798 sv_setpvn(tmpstr,d+1,s-d);
9800 sv_catpvn(herewas,s,bufend-s);
9801 Copy(SvPVX_const(herewas),bufptr,SvCUR(herewas) + 1,char);
9808 while (s < PL_bufend &&
9809 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
9811 CopLINE_inc(PL_curcop);
9813 if (s >= PL_bufend) {
9814 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9815 missingterm(PL_tokenbuf);
9817 sv_setpvn(tmpstr,d+1,s-d);
9819 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
9821 sv_catpvn(herewas,s,PL_bufend-s);
9822 sv_setsv(PL_linestr,herewas);
9823 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
9824 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9825 PL_last_lop = PL_last_uni = Nullch;
9828 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
9829 while (s >= PL_bufend) { /* multiple line string? */
9831 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
9832 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9833 missingterm(PL_tokenbuf);
9835 CopLINE_inc(PL_curcop);
9836 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9837 PL_last_lop = PL_last_uni = Nullch;
9838 #ifndef PERL_STRICT_CR
9839 if (PL_bufend - PL_linestart >= 2) {
9840 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
9841 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
9843 PL_bufend[-2] = '\n';
9845 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
9847 else if (PL_bufend[-1] == '\r')
9848 PL_bufend[-1] = '\n';
9850 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
9851 PL_bufend[-1] = '\n';
9853 if (PERLDB_LINE && PL_curstash != PL_debstash) {
9854 SV *sv = NEWSV(88,0);
9856 sv_upgrade(sv, SVt_PVMG);
9857 sv_setsv(sv,PL_linestr);
9860 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
9862 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
9863 STRLEN off = PL_bufend - 1 - SvPVX_const(PL_linestr);
9864 *(SvPVX(PL_linestr) + off ) = ' ';
9865 sv_catsv(PL_linestr,herewas);
9866 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9867 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
9871 sv_catsv(tmpstr,PL_linestr);
9876 PL_multi_end = CopLINE(PL_curcop);
9877 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
9878 SvPV_shrink_to_cur(tmpstr);
9880 SvREFCNT_dec(herewas);
9882 if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
9884 else if (PL_encoding)
9885 sv_recode_to_utf8(tmpstr, PL_encoding);
9887 PL_lex_stuff = tmpstr;
9888 yylval.ival = op_type;
9893 takes: current position in input buffer
9894 returns: new position in input buffer
9895 side-effects: yylval and lex_op are set.
9900 <FH> read from filehandle
9901 <pkg::FH> read from package qualified filehandle
9902 <pkg'FH> read from package qualified filehandle
9903 <$fh> read from filehandle in $fh
9909 S_scan_inputsymbol(pTHX_ char *start)
9911 register char *s = start; /* current position in buffer */
9917 d = PL_tokenbuf; /* start of temp holding space */
9918 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
9919 end = strchr(s, '\n');
9922 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
9924 /* die if we didn't have space for the contents of the <>,
9925 or if it didn't end, or if we see a newline
9928 if (len >= sizeof PL_tokenbuf)
9929 Perl_croak(aTHX_ "Excessively long <> operator");
9931 Perl_croak(aTHX_ "Unterminated <> operator");
9936 Remember, only scalar variables are interpreted as filehandles by
9937 this code. Anything more complex (e.g., <$fh{$num}>) will be
9938 treated as a glob() call.
9939 This code makes use of the fact that except for the $ at the front,
9940 a scalar variable and a filehandle look the same.
9942 if (*d == '$' && d[1]) d++;
9944 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
9945 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
9948 /* If we've tried to read what we allow filehandles to look like, and
9949 there's still text left, then it must be a glob() and not a getline.
9950 Use scan_str to pull out the stuff between the <> and treat it
9951 as nothing more than a string.
9954 if (d - PL_tokenbuf != len) {
9955 yylval.ival = OP_GLOB;
9957 s = scan_str(start,FALSE,FALSE);
9959 Perl_croak(aTHX_ "Glob not terminated");
9963 bool readline_overriden = FALSE;
9964 GV *gv_readline = Nullgv;
9966 /* we're in a filehandle read situation */
9969 /* turn <> into <ARGV> */
9971 Copy("ARGV",d,5,char);
9973 /* Check whether readline() is overriden */
9974 if (((gv_readline = gv_fetchpv("readline", FALSE, SVt_PVCV))
9975 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
9977 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
9978 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
9979 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
9980 readline_overriden = TRUE;
9982 /* if <$fh>, create the ops to turn the variable into a
9988 /* try to find it in the pad for this block, otherwise find
9989 add symbol table ops
9991 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
9992 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
9993 HV *stash = PAD_COMPNAME_OURSTASH(tmp);
9994 HEK *stashname = HvNAME_HEK(stash);
9995 SV *sym = sv_2mortal(newSVhek(stashname));
9996 sv_catpvn(sym, "::", 2);
10002 OP *o = newOP(OP_PADSV, 0);
10004 PL_lex_op = readline_overriden
10005 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
10006 append_elem(OP_LIST, o,
10007 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
10008 : (OP*)newUNOP(OP_READLINE, 0, o);
10017 ? (GV_ADDMULTI | GV_ADDINEVAL)
10020 PL_lex_op = readline_overriden
10021 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
10022 append_elem(OP_LIST,
10023 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
10024 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
10025 : (OP*)newUNOP(OP_READLINE, 0,
10026 newUNOP(OP_RV2SV, 0,
10027 newGVOP(OP_GV, 0, gv)));
10029 if (!readline_overriden)
10030 PL_lex_op->op_flags |= OPf_SPECIAL;
10031 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
10032 yylval.ival = OP_NULL;
10035 /* If it's none of the above, it must be a literal filehandle
10036 (<Foo::BAR> or <FOO>) so build a simple readline OP */
10038 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
10039 PL_lex_op = readline_overriden
10040 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
10041 append_elem(OP_LIST,
10042 newGVOP(OP_GV, 0, gv),
10043 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
10044 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
10045 yylval.ival = OP_NULL;
10054 takes: start position in buffer
10055 keep_quoted preserve \ on the embedded delimiter(s)
10056 keep_delims preserve the delimiters around the string
10057 returns: position to continue reading from buffer
10058 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
10059 updates the read buffer.
10061 This subroutine pulls a string out of the input. It is called for:
10062 q single quotes q(literal text)
10063 ' single quotes 'literal text'
10064 qq double quotes qq(interpolate $here please)
10065 " double quotes "interpolate $here please"
10066 qx backticks qx(/bin/ls -l)
10067 ` backticks `/bin/ls -l`
10068 qw quote words @EXPORT_OK = qw( func() $spam )
10069 m// regexp match m/this/
10070 s/// regexp substitute s/this/that/
10071 tr/// string transliterate tr/this/that/
10072 y/// string transliterate y/this/that/
10073 ($*@) sub prototypes sub foo ($)
10074 (stuff) sub attr parameters sub foo : attr(stuff)
10075 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
10077 In most of these cases (all but <>, patterns and transliterate)
10078 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
10079 calls scan_str(). s/// makes yylex() call scan_subst() which calls
10080 scan_str(). tr/// and y/// make yylex() call scan_trans() which
10083 It skips whitespace before the string starts, and treats the first
10084 character as the delimiter. If the delimiter is one of ([{< then
10085 the corresponding "close" character )]}> is used as the closing
10086 delimiter. It allows quoting of delimiters, and if the string has
10087 balanced delimiters ([{<>}]) it allows nesting.
10089 On success, the SV with the resulting string is put into lex_stuff or,
10090 if that is already non-NULL, into lex_repl. The second case occurs only
10091 when parsing the RHS of the special constructs s/// and tr/// (y///).
10092 For convenience, the terminating delimiter character is stuffed into
10097 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
10099 SV *sv; /* scalar value: string */
10100 char *tmps; /* temp string, used for delimiter matching */
10101 register char *s = start; /* current position in the buffer */
10102 register char term; /* terminating character */
10103 register char *to; /* current position in the sv's data */
10104 I32 brackets = 1; /* bracket nesting level */
10105 bool has_utf8 = FALSE; /* is there any utf8 content? */
10106 I32 termcode; /* terminating char. code */
10107 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
10108 STRLEN termlen; /* length of terminating string */
10109 char *last = NULL; /* last position for nesting bracket */
10111 /* skip space before the delimiter */
10115 /* mark where we are, in case we need to report errors */
10118 /* after skipping whitespace, the next character is the terminator */
10121 termcode = termstr[0] = term;
10125 termcode = utf8_to_uvchr((U8*)s, &termlen);
10126 Copy(s, termstr, termlen, U8);
10127 if (!UTF8_IS_INVARIANT(term))
10131 /* mark where we are */
10132 PL_multi_start = CopLINE(PL_curcop);
10133 PL_multi_open = term;
10135 /* find corresponding closing delimiter */
10136 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
10137 termcode = termstr[0] = term = tmps[5];
10139 PL_multi_close = term;
10141 /* create a new SV to hold the contents. 87 is leak category, I'm
10142 assuming. 79 is the SV's initial length. What a random number. */
10144 sv_upgrade(sv, SVt_PVIV);
10145 SvIV_set(sv, termcode);
10146 (void)SvPOK_only(sv); /* validate pointer */
10148 /* move past delimiter and try to read a complete string */
10150 sv_catpvn(sv, s, termlen);
10153 if (PL_encoding && !UTF) {
10157 int offset = s - SvPVX_const(PL_linestr);
10158 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
10159 &offset, (char*)termstr, termlen);
10160 const char *ns = SvPVX_const(PL_linestr) + offset;
10161 char *svlast = SvEND(sv) - 1;
10163 for (; s < ns; s++) {
10164 if (*s == '\n' && !PL_rsfp)
10165 CopLINE_inc(PL_curcop);
10168 goto read_more_line;
10170 /* handle quoted delimiters */
10171 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
10173 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
10175 if ((svlast-1 - t) % 2) {
10176 if (!keep_quoted) {
10177 *(svlast-1) = term;
10179 SvCUR_set(sv, SvCUR(sv) - 1);
10184 if (PL_multi_open == PL_multi_close) {
10192 for (t = w = last; t < svlast; w++, t++) {
10193 /* At here, all closes are "was quoted" one,
10194 so we don't check PL_multi_close. */
10196 if (!keep_quoted && *(t+1) == PL_multi_open)
10201 else if (*t == PL_multi_open)
10209 SvCUR_set(sv, w - SvPVX_const(sv));
10212 if (--brackets <= 0)
10217 if (!keep_delims) {
10218 SvCUR_set(sv, SvCUR(sv) - 1);
10224 /* extend sv if need be */
10225 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
10226 /* set 'to' to the next character in the sv's string */
10227 to = SvPVX(sv)+SvCUR(sv);
10229 /* if open delimiter is the close delimiter read unbridle */
10230 if (PL_multi_open == PL_multi_close) {
10231 for (; s < PL_bufend; s++,to++) {
10232 /* embedded newlines increment the current line number */
10233 if (*s == '\n' && !PL_rsfp)
10234 CopLINE_inc(PL_curcop);
10235 /* handle quoted delimiters */
10236 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
10237 if (!keep_quoted && s[1] == term)
10239 /* any other quotes are simply copied straight through */
10243 /* terminate when run out of buffer (the for() condition), or
10244 have found the terminator */
10245 else if (*s == term) {
10248 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
10251 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
10257 /* if the terminator isn't the same as the start character (e.g.,
10258 matched brackets), we have to allow more in the quoting, and
10259 be prepared for nested brackets.
10262 /* read until we run out of string, or we find the terminator */
10263 for (; s < PL_bufend; s++,to++) {
10264 /* embedded newlines increment the line count */
10265 if (*s == '\n' && !PL_rsfp)
10266 CopLINE_inc(PL_curcop);
10267 /* backslashes can escape the open or closing characters */
10268 if (*s == '\\' && s+1 < PL_bufend) {
10269 if (!keep_quoted &&
10270 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
10275 /* allow nested opens and closes */
10276 else if (*s == PL_multi_close && --brackets <= 0)
10278 else if (*s == PL_multi_open)
10280 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
10285 /* terminate the copied string and update the sv's end-of-string */
10287 SvCUR_set(sv, to - SvPVX_const(sv));
10290 * this next chunk reads more into the buffer if we're not done yet
10294 break; /* handle case where we are done yet :-) */
10296 #ifndef PERL_STRICT_CR
10297 if (to - SvPVX_const(sv) >= 2) {
10298 if ((to[-2] == '\r' && to[-1] == '\n') ||
10299 (to[-2] == '\n' && to[-1] == '\r'))
10303 SvCUR_set(sv, to - SvPVX_const(sv));
10305 else if (to[-1] == '\r')
10308 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
10313 /* if we're out of file, or a read fails, bail and reset the current
10314 line marker so we can report where the unterminated string began
10317 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
10319 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
10322 /* we read a line, so increment our line counter */
10323 CopLINE_inc(PL_curcop);
10325 /* update debugger info */
10326 if (PERLDB_LINE && PL_curstash != PL_debstash) {
10327 SV *sv = NEWSV(88,0);
10329 sv_upgrade(sv, SVt_PVMG);
10330 sv_setsv(sv,PL_linestr);
10331 (void)SvIOK_on(sv);
10333 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
10336 /* having changed the buffer, we must update PL_bufend */
10337 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
10338 PL_last_lop = PL_last_uni = Nullch;
10341 /* at this point, we have successfully read the delimited string */
10343 if (!PL_encoding || UTF) {
10345 sv_catpvn(sv, s, termlen);
10348 if (has_utf8 || PL_encoding)
10351 PL_multi_end = CopLINE(PL_curcop);
10353 /* if we allocated too much space, give some back */
10354 if (SvCUR(sv) + 5 < SvLEN(sv)) {
10355 SvLEN_set(sv, SvCUR(sv) + 1);
10356 SvPV_renew(sv, SvLEN(sv));
10359 /* decide whether this is the first or second quoted string we've read
10372 takes: pointer to position in buffer
10373 returns: pointer to new position in buffer
10374 side-effects: builds ops for the constant in yylval.op
10376 Read a number in any of the formats that Perl accepts:
10378 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
10379 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
10382 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
10384 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
10387 If it reads a number without a decimal point or an exponent, it will
10388 try converting the number to an integer and see if it can do so
10389 without loss of precision.
10393 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
10395 register const char *s = start; /* current position in buffer */
10396 register char *d; /* destination in temp buffer */
10397 register char *e; /* end of temp buffer */
10398 NV nv; /* number read, as a double */
10399 SV *sv = Nullsv; /* place to put the converted number */
10400 bool floatit; /* boolean: int or float? */
10401 const char *lastub = 0; /* position of last underbar */
10402 static char const number_too_long[] = "Number too long";
10404 /* We use the first character to decide what type of number this is */
10408 Perl_croak(aTHX_ "panic: scan_num");
10410 /* if it starts with a 0, it could be an octal number, a decimal in
10411 0.13 disguise, or a hexadecimal number, or a binary number. */
10415 u holds the "number so far"
10416 shift the power of 2 of the base
10417 (hex == 4, octal == 3, binary == 1)
10418 overflowed was the number more than we can hold?
10420 Shift is used when we add a digit. It also serves as an "are
10421 we in octal/hex/binary?" indicator to disallow hex characters
10422 when in octal mode.
10427 bool overflowed = FALSE;
10428 bool just_zero = TRUE; /* just plain 0 or binary number? */
10429 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
10430 static const char* const bases[5] =
10431 { "", "binary", "", "octal", "hexadecimal" };
10432 static const char* const Bases[5] =
10433 { "", "Binary", "", "Octal", "Hexadecimal" };
10434 static const char* const maxima[5] =
10436 "0b11111111111111111111111111111111",
10440 const char *base, *Base, *max;
10442 /* check for hex */
10447 } else if (s[1] == 'b') {
10452 /* check for a decimal in disguise */
10453 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
10455 /* so it must be octal */
10462 if (ckWARN(WARN_SYNTAX))
10463 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10464 "Misplaced _ in number");
10468 base = bases[shift];
10469 Base = Bases[shift];
10470 max = maxima[shift];
10472 /* read the rest of the number */
10474 /* x is used in the overflow test,
10475 b is the digit we're adding on. */
10480 /* if we don't mention it, we're done */
10484 /* _ are ignored -- but warned about if consecutive */
10486 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10487 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10488 "Misplaced _ in number");
10492 /* 8 and 9 are not octal */
10493 case '8': case '9':
10495 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
10499 case '2': case '3': case '4':
10500 case '5': case '6': case '7':
10502 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
10505 case '0': case '1':
10506 b = *s++ & 15; /* ASCII digit -> value of digit */
10510 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
10511 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
10512 /* make sure they said 0x */
10515 b = (*s++ & 7) + 9;
10517 /* Prepare to put the digit we have onto the end
10518 of the number so far. We check for overflows.
10524 x = u << shift; /* make room for the digit */
10526 if ((x >> shift) != u
10527 && !(PL_hints & HINT_NEW_BINARY)) {
10530 if (ckWARN_d(WARN_OVERFLOW))
10531 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
10532 "Integer overflow in %s number",
10535 u = x | b; /* add the digit to the end */
10538 n *= nvshift[shift];
10539 /* If an NV has not enough bits in its
10540 * mantissa to represent an UV this summing of
10541 * small low-order numbers is a waste of time
10542 * (because the NV cannot preserve the
10543 * low-order bits anyway): we could just
10544 * remember when did we overflow and in the
10545 * end just multiply n by the right
10553 /* if we get here, we had success: make a scalar value from
10558 /* final misplaced underbar check */
10559 if (s[-1] == '_') {
10560 if (ckWARN(WARN_SYNTAX))
10561 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
10566 if (n > 4294967295.0 && ckWARN(WARN_PORTABLE))
10567 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
10568 "%s number > %s non-portable",
10574 if (u > 0xffffffff && ckWARN(WARN_PORTABLE))
10575 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
10576 "%s number > %s non-portable",
10581 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
10582 sv = new_constant(start, s - start, "integer",
10584 else if (PL_hints & HINT_NEW_BINARY)
10585 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
10590 handle decimal numbers.
10591 we're also sent here when we read a 0 as the first digit
10593 case '1': case '2': case '3': case '4': case '5':
10594 case '6': case '7': case '8': case '9': case '.':
10597 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
10600 /* read next group of digits and _ and copy into d */
10601 while (isDIGIT(*s) || *s == '_') {
10602 /* skip underscores, checking for misplaced ones
10606 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10607 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10608 "Misplaced _ in number");
10612 /* check for end of fixed-length buffer */
10614 Perl_croak(aTHX_ number_too_long);
10615 /* if we're ok, copy the character */
10620 /* final misplaced underbar check */
10621 if (lastub && s == lastub + 1) {
10622 if (ckWARN(WARN_SYNTAX))
10623 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
10626 /* read a decimal portion if there is one. avoid
10627 3..5 being interpreted as the number 3. followed
10630 if (*s == '.' && s[1] != '.') {
10635 if (ckWARN(WARN_SYNTAX))
10636 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10637 "Misplaced _ in number");
10641 /* copy, ignoring underbars, until we run out of digits.
10643 for (; isDIGIT(*s) || *s == '_'; s++) {
10644 /* fixed length buffer check */
10646 Perl_croak(aTHX_ number_too_long);
10648 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10649 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10650 "Misplaced _ in number");
10656 /* fractional part ending in underbar? */
10657 if (s[-1] == '_') {
10658 if (ckWARN(WARN_SYNTAX))
10659 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10660 "Misplaced _ in number");
10662 if (*s == '.' && isDIGIT(s[1])) {
10663 /* oops, it's really a v-string, but without the "v" */
10669 /* read exponent part, if present */
10670 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
10674 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
10675 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
10677 /* stray preinitial _ */
10679 if (ckWARN(WARN_SYNTAX))
10680 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10681 "Misplaced _ in number");
10685 /* allow positive or negative exponent */
10686 if (*s == '+' || *s == '-')
10689 /* stray initial _ */
10691 if (ckWARN(WARN_SYNTAX))
10692 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10693 "Misplaced _ in number");
10697 /* read digits of exponent */
10698 while (isDIGIT(*s) || *s == '_') {
10701 Perl_croak(aTHX_ number_too_long);
10705 if (((lastub && s == lastub + 1) ||
10706 (!isDIGIT(s[1]) && s[1] != '_'))
10707 && ckWARN(WARN_SYNTAX))
10708 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10709 "Misplaced _ in number");
10716 /* make an sv from the string */
10720 We try to do an integer conversion first if no characters
10721 indicating "float" have been found.
10726 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
10728 if (flags == IS_NUMBER_IN_UV) {
10730 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
10733 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
10734 if (uv <= (UV) IV_MIN)
10735 sv_setiv(sv, -(IV)uv);
10742 /* terminate the string */
10744 nv = Atof(PL_tokenbuf);
10748 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
10749 (PL_hints & HINT_NEW_INTEGER) )
10750 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
10751 (floatit ? "float" : "integer"),
10755 /* if it starts with a v, it could be a v-string */
10758 sv = NEWSV(92,5); /* preallocate storage space */
10759 s = scan_vstring(s,sv);
10763 /* make the op for the constant and return */
10766 lvalp->opval = newSVOP(OP_CONST, 0, sv);
10768 lvalp->opval = Nullop;
10774 S_scan_formline(pTHX_ register char *s)
10776 register char *eol;
10778 SV *stuff = newSVpvn("",0);
10779 bool needargs = FALSE;
10780 bool eofmt = FALSE;
10782 while (!needargs) {
10784 #ifdef PERL_STRICT_CR
10785 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
10787 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
10789 if (*t == '\n' || t == PL_bufend) {
10794 if (PL_in_eval && !PL_rsfp) {
10795 eol = (char *) memchr(s,'\n',PL_bufend-s);
10800 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
10802 for (t = s; t < eol; t++) {
10803 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
10805 goto enough; /* ~~ must be first line in formline */
10807 if (*t == '@' || *t == '^')
10811 sv_catpvn(stuff, s, eol-s);
10812 #ifndef PERL_STRICT_CR
10813 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
10814 char *end = SvPVX(stuff) + SvCUR(stuff);
10817 SvCUR_set(stuff, SvCUR(stuff) - 1);
10826 s = filter_gets(PL_linestr, PL_rsfp, 0);
10827 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
10828 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
10829 PL_last_lop = PL_last_uni = Nullch;
10838 if (SvCUR(stuff)) {
10841 PL_lex_state = LEX_NORMAL;
10842 PL_nextval[PL_nexttoke].ival = 0;
10846 PL_lex_state = LEX_FORMLINE;
10848 if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
10850 else if (PL_encoding)
10851 sv_recode_to_utf8(stuff, PL_encoding);
10853 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
10855 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
10859 SvREFCNT_dec(stuff);
10861 PL_lex_formbrack = 0;
10872 PL_cshlen = strlen(PL_cshname);
10877 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
10879 const I32 oldsavestack_ix = PL_savestack_ix;
10880 CV* outsidecv = PL_compcv;
10883 assert(SvTYPE(PL_compcv) == SVt_PVCV);
10885 SAVEI32(PL_subline);
10886 save_item(PL_subname);
10887 SAVESPTR(PL_compcv);
10889 PL_compcv = (CV*)NEWSV(1104,0);
10890 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
10891 CvFLAGS(PL_compcv) |= flags;
10893 PL_subline = CopLINE(PL_curcop);
10894 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
10895 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
10896 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
10898 return oldsavestack_ix;
10902 #pragma segment Perl_yylex
10905 Perl_yywarn(pTHX_ const char *s)
10907 PL_in_eval |= EVAL_WARNONLY;
10909 PL_in_eval &= ~EVAL_WARNONLY;
10914 Perl_yyerror(pTHX_ const char *s)
10916 const char *where = NULL;
10917 const char *context = NULL;
10921 if (!yychar || (yychar == ';' && !PL_rsfp))
10923 else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
10924 PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
10925 PL_oldbufptr != PL_bufptr) {
10928 The code below is removed for NetWare because it abends/crashes on NetWare
10929 when the script has error such as not having the closing quotes like:
10930 if ($var eq "value)
10931 Checking of white spaces is anyway done in NetWare code.
10934 while (isSPACE(*PL_oldoldbufptr))
10937 context = PL_oldoldbufptr;
10938 contlen = PL_bufptr - PL_oldoldbufptr;
10940 else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
10941 PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
10944 The code below is removed for NetWare because it abends/crashes on NetWare
10945 when the script has error such as not having the closing quotes like:
10946 if ($var eq "value)
10947 Checking of white spaces is anyway done in NetWare code.
10950 while (isSPACE(*PL_oldbufptr))
10953 context = PL_oldbufptr;
10954 contlen = PL_bufptr - PL_oldbufptr;
10956 else if (yychar > 255)
10957 where = "next token ???";
10958 else if (yychar == -2) { /* YYEMPTY */
10959 if (PL_lex_state == LEX_NORMAL ||
10960 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
10961 where = "at end of line";
10962 else if (PL_lex_inpat)
10963 where = "within pattern";
10965 where = "within string";
10968 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
10970 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
10971 else if (isPRINT_LC(yychar))
10972 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
10974 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
10975 where = SvPVX_const(where_sv);
10977 msg = sv_2mortal(newSVpv(s, 0));
10978 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
10979 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
10981 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
10983 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
10984 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
10985 Perl_sv_catpvf(aTHX_ msg,
10986 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
10987 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
10990 if (PL_in_eval & EVAL_WARNONLY && ckWARN_d(WARN_SYNTAX))
10991 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, msg);
10994 if (PL_error_count >= 10) {
10995 if (PL_in_eval && SvCUR(ERRSV))
10996 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
10997 ERRSV, OutCopFILE(PL_curcop));
10999 Perl_croak(aTHX_ "%s has too many errors.\n",
11000 OutCopFILE(PL_curcop));
11003 PL_in_my_stash = NULL;
11007 #pragma segment Main
11011 S_swallow_bom(pTHX_ U8 *s)
11013 const STRLEN slen = SvCUR(PL_linestr);
11016 if (s[1] == 0xFE) {
11017 /* UTF-16 little-endian? (or UTF32-LE?) */
11018 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
11019 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
11020 #ifndef PERL_NO_UTF16_FILTER
11021 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
11024 if (PL_bufend > (char*)s) {
11028 filter_add(utf16rev_textfilter, NULL);
11029 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
11030 utf16_to_utf8_reversed(s, news,
11031 PL_bufend - (char*)s - 1,
11033 sv_setpvn(PL_linestr, (const char*)news, newlen);
11035 SvUTF8_on(PL_linestr);
11036 s = (U8*)SvPVX(PL_linestr);
11037 PL_bufend = SvPVX(PL_linestr) + newlen;
11040 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
11045 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
11046 #ifndef PERL_NO_UTF16_FILTER
11047 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
11050 if (PL_bufend > (char *)s) {
11054 filter_add(utf16_textfilter, NULL);
11055 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
11056 utf16_to_utf8(s, news,
11057 PL_bufend - (char*)s,
11059 sv_setpvn(PL_linestr, (const char*)news, newlen);
11061 SvUTF8_on(PL_linestr);
11062 s = (U8*)SvPVX(PL_linestr);
11063 PL_bufend = SvPVX(PL_linestr) + newlen;
11066 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
11071 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
11072 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
11073 s += 3; /* UTF-8 */
11079 if (s[2] == 0xFE && s[3] == 0xFF) {
11080 /* UTF-32 big-endian */
11081 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
11084 else if (s[2] == 0 && s[3] != 0) {
11087 * are a good indicator of UTF-16BE. */
11088 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
11093 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
11096 * are a good indicator of UTF-16LE. */
11097 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
11106 * Restore a source filter.
11110 restore_rsfp(pTHX_ void *f)
11112 PerlIO * const fp = (PerlIO*)f;
11114 if (PL_rsfp == PerlIO_stdin())
11115 PerlIO_clearerr(PL_rsfp);
11116 else if (PL_rsfp && (PL_rsfp != fp))
11117 PerlIO_close(PL_rsfp);
11121 #ifndef PERL_NO_UTF16_FILTER
11123 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
11125 const STRLEN old = SvCUR(sv);
11126 const I32 count = FILTER_READ(idx+1, sv, maxlen);
11127 DEBUG_P(PerlIO_printf(Perl_debug_log,
11128 "utf16_textfilter(%p): %d %d (%d)\n",
11129 utf16_textfilter, idx, maxlen, (int) count));
11133 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
11134 Copy(SvPVX_const(sv), tmps, old, char);
11135 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
11136 SvCUR(sv) - old, &newlen);
11137 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
11139 DEBUG_P({sv_dump(sv);});
11144 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
11146 const STRLEN old = SvCUR(sv);
11147 const I32 count = FILTER_READ(idx+1, sv, maxlen);
11148 DEBUG_P(PerlIO_printf(Perl_debug_log,
11149 "utf16rev_textfilter(%p): %d %d (%d)\n",
11150 utf16rev_textfilter, idx, maxlen, (int) count));
11154 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
11155 Copy(SvPVX_const(sv), tmps, old, char);
11156 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
11157 SvCUR(sv) - old, &newlen);
11158 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
11160 DEBUG_P({ sv_dump(sv); });
11166 Returns a pointer to the next character after the parsed
11167 vstring, as well as updating the passed in sv.
11169 Function must be called like
11172 s = scan_vstring(s,sv);
11174 The sv should already be large enough to store the vstring
11175 passed in, for performance reasons.
11180 Perl_scan_vstring(pTHX_ const char *s, SV *sv)
11182 const char *pos = s;
11183 const char *start = s;
11184 if (*pos == 'v') pos++; /* get past 'v' */
11185 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
11187 if ( *pos != '.') {
11188 /* this may not be a v-string if followed by => */
11189 const char *next = pos;
11190 while (next < PL_bufend && isSPACE(*next))
11192 if ((PL_bufend - next) >= 2 && *next == '=' && next[1] == '>' ) {
11193 /* return string not v-string */
11194 sv_setpvn(sv,(char *)s,pos-s);
11195 return (char *)pos;
11199 if (!isALPHA(*pos)) {
11200 U8 tmpbuf[UTF8_MAXBYTES+1];
11202 if (*s == 'v') s++; /* get past 'v' */
11204 sv_setpvn(sv, "", 0);
11210 /* this is atoi() that tolerates underscores */
11211 const char *end = pos;
11213 while (--end >= s) {
11218 rev += (*end - '0') * mult;
11220 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
11221 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
11222 "Integer overflow in decimal number");
11226 if (rev > 0x7FFFFFFF)
11227 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
11229 /* Append native character for the rev point */
11230 tmpend = uvchr_to_utf8(tmpbuf, rev);
11231 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
11232 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
11234 if (pos + 1 < PL_bufend && *pos == '.' && isDIGIT(pos[1]))
11240 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
11244 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
11252 * c-indentation-style: bsd
11253 * c-basic-offset: 4
11254 * indent-tabs-mode: t
11257 * ex: set ts=8 sts=4 sw=4 noet: