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 $foo *foo &foo etc, but the program
1007 * text only contains the "foo" portion. The first argument is a pointer
1008 * to the "foo", and the second argument is the type symbol to prefix.
1009 * Forces the next token to be a "WORD".
1010 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
1014 S_force_ident(pTHX_ register const char *s, int kind)
1017 OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
1018 PL_nextval[PL_nexttoke].opval = o;
1021 o->op_private = OPpCONST_ENTERED;
1022 /* XXX see note in pp_entereval() for why we forgo typo
1023 warnings if the symbol must be introduced in an eval.
1025 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : GV_ADD,
1026 kind == '$' ? 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
2023 && gv_fetchpv(tmpbuf, 0, SVt_PV))
2028 else if (*s == '$' && s[1] &&
2029 strchr("[#!%*<>()-=",s[1])) {
2030 if (/*{*/ strchr("])} =",s[2]))
2039 if (strchr("wds]",s[1]))
2041 else if (seen['\''] || seen['"'])
2043 else if (strchr("rnftbxcav",s[1]))
2045 else if (isDIGIT(s[1])) {
2047 while (s[1] && isDIGIT(s[1]))
2057 if (strchr("aA01! ",last_un_char))
2059 if (strchr("zZ79~",s[1]))
2061 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
2062 weight -= 5; /* cope with negative subscript */
2065 if (!isALNUM(last_un_char)
2066 && !(last_un_char == '$' || last_un_char == '@'
2067 || last_un_char == '&')
2068 && isALPHA(*s) && s[1] && isALPHA(s[1])) {
2073 if (keyword(tmpbuf, d - tmpbuf))
2076 if (un_char == last_un_char + 1)
2078 weight -= seen[un_char];
2083 if (weight >= 0) /* probably a character class */
2093 * Does all the checking to disambiguate
2095 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
2096 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
2098 * First argument is the stuff after the first token, e.g. "bar".
2100 * Not a method if bar is a filehandle.
2101 * Not a method if foo is a subroutine prototyped to take a filehandle.
2102 * Not a method if it's really "Foo $bar"
2103 * Method if it's "foo $bar"
2104 * Not a method if it's really "print foo $bar"
2105 * Method if it's really "foo package::" (interpreted as package->foo)
2106 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
2107 * Not a method if bar is a filehandle or package, but is quoted with
2112 S_intuit_method(pTHX_ char *start, GV *gv, CV *cv)
2114 char *s = start + (*start == '$');
2115 char tmpbuf[sizeof PL_tokenbuf];
2120 if (SvTYPE(gv) == SVt_PVGV && GvIO(gv))
2124 const char *proto = SvPVX_const(cv);
2135 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
2136 /* start is the beginning of the possible filehandle/object,
2137 * and s is the end of it
2138 * tmpbuf is a copy of it
2141 if (*start == '$') {
2142 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
2147 return *s == '(' ? FUNCMETH : METHOD;
2149 if (!keyword(tmpbuf, len)) {
2150 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
2155 indirgv = gv_fetchpv(tmpbuf, 0, SVt_PVCV);
2156 if (indirgv && GvCVu(indirgv))
2158 /* filehandle or package name makes it a method */
2159 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
2161 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
2162 return 0; /* no assumptions -- "=>" quotes bearword */
2164 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
2165 newSVpvn(tmpbuf,len));
2166 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
2170 return *s == '(' ? FUNCMETH : METHOD;
2178 * Return a string of Perl code to load the debugger. If PERL5DB
2179 * is set, it will return the contents of that, otherwise a
2180 * compile-time require of perl5db.pl.
2187 const char * const pdb = PerlEnv_getenv("PERL5DB");
2191 SETERRNO(0,SS_NORMAL);
2192 return "BEGIN { require 'perl5db.pl' }";
2198 /* Encoded script support. filter_add() effectively inserts a
2199 * 'pre-processing' function into the current source input stream.
2200 * Note that the filter function only applies to the current source file
2201 * (e.g., it will not affect files 'require'd or 'use'd by this one).
2203 * The datasv parameter (which may be NULL) can be used to pass
2204 * private data to this instance of the filter. The filter function
2205 * can recover the SV using the FILTER_DATA macro and use it to
2206 * store private buffers and state information.
2208 * The supplied datasv parameter is upgraded to a PVIO type
2209 * and the IoDIRP/IoANY field is used to store the function pointer,
2210 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
2211 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
2212 * private use must be set using malloc'd pointers.
2216 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
2221 if (!PL_rsfp_filters)
2222 PL_rsfp_filters = newAV();
2224 datasv = NEWSV(255,0);
2225 SvUPGRADE(datasv, SVt_PVIO);
2226 IoANY(datasv) = FPTR2DPTR(void *, funcp); /* stash funcp into spare field */
2227 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2228 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2229 IoANY(datasv), SvPV_nolen(datasv)));
2230 av_unshift(PL_rsfp_filters, 1);
2231 av_store(PL_rsfp_filters, 0, datasv) ;
2236 /* Delete most recently added instance of this filter function. */
2238 Perl_filter_del(pTHX_ filter_t funcp)
2243 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", FPTR2DPTR(XPVIO *, funcp)));
2245 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2247 /* if filter is on top of stack (usual case) just pop it off */
2248 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2249 if (IoANY(datasv) == FPTR2DPTR(void *, funcp)) {
2250 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2251 IoANY(datasv) = (void *)NULL;
2252 sv_free(av_pop(PL_rsfp_filters));
2256 /* we need to search for the correct entry and clear it */
2257 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2261 /* Invoke the idxth filter function for the current rsfp. */
2262 /* maxlen 0 = read one text line */
2264 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2269 if (!PL_rsfp_filters)
2271 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
2272 /* Provide a default input filter to make life easy. */
2273 /* Note that we append to the line. This is handy. */
2274 DEBUG_P(PerlIO_printf(Perl_debug_log,
2275 "filter_read %d: from rsfp\n", idx));
2279 const int old_len = SvCUR(buf_sv);
2281 /* ensure buf_sv is large enough */
2282 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
2283 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2284 if (PerlIO_error(PL_rsfp))
2285 return -1; /* error */
2287 return 0 ; /* end of file */
2289 SvCUR_set(buf_sv, old_len + len) ;
2292 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2293 if (PerlIO_error(PL_rsfp))
2294 return -1; /* error */
2296 return 0 ; /* end of file */
2299 return SvCUR(buf_sv);
2301 /* Skip this filter slot if filter has been deleted */
2302 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
2303 DEBUG_P(PerlIO_printf(Perl_debug_log,
2304 "filter_read %d: skipped (filter deleted)\n",
2306 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2308 /* Get function pointer hidden within datasv */
2309 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
2310 DEBUG_P(PerlIO_printf(Perl_debug_log,
2311 "filter_read %d: via function %p (%s)\n",
2312 idx, datasv, SvPV_nolen_const(datasv)));
2313 /* Call function. The function is expected to */
2314 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2315 /* Return: <0:error, =0:eof, >0:not eof */
2316 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2320 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2322 #ifdef PERL_CR_FILTER
2323 if (!PL_rsfp_filters) {
2324 filter_add(S_cr_textfilter,NULL);
2327 if (PL_rsfp_filters) {
2329 SvCUR_set(sv, 0); /* start with empty line */
2330 if (FILTER_READ(0, sv, 0) > 0)
2331 return ( SvPVX(sv) ) ;
2336 return (sv_gets(sv, fp, append));
2340 S_find_in_my_stash(pTHX_ const char *pkgname, I32 len)
2344 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2348 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2349 (gv = gv_fetchpv(pkgname, 0, SVt_PVHV)))
2351 return GvHV(gv); /* Foo:: */
2354 /* use constant CLASS => 'MyClass' */
2355 if ((gv = gv_fetchpv(pkgname, 0, SVt_PVCV))) {
2357 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2358 pkgname = SvPV_nolen_const(sv);
2362 return gv_stashpv(pkgname, FALSE);
2366 S_tokenize_use(pTHX_ int is_use, char *s) {
2367 if (PL_expect != XSTATE)
2368 yyerror(Perl_form(aTHX_ "\"%s\" not allowed in expression",
2369 is_use ? "use" : "no"));
2371 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
2372 s = force_version(s, TRUE);
2373 if (*s == ';' || (s = skipspace(s), *s == ';')) {
2374 PL_nextval[PL_nexttoke].opval = Nullop;
2377 else if (*s == 'v') {
2378 s = force_word(s,WORD,FALSE,TRUE,FALSE);
2379 s = force_version(s, FALSE);
2383 s = force_word(s,WORD,FALSE,TRUE,FALSE);
2384 s = force_version(s, FALSE);
2386 yylval.ival = is_use;
2390 static const char* const exp_name[] =
2391 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2392 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
2399 Works out what to call the token just pulled out of the input
2400 stream. The yacc parser takes care of taking the ops we return and
2401 stitching them into a tree.
2407 if read an identifier
2408 if we're in a my declaration
2409 croak if they tried to say my($foo::bar)
2410 build the ops for a my() declaration
2411 if it's an access to a my() variable
2412 are we in a sort block?
2413 croak if my($a); $a <=> $b
2414 build ops for access to a my() variable
2415 if in a dq string, and they've said @foo and we can't find @foo
2417 build ops for a bareword
2418 if we already built the token before, use it.
2423 #pragma segment Perl_yylex
2428 register char *s = PL_bufptr;
2434 SV* tmp = newSVpvn("", 0);
2435 PerlIO_printf(Perl_debug_log, "### %"IVdf":LEX_%s/X%s %s\n",
2436 (IV)CopLINE(PL_curcop),
2437 lex_state_names[PL_lex_state],
2438 exp_name[PL_expect],
2439 pv_display(tmp, s, strlen(s), 0, 60));
2442 /* check if there's an identifier for us to look at */
2443 if (PL_pending_ident)
2444 return REPORT(S_pending_ident(aTHX));
2446 /* no identifier pending identification */
2448 switch (PL_lex_state) {
2450 case LEX_NORMAL: /* Some compilers will produce faster */
2451 case LEX_INTERPNORMAL: /* code if we comment these out. */
2455 /* when we've already built the next token, just pull it out of the queue */
2458 yylval = PL_nextval[PL_nexttoke];
2460 PL_lex_state = PL_lex_defer;
2461 PL_expect = PL_lex_expect;
2462 PL_lex_defer = LEX_NORMAL;
2464 return REPORT(PL_nexttype[PL_nexttoke]);
2466 /* interpolated case modifiers like \L \U, including \Q and \E.
2467 when we get here, PL_bufptr is at the \
2469 case LEX_INTERPCASEMOD:
2471 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2472 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2474 /* handle \E or end of string */
2475 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2477 if (PL_lex_casemods) {
2478 const char oldmod = PL_lex_casestack[--PL_lex_casemods];
2479 PL_lex_casestack[PL_lex_casemods] = '\0';
2481 if (PL_bufptr != PL_bufend
2482 && (oldmod == 'L' || oldmod == 'U' || oldmod == 'Q')) {
2484 PL_lex_state = LEX_INTERPCONCAT;
2488 if (PL_bufptr != PL_bufend)
2490 PL_lex_state = LEX_INTERPCONCAT;
2494 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2495 "### Saw case modifier\n"); });
2497 if (s[1] == '\\' && s[2] == 'E') {
2499 PL_lex_state = LEX_INTERPCONCAT;
2504 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2505 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
2506 if ((*s == 'L' || *s == 'U') &&
2507 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
2508 PL_lex_casestack[--PL_lex_casemods] = '\0';
2511 if (PL_lex_casemods > 10)
2512 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2513 PL_lex_casestack[PL_lex_casemods++] = *s;
2514 PL_lex_casestack[PL_lex_casemods] = '\0';
2515 PL_lex_state = LEX_INTERPCONCAT;
2516 PL_nextval[PL_nexttoke].ival = 0;
2519 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2521 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2523 PL_nextval[PL_nexttoke].ival = OP_LC;
2525 PL_nextval[PL_nexttoke].ival = OP_UC;
2527 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2529 Perl_croak(aTHX_ "panic: yylex");
2533 if (PL_lex_starts) {
2536 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2537 if (PL_lex_casemods == 1 && PL_lex_inpat)
2546 case LEX_INTERPPUSH:
2547 return REPORT(sublex_push());
2549 case LEX_INTERPSTART:
2550 if (PL_bufptr == PL_bufend)
2551 return REPORT(sublex_done());
2552 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2553 "### Interpolated variable\n"); });
2555 PL_lex_dojoin = (*PL_bufptr == '@');
2556 PL_lex_state = LEX_INTERPNORMAL;
2557 if (PL_lex_dojoin) {
2558 PL_nextval[PL_nexttoke].ival = 0;
2560 force_ident("\"", '$');
2561 PL_nextval[PL_nexttoke].ival = 0;
2563 PL_nextval[PL_nexttoke].ival = 0;
2565 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2568 if (PL_lex_starts++) {
2570 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2571 if (!PL_lex_casemods && PL_lex_inpat)
2578 case LEX_INTERPENDMAYBE:
2579 if (intuit_more(PL_bufptr)) {
2580 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2586 if (PL_lex_dojoin) {
2587 PL_lex_dojoin = FALSE;
2588 PL_lex_state = LEX_INTERPCONCAT;
2591 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2592 && SvEVALED(PL_lex_repl))
2594 if (PL_bufptr != PL_bufend)
2595 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2596 PL_lex_repl = Nullsv;
2599 case LEX_INTERPCONCAT:
2601 if (PL_lex_brackets)
2602 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2604 if (PL_bufptr == PL_bufend)
2605 return REPORT(sublex_done());
2607 if (SvIVX(PL_linestr) == '\'') {
2608 SV *sv = newSVsv(PL_linestr);
2611 else if ( PL_hints & HINT_NEW_RE )
2612 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2613 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2617 s = scan_const(PL_bufptr);
2619 PL_lex_state = LEX_INTERPCASEMOD;
2621 PL_lex_state = LEX_INTERPSTART;
2624 if (s != PL_bufptr) {
2625 PL_nextval[PL_nexttoke] = yylval;
2628 if (PL_lex_starts++) {
2629 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2630 if (!PL_lex_casemods && PL_lex_inpat)
2643 PL_lex_state = LEX_NORMAL;
2644 s = scan_formline(PL_bufptr);
2645 if (!PL_lex_formbrack)
2651 PL_oldoldbufptr = PL_oldbufptr;
2657 if (isIDFIRST_lazy_if(s,UTF))
2659 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2662 goto fake_eof; /* emulate EOF on ^D or ^Z */
2667 if (PL_lex_brackets) {
2668 yyerror(PL_lex_formbrack
2669 ? "Format not terminated"
2670 : "Missing right curly or square bracket");
2672 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2673 "### Tokener got EOF\n");
2677 if (s++ < PL_bufend)
2678 goto retry; /* ignore stray nulls */
2681 if (!PL_in_eval && !PL_preambled) {
2682 PL_preambled = TRUE;
2683 sv_setpv(PL_linestr,incl_perldb());
2684 if (SvCUR(PL_linestr))
2685 sv_catpvn(PL_linestr,";", 1);
2687 while(AvFILLp(PL_preambleav) >= 0) {
2688 SV *tmpsv = av_shift(PL_preambleav);
2689 sv_catsv(PL_linestr, tmpsv);
2690 sv_catpvn(PL_linestr, ";", 1);
2693 sv_free((SV*)PL_preambleav);
2694 PL_preambleav = NULL;
2696 if (PL_minus_n || PL_minus_p) {
2697 sv_catpv(PL_linestr, "LINE: while (<>) {");
2699 sv_catpv(PL_linestr,"chomp;");
2702 if ((*PL_splitstr == '/' || *PL_splitstr == '\''
2703 || *PL_splitstr == '"')
2704 && strchr(PL_splitstr + 1, *PL_splitstr))
2705 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
2707 /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
2708 bytes can be used as quoting characters. :-) */
2709 /* The count here deliberately includes the NUL
2710 that terminates the C string constant. This
2711 embeds the opening NUL into the string. */
2712 const char *splits = PL_splitstr;
2713 sv_catpvn(PL_linestr, "our @F=split(q", 15);
2716 if (*splits == '\\')
2717 sv_catpvn(PL_linestr, splits, 1);
2718 sv_catpvn(PL_linestr, splits, 1);
2719 } while (*splits++);
2720 /* This loop will embed the trailing NUL of
2721 PL_linestr as the last thing it does before
2723 sv_catpvn(PL_linestr, ");", 2);
2727 sv_catpv(PL_linestr,"our @F=split(' ');");
2730 sv_catpvn(PL_linestr, "\n", 1);
2731 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2732 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2733 PL_last_lop = PL_last_uni = Nullch;
2734 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2735 SV * const sv = NEWSV(85,0);
2737 sv_upgrade(sv, SVt_PVMG);
2738 sv_setsv(sv,PL_linestr);
2741 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2746 bof = PL_rsfp ? TRUE : FALSE;
2747 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2750 if (PL_preprocess && !PL_in_eval)
2751 (void)PerlProc_pclose(PL_rsfp);
2752 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2753 PerlIO_clearerr(PL_rsfp);
2755 (void)PerlIO_close(PL_rsfp);
2757 PL_doextract = FALSE;
2759 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2760 sv_setpv(PL_linestr,PL_minus_p
2761 ? ";}continue{print;}" : ";}");
2762 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2763 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2764 PL_last_lop = PL_last_uni = Nullch;
2765 PL_minus_n = PL_minus_p = 0;
2768 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2769 PL_last_lop = PL_last_uni = Nullch;
2770 sv_setpvn(PL_linestr,"",0);
2771 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2773 /* If it looks like the start of a BOM or raw UTF-16,
2774 * check if it in fact is. */
2780 #ifdef PERLIO_IS_STDIO
2781 # ifdef __GNU_LIBRARY__
2782 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2783 # define FTELL_FOR_PIPE_IS_BROKEN
2787 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2788 # define FTELL_FOR_PIPE_IS_BROKEN
2793 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2794 /* This loses the possibility to detect the bof
2795 * situation on perl -P when the libc5 is being used.
2796 * Workaround? Maybe attach some extra state to PL_rsfp?
2799 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2801 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
2804 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2805 s = swallow_bom((U8*)s);
2809 /* Incest with pod. */
2810 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2811 sv_setpvn(PL_linestr, "", 0);
2812 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2813 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2814 PL_last_lop = PL_last_uni = Nullch;
2815 PL_doextract = FALSE;
2819 } while (PL_doextract);
2820 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2821 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2822 SV * const sv = NEWSV(85,0);
2824 sv_upgrade(sv, SVt_PVMG);
2825 sv_setsv(sv,PL_linestr);
2828 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2830 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2831 PL_last_lop = PL_last_uni = Nullch;
2832 if (CopLINE(PL_curcop) == 1) {
2833 while (s < PL_bufend && isSPACE(*s))
2835 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2839 if (*s == '#' && *(s+1) == '!')
2841 #ifdef ALTERNATE_SHEBANG
2843 static char const as[] = ALTERNATE_SHEBANG;
2844 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2845 d = s + (sizeof(as) - 1);
2847 #endif /* ALTERNATE_SHEBANG */
2856 while (*d && !isSPACE(*d))
2860 #ifdef ARG_ZERO_IS_SCRIPT
2861 if (ipathend > ipath) {
2863 * HP-UX (at least) sets argv[0] to the script name,
2864 * which makes $^X incorrect. And Digital UNIX and Linux,
2865 * at least, set argv[0] to the basename of the Perl
2866 * interpreter. So, having found "#!", we'll set it right.
2869 = GvSV(gv_fetchpv("\030", GV_ADD, SVt_PV)); /* $^X */
2870 assert(SvPOK(x) || SvGMAGICAL(x));
2871 if (sv_eq(x, CopFILESV(PL_curcop))) {
2872 sv_setpvn(x, ipath, ipathend - ipath);
2878 const char *bstart = SvPV_const(CopFILESV(PL_curcop),blen);
2879 const char * const lstart = SvPV_const(x,llen);
2881 bstart += blen - llen;
2882 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
2883 sv_setpvn(x, ipath, ipathend - ipath);
2888 TAINT_NOT; /* $^X is always tainted, but that's OK */
2890 #endif /* ARG_ZERO_IS_SCRIPT */
2895 d = instr(s,"perl -");
2897 d = instr(s,"perl");
2899 /* avoid getting into infinite loops when shebang
2900 * line contains "Perl" rather than "perl" */
2902 for (d = ipathend-4; d >= ipath; --d) {
2903 if ((*d == 'p' || *d == 'P')
2904 && !ibcmp(d, "perl", 4))
2914 #ifdef ALTERNATE_SHEBANG
2916 * If the ALTERNATE_SHEBANG on this system starts with a
2917 * character that can be part of a Perl expression, then if
2918 * we see it but not "perl", we're probably looking at the
2919 * start of Perl code, not a request to hand off to some
2920 * other interpreter. Similarly, if "perl" is there, but
2921 * not in the first 'word' of the line, we assume the line
2922 * contains the start of the Perl program.
2924 if (d && *s != '#') {
2925 const char *c = ipath;
2926 while (*c && !strchr("; \t\r\n\f\v#", *c))
2929 d = Nullch; /* "perl" not in first word; ignore */
2931 *s = '#'; /* Don't try to parse shebang line */
2933 #endif /* ALTERNATE_SHEBANG */
2934 #ifndef MACOS_TRADITIONAL
2939 !instr(s,"indir") &&
2940 instr(PL_origargv[0],"perl"))
2947 while (s < PL_bufend && isSPACE(*s))
2949 if (s < PL_bufend) {
2950 Newxz(newargv,PL_origargc+3,char*);
2952 while (s < PL_bufend && !isSPACE(*s))
2955 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2958 newargv = PL_origargv;
2961 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2963 Perl_croak(aTHX_ "Can't exec %s", ipath);
2967 const U32 oldpdb = PL_perldb;
2968 const bool oldn = PL_minus_n;
2969 const bool oldp = PL_minus_p;
2971 while (*d && !isSPACE(*d)) d++;
2972 while (SPACE_OR_TAB(*d)) d++;
2975 const bool switches_done = PL_doswitches;
2977 if (*d == 'M' || *d == 'm' || *d == 'C') {
2978 const char * const m = d;
2979 while (*d && !isSPACE(*d)) d++;
2980 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2983 d = moreswitches(d);
2985 if (PL_doswitches && !switches_done) {
2986 int argc = PL_origargc;
2987 char **argv = PL_origargv;
2990 } while (argc && argv[0][0] == '-' && argv[0][1]);
2991 init_argv_symbols(argc,argv);
2993 if ((PERLDB_LINE && !oldpdb) ||
2994 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2995 /* if we have already added "LINE: while (<>) {",
2996 we must not do it again */
2998 sv_setpvn(PL_linestr, "", 0);
2999 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3000 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3001 PL_last_lop = PL_last_uni = Nullch;
3002 PL_preambled = FALSE;
3004 (void)gv_fetchfile(PL_origfilename);
3007 if (PL_doswitches && !switches_done) {
3008 int argc = PL_origargc;
3009 char **argv = PL_origargv;
3012 } while (argc && argv[0][0] == '-' && argv[0][1]);
3013 init_argv_symbols(argc,argv);
3019 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
3021 PL_lex_state = LEX_FORMLINE;
3026 #ifdef PERL_STRICT_CR
3027 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
3029 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
3031 case ' ': case '\t': case '\f': case 013:
3032 #ifdef MACOS_TRADITIONAL
3039 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
3040 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
3041 /* handle eval qq[#line 1 "foo"\n ...] */
3042 CopLINE_dec(PL_curcop);
3046 while (s < d && *s != '\n')
3050 else if (s > d) /* Found by Ilya: feed random input to Perl. */
3051 Perl_croak(aTHX_ "panic: input overflow");
3053 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
3055 PL_lex_state = LEX_FORMLINE;
3065 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
3073 while (s < PL_bufend && SPACE_OR_TAB(*s))
3076 if (strnEQ(s,"=>",2)) {
3077 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
3078 DEBUG_T( { S_printbuf(aTHX_
3079 "### Saw unary minus before =>, forcing word %s\n", s);
3081 OPERATOR('-'); /* unary minus */
3083 PL_last_uni = PL_oldbufptr;
3085 case 'r': ftst = OP_FTEREAD; break;
3086 case 'w': ftst = OP_FTEWRITE; break;
3087 case 'x': ftst = OP_FTEEXEC; break;
3088 case 'o': ftst = OP_FTEOWNED; break;
3089 case 'R': ftst = OP_FTRREAD; break;
3090 case 'W': ftst = OP_FTRWRITE; break;
3091 case 'X': ftst = OP_FTREXEC; break;
3092 case 'O': ftst = OP_FTROWNED; break;
3093 case 'e': ftst = OP_FTIS; break;
3094 case 'z': ftst = OP_FTZERO; break;
3095 case 's': ftst = OP_FTSIZE; break;
3096 case 'f': ftst = OP_FTFILE; break;
3097 case 'd': ftst = OP_FTDIR; break;
3098 case 'l': ftst = OP_FTLINK; break;
3099 case 'p': ftst = OP_FTPIPE; break;
3100 case 'S': ftst = OP_FTSOCK; break;
3101 case 'u': ftst = OP_FTSUID; break;
3102 case 'g': ftst = OP_FTSGID; break;
3103 case 'k': ftst = OP_FTSVTX; break;
3104 case 'b': ftst = OP_FTBLK; break;
3105 case 'c': ftst = OP_FTCHR; break;
3106 case 't': ftst = OP_FTTTY; break;
3107 case 'T': ftst = OP_FTTEXT; break;
3108 case 'B': ftst = OP_FTBINARY; break;
3109 case 'M': case 'A': case 'C':
3110 gv_fetchpv("\024",GV_ADD, SVt_PV);
3112 case 'M': ftst = OP_FTMTIME; break;
3113 case 'A': ftst = OP_FTATIME; break;
3114 case 'C': ftst = OP_FTCTIME; break;
3122 PL_last_lop_op = (OPCODE)ftst;
3123 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3124 "### Saw file test %c\n", (int)tmp);
3129 /* Assume it was a minus followed by a one-letter named
3130 * subroutine call (or a -bareword), then. */
3131 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3132 "### '-%c' looked like a file test but was not\n",
3139 const char tmp = *s++;
3142 if (PL_expect == XOPERATOR)
3147 else if (*s == '>') {
3150 if (isIDFIRST_lazy_if(s,UTF)) {
3151 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
3159 if (PL_expect == XOPERATOR)
3162 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
3164 OPERATOR('-'); /* unary minus */
3170 const char tmp = *s++;
3173 if (PL_expect == XOPERATOR)
3178 if (PL_expect == XOPERATOR)
3181 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
3188 if (PL_expect != XOPERATOR) {
3189 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3190 PL_expect = XOPERATOR;
3191 force_ident(PL_tokenbuf, '*');
3204 if (PL_expect == XOPERATOR) {
3208 PL_tokenbuf[0] = '%';
3209 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
3210 if (!PL_tokenbuf[1]) {
3213 PL_pending_ident = '%';
3224 && (PL_expect == XOPERATOR || PL_expect == XTERMORDORDOR)
3225 && FEATURE_IS_ENABLED("~~", 2))
3232 const char tmp = *s++;
3238 goto just_a_word_zero_gv;
3241 switch (PL_expect) {
3244 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
3246 PL_bufptr = s; /* update in case we back off */
3252 PL_expect = XTERMBLOCK;
3256 while (isIDFIRST_lazy_if(s,UTF)) {
3258 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3259 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
3260 if (tmp < 0) tmp = -tmp;
3276 d = scan_str(d,TRUE,TRUE);
3278 /* MUST advance bufptr here to avoid bogus
3279 "at end of line" context messages from yyerror().
3281 PL_bufptr = s + len;
3282 yyerror("Unterminated attribute parameter in attribute list");
3285 return REPORT(0); /* EOF indicator */
3289 SV *sv = newSVpvn(s, len);
3290 sv_catsv(sv, PL_lex_stuff);
3291 attrs = append_elem(OP_LIST, attrs,
3292 newSVOP(OP_CONST, 0, sv));
3293 SvREFCNT_dec(PL_lex_stuff);
3294 PL_lex_stuff = Nullsv;
3297 if (len == 6 && strnEQ(s, "unique", len)) {
3298 if (PL_in_my == KEY_our)
3300 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
3302 ; /* skip to avoid loading attributes.pm */
3305 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
3308 /* NOTE: any CV attrs applied here need to be part of
3309 the CVf_BUILTIN_ATTRS define in cv.h! */
3310 else if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3311 CvLVALUE_on(PL_compcv);
3312 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3313 CvLOCKED_on(PL_compcv);
3314 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3315 CvMETHOD_on(PL_compcv);
3316 else if (!PL_in_my && len == 9 && strnEQ(s, "assertion", len))
3317 CvASSERTION_on(PL_compcv);
3318 /* After we've set the flags, it could be argued that
3319 we don't need to do the attributes.pm-based setting
3320 process, and shouldn't bother appending recognized
3321 flags. To experiment with that, uncomment the
3322 following "else". (Note that's already been
3323 uncommented. That keeps the above-applied built-in
3324 attributes from being intercepted (and possibly
3325 rejected) by a package's attribute routines, but is
3326 justified by the performance win for the common case
3327 of applying only built-in attributes.) */
3329 attrs = append_elem(OP_LIST, attrs,
3330 newSVOP(OP_CONST, 0,
3334 if (*s == ':' && s[1] != ':')
3337 break; /* require real whitespace or :'s */
3341 = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3342 if (*s != ';' && *s != '}' && *s != tmp
3343 && (tmp != '=' || *s != ')')) {
3344 const char q = ((*s == '\'') ? '"' : '\'');
3345 /* If here for an expression, and parsed no attrs, back
3347 if (tmp == '=' && !attrs) {
3351 /* MUST advance bufptr here to avoid bogus "at end of line"
3352 context messages from yyerror().
3356 ? Perl_form(aTHX_ "Invalid separator character "
3357 "%c%c%c in attribute list", q, *s, q)
3358 : "Unterminated attribute list" );
3366 PL_nextval[PL_nexttoke].opval = attrs;
3374 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3375 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3383 const char tmp = *s++;
3388 const char tmp = *s++;
3396 if (PL_lex_brackets <= 0)
3397 yyerror("Unmatched right square bracket");
3400 if (PL_lex_state == LEX_INTERPNORMAL) {
3401 if (PL_lex_brackets == 0) {
3402 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3403 PL_lex_state = LEX_INTERPEND;
3410 if (PL_lex_brackets > 100) {
3411 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
3413 switch (PL_expect) {
3415 if (PL_lex_formbrack) {
3419 if (PL_oldoldbufptr == PL_last_lop)
3420 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3422 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3423 OPERATOR(HASHBRACK);
3425 while (s < PL_bufend && SPACE_OR_TAB(*s))
3428 PL_tokenbuf[0] = '\0';
3429 if (d < PL_bufend && *d == '-') {
3430 PL_tokenbuf[0] = '-';
3432 while (d < PL_bufend && SPACE_OR_TAB(*d))
3435 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3436 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3438 while (d < PL_bufend && SPACE_OR_TAB(*d))
3441 const char minus = (PL_tokenbuf[0] == '-');
3442 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3450 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3455 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3460 if (PL_oldoldbufptr == PL_last_lop)
3461 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3463 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3466 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3468 /* This hack is to get the ${} in the message. */
3470 yyerror("syntax error");
3473 OPERATOR(HASHBRACK);
3475 /* This hack serves to disambiguate a pair of curlies
3476 * as being a block or an anon hash. Normally, expectation
3477 * determines that, but in cases where we're not in a
3478 * position to expect anything in particular (like inside
3479 * eval"") we have to resolve the ambiguity. This code
3480 * covers the case where the first term in the curlies is a
3481 * quoted string. Most other cases need to be explicitly
3482 * disambiguated by prepending a "+" before the opening
3483 * curly in order to force resolution as an anon hash.
3485 * XXX should probably propagate the outer expectation
3486 * into eval"" to rely less on this hack, but that could
3487 * potentially break current behavior of eval"".
3491 if (*s == '\'' || *s == '"' || *s == '`') {
3492 /* common case: get past first string, handling escapes */
3493 for (t++; t < PL_bufend && *t != *s;)
3494 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3498 else if (*s == 'q') {
3501 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3504 /* skip q//-like construct */
3506 char open, close, term;
3509 while (t < PL_bufend && isSPACE(*t))
3511 /* check for q => */
3512 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
3513 OPERATOR(HASHBRACK);
3517 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3521 for (t++; t < PL_bufend; t++) {
3522 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3524 else if (*t == open)
3528 for (t++; t < PL_bufend; t++) {
3529 if (*t == '\\' && t+1 < PL_bufend)
3531 else if (*t == close && --brackets <= 0)
3533 else if (*t == open)
3540 /* skip plain q word */
3541 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3544 else if (isALNUM_lazy_if(t,UTF)) {
3546 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3549 while (t < PL_bufend && isSPACE(*t))
3551 /* if comma follows first term, call it an anon hash */
3552 /* XXX it could be a comma expression with loop modifiers */
3553 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3554 || (*t == '=' && t[1] == '>')))
3555 OPERATOR(HASHBRACK);
3556 if (PL_expect == XREF)
3559 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3565 yylval.ival = CopLINE(PL_curcop);
3566 if (isSPACE(*s) || *s == '#')
3567 PL_copline = NOLINE; /* invalidate current command line number */
3572 if (PL_lex_brackets <= 0)
3573 yyerror("Unmatched right curly bracket");
3575 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3576 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3577 PL_lex_formbrack = 0;
3578 if (PL_lex_state == LEX_INTERPNORMAL) {
3579 if (PL_lex_brackets == 0) {
3580 if (PL_expect & XFAKEBRACK) {
3581 PL_expect &= XENUMMASK;
3582 PL_lex_state = LEX_INTERPEND;
3584 return yylex(); /* ignore fake brackets */
3586 if (*s == '-' && s[1] == '>')
3587 PL_lex_state = LEX_INTERPENDMAYBE;
3588 else if (*s != '[' && *s != '{')
3589 PL_lex_state = LEX_INTERPEND;
3592 if (PL_expect & XFAKEBRACK) {
3593 PL_expect &= XENUMMASK;
3595 return yylex(); /* ignore fake brackets */
3604 if (PL_expect == XOPERATOR) {
3605 if (PL_bufptr == PL_linestart && ckWARN(WARN_SEMICOLON)
3606 && isIDFIRST_lazy_if(s,UTF))
3608 CopLINE_dec(PL_curcop);
3609 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3610 CopLINE_inc(PL_curcop);
3615 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3617 PL_expect = XOPERATOR;
3618 force_ident(PL_tokenbuf, '&');
3622 yylval.ival = (OPpENTERSUB_AMPER<<8);
3634 const char tmp = *s++;
3641 if (tmp && isSPACE(*s) && ckWARN(WARN_SYNTAX)
3642 && strchr("+-*/%.^&|<",tmp))
3643 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3644 "Reversed %c= operator",(int)tmp);
3646 if (PL_expect == XSTATE && isALPHA(tmp) &&
3647 (s == PL_linestart+1 || s[-2] == '\n') )
3649 if (PL_in_eval && !PL_rsfp) {
3654 if (strnEQ(s,"=cut",4)) {
3668 PL_doextract = TRUE;
3672 if (PL_lex_brackets < PL_lex_formbrack) {
3674 #ifdef PERL_STRICT_CR
3675 for (t = s; SPACE_OR_TAB(*t); t++) ;
3677 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3679 if (*t == '\n' || *t == '#') {
3690 const char tmp = *s++;
3692 /* was this !=~ where !~ was meant?
3693 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
3695 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
3696 const char *t = s+1;
3698 while (t < PL_bufend && isSPACE(*t))
3701 if (*t == '/' || *t == '?' ||
3702 ((*t == 'm' || *t == 's' || *t == 'y')
3703 && !isALNUM(t[1])) ||
3704 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
3705 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3706 "!=~ should be !~");
3716 if (PL_expect != XOPERATOR) {
3717 if (s[1] != '<' && !strchr(s,'>'))
3720 s = scan_heredoc(s);
3722 s = scan_inputsymbol(s);
3723 TERM(sublex_start());
3729 SHop(OP_LEFT_SHIFT);
3743 const char tmp = *s++;
3745 SHop(OP_RIGHT_SHIFT);
3755 if (PL_expect == XOPERATOR) {
3756 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3759 return REPORT(','); /* grandfather non-comma-format format */
3763 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3764 PL_tokenbuf[0] = '@';
3765 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3766 sizeof PL_tokenbuf - 1, FALSE);
3767 if (PL_expect == XOPERATOR)
3768 no_op("Array length", s);
3769 if (!PL_tokenbuf[1])
3771 PL_expect = XOPERATOR;
3772 PL_pending_ident = '#';
3776 PL_tokenbuf[0] = '$';
3777 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3778 sizeof PL_tokenbuf - 1, FALSE);
3779 if (PL_expect == XOPERATOR)
3781 if (!PL_tokenbuf[1]) {
3783 yyerror("Final $ should be \\$ or $name");
3787 /* This kludge not intended to be bulletproof. */
3788 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3789 yylval.opval = newSVOP(OP_CONST, 0,
3790 newSViv(PL_compiling.cop_arybase));
3791 yylval.opval->op_private = OPpCONST_ARYBASE;
3797 const char tmp = *s;
3798 if (PL_lex_state == LEX_NORMAL)
3801 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop)
3802 && intuit_more(s)) {
3804 PL_tokenbuf[0] = '@';
3805 if (ckWARN(WARN_SYNTAX)) {
3808 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3811 PL_bufptr = skipspace(PL_bufptr);
3812 while (t < PL_bufend && *t != ']')
3814 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3815 "Multidimensional syntax %.*s not supported",
3816 (t - PL_bufptr) + 1, PL_bufptr);
3820 else if (*s == '{') {
3822 PL_tokenbuf[0] = '%';
3823 if (strEQ(PL_tokenbuf+1, "SIG") && ckWARN(WARN_SYNTAX)
3824 && (t = strchr(s, '}')) && (t = strchr(t, '=')))
3826 char tmpbuf[sizeof PL_tokenbuf];
3827 for (t++; isSPACE(*t); t++) ;
3828 if (isIDFIRST_lazy_if(t,UTF)) {
3830 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE,
3832 for (; isSPACE(*t); t++) ;
3833 if (*t == ';' && get_cv(tmpbuf, FALSE))
3834 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3835 "You need to quote \"%s\"",
3842 PL_expect = XOPERATOR;
3843 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3844 const bool islop = (PL_last_lop == PL_oldoldbufptr);
3845 if (!islop || PL_last_lop_op == OP_GREPSTART)
3846 PL_expect = XOPERATOR;
3847 else if (strchr("$@\"'`q", *s))
3848 PL_expect = XTERM; /* e.g. print $fh "foo" */
3849 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3850 PL_expect = XTERM; /* e.g. print $fh &sub */
3851 else if (isIDFIRST_lazy_if(s,UTF)) {
3852 char tmpbuf[sizeof PL_tokenbuf];
3854 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3855 if ((t2 = keyword(tmpbuf, len))) {
3856 /* binary operators exclude handle interpretations */
3868 PL_expect = XTERM; /* e.g. print $fh length() */
3873 PL_expect = XTERM; /* e.g. print $fh subr() */
3876 else if (isDIGIT(*s))
3877 PL_expect = XTERM; /* e.g. print $fh 3 */
3878 else if (*s == '.' && isDIGIT(s[1]))
3879 PL_expect = XTERM; /* e.g. print $fh .3 */
3880 else if ((*s == '?' || *s == '-' || *s == '+')
3881 && !isSPACE(s[1]) && s[1] != '=')
3882 PL_expect = XTERM; /* e.g. print $fh -1 */
3883 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '='
3885 PL_expect = XTERM; /* e.g. print $fh /.../
3886 XXX except DORDOR operator
3888 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2])
3890 PL_expect = XTERM; /* print $fh <<"EOF" */
3893 PL_pending_ident = '$';
3897 if (PL_expect == XOPERATOR)
3899 PL_tokenbuf[0] = '@';
3900 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3901 if (!PL_tokenbuf[1]) {
3904 if (PL_lex_state == LEX_NORMAL)
3906 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3908 PL_tokenbuf[0] = '%';
3910 /* Warn about @ where they meant $. */
3911 if (*s == '[' || *s == '{') {
3912 if (ckWARN(WARN_SYNTAX)) {
3913 const char *t = s + 1;
3914 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3916 if (*t == '}' || *t == ']') {
3918 PL_bufptr = skipspace(PL_bufptr);
3919 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3920 "Scalar value %.*s better written as $%.*s",
3921 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3926 PL_pending_ident = '@';
3929 case '/': /* may be division, defined-or, or pattern */
3930 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
3934 case '?': /* may either be conditional or pattern */
3935 if(PL_expect == XOPERATOR) {
3943 /* A // operator. */
3953 /* Disable warning on "study /blah/" */
3954 if (PL_oldoldbufptr == PL_last_uni
3955 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3956 || memNE(PL_last_uni, "study", 5)
3957 || isALNUM_lazy_if(PL_last_uni+5,UTF)
3960 s = scan_pat(s,OP_MATCH);
3961 TERM(sublex_start());
3965 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3966 #ifdef PERL_STRICT_CR
3969 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3971 && (s == PL_linestart || s[-1] == '\n') )
3973 PL_lex_formbrack = 0;
3977 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3983 yylval.ival = OPf_SPECIAL;
3989 if (PL_expect != XOPERATOR)
3994 case '0': case '1': case '2': case '3': case '4':
3995 case '5': case '6': case '7': case '8': case '9':
3996 s = scan_num(s, &yylval);
3997 DEBUG_T( { S_printbuf(aTHX_ "### Saw number in %s\n", s); } );
3998 if (PL_expect == XOPERATOR)
4003 s = scan_str(s,FALSE,FALSE);
4004 DEBUG_T( { S_printbuf(aTHX_ "### Saw string before %s\n", s); } );
4005 if (PL_expect == XOPERATOR) {
4006 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
4009 return REPORT(','); /* grandfather non-comma-format format */
4015 missingterm((char*)0);
4016 yylval.ival = OP_CONST;
4017 TERM(sublex_start());
4020 s = scan_str(s,FALSE,FALSE);
4021 DEBUG_T( { S_printbuf(aTHX_ "### Saw string before %s\n", s); } );
4022 if (PL_expect == XOPERATOR) {
4023 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
4026 return REPORT(','); /* grandfather non-comma-format format */
4032 missingterm((char*)0);
4033 yylval.ival = OP_CONST;
4034 /* FIXME. I think that this can be const if char *d is replaced by
4035 more localised variables. */
4036 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
4037 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
4038 yylval.ival = OP_STRINGIFY;
4042 TERM(sublex_start());
4045 s = scan_str(s,FALSE,FALSE);
4046 DEBUG_T( { S_printbuf(aTHX_ "### Saw backtick string before %s\n", s); } );
4047 if (PL_expect == XOPERATOR)
4048 no_op("Backticks",s);
4050 missingterm((char*)0);
4051 yylval.ival = OP_BACKTICK;
4053 TERM(sublex_start());
4057 if (PL_lex_inwhat && isDIGIT(*s) && ckWARN(WARN_SYNTAX))
4058 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
4060 if (PL_expect == XOPERATOR)
4061 no_op("Backslash",s);
4065 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
4066 char *start = s + 2;
4067 while (isDIGIT(*start) || *start == '_')
4069 if (*start == '.' && isDIGIT(start[1])) {
4070 s = scan_num(s, &yylval);
4073 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
4074 else if (!isALPHA(*start) && (PL_expect == XTERM
4075 || PL_expect == XREF || PL_expect == XSTATE
4076 || PL_expect == XTERMORDORDOR)) {
4077 const char c = *start;
4080 gv = gv_fetchpv(s, 0, SVt_PVCV);
4083 s = scan_num(s, &yylval);
4090 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
4126 I32 orig_keyword = 0;
4131 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4133 /* Some keywords can be followed by any delimiter, including ':' */
4134 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
4135 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
4136 (PL_tokenbuf[0] == 'q' &&
4137 strchr("qwxr", PL_tokenbuf[1])))));
4139 /* x::* is just a word, unless x is "CORE" */
4140 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
4144 while (d < PL_bufend && isSPACE(*d))
4145 d++; /* no comments skipped here, or s### is misparsed */
4147 /* Is this a label? */
4148 if (!tmp && PL_expect == XSTATE
4149 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
4151 yylval.pval = savepv(PL_tokenbuf);
4156 /* Check for keywords */
4157 tmp = keyword(PL_tokenbuf, len);
4159 /* Is this a word before a => operator? */
4160 if (*d == '=' && d[1] == '>') {
4163 = (OP*)newSVOP(OP_CONST, 0,
4164 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
4165 yylval.opval->op_private = OPpCONST_BARE;
4169 if (tmp < 0) { /* second-class keyword? */
4170 GV *ogv = Nullgv; /* override (winner) */
4171 GV *hgv = Nullgv; /* hidden (loser) */
4172 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
4174 if ((gv = gv_fetchpv(PL_tokenbuf, 0, SVt_PVCV)) &&
4177 if (GvIMPORTED_CV(gv))
4179 else if (! CvMETHOD(cv))
4183 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
4184 (gv = *gvp) != (GV*)&PL_sv_undef &&
4185 GvCVu(gv) && GvIMPORTED_CV(gv))
4192 tmp = 0; /* overridden by import or by GLOBAL */
4195 && -tmp==KEY_lock /* XXX generalizable kludge */
4197 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
4199 tmp = 0; /* any sub overrides "weak" keyword */
4204 && PL_expect != XOPERATOR
4205 && PL_expect != XTERMORDORDOR)
4207 /* any sub overrides the "err" keyword, except when really an
4208 * operator is expected */
4211 else { /* no override */
4213 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
4214 Perl_warner(aTHX_ packWARN(WARN_MISC),
4215 "dump() better written as CORE::dump()");
4219 if (hgv && tmp != KEY_x && tmp != KEY_CORE
4220 && ckWARN(WARN_AMBIGUOUS)) /* never ambiguous */
4221 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4222 "Ambiguous call resolved as CORE::%s(), %s",
4223 GvENAME(hgv), "qualify as such or use &");
4230 default: /* not a keyword */
4231 /* Trade off - by using this evil construction we can pull the
4232 variable gv into the block labelled keylookup. If not, then
4233 we have to give it function scope so that the goto from the
4234 earlier ':' case doesn't bypass the initialisation. */
4236 just_a_word_zero_gv:
4243 const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
4246 /* Get the rest if it looks like a package qualifier */
4248 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
4250 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
4253 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
4254 *s == '\'' ? "'" : "::");
4259 if (PL_expect == XOPERATOR) {
4260 if (PL_bufptr == PL_linestart) {
4261 CopLINE_dec(PL_curcop);
4262 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
4263 CopLINE_inc(PL_curcop);
4266 no_op("Bareword",s);
4269 /* Look for a subroutine with this name in current package,
4270 unless name is "Foo::", in which case Foo is a bearword
4271 (and a package name). */
4274 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
4276 if (ckWARN(WARN_BAREWORD)
4277 && ! gv_fetchpv(PL_tokenbuf, 0, SVt_PVHV))
4278 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
4279 "Bareword \"%s\" refers to nonexistent package",
4282 PL_tokenbuf[len] = '\0';
4289 /* Mustn't actually add anything to a symbol table.
4290 But also don't want to "initialise" any placeholder
4291 constants that might already be there into full
4292 blown PVGVs with attached PVCV. */
4293 gv = gv_fetchpv(PL_tokenbuf, GV_NOADD_NOINIT,
4298 /* if we saw a global override before, get the right name */
4301 sv = newSVpvn("CORE::GLOBAL::",14);
4302 sv_catpv(sv,PL_tokenbuf);
4305 /* If len is 0, newSVpv does strlen(), which is correct.
4306 If len is non-zero, then it will be the true length,
4307 and so the scalar will be created correctly. */
4308 sv = newSVpv(PL_tokenbuf,len);
4311 /* Presume this is going to be a bareword of some sort. */
4314 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
4315 yylval.opval->op_private = OPpCONST_BARE;
4316 /* UTF-8 package name? */
4317 if (UTF && !IN_BYTES &&
4318 is_utf8_string((U8*)SvPVX_const(sv), SvCUR(sv)))
4321 /* And if "Foo::", then that's what it certainly is. */
4326 /* Do the explicit type check so that we don't need to force
4327 the initialisation of the symbol table to have a real GV.
4328 Beware - gv may not really be a PVGV, cv may not really be
4329 a PVCV, (because of the space optimisations that gv_init
4330 understands) But they're true if for this symbol there is
4331 respectively a typeglob and a subroutine.
4333 cv = gv ? ((SvTYPE(gv) == SVt_PVGV)
4334 /* Real typeglob, so get the real subroutine: */
4336 /* A proxy for a subroutine in this package? */
4337 : SvOK(gv) ? (CV *) gv : NULL)
4340 /* See if it's the indirect object for a list operator. */
4342 if (PL_oldoldbufptr &&
4343 PL_oldoldbufptr < PL_bufptr &&
4344 (PL_oldoldbufptr == PL_last_lop
4345 || PL_oldoldbufptr == PL_last_uni) &&
4346 /* NO SKIPSPACE BEFORE HERE! */
4347 (PL_expect == XREF ||
4348 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
4350 bool immediate_paren = *s == '(';
4352 /* (Now we can afford to cross potential line boundary.) */
4355 /* Two barewords in a row may indicate method call. */
4357 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') &&
4358 (tmp = intuit_method(s, gv, cv)))
4361 /* If not a declared subroutine, it's an indirect object. */
4362 /* (But it's an indir obj regardless for sort.) */
4363 /* Also, if "_" follows a filetest operator, it's a bareword */
4366 ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
4368 (PL_last_lop_op != OP_MAPSTART &&
4369 PL_last_lop_op != OP_GREPSTART))))
4370 || (PL_tokenbuf[0] == '_' && PL_tokenbuf[1] == '\0'
4371 && ((PL_opargs[PL_last_lop_op] & OA_CLASS_MASK) == OA_FILESTATOP))
4374 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
4379 PL_expect = XOPERATOR;
4382 /* Is this a word before a => operator? */
4383 if (*s == '=' && s[1] == '>' && !pkgname) {
4385 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
4386 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
4387 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
4391 /* If followed by a paren, it's certainly a subroutine. */
4395 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
4396 if (*d == ')' && (sv = gv_const_sv(gv))) {
4401 PL_nextval[PL_nexttoke].opval = yylval.opval;
4402 PL_expect = XOPERATOR;
4408 /* If followed by var or block, call it a method (unless sub) */
4410 if ((*s == '$' || *s == '{') && (!gv || !cv)) {
4411 PL_last_lop = PL_oldbufptr;
4412 PL_last_lop_op = OP_METHOD;
4416 /* If followed by a bareword, see if it looks like indir obj. */
4419 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
4420 && (tmp = intuit_method(s, gv, cv)))
4423 /* Not a method, so call it a subroutine (if defined) */
4426 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4427 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4428 "Ambiguous use of -%s resolved as -&%s()",
4429 PL_tokenbuf, PL_tokenbuf);
4430 /* Check for a constant sub */
4431 if ((sv = gv_const_sv(gv))) {
4433 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4434 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4435 yylval.opval->op_private = 0;
4439 /* Resolve to GV now. */
4440 if (SvTYPE(gv) != SVt_PVGV) {
4441 gv = gv_fetchpv(PL_tokenbuf, 0, SVt_PVCV);
4442 assert (SvTYPE(gv) == SVt_PVGV);
4443 /* cv must have been some sort of placeholder, so
4444 now needs replacing with a real code reference. */
4448 op_free(yylval.opval);
4449 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4450 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4451 PL_last_lop = PL_oldbufptr;
4452 PL_last_lop_op = OP_ENTERSUB;
4453 /* Is there a prototype? */
4456 const char *proto = SvPV_const((SV*)cv, len);
4459 if (*proto == '$' && proto[1] == '\0')
4461 while (*proto == ';')
4463 if (*proto == '&' && *s == '{') {
4464 sv_setpv(PL_subname, PL_curstash ?
4465 "__ANON__" : "__ANON__::__ANON__");
4469 PL_nextval[PL_nexttoke].opval = yylval.opval;
4475 /* Call it a bare word */
4477 if (PL_hints & HINT_STRICT_SUBS)
4478 yylval.opval->op_private |= OPpCONST_STRICT;
4481 if (lastchar != '-') {
4482 if (ckWARN(WARN_RESERVED)) {
4483 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4484 if (!*d && !gv_stashpv(PL_tokenbuf,FALSE))
4485 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
4492 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')
4493 && ckWARN_d(WARN_AMBIGUOUS)) {
4494 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4495 "Operator or semicolon missing before %c%s",
4496 lastchar, PL_tokenbuf);
4497 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4498 "Ambiguous use of %c resolved as operator %c",
4499 lastchar, lastchar);
4505 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4506 newSVpv(CopFILE(PL_curcop),0));
4510 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4511 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4514 case KEY___PACKAGE__:
4515 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4517 ? newSVhek(HvNAME_HEK(PL_curstash))
4524 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4525 const char *pname = "main";
4526 if (PL_tokenbuf[2] == 'D')
4527 pname = HvNAME_get(PL_curstash ? PL_curstash : PL_defstash);
4528 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), GV_ADD,
4532 GvIOp(gv) = newIO();
4533 IoIFP(GvIOp(gv)) = PL_rsfp;
4534 #if defined(HAS_FCNTL) && defined(F_SETFD)
4536 const int fd = PerlIO_fileno(PL_rsfp);
4537 fcntl(fd,F_SETFD,fd >= 3);
4540 /* Mark this internal pseudo-handle as clean */
4541 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4543 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4544 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4545 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4547 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4548 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4549 /* if the script was opened in binmode, we need to revert
4550 * it to text mode for compatibility; but only iff it has CRs
4551 * XXX this is a questionable hack at best. */
4552 if (PL_bufend-PL_bufptr > 2
4553 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4556 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4557 loc = PerlIO_tell(PL_rsfp);
4558 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4561 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4563 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4564 #endif /* NETWARE */
4565 #ifdef PERLIO_IS_STDIO /* really? */
4566 # if defined(__BORLANDC__)
4567 /* XXX see note in do_binmode() */
4568 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4572 PerlIO_seek(PL_rsfp, loc, 0);
4576 #ifdef PERLIO_LAYERS
4579 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4580 else if (PL_encoding) {
4587 XPUSHs(PL_encoding);
4589 call_method("name", G_SCALAR);
4593 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
4594 Perl_form(aTHX_ ":encoding(%"SVf")",
4612 if (PL_expect == XSTATE) {
4619 if (*s == ':' && s[1] == ':') {
4622 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4623 if (!(tmp = keyword(PL_tokenbuf, len)))
4624 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4627 else if (tmp == KEY_require || tmp == KEY_do)
4628 /* that's a way to remember we saw "CORE::" */
4641 LOP(OP_ACCEPT,XTERM);
4647 LOP(OP_ATAN2,XTERM);
4653 LOP(OP_BINMODE,XTERM);
4656 LOP(OP_BLESS,XTERM);
4665 /* When 'use switch' is in effect, continue has a dual
4666 life as a control operator. */
4668 if (!FEATURE_IS_ENABLED("switch", 6))
4671 /* We have to disambiguate the two senses of
4672 "continue". If the next token is a '{' then
4673 treat it as the start of a continue block;
4674 otherwise treat it as a control operator.
4685 (void)gv_fetchpv("ENV", GV_ADD, SVt_PVHV); /* may use HOME */
4702 if (!PL_cryptseen) {
4703 PL_cryptseen = TRUE;
4707 LOP(OP_CRYPT,XTERM);
4710 LOP(OP_CHMOD,XTERM);
4713 LOP(OP_CHOWN,XTERM);
4716 LOP(OP_CONNECT,XTERM);
4735 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4736 if (orig_keyword == KEY_do) {
4745 PL_hints |= HINT_BLOCK_SCOPE;
4755 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4756 LOP(OP_DBMOPEN,XTERM);
4762 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4769 yylval.ival = CopLINE(PL_curcop);
4783 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4784 UNIBRACK(OP_ENTEREVAL);
4802 case KEY_endhostent:
4808 case KEY_endservent:
4811 case KEY_endprotoent:
4822 yylval.ival = CopLINE(PL_curcop);
4824 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4826 if ((PL_bufend - p) >= 3 &&
4827 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4829 else if ((PL_bufend - p) >= 4 &&
4830 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4833 if (isIDFIRST_lazy_if(p,UTF)) {
4834 p = scan_ident(p, PL_bufend,
4835 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4839 Perl_croak(aTHX_ "Missing $ on loop variable");
4844 LOP(OP_FORMLINE,XTERM);
4850 LOP(OP_FCNTL,XTERM);
4856 LOP(OP_FLOCK,XTERM);
4865 LOP(OP_GREPSTART, XREF);
4868 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4883 case KEY_getpriority:
4884 LOP(OP_GETPRIORITY,XTERM);
4886 case KEY_getprotobyname:
4889 case KEY_getprotobynumber:
4890 LOP(OP_GPBYNUMBER,XTERM);
4892 case KEY_getprotoent:
4904 case KEY_getpeername:
4905 UNI(OP_GETPEERNAME);
4907 case KEY_gethostbyname:
4910 case KEY_gethostbyaddr:
4911 LOP(OP_GHBYADDR,XTERM);
4913 case KEY_gethostent:
4916 case KEY_getnetbyname:
4919 case KEY_getnetbyaddr:
4920 LOP(OP_GNBYADDR,XTERM);
4925 case KEY_getservbyname:
4926 LOP(OP_GSBYNAME,XTERM);
4928 case KEY_getservbyport:
4929 LOP(OP_GSBYPORT,XTERM);
4931 case KEY_getservent:
4934 case KEY_getsockname:
4935 UNI(OP_GETSOCKNAME);
4937 case KEY_getsockopt:
4938 LOP(OP_GSOCKOPT,XTERM);
4953 yylval.ival = CopLINE(PL_curcop);
4964 yylval.ival = CopLINE(PL_curcop);
4968 LOP(OP_INDEX,XTERM);
4974 LOP(OP_IOCTL,XTERM);
4986 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5018 LOP(OP_LISTEN,XTERM);
5027 s = scan_pat(s,OP_MATCH);
5028 TERM(sublex_start());
5031 LOP(OP_MAPSTART, XREF);
5034 LOP(OP_MKDIR,XTERM);
5037 LOP(OP_MSGCTL,XTERM);
5040 LOP(OP_MSGGET,XTERM);
5043 LOP(OP_MSGRCV,XTERM);
5046 LOP(OP_MSGSND,XTERM);
5052 if (isIDFIRST_lazy_if(s,UTF)) {
5053 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
5054 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
5056 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
5057 if (!PL_in_my_stash) {
5060 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
5068 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5075 s = tokenize_use(0, s);
5079 if (*s == '(' || (s = skipspace(s), *s == '('))
5086 if (isIDFIRST_lazy_if(s,UTF)) {
5088 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
5089 for (t=d; *t && isSPACE(*t); t++) ;
5090 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
5092 && !(t[0] == '=' && t[1] == '>')
5094 int len = (int)(d-s);
5095 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
5096 "Precedence problem: open %.*s should be open(%.*s)",
5103 yylval.ival = OP_OR;
5113 LOP(OP_OPEN_DIR,XTERM);
5116 checkcomma(s,PL_tokenbuf,"filehandle");
5120 checkcomma(s,PL_tokenbuf,"filehandle");
5139 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5143 LOP(OP_PIPE_OP,XTERM);
5146 s = scan_str(s,FALSE,FALSE);
5148 missingterm((char*)0);
5149 yylval.ival = OP_CONST;
5150 TERM(sublex_start());
5156 s = scan_str(s,FALSE,FALSE);
5158 missingterm((char*)0);
5159 PL_expect = XOPERATOR;
5161 if (SvCUR(PL_lex_stuff)) {
5164 d = SvPV_force(PL_lex_stuff, len);
5167 for (; isSPACE(*d) && len; --len, ++d) ;
5170 if (!warned && ckWARN(WARN_QW)) {
5171 for (; !isSPACE(*d) && len; --len, ++d) {
5173 Perl_warner(aTHX_ packWARN(WARN_QW),
5174 "Possible attempt to separate words with commas");
5177 else if (*d == '#') {
5178 Perl_warner(aTHX_ packWARN(WARN_QW),
5179 "Possible attempt to put comments in qw() list");
5185 for (; !isSPACE(*d) && len; --len, ++d) ;
5187 sv = newSVpvn(b, d-b);
5188 if (DO_UTF8(PL_lex_stuff))
5190 words = append_elem(OP_LIST, words,
5191 newSVOP(OP_CONST, 0, tokeq(sv)));
5195 PL_nextval[PL_nexttoke].opval = words;
5200 SvREFCNT_dec(PL_lex_stuff);
5201 PL_lex_stuff = Nullsv;
5207 s = scan_str(s,FALSE,FALSE);
5209 missingterm((char*)0);
5210 yylval.ival = OP_STRINGIFY;
5211 if (SvIVX(PL_lex_stuff) == '\'')
5212 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
5213 TERM(sublex_start());
5216 s = scan_pat(s,OP_QR);
5217 TERM(sublex_start());
5220 s = scan_str(s,FALSE,FALSE);
5222 missingterm((char*)0);
5223 yylval.ival = OP_BACKTICK;
5225 TERM(sublex_start());
5233 s = force_version(s, FALSE);
5235 else if (*s != 'v' || !isDIGIT(s[1])
5236 || (s = force_version(s, TRUE), *s == 'v'))
5238 *PL_tokenbuf = '\0';
5239 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5240 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
5241 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
5243 yyerror("<> should be quotes");
5245 if (orig_keyword == KEY_require) {
5253 PL_last_uni = PL_oldbufptr;
5254 PL_last_lop_op = OP_REQUIRE;
5256 return REPORT( (int)REQUIRE );
5262 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5266 LOP(OP_RENAME,XTERM);
5275 LOP(OP_RINDEX,XTERM);
5285 UNIDOR(OP_READLINE);
5298 LOP(OP_REVERSE,XTERM);
5301 UNIDOR(OP_READLINK);
5309 TERM(sublex_start());
5311 TOKEN(1); /* force error */
5314 checkcomma(s,PL_tokenbuf,"filehandle");
5324 LOP(OP_SELECT,XTERM);
5330 LOP(OP_SEMCTL,XTERM);
5333 LOP(OP_SEMGET,XTERM);
5336 LOP(OP_SEMOP,XTERM);
5342 LOP(OP_SETPGRP,XTERM);
5344 case KEY_setpriority:
5345 LOP(OP_SETPRIORITY,XTERM);
5347 case KEY_sethostent:
5353 case KEY_setservent:
5356 case KEY_setprotoent:
5366 LOP(OP_SEEKDIR,XTERM);
5368 case KEY_setsockopt:
5369 LOP(OP_SSOCKOPT,XTERM);
5375 LOP(OP_SHMCTL,XTERM);
5378 LOP(OP_SHMGET,XTERM);
5381 LOP(OP_SHMREAD,XTERM);
5384 LOP(OP_SHMWRITE,XTERM);
5387 LOP(OP_SHUTDOWN,XTERM);
5396 LOP(OP_SOCKET,XTERM);
5398 case KEY_socketpair:
5399 LOP(OP_SOCKPAIR,XTERM);
5402 checkcomma(s,PL_tokenbuf,"subroutine name");
5404 if (*s == ';' || *s == ')') /* probably a close */
5405 Perl_croak(aTHX_ "sort is now a reserved word");
5407 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5411 LOP(OP_SPLIT,XTERM);
5414 LOP(OP_SPRINTF,XTERM);
5417 LOP(OP_SPLICE,XTERM);
5432 LOP(OP_SUBSTR,XTERM);
5438 char tmpbuf[sizeof PL_tokenbuf];
5439 SSize_t tboffset = 0;
5440 expectation attrful;
5441 bool have_name, have_proto, bad_proto;
5442 const int key = tmp;
5446 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
5447 (*s == ':' && s[1] == ':'))
5450 attrful = XATTRBLOCK;
5451 /* remember buffer pos'n for later force_word */
5452 tboffset = s - PL_oldbufptr;
5453 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
5454 if (strchr(tmpbuf, ':'))
5455 sv_setpv(PL_subname, tmpbuf);
5457 sv_setsv(PL_subname,PL_curstname);
5458 sv_catpvn(PL_subname,"::",2);
5459 sv_catpvn(PL_subname,tmpbuf,len);
5466 Perl_croak(aTHX_ "Missing name in \"my sub\"");
5467 PL_expect = XTERMBLOCK;
5468 attrful = XATTRTERM;
5469 sv_setpvn(PL_subname,"?",1);
5473 if (key == KEY_format) {
5475 PL_lex_formbrack = PL_lex_brackets + 1;
5477 (void) force_word(PL_oldbufptr + tboffset, WORD,
5482 /* Look for a prototype */
5486 s = scan_str(s,FALSE,FALSE);
5488 Perl_croak(aTHX_ "Prototype not terminated");
5489 /* strip spaces and check for bad characters */
5490 d = SvPVX(PL_lex_stuff);
5493 for (p = d; *p; ++p) {
5496 if (!strchr("$@%*;[]&\\", *p))
5501 if (bad_proto && ckWARN(WARN_SYNTAX))
5502 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5503 "Illegal character in prototype for %"SVf" : %s",
5505 SvCUR_set(PL_lex_stuff, tmp);
5513 if (*s == ':' && s[1] != ':')
5514 PL_expect = attrful;
5515 else if (*s != '{' && key == KEY_sub) {
5517 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
5519 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, PL_subname);
5523 PL_nextval[PL_nexttoke].opval =
5524 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5525 PL_lex_stuff = Nullsv;
5529 sv_setpv(PL_subname,
5530 PL_curstash ? "__ANON__" : "__ANON__::__ANON__");
5533 (void) force_word(PL_oldbufptr + tboffset, WORD,
5542 LOP(OP_SYSTEM,XREF);
5545 LOP(OP_SYMLINK,XTERM);
5548 LOP(OP_SYSCALL,XTERM);
5551 LOP(OP_SYSOPEN,XTERM);
5554 LOP(OP_SYSSEEK,XTERM);
5557 LOP(OP_SYSREAD,XTERM);
5560 LOP(OP_SYSWRITE,XTERM);
5564 TERM(sublex_start());
5585 LOP(OP_TRUNCATE,XTERM);
5597 yylval.ival = CopLINE(PL_curcop);
5601 yylval.ival = CopLINE(PL_curcop);
5605 LOP(OP_UNLINK,XTERM);
5611 LOP(OP_UNPACK,XTERM);
5614 LOP(OP_UTIME,XTERM);
5620 LOP(OP_UNSHIFT,XTERM);
5623 s = tokenize_use(1, s);
5633 yylval.ival = CopLINE(PL_curcop);
5637 yylval.ival = CopLINE(PL_curcop);
5641 PL_hints |= HINT_BLOCK_SCOPE;
5648 LOP(OP_WAITPID,XTERM);
5657 ctl_l[0] = toCTRL('L');
5659 gv_fetchpv(ctl_l, GV_ADD, SVt_PV);
5662 gv_fetchpv("\f", GV_ADD, SVt_PV); /* Make sure $^L is defined */
5667 if (PL_expect == XOPERATOR)
5673 yylval.ival = OP_XOR;
5678 TERM(sublex_start());
5683 #pragma segment Main
5687 S_pending_ident(pTHX)
5690 register I32 tmp = 0;
5691 /* pit holds the identifier we read and pending_ident is reset */
5692 char pit = PL_pending_ident;
5693 PL_pending_ident = 0;
5695 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5696 "### Pending identifier '%s'\n", PL_tokenbuf); });
5698 /* if we're in a my(), we can't allow dynamics here.
5699 $foo'bar has already been turned into $foo::bar, so
5700 just check for colons.
5702 if it's a legal name, the OP is a PADANY.
5705 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5706 if (strchr(PL_tokenbuf,':'))
5707 yyerror(Perl_form(aTHX_ "No package name allowed for "
5708 "variable %s in \"our\"",
5710 tmp = allocmy(PL_tokenbuf);
5713 if (strchr(PL_tokenbuf,':'))
5714 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5716 yylval.opval = newOP(OP_PADANY, 0);
5717 yylval.opval->op_targ = allocmy(PL_tokenbuf);
5723 build the ops for accesses to a my() variable.
5725 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5726 then used in a comparison. This catches most, but not
5727 all cases. For instance, it catches
5728 sort { my($a); $a <=> $b }
5730 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5731 (although why you'd do that is anyone's guess).
5734 if (!strchr(PL_tokenbuf,':')) {
5736 tmp = pad_findmy(PL_tokenbuf);
5737 if (tmp != NOT_IN_PAD) {
5738 /* might be an "our" variable" */
5739 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
5740 /* build ops for a bareword */
5741 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
5742 HEK * const stashname = HvNAME_HEK(stash);
5743 SV * const sym = newSVhek(stashname);
5744 sv_catpvn(sym, "::", 2);
5745 sv_catpv(sym, PL_tokenbuf+1);
5746 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5747 yylval.opval->op_private = OPpCONST_ENTERED;
5750 ? (GV_ADDMULTI | GV_ADDINEVAL)
5753 ((PL_tokenbuf[0] == '$') ? SVt_PV
5754 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5759 /* if it's a sort block and they're naming $a or $b */
5760 if (PL_last_lop_op == OP_SORT &&
5761 PL_tokenbuf[0] == '$' &&
5762 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5765 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5766 d < PL_bufend && *d != '\n';
5769 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5770 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5776 yylval.opval = newOP(OP_PADANY, 0);
5777 yylval.opval->op_targ = tmp;
5783 Whine if they've said @foo in a doublequoted string,
5784 and @foo isn't a variable we can find in the symbol
5787 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5788 GV *gv = gv_fetchpv(PL_tokenbuf+1, 0, SVt_PVAV);
5789 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5790 && ckWARN(WARN_AMBIGUOUS))
5792 /* Downgraded from fatal to warning 20000522 mjd */
5793 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5794 "Possible unintended interpolation of %s in string",
5799 /* build ops for a bareword */
5800 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5801 yylval.opval->op_private = OPpCONST_ENTERED;
5805 ? (GV_ADDMULTI | GV_ADDINEVAL)
5806 /* If the identifier refers to a stash, don't autovivify it.
5807 * Change 24660 had the side effect of causing symbol table
5808 * hashes to always be defined, even if they were freshly
5809 * created and the only reference in the entire program was
5810 * the single statement with the defined %foo::bar:: test.
5811 * It appears that all code in the wild doing this actually
5812 * wants to know whether sub-packages have been loaded, so
5813 * by avoiding auto-vivifying symbol tables, we ensure that
5814 * defined %foo::bar:: continues to be false, and the existing
5815 * tests still give the expected answers, even though what
5816 * they're actually testing has now changed subtly.
5818 : !(*PL_tokenbuf == '%' && *(d = PL_tokenbuf + strlen(PL_tokenbuf) - 1) == ':' && d[-1] == ':'),
5819 ((PL_tokenbuf[0] == '$') ? SVt_PV
5820 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5826 * The following code was generated by perl_keyword.pl.
5830 Perl_keyword (pTHX_ const char *name, I32 len)
5834 case 1: /* 5 tokens of length 1 */
5866 case 2: /* 18 tokens of length 2 */
6012 case 3: /* 29 tokens of length 3 */
6016 if (name[1] == 'N' &&
6079 if (name[1] == 'i' &&
6119 if (name[1] == 'o' &&
6128 if (name[1] == 'e' &&
6137 if (name[1] == 'n' &&
6146 if (name[1] == 'o' &&
6155 if (name[1] == 'a' &&
6164 if (name[1] == 'o' &&
6226 if (name[1] == 'e' &&
6240 return (FEATURE_IS_ENABLED("say", 3) ? -KEY_say : 0);
6266 if (name[1] == 'i' &&
6275 if (name[1] == 's' &&
6284 if (name[1] == 'e' &&
6293 if (name[1] == 'o' &&
6305 case 4: /* 41 tokens of length 4 */
6309 if (name[1] == 'O' &&
6319 if (name[1] == 'N' &&
6329 if (name[1] == 'i' &&
6339 if (name[1] == 'h' &&
6349 if (name[1] == 'u' &&
6362 if (name[2] == 'c' &&
6371 if (name[2] == 's' &&
6380 if (name[2] == 'a' &&
6416 if (name[1] == 'o' &&
6429 if (name[2] == 't' &&
6438 if (name[2] == 'o' &&
6447 if (name[2] == 't' &&
6456 if (name[2] == 'e' &&
6469 if (name[1] == 'o' &&
6482 if (name[2] == 'y' &&
6491 if (name[2] == 'l' &&
6507 if (name[2] == 's' &&
6516 if (name[2] == 'n' &&
6525 if (name[2] == 'c' &&
6538 if (name[1] == 'e' &&
6548 if (name[1] == 'p' &&
6561 if (name[2] == 'c' &&
6570 if (name[2] == 'p' &&
6579 if (name[2] == 's' &&
6595 if (name[2] == 'n' &&
6665 if (name[2] == 'r' &&
6674 if (name[2] == 'r' &&
6683 if (name[2] == 'a' &&
6699 if (name[2] == 'l' &&
6761 if (name[2] == 'e' &&
6764 return (FEATURE_IS_ENABLED("switch", 6) ? KEY_when : 0);
6777 case 5: /* 38 tokens of length 5 */
6781 if (name[1] == 'E' &&
6792 if (name[1] == 'H' &&
6806 if (name[2] == 'a' &&
6816 if (name[2] == 'a' &&
6833 if (name[2] == 'e' &&
6843 if (name[2] == 'e' &&
6847 return (FEATURE_IS_ENABLED("switch", 6) ? -KEY_break : 0);
6863 if (name[3] == 'i' &&
6872 if (name[3] == 'o' &&
6908 if (name[2] == 'o' &&
6918 if (name[2] == 'y' &&
6932 if (name[1] == 'l' &&
6946 if (name[2] == 'n' &&
6956 if (name[2] == 'o' &&
6970 if (name[1] == 'i' &&
6975 return (FEATURE_IS_ENABLED("switch", 6) ? KEY_given : 0);
6984 if (name[2] == 'd' &&
6994 if (name[2] == 'c' &&
7011 if (name[2] == 'c' &&
7021 if (name[2] == 't' &&
7035 if (name[1] == 'k' &&
7046 if (name[1] == 'r' &&
7060 if (name[2] == 's' &&
7070 if (name[2] == 'd' &&
7087 if (name[2] == 'm' &&
7097 if (name[2] == 'i' &&
7107 if (name[2] == 'e' &&
7117 if (name[2] == 'l' &&
7127 if (name[2] == 'a' &&
7137 if (name[2] == 'u' &&
7151 if (name[1] == 'i' &&
7165 if (name[2] == 'a' &&
7178 if (name[3] == 'e' &&
7213 if (name[2] == 'i' &&
7230 if (name[2] == 'i' &&
7240 if (name[2] == 'i' &&
7257 case 6: /* 33 tokens of length 6 */
7261 if (name[1] == 'c' &&
7276 if (name[2] == 'l' &&
7287 if (name[2] == 'r' &&
7302 if (name[1] == 'e' &&
7317 if (name[2] == 's' &&
7322 if(ckWARN_d(WARN_SYNTAX))
7323 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
7329 if (name[2] == 'i' &&
7347 if (name[2] == 'l' &&
7358 if (name[2] == 'r' &&
7373 if (name[1] == 'm' &&
7388 if (name[2] == 'n' &&
7399 if (name[2] == 's' &&
7414 if (name[1] == 's' &&
7420 if (name[4] == 't' &&
7429 if (name[4] == 'e' &&
7438 if (name[4] == 'c' &&
7447 if (name[4] == 'n' &&
7463 if (name[1] == 'r' &&
7481 if (name[3] == 'a' &&
7491 if (name[3] == 'u' &&
7505 if (name[2] == 'n' &&
7523 if (name[2] == 'a' &&
7537 if (name[3] == 'e' &&
7550 if (name[4] == 't' &&
7559 if (name[4] == 'e' &&
7581 if (name[4] == 't' &&
7590 if (name[4] == 'e' &&
7606 if (name[2] == 'c' &&
7617 if (name[2] == 'l' &&
7628 if (name[2] == 'b' &&
7639 if (name[2] == 's' &&
7662 if (name[4] == 's' &&
7671 if (name[4] == 'n' &&
7684 if (name[3] == 'a' &&
7701 if (name[1] == 'a' &&
7716 case 7: /* 29 tokens of length 7 */
7720 if (name[1] == 'E' &&
7733 if (name[1] == '_' &&
7746 if (name[1] == 'i' &&
7753 return -KEY_binmode;
7759 if (name[1] == 'o' &&
7766 return -KEY_connect;
7775 if (name[2] == 'm' &&
7781 return -KEY_dbmopen;
7792 if (name[4] == 'u' &&
7796 return (FEATURE_IS_ENABLED("switch", 6) ? KEY_default : 0);
7802 if (name[4] == 'n' &&
7823 if (name[1] == 'o' &&
7836 if (name[1] == 'e' &&
7843 if (name[5] == 'r' &&
7846 return -KEY_getpgrp;
7852 if (name[5] == 'i' &&
7855 return -KEY_getppid;
7868 if (name[1] == 'c' &&
7875 return -KEY_lcfirst;
7881 if (name[1] == 'p' &&
7888 return -KEY_opendir;
7894 if (name[1] == 'a' &&
7912 if (name[3] == 'd' &&
7917 return -KEY_readdir;
7923 if (name[3] == 'u' &&
7934 if (name[3] == 'e' &&
7939 return -KEY_reverse;
7958 if (name[3] == 'k' &&
7963 return -KEY_seekdir;
7969 if (name[3] == 'p' &&
7974 return -KEY_setpgrp;
7984 if (name[2] == 'm' &&
7990 return -KEY_shmread;
7996 if (name[2] == 'r' &&
8002 return -KEY_sprintf;
8011 if (name[3] == 'l' &&
8016 return -KEY_symlink;
8025 if (name[4] == 'a' &&
8029 return -KEY_syscall;
8035 if (name[4] == 'p' &&
8039 return -KEY_sysopen;
8045 if (name[4] == 'e' &&
8049 return -KEY_sysread;
8055 if (name[4] == 'e' &&
8059 return -KEY_sysseek;
8077 if (name[1] == 'e' &&
8084 return -KEY_telldir;
8093 if (name[2] == 'f' &&
8099 return -KEY_ucfirst;
8105 if (name[2] == 's' &&
8111 return -KEY_unshift;
8121 if (name[1] == 'a' &&
8128 return -KEY_waitpid;
8137 case 8: /* 26 tokens of length 8 */
8141 if (name[1] == 'U' &&
8149 return KEY_AUTOLOAD;
8160 if (name[3] == 'A' &&
8166 return KEY___DATA__;
8172 if (name[3] == 'I' &&
8178 return -KEY___FILE__;
8184 if (name[3] == 'I' &&
8190 return -KEY___LINE__;
8206 if (name[2] == 'o' &&
8213 return -KEY_closedir;
8219 if (name[2] == 'n' &&
8226 return -KEY_continue;
8236 if (name[1] == 'b' &&
8244 return -KEY_dbmclose;
8250 if (name[1] == 'n' &&
8256 if (name[4] == 'r' &&
8261 return -KEY_endgrent;
8267 if (name[4] == 'w' &&
8272 return -KEY_endpwent;
8285 if (name[1] == 'o' &&
8293 return -KEY_formline;
8299 if (name[1] == 'e' &&
8310 if (name[6] == 'n' &&
8313 return -KEY_getgrent;
8319 if (name[6] == 'i' &&
8322 return -KEY_getgrgid;
8328 if (name[6] == 'a' &&
8331 return -KEY_getgrnam;
8344 if (name[4] == 'o' &&
8349 return -KEY_getlogin;
8360 if (name[6] == 'n' &&
8363 return -KEY_getpwent;
8369 if (name[6] == 'a' &&
8372 return -KEY_getpwnam;
8378 if (name[6] == 'i' &&
8381 return -KEY_getpwuid;
8401 if (name[1] == 'e' &&
8408 if (name[5] == 'i' &&
8415 return -KEY_readline;
8420 return -KEY_readlink;
8431 if (name[5] == 'i' &&
8435 return -KEY_readpipe;
8456 if (name[4] == 'r' &&
8461 return -KEY_setgrent;
8467 if (name[4] == 'w' &&
8472 return -KEY_setpwent;
8488 if (name[3] == 'w' &&
8494 return -KEY_shmwrite;
8500 if (name[3] == 't' &&
8506 return -KEY_shutdown;
8516 if (name[2] == 's' &&
8523 return -KEY_syswrite;
8533 if (name[1] == 'r' &&
8541 return -KEY_truncate;
8550 case 9: /* 8 tokens of length 9 */
8554 if (name[1] == 'n' &&
8563 return -KEY_endnetent;
8569 if (name[1] == 'e' &&
8578 return -KEY_getnetent;
8584 if (name[1] == 'o' &&
8593 return -KEY_localtime;
8599 if (name[1] == 'r' &&
8608 return KEY_prototype;
8614 if (name[1] == 'u' &&
8623 return -KEY_quotemeta;
8629 if (name[1] == 'e' &&
8638 return -KEY_rewinddir;
8644 if (name[1] == 'e' &&
8653 return -KEY_setnetent;
8659 if (name[1] == 'a' &&
8668 return -KEY_wantarray;
8677 case 10: /* 9 tokens of length 10 */
8681 if (name[1] == 'n' &&
8687 if (name[4] == 'o' &&
8694 return -KEY_endhostent;
8700 if (name[4] == 'e' &&
8707 return -KEY_endservent;
8720 if (name[1] == 'e' &&
8726 if (name[4] == 'o' &&
8733 return -KEY_gethostent;
8742 if (name[5] == 'r' &&
8748 return -KEY_getservent;
8754 if (name[5] == 'c' &&
8760 return -KEY_getsockopt;
8785 if (name[4] == 'o' &&
8792 return -KEY_sethostent;
8801 if (name[5] == 'r' &&
8807 return -KEY_setservent;
8813 if (name[5] == 'c' &&
8819 return -KEY_setsockopt;
8836 if (name[2] == 'c' &&
8845 return -KEY_socketpair;
8858 case 11: /* 8 tokens of length 11 */
8862 if (name[1] == '_' &&
8873 return -KEY___PACKAGE__;
8879 if (name[1] == 'n' &&
8890 return -KEY_endprotoent;
8896 if (name[1] == 'e' &&
8905 if (name[5] == 'e' &&
8912 return -KEY_getpeername;
8921 if (name[6] == 'o' &&
8927 return -KEY_getpriority;
8933 if (name[6] == 't' &&
8939 return -KEY_getprotoent;
8953 if (name[4] == 'o' &&
8961 return -KEY_getsockname;
8974 if (name[1] == 'e' &&
8982 if (name[6] == 'o' &&
8988 return -KEY_setpriority;
8994 if (name[6] == 't' &&
9000 return -KEY_setprotoent;
9016 case 12: /* 2 tokens of length 12 */
9017 if (name[0] == 'g' &&
9029 if (name[9] == 'd' &&
9032 { /* getnetbyaddr */
9033 return -KEY_getnetbyaddr;
9039 if (name[9] == 'a' &&
9042 { /* getnetbyname */
9043 return -KEY_getnetbyname;
9055 case 13: /* 4 tokens of length 13 */
9056 if (name[0] == 'g' &&
9063 if (name[4] == 'o' &&
9072 if (name[10] == 'd' &&
9075 { /* gethostbyaddr */
9076 return -KEY_gethostbyaddr;
9082 if (name[10] == 'a' &&
9085 { /* gethostbyname */
9086 return -KEY_gethostbyname;
9099 if (name[4] == 'e' &&
9108 if (name[10] == 'a' &&
9111 { /* getservbyname */
9112 return -KEY_getservbyname;
9118 if (name[10] == 'o' &&
9121 { /* getservbyport */
9122 return -KEY_getservbyport;
9141 case 14: /* 1 tokens of length 14 */
9142 if (name[0] == 'g' &&
9156 { /* getprotobyname */
9157 return -KEY_getprotobyname;
9162 case 16: /* 1 tokens of length 16 */
9163 if (name[0] == 'g' &&
9179 { /* getprotobynumber */
9180 return -KEY_getprotobynumber;
9194 S_checkcomma(pTHX_ register char *s, const char *name, const char *what)
9198 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
9199 if (ckWARN(WARN_SYNTAX)) {
9201 for (w = s+2; *w && level; w++) {
9208 for (; *w && isSPACE(*w); w++) ;
9209 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
9210 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
9211 "%s (...) interpreted as function",name);
9214 while (s < PL_bufend && isSPACE(*s))
9218 while (s < PL_bufend && isSPACE(*s))
9220 if (isIDFIRST_lazy_if(s,UTF)) {
9222 while (isALNUM_lazy_if(s,UTF))
9224 while (s < PL_bufend && isSPACE(*s))
9228 *s = '\0'; /* XXX If we didn't do this, we could const a lot of toke.c */
9229 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
9233 Perl_croak(aTHX_ "No comma allowed after %s", what);
9238 /* Either returns sv, or mortalizes sv and returns a new SV*.
9239 Best used as sv=new_constant(..., sv, ...).
9240 If s, pv are NULL, calls subroutine with one argument,
9241 and type is used with error messages only. */
9244 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, SV *sv, SV *pv,
9248 HV * const table = GvHV(PL_hintgv); /* ^H */
9252 const char *why1 = "", *why2 = "", *why3 = "";
9254 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
9257 why2 = strEQ(key,"charnames")
9258 ? "(possibly a missing \"use charnames ...\")"
9260 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
9261 (type ? type: "undef"), why2);
9263 /* This is convoluted and evil ("goto considered harmful")
9264 * but I do not understand the intricacies of all the different
9265 * failure modes of %^H in here. The goal here is to make
9266 * the most probable error message user-friendly. --jhi */
9271 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
9272 (type ? type: "undef"), why1, why2, why3);
9274 yyerror(SvPVX_const(msg));
9278 cvp = hv_fetch(table, key, strlen(key), FALSE);
9279 if (!cvp || !SvOK(*cvp)) {
9282 why3 = "} is not defined";
9285 sv_2mortal(sv); /* Parent created it permanently */
9288 pv = sv_2mortal(newSVpvn(s, len));
9290 typesv = sv_2mortal(newSVpv(type, 0));
9292 typesv = &PL_sv_undef;
9294 PUSHSTACKi(PERLSI_OVERLOAD);
9306 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
9310 /* Check the eval first */
9311 if (!PL_in_eval && SvTRUE(ERRSV)) {
9312 sv_catpv(ERRSV, "Propagated");
9313 yyerror(SvPV_nolen_const(ERRSV)); /* Duplicates the message inside eval */
9315 res = SvREFCNT_inc(sv);
9319 (void)SvREFCNT_inc(res);
9328 why1 = "Call to &{$^H{";
9330 why3 = "}} did not return a defined value";
9338 /* Returns a NUL terminated string, with the length of the string written to
9342 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
9344 register char *d = dest;
9345 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
9348 Perl_croak(aTHX_ ident_too_long);
9349 if (isALNUM(*s)) /* UTF handled below */
9351 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
9356 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
9360 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
9361 char *t = s + UTF8SKIP(s);
9362 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
9364 if (d + (t - s) > e)
9365 Perl_croak(aTHX_ ident_too_long);
9366 Copy(s, d, t - s, char);
9379 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
9383 char *bracket = Nullch;
9389 e = d + destlen - 3; /* two-character token, ending NUL */
9391 while (isDIGIT(*s)) {
9393 Perl_croak(aTHX_ ident_too_long);
9400 Perl_croak(aTHX_ ident_too_long);
9401 if (isALNUM(*s)) /* UTF handled below */
9403 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
9408 else if (*s == ':' && s[1] == ':') {
9412 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
9413 char *t = s + UTF8SKIP(s);
9414 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
9416 if (d + (t - s) > e)
9417 Perl_croak(aTHX_ ident_too_long);
9418 Copy(s, d, t - s, char);
9429 if (PL_lex_state != LEX_NORMAL)
9430 PL_lex_state = LEX_INTERPENDMAYBE;
9433 if (*s == '$' && s[1] &&
9434 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
9447 if (*d == '^' && *s && isCONTROLVAR(*s)) {
9452 if (isSPACE(s[-1])) {
9454 const char ch = *s++;
9455 if (!SPACE_OR_TAB(ch)) {
9461 if (isIDFIRST_lazy_if(d,UTF)) {
9465 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
9467 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
9470 Copy(s, d, e - s, char);
9475 while ((isALNUM(*s) || *s == ':') && d < e)
9478 Perl_croak(aTHX_ ident_too_long);
9481 while (s < send && SPACE_OR_TAB(*s)) s++;
9482 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
9483 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
9484 const char *brack = *s == '[' ? "[...]" : "{...}";
9485 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
9486 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
9487 funny, dest, brack, funny, dest, brack);
9490 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
9494 /* Handle extended ${^Foo} variables
9495 * 1999-02-27 mjd-perl-patch@plover.com */
9496 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
9500 while (isALNUM(*s) && d < e) {
9504 Perl_croak(aTHX_ ident_too_long);
9509 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
9510 PL_lex_state = LEX_INTERPEND;
9515 if (PL_lex_state == LEX_NORMAL) {
9516 if (ckWARN(WARN_AMBIGUOUS) &&
9517 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
9519 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
9520 "Ambiguous use of %c{%s} resolved to %c%s",
9521 funny, dest, funny, dest);
9526 s = bracket; /* let the parser handle it */
9530 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
9531 PL_lex_state = LEX_INTERPEND;
9536 Perl_pmflag(pTHX_ U32* pmfl, int ch)
9541 *pmfl |= PMf_GLOBAL;
9543 *pmfl |= PMf_CONTINUE;
9547 *pmfl |= PMf_MULTILINE;
9549 *pmfl |= PMf_SINGLELINE;
9551 *pmfl |= PMf_EXTENDED;
9555 S_scan_pat(pTHX_ char *start, I32 type)
9558 char *s = scan_str(start,FALSE,FALSE);
9561 char * const delimiter = skipspace(start);
9562 Perl_croak(aTHX_ *delimiter == '?'
9563 ? "Search pattern not terminated or ternary operator parsed as search pattern"
9564 : "Search pattern not terminated" );
9567 pm = (PMOP*)newPMOP(type, 0);
9568 if (PL_multi_open == '?')
9569 pm->op_pmflags |= PMf_ONCE;
9571 while (*s && strchr("iomsx", *s))
9572 pmflag(&pm->op_pmflags,*s++);
9575 while (*s && strchr("iogcmsx", *s))
9576 pmflag(&pm->op_pmflags,*s++);
9578 /* issue a warning if /c is specified,but /g is not */
9579 if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL)
9580 && ckWARN(WARN_REGEXP))
9582 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless without /g" );
9585 pm->op_pmpermflags = pm->op_pmflags;
9587 PL_lex_op = (OP*)pm;
9588 yylval.ival = OP_MATCH;
9593 S_scan_subst(pTHX_ char *start)
9601 yylval.ival = OP_NULL;
9603 s = scan_str(start,FALSE,FALSE);
9606 Perl_croak(aTHX_ "Substitution pattern not terminated");
9608 if (s[-1] == PL_multi_open)
9611 first_start = PL_multi_start;
9612 s = scan_str(s,FALSE,FALSE);
9615 SvREFCNT_dec(PL_lex_stuff);
9616 PL_lex_stuff = Nullsv;
9618 Perl_croak(aTHX_ "Substitution replacement not terminated");
9620 PL_multi_start = first_start; /* so whole substitution is taken together */
9622 pm = (PMOP*)newPMOP(OP_SUBST, 0);
9628 else if (strchr("iogcmsx", *s))
9629 pmflag(&pm->op_pmflags,*s++);
9634 if ((pm->op_pmflags & PMf_CONTINUE) && ckWARN(WARN_REGEXP)) {
9635 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless in s///" );
9640 PL_sublex_info.super_bufptr = s;
9641 PL_sublex_info.super_bufend = PL_bufend;
9643 pm->op_pmflags |= PMf_EVAL;
9644 repl = newSVpvn("",0);
9646 sv_catpv(repl, es ? "eval " : "do ");
9647 sv_catpvn(repl, "{ ", 2);
9648 sv_catsv(repl, PL_lex_repl);
9649 sv_catpvn(repl, " };", 2);
9651 SvREFCNT_dec(PL_lex_repl);
9655 pm->op_pmpermflags = pm->op_pmflags;
9656 PL_lex_op = (OP*)pm;
9657 yylval.ival = OP_SUBST;
9662 S_scan_trans(pTHX_ char *start)
9671 yylval.ival = OP_NULL;
9673 s = scan_str(start,FALSE,FALSE);
9675 Perl_croak(aTHX_ "Transliteration pattern not terminated");
9676 if (s[-1] == PL_multi_open)
9679 s = scan_str(s,FALSE,FALSE);
9682 SvREFCNT_dec(PL_lex_stuff);
9683 PL_lex_stuff = Nullsv;
9685 Perl_croak(aTHX_ "Transliteration replacement not terminated");
9688 complement = del = squash = 0;
9692 complement = OPpTRANS_COMPLEMENT;
9695 del = OPpTRANS_DELETE;
9698 squash = OPpTRANS_SQUASH;
9707 Newx(tbl, complement&&!del?258:256, short);
9708 o = newPVOP(OP_TRANS, 0, (char*)tbl);
9709 o->op_private &= ~OPpTRANS_ALL;
9710 o->op_private |= del|squash|complement|
9711 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
9712 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
9715 yylval.ival = OP_TRANS;
9720 S_scan_heredoc(pTHX_ register char *s)
9723 I32 op_type = OP_SCALAR;
9727 const char newline[] = "\n";
9728 const char *found_newline;
9732 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
9736 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
9739 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
9740 if (*peek == '`' || *peek == '\'' || *peek =='"') {
9743 s = delimcpy(d, e, s, PL_bufend, term, &len);
9753 if (!isALNUM_lazy_if(s,UTF))
9754 deprecate_old("bare << to mean <<\"\"");
9755 for (; isALNUM_lazy_if(s,UTF); s++) {
9760 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
9761 Perl_croak(aTHX_ "Delimiter for here document is too long");
9764 len = d - PL_tokenbuf;
9765 #ifndef PERL_STRICT_CR
9766 d = strchr(s, '\r');
9768 char * const olds = s;
9770 while (s < PL_bufend) {
9776 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
9785 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
9789 if ( outer || !(found_newline = ninstr(s,PL_bufend,newline,newline+1)) ) {
9790 herewas = newSVpvn(s,PL_bufend-s);
9794 herewas = newSVpvn(s,found_newline-s);
9796 s += SvCUR(herewas);
9798 tmpstr = NEWSV(87,79);
9799 sv_upgrade(tmpstr, SVt_PVIV);
9802 SvIV_set(tmpstr, -1);
9804 else if (term == '`') {
9805 op_type = OP_BACKTICK;
9806 SvIV_set(tmpstr, '\\');
9810 PL_multi_start = CopLINE(PL_curcop);
9811 PL_multi_open = PL_multi_close = '<';
9812 term = *PL_tokenbuf;
9813 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
9814 char *bufptr = PL_sublex_info.super_bufptr;
9815 char *bufend = PL_sublex_info.super_bufend;
9816 char * const olds = s - SvCUR(herewas);
9817 s = strchr(bufptr, '\n');
9821 while (s < bufend &&
9822 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
9824 CopLINE_inc(PL_curcop);
9827 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9828 missingterm(PL_tokenbuf);
9830 sv_setpvn(herewas,bufptr,d-bufptr+1);
9831 sv_setpvn(tmpstr,d+1,s-d);
9833 sv_catpvn(herewas,s,bufend-s);
9834 Copy(SvPVX_const(herewas),bufptr,SvCUR(herewas) + 1,char);
9841 while (s < PL_bufend &&
9842 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
9844 CopLINE_inc(PL_curcop);
9846 if (s >= PL_bufend) {
9847 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9848 missingterm(PL_tokenbuf);
9850 sv_setpvn(tmpstr,d+1,s-d);
9852 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
9854 sv_catpvn(herewas,s,PL_bufend-s);
9855 sv_setsv(PL_linestr,herewas);
9856 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
9857 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9858 PL_last_lop = PL_last_uni = Nullch;
9861 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
9862 while (s >= PL_bufend) { /* multiple line string? */
9864 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
9865 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9866 missingterm(PL_tokenbuf);
9868 CopLINE_inc(PL_curcop);
9869 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9870 PL_last_lop = PL_last_uni = Nullch;
9871 #ifndef PERL_STRICT_CR
9872 if (PL_bufend - PL_linestart >= 2) {
9873 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
9874 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
9876 PL_bufend[-2] = '\n';
9878 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
9880 else if (PL_bufend[-1] == '\r')
9881 PL_bufend[-1] = '\n';
9883 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
9884 PL_bufend[-1] = '\n';
9886 if (PERLDB_LINE && PL_curstash != PL_debstash) {
9887 SV *sv = NEWSV(88,0);
9889 sv_upgrade(sv, SVt_PVMG);
9890 sv_setsv(sv,PL_linestr);
9893 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
9895 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
9896 STRLEN off = PL_bufend - 1 - SvPVX_const(PL_linestr);
9897 *(SvPVX(PL_linestr) + off ) = ' ';
9898 sv_catsv(PL_linestr,herewas);
9899 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9900 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
9904 sv_catsv(tmpstr,PL_linestr);
9909 PL_multi_end = CopLINE(PL_curcop);
9910 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
9911 SvPV_shrink_to_cur(tmpstr);
9913 SvREFCNT_dec(herewas);
9915 if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
9917 else if (PL_encoding)
9918 sv_recode_to_utf8(tmpstr, PL_encoding);
9920 PL_lex_stuff = tmpstr;
9921 yylval.ival = op_type;
9926 takes: current position in input buffer
9927 returns: new position in input buffer
9928 side-effects: yylval and lex_op are set.
9933 <FH> read from filehandle
9934 <pkg::FH> read from package qualified filehandle
9935 <pkg'FH> read from package qualified filehandle
9936 <$fh> read from filehandle in $fh
9942 S_scan_inputsymbol(pTHX_ char *start)
9944 register char *s = start; /* current position in buffer */
9950 d = PL_tokenbuf; /* start of temp holding space */
9951 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
9952 end = strchr(s, '\n');
9955 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
9957 /* die if we didn't have space for the contents of the <>,
9958 or if it didn't end, or if we see a newline
9961 if (len >= sizeof PL_tokenbuf)
9962 Perl_croak(aTHX_ "Excessively long <> operator");
9964 Perl_croak(aTHX_ "Unterminated <> operator");
9969 Remember, only scalar variables are interpreted as filehandles by
9970 this code. Anything more complex (e.g., <$fh{$num}>) will be
9971 treated as a glob() call.
9972 This code makes use of the fact that except for the $ at the front,
9973 a scalar variable and a filehandle look the same.
9975 if (*d == '$' && d[1]) d++;
9977 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
9978 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
9981 /* If we've tried to read what we allow filehandles to look like, and
9982 there's still text left, then it must be a glob() and not a getline.
9983 Use scan_str to pull out the stuff between the <> and treat it
9984 as nothing more than a string.
9987 if (d - PL_tokenbuf != len) {
9988 yylval.ival = OP_GLOB;
9990 s = scan_str(start,FALSE,FALSE);
9992 Perl_croak(aTHX_ "Glob not terminated");
9996 bool readline_overriden = FALSE;
9997 GV *gv_readline = Nullgv;
9999 /* we're in a filehandle read situation */
10002 /* turn <> into <ARGV> */
10004 Copy("ARGV",d,5,char);
10006 /* Check whether readline() is overriden */
10007 if (((gv_readline = gv_fetchpv("readline", 0, SVt_PVCV))
10008 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
10010 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
10011 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
10012 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
10013 readline_overriden = TRUE;
10015 /* if <$fh>, create the ops to turn the variable into a
10021 /* try to find it in the pad for this block, otherwise find
10022 add symbol table ops
10024 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
10025 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
10026 HV *stash = PAD_COMPNAME_OURSTASH(tmp);
10027 HEK *stashname = HvNAME_HEK(stash);
10028 SV *sym = sv_2mortal(newSVhek(stashname));
10029 sv_catpvn(sym, "::", 2);
10030 sv_catpv(sym, d+1);
10035 OP *o = newOP(OP_PADSV, 0);
10037 PL_lex_op = readline_overriden
10038 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
10039 append_elem(OP_LIST, o,
10040 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
10041 : (OP*)newUNOP(OP_READLINE, 0, o);
10050 ? (GV_ADDMULTI | GV_ADDINEVAL)
10053 PL_lex_op = readline_overriden
10054 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
10055 append_elem(OP_LIST,
10056 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
10057 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
10058 : (OP*)newUNOP(OP_READLINE, 0,
10059 newUNOP(OP_RV2SV, 0,
10060 newGVOP(OP_GV, 0, gv)));
10062 if (!readline_overriden)
10063 PL_lex_op->op_flags |= OPf_SPECIAL;
10064 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
10065 yylval.ival = OP_NULL;
10068 /* If it's none of the above, it must be a literal filehandle
10069 (<Foo::BAR> or <FOO>) so build a simple readline OP */
10071 GV *gv = gv_fetchpv(d, GV_ADD, SVt_PVIO);
10072 PL_lex_op = readline_overriden
10073 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
10074 append_elem(OP_LIST,
10075 newGVOP(OP_GV, 0, gv),
10076 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
10077 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
10078 yylval.ival = OP_NULL;
10087 takes: start position in buffer
10088 keep_quoted preserve \ on the embedded delimiter(s)
10089 keep_delims preserve the delimiters around the string
10090 returns: position to continue reading from buffer
10091 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
10092 updates the read buffer.
10094 This subroutine pulls a string out of the input. It is called for:
10095 q single quotes q(literal text)
10096 ' single quotes 'literal text'
10097 qq double quotes qq(interpolate $here please)
10098 " double quotes "interpolate $here please"
10099 qx backticks qx(/bin/ls -l)
10100 ` backticks `/bin/ls -l`
10101 qw quote words @EXPORT_OK = qw( func() $spam )
10102 m// regexp match m/this/
10103 s/// regexp substitute s/this/that/
10104 tr/// string transliterate tr/this/that/
10105 y/// string transliterate y/this/that/
10106 ($*@) sub prototypes sub foo ($)
10107 (stuff) sub attr parameters sub foo : attr(stuff)
10108 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
10110 In most of these cases (all but <>, patterns and transliterate)
10111 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
10112 calls scan_str(). s/// makes yylex() call scan_subst() which calls
10113 scan_str(). tr/// and y/// make yylex() call scan_trans() which
10116 It skips whitespace before the string starts, and treats the first
10117 character as the delimiter. If the delimiter is one of ([{< then
10118 the corresponding "close" character )]}> is used as the closing
10119 delimiter. It allows quoting of delimiters, and if the string has
10120 balanced delimiters ([{<>}]) it allows nesting.
10122 On success, the SV with the resulting string is put into lex_stuff or,
10123 if that is already non-NULL, into lex_repl. The second case occurs only
10124 when parsing the RHS of the special constructs s/// and tr/// (y///).
10125 For convenience, the terminating delimiter character is stuffed into
10130 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
10132 SV *sv; /* scalar value: string */
10133 char *tmps; /* temp string, used for delimiter matching */
10134 register char *s = start; /* current position in the buffer */
10135 register char term; /* terminating character */
10136 register char *to; /* current position in the sv's data */
10137 I32 brackets = 1; /* bracket nesting level */
10138 bool has_utf8 = FALSE; /* is there any utf8 content? */
10139 I32 termcode; /* terminating char. code */
10140 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
10141 STRLEN termlen; /* length of terminating string */
10142 char *last = NULL; /* last position for nesting bracket */
10144 /* skip space before the delimiter */
10148 /* mark where we are, in case we need to report errors */
10151 /* after skipping whitespace, the next character is the terminator */
10154 termcode = termstr[0] = term;
10158 termcode = utf8_to_uvchr((U8*)s, &termlen);
10159 Copy(s, termstr, termlen, U8);
10160 if (!UTF8_IS_INVARIANT(term))
10164 /* mark where we are */
10165 PL_multi_start = CopLINE(PL_curcop);
10166 PL_multi_open = term;
10168 /* find corresponding closing delimiter */
10169 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
10170 termcode = termstr[0] = term = tmps[5];
10172 PL_multi_close = term;
10174 /* create a new SV to hold the contents. 87 is leak category, I'm
10175 assuming. 79 is the SV's initial length. What a random number. */
10177 sv_upgrade(sv, SVt_PVIV);
10178 SvIV_set(sv, termcode);
10179 (void)SvPOK_only(sv); /* validate pointer */
10181 /* move past delimiter and try to read a complete string */
10183 sv_catpvn(sv, s, termlen);
10186 if (PL_encoding && !UTF) {
10190 int offset = s - SvPVX_const(PL_linestr);
10191 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
10192 &offset, (char*)termstr, termlen);
10193 const char *ns = SvPVX_const(PL_linestr) + offset;
10194 char *svlast = SvEND(sv) - 1;
10196 for (; s < ns; s++) {
10197 if (*s == '\n' && !PL_rsfp)
10198 CopLINE_inc(PL_curcop);
10201 goto read_more_line;
10203 /* handle quoted delimiters */
10204 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
10206 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
10208 if ((svlast-1 - t) % 2) {
10209 if (!keep_quoted) {
10210 *(svlast-1) = term;
10212 SvCUR_set(sv, SvCUR(sv) - 1);
10217 if (PL_multi_open == PL_multi_close) {
10225 for (t = w = last; t < svlast; w++, t++) {
10226 /* At here, all closes are "was quoted" one,
10227 so we don't check PL_multi_close. */
10229 if (!keep_quoted && *(t+1) == PL_multi_open)
10234 else if (*t == PL_multi_open)
10242 SvCUR_set(sv, w - SvPVX_const(sv));
10245 if (--brackets <= 0)
10250 if (!keep_delims) {
10251 SvCUR_set(sv, SvCUR(sv) - 1);
10257 /* extend sv if need be */
10258 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
10259 /* set 'to' to the next character in the sv's string */
10260 to = SvPVX(sv)+SvCUR(sv);
10262 /* if open delimiter is the close delimiter read unbridle */
10263 if (PL_multi_open == PL_multi_close) {
10264 for (; s < PL_bufend; s++,to++) {
10265 /* embedded newlines increment the current line number */
10266 if (*s == '\n' && !PL_rsfp)
10267 CopLINE_inc(PL_curcop);
10268 /* handle quoted delimiters */
10269 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
10270 if (!keep_quoted && s[1] == term)
10272 /* any other quotes are simply copied straight through */
10276 /* terminate when run out of buffer (the for() condition), or
10277 have found the terminator */
10278 else if (*s == term) {
10281 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
10284 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
10290 /* if the terminator isn't the same as the start character (e.g.,
10291 matched brackets), we have to allow more in the quoting, and
10292 be prepared for nested brackets.
10295 /* read until we run out of string, or we find the terminator */
10296 for (; s < PL_bufend; s++,to++) {
10297 /* embedded newlines increment the line count */
10298 if (*s == '\n' && !PL_rsfp)
10299 CopLINE_inc(PL_curcop);
10300 /* backslashes can escape the open or closing characters */
10301 if (*s == '\\' && s+1 < PL_bufend) {
10302 if (!keep_quoted &&
10303 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
10308 /* allow nested opens and closes */
10309 else if (*s == PL_multi_close && --brackets <= 0)
10311 else if (*s == PL_multi_open)
10313 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
10318 /* terminate the copied string and update the sv's end-of-string */
10320 SvCUR_set(sv, to - SvPVX_const(sv));
10323 * this next chunk reads more into the buffer if we're not done yet
10327 break; /* handle case where we are done yet :-) */
10329 #ifndef PERL_STRICT_CR
10330 if (to - SvPVX_const(sv) >= 2) {
10331 if ((to[-2] == '\r' && to[-1] == '\n') ||
10332 (to[-2] == '\n' && to[-1] == '\r'))
10336 SvCUR_set(sv, to - SvPVX_const(sv));
10338 else if (to[-1] == '\r')
10341 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
10346 /* if we're out of file, or a read fails, bail and reset the current
10347 line marker so we can report where the unterminated string began
10350 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
10352 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
10355 /* we read a line, so increment our line counter */
10356 CopLINE_inc(PL_curcop);
10358 /* update debugger info */
10359 if (PERLDB_LINE && PL_curstash != PL_debstash) {
10360 SV *sv = NEWSV(88,0);
10362 sv_upgrade(sv, SVt_PVMG);
10363 sv_setsv(sv,PL_linestr);
10364 (void)SvIOK_on(sv);
10366 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
10369 /* having changed the buffer, we must update PL_bufend */
10370 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
10371 PL_last_lop = PL_last_uni = Nullch;
10374 /* at this point, we have successfully read the delimited string */
10376 if (!PL_encoding || UTF) {
10378 sv_catpvn(sv, s, termlen);
10381 if (has_utf8 || PL_encoding)
10384 PL_multi_end = CopLINE(PL_curcop);
10386 /* if we allocated too much space, give some back */
10387 if (SvCUR(sv) + 5 < SvLEN(sv)) {
10388 SvLEN_set(sv, SvCUR(sv) + 1);
10389 SvPV_renew(sv, SvLEN(sv));
10392 /* decide whether this is the first or second quoted string we've read
10405 takes: pointer to position in buffer
10406 returns: pointer to new position in buffer
10407 side-effects: builds ops for the constant in yylval.op
10409 Read a number in any of the formats that Perl accepts:
10411 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
10412 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
10415 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
10417 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
10420 If it reads a number without a decimal point or an exponent, it will
10421 try converting the number to an integer and see if it can do so
10422 without loss of precision.
10426 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
10428 register const char *s = start; /* current position in buffer */
10429 register char *d; /* destination in temp buffer */
10430 register char *e; /* end of temp buffer */
10431 NV nv; /* number read, as a double */
10432 SV *sv = Nullsv; /* place to put the converted number */
10433 bool floatit; /* boolean: int or float? */
10434 const char *lastub = 0; /* position of last underbar */
10435 static char const number_too_long[] = "Number too long";
10437 /* We use the first character to decide what type of number this is */
10441 Perl_croak(aTHX_ "panic: scan_num");
10443 /* if it starts with a 0, it could be an octal number, a decimal in
10444 0.13 disguise, or a hexadecimal number, or a binary number. */
10448 u holds the "number so far"
10449 shift the power of 2 of the base
10450 (hex == 4, octal == 3, binary == 1)
10451 overflowed was the number more than we can hold?
10453 Shift is used when we add a digit. It also serves as an "are
10454 we in octal/hex/binary?" indicator to disallow hex characters
10455 when in octal mode.
10460 bool overflowed = FALSE;
10461 bool just_zero = TRUE; /* just plain 0 or binary number? */
10462 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
10463 static const char* const bases[5] =
10464 { "", "binary", "", "octal", "hexadecimal" };
10465 static const char* const Bases[5] =
10466 { "", "Binary", "", "Octal", "Hexadecimal" };
10467 static const char* const maxima[5] =
10469 "0b11111111111111111111111111111111",
10473 const char *base, *Base, *max;
10475 /* check for hex */
10480 } else if (s[1] == 'b') {
10485 /* check for a decimal in disguise */
10486 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
10488 /* so it must be octal */
10495 if (ckWARN(WARN_SYNTAX))
10496 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10497 "Misplaced _ in number");
10501 base = bases[shift];
10502 Base = Bases[shift];
10503 max = maxima[shift];
10505 /* read the rest of the number */
10507 /* x is used in the overflow test,
10508 b is the digit we're adding on. */
10513 /* if we don't mention it, we're done */
10517 /* _ are ignored -- but warned about if consecutive */
10519 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10520 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10521 "Misplaced _ in number");
10525 /* 8 and 9 are not octal */
10526 case '8': case '9':
10528 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
10532 case '2': case '3': case '4':
10533 case '5': case '6': case '7':
10535 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
10538 case '0': case '1':
10539 b = *s++ & 15; /* ASCII digit -> value of digit */
10543 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
10544 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
10545 /* make sure they said 0x */
10548 b = (*s++ & 7) + 9;
10550 /* Prepare to put the digit we have onto the end
10551 of the number so far. We check for overflows.
10557 x = u << shift; /* make room for the digit */
10559 if ((x >> shift) != u
10560 && !(PL_hints & HINT_NEW_BINARY)) {
10563 if (ckWARN_d(WARN_OVERFLOW))
10564 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
10565 "Integer overflow in %s number",
10568 u = x | b; /* add the digit to the end */
10571 n *= nvshift[shift];
10572 /* If an NV has not enough bits in its
10573 * mantissa to represent an UV this summing of
10574 * small low-order numbers is a waste of time
10575 * (because the NV cannot preserve the
10576 * low-order bits anyway): we could just
10577 * remember when did we overflow and in the
10578 * end just multiply n by the right
10586 /* if we get here, we had success: make a scalar value from
10591 /* final misplaced underbar check */
10592 if (s[-1] == '_') {
10593 if (ckWARN(WARN_SYNTAX))
10594 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
10599 if (n > 4294967295.0 && ckWARN(WARN_PORTABLE))
10600 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
10601 "%s number > %s non-portable",
10607 if (u > 0xffffffff && ckWARN(WARN_PORTABLE))
10608 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
10609 "%s number > %s non-portable",
10614 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
10615 sv = new_constant(start, s - start, "integer",
10617 else if (PL_hints & HINT_NEW_BINARY)
10618 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
10623 handle decimal numbers.
10624 we're also sent here when we read a 0 as the first digit
10626 case '1': case '2': case '3': case '4': case '5':
10627 case '6': case '7': case '8': case '9': case '.':
10630 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
10633 /* read next group of digits and _ and copy into d */
10634 while (isDIGIT(*s) || *s == '_') {
10635 /* skip underscores, checking for misplaced ones
10639 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10640 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10641 "Misplaced _ in number");
10645 /* check for end of fixed-length buffer */
10647 Perl_croak(aTHX_ number_too_long);
10648 /* if we're ok, copy the character */
10653 /* final misplaced underbar check */
10654 if (lastub && s == lastub + 1) {
10655 if (ckWARN(WARN_SYNTAX))
10656 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
10659 /* read a decimal portion if there is one. avoid
10660 3..5 being interpreted as the number 3. followed
10663 if (*s == '.' && s[1] != '.') {
10668 if (ckWARN(WARN_SYNTAX))
10669 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10670 "Misplaced _ in number");
10674 /* copy, ignoring underbars, until we run out of digits.
10676 for (; isDIGIT(*s) || *s == '_'; s++) {
10677 /* fixed length buffer check */
10679 Perl_croak(aTHX_ number_too_long);
10681 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10682 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10683 "Misplaced _ in number");
10689 /* fractional part ending in underbar? */
10690 if (s[-1] == '_') {
10691 if (ckWARN(WARN_SYNTAX))
10692 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10693 "Misplaced _ in number");
10695 if (*s == '.' && isDIGIT(s[1])) {
10696 /* oops, it's really a v-string, but without the "v" */
10702 /* read exponent part, if present */
10703 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
10707 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
10708 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
10710 /* stray preinitial _ */
10712 if (ckWARN(WARN_SYNTAX))
10713 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10714 "Misplaced _ in number");
10718 /* allow positive or negative exponent */
10719 if (*s == '+' || *s == '-')
10722 /* stray initial _ */
10724 if (ckWARN(WARN_SYNTAX))
10725 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10726 "Misplaced _ in number");
10730 /* read digits of exponent */
10731 while (isDIGIT(*s) || *s == '_') {
10734 Perl_croak(aTHX_ number_too_long);
10738 if (((lastub && s == lastub + 1) ||
10739 (!isDIGIT(s[1]) && s[1] != '_'))
10740 && ckWARN(WARN_SYNTAX))
10741 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10742 "Misplaced _ in number");
10749 /* make an sv from the string */
10753 We try to do an integer conversion first if no characters
10754 indicating "float" have been found.
10759 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
10761 if (flags == IS_NUMBER_IN_UV) {
10763 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
10766 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
10767 if (uv <= (UV) IV_MIN)
10768 sv_setiv(sv, -(IV)uv);
10775 /* terminate the string */
10777 nv = Atof(PL_tokenbuf);
10781 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
10782 (PL_hints & HINT_NEW_INTEGER) )
10783 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
10784 (floatit ? "float" : "integer"),
10788 /* if it starts with a v, it could be a v-string */
10791 sv = NEWSV(92,5); /* preallocate storage space */
10792 s = scan_vstring(s,sv);
10796 /* make the op for the constant and return */
10799 lvalp->opval = newSVOP(OP_CONST, 0, sv);
10801 lvalp->opval = Nullop;
10807 S_scan_formline(pTHX_ register char *s)
10809 register char *eol;
10811 SV *stuff = newSVpvn("",0);
10812 bool needargs = FALSE;
10813 bool eofmt = FALSE;
10815 while (!needargs) {
10817 #ifdef PERL_STRICT_CR
10818 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
10820 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
10822 if (*t == '\n' || t == PL_bufend) {
10827 if (PL_in_eval && !PL_rsfp) {
10828 eol = (char *) memchr(s,'\n',PL_bufend-s);
10833 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
10835 for (t = s; t < eol; t++) {
10836 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
10838 goto enough; /* ~~ must be first line in formline */
10840 if (*t == '@' || *t == '^')
10844 sv_catpvn(stuff, s, eol-s);
10845 #ifndef PERL_STRICT_CR
10846 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
10847 char *end = SvPVX(stuff) + SvCUR(stuff);
10850 SvCUR_set(stuff, SvCUR(stuff) - 1);
10859 s = filter_gets(PL_linestr, PL_rsfp, 0);
10860 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
10861 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
10862 PL_last_lop = PL_last_uni = Nullch;
10871 if (SvCUR(stuff)) {
10874 PL_lex_state = LEX_NORMAL;
10875 PL_nextval[PL_nexttoke].ival = 0;
10879 PL_lex_state = LEX_FORMLINE;
10881 if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
10883 else if (PL_encoding)
10884 sv_recode_to_utf8(stuff, PL_encoding);
10886 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
10888 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
10892 SvREFCNT_dec(stuff);
10894 PL_lex_formbrack = 0;
10905 PL_cshlen = strlen(PL_cshname);
10910 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
10912 const I32 oldsavestack_ix = PL_savestack_ix;
10913 CV* outsidecv = PL_compcv;
10916 assert(SvTYPE(PL_compcv) == SVt_PVCV);
10918 SAVEI32(PL_subline);
10919 save_item(PL_subname);
10920 SAVESPTR(PL_compcv);
10922 PL_compcv = (CV*)NEWSV(1104,0);
10923 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
10924 CvFLAGS(PL_compcv) |= flags;
10926 PL_subline = CopLINE(PL_curcop);
10927 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
10928 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
10929 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
10931 return oldsavestack_ix;
10935 #pragma segment Perl_yylex
10938 Perl_yywarn(pTHX_ const char *s)
10940 PL_in_eval |= EVAL_WARNONLY;
10942 PL_in_eval &= ~EVAL_WARNONLY;
10947 Perl_yyerror(pTHX_ const char *s)
10949 const char *where = NULL;
10950 const char *context = NULL;
10954 if (!yychar || (yychar == ';' && !PL_rsfp))
10956 else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
10957 PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
10958 PL_oldbufptr != PL_bufptr) {
10961 The code below is removed for NetWare because it abends/crashes on NetWare
10962 when the script has error such as not having the closing quotes like:
10963 if ($var eq "value)
10964 Checking of white spaces is anyway done in NetWare code.
10967 while (isSPACE(*PL_oldoldbufptr))
10970 context = PL_oldoldbufptr;
10971 contlen = PL_bufptr - PL_oldoldbufptr;
10973 else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
10974 PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
10977 The code below is removed for NetWare because it abends/crashes on NetWare
10978 when the script has error such as not having the closing quotes like:
10979 if ($var eq "value)
10980 Checking of white spaces is anyway done in NetWare code.
10983 while (isSPACE(*PL_oldbufptr))
10986 context = PL_oldbufptr;
10987 contlen = PL_bufptr - PL_oldbufptr;
10989 else if (yychar > 255)
10990 where = "next token ???";
10991 else if (yychar == -2) { /* YYEMPTY */
10992 if (PL_lex_state == LEX_NORMAL ||
10993 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
10994 where = "at end of line";
10995 else if (PL_lex_inpat)
10996 where = "within pattern";
10998 where = "within string";
11001 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
11003 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
11004 else if (isPRINT_LC(yychar))
11005 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
11007 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
11008 where = SvPVX_const(where_sv);
11010 msg = sv_2mortal(newSVpv(s, 0));
11011 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
11012 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
11014 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
11016 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
11017 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
11018 Perl_sv_catpvf(aTHX_ msg,
11019 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
11020 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
11023 if (PL_in_eval & EVAL_WARNONLY && ckWARN_d(WARN_SYNTAX))
11024 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, msg);
11027 if (PL_error_count >= 10) {
11028 if (PL_in_eval && SvCUR(ERRSV))
11029 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
11030 ERRSV, OutCopFILE(PL_curcop));
11032 Perl_croak(aTHX_ "%s has too many errors.\n",
11033 OutCopFILE(PL_curcop));
11036 PL_in_my_stash = NULL;
11040 #pragma segment Main
11044 S_swallow_bom(pTHX_ U8 *s)
11046 const STRLEN slen = SvCUR(PL_linestr);
11049 if (s[1] == 0xFE) {
11050 /* UTF-16 little-endian? (or UTF32-LE?) */
11051 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
11052 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
11053 #ifndef PERL_NO_UTF16_FILTER
11054 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
11057 if (PL_bufend > (char*)s) {
11061 filter_add(utf16rev_textfilter, NULL);
11062 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
11063 utf16_to_utf8_reversed(s, news,
11064 PL_bufend - (char*)s - 1,
11066 sv_setpvn(PL_linestr, (const char*)news, newlen);
11068 SvUTF8_on(PL_linestr);
11069 s = (U8*)SvPVX(PL_linestr);
11070 PL_bufend = SvPVX(PL_linestr) + newlen;
11073 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
11078 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
11079 #ifndef PERL_NO_UTF16_FILTER
11080 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
11083 if (PL_bufend > (char *)s) {
11087 filter_add(utf16_textfilter, NULL);
11088 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
11089 utf16_to_utf8(s, news,
11090 PL_bufend - (char*)s,
11092 sv_setpvn(PL_linestr, (const char*)news, newlen);
11094 SvUTF8_on(PL_linestr);
11095 s = (U8*)SvPVX(PL_linestr);
11096 PL_bufend = SvPVX(PL_linestr) + newlen;
11099 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
11104 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
11105 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
11106 s += 3; /* UTF-8 */
11112 if (s[2] == 0xFE && s[3] == 0xFF) {
11113 /* UTF-32 big-endian */
11114 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
11117 else if (s[2] == 0 && s[3] != 0) {
11120 * are a good indicator of UTF-16BE. */
11121 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
11126 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
11129 * are a good indicator of UTF-16LE. */
11130 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
11139 * Restore a source filter.
11143 restore_rsfp(pTHX_ void *f)
11145 PerlIO * const fp = (PerlIO*)f;
11147 if (PL_rsfp == PerlIO_stdin())
11148 PerlIO_clearerr(PL_rsfp);
11149 else if (PL_rsfp && (PL_rsfp != fp))
11150 PerlIO_close(PL_rsfp);
11154 #ifndef PERL_NO_UTF16_FILTER
11156 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
11158 const STRLEN old = SvCUR(sv);
11159 const I32 count = FILTER_READ(idx+1, sv, maxlen);
11160 DEBUG_P(PerlIO_printf(Perl_debug_log,
11161 "utf16_textfilter(%p): %d %d (%d)\n",
11162 utf16_textfilter, idx, maxlen, (int) count));
11166 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
11167 Copy(SvPVX_const(sv), tmps, old, char);
11168 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
11169 SvCUR(sv) - old, &newlen);
11170 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
11172 DEBUG_P({sv_dump(sv);});
11177 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
11179 const STRLEN old = SvCUR(sv);
11180 const I32 count = FILTER_READ(idx+1, sv, maxlen);
11181 DEBUG_P(PerlIO_printf(Perl_debug_log,
11182 "utf16rev_textfilter(%p): %d %d (%d)\n",
11183 utf16rev_textfilter, idx, maxlen, (int) count));
11187 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
11188 Copy(SvPVX_const(sv), tmps, old, char);
11189 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
11190 SvCUR(sv) - old, &newlen);
11191 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
11193 DEBUG_P({ sv_dump(sv); });
11199 Returns a pointer to the next character after the parsed
11200 vstring, as well as updating the passed in sv.
11202 Function must be called like
11205 s = scan_vstring(s,sv);
11207 The sv should already be large enough to store the vstring
11208 passed in, for performance reasons.
11213 Perl_scan_vstring(pTHX_ const char *s, SV *sv)
11215 const char *pos = s;
11216 const char *start = s;
11217 if (*pos == 'v') pos++; /* get past 'v' */
11218 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
11220 if ( *pos != '.') {
11221 /* this may not be a v-string if followed by => */
11222 const char *next = pos;
11223 while (next < PL_bufend && isSPACE(*next))
11225 if ((PL_bufend - next) >= 2 && *next == '=' && next[1] == '>' ) {
11226 /* return string not v-string */
11227 sv_setpvn(sv,(char *)s,pos-s);
11228 return (char *)pos;
11232 if (!isALPHA(*pos)) {
11233 U8 tmpbuf[UTF8_MAXBYTES+1];
11235 if (*s == 'v') s++; /* get past 'v' */
11237 sv_setpvn(sv, "", 0);
11243 /* this is atoi() that tolerates underscores */
11244 const char *end = pos;
11246 while (--end >= s) {
11251 rev += (*end - '0') * mult;
11253 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
11254 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
11255 "Integer overflow in decimal number");
11259 if (rev > 0x7FFFFFFF)
11260 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
11262 /* Append native character for the rev point */
11263 tmpend = uvchr_to_utf8(tmpbuf, rev);
11264 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
11265 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
11267 if (pos + 1 < PL_bufend && *pos == '.' && isDIGIT(pos[1]))
11273 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
11277 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
11285 * c-indentation-style: bsd
11286 * c-basic-offset: 4
11287 * indent-tabs-mode: t
11290 * ex: set ts=8 sts=4 sw=4 noet: